DETAILED ACTION
This is a Final Office action is in response to communications filed on March 30th, 2026. Claim 1-2, 5-7, 11, 15, 16-17, and 20 is amended and claims 4, 8, and 19 is/are cancelled. Claims 1-3, 5-7, 9-18, and 20 have been examined in this application.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-3, 5-7, 9-18, and 20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e. an abstract idea) without significantly more.
Step 1: Claims 1-15 is/are drawn to method (i.e., a process) and claims 13-14 is/are drawn to electronic device (i.e. system) (i.e., a manufacture). (Step 1: YES).
Step 2A - Prong One: In prong one of step 2A, the claim(s) is/are analyzed to evaluate whether it/they recite(s) a judicial exception.
Claim 1: A method for providing product objects information comprising:
generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first artificial intelligence (AI) model, wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions;
identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content;
determining national or regional attribute information of the target user;
processing the original descriptive information to adapt to local expression based on the target user's national or regional attribute information to generate target descriptive information, wherein the processing comprises:
processing the original textual content via an AI large-scale parameter model to generate target textual content by replacing keywords in the original textual content with corresponding local common terms queried from the pre-established knowledge base,
and adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information;
and providing the target descriptive information corresponding to the at least one target product object to a client device of the target user to provide the target descriptive information on a designated webpage.
Claim 15: A method for providing product object information, comprising:
receiving target descriptive information of at least one target product object provided for a target user from a server, the target descriptive information is generated by;
processing original textual content of the target product object via an artificial intelligence (AI) large-scale parameter model to generate target textual content by replacing keywords in the original textual content with corresponding local common terms queried from a pre-established knowledge base,
and adapting the target textual content to conform to local grammatical expression habits corresponding to national or regional attribute information of the target user,
wherein the pre-established knowledge base is generated by inputting small-scale samples of regional terms into a first AI model, wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with the local common terms and the local grammatical expression habits for multiple regions;
and displaying the target descriptive information of at least one target product object comprising the target textual content on a designated webpage.
(Examiner notes: The underlined claim terms above are interpreted as additional elements beyond the abstract idea and are further analyzed under Step 2A - Prong Two)
Under their broadest reasonable interpretation, the claims recite a method of generating a pre-established knowledge base from regional terms, identifying a target product object and original descriptive information, determining national or regional attribute information of a target user, processing the original descriptive information to adapt it to local expression based on the user’s national or regional attribute information, replacing keywords with local common terms, adapting textual content to local grammatical expression habits, and providing the resulting target descriptive information to a client device for display on a designated webpage i.e., collecting information (user region), analyzing/processing information to produce a localized version, and presenting the results, which constitutes an abstract idea in the form of mental processes and/or certain methods of organizing human activity (marketing/personalized content presentation). Similarly, claim 15 recites receiving target descriptive information from a server, where the target descriptive information is generated by processing original textual content using an AI large-scale parameter model, replacing keywords with local common terms queried from a pre-established knowledge base, adapting the target textual content to local grammatical expression habits corresponding to national or regional attribute information of a target user, and displaying the target descriptive information on a designated webpage that can be performed as a mental process (conceptually deciding how to rewrite product descriptions for a locale presentation) and/or as a method of organizing activity such as localized merchandising/marketing and customer-facing information presentation. From applicant’s specification, the claimed invention is implemented a product information service system, “customized experience for each individual” typically refers to the strategy of tailoring different product recommendations for different consumers to achieve better sales outcomes. Traditional product recommendations are based on the attributes of the products themselves, such as category, price, sales volume, and other factors. (see at least [0003] of instant specification). Under the broadest reasonable interpretation, the independent claims recite the abstract idea of organizing, modifying, and presenting product information based on user attributes. More specifically, the claims are directed to collecting or identifying information, including product information and user regional information; analyzing that information to determine how product text should be localized; transforming the product information into localized descriptive content; and presenting the transformed information to the user. Such activities fall within the abstract idea groupings of mental processes and certain methods of organizing human activity, including marketing, advertising, sales activity, and personalized content presentation. The recited localization of product descriptions could be performed conceptually by a person determining a user’s region, selecting locally appropriate terms for product text, revising the product text according to regional grammar or style preferences, and presenting the revised product description to the user. The Examiner notes that although the claim limitations are summarized, the analysis regarding subject matter eligibility considers the entirety of the claim and all of the claim elements individually, as a whole, and in ordered combination.
And the dependent claims 2-3, 5-7, 9-18, and 20 recites an abstract idea, recite using the AI large-scale parameter model to comprehend and process descriptive information. Claims 3 and 18 recite applying the localization to products associated with multiple SKUs. Claims 5 and 20 recite converting keywords related to product names or adjectives into local common terms. Claim 6 recites processing title text based on attribute preferences of a demographic. Claim 7 recites reordering user reviews to prioritize local reviews. Claims 9-11 recite transforming product attribute or parameter descriptions based on national or regional attribute information. Claims 12-14 recite transforming rich media, image information, or audio information based on national or regional attribute information. These limitations remain directed to selecting, transforming, reordering, or presenting descriptive product content based on user region or locale, which is a form of information analysis, content customization, and localized marketing or presentation. Such activities fall within the abstract idea groupings of mental processes and certain methods of organizing human activity, including marketing, advertising, sales activity, and personalized content presentation.
Independent claim(s) 16 recite/describe nearly identical steps (and therefore also recite limitations that fall within this subject matter grouping of abstract ideas), and this/these claim(s) is/are therefore determined to recite an abstract idea under the same analysis.
As such, the Examiner concludes that claim 1 recites an abstract idea (Step 2A – Prong One: YES).
Step 2A - Prong Two: In prong two of step 2A, an evaluation is made whether a claim recites any additional element, or combination of additional elements, that integrate the exception into a practical application of that exception. An “addition element” is an element that is recited in the claim in addition to (beyond) the judicial exception (i.e., an element/limitation that sets forth an abstract idea is not an additional element). The phrase “integration into a practical application” is defined as requiring an additional element or a combination of additional elements in the claim to apply, rely on, or use the judicial exception in a manner that imposes a meaningful limit on the judicial exception, such that it is more than a drafting effort designed to monopolize the exception.
The requirement to execute the claimed steps/functions using a server, client device, processors; AI models, and one or more computer-readable memories, etc. (Claims 1 and 15-16) is/are equivalent to adding the words “apply it” on a generic computer and/or mere instructions to implement the abstract idea on a generic computer.
Similarly, the limitations of using a server, client device, processors; and one or more computer-readable memories, etc. (Claims 1 and 15-16, and dependent claims 2-3, 5-7, 9-14, 17-18, and 20) are recited at a high level of generality and amount to no more than mere instructions to apply the exception using generic computer components. This/these limitation(s) do/does not impose any meaningful limits on practicing the abstract idea, and therefore do/does not integrate the abstract idea into a practical application (see MPEP 2106.05(f)).
Further, the additional limitations beyond the abstract idea identified above, serves merely to generally link the use of the judicial exception to a particular technological environment or field of use. Specifically, it/they serve(s) to limit the application of the abstract idea to computerized environments (e.g., receive, identify, generate, determine, display, etc. steps performed by a server, client device, processors; and one or more computer-readable memories, etc.). This reasoning was demonstrated in Intellectual Ventures I LLC v. Capital One Bank (Fed. Cir. 2015), where the court determined "an abstract idea does not become nonabstract by limiting the invention to a particular field of use or technological environment, such as the Internet [or] a computer"). This/these limitation(s) do/does not impose any meaningful limits on practicing the abstract idea, and therefore do/does not integrate the abstract idea into a practical application (see MPEP 2106.05(h)).
The recited additional element(s) of first artificial intelligence model and AI large-scale parameter model do not integrate the abstract idea into a practical application. The claims do not recite any particular AI-model architecture, training method, objective function, parameter-update process, neural-network structure, prompt-engineering technique, retrieval algorithm, or inference optimization that improves the functioning of the computer or the AI model itself. Rather, the AI models are recited functionally as tools used to generate, process, or adapt product text according to regional terminology and grammar preferences. (Independent Claims 1 and 15-16), additionally and/or alternatively simply append insignificant extra-solution activity to the judicial exception, (e.g., mere pre-solution activity, such as data gathering, in conjunction with an abstract idea). This/these limitation(s) do/does not impose any meaningful limits on practicing the abstract idea, and therefore do/does not integrate the abstract idea into a practical application. (See MPEP 2106.05(g)).
Dependent claims 2-3, 5-7, 9-14, 17-18, and 20 fail to include any additional elements. In other words, each of the limitations/elements recited in respective dependent claims is/are further part of the abstract idea as identified by the Examiner for each respective dependent claim (i.e., they are part of the abstract idea recited in each respective claim).
The Examiner has therefore determined that the additional elements, or combination of additional elements, do not integrate the abstract idea into a practical application. Accordingly, the claim(s) is/are directed to an abstract idea (Step 2A – Prong two: NO).
Step 2B: In step 2B, the claims are analyzed to determine whether any additional element, or combination of additional elements, is/are sufficient to ensure that the claims amount to significantly more than the judicial exception. This analysis is also termed a search for an "inventive concept." An "inventive concept" is furnished by an element or combination of elements that is recited in the claim in addition to (beyond) the judicial exception, and is sufficient to ensure that the claim as a whole amounts to significantly more than the judicial exception itself. Alice Corp., 134 S. Ct. at 2355, 110 USPQ2d at 1981 (citing Mayo, 566 U.S. at 72-73, 101 USPQ2d at 1966).
As discussed above in “Step 2A – Prong 2”, the identified additional elements in independent Claim(s) 1 and 15-16, and dependent claims 2-3, 5-7, 9-14, 17-18, and 20 are equivalent to adding the words “apply it” on a generic computer, and/or generally link the use of the judicial exception to a particular technological environment or field of use. Therefore, the claims as a whole do not amount to significantly more than the judicial exception itself.
The recited additional element(s) of first artificial intelligence model and AI large-scale parameter model do not integrate the abstract idea into a practical application. The claims do not recite any particular AI-model architecture, training method, objective function, parameter-update process, neural-network structure, prompt-engineering technique, retrieval algorithm, or inference optimization that improves the functioning of the computer or the AI model itself. Rather, the AI models are recited functionally as tools used to generate, process, or adapt product text according to regional terminology and grammar preferences (Independent Claims 1, and 15-16), additionally and/or alternatively simply append insignificant extra-solution activity to the judicial exception, (e.g., mere pre-solution activity, such as data gathering, in conjunction with an abstract idea) i.e. localization of product descriptions could be performed conceptually by a person determining a user’s region, selecting locally appropriate terms for product text, revising the product text according to regional grammar or style preferences, and presenting the revised product description to the user is similar to “Receiving or transmitting data over a network, e.g., using the Internet to gather data”, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information), “Storing and retrieving information in memory”, Versata Dev. Group, Inc. v. SAP Am., Inc., 793 F.3d 1306, 1334, 115 USPQ2d 1681, 1701 (Fed. Cir. 2015); OIP Techs., 788 F.3d at 1363, 115 USPQ2d at 1092-93; “Presenting offers to potential customers and gathering statistics generated based on the testing about how potential customers responded to the offers; the statistics are then used to calculate an optimized price”, OIP Technologies, 788 F.3d at 1363, 115 USPQ2d at 1092-93, Determining an estimated outcome and setting a price, OIP Techs., 788 F.3d at 1362-63, 115 USPQ2d at 1092-93, is a well-understood, routine, and conventional function when it is claimed in a merely generic manner (as it is here) (See MPEP 2106.05(d) (II)).
This conclusion is based on a factual determination. Applicant’s own disclosure at paragraph [0104, 0196] acknowledges that “target product is localized based on the national or reginal attributes of the target user to generate the target text to displayed on the target page” and “the description of the above embodiments, it can be understood by those skilled in the art that the present application can be implemented by means of software combined with the necessary general hardware platform” (i.e., conventional nature of receiving and transmitting data/messages over a network). This additional element therefore do not ensure the claim amounts to significantly more than the abstract idea.
Viewing the additional limitations in combination also shows that they fail to ensure the claims amount to significantly more than the abstract idea. When considered as an ordered combination, the additional components of the claims add nothing that is not already present when considered separately, and thus simply append the abstract idea with words equivalent to “apply it” on a generic computer and/or mere instructions to implement the abstract idea on a generic computer or/and append the abstract idea with insignificant extra solution activity associated with the implementation of the judicial exception, (e.g., mere data gathering, post-solution activity) and/or simply appending well-understood, routine, conventional activities previously known to the industry, specified at a high level of generality, to the judicial exception.
The dependent claims 2-3, 5-7, 9-14, 17-18, and 20 fail to include any additional elements. In other words, each of the limitations/elements recited in respective independent claims is/are further part of the abstract idea as identified by the Examiner for each respective dependent claim (i.e., they are part of the abstract idea recited in each respective claim).
Claims 2-3, 5-7, 9-14, 17-18, and 20 merely further limit the abstract idea by reciting the use of an “artificial intelligence (AI) large-scale parameter model” to comprehend, transform, or adapt original descriptive information based on national or regional attribute information, the target product object is associated with multiple Stock Keeping Units (SKUs), and that descriptive information or SKUs are processed or reordered to prioritize locally common expressions or attribute values, the original descriptive information comprises textual content, including titles, keywords, adjectives, and user reviews, and that such textual content is transformed, converted, or reordered based on national or regional attribute information, transforming original attribute or parameter descriptions of a product object based on national or regional attribute information to generate localized attribute or parameter descriptions for display, the original descriptive information comprises rich media information, including image and audio information, and that such media is transformed based on national or regional attribute information to align with local preferences. Collectively, these dependent claims add only broadly claim transforming or selecting media content based on user region, which is a form of content presentation and customization, all of which constitute well-understood, routine, and conventional activities performed by generic computer components and therefore fail to integrate the abstract concept into a practical application and it is recited at a high level of generality and does not integrate the judicial exception into a practical application.
The Examiner has therefore determined that no additional element, or combination of additional claims elements is/are sufficient to ensure the claim(s) amount to significantly more than the abstract idea identified above (Step 2B: NO).
Therefore, claims 1-3, 5-7, 9-18, and 20 are not eligible subject matter under 35 USC 101.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status:
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
Determining the scope and contents of the prior art.
Ascertaining the differences between the prior art and the claims at issue.
Resolving the level of ordinary skill in the pertinent art.
Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 3, 5-6, 9, 11-16, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Pat. 9684653 (“Bhagat”) in view U.S. Pub. 20230259692 (“Wright”) in view U.S. Pub. 20220027577 (“Duan”) in view U.S. Pub. 20150206189 (“Norwood”).
As per claims 1 and 16, Bhagat discloses, generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first artificial intelligence (AI) model, wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions (Examiner notes that underlined limitation is disclosed by prior art. Examiner notes that the product information provided by sellers may include values for various product attributes, such as a product identifier, stock keeping unit number, title or description of the product, technical specifications, purchase price, product availability, shipping parameters and costs, geographic region in which the product is available, locale of the seller’s fulfillment center, and other product details. Bhagat further discloses that such product information may be used to generate product listings in a product catalog made available through the online e-commerce marketplace for searching and browsing by customers) (“The product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog. The product listings are made available through the online e-commerce marketplace for searching and browsing by customers wishing to purchase the corresponding products from the sellers” and “For each of the words in the field 308B, the translation dictionary 304 also includes a corresponding probability in the field 308B that the word in the field 308B is a translation of the word in the field 308A. For example, the translation dictionary 304 might include the English word “shoe” in the column 308A. For the word “shoe”, the field 308B might include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the field 308C might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%. It should be appreciated that the structure illustrated in FIG. 3 is merely illustrative and that the data described above might be stored in different ways in other embodiments”) (Col. 1 Ln. 18-31 and Col. 10 Ln. 22-47);
identifying at least one target product object and its original descriptive information to be provided to a target user (Examiner interprets discloses identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, because Bhagat’s product information includes product title and description information, technical specifications, and other textual product details used to generate product listings) (“product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog” and “a Web server module 602 is a software application that is configured to receive and respond for requests for Web pages and other resources. The resources may be stored in the Web site resources database 608 or generated dynamically.”) (Col. 1 Ln. 18-28 and Col. 13 Ln. 40-45), wherein the original descriptive information comprises original textual content (“The product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog. The product listings are made available through the online e-commerce marketplace for searching and browsing by customers wishing to purchase the corresponding products from the sellers.”) (Col. 1 Ln. 18-31);
determining national or regional attribute information of the target user (Examiner interprets determining national or regional attribute information of the target user, because Bhagat teaches merchant systems geographically distributed to provide an e-commerce marketplace to customers located around the world, such as an English-language version for customers in the United States and other versions for customers located in Europe, China, Japan, and Australia) (“the merchant system 108 might be geographically distributed in order to provide the e-commerce marketplace 118 to customers located around the world. For example, a merchant system 108A might be located in the United States in order to provide an English language version of the e-commerce marketplace to customers located in the U.S. Similarly, a merchant system 108B might be located in Europe to serve customers located in that region. Likewise, merchant systems 108C, 108D, and 108E might be located in China, Japan, and Australia, respectively, to serve customers located in those areas …”) (Col. 8 Ln. 37-47, Col. 1 Ln. 33-40);
processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
wherein the processing comprises: processing the original textual content via an AI large-scale parameter model to generate target textual content by replacing keywords in the original textual content with corresponding local common terms queried from the pre-established knowledge base (Examiner notes that the underlined limitation is disclosed by another prior art. (Examiner interprets translation dictionary corresponds to a pre-established knowledge base because it is a stored repository of language-mapping information used to convert original product text into localized text. Bhagat’s stored translation terms and probabilities correspond to local common terms, because the dictionary stores foreign-language terms associated with product information and uses those stored terms to translate or localize product-related content, further replacing keywords in original textual content with corresponding local common terms queried from the pre-established knowledge base, because Bhagat teaches tokenizing search queries or website resources and translating current tokens using the translation dictionary) (“As shown in FIG. 3, the translation dictionary 304 in one embodiment includes a field 308A that stores the words found in the product information 306A (i.e. the English product information in the example shown in FIG. 3). For each of the words in the product information 306A, the translation dictionary 304 also includes one or more words in a field 308B from the foreign language product information (i.e. the product information in Spanish in the example shown in FIG. 3) … For each of the words in the field 308B, the translation dictionary 304 also includes a corresponding probability in the field 308B that the word in the field 308B is a translation of the word in the field 308A. For example, the translation dictionary 304 might include the English word “shoe” in the column 308A. For the word “shoe”, the field 308B might include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the field 308C might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%. It should be appreciated that the structure illustrated in FIG. 3 is merely illustrative and that the data described above might be stored in different ways in other embodiments”) (Col. 10 Ln. 22-47, Col. 11 Ln. 1-20, Col. 13 Ln. 20-45).
Bhagat discloses, processing the original textual content via an AI large-scale parameter model, but specifically doesn’t discloses, generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first artificial intelligence (AI) model and processing the original textual content via an AI large-scale parameter model to generate target textual content, however Wright discloses, generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first artificial intelligence (AI) model (Examiner interprets Wright’s use of example product title-description pairs corresponds to sample textual inputs used by an AI model to generate improved or style-consistent product text. Wright’s generative language model corresponds to an AI large-scale parameter model, because Wright expressly identifies transformer models including BERT and GPT, which are large-scale machine learning language models used to generate natural language product descriptions) (“the merchant may enter one or more example product title and product description pairs, such as “linen halter top” and its corresponding product description, “corduroy cargo pants” and its corresponding product description, “embroidered poplin top” and its corresponding product description, followed by the words “ribbed crop tank top”, as input 502. The one or more example product title and product description pairs may be chosen by the merchant to form part of input 502 because the merchant favours the example product descriptions over other product descriptions. The merchant may favour the example product descriptions over other product descriptions due to the wording, or grammar, or tone, or length, or flow, or any other characteristic possessed by the example product descriptions. By providing the additional example product titles and/or example product descriptions, the generative language model 510 may generate a product description that is different from that generated in the absence of the example product title and/or example product description. The different product description may be generated taking into consideration the example product title and/or example product description and result in a generated product description that possibly has fewer inaccuracies and/or is more consistent with the preferred style of the merchant … Once the merchant provides the input 502 via the user interface 428, the input may be delivered to the generative language model 510, e.g. by transmitting the input from the merchant device 420 to the product description generator 410, which executes the generative language model 510. The generative language model 510 may be any type of natural language processing machine learning or deep learning model, for example: a recurrent neural network (RNN) model such as the long short-term memory (LSTM) model or the gated recurrent neural network, or a transformer model such as the Bidirectional Encoder Representations from Transformers (BERT) or the Generative Pre-trained Transformer (GPT) model.”) (0080-0082),
processing the original textual content via an AI large-scale parameter model to generate target textual content (Examiner interprets that a merchant may provide one or more example product title and product description pairs as input, and that the examples may be selected because the merchant prefers their wording, grammar, tone, length, flow, or other characteristics. Wright further teaches that the generative language model may generate a product description taking those examples into consideration, resulting in generated descriptions that are more consistent with the preferred style of the merchant. Wright also discloses that the generative language model may be a natural language processing machine learning or deep learning model, including a recurrent neural network, LSTM, gated recurrent neural network, transformer model, BERT model, or GPT model) (“a merchant may wish to be provided with computer-generated product descriptions for their products. For example, a natural language processing model such as a generative language model may be used to generate product descriptions for the merchant … Once the merchant provides the input 502 via the user interface 428, the input may be delivered to the generative language model 510, e.g. by transmitting the input from the merchant device 420 to the product description generator 410, which executes the generative language model 510. The generative language model 510 may be any type of natural language processing machine learning or deep learning model, for example: a recurrent neural network (RNN) model such as the long short-term memory (LSTM) model or the gated recurrent neural network, or a transformer model such as the Bidirectional Encoder Representations from Transformers (BERT) or the Generative Pre-trained Transformer (GPT) model.”) (0077-0082, 0089-0090)
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions, identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, determining national or regional attribute information of the target user, processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first artificial intelligence (AI) model and processing the original textual content via an AI large-scale parameter model to generate target textual content, processing the original textual content via an AI large-scale parameter model to generate target textual content, as taught by Wright for the purpose to utilize generative language models to produce product descriptions that better match preferred wording, grammar, tone, length, and flow thus to improve the quality and naturalness of localized product descriptions, reduce manual drafting or translation effort, and generate product descriptions more consistent with preferred local or merchant-specific language conventions.
Bhagat specifically doesn’t discloses, and adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information, however Duan discloses, and adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information (Examiner interprets Duan’s vocabulary, glossary, word choice, sentence structures, and paragraph structures correspond to grammatical expression habits, because these features define how language is locally or stylistically expressed. When applied to Bhagat’s regional localization system and Wright’s product-description generation model, Duan teaches adapting generated product text to conform to local grammar, word choice, and expression style) (“a method of converting a natural language input or text in one style into a nature language or text in another style (which is also referred to as stylization). In a text, “style” refers to text features and/or language features for embodying how ideas or information is presented, such as length, vocabulary or glossary, writing style, and the like. Different styles can be distinguished from one another. More specifically, the style can include word choice, sentence structures, paragraph structures, and the like … When stylizing a text, the computing device 100 can receive a text (or a natural language sentence) 170 via the input device 150. The text generalization module 122 can process the text 170 and generate, based on the text 170, a text in another style which at least reflects, embodies, or indicates information contained or conveyed by the text 170. The text can be fed to the output device 160 and provided to a user as an output 180. For example, the text can be displayed on a display for the user”) (0022-0025, 0035-0038).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions, identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, determining national or regional attribute information of the target user, processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information, as taught by Duan for the purpose to apply text-stylization features would improve the readability and local appropriateness of the generated product description while preserving the original product information.
Bhagat discloses, generating translated resources dynamically in response to a request for website resources, including product listings, but specifically doesn’t discloses, providing the target descriptive information corresponding to the at least one target product object to a client device of the target user to provide the target descriptive information on a designated webpage, however Norwood discloses, and providing the target descriptive information corresponding to the at least one target product object to a client device of the target user to provide the target descriptive information on a designated webpage (Examiner interprets a localization engine that performs localization of content contained in offer listings specified by a merchant client. Norwood teaches that the localization engine may generate one or more localized offer listings and that a network page server serves the localized offer listings to customer clients in the form of localized network pages or other forms of network content. Norwood further discloses that the localized offer listings are translated into one or more languages based at least in part on the locale preferred by users of the customer client) (“localization engine 124 performs the localization of the content contained in one or more offer listings specified by the merchant client 106 using the user interface generated by the merchant interface application 121. Localization of the content is performed using the information obtained by the merchant interface application 121, wherein the localization engine 124 may generate one or more localized offer listings. The network page server 127 serves up the localized offer listings to one or more customer clients 116 in the form of localized network pages or other forms of network content, where the localized offer listings have been translated into one or more languages based at least in part on the locale preferred by one or more users of the customer client 116” and “a selection of one or more locales in which target offer listings associated with the merchant client 106 are to be translated and localized for customers in the selected locales. The item information 147 retrieved by the merchant interface application 121 from the merchant client 106 includes such information as a specific item(s) and a stock keeping unit (SKU) or other identifier associated with a merchant, where the merchant SKU identifies particular inventory from which transactions are to be fulfilled. The merchant SKU may be associated, for example, with inventory in a particular location. The item information 147 also includes pricing information relating to the specified item(s) …”) (0014-0015 and 0017-0023, 0038).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions, identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, determining national or regional attribute information of the target user, processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, providing the target descriptive information corresponding to the at least one target product object to a client device of the target user to provide the target descriptive information on a designated webpage, as taught by Norwood for the purpose to provide a known architecture for serving localized product pages to customer client devices. The motivation would have been to deliver the generated localized product description to the appropriate customer-facing webpage based on the customer’s locale.
As per claims 3 and 18, Bhagat discloses, the target product object is associated with multiple Stock Keeping Units (SKUs) (“sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog. The product listings are made available through the online e-commerce marketplace for searching and browsing by customers wishing to purchase the corresponding products from the sellers”) (Col. 1 Ln. 18-32);
and wherein the processing the original descriptive information to adapt to the local expression based on the target user’s national or regional attribute information to generate the target descriptive information comprises (“one embodiment the translation dictionary creation module 302 iteratively applies an Expectation-Maximization (“EM”) algorithm to determine the probability that words in the product information 306A are translations of words in the product information 306B. As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 10 Ln. 7-21, Col. 3 Ln. 25-29):
processing the original descriptive information corresponding to the SKUs to adapt to local expression to generate the target descriptive information, thereby to provide different target descriptive information for the same SKU to users with different national or regional attribute information, based on various local expression (“multi-lingual product information provided by sellers may be utilized to provide versions of the e-commerce marketplace 118 in various languages. For instance, product information provided in Spanish may be utilized to provide a Spanish language version of the e-commerce marketplace 118 to customers in Spain and other Spanish-speaking countries. Product information provided in other languages might be utilized to provide versions of the e-commerce marketplace 118 in languages appropriate for different geographic regions”) (Col. 8 Ln. 61-67).
As per claims 5 and 20, Bhagat discloses, wherein the processing the original textual content comprises (“one embodiment the translation dictionary creation module 302 iteratively applies an Expectation-Maximization (“EM”) algorithm to determine the probability that words in the product information 306A are translations of words in the product information 306B. As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 10 Ln. 7-21, Col. 3 Ln. 25-29):
converting keywords related to a product name and/or adjectives included in the original textual content into local common terms corresponding to the national or regional attribute information (Examiner interprets that tokenizing terms and using a translation dictionary with word/probability mapping) (“a seller might provide a description for a product in English and provide the same product description for the product in Spanish. The merchant system might store and utilize this information to provide localized versions of the online e-commerce marketplace in languages appropriate for a particular locale. For example, Spanish language product information might be utilized to provide a Spanish version of the online e-commerce marketplace to customers located in Spain and/or other Spanish speaking countries”) (Col. 3 Ln.8-17).
As per claims 6, Bhagat discloses, wherein the original textual content comprises original title textual content user (”product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog” and “a Web server module 602 is a software application that is configured to receive and respond for requests for Web pages and other resources. The resources may be stored in the Web site resources database 608 or generated dynamically.”) (Col. 1 Ln. 18-28 and Col. 13 Ln. 40-45);
and wherein the processing the original textual content comprises (Col. 10 Ln. 7-21, Col. 3 Ln. 25-29):
processing the original title textual content to adapt to a textual content related to an expression of product attributes based on attribute preferences of a demographic corresponding to the national or regional attribute information for an category of products to which the target product object belongs (Examiner notes that attribute preferences of a demographic broadly as locale-based preferences (e.g., what details are emphasized in a region/country), which is consistent with localization systems) (“one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
As per claims 9, Bhagat discloses, wherein the original descriptive information comprises an original attribute/parameter description of the target product object (“product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog” and “a Web server module 602 is a software application that is configured to receive and respond for requests for Web pages and other resources. The resources may be stored in the Web site resources database 608 or generated dynamically.”) (Col. 1 Ln. 18-28 and Col. 13 Ln. 40-45);
and wherein the processing the original descriptive information to adapt to the local expression based on the target user’s national or regional attribute information to generate the target descriptive information comprises (“generate a translation dictionary from product information that has been received in multiple languages. In particular, a statistical analysis may be performed on product information that has been provided in multiple languages in order to generate the translation dictionary” and “one embodiment the translation dictionary creation module 302 iteratively applies an Expectation-Maximization (“EM”) algorithm to determine the probability that words in the product information 306A are translations of words in the product information 306B. As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 3 Ln. 25-29 and Col. 10 Ln. 7-21):
transforming the original attribute/parameter description of the target product object based on the national or regional attribute information of the target user to adapt to the local expression to generate target attribute/parameter description, for display on a designated webpage (Examiner notes, translation dictionary is text-centric) (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
As per claims 11, Bhagat discloses, wherein the transforming the original attribute/parameter description of the target product object based on the national or regional attribute information of the target user to adapt to the local expression comprises (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27);
inputting the original attribute/parameter description information and the national or regional attribute information of the target user into an AI large-scale parameter model, so that the AI large-scale parameter model performs the transforming the original attribute/parameter description of the target product object based on the national or regional attribute information of the target user to adapt to into the local expression (“As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 10 Ln. 11-21).
As per claims 12, Bhagat discloses, wherein the original descriptive information comprises original rich media information of the target product object (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27);
and wherein the processing the original descriptive information to adapt to the local expression based on the target user’s national or regional attribute information to generate the target descriptive information comprises (Col. 3 Ln. 25-29 and Col. 10 Ln. 7-21):
transforming the original rich media information of the target product object based on the national or regional attribute information of the target user to adapt to the local expression to generate target rich media information, for display on the designated webpage (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
As per claims 13, Bhagat discloses, wherein the original rich media information comprises original image information (“online shopping module 112 might utilize pre-stored or dynamically created resources to generate the e-commerce marketplace 118. For instance, Web pages, images, text files, program code for generating Web pages, metadata, scripts, executable code, and other types of data utilized to create and/or provide a Web page might be stored or dynamically generated. Other types of resources might also be stored or generated dynamically by the online shopping module 112 to provide the e-commerce marketplace 118”) (Col. 6 Ln. 60-67);
and wherein the transforming the original rich media information of the target product object based on the national or regional attribute information of the target user to adapt to the local expression comprises (Col. 3 Ln. 25-29 and Col. 10 Ln. 7-21):
transforming composition style, model type, and/or atmospheric elements of the original image information based on the national or regional attribute information of the target user, to generate target image information that aligns with local preferences corresponding to the national or regional attribute information (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
As per claims 14, Bhagat discloses, and wherein the transforming the original rich media information of the target product object based on the national or regional attribute information of the target user to adapt to the local expression comprises (Col. 3 Ln. 25-29 and Col. 10 Ln. 7-21): transforming the original audio information based on the national or regional attribute information of the target user, to generate target audio information that aligns with local preferences corresponding to the national or regional attribute information (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
Bhagat discloses, generating a response to the Web page request that includes the translated resources dynamically in response to a request for the resources generated dynamically by the online shopping retrieve a product listing for a particular product offered for sale by the online merchant or another seller from a product catalog, but specifically doesn’t discloses, wherein the original rich media information comprises original audio information, however Norwood discloses, wherein the original rich media information comprises original audio information (“network page server 127 is executed to serve up localized offer listings to the customer clients 116 and operates in conjunction with the electronic commerce system 129 to perform purchase transactions. The localized offer listings may include such offer listing source data as hypertext markup language (HTML), extensible markup language (XML), extensible HTML (XHTML), mathematical markup language (MathML), scalable vector graphics (SVG), cascading style sheets (CSS), images, audio, video, graphics, text, and/or any other data that may be used in serving up or generating the offer listings”) (0023).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for identifying at least one target product object and its original descriptive information to be provided to a target user; determining national or regional attribute information of the target user; processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, wherein the original rich media information comprises original audio information, as taught by Norwood for the purpose for providing localization architecture for selecting and delivering localized product pages.
As per claims 15, Bhagat discloses, receiving target descriptive information of at least one target product object provided for a target user from a server (Examiner interprets that Bhagat discloses, a method for providing product object information in an e-commerce marketplace, wherein product information may include a product identifier, SKU number, title or description, technical specifications, purchase price, availability, shipping parameters, geographic region, seller fulfillment locale, and other product details, and wherein such product information may be used to generate product listings in a product catalog) (”product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog” and “a Web server module 602 is a software application that is configured to receive and respond for requests for Web pages and other resources. The resources may be stored in the Web site resources database 608 or generated dynamically.”) (Col. 1 Ln. 18-28 and Col. 13 Ln. 40-45),
the target descriptive information is generated by: processing original textual content of the target product object via an artificial intelligence (AI) large-scale parameter model to generate target textual content by replacing keywords in the original textual content with corresponding local common terms queried from a pre-established knowledge base (Examiner notes that the underlined limitation is disclosed by another prior art. Examiner interprets generating and using a pre-established translation dictionary from product information. In particular, Bhagat discloses that a translation dictionary creation module creates translation dictionary 304 using product information in different languages, and that statistical analysis, including an EM algorithm and GIZA/GIZA++, may be used to determine probabilities that words in product information in one language are translations of words in product information in another language i.e. the translation dictionary stores words from product information and corresponding foreign-language words, such as “shoe” corresponding to “zapato” and “herradura,” along with probabilities, and tokenizing text/search terms and translating the tokenized terms using the translation dictionary. Bhagat’s translation dictionary corresponds to the claimed pre-established knowledge base, and the stored foreign-language product terms correspond to local common terms) (“The translation dictionary creation module 302 is configured to generate a translation dictionary 304 that can be utilized to translate human readable language. The translation dictionary creation module 302 creates the translation dictionary 304 using product information stored in the merchant marketplace product catalog 122. In the example shown in FIG. 3, the translation dictionary creation module 302 creates the translation dictionary 304 using product information 306A specified for a product in the English language and product information 306B specified for the same product in the Spanish language. When the translation dictionary 304 is created using English and Spanish product information, the translation dictionary 304 can be utilized to translate words from Spanish to English. When the translation dictionary 304 is created using product information expressed using other languages, the translation dictionary 304 may be utilized to translate words between the other languages. For example, the product information 306A and 306B might be utilized to create a translation dictionary 304 for translating from English to Spanish or another language …”) (Col. 9 Ln. 35-67 – Col. 10 Ln. 1-47),
wherein the pre-established knowledge base is generated by inputting small-scale samples of regional terms into a first AI model (“The translation dictionary creation module 302 is configured to generate a translation dictionary 304 that can be utilized to translate human readable language. The translation dictionary creation module 302 creates the translation dictionary 304 using product information stored in the merchant marketplace product catalog 122. In the example shown in FIG. 3, the translation dictionary creation module 302 creates the translation dictionary 304 using product information 306A specified for a product in the English language and product information 306B specified for the same product in the Spanish language. When the translation dictionary 304 is created using English and Spanish product information, the translation dictionary 304 can be utilized to translate words from Spanish to English. When the translation dictionary 304 is created using product information expressed using other languages, the translation dictionary 304 may be utilized to translate words between the other languages. For example, the product information 306A and 306B might be utilized to create a translation dictionary 304 for translating from English to Spanish or another language …”) (Col. 9 Ln.35-67, Col. 10 Ln. 1-47),
Bhagat discloses, generating a translation dictionary 304 from product information, including English/Spanish product information, and product category specifics dictionaries by utilizing statistical analysis/EM/ GIZA++ to learn bilingual dictionaries, but specifically doesn’t discloses, processing original textual content of the target product object via an artificial intelligence (AI) large-scale parameter model to generate target textual content and wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with the local common terms and the local grammatical expression habits for multiple regions, however Wright discloses, processing original textual content of the target product object via an artificial intelligence (AI) large-scale parameter model to generate target textual content (Examiner notes that natural language processing model, such as a generative language model, may be used to generate product descriptions. Wright further discloses that the generative language model may be a machine learning or deep learning model, including transformer models such as BERT or GPT i.e. input to the generative language model may include a product title or example product title/product-description pairs, and that the generative language model may generate a product description consistent with example wording, grammar, tone, length, and flow and Wright’s generative language model corresponds to the claimed AI large-scale parameter model, and Wright’s example product title/product-description pairs correspond to sample textual inputs used by the model) (“a merchant may wish to be provided with computer-generated product descriptions for their products. For example, a natural language processing model such as a generative language model may be used to generate product descriptions for the merchant … Once the merchant provides the input 502 via the user interface 428, the input may be delivered to the generative language model 510, e.g. by transmitting the input from the merchant device 420 to the product description generator 410, which executes the generative language model 510. The generative language model 510 may be any type of natural language processing machine learning or deep learning model, for example: a recurrent neural network (RNN) model such as the long short-term memory (LSTM) model or the gated recurrent neural network, or a transformer model such as the Bidirectional Encoder Representations from Transformers (BERT) or the Generative Pre-trained Transformer (GPT) model.” and “the merchant may enter one or more example product title and product description pairs, such as “linen halter top” and its corresponding product description, “corduroy cargo pants” and its corresponding product description, “embroidered poplin top” and its corresponding product description, followed by the words “ribbed crop tank top”, as input 502. The one or more example product title and product description pairs may be chosen by the merchant to form part of input 502 because the merchant favours the example product descriptions over other product descriptions. The merchant may favour the example product descriptions over other product descriptions due to the wording, or grammar, or tone, or length, or flow, or any other characteristic possessed by the example product descriptions. By providing the additional example product titles and/or example product descriptions, the generative language model 510 may generate a product description that is different from that generated in the absence of the example product title and/or example product description. The different product description may be generated taking into consideration the example product title and/or example product description and result in a generated product description that possibly has fewer inaccuracies and/or is more consistent with the preferred style of the merchant … Once the merchant provides the input 502 via the user interface 428, the input may be delivered to the generative language model 510, e.g. by transmitting the input from the merchant device 420 to the product description generator 410, which executes the generative language model 510. The generative language model 510 may be any type of natural language processing machine learning or deep learning model, for example: a recurrent neural network (RNN) model such as the long short-term memory (LSTM) model or the gated recurrent neural network, or a transformer model such as the Bidirectional Encoder Representations from Transformers (BERT) or the Generative Pre-trained Transformer (GPT) model.”) (0080-0082 and 0077-0082, 0089-0090),
wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with the local common terms and the local grammatical expression habits for multiple regions (Examiner notes that the Wright discloses using one or more example product title/product description pairs as input to the generative language model, under BRI this corresponds to small-scale samples because the claim does not define a numeric threshold for “small-scale”. Further delivering input 502 to a generative language model 510, which may be an NLP machine learning/deep learning model, including RNN, LSTM, BERT, or GPT) (0080-0082 and 0077-0082, 0089-0090).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions, identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, determining national or regional attribute information of the target user, processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, processing original textual content of the target product object via an artificial intelligence (AI) large-scale parameter model to generate target textual content and wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with the local common terms and the local grammatical expression habits for multiple regions, as taught by Wright for the purpose for the purpose to utilize generative language models to produce product descriptions that better match preferred wording, grammar, tone, length, and flow thus to improve the quality and naturalness of localized product descriptions, reduce manual drafting or translation effort, and generate product descriptions more consistent with preferred local or merchant-specific language conventions.
Bhagat specifically doesn’t discloses, and adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information, however Duan discloses, and adapting the target textual content to conform to local grammatical expression habits corresponding to national or regional attribute information of the target user (Examiner interprets converting natural-language text from one style into another style, wherein style includes vocabulary or glossary, word choice, sentence structures, and paragraph structures. Under BRI, Duan’s vocabulary, word choice, sentence structures, and paragraph structures correspond to the claimed grammatical expression habits) (“a method of converting a natural language input or text in one style into a nature language or text in another style (which is also referred to as stylization). In a text, “style” refers to text features and/or language features for embodying how ideas or information is presented, such as length, vocabulary or glossary, writing style, and the like. Different styles can be distinguished from one another. More specifically, the style can include word choice, sentence structures, paragraph structures, and the like … When stylizing a text, the computing device 100 can receive a text (or a natural language sentence) 170 via the input device 150. The text generalization module 122 can process the text 170 and generate, based on the text 170, a text in another style which at least reflects, embodies, or indicates information contained or conveyed by the text 170. The text can be fed to the output device 160 and provided to a user as an output 180. For example, the text can be displayed on a display for the user”) (0022-0025, 0035-0038).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for wherein the first AI model generalizes from the small-scale samples to populate the pre-established knowledge base with local common terms and local grammatical expression habits for multiple regions, identifying at least one target product object and its original descriptive information to be provided to a target user, wherein the original descriptive information comprises original textual content, determining national or regional attribute information of the target user, processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, adapting the target textual content to conform to the local grammatical expression habits corresponding to the national or regional attribute information, thereby generating the target descriptive information, as taught by Duan for the purpose to apply text-stylization features would improve the readability and local appropriateness of the generated product description while preserving the original product information.
Bhagat discloses, displaying the target descriptive information of at least one target product object on a designated webpage, however Norwood discloses, and displaying the target descriptive information of at least one target product object comprising the target textual content on a designated webpage (Examiner interprets a localization engine that generates localized offer listings and a network page server that serves the localized offer listings to customer clients in the form of localized network pages or other network content. Norwood further discloses that the localized offer listings have been translated into one or more languages based at least in part on the locale preferred by users of the customer client) (“localization engine 124 performs the localization of the content contained in one or more offer listings specified by the merchant client 106 using the user interface generated by the merchant interface application 121. Localization of the content is performed using the information obtained by the merchant interface application 121, wherein the localization engine 124 may generate one or more localized offer listings. The network page server 127 serves up the localized offer listings to one or more customer clients 116 in the form of localized network pages or other forms of network content, where the localized offer listings have been translated into one or more languages based at least in part on the locale preferred by one or more users of the customer client 116” and “a selection of one or more locales in which target offer listings associated with the merchant client 106 are to be translated and localized for customers in the selected locales. The item information 147 retrieved by the merchant interface application 121 from the merchant client 106 includes such information as a specific item(s) and a stock keeping unit (SKU) or other identifier associated with a merchant, where the merchant SKU identifies particular inventory from which transactions are to be fulfilled. The merchant SKU may be associated, for example, with inventory in a particular location. The item information 147 also includes pricing information relating to the specified item(s) …”) (0014-0015 and 0017-0023, 0038).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for identifying at least one target product object and its original descriptive information to be provided to a target user; determining national or regional attribute information of the target user; processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, displaying the target descriptive information of at least one target product object on a designated webpage, as taught by Norwood for the purpose for serving localized offer listings to customer client devices as localized network pages thus to improve localized product-description quality, reduce manual drafting/translation burden, and deliver localized product content to users through known e-commerce webpages.
Claims 2, 7, 10, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Pat. 9684653 (“Bhagat”) in view U.S. Pub. 20230259692 (“Wright”) in view U.S. Pub. 20220027577 (“Duan”) in view U.S. Pub. 20150206189 (“Norwood”) in further view of U.S. Pub. 20240346233 (“Paulino”).
As per claims 2 and 17, Bhagat discloses, wherein the processing the original descriptive information to adapt to the local expression based on the target user’s national or regional attribute information to generate the target descriptive information comprises (“generate a translation dictionary from product information that has been received in multiple languages. In particular, a statistical analysis may be performed on product information that has been provided in multiple languages in order to generate the translation dictionary” and “one embodiment the translation dictionary creation module 302 iteratively applies an Expectation-Maximization (“EM”) algorithm to determine the probability that words in the product information 306A are translations of words in the product information 306B. As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 3 Ln. 25-29 and Col. 10 Ln. 7-21):
utilizing the (AI) large-scale parameter model to comprehend the original descriptive information, and processing the original descriptive information to adapt to the local expression based on the national or regional attribute information of the target user to generate the target descriptive information (Examiner notes that the underlined limitation is disclose by another reference) (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
Bhagat discloses, allowing the model to focus on the most contextually relevant parts of the input for the task for handling sequence data and their capacity for parallel computation, often serve as foundational elements in constructing large generative AI models (LXM), but specifically doesn’t discloses, utilizing an artificial intelligence (AI) large-scale parameter model to comprehend the original descriptive information, however Paulino discloses, utilizing the (AI) large-scale parameter model to comprehend the original descriptive information (Examiner notes that AI large-scale parameter model can encompass trained statistical models used in machine translation including probabilistic models to process and generate localized descriptions)(“a review summary that summarizes the set of reviews associated with the item is generated, using the trained large language model. In embodiments, the review summary is generated in accordance with the set of weights that correspond with the set of reviews. At block 706, the review summary is provided to a data store for subsequent presentation. In some cases, the review summary is subsequently provided for display based on a request to view the review summary for the item”) (0091).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for identifying at least one target product object and its original descriptive information to be provided to a target user; determining national or regional attribute information of the target user; processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, utilizing an artificial intelligence (AI) large-scale parameter model to comprehend the original descriptive information, as taught by Paulino for the purpose to have more influence or provide more constructive insight to potential consumers and item providers.
As per claims 7, Bhagat discloses, wherein the original textual content comprises an original user review textual content (”product information provided by sellers might include values for various attributes of a product, such as an identifier for a product, like a stock keeping unit (“SKU”) number; a title or description of the product; technical specifications for the product; the purchase price; product availability, such as whether the product is “in stock”; shipping parameters and costs; the geographic region in which the product is available to customers; the locale of the seller's fulfillment center; and other types of product details. The product information may then be utilized to generate product listings in a product catalog” and “a Web server module 602 is a software application that is configured to receive and respond for requests for Web pages and other resources. The resources may be stored in the Web site resources database 608 or generated dynamically.”) (Col. 1 Ln. 18-28 and Col. 13 Ln. 40-45);
and wherein the processing the original textual content of the target product object based on the national or regional attribute information of the target user to adapt to the local expression comprises (“one embodiment the translation dictionary creation module 302 iteratively applies an Expectation-Maximization (“EM”) algorithm to determine the probability that words in the product information 306A are translations of words in the product information 306B. As known to those skilled in the art, the EM algorithm may be used for the unsupervised learning of bilingual translation dictionaries for statistical machine translation systems. The open source tools GIZA and GIZA++ are implementations of the EM algorithm that may be utilized in embodiments to create the translation dictionary 304. Other tools and/or components might be utilized to implement the EM algorithm in other embodiments. Other mechanisms might also be utilized to create the translation dictionary 304 from the bilingual product information 306”) (Col. 10 Ln. 7-21, Col. 3 Ln. 25-29).
Bhagat discloses, allowing the model to focus on the most contextually relevant parts of the input for the task for handling sequence data and their capacity for parallel computation, often serve as foundational elements in constructing large generative AI models (LXM), but specifically doesn’t discloses, reordering the original user review textual content based on the national or regional attribute information of the target user, in order to prioritize the display of local user reviews according to the national or regional attribute information, however Paulina discloses, reordering the original user review textual content based on the national or regional attribute information of the target user (Examiner notes that under BRI, “reordering” includes sorting/filtering review display order (e.g., “show most relevant first,” “show local first”), i.e. local user reviews according to national/regional attribute information can be interpreted as reviews written in the user’s language/region, which is a predictable extension of Bhagat’s region/language segmentation when combined with a review-ranking reference) (“existing reviews may be distributed across multiple locations, such as different websites, and also may contain a large volume of reviews that are not helpful for a potential consumer in that these reviews do not provide explanation or supporting detail regarding the features of the product that led to the review. For instance, reviews such “it's good” or “I hated it,” reviews that contain obscenities, or reviews that are not relevant to the product, do not help the potential customer to better understand the product. Further still, many of these reviews can be outdated and regard older versions of the product, thus requiring a user to look at the date of the review and have knowledge regarding the current version of the product. Accordingly, reviewing numerous product reviews to locate those reviews that are relevant and helpful to a current product places a burden on the user that can unnecessarily consume computing resources, as well as consume the user's time”) (0016), in order to prioritize the display of local user reviews according to the national or regional attribute information (“data selection may be based on any of a variety of aspects, such as data of reviews, weights of reviews, rankings associated with reviews, reviewers providing the review, and/or the like. As one example, the prompt generator 224 can first call for the input size constraint of tokens. Responsively, the prompt generator 224 can then tokenize each of the review candidates to generate tokens and, thereafter, responsively and progressively add each data set ranked/weighted from highest to lowest if and until the token threshold (indicating the input size constraint) is met or exceeded, at which point the prompt generator 224 stops”) (0076).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for identifying at least one target product object and its original descriptive information to be provided to a target user; determining national or regional attribute information of the target user; processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, reordering the original user review textual content based on the national or regional attribute information of the target user, in order to prioritize the display of local user reviews according to the national or regional attribute information, as taught by Paulino for the purpose to have more influence or provide more constructive insight to potential consumers and item providers.
As per claims 10, Bhagat discloses, wherein the transforming the original attribute/parameter description of the target product object based on the national or regional attribute information of the target user to adapt to the local expression comprises (“the translation dictionary includes the words found in the product information in one language (e.g. English). For each of the words in the first language, the translation dictionary also includes one or more words from the foreign language product information (e.g. Spanish), along with a corresponding probability that the word is a translation of the word in the first language. For example, the translation dictionary might include the English word “shoe.” For the word “shoe”, the translation dictionary might also include the Spanish words “zapato” and “herradura.” For the words “zapato” and “herradura”, the translation dictionary might indicate that the probability that the Spanish word “zapato” is a translation of the English word “shoe” is 85%, and that the probability that the Spanish word “herradura” is a translation of the English word “shoe” is 15%” and “The translated search queries can then be utilized to search the marketplace 118 in its native language, thereby likely providing better search results than if the foreign language search query had been processed without translation. Details regarding the structure of and functionality provided by the language translation system 114 will be provided below with respect to FIGS. 3-9”) (Col. 3 Ln. 30-45 and Col. 9 Ln. 21-27).
Bhagat discloses, allowing the model to focus on the most contextually relevant parts of the input for the task for handling sequence data and their capacity for parallel computation, often serve as foundational elements in constructing large generative AI models (LXM), but specifically doesn’t discloses, if the target product object is associated with multiple Stock Keeping Units (SKUs) having different attribute values/parameter values, reordering the SKUs to prioritize the display of the SKUs with locally commonly used attribute values/parameter values, corresponding to the national or regional attribute information, however Paulina discloses, if the target product object is associated with multiple Stock Keeping Units (SKUs) having different attribute values/parameter values, reordering the SKUs to prioritize the display of the SKUs with locally commonly used attribute values/parameter values, corresponding to the national or regional attribute information (Examiner notes (“a user viewing an item (e.g., a potential product to purchase) may select a link or icon to view a review summary associated with the item. In other cases, a user may indirectly or implicitly select to generate or view a review summary(s) related to an item. For instance, a user may navigate to a media store application or website. Based on the navigation to the media store application or website, the user may indirectly indicate to generate or view a review summary. In some cases, such an indication may be based on generally navigating to the application or website. For instance, a review summary may be requested for each item to be presented in the application or website or for a particular item(s) being features or promoted. In other cases, such an indication may be based on selecting a particular product” and “existing reviews may be distributed across multiple locations, such as different websites, and also may contain a large volume of reviews that are not helpful for a potential consumer in that these reviews do not provide explanation or supporting detail regarding the features of the product that led to the review. For instance, reviews such “it's good” or “I hated it,” reviews that contain obscenities, or reviews that are not relevant to the product, do not help the potential customer to better understand the product. Further still, many of these reviews can be outdated and regard older versions of the product, thus requiring a user to look at the date of the review and have knowledge regarding the current version of the product”) (0032 and 0016).
It would have been obvious to a person of ordinary skill in the art before the effective filling date of the applicant’s invention for identifying at least one target product object and its original descriptive information to be provided to a target user; determining national or regional attribute information of the target user; processing the original descriptive information to adapt to local expression based on the target user’s national or regional attribute information to generate target descriptive information, as disclosed by Bhagat, if the target product object is associated with multiple Stock Keeping Units (SKUs) having different attribute values/parameter values, reordering the SKUs to prioritize the display of the SKUs with locally commonly used attribute values/parameter values, corresponding to the national or regional attribute information, as taught by Paulino for the purpose to have more influence or provide more constructive insight to potential consumers and item providers and to increase conversion and reduce browsing time, this is a known merchandising optimization in localized marketplaces.
Response to Arguments
With regards to § 101 rejections:
The arguments filed on March 30th, 2026, with respect to the rejection(s) of claims 1-3, 5-7, 9-18, and 20 under 35 U.S.C 101 have been fully considered but are unpersuasive.
Applicant states that the amended claims recite a specific, multi-phase artificial intelligence architecture that improves the functioning of automated localization and natural language processing systems by performing knowledge-base generation offline and later using the generated knowledge base during real-time processing. However, the claims do not recite the specific technical details necessary to support the alleged improvement to computer functionality. The claims recite, at a high level of generality, generating a pre-established knowledge base by inputting small-scale samples of regional terms into a first AI model, where the model generalizes from the samples to populate the knowledge base with local common terms and local grammatical expression habits. The claims further recite using an AI large-scale parameter model to generate target textual content by replacing keywords with local common terms queried from the knowledge base and adapting the target textual content to local grammatical expression habits. These limitations describe the information-processing result to be achieved, namely generating and using localized linguistic information to produce localized product descriptions. The claims do not recite a particular AI architecture, neural-network configuration, training algorithm, objective function, parameter-update process, knowledge-base data structure, retrieval index, query mechanism, memory-management technique, latency-reduction mechanism, or other specific technological implementation. Applicant’s asserted improvement is therefore not adequate with the scope of the claims. Although the Specification may describe advantages such as reducing latency or avoiding repeated real-time calculation of regional slang, the claims do not require any particular technical operation that achieves such latency reduction or computational improvement. The claims do not recite any measured reduction in processing time, memory use, network traffic, database latency, model-inference cost, or processor workload. Instead, the claims merely state that localized terms and grammatical expression habits are generated, stored, queried, and used to modify product text. Accordingly, the claimed AI model and knowledge base are used as tools to perform the abstract idea of organizing, modifying, and presenting product information based on user regional attributes. Reciting an AI model to perform text processing does not, by itself, convert the abstract idea into a technological improvement. Likewise, reciting a pre-established knowledge base does not integrate the abstract idea into a practical application where the knowledge base is claimed only as a repository of linguistic information used for keyword replacement and grammatical adaptation.
Applicant further states that the ordered combination provides an inventive concept because it is allegedly not well-understood, routine, or conventional to automatically generalize a knowledge base from a small seed of samples and then use that knowledge base with a large-scale parameter model for real-time grammatical adaptation. This argument is not persuasive. The ordered combination still amounts to the abstract sequence of collecting or generating linguistic information, storing that information, retrieving the information, applying it to product text, and displaying the resulting localized product description. The claims do not recite an unconventional arrangement of computer components or an unconventional technical implementation of AI, database, memory, processor, or network functionality. Rather, the claims recite generic AI models, a generic knowledge base, a server, a client device, processors, memories, and a webpage, each used for its ordinary purpose of processing, storing, transmitting, and displaying information. The focus of the claims remains the localization and presentation of product content based on a user’s national or regional attribute information. Any alleged improvement lies in the content, accuracy, naturalness, or usefulness of the localized product description, not in an improvement to the functioning of a computer, AI model, database, network, or other technology. Improving the linguistic quality or marketing suitability of product descriptions is an improvement to information content and customer-facing presentation, not a technological improvement under the eligibility analysis. When considered individually and as an ordered combination, the additional elements do not impose meaningful limits on the abstract idea. The claims do not do more than implement localized product-description generation using generic computer components and functionally recited AI tools. Therefore, the amended claims do not integrate the abstract idea into a practical application under Step 2A, Prong Two, and do not recite significantly more than the abstract idea under Step 2B. Accordingly, the § 101 rejection is maintained. See MPEP §§ 2106.05(a)–(c), (e)– (h).
With regards to § 103 rejections:
Applicant's arguments, see pages 11-12, filed March 30th, 2026, with respect to the rejection(s) of claims 1-3, 5-7, 9-18, and 20 under 35 U.S.C 102/103 have been fully considered but are unpersuasive/moots on new ground of rejection.
Applicant’s arguments have been considered but are not persuasive. Applicant states that the previously cited references fail to teach the newly claimed multi-phase artificial intelligence architecture, including generating a pre-established knowledge base from small-scale samples using a first AI model and subsequently using that knowledge base during generation of localized product text. However, the present rejection does not rely on Bhagat, Norwood, and Paulino alone for these limitations. However, Bhagat is relied upon for teaching an e-commerce product-information localization system, including product information such as product identifiers, SKU information, product titles and descriptions, technical specifications, availability, geographic region, seller fulfillment locale, and other product details. Bhagat further teaches generating and using a translation dictionary from multilingual product information. Under the broadest reasonable interpretation, Bhagat’s translation dictionary corresponds to a pre-established knowledge base because it is a stored repository of language-mapping information used to translate or localize product-related textual content. Bhagat’s stored foreign-language product terms and associated probabilities correspond to local common terms because they are stored terms used to replace or translate original product terms into localized expressions. Applicant further states that Bhagat does not teach using an AI model to generalize from a small set of samples. The Examiner agrees that Bhagat is not relied upon alone for that feature. Wright discloses providing one or more example product title/product-description pairs as input to a generative language model. Wright further teaches that the examples may be selected based on wording, grammar, tone, length, flow, or other characteristics, and that the generative language model generates a product description taking those examples into consideration. Wright also teaches that the generative language model may be a machine-learning or deep-learning natural language processing model, including a transformer model such as BERT or GPT. Thus, Wright teaches using sample product-language inputs with an AI language model to generate product text consistent with the wording, grammar, tone, length, and flow of the examples. Applicant also states that the prior art does not teach local grammatical expression habits. However, Duan teaches converting natural-language text from one style into another style, where style includes vocabulary or glossary, word choice, sentence structures, and paragraph structures. Under the broadest reasonable interpretation, Duan’s vocabulary, word choice, sentence structures, and paragraph structures correspond to grammatical expression habits because they define how language is expressed in a particular style or linguistic form. Applying Duan’s style-conversion teachings to the localized product-description system of Bhagat and Wright would be to improve the local readability and grammatical naturalness of generated product descriptions. Further, Norwood is relied upon for teaching delivery of localized offer listings to customer clients as localized network pages. Norwood teaches a localization engine that generates localized offer listings and a network page server that serves those localized offer listings to customer clients in the form of localized network pages or other network content. Thus, Norwood teaches providing localized descriptive product information to a client device for display on a designated webpage. Accordingly, Applicant’s argument that the cited art fails to teach or suggest the claimed multi-phase AI architecture is not persuasive because the claimed arrangement is an obvious combination of known localization, generative-language-model, text-stylization, and localized webpage-delivery techniques.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. US. Pat. 11222362 (“Whiteman”).
Whiteman discloses, a method of localizing an element present in a piece of content having a plurality of elements. A cost of localizing an element with respect to each of one or more localization sources is first computed. At least one criterion based on which a localization source for localizing the element is to be determined is obtained. A localization source for localizing the element is then selected based on an assessment with respect to the at least one criterion. The element of the content is then localized using the selected localization source.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to GAUTAM UBALE whose telephone number is (571)272-9861. The examiner can normally be reached Mon-Fri. 7:00 AM- 6:30 PM PST.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Marissa Thein can be reached at (571) 272-6764. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/GAUTAM UBALE/Primary Examiner, Art Unit 3689
Prosecution Insights