506
DETAILED ACTION
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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 06/03/2026 has been entered.
Status of Claims
Claims 1, 8, and 15 have been amended by Applicant. No claims have been cancelled or added. Claims 1-20 are currently pending.
Response to Arguments
Claim Rejections under 35 U.S.C. 103
The rejection of claims 1-20 under 35 U.S.C. 103 have been withdrawn in view of Applicant’s amendments to independent claims 1, 8, and 15. However, upon further consideration and in view of said amendments, a new grounds of rejection has been made herein.
Applicant argues (in pages 13-14 of Applicant’s remarks) that there is no motivation to combine Gupta and Gautam with Nakano. In support, Applicant argues that the incorporation of Nakano to Gupta would defeat Gupta’s purpose and require a fundamental redesign of Gupta’s architecture. Similarly, Applicant argues that the same would apply to Gautam.
In response to applicant's argument that above, the test for obviousness is not whether the features of a secondary reference may be bodily incorporated into the structure of the primary reference; nor is it that the claimed invention must be expressly suggested in any one or all of the references. Rather, the test is what the combined teachings of the references would have suggested to those of ordinary skill in the art. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981). In this case, Nakano teaches fine-tuning a GPT model to answer long-form questions using a text-based web-browsing environment, which allows the model to search and navigate the web. (Nakano, Abstract) On the other hand, Gupta teaches methods that generate a dynamic cross-platform ask interface and utilizes a cross-platform language processing model to provide, platform-specific, contextually based responses to natural language digital text queries, utilizing machine learning models. (Gupta, Abstract). Furthermore, Gautam teaches methods and systems for providing real-time business intelligence using natural language queries to facilitate a user’s search, wherein Gautam also teaches employing machine learning within its structure. (Gautam, Abstract) Hence, it is Examiner’s position that the three reference are combinable as they all address the same objectives, each providing an advantage when they are combined, as stated in the motivation statements in the rejection of claim 1, as amended.
Applicant’s remaining arguments with respect to claim(s) 1, 8, and 15 (as amended) have been considered but are moot because the new ground of rejection does not rely on any reference applied in the prior rejection of record for any teaching or matter specifically challenged in the argument.
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 (i.e., changing from AIA to pre-AIA ) 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, 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.
Claims 1-2, 5-9, 12-16, and 19-20 (as amended) are rejected under 35 U.S.C. 103 as being unpatentable over Gupta et al. (US 20230021797 A1 filed Jul. 22, 2021 and published Jan. 26, 2023) in view of Gautam et al. (US 10,657,125 B1, published May 19, 2020), Nakano et al., “WebGPT: Browser-assisted question-answering with human feedback” (June 1, 2022), Mahapatra et al.,(US 20230020886 A1, filed Jul. 8, 2021 and published Jan. 19, 2023) , and Sianez (US 20210191925 A1, filed Dec. 18, 2020 and published Jun. 24, 2021)
Regarding Claim 1 (as amended),
Gupta teaches a processor-implemented method of neural network based text generation server using…search results, the method comprising (Gupta, Fig. 6F teaches using a cross-platform search system to generate textual responses using search results; pg. 5 [0046]-[0047] teaches neural network based text generation; pg. 16 [0136] teaches processor-implemented; Gupta, Abstract, teaches the present disclosure relates to systems, non-transitory computer-readable media, and methods that generate a dynamic cross-platform ask interface; Fig. 3 teaches answer providing servers 102b, 102c):
receiving, at the neural network based text generation server comprising a generative neural network model and from a user interface on a user device, a user input inquiring on a topic relating to real-time information that is not contained in prior training data the generative neural network model has been trained on (Gupta, pg. 7 [0067]: “the cross-platform search system 108 generates platform-specific requests in response to receiving a digital text query from a client device (e.g., the client device 110)… Additionally or alternatively, the cross-platform search system 108 receives the digital text query 302 requesting information specific to the software platform system 114a, or information specific to a general help topic.…the cross-platform search system 108 receives a digital text query 302…the cross-platform search system 108 receives the digital text query 302 from the software platform application 112 a (e.g., the native application installed on the client device 110 that coordinates with the software platform system 114 a), where the digital text query 302 requests information specific to the software platform system 114 b in natural language” teaches the cross-platform search system receives a digital text query in natural language (corresponds to natural language input) from a software platform application (corresponds to user interface) on a client device (corresponds to user device); pg. 6 [0061]: “Indeed, in one or more implementations, the server(s) 102 a includes all, or a portion of, the cross-platform search system 108” teaches the cross-platform search system can be implemented at a server; Gupta [0091]: teaching the response to the digital text query 312 causes the client device 110 to generate results in real-time; see also Figs. 6F; Note: As stated further below, Gupta, [0022] teaches the cross-platform search system builds the cross-platform language processing model utilizing the previous digital text queries and corresponding ground truth intents included in the training data [previous digital text queries as in not real-time information];Gupta, Paragraph [0047] teaches machine learning model can include generative adversarial neural networks.);
generating, by a query processing module at the neural network based text generation server, one or more search queries based on the user input (Gupta, pg. 8 [0073]: “in response to a digital text query requesting an update on the performance of a particular campaign over the last week, the cross-platform search system 108 extracts one or more parameter values from the digital text query including the name of the campaign and the date range including the last week…the cross-platform language processing model 304 extracts parameter values along with a registered intent from the digital text query 302…the cross-platform search system 108 utilizes the already-extracted parameter values to generate the platform-specific request” teaches the cross-platform search system generates a platform-specific request (corresponds to search query) based on the digital text query in natural language (corresponds to natural language input; Gupta, Paragraph [0008] teaches the disclosed systems offer a wide range of actions, insights, and content/links, all accessible from a single, dynamic user interface; Gupta, Paragraph [0025] further teaches the cross-platform search system further receives run-time digital text queries from a dynamic user interface.); pg. 6 [0061]: “Indeed, in one or more implementations, the server(s) 102 a includes all, or a portion of, the cross-platform search system 108” teaches the cross-platform search system can be implemented at a server; Gupta, Paragraph [0007] teaches in response to receiving a digital text query on a first software platform, the disclosed systems utilize a cross-platform language processing model (trained to recognize and understand terminology across a variety of specific software platforms) to extract at least one intent from the digital text query and identify a platform-specific configuration corresponding to the extracted intent.; Fig. 3 teaches answer providing servers 102b, 102c);
generating, by the generative neural network model that is deployed at the neural network based text generation server pretrained to process a variety of natural language processing tasks (Gupta, Paragraph [0007] teaches in response to receiving a digital text query on a first software platform, the disclosed systems utilize a cross-platform language processing model [trained to recognize and understand terminology across a variety of specific software platforms]), an output text describing the topic by feeding a natural language input combining texts from the one or more search results to the generative artificial intelligence AI neural network (Gupta, Paragraph [0046] teaches the machine learning model learns to approximate complex functions and generate outputs based on inputs provided by the model; Gupta, Paragraph [0048] further teaches an artificial neural network uses sequential information associated with words in a text input (e.g., a sentence) and in which an output of a current word is dependent on computations for previous words; Gupta, Abstract, teaches the cross-platform processing model provides platform-specific, contextually based responses to natural language digital text queries; pg. 13 [0110]: “the cross-platform search server system 104 is operable on the server(s) 102 a…the cross-platform search server system 104 (or mirrored cross-platform search system 108) includes…the cross-platform language processing model 304” teaches the server implements the cross-platform language processing model; pg. 5 [0046]-[0047] teaches the cross-platform language processing model can be a generative adversarial neural networks,…; pg. 8 [0075]: “In response to receiving a response to the platform-specific request from the answer provider (e.g., from the answer provider server(s) 102 c), the cross-platform search system 108 generates a response to the digital text query 312” teaches generating a response (corresponds to output) based on an answer (corresponds to search result; see also [0099], which recites “search result”) from an answer provider; pg. 12 [0105]: “the cross-platform search system 108 generates the query response items 624a, 624b including one or more of a media player associated with a digital video, a document preview (e.g., a PDF document preview, a WORD document preview), a link to a digital content item, and/or a digital image preview or link” teaches the response (corresponds to output) includes a link to a digital content item (corresponds to reference to a data source server) an output text describing the topic, wherein the generating comprises transforming a natural language input combining the user input and the one or more search results from the real-time Internet search into output text through layers of weights associated with neurons and non-linear activation functions in the generative neural network model (Gupta [0091]: teaching in one or more embodiments, the cross-platform search system 108 provides the response to the digital text query 312 [i.e., user input] to the client device 110 to cause the client device 110 to display the response to the digital text query 312 according to the display instructions. In at least one embodiment, for example, the cross-platform search system 108 does not render the digital text query response at the server-level. Instead, the cross-platform search system 108 generates the response to the digital text query 312 including raw data received from the software platform system 114b and the display instructions associated with at least one rendition type. Thus, when provided to the client device 110, the response to the digital text query 312 causes the client device 110 to generate a display of the raw data according to the display instructions. In this way, the response to the digital text query 312 causes the client device 110 to generate results in real-time, rather than providing a pre-rendered display.; see also Figs. 6F illustrating the user query – i.e., user input 612b concurrently displayed [combined] with the real-time Internet search results – 616b comprising - 624a and 624b; Gupta, Paragraph [0080] teaches the cross-platform language processing model 304 utilizes the knowledge graph as part of a graph neural network (e.g., with one or more neural network layers that generate an intent prediction based on weights between nodes within the knowledge graph neural network). Moreover, the cross-platform search system 108 performs the act 410 by modifying internal parameters and/or the weighted edges in the knowledge graph to reflect the terms or phrases of the previous digital text queries that correspond with the ground truth intents.; Gupta, Paragraph [0046] further teaches the cross-platform language processing model is a machine learning model with one or more parameters that can be built or tuned (e.g., trained) based on inputs to approximate unknown functions. In particular, the term machine learning model learns to approximate complex functions and generate outputs based on inputs provided to the model.), wherein the generative neural network model has not been trained on any training data containing the real time information (Gupta, [0022]: the cross-platform search system builds the cross-platform language processing model utilizing the previous digital text queries and corresponding ground truth intents included in the training data [previous digital text queries as in not real-time information].; Gupta, [0059]: a user of the administrator client device 106 utilizes one or more user interfaces of the cross-platform search system 108 on the administrator client device 106 to submit platform-specific configurations associated with the software platform systems 114a, 114b along with previous digital text queries for information specific to the software platform system 114a, 114b, respectively, and ground truth intents for the previous digital text queries. Furthermore, the cross-platform search system 108 trains a cross-platform language processing model with this training data,);
However, Gupta does not distinctly disclose:
performing by a search engine at the neural network based text generation server real-time Internet search through web search on web indexing of one or more Internet data source servers based on the generated one or more search queries, wherein each of the generated one or more search queries is customized for a respective one of the one or more Internet data source servers;
obtaining one or more search result through the real-time Internet search from the one or more Internet data source servers;
and wherein the output text includes one or more sentences that are different from the user input, wherein the output text generated by the generative neural network model is modified based at least in part on one or more user-configured parameters specified by the user input, and wherein the one or more user-configured parameters comprise at least one of a type of output, an intended audience, a tone of the output, a format of the output, and a length of the output; and
and inserting, by the generative neural network model a citation in a form of a user-clickable user interface (UI) element that redirects to a webpage of at least one search result from at least on Internet data source server that provides reference authority to the one or more sentences in the output text; and
transmitting, from the server to the user device, the output text comprising the one or more sentences and the user-clickable UI element to be displayed at the user interface.
Nevertheless, Gautam teaches text generation using real-time search results… obtaining one or more search results through a real-time search at one or more data source servers based on the one or more search queries, as provided below.
obtaining one or more search result through the real-time Internet search from the one or more Internet data source servers; (Gautam et al., Col. 8 lines 54-65: “the analytics engine module 209…transforms the analytics query into a search query and sends a signal representing the search query to the search engine controller module 205. The search engine controller module 205 can perform a parallel distribution of the search query to one or more search engine server(s) (shown as 109 in FIG. 1), for example via an output signal 223 (shown as 123 in FIG. 1). The search engine(s) 109 performs search on various distributed search index nodes 215 a-215 m and sends a signal (not shown) representing the search results to the analytics engine module 209” [Note: Gautam teaches the real-time business intelligence platform performs real-time search of data source servers to produce search results; Fig. 6 teaches the output graph includes text])
performing by a search engine at the neural network based text generation server real-time Internet search through web search on web indexing of one or more Internet data source servers based on the generated one or more search queries, …;(Gautam, Abstract, teaches providing real-time business intelligence using national language queries facilitate a user to search within a data warehouse using a natural language question. Such business intelligence platform may receive a natural language based question, extract one or more key words from the natural language based question; Gautam, Col. 2, lines 48-51 teaches a real time discovery and business intelligence platform using natural language queries allows a user to search within a data warehouse and other data sources using a natural language question; Gautam, Col. 3, lines 9-11, teaches in one implementation, business intelligence may include the scope of analysis in Internet of Things (IoT) and/or Internet of Everything (IoE). Fig. 1, teaches Communication Network 105 and Col. 5, lines 42-47 teaches Communication Network 105 can be any communication network, such as the Internet, configurable to allow the one or more UEs 101, the one or more search servers 109 and business intelligence platform 103 to communicate with communication network 105.; Fig. 3 teaches Search Engine Indexer 315; Gautam, Col. 3, lines 63-6 and Col. 4, lines 1-3 teach the business intelligence platform indexes data from disparate sources into a computation search engine designed for real-time ad-hoc multi-dimensional analysis. The search engine can be used as an underlying data storage mechanism that enables fast multi-dimensional lookups in real-time, which enables real-time processing of a natural language question using cross functional-dependency algorithms without a time lapse.; Gautam, Col. 5, lines 4-10 teaches the business intelligence platform can use index files based on computational data search engine technology. Such a platform can provide natural language and search-based interfaces to analyze data and generate reports substantially in real-time without requiring a user to write queries in a query language (e.g., SQL) or use software configurations for generating reports.; Gautam, Col. 13, lines 50-56, teaches a search engine indexer 315 (similar to the search engine control module 205) can use the data sets 313 to define distributed search index nodes 317. The distributed search index nodes 317 can be similar to search index nodes 215a-215m. The search index nodes 317 can include data extracted, transformed, and loaded by the ETL layer 319.; Gautam et al., Col. 8 lines 54-65 further teach: “the analytics engine module 209…transforms the analytics query into a search query and sends a signal representing the search query to the search engine controller module 205. The search engine controller module 205 can perform a parallel distribution of the search query to one or more search engine server(s) (shown as 109 in FIG. 1)).
It would have been obvious for one of ordinary skill in the arts before the effective filing date of the claimed invention to incorporate the limitation(s) above as taught by Gautam et al. to the disclosed invention of Gupta et al.
One of ordinary skill in the arts would have been motivated to make this modification to leverage “real-time data analysis, reporting and business intelligence related to data stored in various sources, and more particularly, to providing real-time business intelligence to users, irrespective of users' technical knowledge, using natural language interfaces” (Gautam et al. Col. 1 lines 59-63).
However, the combination of Gupta in view of Gautam does not distinctly disclose:
…wherein each of the generated one or more search queries is customized for a respective one of the one or more Internet data source servers;
and wherein the output text includes one or more sentences that are different from the user input, wherein the output text generated by the generative neural network model is modified based at least in part on one or more user-configured parameters specified by the user input, and wherein the one or more user-configured parameters comprise at least one of a type of output, an intended audience, a tone of the output, a format of the output, and a length of the output; and
and inserting, by the generative neural network model a citation in a form of a user-clickable user interface (UI) element that redirects to a webpage of at least one search result from at least on Internet data source server that provides reference authority to the one or more sentences in the output text; and
transmitting, from the server to the user device, the output text comprising the one or more sentences and the user-clickable UI element to be displayed at the user interface.
Nevertheless, Nakano teaches:
and wherein the output text includes one or more sentences that are different from the user input, …(See user interface illustrated in Nakano Figure 1 where the user input is “how to train crows to bring…” and the output texts are shown below and to the right of those results there is a user-clickable citation/reference);
and inserting, by the generative neural network model a citation in a form of a user-clickable user interface (UI) element that redirects to a webpage of at least one search result from at least on Internet data source server that provides reference authority to the one or more sentences in the output text (Nakano, Figure 1, teaches observation from a text-based web-browser environment [from fined-tuned GPT3 –generative neural network model, see Abstract] as shown to human demonstrators (left) and models (right),. Figure 1, as shown, teaches link that redirects to webpage <www.birdsoutsidemywindow.org> shown as a citation/reference at the right of the search results to the left; Nakano, pg. 2 further teaches a text-based web-browsing environment that a fine-tuned language model can interact with. This allows to improve both retrieval and synthesis in an end-to end fashion using general methods such as imitation learning and reinforcement learning. And, further teaches generating answers with references: passages extracted by the model from webpages while browsing. This is crucial for allowing to judge the factual accuracy of answers without engaging in a difficult and subjective process of independent search.); and
transmitting, from the server to the user device, the output text comprising the one or more sentences and the user-clickable UI element to be displayed at the user interface (Figure 1, teaches observation from a text-based web-browser environment [from fined-tuned GPT3 –generative neural network model, see Abstract] as shown to human demonstrators (left) and models (right),. Figure 1, as shown, teaches link that redirects to webpage <www.birdsoutsidemywindow.org> shown as a citation/reference at the right of the search results to the left;).
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to have modified the dynamic cross-platform ask interface and natural language processing model, as taught by Gupta in view of Gautam, to further include the -----browser-assisted generative neural network model, as taught by Nakano, in order to improve both retrieval and synthesis in and end-to-end fashion using general methods such as imitation learning and reinforcement learning. And, further generating answers with reference:; passages extracted by the model from webpages while browsing. This is crucial for allowing to judge the factual accuracy of answers without engaging in a difficult and subjective process of independent search.(Nakano, pg. 2)
However, the combination does not distinctly disclose wherein the output text generated by the generative neural network model is modified based at least in part on one or more user-configured parameters specified by the user input, and wherein the one or more user-configured parameters comprise at least one of a type of output, an intended audience, a tone of the output, a format of the output, and a length of the output; and …wherein each of the generated one or more search queries is customized for a respective one of the one or more Internet data source servers.
Nevertheless, Mahapatra teaches wherein the output text generated by the generative neural network model is modified based at least in part on one or more user-configured parameters specified by the user input, and wherein the one or more user-configured parameters comprise at least one of a type of output, an intended audience, a tone of the output, a format of the output, and a length of the output (Mahapatra, [0031] teaches user interface module 120 provides one or more user interfaces enabling a user to interact with the text summarization system 104. Among other things, the user interface module 120 provides user interfaces allowing a user to provide inputs that control aspects regarding generation of a text summarization model. FIG. 4 provides an example of a user interface 400 that facilitates a user providing input for generating a text summarization model. As shown in FIG. 4, the user interface 400 allows a user to enter a model name 402 and provide an input dataset 404. In some configurations, no input beyond an input dataset is needed for generating a text summarization model. However, in some configurations, the text summarization system 104 enables a user to provide additional input to control various aspects of the model generation. For instance, the user interface 400 allows a user to specify various parameters, such as: the type of summarization task 406, the size of the summary 408, the dataset type 410, the model size 412, the training time 414, the number of layers of the text summarization model 416, and the number of epochs 418. It should be understood that the parameters shown in FIG. 4 are provided by way of example only and not limitation.; Mahapatra [0021] teaches embodiments of the present invention address the shortcomings of prior text summarization approaches by providing a text summarization system that auto-generates text summarization models for extractive and abstractive summarization. The text summarization system uses a combination of neural architecture search and knowledge distillation. An input dataset is provided as input to the text summarization system for generating a text summarization model. The input dataset may come from a specific domain providing examples that guide the text summarization system to learn the terminology from the given domain. Additional input may be provided to guide the model generation process, such as an indication of the text summarization task, the summary size, the model size, the number of layers, and the number of epochs, among other possible parameters that may be specified.; Mahapatra [0022] teaches given the input dataset, a language model (which may comprise, for instance, a large transformer-based model) is fine-tuned for a specific text summarization task (i.e., extractive or abstractive summarization) using the input dataset. See also, Fig. 2.);
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to have modified the dynamic cross-platform ask interface and natural language processing model, as taught by Gupta in view of Gautam and Nakano, to further include the text summarization by a large transformer-based model, as taught by Mahapatra, in order to enable the generation of text summarization models that are custom-tailored to specific content (e.g., content having unique terminology). Additionally, the text summarization models that are custom-created by the technology described herein achieve near state-of the-art results on accuracy, while being extremely cost efficient by decreasing the model size, disk-space, and inference time relative to existing text summarization models. (Mahapatra, [0023])
However, the combination does not distinctly disclose:
…wherein each of the generated one or more search queries is customized for a respective one of the one or more Internet data source servers.
Nevertheless, Sianez teaches …wherein each of the generated one or more search queries is customized for a respective one of the one or more Internet data source servers (Sianez [0102] At 521, the method 500 includes identifying sources of responses to the set of query terms. As shown in FIG. 5B, example sources of responses can include a set of user specified sources 521A, a database of sources 521B, a cache of anticipated sources 521C, and a set of internet search engines 521D. In some embodiments, the set of internet search engines 521D is predetermined. In some embodiments, the method 500 can include executing a set of search engines (e.g., the set of search engines 521D) and/or similar information retrieval system based on the set of query terms to identify the sources of responses. For example, the sources of responses can be extracted from a set of search results generated by the executed search engine (e.g., as a list of links to webpages). In some embodiments, the search engine can generate a set of search results by interrogating a curated knowledge base, such as the database of sources 521B. The curated knowledge base can be designed to provide commonly used sources by the user and/or a set of users. In some embodiments, the user can specify one or more sources (e.g., the user specified sources 521A), such as a website URL, which can be identified as a source of responses. For instance, if a user wants to learn about photosynthesis, and knows of a website about photosynthesis, the user can select to use that website as a source when executing a search. In some embodiments, the user can identify one or more files to be a source of responses (e.g., as part of the user specified sources 521A). For example, the files can be hypertext markup language (HTML) content of webpages, HTML files, rich-text formatted files, portable document format (PDF) files, word processing documents, emails, e-books, images of text, or any other suitable source of response. The method 500 can thus use any suitable source of responses provided by the user…The set of internet search engines 521D can be a set of any internet search engine, or a set of internet search engines specified by the user. The user can also specify sources of responses that should not be used when executing a search. The user may also configure a custom search engine by providing a URL where the query terms can be inserted.; Sianez [0015] teaches methods and apparatus are disclosed for providing a natural language response to a user's natural language query of a search engine or similar information retrieval system. The methods and apparatus involve using a machine learning model to transform the natural language query to a search query compatible with the search engine, identify pertinent information (also referred to herein as “prominent content”) within the search results of the predetermined search engine, and transform a portion of the pertinent information as the natural language response to the natural language query. The methods and apparatus can thus reduce the user's responsibility to manually sort through various search results to identify pertinent information. The use of natural language and the efficient presentation of search results are broadly accessible to a wide range of users, particularly for users with physical, intellectual, visual, or hearing impairments. Further, the method and apparatus can prevent threats to the user's information privacy and information security by preventing the user from accessing websites or programs that may compromise their user data; Sianez [0031] The processor 104 can optionally include one or more machine learning models);
Before the effective filing date of the claimed invention, it would have been obvious to one of ordinary skill in the art to have modified the dynamic cross-platform ask interface and natural language processing model, as taught by Gupta in view of Gautam, Nakano, and Mahapatra, to further include methods for providing a natural language response to a user’s natural query of a search engine, as taught by Sianez. The methods and apparatus can reduce the user's responsibility to manually sort through various search results to identify pertinent information. The use of natural language and the efficient presentation of search results are broadly accessible to a wide range of users, particularly for users with physical, intellectual, visual, or hearing impairments. Further, the method and apparatus can prevent threats to the user's information privacy and information security by preventing the user from accessing websites or programs that may compromise their user data. (Sianez, [0015])
Regarding Claim 2,
The combination of Gupta in view of Gautam, Nakano, Mahapatra, and Sianez teaches all of the limitations of claim 1, the combination further teaches wherein the output text is generated by the generative AI neural network model at the server, and the generative AI neural network model is a language model (Nakano, Abstract, Figure 1, and pg. 2 teach fine-tuned GPT3 – a browser assisted generative neural network model that is a large language model, wherein Figure 1 illustrates the generated output text)
Regarding Claim 5,
The combination of Gupta in view of Gautam Nakano, Mahapatra, and Sianez teaches all of the limitations of claim 1, and Gautam further teaches wherein the obtaining the one or more search results through the real-time search comprises obtaining content from a web file via a link in the one or more search results (Gautam, Col. 8 lines 54-65: “the analytics engine module 209…transforms the analytics query into a search query and sends a signal representing the search query to the search engine controller module 205. The search engine controller module 205 can perform a parallel distribution of the search query to one or more search engine server(s) (shown as 109 in FIG. 1), for example via an output signal 223 (shown as 123 in FIG. 1). The search engine(s) 109 performs search on various distributed search index nodes 215 a-215 m and sends a signal (not shown) representing the search results to the analytics engine module 209” teaches the real-time business intelligence platform performs real-time search of data source servers to produce search results; Col. 6 line 45 to Col. 7 line 17: “The search engine server(s) 109 each can be, for example, a web server configured to provide search capabilities to electronic devices, such as UEs 101…The UEs 101 each can include a web browser configured to access a webpage or website hosted on or accessible via the business intelligence platform 103 over communication network 105…a user of a UE 101 can access a search engine server 109 via a URL designated for the search engine server 109 ” teaches obtaining search results includes obtaining content from web files through a URL (link)).
Motivation to combine same as stated for claim 1.
Regarding Claim 6,
The combination of Gupta in view of Gautam Nakano, Mahapatra, and Sianez teaches all of the limitations of claim 1, and the combination further teaches wherein the output text is a summary of the one or more search results, and the user-clickable citation comprises a reference to the at least one search result from at least one Internet data source server indicates the one or more sentences relates to the at least one Internet data source server (Gupta, Fig. 6F and pg. 12 [0105]: “the cross-platform search system 108 generates the query response items 624a, 624b including one or more of a media player associated with a digital video, a document preview (e.g., a PDF document preview, a WORD document preview), a link to a digital content item, and/or a digital image preview or link” teaches the response (corresponds to output) includes a link to a digital content item (corresponds to reference to a data source server) wherein Fig. 6F teaches the output (as a reference to at least one data source server) indicates a portion of text that is a summary of the search result;).
Motivation to combine same as stated above for claim 1.
Regarding Claim 7,
The combination of Gupta in view of Gautam Nakano, Mahapatra, and Sianez teaches all of the limitations of claim 1, and Gupta further teaches wherein the user input comprises one or more of a text input, an audio input, an image input, and a video input (Gupta pg. 4 [0039]: “As used herein, a "digital text query" refers to one or more words and/or phrases that form a request for information. The cross-platform search system can identify a digital text query based on user input via a search interface, such as text input via a search bar or audio input device” teaches the digital text query in natural language (corresponds to natural language input) can be a text input or audio input; Gupta, Paragraph [0008] teaches the disclosed systems offer a wide range of actions, insights, and content/links, all accessible from a single, dynamic user interface; Gupta, Paragraph [0025] further teaches the cross-platform search system further receives run-time digital text queries from a dynamic user interface.); .
Regarding Claim 8 (as amended), the claim recites the same and/or analogous limitations to claim 1 (as amended). Therefore, claim 8 is rejected based on the same rationale and motivation as claim 1.
Gupta also teaches a system…the system comprising: a communication interface (Gupta Fig. 3 and pg. 7 [0067]: “the cross-platform search system 108 generates platform-specific requests in response to receiving a digital text query from a client device (e.g., the client device 110)…the cross-platform search system 108 receives a digital text query 302…the cross-platform search system 108 receives the digital text query 302 from the software platform application 112 a (e.g., the native application installed on the client device 110 that coordinates with the software platform system 114 a), where the digital text query 302 requests information specific to the software platform system 114 b in natural language” teach the answer provider has a “gateway” (corresponds to a communication interface) that receives a digital text query in natural language (corresponds to natural language input) from a software platform application (corresponds to user interface) on a client device (corresponds to user device); pg. 6 [0061]: “Indeed, in one or more implementations, the server(s) 102 a includes all, or a portion of, the cross-platform search system 108” teaches the cross-platform search system can be implemented at a server; Gupta, Abstract, teaches the present disclosure relates to systems, non-transitory computer-readable media, and methods that generate a dynamic cross-platform ask interface; Fig. 3 teaches answer providing servers 102b, 102c.; Paragraph [0047] teaches machine learning model can be a generative adversarial neural network model.);
a server implementing a generative artificial intelligence (AI) neural network and a plurality of processor-executable instructions; and one or more processors executing the instructions to perform operations comprising (Gupta, pg. 13 [0110]: “the cross-platform search server system 104 is operable on the server(s) 102 a…the cross-platform search server system 104 (or mirrored cross-platform search system 108) includes…the cross-platform language processing model 304” teaches the server implements the cross-platform language processing model; pg. 5 [0046]-[0047] teaches the cross-platform language processing model can be a generative adversarial neural networks (corresponds to generative artificial intelligence (AI) neural network); pg. 16 [0136] teaches processors executing the instructions).
Regarding Claim 9,
Claim 9 recites the same and/or analogous limitations to claim 2. Therefore, claim 9 is rejected based on the same rationale as claim 2.
Regarding Claim 12,
Claim 12 recites the same and/or analogous limitations to claim 5. Therefore, claim 12 is rejected based on the same rationale as claim 5.
Regarding Claim 13,
Claim 13 recites the same and/or analogous limitations to claim 6. Therefore, claim 13 is rejected based on the same rationale as claim 6.
Regarding Claim 14,
Claim 14 recites the same and/or analogous limitations to claim 7. Therefore, claim 14 is rejected based on the same rationale as claim 7.
Regarding Claim 15,
Claim 15 (as amended) recites the same and/or analogous limitations to claim 1 (as amended). Therefore, claim 15 is rejected based on the same rationale and motivation as claim 1.
Gupta also teaches a processor-readable non-transitory storage medium storing a plurality of processor-executable instructions…the instructions being executed by one or more processors to perform operations comprising (Gupta, pg. 16 [0136] teaches a processor-readable non-transitory storage medium storing a plurality of processor-executable instructions).
Regarding Claim 16,
Claim 16 recites analogous limitations to claim 2. Therefore, claim 16 is rejected based on the same rationale as claim 2.
Regarding Claim 19,
Claim 19 recites the same and/or analogous limitations to claim 5. Therefore, claim 19 is rejected based on the same rationale as claim 5.
Regarding Claim 20,
Claim 20 (as amended) recites the same and/or analogous limitations to claim 6. Therefore, claim 20 is rejected based on the same rationale as claim 6.
Claims 3, 10, and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Gupta in view of Gautam Nakano, Mahapatra, and Sianez, as applied to claims 1, 8, and 15, and further in view of Chai et al. (US 2023/0205832 A1, filed Dec. 29, 2021 and published Jun. 29, 2023)
Regarding Claim 3,
The combination of Gupta in view of Gautam Nakano, Mahapatra, and Sianez teaches all of the limitations of claim 1, however, the combination does not distinctly disclose further comprising: transmitting a text generation input comprising the one or more search results to an external server hosting a language model; and obtaining the output text from the external server.
Nevertheless, Chai teaches further comprising: transmitting a text generation input comprising the one or more search results to an external server hosting a language model (Chai, Fig. 1 and pg. 3 [0026]: “The asset generator system 106 includes a document understanding module 114 that obtains the webpage 110 and recognizes portions of the webpage 110. The webpage 110 includes content 115, where the content 115 can include a title, section headers, body text, etc. The document understanding module 114 detects a language of the content 115 in the webpage 110, extracts a title from HTML of the webpage 110…the document understanding module 114 can employ natural language processing (NLP) technologies, image analysis technologies, and the like in connection with recognizing and extracting text from the content 115 of the webpage 110” teaches transmitting the webpages (correspond to search results) as input to an external computing system (server) that hosts the generator model (corresponds to language model)); and
obtaining the output text from the external server (Chai, Fig. 1 teaches the external computing system (server) provides asset outputs; see also Fig. 5 Step 512; Chai, Paragraph [0021]: Electronic summary documents are conventionally presented on search engine results pages (SERPs) in response to receipt of user queries. In a non-limiting example, an electronic summary document is a text advertisement.).
It would have been obvious for one of ordinary skill in the arts before the effective filing date of the claimed invention to incorporate the limitation(s) above as taught by Chai et al. to the disclosed invention of Gupta in view of Gautam Nakano, Mahapatra, and Sianez.
One of ordinary skill in the arts would have been motivated to make this modification to leverage the “generation of multiple, diverse electronic summary documents for a webpage” (Chai et al. pg. 1 [0005]).
Regarding Claim 10,
Claim 10 recites the same and/or analogous limitations to claim 3. Therefore, claim 10 is rejected based on the same rationale and motivation as claim 3.
Regarding Claim 17,
Claim 17 recites the same and/or analogous limitations to claim 3. Therefore, claim 17 is rejected based on the same rationale and motivation as claim 3.
Conclusion
The following prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
Stiennon et al., “Learning to summarize from human feedback” (2020)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to BEATRIZ RAMIREZ BRAVO whose telephone number is 571-272-2156. The examiner can normally be reached Mon. - Fri. 7:30a.m.-5:00p.m..
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, USMAAN SAEED can be reached at 571-272-4046. 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.
/B.R.B./Examiner, Art Unit 2146
/USMAAN SAEED/Supervisory Patent Examiner, Art Unit 2146
Prosecution Insights