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 .
Election/Restrictions
1. Applicant’s response to Restriction/Election requirement received 09/08/2026 is entered. Examiner agrees with the Applicant’s arguments with traverse and Requirement for Restriction/Election is now withdrawn.
Claims 1-20 filed 02/25/2025 are pending for examination.
2. Continuity: This application filed 02/25/2025 is a Continuation of PCT/CN2023/114023 , filed 08/21/2023 claims foreign priority to 202211032748.0, filed 08/26/2022.
Claim Rejections - 35 USC § 101
3. 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--20 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more, when analyzed as per MPEP 2106.
Step 1 analysis:
Claims 1-16 are to a process comprising a series of steps, and clams 17-20 to a device, which are statutory (Step 1: Yes).
Step 2A Analysis:
Claim 1 recites:
1. A recommendation method applied to an electronic device, wherein the method comprises:
(i) obtaining a first feature representation and a second feature representation, wherein the first feature representation corresponds to attribute information of a target user, the second feature representation corresponds to attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations;
(ii) determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations;
(iii) fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and
(iv) based on the recommendation information meeting a preset condition, recommending the target item to the target user.
Step 2A Prong 1 analysis: This part of the eligibility analysis evaluates whether the claim recites a judicial exception. As explained in MPEP 2106.04, subsection II, a claim “recites” a judicial exception when the judicial exception is “set forth” or “described” in the claim.
Claims 1-20 recite abstract idea.
The highlighted limitations comprising , “ determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and based on the recommendation information meeting a preset condition, recommending the target item to the target user”, under their broadest reasonable interpretation relates to a commercial activity, which benefits entities providing items and improving user satisfaction, see Specification para 0004, “ The selectivity prediction can maximize benefits of an enterprise and improve user satisfaction. The recommendation system needs to consider both a selectivity of an item to a user and an item price. The selectivity is predicted by the recommendation system based on historical behavior of the user, and the item price represents benefits of the system that are obtained after the item is selected/downloaded. . “ and fall within , Certain Methods of Organizing Human Activity”, grouping of abstract ideas. See MPEP 2106.04(a)(2), subsection II.
The limitations, ““ determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and based on the recommendation information meeting a preset condition, recommending the target item to the target user”, under their broadest reasonable interpretation, fall within the mental process groupings of abstract ideas because they cover concepts performed in the human mind, including observation, evaluation, judgment, and opinion. See MPEP 2106.04(a)(2), subsection III. These limitations, as drafted, is a simple process that covers performance of the limitations in the mind but for the recitation of “by a device”. That is, other than reciting “by a device” nothing in the claim elements precludes the steps from practically being performed in the mind. For example, but for the “by a device” language, the claim encompasses a person looking at collected data, analyzing the same by comparing them and finding the similarities between the attributes of a target user and the target item and finding that the recommendation information meets a preset condition for recommending the target item to the target user. The mere nominal recitation of by a device does not take the claim limitations out of the mental process grouping. Thus, the claim 1 recites a mental process. See MPEP 2106.04(a)(2) Abstract Idea Groupings [R-07.2022] II. MENTAL PROCESSES: claims do recite a mental process when they contain limitations that can practically be performed in the human mind, including for example, observations, evaluations, judgments, and opinions. Examples of claims that recite mental processes include:• a claim to "collecting information, analyzing it, and displaying certain results of the collection and analysis," where the data analysis steps are recited at a high level of generality such that they could practically be performed in the human mind, Electric Power Group v. Alstom, S.A., 830 F.3d 1350, 1353-54, 119 USPQ2d 1739, 1741-42 (Fed. Cir. 2016); • a claim to collecting and comparing known information (claim 1), which are steps that can be practically performed in the human mind, Classen Immunotherapies, Inc. v. Biogen IDEC, 659 F.3d 1057, 1067, 100 USPQ2d 1492, 1500 (Fed. Cir. 2011);
Two Abstract Ideas:
”. If a claim that includes two or more abstract ideas groupings per Step 2A, Prong One , as per MPEP 2106.04, subsection IIB, under such circumstances, the Supreme Court has treated such claims in the same manner as claims reciting a single judicial exception. Id. (discussing Bilski v. Kappos, 561 U.S. 593 (2010)).
Here, steps (b), (d), and (e) fall within the mental process grouping of abstract ideas, and steps (b) and (c) fall within the mathematical concepts grouping of abstract ideas. Limitations (b)-(e) are considered together as a single abstract idea for further analysis. (Step 2A, Prong One: YES
Two Abstract Ideas:
If a claim that includes two or more abstract ideas groupings per Step 2A, Prong One , as per MPEP 2106.04, subsection IIB, under such circumstances, the Supreme Court has treated such claims in the same manner as claims reciting a single judicial exception. Id. (discussing Bilski v. Kappos, 561 U.S. 593 (2010)). Here, the limitations, as discussed above, fall within the certain methods of organizing human activity and mental process groupings of abstract ideas, which are considered together as a single abstract idea for further analysis. (Step 2A, Prong One: YES).
Sinc the other two independent claims 11 and 17 recite similar limitations, they are analyzed on the same basis reciting an abstract idea.
Thus, claims 1, 11, and 17 with their dependent claims 2-10, 12-17, and 18-20 recite an abstract idea (Step 2A, Prong One: YES).
Step 2A Prong 2 analysis: This part of the eligibility analysis evaluates whether the claim as a whole integrates the recited judicial exception into a practical application of the exception or whether the claim is “directed to” the judicial exception. This evaluation is performed by (1) identifying whether there are any additional elements recited in the claim beyond the judicial exception, and (2) evaluating those additional elements individually and in combination to determine whether the claim as a whole integrates the exception into a practical application. See MPEP 2106.04(d).
Claims 1-20: The judicial exception is not integrated into a practical application.
Claim 1 recites the additional limitations of using a generic computer device implementing the steps of:”(i) obtaining a first feature representation and a second feature representation, wherein the first feature representation corresponds to attribute information of a target user, the second feature representation corresponds to attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations; (ii) determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; (iii) fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and (iv) based on the recommendation information meeting a preset condition, recommending the target item to the target user. “.
The limitations “(:”(i) obtaining a first feature representation and a second feature representation, wherein the first feature representation corresponds to attribute information of a target user, the second feature representation corresponds to attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations” are mere data gathering recited at a high level of generality, and thus are insignificant extra-solution activity. See MPEP 2106.05(g) (“whether the limitation is significant”). In addition, all uses of the recited judicial exceptions require such data gathering and output, and, as such, these limitations do not impose any meaningful limits on the claim. These limitations amount to necessary data gathering and outputting. See MPEP 2106.05. Further, these limitations are recited as being performed by a computer device recited at a high level of generality and the computer device is used as a tool to perform the generic computer function of receiving data. See MPEP 2106.05(f).
In limitations “ii)determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; (iii) fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and (iv) based on the recommendation information meeting a preset condition, recommending the target item to the target user”, the computer device is used to perform an abstract idea, as discussed above in Step 2A, Prong One, such that it amounts to no more than mere instructions to apply the exception using a generic computer. See MPEP 2106.05(f).
Even when viewed individually and in combination, these additional elements in claim 1 do not integrate the recited judicial exception into a practical application (Step 2A, Prong Two: NO), and the claim is directed to the judicial exception. (Step 2A: YES). Since the limitations of the independent claim 17 recites similar limitations as claim 1, they are analyzed on the same basis as directed to an abstract idea.
The independent claim 11 recites additional elements comprising::
obtaining a first feature representation and a second feature representation, wherein the first feature representation is obtained through feature extraction by a first encoder on attribute information of a target user, the second feature representation is obtained through feature extraction by a second encoder on attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations [These limitations recite using first and second encoders to extract features from the user and item information merely amount to generally linking the use of judicial exception and the insignificant extra=solution activity of obtaining data to a particularly technological environment or field of use [the encoders] ; (ii)determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; and (iii) fusing the plurality of similarities to obtain recommendation information between the target user and the target item [In these two steps the computer device is used to perform an abstract idea, as discussed above in Step 2A, Prong One, such that it amounts to no more than mere instructions to apply the exception using a generic computer. See MPEP 2106.05(f).
The limitations, “wherein the recommendation information and a corresponding similarity annotation are used to determine a first loss, the first loss is used to update the first encoder and the second encoder, and the similarity annotation is obtained based on a real operation log of the target user “, recite determining first loss for the encoders which is a long standing practice of using calculating a loss function to find errors and to update results and the limitations , as recited do not reflect any technical improvement in the calculating the loss function or to any hardware or software. Therefore, the step of calculating first loss is recited in a nominal manner and does not integrate the abstract idea into a practical application because they do not add meaningful limits on practicing the abstract idea.
Even when viewed individually and in combination, these additional elements in claim 11 do not integrate the recited judicial exception into a practical application (Step 2A, Prong Two: NO), and the claim is directed to the judicial exception. (Step 2A: YES).
Dependent claims 2-10, 12-16, and 18-20:
Claim 2 recites segmenting feature representations, claim 3 recites describing the similarities, and the fusing act by fusing maximum values of the plurality of similarities, claims 9-10 recite describing the attribute information of the target user and the target item. These limitations recite mental processes and non-functional descriptive subject matter which do not integrate the abstract idea into a practical application, because they do not add any meaningful limits on practicing the abstract idea. Claim 4-8 recite using encoders for obtaining feature representations including attribute information in form of embedding vectors and then using neural network layers including a connected layer to determine weights of the attributes. The use of encoders and neural network is recited in a nominal manner. Using encoders which are neural network components include multiple layers to turn text such as attributes of the user and the item into numerical representations such as embedding vectors, wherein the dense [fully connected layer] is used for data such as demographics, age of a user and for attributes of an item. The encoder may comprise stacking multiple layers for passing data through a chain of mathematical transformations such as embeddings vectors. Limitations in claims 4-8, as recited, do not provide details reflecting technical improvement in the function of encoders , or hardware or software triggering specific mathematical transformation. Therefore, the limitations of claims 4-8, as recited, represent as mere long standing practice before the effective date of the claimed invention and the claim limitations are not directed to an improvement in the functioning of encoders. The recitation of “using encoders” in limitations of claims 4-8 also merely indicates a field of use or technological environment in which the judicial exception is performed. Although the additional element “using encoders” limit the identified judicial exceptions “of “obtaining feature representations of a target user and a target item from text data using encoders ,” this type of limitation merely confines the use of the abstract idea to a particular technological environment (encoders -neural networks) and thus fails to add an inventive concept to the claims. See MPEP 2106.05(h). Thus, limitations of claims 4-8 do not integrate the abstract idea into a practical application, because they do not add any meaningful limits on practicing the abstract idea.
Limitations of dependent claim 12 are directed to determining second loss are similar as discussed for the base claim 11 and for the same reasons do not integrate the abstract idea into a practical application, because they do not add any meaningful limits on practicing the abstract idea. Since limitations of claim 13-16 are similar to the limitations of claims 2-8, and limitations of claims 18-20 are similar to the limitations of claims 2-4, they are analyzed on the same basis as directed to an abstract idea.
Even when viewed individually and in combination, the additional elements in claims 1-20 do not integrate the recited judicial exception into a practical application (Step 2A, Prong Two: NO), and the claim is directed to the judicial exception. (Step 2A: YES).
Step 2B analysis: This part of the eligibility analysis evaluates whether the claim as a whole amounts to significantly more than the recited exception i.e., whether any additional element, or combination of additional elements, adds an inventive concept to the claim. See MPEP 2106.05.
The claims 1-20 do not include additional elements that are sufficient to amount to significantly more than the judicial exception.
Since claims are as per Step 2A are directed to an abstract idea, they have to be analyzed per Step 2B, if they recite an inventive step, i.e., the claims recite additional elements or a combination of elements that amount to “Significantly More” than the judicial exception in the claim.
As discussed above with respect to Step 2A Prong Two, the additional elements in the claims 1-20 amount to no more than mere instructions to apply the exception using a generic computer components, and generally linking the judicial exception to a particular technological environment or field of use. The same analysis applies here in 2B, i.e., mere instructions to apply the exception using a generic computer components, and generally linking the judicial exception to a particular technological environment or field of use using a generic computer components cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B. The additional elements using encoders [neural network consisting multiple layers including calculating the loss function] in claims 4-8, 11-12, 15-16 and 20, as analyzed for Step 2A merely amount to Simply appending well-understood, routine, conventional activities previously known to the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions that are well-understood, routine and conventional activities previously known to the industry, as discussed in Alice Corp., 573 U.S. at 225, 110 USPQ2d at 1984 (see MPEP § 2106.05(d)).
As per MPEP 2106 , a conclusion that an additional element or elements is/are extra-solution activity, or are well-understood, conventional and routine activity in step 2A should be re-evaluated in step 2B. Here the steps of obtaining data and transmitting/output data were considered are extra-solution activity, or are well-understood, conventional and routine activity activities in step 2A and thus it is re-evaluated in step 2B to determine if it is more than what is well-understood, routine, conventional activity in the field. The background of the example does not provide any indication that the computer components are anything other than a generic, off the shelf computer component and the Symantec, TLI, OIP Techs, Versata court decisions cited in MPEP 2106.05(d) (ii) indicate that mere obtaining/receiving/acquiring and transmitting data using a generic computer is a well-understood, routine, conventional function when it is claimed in a merely generic manner (as it is here). Accordingly, a conclusion that the receiving, acquiring, transmitting, and displaying steps are well-understood, routine conventional activities are supported under Berkheimer Option 2. See MPEP 2106.05 (f) 2: Whether the claim invokes computers or other machinery merely as a tool to perform an existing process. Use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more. See Affinity Labs v. DirecTV, 838 F.3d 1253, 1262, 120 USPQ2d 1201, 1207 (Fed. Cir. 2016) (cellular telephone); TLI Communications LLC v. AV Auto, LLC, 823 F.3d 607, 613, 118 USPQ2d 1744, 1748 (Fed. Cir. 2016) (computer server and telephone unit). Similarly, "claiming the improved speed or efficiency inherent with applying the abstract idea on a computer" does not integrate a judicial exception into a practical application or provide an inventive concept. Intellectual Ventures I LLC v. Capital One Bank (USA), 792 F.3d 1363, 1367, 115 USPQ2d 1636, 1639 (Fed. Cir. 2015). In contrast, a claim that purports to improve computer capabilities or to improve an existing technology may integrate a judicial exception into a practical application or provide significantly more. McRO, Inc. v. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1314-15, 120 USPQ2d 1091, 1101-02 (Fed. Cir. 2016); Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335-36, 118 USPQ2d 1684, 1688-89 (Fed. Cir. 2016). See MPEP §§ 2106.04(d)(1) and 2106.05(a) for a discussion of improvements to the functioning of a computer or to another technology or technical field.
Even when considered in combination, the additional elements in claims 1-20 represent mere instructions to implement an abstract idea or other exception on a computer and insignificant extra-solution activity, which do not provide an inventive concept. (Step 2B: NO).
Claims 1-20, as drafted, are not patent eligible.
Claim Rejections - 35 USC § 102
4. 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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1-5, 9-10, 11-15, 17-20 are rejected under 35 U.S.C. 102(a)(2) as being anticipated by Fan et al. [US 2023/0401464 A1], hereinafter Fan.
Note: The bolded text represents “[[emphasis added to the relevant text]] direct quotes from the references are in a smaller and italic font.
Regarding claim 1, Fan teaches a recommendation method applied to an electronic device, wherein the method comprises [See Abstract, “ a method includes obtaining a pre-trained recommender model that has been trained using contrastive learning [is a machine learning technique for training models to understand data by retrieving similar items closer together and pushing dissimilar items farther away in an embedding space] and with feature-level augmentation and instance-level augmentation”, and paras 0009—0010, describe that such a method 700 is implemented by an electronic device including one or more processors and memory storing one or more programs. The one or more programs include instructions for performing any of the methods described herein (e.g., the method 700):
obtaining a first feature representation and a second feature representation, wherein the first feature representation corresponds to attribute information of a target user, the second feature representation corresponds to attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations; determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub- representations; fusing the plurality of similarities to obtain recommendation information between the target user and the target item; and based on the recommendation information meeting a preset condition, recommending the target item to the target user [ See Abstract, para 0009—0010, Figs. 5-6 and paras 0068—0069 and 0087—0088:
Abstract: In one aspect, a method includes obtaining a pre-trained recommender model that has been trained using contrastive learning with feature-level augmentation and instance-level augmentation. The method further includes generating, via the model, a user embedding based on features of the user and generating, via the model, a respective episode embedding for each episode of a plurality of episodes, each respective episode embedding based on features of the corresponding episode. The method also includes generating, via the model, a respective similarity score (corresponding to a latent similarity between the user embedding and each respective episode embedding) for each episode, the respective similarity score, and ranking the episodes in accordance with the respective similarity scores. The method further includes recommending the highest ranked episode to the user.: The method further includes generating, via the model, a user embedding based on features of the user and generating, via the model, a respective episode embedding for each episode of a plurality of episodes, each respective episode embedding based on features of the corresponding episode. The method also includes generating, via the model, a respective similarity score (corresponding to a latent similarity between the user embedding and each respective episode embedding) for each episode, the respective similarity score, and ranking the episodes in accordance with the respective similarity scores. The method further includes recommending the highest ranked episode to the user]”, and Paragraphs :0009—0010, 0068-0069, 0087—0088, and 00959: {0009] ……. an electronic device is provided. The electronic device includes one or more processors and memory ….. for performing any of the methods described herein (e.g., the method 700). [0010] …….. a non-transitory computer -readable storage medium is provided ……. for execution by an electronic device with one or more processors. The one or more programs comprising instructions for performing any of the methods described herein (e.g., the method 700). ………[0068] FIG. 5 is a block diagram illustrating a recommender framework 500 ….The recommender framework 500 includes two paths (towers). One path includes obtaining user features 504 (e.g., a user feature vector) from a user profile 502 (e.g., preferences and usage history). In some embodiments, the user features 504 include one or more of: gender, age, location, preferences, and usage history (e.g., the historical interactions 404). …….., the user features include one or more of: gender, age, country, podcast topics liked in a previous time period (e.g., 30, 60, or days), a user language, a pre-trained collaborative filtering embedding vector, a user embedding pre-trained with podcast interactions, and an averaged streaming time. A user embedding 506 is generated from the user features 504. In some embodiments, a multilayer perceptron (MLP) or feedforward artificial neural network (ANN) is used for user feature 504 transforms. [0069] The other path includes obtaining episode features 510 (e.g., an episode feature vector) for each episode of one or more episodes 508. In some embodiments, the episode features 510 include one or more of: episode topic, episode title, episode genre, and episode consumption data. In some embodiments, the episode features for a podcast include one or more of: topics, country, collaborative filtering pre-trained embeddings, and pre-trained semantic embeddings of the podcast. A respective episode embedding 512 is generated from each set of episode features 510. In some embodiments, an MLP or feedforward ANN is used for episode feature 510 transforms. A similarity score 514 (e.g., a preference score) is generated for the user embedding 506 and each episode embedding 512. In some embodiments, for discrete features, the features are encoded as one-hot or multi-hot vectors. In some embodiments, for continuous features, such as pre-trained embedding vectors, the features are input directly. In some embodiments, after encoding, all of the features for the user are concatenated and all of the features for the episode are concatenated………… [0087] The system generates (718), via the pre-trained recommender model, a user embedding (e.g., the user embedding 506) based on a plurality of features of the user. In some embodiments, the plurality of features of the user includes (720) one or more of: a gender, an age, a country, a language, a recent topic liked, a streaming statistic, and a collaborative filtering vector. [0088] The system generates (722), via the pre-trained recommender model, a respective episode embedding (e.g., the episode embedding 512) for each episode of a plurality of episodes, each respective episode embedding based on a plurality of features of the corresponding episode. In some embodiments, the plurality of features of the episode includes (724) one or more of: a topic, a country, a language, a licensor, a publisher, a collaborative filtering vector, and a semantic embedding. [0095] the respective similarity score corresponding to a latent similarity between the user embedding and each respective episode embedding; (5) ranking the plurality of episodes (e.g., using the ranker 329) in accordance with the respective similarity scores; and (6) recommending a highest ranked episode of the plurality of episodes to the user (e.g., presenting the highest ranked episode at the electronic device 102). As an example, the recommended episodes 412 are presented to the user at the electronic device 102..”].
The above excerpts read on the claimed limitations as they describe obtaining features including attributes of a specific user with sub-representations such as, gender, age, location, preferences, and usage history (e.g., the historical interactions 404). … country, podcast topics liked in a previous time period (e.g., 30, 60, or days), a user language, a pre-trained collaborative filtering embedding vector, a user embedding pre-trained with podcast interactions, and an averaged streaming time, and the features including attributes of a target episode [corresponds to the claimed target item], such as episode topic, episode title, episode genre, and episode consumption data. …country, collaborative filtering pre-trained embeddings, and pre-trained semantic embeddings of the podcast and these feature sub representations are combined and compared to determine the similarities based on the embedding vectors to obtain a recommendation based on a similarity score corresponding to a latency similarity a threshold and then recommending he highest ranked episode [the target item] to the target user.
Regarding claim 2, the limitations, “ that the method according to claim 1, wherein the plurality of first feature sub-representations are obtained by segmenting the first feature representation, and the plurality of second feature sub-representations are obtained by segmenting the second feature representation”, are already covered in the analysis of claim 1 wherein the first feature representation can be segmented into sub-representations , such as gender, age, location, preferences, and usage history (e.g., the historical interactions 404). … country, podcast topics liked in a previous time period (e.g., 30, 60, or days), a user language, and the second feature sub-representations [target item that is episode] by segmenting into an episode topic, an episode title, an episode genre, and episode consumption data. …country, collaborative filtering pre-trained embeddings, and pre-trained semantic embeddings of the podcast.
Regarding claim 3, the limitations, “ 3. (Previously Presented) The method according to claim 1, wherein the plurality of similarities comprise a plurality of groups of similarities, and similarities in each group are: similarities between one of the first feature sub-representations and the second feature sub- representations; or similarities between one of the second feature sub-representations and the first feature sub- representations; and wherein the fusing the plurality of similarities comprises: fusing maximum values of the plurality of similarities comprised in all the groups of similarities.”, are already covered in the analysis of claim , wherein the recommendation system considers the highest ranked episode based on the closes embedding vectors of the user and the target item [episode].
Regarding claims 4-5, Fan teaches the limitations, “ 4. (Original) The method according to claim 1, wherein the first feature representation is obtained by a first encoder based on the attribute information of the target user, the first encoder comprises a plurality of first fully-connected layers, the first feature representation is obtained based on outputs of M first fully-connected layers in the plurality of first fully-connected layers, and M is a positive number greater than 1, and the method according to claim 1, wherein the second feature representation is obtained by a second encoder based on the attribute information of the target item, the second encoder comprises a plurality of second fully-connected layers, and the second feature representation is obtained based on an output of a second fully-connected layer that is in the plurality of second fully-connected layers and that is closest to an output layer [ See paras 0007, 0021, and 0070—0073:
[0007] ………. The recommendation system described herein uses (i) a two-tower model, and (ii) a contrastive learning approach to improve performance and help users discover new shows in accordance with some embodiments. For example, a two-tower model determines a latent similarity between a user embedding and an episode embedding. A contrastive learning approach can include feature-level augmentation (e.g., feature dropout layer(s)) to obtain augmented episode embeddings. The contrastive learning approach can also include instance-level augmentation (e.g., identifying similar episodes using semantics, cosine similarity, and/or knowledge graph information) to obtain correlated episode embeddings. Thus, the contrastive learning approach can increase the size of the training set for the two-tower model (e.g., to overcome the sparsity problem). [0021] The disclosed embodiments include a two-tower recommender framework with contrastive learning to improve recommendations for discovery and exploration. Contrastive learning involves learning to compare items and can be used to enhance an encoder network. [0071] In some embodiments, the neural networks F.sub.u and F.sub.i are defined by:
F.sub.u.sup.L=ReLU(F.sub.u.sup.L−1W.sub.1.sup.L+b.sub.1.sup.L)
F.sub.i.sup.L=ReLU(F.sub.i.sup.L−1W.sub.2.sup.L+b.sub.2.sup.L) Equations 2: User and Episode Neural Networks
where ReLU( ) represents a rectified linear unit (ReLU) activation function, W.sub.*.sup.L represents a linear transformation, and b.sub.*.sup.L represents a bias. For example, the 0-th layer of each tower is the input feature vector of f.sub.u and f.sub.i respectively. In some embodiments, the last layer output embedding is used to make predictions, e.g., as in Equation 1 above. In some embodiments, the neural networks in Equations 2 are L-th layers fully connected neural networks…..…..[0073] FIGS. 6A-6B are block diagrams illustrating augmentation frameworks in accordance with some embodiments. …….. An augmented episode embedding 608 is generated from the masked features 606. In some embodiments, the augmented episode embedding 608 is generated using a same encoder as used for generating the episode embedding(s) 512.] . These excerpts teach using a two tower model neural networks using encoders for generating user and episode [the target item] embeddings with a contrastive learning and the neural networks are L-th fully connected layers [reads on the claimed M first fully connected layers in the plurality of fully connected layers where M is a positive number. The two-model is a neural network considering of two separate sub-networks , which function as encoders and each tower includes a stack of fully connected layers , where the last layer provides the output. The contrastive learning helps in bringing the matching pairs closer and pushing the non-matching pairs farther from each other.
Regarding claim 9, Fan teaches that The method according to claim 1, wherein the attribute information of the target user comprises at least one of the following: a gender, an age, an occupation, an income, a hobby, and an education level [See para 0068, “ The recommender framework 500 includes two paths (towers). One path includes obtaining user features 504 (e.g., a user feature vector) from a user profile 502 (e.g., preferences and usage history). In some embodiments, the user features 504 include one or more of: gender, age, location, preferences, and usage history (e.g., the historical interactions 404). …….., the user features include one or more of: gender, age, country, podcast topics liked in a previous time period (e.g., 30, 60, or days), a user language, a pre-trained collaborative filtering embedding vector, a user embedding pre-trained with podcast interactions, and an averaged streaming time.”, and [0069] The other path includes obtaining episode features 510 (e.g., an episode feature vector) for each episode of one or more episodes 508. In some embodiments, the episode features 510 include one or more of: episode topic, episode title, episode genre, and episode consumption data. In some embodiments, the episode features for a podcast include one or more of: topics, country, collaborative filtering pre-trained embeddings, and pre-trained semantic embeddings of the podcast.
Regarding claim 10, Fan teaches that the method according to claim 1, wherein the attribute information of the target item comprises at least one of the following: an item name, a developer, an installation package size, a category, and a positivity rating [See para 0069, “ [0069] The other path includes obtaining episode features 510 (e.g., an episode feature vector) for each episode of one or more episodes 508. In some embodiments, the episode features 510 include one or more of: episode topic, episode title, episode genre, and episode consumption data. In some embodiments, the episode features for a podcast include one or more of: topics, country, collaborative filtering pre-trained embeddings, and pre-trained semantic embeddings of the podcast.].
Regarding claims 17-20, their limitations are already covered in the analysis of claims 1-4 as anticipated by Fan.
Claim Rejections - 35 USC § 103
5. 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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
Note: The bolded text represents “[[emphasis added to the relevant text]] direct quotes from the references are in a smaller and italic font.
5.1. Claims 6-8, are rejected under 35 U.S.C. 103 as being unpatentable over Fan in view of Mukherjee et al. [US 20230420085 A1]; hereinafter Mukherjee.
Regarding claim 6, Fan teaches the limitations of claims 1 and 4, as analyzed above using Two-tower model including neural networks encoders including the limitations and including the limitations that wherein the attribute information of the target user comprises user attributes in a plurality of dimensions, and the first feature representation is obtained by processing, by the first encoder, a first target embedding representation corresponding to the attribute information of the target user; and the method further comprises: obtaining a first initial embedding representation corresponding to the attribute information of the user, wherein the first initial embedding representation comprises a plurality of first embedding vectors, and each first embedding vector corresponds to a user attribute in one dimension, as already analyzed for claims 4-6 above, but fails to disclose explicitly processing the first initial embedding representation over a weight determining network, to obtain first weights of all the user attributes, and adjusting the plurality of first embedding vectors based on the first weights, to obtain the first target embedding representation. It is a routine practice that weights are learned during training a model wherein the weights learned can be positive or negative deepening when a feature helps or hurts the recommendation, see Mukherjee. Mukherjee, in the same field of endeavor [See Abstract and para 0005] using a two tower model with contrastive learning using neural networks encoders to convert data to embedding in a shared embedding space teaches obtaining weights of the attributes of an object [which can include a user or an item] for generating embeddings of both objects [See Mukherjee paras 0036, “ the reaction encoder 210 and the procedure encoder 212 of the machine learning system 100 are trained with labeled pair-wise data through contrastive learning. Training enables the encoders to learn the weights for generating the reaction embeddings 214 and the procedure embeddings 216. “, wherein the reaction and procedure embedding correspond to the user and the episode embeddings in Fan and the user and item embedding in the Applicant’s claim. Therefore, in view of the teachings of Mukherjee, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Fan to incorporate obtaining weights for the user and item attributes to generate or adjust the user and the item embeddings, because the weights can indicate that they are positive or negative that is if an attribute would help or hurt making a recommendation, and secondly since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Regarding claim 7, its limitations are already covered in the analysis of claims6 above.
Regarding claim 8, the limitations, “ The method according to claim 6, wherein the weight determining network comprises only one fully-connected (FC) layer”, are already covered in the analysis of claims 4-5 above, see Fan para 0071, “ the neural networks are L-th fully connected layers”]
5.2. Claim 11-15 are rejected under 35 U.S.C. 103 as being unpatentable over Fan in view of CN 111696678 A, hereinafter CN;678A.
Regarding claim 11, Fan teaches all the limitations, “A data processing method, wherein the method comprises: obtaining a first feature representation and a second feature representation, wherein the first feature representation is obtained through feature extraction by a first encoder on attribute information of a target user, the second feature representation is obtained through feature extraction by a second encoder on attribute information of a target item, the first feature representation comprises a plurality of first feature sub-representations, and the second feature representation comprises a plurality of second feature sub-representations; determining a plurality of similarities based on the plurality of first feature sub- representations and the plurality of second feature sub-representations, wherein each similarity is a similarity between one of the first feature sub-representations and one of the second feature sub-representations; and fusing the plurality of similarities to obtain recommendation information between the target user and the target”, are already covered in the analysis of claim 1 as anticipated by Fan. Fan does not teach calculating loss function to determine errors and helps to update the recommendation. CN’678A, in the same field of applying neural network encoder, teaches determining a loss function using recommendation information[correspond to the recommendation information obtained via encoder] and corresponding actual information [corresponds to the similarity annotation obtained based on a real operation log of the target user] to update the first encoder [for the user] and the second encoder [for the target item or the episodes in Fan reference] [See CN’ 678A page 12, “ the invention define the loss function of the effect prediction model, so as to train the effect prediction model through the loss function. the loss function is composed of the minimum reconstruction error of the SDAE self-encoder in the characteristic extraction process and the prediction error of the curative effect prediction model, the prediction error is determined according to the prediction effect evaluation information and the actual effect evaluation information. “]. Therefore, in view of the teachings of CN’678A in the same field of endeavor, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified FAN to incorporate the concept that the recommendation information and a corresponding similarity annotation are used to detinning a first loss, the first loss is used to update the first encoder and the second encoder, and the similarity annotation is obtained based on a real operation log of the target user, because, as shown in CN’678A, it would help in updating the first and the second encoder by turning prediction errors into a numerical score for guiding the training of the encoders and secondly, since the claimed invention is merely a combination of old elements, and in the combination each element merely would have performed the same function as it did separately, and one of ordinary skill in the art would have recognized that the results of the combination were predictable.
Regarding claims 12- 15 their limitations are already covered in the analysis of claims 1-6 as anticipated by Fan.
5.3. Claim 16 is rejected under 35 U.S.C. 103 as being unpatentable over Fan in view of C’678A in view of Mukherjee.
Regarding claim 16, its limitations , “ The method according to claim 11, wherein the attribute information of the target user comprises user attributes in a plurality of dimensions, and the first feature representation is obtained by processing, by the first encoder, a first target embedding 4 representation corresponding to the attribute information of the target user; and the method further comprises: obtaining a first initial embedding representation corresponding to the attribute information of the user, wherein the first initial embedding representation comprises a plurality of first embedding vectors, and each first embedding vector corresponds to a user attribute in one dimension; processing the first initial embedding representation over a weight determining network, to obtain first weights of all the user attributes; and adjusting the plurality of first embedding vectors based on the first weights, to obtain the first target embedding representation [are already covered in the analysis of claims 6-7 above] as being unpatentable over Fan in view of Mukherjee and the limitations that the first loss is further used to update the weight determining network [is already covered in the analysis of claim 11 above in view of CN’678A] .
Conclusion
6. The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Foreign reference:
i. WO 2022109134 A1 [See paras 0023—0024, “ describes contrastive learning that learns similar/dissimilar representations from data, based on a similarity measure and a contrastive loss function, wherein the contrastive loss includes a contrastive loss as a regularizer, which is joint optimized with the encoder loss .
NPL reference:
ii. W. Yan, D. Wang, M. Cao and J. Liu, "Deep Auto Encoder Model With Convolutional Text Networks for Video Recommendation," in IEEE Access, vol. 7, pp. 40333-40346, 2019, retrieved from IP.COM on 10/10/2026 describes deep learning-based recommendation model to solve the sparsity and cold start recommendation problems by mining the auxiliary data of users' viewing behavior datasets (e.g., the user attribute features information and video item attribute features information) and to deeply mine the latent information and their correlations of the user features and item features. First of all, the user features and video item features are processed and deeply mined by the data preprocessing layer, embedding dense layer, convolution network layer, share layer, and the auto encoder layer of the proposed model. After that, the final predictive rating process is conducted in multi-layer perception by combining with the target rating vector data and the processed user and item feature data, which is deeply mined by the above-mentioned submodels of our proposed algorithm model. The user’s preference for an item is represented by a 0-1 value and . Finally, the items corresponding to the predicted values of the output layer nodes in the model are sequentially recommended to the user.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to YOGESH C GARG whose telephone number is (571)272-6756. The examiner can normally be reached Max-Flex.
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, Maria-Teresa 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.
YOGESH C. GARG
Primary Examiner
Art Unit 3688
/YOGESH C GARG/Primary Examiner, Art Unit 3688
