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 .
Status of Claims
The present application is being examined under the claims filed 11/21/2025.
Claims 1-20 are pending.
Response to Amendment
This Office Action is in response to Applicant’s communication filed 11/21/2025 in response to office action mailed 08/21/2025. The Applicant’s remarks and any amendments to the claims or specification have been considered with the results that follow.
Response to Arguments
Regarding Objections and Informalities
In Remarks page 11, Argument 1
(Examiner summarizes Applicant’s arguments) Applicant argues that amendments obviate the objection to claim 11.
Examiner’s response to Argument 1
Examiner agrees that the amendments obviate the objection and the objection is withdrawn.
Regarding 35 U.S.C. 112(f)
In Remarks page 11, Argument 2
(Examiner summarizes Applicant’s arguments) Applicant argues that the claims have been amended to avoid interpretation under 35 U.S.C. 112(f).
Examiner’s response to Argument 2
Examiner agrees that the claims no longer invoke 35 U.S.C. 112(f) and thus the interpretations are withdrawn.
Regarding 35 U.S.C. 101
In Remarks page 12, Argument 3
(Examiner summarizes Applicant’s arguments) Applicant argues that, under step 1, each of the claims are directed to a statutory category.
Examiner’s response to Argument 3
Examiner agrees that each of the claims is directed to a statutory category under 35 U.S.C. 101.
In Remarks page 13, Argument 4
(Examiner summarizes Applicant’s arguments) Applicant argues that recitation of computing nonnegative matrices has been removed from claim 1 and thus claim 1 is not directed to a mathematical process. Applicant argues that recitation of computing a probability of an additional edge and is thus not directed to a mental process, and further that the limitations are different from matrix multiplications and activation functions. Applicant argues that the newly added limitations are not directed to mental processes because the method uses a specialized machine learning model for content recommendation.
Examiner’s response to Argument 4
While Examiner agrees that the limitations of claim 1 no longer explicitly recite any mathematical concepts, the newly added limitations of claim 1 still recite abstract ideas in the form of mental processes. Merely performing a mental process with a generic machine learning model in the technological environment of content recommendation does not render the limitations non-abstract ideas. For example, “determining a homophilous cluster affinity among the plurality of users” amounts to performing an evaluation of the similarity between data points. A human being could observe that two people (users) who talk to each other frequently online are male and determine that they are friends (a homophilous cluster affinity).
In Remarks page 15, Argument 5
(Examiner summarizes Applicant’s arguments) Applicant argues that the additional elements of receiving a dataset of interactions between users and content items, and providing the content item to a user integrate claim 1 into a practical application. Applicant further argues that the elements of claim 1 are integrated into a practical application of providing more accurate and efficient data augmentation and content recommendation via a specialized machine learning model, thus improving content recommendation to provide content to a user based on a predicted interaction. Applicant argues that technological improvements are not limited only to the functioning of computer hardware. Applicant further argues that the additional elements recite a unique and specific implementation of a machine learning model resulting in improved data augmentation and content recommendations.
Examiner’s response to Argument 6
Examiner disagrees with Applicant’s analysis. The claims do recite limitations related to content and user interactions. However, these additional elements do not integrate the claim into a judicial exception as they are recited at a high level of generality and amount to merely receiving data, outputting data, and using machine learning model without reciting any details about how the data is received, how the content is delivered to the user, nor the specific and novel inner details of the machine learning model (for example, architecture or training). Merely reciting inputting and outputting data in an ordinary way amounts to mere data gathering and outputting (see MPEP 2106.05(g)), and using an ordinary machine learning model in its ordinary capacity to apply the mental processes (see MPEP 2106.05(f)) cannot integrate the abstract idea into an abstract idea.
Regarding technical improvements, examiner’s analysis is similar. MPEP 2106.05(a) recites:
After the examiner has consulted the specification and determined that the disclosed invention improves technology, the claim must be evaluated to ensure the claim itself reflects the disclosed improvement in technology. Intellectual Ventures I LLC v. Symantec Corp., 838 F.3d 1307, 1316, 120 USPQ2d 1353, 1359 (Fed. Cir. 2016) (patent owner argued that the claimed email filtering system improved technology by shrinking the protection gap and mooting the volume problem, but the court disagreed because the claims themselves did not have any limitations that addressed these issues). That is, the claim must include the components or steps of the invention that provide the improvement described in the specification.
Though applicant insists that the claims would provide technological improvements to data augmentation and content recommendation, the additional elements merely include bare details about inputting and outputting of data in an ordinary way without any of the components or steps necessary to solve a particular problem in these fields.
Examiner further points to Recentive Analytics v. Fox Corp. The claims at issue in Recentive were directed to using and training a machine learning model for live event scheduling. It was decided that (page 13) “That is, the claims do not delineate steps through which the machine learning technology achieves an improvement” and (page 18) “Today, we hold only that patents that do no more than claim the application of generic machine learning to new data environments, without disclosing improvements to the machine learning models to be applied, are patent ineligible under § 101.” Similarly, the instant application is directed to using and training a machine learning model for the data environment of content suggestions. However, the claims do not delineate how the machine learning model improves technology.
Thus the claims do not integrate the judicial exception into a practical application under step 2A prong 2. See claim rejections under 35 U.S.C. 101 below for a complete analysis.
In Remarks page 20, Argument 7
(Examiner summarizes Applicant’s arguments) Applicant argues that the claimed subject matter are unconventional operations that are not well-understood routine and conventional, citing to elements of determining homophilous and heterophilous cluster affinity and generating a predicted interaction based on the cluster affinities. Applicant argues that conventional models do not perform well with missing data and the present application can identify missing links to generate new edges. Applicant further argues that providing the content item to the user via website provides an improvement to data augmentation and content recommendation.
Examiner’s response to Argument 7
Examiner disagrees. Examiner notes that abstract idea limitations are not evaluated in step 2B. Only additional elements are evaluated. Arguments related to determining cluster affinity and generating a predicted interaction are rendered moot because those limitations are directed to mental processes with mere tangential additions added after the fact. Thus, it is not relevant whether abstract idea limitations in isolation are conventional or not.
Moreover, claim limitations which do no more than provide mere instructions to apply a judicial exception (MPEP 2106.05(f)) and limit a judicial exception to a particular technological environment (MPEP 2106.05(h)) cannot amount to significantly more.
Examiner further notes that the courts have recognized certain kinds of limitations as well-understood, routine, and conventional activity. These limitations include (MPEP 2106.05(d) II.) “Receiving or transmitting data over a network, e.g., using the Internet to gather data”. Receiving a dataset and providing content items over the internet, as recited in the claims, include no details about how the data is received nor how the content items are provided in any unconventional way.
Thus the claims do not amount to significantly more under step 2B. See claim rejections under 35 U.S.C. 101 below for a complete analysis.
In Remarks page 22, Argument 8
(Examiner summarizes Applicant’s arguments) Applicant argues that analogous independent claims 8 and 17 and all dependent claims are patent eligible by virtue of claim 1.
Examiner’s response to Argument 8
Examiner disagrees. Claim 1 is not deemed eligible, and the analysis of the dependent claims has not changed.
Regarding Art-Based Rejections
In Remarks page 23, Argument 9
(Examiner summarizes Applicant’s arguments) Applicant points to numerous specification paragraphs defining the cluster affinity metrics among users, and affinity clustering (i.e. figure 6), and further points to the cited sections of Zhu. Applicant concludes that Zhu is silent about determining a homophilous cluster affinity and a heterophilous cluster affinity.
Examiner’s response to Argument 9
Examiner agrees that Zhu does not teach on claim 8 in its entirety and therefore the rejections under 35 U.S.C. 102 are withdrawn. However, a new rejection is issued under 35 U.S.C. 103. Though Examiner acknowledges that Zhu does not teach receiving a graph with the user-item interactions, updated search revealed new art to teach on this portion of the claims. Moreover, Examiner argues that Zhu’s matrix H describes homophily (a homophilous cluster affinity) and heterophily (a heterophilous cluster affinity) within a graph and thus teaches some of the newly added limitations.
Examiner further notes that the claims as currently amended do not recite the details of how clustering would be performed by the machine learning model (e.g. specification Fig. 6 and corresponding explanatory paragraphs). Limitations are not to be imported from the specification into the claims. However, Applicant may wish to amend the claims to include these details which may change the claim interpretation.
In Remarks page 29, Argument 10
(Examiner summarizes Applicant’s arguments) Applicant cites several paragraphs from the specification and several paragraphs from Zhu, concluding that Zhu does not teach the limitation of “generating, using a machine learning model, a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity, wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users.
Examiner’s response to Argument 10
Examiner maintains that Zhu does teach:
Generating, using a machine learning model, a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity
Zhu teaches generating, by machine learning prediction, a matrix (H) which describes the probability that nodes in a graph are linked. These probabilities represent predicted interactions and are based on the mapping for homophilous and heterophilous cluster affinity.
However, Examiner agrees that Zhu does not teach the newly added limitation:
Wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users
The rejections under 35 U.S.C. 102 are withdrawn, however a new rejection under 35 U.S.C. 103 is issued using new art to teach this portion of the limitation.
In Remarks page 32, Argument 11
(Examiner summarizes Applicant’s arguments) Regarding 35 U.S.C. 103, Applicant argues that claim 1 was amended and Zhu does not teach the amended limitations alone nor in combination with Kim. Applicant further argues that analogous independent claims and all dependent claims are allowable for the same reasons.
Examiner’s response to Argument 11
Examiner maintains that Zhu still teaches many limitations of the independent claims (see arguments above), and a new reference is relied upon to teach the remaining limitations. See rejections under 35 U.S.C. 103 for the complete analysis. Thus all of the claims (including analogous independent and dependent claims) are further rejected under 35 U.S.C. 103.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-20 are rejected under 35 U.S.C. 101 for containing an abstract idea without significantly more.
Regarding Claim 1:
Step 1 – Is the claim to a process, machine, manufacture, or composition of matter?
Yes, the claim is to a process.
Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites the abstract ideas of:
determining a homophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating the similarity between clustered data points.
determining a heterophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating the difference between clustered data points.
generating, […], a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity, wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating known values to form an opinion on how a user would interact with particular content.
Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No, the claim does not recite additional elements that integrate the judicial exception into a practical application. The additional elements:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
using a machine learning model — This limitation is directed to mere instructions to apply a judicial exception. Using generic machine learning to apply a judicial exception (see MPEP 2106.05(f)) is insufficient to integrate the judicial exception into a practical application. Even if the machine learning is implemented on a generic computer (see MPEP 2106.05(f)(2), 2106.04(d)), the limitation does not integrate the judicial exception into a practical application.
and providing, via the website, the content item to the user based on the predicted interaction. —This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
Step 2B – Does the claim recite additional elements that amount to significantly more than the abstract idea itself?
No, the claim does not recite additional elements which amount to significantly more than the abstract idea itself. The additional elements as identified in step 2A prong 2:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website — This limitation is recited at a high level of generality and amounts to mere data gathering of transmitting and receiving data over a network, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
using a machine learning model — Mere instructions to apply a judicial exception (see MPEP 2106.05(f)) and using a generic computer as a tool (see MPEP 2106.05(f)(2), 2106.05(d)) cannot amount to significantly more than the judicial exception itself.
and providing, via the website, the content item to the user based on the predicted interaction — This limitation is recited at a high level of generality and amounts to mere data gathering of transmitting and receiving data over a network, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
Regarding Claim 2
Claim 2 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim recites the additional limitations:
Step 2A Prong 2:
the dataset includes a plurality of nodes corresponding to the plurality of users and the plurality of content items and a plurality of edges corresponding to a plurality of interactions — This limitation is directed to merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) as it merely limits the field of the type of data in the dataset.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
the dataset includes a plurality of nodes corresponding to the plurality of users and the plurality of content items and a plurality of edges corresponding to a plurality of interactions — Merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 3
Claim 3 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim recites the additional limitations:
Step 2A Prong 2:
further comprising: adding the additional edge to the dataset to obtain the augmented dataset — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
further comprising: adding the additional edge to the dataset to obtain the augmented dataset — This limitation is recited at a high level of generality and amounts to mere data gathering of storing and retrieving information in memory, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 4
Claim 4 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim merely recites the additional abstract ideas
Step 2A Prong 1:
computing a first nonnegative matrix representing the homophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix.
computing a second nonnegative matrix representing the heterophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix.
further comprising: computing a first product of the first nonnegative matrix and a transpose of the first nonnegative matrix; computing a second product of the second nonnegative matrix and a transpose of the second nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating matrix products and transposes.
computing a difference between the first product and the second product to obtain a symmetric difference matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of matrix subtraction.
and applying a nonnegative nonlinear function to the symmetric difference matrix, wherein the predicted interaction is based on the nonnegative nonlinear function — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of applying a function.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 5
Claim 5 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim merely recites the additional abstract ideas:
Step 2A Prong 1:
further comprising: identifying a number of clusters, wherein a sum of a dimension of the first nonnegative matrix and a dimension of the second nonnegative matrix is equal to the number of clusters — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of adding numbers.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 6
Claim 6 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim merely recites the additional abstract ideas:
Step 2A Prong 1:
further comprising: computing a first factor matrix and a second factor matrix, wherein the first factor matrix or the second factor matrix includes a negative value, and wherein the first nonnegative matrix and the second nonnegative matrix are computed based on the first factor matrix and the second factor matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of matrix factorization.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 7
Claim 7 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 1 which included an abstract idea (see rejection for claim 1). The claim merely recites the additional abstract idea:
Step 2A Prong 1:
further comprising: computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix, wherein the machine learning model includes the cluster affinity matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix, which can be performed by equation 1 (see equation 1 and paragraph 95).
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 8:
Step 1 – Is the claim to a process, machine, manufacture, or composition of matter?
Yes, the claim is to a process.
Step 2A – Prong 1 – Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Yes, the claim recites the abstract ideas of:
determining a homophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating the similarity between clustered data points.
determining a heterophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating the difference between clustered data points.
generating, […], a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity, wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to evaluating known values to form an opinion on how a user would interact with particular content.
Step 2A – Prong 2 – Does the claim recite additional elements that integrate the judicial exception into a practical application?
No, the claim does not recite additional elements that integrate the judicial exception into a practical application. The additional elements:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
using a machine learning model — This limitation is directed to mere instructions to apply a judicial exception. Using generic machine learning to apply a judicial exception (see MPEP 2106.05(f)) is insufficient to integrate the judicial exception into a practical application. Even if the machine learning is implemented on a generic computer (see MPEP 2106.05(f)(2), 2106.04(d)), the limitation does not integrate the judicial exception into a practical application.
and updating parameters of the machine learning model based on the predicted interaction — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
Step 2B – Does the claim recite additional elements that amount to significantly more than the abstract idea itself?
No, the claim does not recite additional elements which amount to significantly more than the abstract idea itself. The additional elements as identified in step 2A prong 2:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website — This limitation is recited at a high level of generality and amounts to mere data gathering of transmitting and receiving data over a network, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
using a machine learning model — Mere instructions to apply a judicial exception (see MPEP 2106.05(f)) and using a generic computer as a tool (see MPEP 2106.05(f)(2), 2106.05(d)) cannot amount to significantly more than the judicial exception itself.
and updating parameters of the machine learning model based on the predicted interaction — This limitation is recited at a high level of generality and amounts to mere data gathering of storing and retrieving information in memory, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
Regarding Claim 9
Dependent claim 9 is a method claim corresponding to method claim 3, which was directed to an abstract idea, therefore the same rejection and rationale applies.
Regarding Claim 10
Dependent claim 10 is a method claim corresponding to method claim 4, which was directed to an abstract idea, therefore the same rejection and rationale applies.
Regarding Claim 11
Dependent claim 11 is a method claim corresponding to method claim 5, which was directed to an abstract idea, therefore the same rejection and rationale applies.
Regarding Claim 12
Dependent claim 12 is a method claim corresponding to method claim 6, which was directed to an abstract idea, therefore the same rejection and rationale applies.
Regarding Claim 13
Dependent claim 13 is a method claim corresponding to method claim 7, which was directed to an abstract idea, therefore the same rejection and rationale applies.
Regarding Claim 14
Claim 14 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 10 which included an abstract idea (see rejection for claim 10). The claim recites the additional limitations:
Step 2A Prong 1:
selecting a regularization term; applying the regularization term to the first nonnegative matrix to obtain a regularized first nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of matrix regularization.
and applying the regularization term to the second nonnegative matrix to obtain a regularized second nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of matrix regularization.
Step 2A Prong 2:
wherein the parameters of the machine learning model are updated based on the regularized first nonnegative matrix and the regularized second nonnegative matrix — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
wherein the parameters of the machine learning model are updated based on the regularized first nonnegative matrix and the regularized second nonnegative matrix — This limitation is recited at a high level of generality and amounts to mere data gathering of storing and retrieving information in memory, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 15
Claim 15 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 8 which included an abstract idea (see rejection for claim 8). The claim recites the additional limitations:
Step 2A Prong 1:
computing an L2 norm of a plurality of columns of the first nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating an L2 norm.
and ranking the plurality of columns based on the L2 norm — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to performing a judgement of the proper ordering of columns based on given metrics.
Step 2A Prong 2:
wherein the interaction is based on the ranking — This limitation is directed to merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) as it merely limits the field of the type of data of the interaction.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
wherein the interaction is based on the ranking — Merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 16
Claim 16 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 8 which included an abstract idea (see rejection for claim 8). The claim recites the additional limitations:
Step 2A Prong 1:
computing an L2 norm of a plurality of columns of the second nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating an L2 norm.
and ranking the plurality of columns of the second nonnegative matrix based on the L2 norm — This limitation is directed to the abstract idea of a mental process (including an observation, evaluation, judgement, opinion) which can be performed by the human mind, or by a human using pen and paper (see MPEP 2106.04(a)(2) III. C.). The limitation is directed to a mental process because it amounts to performing a judgement of the proper ordering of columns based on given metrics.
Step 2A Prong 2:
wherein the interaction is based on the ranking — This limitation is directed to merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) as it merely limits the field of the type of data of the interaction.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
wherein the interaction is based on the ranking — Merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 17
Independent claim 19 is an apparatus claim corresponding to method claim 1, which was directed to an abstract idea, therefore the same rejection and rationale applies. The only difference is that claim 1 recites the following additional elements treated under step 2A prong 2 and step 2B:
Step 2A Prong 2:
An apparatus comprising: a processor; and a memory including instructions executable by the processor to perform operations comprising: — This limitation is directed to merely applying an abstract idea using a generic computer as a tool (see MPEP 2106.05(f)(2), 2106.04(d)).
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
An apparatus comprising: a processor; and a memory including instructions executable by the processor to perform operations comprising: — Using a generic computer as a tool (see MPEP 2106.05(f)(2), 2106.05(d)) cannot amount to significantly more than the judicial exception itself.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 18
Claim 18 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 17 which included an abstract idea (see rejection for claim 17). The claim recites the additional limitations:
Step 2A Prong 1:
computing a first nonnegative matrix representing the homophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix.
computing a second nonnegative matrix representing the heterophilous cluster affinity among the plurality of users — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix.
and computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix — This limitation is directed to the abstract idea of a mathematical process, and mathematical calculations in particular (MPEP 2106.04(a)(2) I. C.). The claim describes the mathematical operations of calculating a matrix.
Step 2A Prong 2:
wherein the machine learning model includes the cluster affinity matrix — This limitation is directed to merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) as it merely limits the field of the machine learning model.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
wherein the machine learning model includes the cluster affinity matrix — Merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 19
Claim 19 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 17 which included an abstract idea (see rejection for claim 17). The claim recites the additional limitations:
Step 2A Prong 2:
updating parameters of the machine learning model based on the probability of the predicted interaction — This limitation is directed to mere data gathering and outputting which has been recognized by the courts (as per Ultramercial, 772 F.3d at 715, 112 USPQ2d at 1754) as insignificant extra-solution activity (see MPEP 2106.05(g)).
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
updating parameters of the machine learning model based on the probability of the predicted interaction — This limitation is recited at a high level of generality and amounts to mere data gathering of storing and retrieving information in memory, which is well-understood, routine, and conventional activity (see MPEP 2106.05(d) II.), which cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Regarding Claim 20
Claim 20 is rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more. The claim is dependent on claim 17 which included an abstract idea (see rejection for claim 17). The claim recites the additional limitations:
Step 2A Prong 2:
machine learning model comprises a logistic principal components analysis (LPCA) model — This limitation is directed to merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) as it merely limits the field of the training component to a particular type of model.
Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) I.), failing step 2A prong 2.
Step 2B:
The additional elements as identified in step 2A prong 2:
machine learning model comprises a logistic principal components analysis (LPCA) model — Merely limiting a judicial exception to a particular field of use (see MPEP 2106.05(h)) cannot amount to significantly more than the judicial exception.
Thus, the claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B.
Claim Rejections - 35 USC § 103
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, 8, 17, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. “Graph Neural Networks with Heterophily” herein referred to as Zhu in view of NPL reference Salamat et al. “HeteroGraphRec: A heterogeneous graph-based neural networks for social recommendation” herein referred to as Salamat.
Regarding Claim 1
Zhu teaches:
determining a homophilous cluster affinity among the plurality of users;
(page 11169 column 2 last paragraph) “More generally, H can be used to model any discrete attribute; in that case, Hij is the probability that a node with attribute value i connects with a node with value j.”; [*Examiner notes: The homophilous cluster affinity is mapped to the values in H that are of high probability of being linked, i.e. at or above 50%]
determining a heterophilous cluster affinity among the plurality of users;
(page 11169 column 2 last paragraph) “More generally, H can be used to model any discrete attribute; in that case, Hij is the probability that a node with attribute value i connects with a node with value j. […] In the following sections, we propose CPGNN, which is capable of learning H in an end-to-end way based on a rough initial estimation”; [*Examiner notes: The heterophilous cluster affinity is mapped to the values in H that are of low probability of being linked, i.e. below 50%]
generating, using a machine learning model, a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity
(page 11171 column 1 paragraph 1) “Parameter H in CPGNN can be easily understood: it captures the probability that node pairs in specific classes connect with each other[*Examiner notes: predicted interaction].”; [*Examiner notes: The parameter H predicts which nodes are likely to connect (a predicted interaction) based on the values in H (homophilous and heterophilous cluster affinity)]
Zhu does not explicitly teach:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website;
wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users;
and providing, via the website, the content item to the user based on the predicted interaction.
However, Salamat teaches:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website;
(page 6 column 1 bullet point 2) “Douban Movie is a Chinese website that allows Internet users to share their comments and viewpoints about movies. Users can post short or long comments on movies and give them marks.”
wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users;
(page 7 column 1 section 4.3.2 bullet point 2) “user–item: This component of the model quantifies the value of the user’s item interaction history and how much it identifies the user’s preferences.”; Fig. 2
PNG
media_image1.png
553
656
media_image1.png
Greyscale
and providing, via the website, the content item to the user based on the predicted interaction.
(page 1 abstract) “Recommender systems in social networks are widely used for connecting users to their desired items from a vast catalog of available items.”; (page 7 column 2 last paragraph) “These global parameters are effective at analyzing the most popular or the most well-received items in the entire network, which helps provide accurate recommendations.”
Zhu, Salamat, and the instant application are analogous because they are all directed to graphs and machine learning.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous and heterophilous affinity techniques of Zhu with the user-item interactions taught by Salamat because (Salamat page 7) “This new social network model naturally contains item–item connections based on similarity or similar metrics. These connections contain information about the structure of the items in the network. We then proposed a neural network architecture that utilizes the various connections and uses the structured input to improve the recommender system’s performance”.
Regarding Claim 2
Zhu in view of Salamat teaches:
The method of claim 1
(see rejection of claim 1)
And Salamat further teaches:
wherein: the dataset includes a plurality of nodes corresponding to the plurality of users and the plurality of content items and a plurality of edges corresponding to a plurality of interactions.
(page 3 column 2 last paragraph before bullet points) “The heterogeneous graph consists of two types of entities — users and items, and three different kinds of edges, which are the user–user, item–item, and user–item connections, shown as Fig. 2.”; Fig. 2
PNG
media_image2.png
667
672
media_image2.png
Greyscale
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to combine Zhu and Salamat for the same reasons given in claim 1 above.
Regarding Claim 8
Zhu teaches:
determining a homophilous cluster affinity among the plurality of users;
(page 11169 column 2 last paragraph) “More generally, H can be used to model any discrete attribute; in that case, Hij is the probability that a node with attribute value i connects with a node with value j.”; [*Examiner notes: The homophilous cluster affinity is mapped to the values in H that are of high probability of being linked, i.e. at or above 50%]
determining a heterophilous cluster affinity among the plurality of users;
(page 11169 column 2 last paragraph) “More generally, H can be used to model any discrete attribute; in that case, Hij is the probability that a node with attribute value i connects with a node with value j. […] In the following sections, we propose CPGNN, which is capable of learning H in an end-to-end way based on a rough initial estimation”; [*Examiner notes: The heterophilous cluster affinity is mapped to the values in H that are of low probability of being linked, i.e. below 50%]
generating, using a machine learning model, a predicted interaction based on the homophilous cluster affinity and the heterophilous cluster affinity
(page 11171 column 1 paragraph 1) “Parameter H in CPGNN can be easily understood: it captures the probability that node pairs in specific classes connect with each other[*Examiner notes: predicted interaction].”; [*Examiner notes: The parameter H predicts which nodes are likely to connect (a predicted interaction) based on the values in H (homophilous and heterophilous cluster affinity)]
and updating parameters of the machine learning model based on the predicted interaction
(page 11170 column 1 below equation 5) “We formulate each layer as: [Equation 6] Each layer propagates and updates the current belief per node[*Examiner notes: updating parameters of ML model] in its neighborhood.”; Equation 6
PNG
media_image3.png
130
676
media_image3.png
Greyscale
Zhu does not explicitly teach:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website;
wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users;
and providing, via the website, the content item to the user based on the predicted interaction.
However, Salamat teaches:
A method comprising: receiving a dataset that includes a plurality of interactions between a plurality of users and a plurality of content items on a website;
(page 6 column 1 bullet point 2) “Douban Movie is a Chinese website that allows Internet users to share their comments and viewpoints about movies. Users can post short or long comments on movies and give them marks.”
wherein the predicted interaction indicates an interaction between a content item of the plurality of content items and a user of the plurality of users;
(page 7 column 1 section 4.3.2 bullet point 2) “user–item: This component of the model quantifies the value of the user’s item interaction history and how much it identifies the user’s preferences.”; Fig. 2
PNG
media_image1.png
553
656
media_image1.png
Greyscale
Zhu, Salamat, and the instant application are analogous because they are all directed to graphs and machine learning.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous and heterophilous affinity techniques of Zhu with the user-item interactions taught by Salamat because (Salamat page 7) “This new social network model naturally contains item–item connections based on similarity or similar metrics. These connections contain information about the structure of the items in the network. We then proposed a neural network architecture that utilizes the various connections and uses the structured input to improve the recommender system’s performance”.
Regarding Claim 17
Claim 17 is a computer apparatus claim corresponding to method claim 1. The only difference is that claim 17 recites an apparatus with a processor and memory.
Zhu teaches:
An apparatus comprising: a processor; a memory including instructions executable by the processor
(page 11175 column 1 acknowledgements) “We gratefully acknowledge the support of NVIDIA Corporation with the donation of the Quadro P6000 GPU used for this research.”
The remaining limitations of the claim are taught by the rejection of claim 1.
Regarding Claim 19
Zhu in view of Salamat teaches:
The apparatus of claim 17
(see rejection of claim 17)
And Zhu further teaches:
further comprising: updating parameters of the machine learning model based on the predicted interaction.
(page 11170 column 1 below equation 5) “We formulate each layer as: [Equation 6] Each layer propagates and updates the current belief per node[*Examiner notes: updating parameters of ML model] in its neighborhood.”; Equation 6
PNG
media_image3.png
130
676
media_image3.png
Greyscale
Claims 3 and 9 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat, and further in view of NPL reference Kim et al. “The Network Completion Problem: Inferring Missing Nodes and Edges in Networks” herein referred to as Kim.
Regarding Claim 3
Zhu in view of Salamat teaches:
The method of claim 1
(see rejection of claim 1)
Zhu in view of Salamat does not explicitly teach:
further comprising: adding an additional edge to the dataset to obtain an augmented dataset
However, Kim teaches:
further comprising: adding an additional edge to the dataset to obtain an augmented dataset
(page 47 column 2 second to last paragraph) “There is a complete network represented by adjacency matrix H and some nodes and corresponding edges are missing from it. We only observe the network (matrix) G, the non-missing part of H, and aim to infer the missing nodes and edges, i.e., the missing part Z.”; (page 47 column 1 abstract) “We address this issue by studying the Network Completion Problem: Given a network with missing nodes and edges, can we complete the missing part?”
Zhu, Salamat, Kim, and the instant application are analogous because they are all directed to machine learning and graphs
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat with the graph completion taught by Kim because (Kim page 47 abstract) “While the social and information networks have become ubiquitous, the challenge of collecting complete network data still persists. Many times the collected network data is incomplete with nodes and edges missing. Commonly, only a part of the network can be observed and we would like to infer the unobserved part of the network.”
Regarding Claim 9
Claim 9 is a method claim corresponding to method claim 3. All of the additional limitations of claim 9 are taught in the rejection of claim 3.
Claim 4-6 and 10-12 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat, and further in view of NPL reference Chanpuriya et al. “An Interpretable Graph Generative Model with Hetrophily”, herein referred to as Chanpuriya.
Regarding Claim 4
Zhu in view of Salamat teaches:
The method of claim 1
(see rejection of claim 1)
Zhu further teaches:
further comprising: computing a first nonnegative matrix representing the homophilous cluster affinity among the plurality of users;
[*Examiner notes: the broadest reasonable interpretation of this matrix includes any nonnegative matrix that describes the graph, which relates to the homophilous cluster affinity]; (page 11169 column 2 definition 2) “Let Y[*Examiner notes: first matrix] ∈ R|V|x|Y| where Yvj = 1 if yv = j, and Yvj = 0[*Examiner notes: nonnegative matrix] otherwise.”
computing a second nonnegative matrix representing the heterophilous cluster affinity among the plurality of users;
[*Examiner notes: the broadest reasonable interpretation of this matrix includes any nonnegative matrix that describes the graph, which relates to the heterophilous cluster affinity]; (page 11169 column 1 second to last paragraph) “For subsequent discussions, we use A ∈ {0, 1}|V|x|V| for the adjacency matrix[*Examiner notes: second nonnegative matrix] with self-loops removed”
Zhu in view of Salamat does not explicitly teach:
computing a first product of the first nonnegative matrix and a transpose of the first nonnegative matrix; computing a second product of the second nonnegative matrix and a transpose of the second nonnegative matrix; computing a difference between the first product and the second product to obtain a symmetric difference matrix; and applying a nonnegative nonlinear function to the symmetric difference matrix, wherein the predicted interaction is based on the nonnegative nonlinear function.
However, Chanpuriya teaches:
computing a first product of the first nonnegative matrix and a transpose of the first nonnegative matrix; computing a second product of the second nonnegative matrix and a transpose of the second nonnegative matrix; computing a difference between the first product and the second product to obtain a symmetric difference matrix; and applying a nonnegative nonlinear function to the symmetric difference matrix, wherein the predicted interaction is based on the nonnegative nonlinear function.
(page 6 section “third stage” paragraph 2) “These remaining k communities are then directly optimized by minimizing the cross-entropy loss of Equation 3 on the following graph model: [Equation 5]”; (page 3 section 2 below equation 1) “where σ is the logistic function[*Examiner notes: nonnegative nonlinear function].”; [*Examiner notes: a logistic function is a type of non-linear function]; Equation 5
PNG
media_image4.png
37
545
media_image4.png
Greyscale
Zhu, Salamat, Chanpuriya, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to combine the homophilous/heterophilous matrix techniques taught by Zhu in view of Kim with the symmetric difference matrix calculations taught by Chanpuriya because (Chanpuriya page 1 abstract) “Our theoretical results demonstrate the expressiveness of our model in its ability to exactly reconstruct a graph using a number of clusters that is linear in the maximum degree, along with its ability to capture both heterophily and homophily in the data. Further, our experiments demonstrate the effectiveness of our model for a variety of important application tasks such as multi-label clustering and link prediction”
Regarding Claim 5
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 4
(see rejection of claim 4)
Chanpuriya further teaches:
further comprising: identifying a number of clusters, wherein a sum of a dimension of the first nonnegative matrix and a dimension of the second nonnegative matrix is equal to the number of clusters
(page 9 last paragraph) “Further, our theoretical results demonstrate the expressiveness of our model in its ability to exactly reconstruct a graph using a number of clusters that is linear in the maximum degree, along with its ability to capture both heterophily and homophily in the data.”
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to combine Zhu and Salamat with Chanpuriya for the same reasons given in claim 4 above.
Regarding Claim 6
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 1
(see rejection of claim 1)
Chanpuriya further teaches:
further comprising: computing a first factor matrix and a second factor matrix, wherein the first factor matrix or the second factor matrix includes a negative value, and wherein the first nonnegative matrix and the second nonnegative matrix are computed based on the first factor matrix and the second factor matrix
(page 4 paragraph 2) “More recently, Peysakhovich & Bottou (2021) propose a decomposition of the form A ≈ D + BB> − CC>, where D ∈ Rn×n is diagonal and B, C ∈ Rn×k are low-rank; the authors discuss how, interestingly, this model separates the homophilous and heterophilous structure into different factors, namely B and C. However, this work does not pursue a clustering interpretation or investigate setting the factors B and C to be nonnegative.”
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to combine Zhu and Salamat with Chanpuriya for the same reasons given in claim 4 above.
Regarding Claim 10
Claim 10 is a method claim corresponding to method claim 4. All of the additional limitations of claim 10 are taught in the rejection of claim 4.
Regarding Claim 11
Claim 11 is a method claim corresponding to method claim 5. All of the additional limitations of claim 11 are taught in the rejection of claim 5.
Regarding Claim 12
Claim 12 is a method claim corresponding to method claim 6. All of the additional limitations of claim 9 are taught in the rejection of claim 6.
Claims 7 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat and Chanpuriya, and further in view of NPL reference Qin et al. “Affinity Matrix Learning Via Nonnegative Matrix Factorization for Hyperspectral Imagery Clustering” herein referred to as Qin.
Regarding Claim 7
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 1
(see rejection of claim 1)
Zhu in view of Salamat and Chanpuriya does not explicitly teach:
further comprising: computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix, wherein the machine learning model includes the cluster affinity matrix
However, Qin teaches:
further comprising: computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix, wherein the machine learning model includes the cluster affinity matrix
(page 403 column 2 paragraph 3) “To simultaneously reduce the computational cost and achieve accurate affinity for HSI clustering, we propose a framework to integrate the spectral-spatial information into the nonnegative matrix factorization (NMF) for affinity matrix learning (NMFAML).”
Zhu, Salamat, Qin, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat and Chanpuriya with the cluster affinity matrix of Qin because (Qin page 402) “Experimental results on three public benchmark HSIs demonstrate that the proposed method is superior to the considered state-of-the-art baseline methods on both the computational cost and clustering accuracy.”
Regarding Claim 13
Claim 13 is a method claim corresponding to method claim 7. All of the additional limitations of claim 13 are taught in the rejection of claim 7.
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat and Chanpuriya, and further in view of NPL reference Yan et al. “Generalizing Graph Networks Beyond Homophily” herein referred to as Yan.
Regarding Claim 14
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 10
(see rejection of claim 10)
Zhu in view of Salamat and Chanpuriya does not explicitly teach:
further comprising: selecting a regularization term; applying the regularization term to the first nonnegative matrix to obtain a regularized first nonnegative matrix; and applying the regularization term to the second nonnegative matrix to obtain a regularized second nonnegative matrix, wherein the parameters of the machine learning model are updated based on the regularized first nonnegative matrix and the regularized second nonnegative matrix.
However, Yan teaches:
further comprising: selecting a regularization term
(page 24 section “Cora Full Benchmark Tuning) “Therefore, we need to re-tune the hyperparameters, especially the regularization weights and learning rates, in order to get reasonable performance.”
applying the regularization term to the first nonnegative matrix to obtain a regularized first nonnegative matrix; and applying the regularization term to the second nonnegative matrix to obtain a regularized second nonnegative matrix, wherein the parameters of the machine learning model are updated based on the regularized first nonnegative matrix and the regularized second nonnegative matrix.
(page 21 section F.1) “For loss function, we calculate the cross entropy between the predicted and the ground-truth labels for nodes within the training set, and add L2 regularization of network parameters We and Wc.”
Zhu, Salamat, Chanpuriya, Yan, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat and Chanpuriya with the regularization of Yan because (Yan page 1 abstract) “Going beyond the traditional benchmarks with strong homophily, our empirical analysis on synthetic and real networks shows that, thanks to the identified designs, H2GCN has consistently strong performance across the full spectrum of low-to-high homophily, unlike competitive prior models without them.”
Claims 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu, Salmat, and Chanpuriya and further in view of Wang et al. (US20220381838A1) herein referred to as Wang.
Regarding Claim 15
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 10
(see rejection of claim 10)
Zhu in view of Salamat and Chanpuriya does not explicitly teach:
further comprising: computing an L2 norm of a plurality of columns of the first nonnegative matrix; and ranking the plurality of columns based on the L2 norm, wherein the predicted interaction is computed based on the ranking
However, Wang teaches:
further comprising: computing an L2 norm of a plurality of columns of the first nonnegative matrix; and ranking the plurality of columns based on the L2 norm, wherein the predicted interaction is computed based on the ranking
(paragraph [0042]) “where G > is the first q−n columns of a matrix {tilde over (G)} obtained by arranging the column vector of G in a descending order according to the Euclidean norm[*Examiner notes: L2 norm]”; [*Examiner notes: The predicted interaction is computed based on the ranking because the predicted interaction is based on the first nonnegative matrix]
Zhu, Salamat, Chanpuriya, Wang, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat and Chanpuriya with the L2 norm ranking of Wang because (Wang paragraph [0116]) “It shows that the fault diagnosis method provided in the present disclosure has the features of high fault detection efficiency and accurate fault diagnosis.”
Regarding Claim 16
Zhu in view of Salamat and Chanpuriya teaches:
The method of claim 10
(see rejection of claim 10)
Zhu in view of Salamat and Chanpuriya does not explicitly teach:
further comprising: computing an L2 norm of a plurality of columns of the second nonnegative matrix; and ranking the plurality of columns of the second nonnegative matrix based on the L2 norm, wherein the predicted interaction is based on the ranking
However, Wang teaches:
further comprising: computing an L2 norm of a plurality of columns of the second nonnegative matrix; and ranking the plurality of columns of the second nonnegative matrix based on the L2 norm, wherein the predicted interaction is based on the ranking
(paragraph [0042]) “where G > is the first q−n columns of a matrix {tilde over (G)} obtained by arranging the column vector of G in a descending order according to the Euclidean norm[*Examiner notes: L2 norm]”; [*Examiner notes: The predicted probability of the edge is computed based on the ranking because the predicted probability of the edge is based on the second nonnegative matrix]
Zhu, Salamat, Chanpuriya, Wang, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat, and Chanpuriya with the L2 norm ranking of Wang because (Wang paragraph [0116]) “It shows that the fault diagnosis method provided in the present disclosure has the features of high fault detection efficiency and accurate fault diagnosis.”
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat and Qin.
Regarding Claim 18
Zhu in view of Salamat teaches:
The apparatus of claim 17
(see rejection of claim 17)
Zhu further teaches:
further comprising: computing a first nonnegative matrix representing the homophilous cluster affinity among the plurality of users;
[*Examiner notes: the broadest reasonable interpretation of this matrix includes any nonnegative matrix that describes the graph, which relates to the homophilous cluster affinity]; (page 11169 column 2 definition 2) “Let Y[*Examiner notes: first matrix] ∈ R|V|x|Y| where Yvj = 1 if yv = j, and Yvj = 0[*Examiner notes: nonnegative matrix] otherwise.”
computing a second nonnegative matrix representing the heterophilous cluster affinity among the plurality of users;
[*Examiner notes: the broadest reasonable interpretation of this matrix includes any nonnegative matrix that describes the graph, which relates to the heterophilous cluster affinity]; (page 11169 column 1 second to last paragraph) “For subsequent discussions, we use A ∈ {0, 1}|V|x|V| for the adjacency matrix[*Examiner notes: second nonnegative matrix] with self-loops removed”
Zhu in view of Salamat does not explicitly teach:
and computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix, wherein the machine learning model includes the cluster affinity matrix
However, Qin teaches:
and computing a cluster affinity matrix based on the first nonnegative matrix and the second nonnegative matrix, wherein the machine learning model includes the cluster affinity matrix
(page 403 column 2 paragraph 3) “To simultaneously reduce the computational cost and achieve accurate affinity for HSI clustering, we propose a framework to integrate the spectral-spatial information into the nonnegative matrix factorization (NMF) for affinity matrix learning (NMFAML).”
Zhu, Salamat, Qin, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Salamat with the cluster affinity matrix of Qin because (Qin page 402) “Experimental results on three public benchmark HSIs demonstrate that the proposed method is superior to the considered state-of-the-art baseline methods on both the computational cost and clustering accuracy.”
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu in view of Salamat, and further in view of NPL reference Lee et al. “Sparse Logistic Principal Components Analysis for Binary Data” herein referred to as Lee.
Regarding Claim 20
Zhu in view of Salamat teaches:
The apparatus of claim 19
(see rejection of claim 19)
Zhu in view of Salamat does not explicitly teach:
wherein: the training component comprises a logistic principal components analysis (LPCA) model
However, Lee teaches:
wherein: the training component comprises a logistic principal components analysis (LPCA) model
(page 1 abstract) “We develop a new principal components analysis (PCA) type dimension reduction method for binary data. Different from the standard PCA which is defined on the observed data, the proposed PCA is defined on the logit transform of the success probabilities of the binary observations.”
Zhu, Kim, Lee, and the instant application are analogous because they are all directed to machine learning and graphs.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the present invention to modify the homophilous/heterophilous matrix techniques taught by Zhu in view of Kim with the LPCA of Lee because (page 1 abstract) “The effectiveness of the proposed sparse logistic PCA method is illustrated by application to a single nucleotide polymorphism data set and a simulation study” and (Lee page 14 section 7 paragraph 1) “The MM algorithm developed for implementation of our method provides a unified solution for dealing with i) the non-quadratic likelihood, ii) the non-differentiable penalty function; iii) presence of missing data”
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Ezra J Baker whose telephone number is (703)756-1087. The examiner can normally be reached Monday - Friday 10:00 am - 8:00 pm ET.
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, David Yi can be reached at (571) 270-7519. 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.
/E.J.B./Examiner, Art Unit 2126
/VAN C MANG/Primary Examiner, Art Unit 2126