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 .
Information Disclosure Statement
An Information Disclosure Statement (IDS) has not been submitted as of the mailing of the last Office Action dated 13 May 2025. Applicant is reminded of the continuing obligation under 37 CFR 1.56 to timely apprise the Office of any information which is material to patentability of the claims under consideration in this application.
Claim Objections
Claim 5 is objected to because of the following informalities: the claim status shows “Original” despite having been amended in the present filing (i.e., “look-a-like” was amended to “lookalike”). Appropriate correction is required (i.e., the future claim status of this claim should be “Previously Presented” instead of “Original”).
Introductory Remarks
In response to communications filed on 12 August 2025, claims 1, 51, 8, 12, 15, and 18 are amended per Applicant's request. Claims 7, 14, and 20 are cancelled. No claims were withdrawn. Claims 24-26 are new. Therefore, claims 1-6, 8-13, 15-19, and 21-26 are presently pending in the application, of which claims 1, 8, and 15 are presented in independent form.
The previously raised 112 rejections of the pending claims is withdrawn in view of the amendments to the claims. A new ground(s) of rejection has been issued.
The previously raised 103 rejection of the pending claims is withdrawn in view of the amendments to the claims. A new ground(s) of rejection has been issued.
Response to Arguments
Applicant’s arguments filed 12 August 2025 with respect to the rejection of the claims under 35 U.S.C. 112 (see Remarks, p. 17-24) have been fully considered but are not persuasive. The amendments raise new issues, which have been addressed in the 112 rejections below.
Applicant’s arguments filed 12 August 2025 with respect to the rejection of the claims under 35 U.S.C. 103 (see Remarks, p. 16-17) have been fully considered but are moot, as Applicant’s arguments solely argue that the prior art did not disclose the amended claim features (of which some of the amended claim features were previously disclosed by the prior art; however, the rejection has been modified below to conform to the amended language), and/or do not apply to the new reference (and thus new combination of references) being used in the current rejection.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1-6, 8-13, 15-19, and 21-26 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claims 1, 8, and 15 recite “generating…a weighted interaction graph” and “inferring…a set of new entity-to-class relationships”, the latter which appears to recite substantially overlapping language as the “generating…a weighted interaction graph” steps. There is a lack of written description that separates these two steps in this manner. See, e.g., the 112(b), indefiniteness rejection below for further detail.
The dependent claims are rejected for at least by virtue of their dependency on their respective independent claims, and for failing to cure the deficiencies of their respective independent claims.
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1-6, 8-13, 15-19, and 21-26 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 1, 8, and 15 recite multiple limitations that appear to be overlapping, yet it is unclear whether the steps are being presented in the order as claimed, whether steps expand on other limitations, and/or the precise relationships between the various steps. These steps are:
“determining that the first entity has the particular likelihood that can be inferred to the second entity” and “inferring…based on the first set of data, the second set of data, the determined similarity between the one or more first interaction classes and the one or more second interaction classes…a set of new entity-to-class relationships comprising…”. It appears that the steps of determining new entity-to-class relationships would encompass the first “determining that the first entity has the particular likelihood that can be inferred to the second entity”, yet this limitation is claimed prior to the “inferring…a set of new entity-to-class relationships” step. Therefore, it is confusing as to whether the latter step is an expansion of the first, or a separate step of the first, i.e., whether there is a relationship between the two, as they appear to overlap, yet are claimed as separate elements.
Furthermore, the “generating…a weighted interaction graph” and “inferring…a set of new entity-to-class relationships” steps appear to be claimed separately. However, the later step of “generating…using the weighted interaction graph, an extended set of data for the particular interaction class…” only refers to the generated weighted interaction graph. The inferring of new entity-to-class relationships step is completely abandoned. Therefore, it is unclear what the role of the “inferring…a set of new entity-to-class relationships” step is, and whether it has any relationship to the generated weighted interaction graph.
Additionally, in a related vein, the “generating…a weighted interaction graph” step includes determining entity-to-class weights based on first entity-to-class weights (for the first entity) and second entity-to-class weights (for the second entity). The step of “inferring…a set of new entity-to-class relationships” also states a new entity-to-class relationship “between the second entity…and the particular interaction class based at least in part on comparing entity-to-class weights…from the first set of data to the second entity-to-class weights of the second entity”. Again, language appears to be substantially similar, yet the relationship between the “generating…a weighted interaction graph” and “inferring…a set of new entity-to-class relationships” is not clear.
Furthermore, the claims recite “inferring…based on the first set of data, the second set of data, the determined similarity between the one or more first interaction classes and the one or more second interaction classes, and the determined similarity between the first entity and the second entity, a set of new entity-to-class relationships”. There is a lack of antecedent basis for such a limitation, as there was no prior limitation describing determining a similarity between the first and second entities.
Additionally, the claims recite “determining…a similarity of entity-to-class relationships between the first entity…and the second entity”. Yet this limitation is not utilized at all later in the claimed steps.
The dependent claims are rejected for at least by virtue of their dependency on their respective independent claims, and for failing to cure the deficiencies of their respective independent claims.
Claims 3-4, 6, 10-11, 13, 16-17, and 19 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 3, 6, 10, 13, 16, and 19 recite “wherein determining the set of new relationships comprises…”. Similarly, Claims 4, 11, and 17 recite “wherein each relationship included in the plurality of relationships comprises a likelihood of the entity interacting with a corresponding class of the plurality of classes”. There is a lack of antecedent basis for such a limitation, as the independent claims had been amended to recite “determining a set of new entity-to-class relationships”.
Furthermore, Claims 3, 10, and 16 recite “receiving…historical interaction data; and determining…using the historical interaction data, a plurality of relationships between the entity and a plurality of classes”. However, it appears that their respective independent claims already utilize “historical interactions” (implicitly “historical interaction data”) to determine these relationships between the entity and a plurality of (interaction) classes, if “interaction” classes were the appropriate term.
Additionally, Claims 3, 10 and 16 recite using “a plurality of classes”, and the independent claims recite “interaction classes”. These appear to be referring to the same item; however, there is a lack of antecedent basis issue with these limitations.
Claims 5, 12, and 18 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
The claims recite “wherein the first set of data includes propensity-scored entities, wherein the second set of data includes lookalike-scored entities”. These limitations are already claimed in their respective independent claims.
The following is a quotation of 35 U.S.C. 112(d):
(d) REFERENCE IN DEPENDENT FORMS.—Subject to subsection (e), a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
The following is a quotation of pre-AIA 35 U.S.C. 112, fourth paragraph:
Subject to the following paragraph [i.e., the fifth paragraph of pre-AIA 35 U.S.C. 112], a claim in dependent form shall contain a reference to a claim previously set forth and then specify a further limitation of the subject matter claimed. A claim in dependent form shall be construed to incorporate by reference all the limitations of the claim to which it refers.
Claims 4, 11, and 17 are rejected under 35 U.S.C. 112(d) or pre-AIA 35 U.S.C. 112, 4th paragraph, as being of improper dependent form for failing to further limit the subject matter of the claim upon which it depends, or for failing to include all the limitations of the claim upon which it depends. The claims recite “wherein each [entity-to-class]2 relationship included in the plurality of relationships comprises a likelihood of the entity interacting with a corresponding class of the plurality of classes”. However, independent claims 1, 8, and 15 already recite “generating…a weighted interaction graph that indicates…a likelihood of each entity interacting with a corresponding interaction class…” Therefore, claims 4, 11, and 17 fail to further limit the subject matter of the independent claims upon which they depend upon. Applicant may cancel the claim(s), amend the claim(s) to place the claim(s) in proper dependent form, rewrite the claim(s) in independent form, or present a sufficient showing that the dependent claim(s) complies with the statutory requirements.
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 1-6, 8-13, 15-17, 19, and 24-26 are rejected under 35 U.S.C. 103 as being unpatentable over Liao et al. (“Liao”) (US 10,687,105 B1, incorporating by reference Juan et al. (“IBR-Juan”) (US 2012/0166532 A1), Li et al. (“IBR-Li”) (US 2013/0124298 A1), Hong et al. (“IBR-Hong”) (US 2015/0332336 A1), and Kirti et al. (“IBR-Kirti”) (US 2017/0262894 A1)), in view of Zheng et al. (“Zheng”) (US 2016/0342705 A1), in further view of Rao (“Rao”) (US 11,250,488 B1).
Regarding claim 1: Liao teaches A computer-implemented method comprising:
receiving, at a computing device and from a first source of data, a request to initiate a communication campaign for a particular interaction class, wherein the communication campaign is based on a first set of data that includes data about a first set of entities, wherein at least part of the first set of data includes historical interactions by the first set of entities, wherein the first set of entities comprises a first entity (IBR-Hong, [0016-0020], where the social networking system 100 receives a request 110 to present an advertisement to the user 105 from client device 107, e.g., the interface identifying one or more opportunities to present advertisements to the user 105, which causes the client device 107 to communicate the request 110 to present the advertisement to the social networking system 100. Based on user characteristics associated with the user 105 (i.e., “first set of data that includes data about a first set of entities, wherein the first set of entities comprises a first entity”), one or more targeting groups and cluster groups are associated with the user. See IBR-Hong, [0023-0024], where ad requests are retrieved and presented based on targeting criteria associated with cluster groups associated with the user (i.e., “communication campaign associated with a particular interaction class”). See also IBR-Juan, [0036], where the disclosed system may perform user interest predictions based on historical user activity and past responses to choices of actions (i.e., “wherein at least part of the first set of data is determined from historical interactions by the first set of entities”));
determining a performance of the first entity for the particular interaction class, wherein the performance indicates a particular likelihood of interaction with the particular interaction class based on one or more historical interactions by the first entity with the particular interaction class (IBR-Juan, [0018], where to predict the likely actions a user may take in a given situation, a measure of affinity for a user may be requested. The measure of affinity may reflect the user’s interest in content, actions, advertisements, or any other objects in the social networking system (i.e., “particular interaction class”). See also IBR-Juan, [0028], where the system computes a set of predictor functions 110, which predict whether a user will perform a set of corresponding actions. Each predictor function 110 may be representative of a user’s interest in a particular action (i.e., “particular interaction class”) associated with the predictor function 110. Historical activity of a user may be used as a signal of a user’s future interest in the same activity. The predictor function may be generated based on a user’s historical activity associated with an action (i.e., “historical interactions with the particular interaction class”), e.g., the predictor function 110 may take as an input the user’s historical activity and output a measure of the likelihood that the user will engage in the corresponding activity);
receiving, by the computing device and from a second source of data that is different than the first source of data, a second set of data that includes data about a second set of entities that is at least partially different than the first set of entities, wherein at least part of the second set of data includes historical interactions by the second set of entities, wherein the second set of entities comprises a second entity that is not related to the particular interaction class (IBR-Li, [0029], where targeting criteria for an advertisement may include a generated set of users that are similar to users that have previously engaged with the advertisement or the advertiser for the advertisement (implying that the “generated set of users” had not previously engaged with the advertisement or advertiser for the advertisement (i.e., “not related to the particular interaction class”), and thus are a second set of entities “that is at least partially different than the first set of entities”, i.e., the first set having previously engaged with the advertisement/advertiser, the second set not having previously engaged with the advertisement/advertiser). See also IBR-Li, [0017], where user profile objects 102 are identified by interactions with systems internal (i.e., “first source of data”) and external to the social networking system 100 (i.e., “a second source of data that is different than the first source of data”). See also IBR-Juan, [0036], where the disclosed system may perform user interest predictions based on historical user activity and past responses to choices of actions (i.e., “wherein at least part of the second set of data is determined from historical interactions by the second set of entities”));
… and
outputting the extended set of data to facilitate communication with a third set of entities that comprises a first plurality of entities from the first set of entities and a second plurality of entities from the second set of entities that are not included in the first set of entities, wherein the second plurality of entities includes the second entity (IBR-Li, [0019], where ad targeting module 114 receives the targeted user profile objects 112 as a targeting cluster 124 of users that the advertisement 118 may be provided to for display (i.e., “outputting the extended set of data to facilitate communication with a third set of entities”). As a result, an advertiser 116 may target an advertisement 118 to a targeting cluster 124 of users of the social networking system 100 that are similar to a training cluster 122 of users as defined by targeting criteria 120 included in the advertisement 118, such as users that have engaged with similar advertisements, users that have performed an interaction with a specified object on the social networking system, and users that have performed a particular action on a system external to the social networking system 100).
Liao does not appear to explicitly teach determining, by the computing device and based on the first set of data and the second set of data, a similarity between one or more first interaction classes from the first set of data and one or more second interaction classes related to one or more entities of the second set of entities in the second set of data; wherein the one or more first interaction classes are separate from and not identical to the one or more second interaction classes; wherein at least one of the one or more first interaction classes from the first set of data is not related to the first entity in the first set of data and wherein the second entity is not represented in the first set of data; determining, by the computing device and based on the first set of data and the second set of data, a similarity of entity-to-class relationships between the first entity of the first set of data and the second entity of the second set of data based at least in part on interaction classes related to the first entity and interaction classes related to the second entity; generating, by the computing device, a weighted interaction graph that indicates, for each entity of the first set of entities and the second set of entities, a likelihood of each entity interacting with a corresponding interaction class, wherein the likelihood of each entity interacting with a corresponding interaction class is used to determine first entity-to-class weights for the first entity and second entity-to-class weights for the second entity; wherein generating the weighted interaction graph comprises: determining that the first entity has the particular likelihood that can be inferred to the second entity, wherein the second set of data did not previously indicate any historical interactions with the particular interaction class by the second entity and did not indicate an association between the first entity and the second entity; and generating a weight between the second entity and the particular interaction class to indicate a likelihood of interaction for the second entity with the particular interaction class based at least in part on the particular likelihood of interaction for the first entity with the particular interaction class; inferring, by the computing device and based on the first set of data, the second set of data, the determined similarity between the one or more first interaction classes and the one or more second interaction classes, and the determined similarity between the first entity and the second entity, a set of new entity-to-class relationships comprising: a first new entity-to-class relationship between the first entity and the at least one of the one or more first interaction classes based at least in part on comparing the first entity-to-class weights from the first entity to entity-to-class weights of a lookalike-scored entity from the second set of data; wherein the at least one of the one or more first interaction classes were not previously related to the first entity in the first set of data; and a second new entity-to-class relationship between the second entity of the second set of entities and the particular interaction class based at least in part on comparing entity-to-class weights from a propensity-scored entity from the first set of data to the second entity-to-class weights of the second entity; wherein the second entity was not previously related to the particular interaction class; [and] generating, by the computing device and using the weighted interaction graph, an extended set of data for the particular interaction class, wherein the extended set of data comprises the second entity.
Zheng teaches determining, by the computing device and based on the first set of data and the second set of data, a similarity between one or more first interaction classes from the first set of data and one or more second interaction classes related to one or more entities of the second set of entities in the second set of data; wherein the one or more first interaction classes are separate from and not identical to the one or more second interaction classes; wherein at least one of the one or more first interaction classes from the first set of data is not related to the first entity in the first set of data and wherein the second entity is not represented in the first set of data (Zheng, [0063], where interests derived from user correspondences may be used to infer the known interests of the seed users to be propagated, e.g., two friends talk about NBA over emails, but never had the chance to talk about football, even though one is a football fan. A model may propagate a known interest on football from one friend (as a seed user) over to a derived interest on the NBA);
generating, by the computing device, a weighted interaction graph that indicates, for each entity of the first set of entities and the second set of entities, a likelihood of each entity interacting with a corresponding interaction class …; wherein generating the weighted interaction graph comprises: determining that the first entity has the particular likelihood that can be inferred to the second entity (Zheng, [0039], where the user interest inference engine 102 collects and processes correspondences among the users, topics of the correspondences, etc. Based on obtained information, user interest inference engine 102 constructs a correspondence graph for the set of users 106, in which each node of the graph corresponds to one of the users 106. For each seed user (a node in the correspondence graph, whose interests are considered known and accurate, e.g., based on declared interests or explicit user actions), the user interest inference engine 102 propagates his/her known interest(s) through the links in the correspondence graph to the neighbors.
See also Zheng, [0047], [0052], [0055], and [0060], with respect to the weights “indicat[ing]…a likelihood of each entity interacting with a corresponding interaction class”), where each link in the correspondence graph is associated with a set of weights over interests, representing the connection strength over interests between respective pair of users. Static weights are calculated for each user/pair connection and aggregated implied interest sets for each user pair/connection. Interest is propagated to neighbors with attenuation or amplification proportional to the weight set of the link, e.g., levels of known interest signals from the seed users may be amplified or attenuated by the weights associated with the links (connections strengths) over the network in the correspondence graph. The levels of the known interests at each neighbor of the seed user are adjusted based on the connection strength between the neighbor and the seed user), wherein the second set of data did not previously indicate any historical interactions with the particular interaction class by the second entity and did not indicate an association between the first entity and the second entity (see Zheng, [0063] above, where a mutual interest in football between two users may result in an interest in NBA from one of the users to be inferred to the other user); …
inferring, by the computing device and based on the first set of data, the second set of data, the determined similarity between the one or more first interaction classes and the one or more second interaction classes, and the determined similarity between the first entity and the second entity, a set of new entity-to-class relationships (see Zheng, [0039], [0047], [0052], [0055], [0060], and [0063] above) comprising:
a first new entity-to-class relationship between the first entity and the at least one of the one or more first interaction classes based at least in part on comparing the first entity-to-class weights from the first entity to entity-to-class weights of a lookalike-scored entity from the second set of data … (Liao, [8:50-67]-[9:1-6], where the cluster group module 240 performs a lookalike expansion algorithm on a seed group of users to obtain a larger group of users who share characteristics with the seed group, thereby generating a larger group of users. See also IBR-Li, [0037], where confidence scores for users is based on a generated user model for an advertiser, where as a user exhibits more features in the user model for an advertiser, the confidence score for that user increases (i.e., “look-a-like-scored entities”)); and
a second new entity-to-class relationship between the second entity of the second set of entities and the particular interaction class based at least in part on comparing entity-to-class weights from a propensity-scored entity from the first set of data to the second entity-to-class weights of the second entity … (IBR-Li, [0031], where users in the training cluster of users have engaged with the advertiser, whose keyword profiles are then used to determine a set of keyword features for the user model. See IBR-Kirti, [0031], where a cluster score is generated for a user, where if the cluster score equals or exceeds a cluster group cutoff score, the user is included in a cluster group associated with the ad request. See also IBR-Kirti, [0046], where users in the cluster group have at least a threshold affinity for the advertisement included in the ad request (i.e., “propensity-scored entities”). See IBR-Li, [0026] and [0029] above with respect to “wherein the second entity was not previously related to the particular interaction class”); [and]
generating, by the computing device and using the weighted interaction graph, an extended set of data for the particular interaction class, wherein the extended set of data comprises the second entity (Zheng, [0061], where for seed user U5, its known interest sets (e.g., cooking: strong; fishing: fair) are propagated to each of neighbors, i.e., users U10, U6, U3, and U1, respectively. The level of each known interest is adjusted based on the weights associated with each link between the seed user U5 and the respective neighbors. The weights over interests are used for adjusting the level of known interests when they are propagated from the seed users to their neighbors).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Liao and Zheng (hereinafter “Liao as modified”) with the motivation of effectively profiling users to deliver relevant content to, without requiring users to explicitly update their interest profile information (i.e., thus increased convenience for users, while greater data accuracy for content providers such as advertisers).
Furthermore, it would have been obvious to one of ordinary skill in the art to have utilized Liao’s disclosure in generating Zheng’s weighted interaction graph by incorporating similarity in user behavior when determining whether to infer user interests from one user to another in unknown situations (e.g., when users did not previously have historical interactions with a particular object/content/topic) with the motivation of broadening the types of interests that can be inferred (e.g., including when information is sparser) while still retaining some degree of accuracy in inferring user interests.
Liao as modified does not appear to explicitly teach determining, by the computing device and based on the first set of data and the second set of data, a similarity of entity-to-class relationships between the first entity of the first set of data and the second entity of the second set of data based at least in part on interaction classes related to the first entity and interaction classes related to the second entity; wherein the likelihood of each entity interacting with a corresponding interaction class is used to determine first entity-to-class weights for the first entity and second entity-to-class weights for the second entity; generating a weight between the second entity and the particular interaction class to indicate a likelihood of interaction for the second entity with the particular interaction class based at least in part on the particular likelihood of interaction for the first entity with the particular interaction class; [wherein the first new entity-to-class relationship between the first entity and the at least one of the one or more first interaction classes was generated] wherein the at least one of the one or more first interaction classes were not previously related to the first entity in the first set of data; and [wherein the second new entity-to-class relationship between the second entity and the particular interaction class was generated] wherein the second entity was not previously related to the particular interaction class.
Rao teaches determining, by the computing device and based on the first set of data and the second set of data, a similarity of entity-to-class relationships between the first entity of the first set of data and the second entity of the second set of data based at least in part on interaction classes related to the first entity and interaction classes related to the second entity (Rao, [6:14-62], where the machine learning model looks for other users with similar purchasing behavior as the particular user and determines what other categories those other users have purchased in that the particular user has not purchased in, and predicts new categories that the particular user is likely to be interested in based on this);
wherein the likelihood of each entity interacting with a corresponding interaction class is used to determine first entity-to-class weights for the first entity and second entity-to-class weights for the second entity; and generating a weight between the second entity and the particular interaction class to indicate a likelihood of interaction for the second entity with the particular interaction class based at least in part on the particular likelihood of interaction for the first entity with the particular interaction class (Rao, [2:44-67]-[3:1-18], where a graph is constructed from historical purchase data comprising a plurality of nodes representing the plurality of categories, where edges connecting two nodes represent at least one instance in which a user of plurality of users purchased items from both categories represented by two nodes, or a “co-purchase” (i.e., thus, the graph being an “interaction graph”). The edges may have respective weights (i.e., “wherein the likelihood of each entity interacting with a corresponding interaction class is used to determine [first and second entity-to-class weights for the first and second entities]”), which increases as more purchases are made in both categories. Only those edges of significance may be retained, and thus recommended to users, e.g., for predicting new categories a particular user is likely to be interested in (Rao, [2:15-43]) (i.e., “a likelihood of each entity interacting with a corresponding interaction class”));
[wherein the first new entity-to-class relationship between the first entity and the at least one of the one or more first interaction classes was generated] wherein the at least one of the one or more first interaction classes were not previously related to the first entity in the first set of data; and [wherein the second new entity-to-class relationship between the second entity and the particular interaction class was generated] wherein the second entity was not previously related to the particular interaction class (Rao, [4:37-50], wherein one or more previous categories a target user has interacted with may be processed through the trained neural network, and output a new category based at least in part on the one or more previous categories the target user interacted with, wherein the new category represents the next category a target user is predicted to interact with which the target user has not interacted with before).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Liao as modified and Rao (hereinafter “Liao as modified”) with the motivation of utilizing a graph to represent the entirety of previous data with less actual data, thereby requiring less memory and other computational resources, particularly in the training of a machine learning model (which is trained on the graph data rather than the raw data) (Rao, [2:15-43]), and allowing for new category recommendations to be personalized for each user thereby increasing user experience and increasing the likelihood that the customer will engage with the recommended items (Rao, [4:10-17]).
Regarding claim 2: Liao as modified teaches The computer-implemented method of claim 1, wherein the first source of data is disjunctive with respect to the second source of data (IBR-Li, [0026], where an action log may track users’ actions on the social networking system 100 (i.e., “first source of data”) as well as external websites that communicate information back to the social networking system. See also IBR-Li, [0031], where the user’s interactions on the social networking system (i.e., “first source of data”) and user’s interactions with external systems (i.e., “is disjunctive with respect to the second source of data”) may be captured by the social networking system).
Regarding claim 3: Liao as modified teaches The computer-implemented method of claim 1, wherein determining the set of new relationships comprises, for each entity included in the first set of entities and in the second set of entities:
receiving, by the computing device, historical interaction data (IBR-Li, [0025], where an action logger receives communications from the web server 208 about user actions on and/or off the social networking system, and records, e.g., viewing histories, advertisements clicked on, purchasing activity, etc.); and
determining, by the computing device and using the historical interaction data, a plurality of relationships between the entity and a plurality of classes (IBR-Li, [0028], where edges (i.e., “relationships”) may be generated between a user and objects (i.e., “plurality of classes”) in the social networking system 100 when users interact with those objects one or more times (i.e., “historical interaction data”)).
Regarding claim 4: Liao as modified teaches The computer-implemented method of claim 3, wherein each relationship included in the plurality of relationships comprises a likelihood of the entity interacting with a corresponding class of the plurality of classes (IBR-Kirti, [0048], where a cluster score between the user and cluster group is determined, which represents a measure of a user’s affinity for content associated with targeting criteria associated with the cluster group (i.e., “plurality of classes”) based on the characteristics of the user, which may be retrieved from the action log 320 and edge store 325. The cluster model predicts an affinity of a user for content associated with targeting criteria associated with the cluster group (i.e., “plurality of classes”), which provides an indication of a likelihood of the user interacting with content associated with targeting criteria associated with the cluster group).
Regarding claim 5: Liao as modified teaches The computer-implemented method of claim 1, wherein the first set of data includes propensity-scored entities (IBR-Li, [0031], where users in the training cluster of users have engaged with the advertiser, whose keyword profiles are then used to determine a set of keyword features for the user model. See IBR-Kirti, [0031], where a cluster score is generated for a user, where if the cluster score equals or exceeds a cluster group cutoff score, the user is included in a cluster group associated with the ad request. See also IBR-Kirti, [0046], where users in the cluster group have at least a threshold affinity for the advertisement included in the ad request (i.e., “propensity-scored entities”)), wherein the second set of data includes look-a-like-scored entities (Liao, [8:50-67]-[9:1-6], where the cluster group module 240 performs a lookalike expansion algorithm on a seed group of users to obtain a larger group of users who share characteristics with the seed group, thereby generating a larger group of users. See also IBR-Li, [0037], where confidence scores for users is based on a generated user model for an advertiser, where as a user exhibits more features in the user model for an advertiser, the confidence score for that user increases (i.e., “look-a-like-scored entities”)), and
wherein the computer-implemented method further comprises receiving, by the computing device, a third set of data, which includes unscored entities and that is at least partially different than the first set of data and the second set of data (Liao, [1:66-67]-[2:1-18], where a new group of users of the online system is defined for a content provider to use as an audience for its content. The online system performs a lookalike expansion on the specified subset to obtain a larger group of users of the online system who are similar to the specified subset. See Liao, [12:52-67]-[13:1-46], where a custom audience is initially used as the seed group in a lookalike expansion algorithm, which uses the users’ weights to tune the resulting expanded group. The system determine whether to include a candidate user in the expanded audience (i.e., “unscored entities and that is at least partially different than the first set of data”)), from a third source of data that is different than the first source of data and the second source of data (IBR-Li, [0019], where an advertiser may target an advertisement to a targeting cluster of users of the social networking system that are similar to a training cluster of users as defined by targeting criteria 120, such as users that have engaged with similar advertisements, users that have performed an interaction with a specified object on the social networking system, as well as users that have performed a particular action on a system external to the social networking system 100 (i.e., “that is different from the first source of data”), in which there may be multiple external systems (IBR-Li, [0017]). See also IBR-Li, [0013] and [0017], where external systems may be multiple).
Although Liao as modified does not appear to explicitly state a “third” source of data (as well as “third” set of entities, etc.), Liao as modified discloses that more than one other external source of information. Therefore, Liao as modified suggests that multiple external sources of information including, e.g., a third source of data as claimed, may exist. Therefore, although Liao as modified does not appear to explicitly state a third source, Liao as modified discloses that the aforementioned steps may be performed for any user that previously had not interacted with the advertisements, which may include, e.g., a third source of data. Therefore, one of ordinary skill in the art would have found it obvious to modify Liao as modified to include such a third source with predictably equivalent operating characteristics—which is that users, regardless of the specific external system that information was derived from, are analyzed and grouped, with the motivation of enabling the system to operate on a wide range of data sources rather than limiting to, e.g., just two data sources as explicitly claimed.
Regarding claim 6: Liao as modified teaches The computer-implemented method of claim 5, wherein determining the set of new relationships between each entity of the first set of entities and the second set of entities and the set of classes inferred from the first set of data and from the second set of data comprises determining, by the computing device, the set of new relationships between each entity of the first set of entities, the second set of entities, and the unscored entities and a different set of classes inferred from the first set of data, from the second set of data, and from the third set of data (Zheng, [0039], where the user interest inference engine 102 collects and processes correspondences among the users 106 (i.e., “determining…a set of relationships between each entity of the first set of entities and the second set of entities”), topics of the correspondences, etc. Based on obtained information, user interest inference engine 102 constructs a correspondence graph for the set of users 106, in which each node of the graph corresponds to one of the users 106. For each seed user (a node in the correspondence graph, whose interests are considered known and accurate, e.g., based on declared interests or explicit user actions), the user interest inference engine 102 propagates his/her known interest(s) through the links in the correspondence graph to the neighbors, e.g., by summing up the contributions from all neighbors with the assumption that users who communicate with each other likely share the same interests. See also Zheng, [0063], with regards to the likelihood of each entity interacting with a corresponding class (see the cited portion, Zheng, [0063] below for further explanation)).
Regarding claim 8: Claim 8 recites substantially the same claim limitations as claim 1, and is rejected for the same reasons.
Note that Liao teaches A non-transitory machine-readable storage medium comprising a computer-program product that includes instructions configured to cause a data processing apparatus to perform operations comprising [the claimed steps] (Liao, [14:32-67], where the disclosed system may be embodied as a computer program stored in a non-transitory, tangible computer readable medium containing computer program code which can be executed by a computer processor to perform the disclosed steps).
Regarding claim 9: Claim 9 recites substantially the same claim limitations as claim 2, and is rejected for the same reasons.
Regarding claim 10: Claim 10 recites substantially the same claim limitations as claim 3, and is rejected for the same reasons.
Regarding claim 11: Claim 11 recites substantially the same claim limitations as claim 4, and is rejected for the same reasons.
Regarding claim 12: Claim 12 recites substantially the same claim limitations as claim 5, and is rejected for the same reasons.
Regarding claim 13: Claim 13 recites substantially the same claim limitations as claim 6, and is rejected for the same reasons.
Regarding claim 15: Claim 15 recites substantially the same claim limitations as claim 1, and is rejected for the same reasons.
Note that Liao teaches A system, comprising: one or more data processors; and a non-transitory computer-readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform operations comprising [the claimed steps] (Liao, [14:32-67], where the disclosed system may be embodied as a computer program stored in a non-transitory, tangible computer readable medium containing computer program code which can be executed by a computer processor to perform the disclosed steps).
Regarding claim 16: Claim 16 recites substantially the same claim limitations as claim 3, and is rejected for the same reasons.
Regarding claim 17: Claim 17 recites substantially the same claim limitations as claim 4, and is rejected for the same reasons.
Regarding claim 19: Claim 19 recites substantially the same claim limitations as claim 6, and is rejected for the same reasons.
Regarding claim 24: Liao as modified teaches The computer-implemented method of claim 1, wherein inferring, by the computing device and based on the first set of data, the second set of data, the determined similarity between the one or more first interaction classes and the one or more second interaction classes, and the determined similarity between the first entity and the second entity, a set of new entity-to-class relationships further comprises:
generating, for the first entity, a first entity-specific weighted interaction graph comprising the first entity and the one or more first interaction classes from the first set of data; generating, for the second entity, a second entity-specific weighted interaction graph comprising the second entity and the one or more second interaction classes from the second set of data (Zheng, [0043], where for different sets of users, their respective correspondence graphs are stored in the correspondence graph database); and
arranging elements of the first entity-specific weighted interaction graph to indicate likelihoods of interaction for the second entity based at least in part on the second entity-specific weighted interaction graph (Zheng, [0043], where the user correspondence graph generator 302 builds and updates a correspondence graph for a set of users based on the correspondences among the set of users, e.g., identifying implicit links among users by mining useful information from the users’ communication, and infer interests of users by determining interests of other users based on known interests of seed user(s) and the set of user’s correspondence graph. See Zheng, [0047], [0052], [0055], and [0060] in claim 1 above, with respect to the “indicate likelihoods of interaction”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have combined the teachings of Liao as modified and Zheng with the motivation of increasing search speed by having individual correspondence graphs for each user, e.g., instead of having a very large graph compr