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 .
This office action is in response to submission of application on 03/15/2024
Claims 1-20 are presented for examination.
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 because the claimed invention is directed to an abstract idea without significantly more.
Step 1: Is the claim to a process, machine, manufacture or composition of matter?
Claims 1-10 are directed to a method, claims 11-19 is directed to a machine, and claim 20 is directed to a non-transitory computer readable medium; therefore, all claims are directed to one of the four statutory categories.
Step 2A Prong One: Does the claim recite an abstract idea, law of nature, or natural phenomenon?
Claim 1 recites limitations of:
computing, using a suggestion retrieval system, a group-to-post relevance score for each post from the group content submissions; - mathematical concept (relationships, formulas or equations, calculations) of computing a relevance score
identifying, using a first machine learning model of the suggestion retrieval system, a set of recommended posts from the group content submissions having the group-to-post relevance score that at least reaches a group-to-post relevance score threshold for a group, wherein the group is associated with a group attribute; - mental process (observation, evaluation, judgement) as a human mind is able to compare a score to a threshold
classifying, using an intent-based ranking system and the group attribute, at least one post from the set of recommended posts as relevant or non-relevant to the group using a second machine learning model of the intent-based ranking system; - mental process (observation, evaluation, judgement) as a human mind is able to judge whether a post is relevant or non-relevant to a group
computing, using the intent-based ranking system, a relevance ranking score of the at least one post from the set of recommended posts classified as relevant to the group using the second machine learning model; - mathematical concept (relationships, formulas or equations, calculations) of computing a relevance ranking score
and identifying, using the intent-based ranking system, a set of suggested posts having the relevance ranking score that at least reaches a relevance ranking score threshold of the group; - mental process (observation, evaluation, judgement) as a human mind is able to compare a score to a threshold and select the posts that reach it
Step 2A Prong Two: Does the claim recite additional elements that integrate the judicial exception into a practical application?
Claim 1 recites limitations of:
accessing group content submissions to an online community platform; - accessing group content submissions merely amounts to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g), item (3), which identifies necessary data gathering and outputting as an example of extra-solution activity.
and causing at least one post of the set of suggested posts to be presented on a device. – causing a post to be presented merely amounts to outputting data which is insignificant extra-solution activity. See MPEP § 2106.05(g), item (3), which identifies necessary data gathering and outputting as an example of extra-solution activity.
Step 2B: Does the claim recite additional elements that amounts to significantly more than the judicial exception?
The additional elements are:
accessing group content submissions to an online community platform; - accessing group content submissions merely amounts to data gathering which is insignificant extra-solution activity. See MPEP § 2106.05(g). Data gathering is well-understood, routine, and conventional. See MPEP 2106.05(d)(II)(iv).
and causing at least one post of the set of suggested posts to be presented on a device. – causing a post to be presented merely amounts to outputting data which is insignificant extra-solution activity. See MPEP § 2106.05(g). Data gathering is well-understood, routine, and conventional. See MPEP 2106.05(d)(II)(iv).
The additional elements do not amount to significantly more than the abstract idea. Therefore,
the claim is not patent eligible.
Independent claim 11 and 20 recites the same relevant limitations and a similar analysis applies. Claim 11 recites the additional elements of, “A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the computing apparatus to:” - components recited at a high level are construed as generic computer components used to implement the abstract idea. See MPEP 2106.05(f)(2). Claim 20 recites the additional elements of, “A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a processor, cause the processor to:” - components recited at a high level are construed as generic computer components used to implement the abstract idea. See MPEP 2106.05(f)(2). They do not integrate the abstract idea into a practical application. Nor do they amount to significantly more. Therefore, the independent claims are not patent eligible.
The above analysis similarly applies to the dependent claims.
Dependent claim 2 and 12 recites,
“training, using the suggestion retrieval system, the first machine learning model based on semantic attribute data from a first analysis of public posts and group description attributes, and statistical attribute data from a second analysis of group member profile attributes of the online community platform,” - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
“wherein the group attribute comprises the group description attributes and group member profile attributes,” - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
“wherein the group content submissions comprise: public posts and pending posts” - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
Dependent claim 3 and 13 recites,
“generating a mapping of groups-to-topic clusters using a group-to-topic clusters machine learning model;” - mathematical concept (relationships, formulas or equations, calculations) of using clustering because its computing the relationships among data points and assigning each group to the cluster
“generating a mapping of post-to-topic clusters using a filtered posts topic cluster machine learning model,” - mathematical concept (relationships, formulas or equations, calculations) as assigning each post to a topic cluster requires computing the post’s relationship to each cluster
“wherein the first machine learning model comprises the group-to-topic clusters machine learning model and the filtered posts topic cluster machine learning model.” - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
Dependent claim 4 and 14 recites,
“generating a mapping of groups-to-posts based on the mapping of groups-to-topic clusters and the mapping of the post-to-topic clusters.” - mental process (observation, evaluation, judgement) as a human mind is able to create a mapping of groups-to-posts
Dependent claim 5 and 15 recites,
“computing, using the mapping of groups-to-posts, the group-to-post relevance score for each post from the public posts and pending posts based on a product of a relevance score of a post-to-topic cluster and a concentration of group posts relevant to a topic cluster.” - mathematical concept (relationships, formulas or equations, calculations) of computing the group-to-post relevance score for each post
Dependent claim 6 and 16 recites,
“identifying the group member profile attributes of the online community platform, wherein the group member profile attributes comprise a skill attribute, an industry attribute, a number of group members attribute, and a number of group members attribute having a predefined skill;” - mental process (observation, evaluation, judgement) as a human mind is able to identify the group member profile attributes
“computing a statistical analysis relevance score of a skill or an industry for the group based on the group member profile attributes;” - mathematical concept (relationships, formulas or equations, calculations) of computing a statistical analysis relevance score
“identifying a skill or an industry with the statistical analysis relevance score of the skill or industry for the group exceeding a skill or industry group to post relevance score threshold for the group;” - mental process (observation, evaluation, judgement) as a human mind is able to identify a skill or an industry
“generating skill and industry group mapping based identifying the skill or the industry with the statistical analysis relevance score of the skill or industry for the group exceeding the skill or industry group to post relevance score threshold for the group,” - mental process (observation, evaluation, judgement) as a human mind is able to create a skill and industry group based on determining the skill or the industry
“wherein the group-to-post relevance score is further based on the skill and industry group mapping”. - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
Dependent claim 7 and 17 recites,
“identifying hashtags from the public posts;” - mental process (observation, evaluation, judgement) as a human mind is able to identify hashtags
“and identifying relevant hashtags from the hashtags based on hashtag to skills mapping data generated based on a statistical correlation between skills and hashtags”. - mental process (observation, evaluation, judgement) as a human mind can identify relevant hashtags
Dependent claim 8 and 18 recites,
“augmenting or sanitizing the mapping of groups-to-topic clusters by using a natural language inference model (NLI) with a group title and description, and textual representation of interest clusters as inputs,” - this element constitutes “mere instructions to apply an exception.” (MPEP § 2106.05(f))
“wherein the natural language inference model outputs a probability of two text samples being related to one another.” - mathematical concept (relationships, formulas or equations, calculations) of computing a probability of two text samples
Dependent claim 9 and 19 recites,
a supervised learning model comprising a Siamese network that assesses a similarity between text of group posts and a group definition; - mathematical concept (relationships, formulas or equations, calculations) of assessing similarity between two texts
an unsupervised learning model that applies a clustering algorithm to organize group posts into distinct clusters, - mathematical concept (relationships, formulas or equations, calculations) as a clustering algorithm computes the relationship between the posts and assigns them to cluster
wherein the intent-based ranking system is configured to combine a first similarity score from the supervised learning model and a second similarity score from the unsupervised learning model to identify a similarity score of published public group content submission or pending group content submission, - mathematical concept (relationships, formulas or equations, calculations) of combining two values into a score
the similarity score exceeding a predefined threshold indicating a submission as relevant to the group. - mental process (observation, evaluation, judgement) as a human mind can compare a score to a threshold and judge the submission to be relevant
Dependent claim 10 recites,
“generating a first feed comprising one or more posts from the pending posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group;” - mental process (observation, evaluation, judgement) as a human mind can compare each post’s score to a threshold and collect the posts that exceed it into a grouping
“generating a second feed comprising one or more posts from the pending posts with a corresponding relevance score lower than the relevance ranking score threshold of the group;” - mental process (observation, evaluation, judgement) as a human mind can collect the posts falling below the threshold
“generating a third feed comprising one or more posts from the public posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group;” - mental process (observation, evaluation, judgement) as a human mind can collect the public posts exceeding the threshold
“generating a graphical user interface comprising the first feed, the second feed, and the third feed; and presenting, at a client device, the graphical user interface.” – generating a graphical user interface that displays the results merely amounts to outputting data which is insignificant extra-solution activity. See MPEP § 2106.05(g). Data gathering is well-understood, routine, and conventional. See MPEP 2106.05(d)(II)(iv).
The dependent claims do not integrate the abstract idea into a practical application, nor do they amount to significantly more than the abstract idea.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The following references are being used:
Seth et al., (GROUP RECOMMENDATION BASED ON TOPIC CLUSTERING, US 20240403694 A1, herein Seth)
Mittal et al., (Group Recommendation For User-generated Content, US 12216720 B2, herein Mittal)
Li et al., (Selecting Content For Presentation To A User Of A Social Networking System Based On A Topic Associated With A Group Of Which The User Is A Member, US 11636165 B1, herein Li)
Gerrard et al., (MEMBER FEATURE SETS, GROUP FEATURE SETS AND TRAINED COEFFICIENTS FOR RECOMMENDING RELEVANT GROUPS, US 20170220935 A1, herein Gerrard)
Yin et al., (Benchmarking Zero-shot Text Classification: Datasets, Evaluation and Entailment Approach, herein Yin)
Reimers et al., (Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks, herein Reimers)
Fang et al., (CHAOTIC-TO-FINE CLUSTERING FOR UNLABELED PLANT DISEASE IMAGES, herein Fang)
Alfred et al., (Method And Apparatus For Providing Control Of Social Networking Sites, US 10778804 B2, herein Alfred)
Claim(s) 1-5 and 11-15 is/are rejected under 35 U.S.C. 103 as being unpatentable over Li, Seth, and Mittal.
Regarding claim 1,
Li teaches,
accessing group content submissions to an online community platform; (Li, claim 9, “the information describing the plurality of objects maintained in the social networking system comprises content posted to the target group by a member of the target group”, note: Li runs on a social network where members join groups and post content to them. The system retrieves the information stored for a group, and that information includes the content members posted there. Content a member posted to a group is a group content submission, and retrieving it is accessing it)
computing, using a suggestion retrieval system, a group-to-post relevance score for each post from the group content submissions; (Li, “A similarity score for each of the candidate content items may be computed 360 (e.g., using the scoring module 255) by the social networking system 140. The social networking system 140 may compute 360 a similarity score for a candidate content item based at least in part on a measure of similarity between one or more topics associated with the candidate content item and one or more topics associated with the group.”, note: Li scores how closely each item’s topics match the group’s topics. A score of how well an item matches a group is a score of that item’s relevance to the group. Li says the score is computed for each candidate item, which meets for each post. The scoring and selection modules perform the claimed function)
identifying, using a first machine learning model of the suggestion retrieval system, a set of recommended posts from the group content submissions having the group-to-post relevance score that (Li, claim 1, “the similarity score is output from a machine learning model that is trained using training data that describes previous actions performed by members of one or more groups with content items associated with one or more other topics” note: Li’s model is trained on members’ past actions with content items and outputs the similarity score, and it sits in the same pipeline, so it is the first ML model.)
and causing at least one post of the set of suggested posts to be presented on a device. (Li, “the social networking system 140 may send 380 the content feed for display to the viewing user via a group page associated with the group. In this example, the group page may be presented on a client device 110 (e.g., a mobile device) associated with the viewing user.”, note: Li puts the selected items into a feed and sends that feed for display on the group page, which the member views on the phone, laptop, or similar device. The items in the feed are the ones Li selected based on their similarity scores, so they are the suggested posts, and sending the feed causes them to be presented on a device)
Li does not teach,
at least reaches a group-to-post relevance score threshold for a group, wherein the group is associated with a group attribute
classifying, using an intent-based ranking system and the group attribute, at least one post from the set of recommended posts as relevant or non-relevant to the group using a second machine learning model of the intent-based ranking system;
computing, using the intent-based ranking system, a relevance ranking score of the at least one post from the set of recommended posts classified as relevant to the group using the second machine learning model;
and identifying, using the intent-based ranking system, a set of suggested posts having the relevance ranking score that at least reaches a relevance ranking score threshold of the group;
Seth teaches,
at least reaches a group-to-post relevance score threshold for a group, wherein the group is associated with a group attribute (Seth, paragraph [0107], “If the post score is above a post threshold then the method flows to operation 1014 to recommend the post for being posted in the selected group;” note: Seth compares the post’s score to a threshold and recommends the post for the group only when the score is above that threshold.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li and Seth because Li selects posts for a group by ranking them against each other, and Seth’s comparison of each post’s score to a threshold gives consistent recommendations.
Seth does not teach,
classifying, using an intent-based ranking system and the group attribute, at least one post from the set of recommended posts as relevant or non-relevant to the group using a second machine learning model of the intent-based ranking system;
computing, using the intent-based ranking system, a relevance ranking score of the at least one post from the set of recommended posts classified as relevant to the group using the second machine learning model;
and identifying, using the intent-based ranking system, a set of suggested posts having the relevance ranking score that at least reaches a relevance ranking score threshold of the group;
Mittal teaches,
classifying, using an intent-based ranking system and the group attribute, at least one post from the set of recommended posts as relevant or non-relevant to the group using a second machine learning model of the intent-based ranking system; (Mittal, claim 1, “using an intent machine-learning (ML) model to calculate the group interest scores, wherein the intent ML model is trained with training data based on past content and assigned labels for corresponding intents” note: Mittal’s intent model is trained on past content labeled with intents and outputs one intent per post. Mittal then eliminates posts with intents like personal updates or congratulations and keeps posts with intents like sharing knowledge. Putting each post in one of those two buckets is classifying it as non-relevant or relevant. That model is separate from the scoring model, so it is the second ML model, and it ties to the group interest scores that describe the group)
computing, using the intent-based ranking system, a relevance ranking score of the at least one post from the set of recommended posts classified as relevant to the group using the second machine learning model; (Mittal, FIG. 10, “The higher the dot score is for a pair, the higher the match scores, which indicates the confidence that the topic of the post is relevant for the topic of the group.”, note: Mittal’s match score measures how relevant a post’s topic is to a group’s topic, and Mittal uses it to rank candidates and pick the best, so it is a ranking score. Because Mittal filters out irrelevant posts before scoring and scores only what remains, the posts that get a score are the ones classified as relevant.)
and identifying, using the intent-based ranking system, a set of suggested posts having the relevance ranking score that at least reaches a relevance ranking score threshold of the group; (Mittal, FIG. 10, “The group recommendation is presented to the user if the match score content-group exceeds a predetermined threshold.”, note: Mittal compares the match score to a threshold and suggests only the posts that clear it. A score that exceeds a threshold has at least reached it. This threshold applies to a different score at a later stage than the cutoff.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching Li, Seth and Mittal because Li matches a post to a group by topic similarity alone, and Mittal’s trained intent model and match-score threshold prevent recommending posts whose intent is against the group’s rules even when the topics match. This allows for more precise, higher-quality group recommendations (Mittal, “In some experiments, it was determined that a threshold of 0.3 for the match score generates 60-70% precise recommendations.”)
Regarding claim 2,
Li teaches,
training, using the suggestion retrieval system, the first machine learning model based on semantic attribute data from a first analysis of public posts and group description attributes, (Li, “the social networking system 140 may train a machine-learning model to compute 360 a similarity score for a content item based on correlations between the actions performed by the members of the groups and the topics associated with the groups and with the content items,” note: the model is trained on the topics of the groups and of the content items. Li gets those topics by pulling keywords out of the group’s title, its tags, and the content posted to the group. Keywords taken from the text describe what the text is about, so they are semantic data. Li’s groups can be set to open, so the posts analyzed include public ones. Training on those topics is training on semantic data from posts and from the group’s descriptive text.
and statistical attribute data from a second analysis of group member profile attributes of the online community platform, (Li, “the social networking system 140 may retrieve information describing a frequency of actions performed by members of different groups with content items, in which both the groups and the content items are associated with various topics”, note: Li counts how often group members act on content, and a count is a statistic, not a semantic feature. Li performs this kind of member analysis elsewhere too. It pulls keywords from member’s demographic, hobby, and occupation information, and it assigns a group an attribute when at least a threshold number or percentage of members did something.)
wherein the group attribute comprises the group description attributes and group member profile attributes, (Li, “the social networking system 140 may store 345 a topic associated with the group in association with a name or other identifier associated with the group in the group store 230.”, note: The group attribute relied on in claim 1 is the topic Li stores for the group. Li builds that topic from two sources, the group’s own descriptive text, and data about the group’s members. Because both feed the stored topic, the group attribute includes both kinds of data.)
wherein the group content submissions comprise: public posts and pending posts (Li, “privacy settings for the group (e.g., “open,” “closed,” or “secret”),” note: Li’s groups have a privacy setting, and “open” means the group’s content is not restricted to members. Posts made in an open group are public posts, and Li retrieves and analyzes those posts as part of the group’s content)
Regarding claim 3,
Seth teaches,
generating a mapping of groups-to-topic clusters using a group-to-topic clusters machine learning model; (Seth, paragraph [0070], “The result is a table that maps each group (e.g., the group ID) to a topic ID” note: Seth first groups similar posts into clusters and gives each cluster a topic ID. It then builds a table tying each group to the topic ID that shows up most often in that group’s posts. That table is a mapping of groups to topic clusters. The topic IDs comes from Seth’s clustering model, which learns the clusters from the post data, so the mapping is produced using a machine learning model)
generating a mapping of post-to-topic clusters using a filtered posts topic cluster machine learning model, (Seth, paragraph [0067], “After the clustering, each post is associated with a topic identifier (ID) that represents the corresponding cluster.”, note: Seth attaches a topic ID to each post, which links that post to its cluster. That link is the post-to-topic-cluster mapping. Seth also trains a post classifier to assign a topic ID to a new post, and a new post reaches that classifier only after Seth screens out posts that are too short and posts whose intent does not fit, so the model operates on filtered posts.)
wherein the first machine learning model comprises the group-to-topic clusters machine learning model and the filtered posts topic cluster machine learning model. (Seth, paragraph [0111], “The social networking server 1112 includes, among other modules, a clustering system 1116, the group membership service 902, and the post classifier model 706.”, note: Seth’s system contains two modes that work together, the clustering system that creates the topic clusters and ties groups to them, and the post classifier that assigns a topic to each post. Both run inside the same service that scores posts against groups, so the modeling used on the retrieval side is made up of both models.)
Regarding claim 4,
Seth teaches,
generating a mapping of groups-to-posts based on the mapping of groups-to-topic clusters and the mapping of the post-to-topic clusters. (Seth, paragraph [0086], “once the topic ID is generated, the topic-to-group table is checked to obtain the relevant group that maps the topic” note: Seth connects posts to groups in two steps using the mapping from claim 3. First it finds the post’s topic ID, which is the post-to-topic-cluster link. Then it checks the topic-to-group table, which is the group-to-topic-cluster-link, to see which group that topic belongs to. Because both mappings share the same topic ID, joining them produces a link between the post and the group. That link is the groups-to-posts.)4
Regarding claim 5,
Mittal teaches,
computing, using the mapping of groups-to-posts, the group-to-post relevance score for each post from the public posts and pending posts based on a product of a relevance score of a post-to-topic cluster and a concentration of group posts relevant to a topic cluster. (Mittal, FIG. 10, “Thus, the result of multiplying the interest scores for cloud computing is 0.97×0.37, which is 0.36.”, note: Mittal calculates a score for how well a post fits a group by taking each topic, multiplying the post’s score for that topic by the group’s score for that topic, and summing across topics. That multiplication is the product. The first factor is the post’s own topic score, which is its relevance to that topic cluster. The second factor is the group’s topic score, which Mittal computes as the average of the topic scores across the group’s posts. An average across the group’s posts measures how much of the group’s content falls in that topic, which is the concentration of group posts relevant to the topic cluster.)
Claims 11-15 are a machine, A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the computing apparatus to: (Li, “Embodiments also may relate to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, and/or it may comprise a general-purpose computing device selectively activated or reconfigured by a computer program stored in the computer.”) that corresponds to method claim 1-5, respectively. Otherwise, they are not patentably distinguishable. Therefore, claims 11-15 are rejected for the same reasons as claims 1-5, respectively.
Claim 20 is a computer readable medium, A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a processor, cause the processor to: (Li, claim 12, ”A computer program product comprising a non-transitory computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to:”) that corresponds to method claim 1. Otherwise, it is not patentably distinguishable. Therefore, claim 20 is rejected for the same reasons as claim 1.
Claim(s) 6 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over Li and Mittal and in further view of Seth and Gerrard.
Regarding claim 6,
Gerrard teaches,
identifying the group member profile attributes of the online community platform, wherein the group member profile attributes comprise a skill attribute, an industry attribute, a number of group members attribute, and a number of group members attribute having a predefined skill; (Gerrard, paragraph [0035], “A feature set includes any one or more of the following profile attributes: current industry, gender, professional experience, country, geographical region, educational degree, occupational function, employer (current and previous) and skill” note: Gerrard reads the profiles of a group’s members and pulls out attributes, and skill and industry are both on its list. Gerrard then counts how many members share a given attribute and divides by the group’s total membership, so it has both the total count and the count of members with that attribute)
computing a statistical analysis relevance score of a skill or an industry for the group based on the group member profile attributes; (Gerrard, paragraph [0041], “the Group Relevance Engine 206 identifies a percentage of member accounts currently subscribed to the group that each have a common type of profile attribute” note: Gerrard works out what share of a group’s members hold a particular skill or industry. That share comes straight from the member profile data and shows how strongly the attribute characterizes the group, which is a statistical relevance score.)
identifying a skill or an industry with the statistical analysis relevance score of the skill or industry for the group exceeding a skill or industry group to post relevance score threshold for the group; (Gerrard, paragraph [0042], “the Group Relevance Engine 206 determines the percentage of member accounts meets a percentage threshold”, note: Gerrard compares the percentage for a group against a cutoff and keeps only the attributes that clear it)
generating skill and industry group mapping based identifying the skill or the industry with the statistical analysis relevance score of the skill or industry for the group exceeding the skill or industry group to post relevance score threshold for the group, (Gerrard, paragraph [0043], “the Group Relevance Engine 206 sets the aggregate group feature to the common type of profile attribute” note: Once a skill or industry clears the threshold, Gerrard stores it as that group’s aggregate feature. Recording which skill or industry characterizes which group is the claimed mapping, and it is produced directly from the threshold result.)
Gerrard does not teach,
wherein the group-to-post relevance score is further based on the skill and industry group mapping.
Li teaches,
wherein the group-to-post relevance score is further based on the skill and industry group mapping. (Li, “the social networking system 140 may extract 325 keywords included in demographic information and information describing hobbies and occupations associated with members of a group (e.g., from the user profile store 205)” note: Li scores each post by comparing it to the group’s topics, and Li builds those topics partly from what its members do for a living. Gerrard’s mapping says which skills and industries characterize the group. Feeding that mapping into the member data Li already uses to build the group’s topics makes the score further based on it.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li, Mittal, Seth, and Gerrard because Li derives a group’s attribute from the text of the group and its members’ profiles without measuring how widely any attribute is shared, and Gerrard’s calculation of the percentage of group members holding a given skill or industry, compared against a threshold, keeps only the attributes that characterize the group, which produces more accurate group attribute and has better matches between posts and groups.
Regarding claim 7,
Mittal teaches,
identifying hashtags from the public posts; (Mittal, FIG. 8, “Post 802 has been posted by user Joe Smith, a developer, with the post text being as follows: Follow this guide to deploy a simple #Spring #Cloud #microservice #application to run on #Cloud.”, note: Mittal’s posts carry hashtags, and Mittal uses hashtags to label posts when building training data. The system must locate the hashtags inside the posts to do that. Mittal’s posts are published and can be public.)
and identifying relevant hashtags from the hashtags based on hashtag to skills mapping data generated based on a statistical correlation between skills and hashtags. (Mittal, FIG. 5, “In some example embodiments, a machine learning model, referred to herein as the interest model, is used to determine the interest associated with a piece of text. The interest model takes as an input the piece of text and generates a list of interests that are associated with the text together with a corresponding score, where the higher the score for an interest, the more relevant is the interest to the text.”, note: Mittal keeps a list of hashtags for each professional interest category, such as the hashtags that go with cloud computing. That stored list is mapping data between hashtags and those categories. Mittal feeds those pairings to a machine learning model as training data, and the model learns from them which text goes with which category and outputs a relevance score. Because the model learns the association from data rather than being told it, the mapping is built on a correlation learned from observation. Keeping the higher-scoring items narrow the full set down to the relevant ones.)
Claim 16-17 are a machine, A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the computing apparatus to: (Li, “Embodiments also may relate to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, and/or it may comprise a general-purpose computing device selectively activated or reconfigured by a computer program stored in the computer.”) that corresponds to method claim 6-7, respectively. Otherwise, they are not patentably distinguishable. Therefore, claim 16-17 are rejected for the same reasons as claims 6-7, respectively.
Claim(s) 8 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Li and Mittal and in further view of Seth and Yin.
Regarding claim 8,
Yin teaches
augmenting or sanitizing the mapping of groups-to-topic clusters by using a natural language inference model (NLI) with a group title and description, and textual representation of interest clusters as inputs, (Yin, page 2, section 1, “we propose to treat 0SHOT-TC as a textual entailment problem.”, note: Yin teaches checking whether a topic fits a piece of text by asking an NLI model whether the topic follows from the text. Applying Yin’s check to Seth’s mapping, with the group’s title and description as one input and the cluster’s text as the other, either confirms the group’s assigned topic or corrects it)
wherein the natural language inference model outputs a probability of two text samples being related to one another. (Yin, page 6, section 5, “If only seen or unseen labels are predicted as positive, we pick the one with the highest probability;” note: Yin’s model looks at a pair of texts, the input text and a sentence built from a label, and decides whether the first entails the second. It outputs a probability for that decision, and Yin picks labels by comparing those probabilities. A probability that one text entails another is a probability that the two are related.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li, Mittal, Seth, and Yin because Seth assigns each group the topic that appears most often in its posts, and spam or off-topic posts in a group can skew that count, and Yin’s entailment model checks whether a candidate topic actually follows from the group’s own title and description, which corrects wrong topic assignments and produces better matches between posts and groups.
Claim 18 is a machine, A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the computing apparatus to: (Li, “Embodiments also may relate to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, and/or it may comprise a general-purpose computing device selectively activated or reconfigured by a computer program stored in the computer.”) that corresponds to method claim 8, respectively. Otherwise, they are not patentably distinguishable. Therefore, claim 18 is rejected for the same reasons as claims 8, respectively.
Claim(s) 9 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Li and Mittal and in further view of Seth, Reimers and Fang.
Regarding claim 9,
Reimers teaches,
a supervised learning model comprising a Siamese network that assesses a similarity between text of group posts and a group definition; (Reimers, abstract, “In this publication, we present Sentence-BERT (SBERT), a modification of the pretrained BERTnetwork that use siamese and triplet network structures to derive semantically mean ingful sentence embeddings that can be compared using cosine-similarity.”, note: SBERT runs two texts through the same network and compares them to get a similarity score, learning from sentence pairs that people labeled. Li computes a similarity between a post’s topics and a group’s topics, and Li builds the group’s topics from its title, tags, and description, which is the group definition. Substituting SBERT’s Siamese comparison for Li’s similarity computation is the model)
an unsupervised learning model that applies a clustering algorithm to organize group posts into distinct clusters, (Reimers, section 8, “For example, clustering of 10,000 sentences with hierarchical clustering requires with BERT about 65 hours, as around 50 Million sentence combinations must be computed. With SBERT, we were able to reduce the effort to about 5 seconds” note: Reimers computes embeddings with SBERT and then runs a separate hierarchical clustering algorithm over those embeddings. That clustering algorithm is the unsupervised model, distinct from the Siamese network. Applied to Li’s group posts, it sorts them into separate clusters without any labels)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li, Mittal, and Reimers because Li and Mittal score a post against a group by comparing lists of topic scores, and Reimers’ Siamese network compares the post’s text directly against the group’s own text and produces a similarity score from the meaning of the two texts, which catches relevance that topic-list matching misses and does it fast enough to compare and cluster large numbers of posts.
Reimers does not teach,
wherein the intent-based ranking system is configured to combine a first similarity score from the supervised learning model and a second similarity score from the unsupervised learning model to identify a similarity score
of published public group content submission or pending group content submission, the similarity score exceeding a predefined threshold indicating a submission as relevant to the group.
Fang teaches,
wherein the intent-based ranking system is configured to combine a first similarity score from the supervised learning model and a second similarity score from the unsupervised learning model to identify a similarity score (Fang, page 7, section 3.4.2, “Final class determination. Through the Siamese network, we get a second similarity score S(xi) for xi in A. Finally, we calculate the average of S(xi) and S(xi) (i.e. the similarity score obtained in Similarity Finder), and the class with the highest average score will be the image’s final class.”, note: Fang scores each item two ways and averages the results into one number that decides where the item belongs. In the combination, the first score is the post-to-group definition score and the second is the post’s closeness to the clusters, averaged as Fang teaches. Fang pairs a score from a trained model with a score measured against unsupervised clusters.)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li, Mittal, Seth, Reimers, and Fang because a single similarity measure can misjudge where an item belongs, and Fang averages the trained Siamese network’s score with the item’s similarity to the clusters formed without supervision, so the two measures check each other and the item is assigned to the right group more accurately.
Fang does not teach,
of published public group content submission or pending group content submission, the similarity score exceeding a predefined threshold indicating a submission as relevant to the group.
Mittal teaches,
of published public group content submission or pending group content submission, the similarity score exceeding a predefined threshold indicating a submission as relevant to the group. (Mittal, FIG. 10, “The group recommendation is presented to the user if the match score content-group exceeds a predetermined threshold. In some experiments, it was determined that a threshold of 0.3 for the match score generates 60-70% precise recommendations” note: Mittal’s post is already published and can be public before it is scores, and whether it goes into any group’s feed is still open when the score is computed. Mittal checks the score against a fixed cutoff such as 0.3 and treats the post as belonging with the group only if it clears.)
Claim 19 is a machine, A computing apparatus comprising: a processor; and a memory storing instructions that, when executed by the processor, configure the computing apparatus to: (Li, “Embodiments also may relate to an apparatus for performing the operations herein. This apparatus may be specially constructed for the required purposes, and/or it may comprise a general-purpose computing device selectively activated or reconfigured by a computer program stored in the computer.”) that corresponds to method claim 9, respectively. Otherwise, they are not patentably distinguishable. Therefore, claim 19 is rejected for the same reasons as claims 9, respectively.
Claim(s) 10 is/are rejected under 35 U.S.C. 103 as being unpatentable over Li and Mittal and in further view of Seth and Alfred.
Mittal teaches
generating a first feed comprising one or more posts from the pending posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group; (Mittal, FIG. 3, “If the user selects option 306, the post will be added to the group feed of the suggested group.”, note: In Mittal, a post is scored against the group while its inclusion is still undecided, so it is pending as to that group. Only posts whose score clears the threshold are suggested, and those are the posts that end up in the group feed. The group feed therefore holds the pending posts that cleared the threshold)
Mittal does not teach,
generating a second feed comprising one or more posts from the pending posts with a corresponding relevance score lower than the relevance ranking score threshold of the group;
generating a third feed comprising one or more posts from the public posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group;
generating a graphical user interface comprising the first feed, the second feed, and the third feed;
and presenting, at a client device, the graphical user interface.
Alfred teaches,
generating a second feed comprising one or more posts from the pending posts with a corresponding relevance score lower than the relevance ranking score threshold of the group; (Alfred, claim 1, “receiving, by a processor, a request from a third party user for permission to post information on a social networking site associated with an owner, wherein the information is not posted on the social networking site associated with the owner until the request is granted” note: Alfred holds every submitted post in a queue and publishes nothing until the owner approves it, so the queue is a collection of posts that are pending. Mittal separately scores posts and splits them at a threshold. Combining the two, the posts that fall below the threshold are the ones left in the review queue for a person to decide on)
It would be obvious to one of ordinary skill in the art before the effective filing date of the
claimed invention to combine the teaching of Li, Mittal, Seth, and Alfred because Li and Mittal publish a post to the group as soon as the system decides it is relevant, and Alfred holds each submitted post in a review queue until the owner approves it, which lets the own stop inappropriate posts before they appear on the page rather than only deleting them afterwards.
Alfred does not teach,
generating a third feed comprising one or more posts from the public posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group;
generating a graphical user interface comprising the first feed, the second feed, and the third feed;
and presenting, at a client device, the graphical user interface.
Li teaches,
generating a third feed comprising one or more posts from the public posts with a corresponding relevance score exceeding the relevance ranking score threshold of the group; (Li, FIG. 3, “once the social networking system 140 has generated 375 a content feed, the social networking system 140 may send 380 the content feed for display to the viewing user via a group page associated with the group.”, note: Li scores every candidate item against the group’s topics, ranks them, selects the once that score well, and puts those into a content feed for the group page. Those items come from content already posted and viewable in the group, so they are public posts. A feed of public posts that scored high enough to be selected is the claimed third feed.)
generating a graphical user interface comprising the first feed, the second feed, and the third feed; (Li, FIG. 3, “the social networking system 140 may send 380 the content feed for display to the viewing user via a group page associated with the group.”, note: Li’s group page is a graphical user interface that displays a content feed for the group. Li shows one feed on that page)
and presenting, at a client device, the graphical user interface. (Li, “In this example, the group page may be presented on a client device 110 (e.g., a mobile device) associated with the viewing user.”, note: Li displays the group page on the member’s client device, which Li describes as a desktop, laptop, mobile telephone, or smartphone. This is presenting the graphical user interface at a client device.)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to UYEN-NHU PHAM TRAN whose telephone number is (571)272-1559. The examiner can normally be reached Monday - Friday 7:30-5.
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, Miranda Huang can be reached at (571) 270-7092. 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.
/U.P.T./ Examiner, Art Unit 2124
/MIRANDA M HUANG/ Supervisory Patent Examiner, Art Unit 2124
Prosecution Insights