DETAILED ACTION
Status of Claims
This communication is the final action on the merits in response to the amendments and arguments filed on March 3 2026 and April 13, 2026. Claims 21-29 were amended. Claims 21-29 are currently pending and have been examined.
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 .
Claim Objections
Claims 22-29 are objected to because of the following informalities:
Claims 22-29: “The system application of claim…” should read “The system of claim…”
Appropriate correction is required.
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.
Claims 21-29 are rejected under 35 U.S.C. 112(a) as failing to comply with the written description requirement. The claims contain 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, at the time the application was filed, had possession of the claimed invention.
Independent Claim 21 recites “the neural network model further trained to improve accuracy of the networking recommendation by optimizing the matching score to identify second user accounts having skills and information aligned with the network objective of the target user account, the improvement based on comparing the target NLP output to the candidate NLP output,” yet Applicant’s specification fails to disclose improving the accuracy of the networking recommendation by optimizing the matching score.
Independent Claim 21 further recites “train the neural network model further based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss,” yet Applicant’s specification fails to disclose iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss. Paragraph [0027] of the specification states that “the user 10 associated with the user device 110 may provide value of “4” for the loss 118 when rating how beneficial the networking recommendation 245 was for the user 10 on a scale of 1–5. The networking application 140 may use the loss 118 to train the neural network 400.” However, the specification does not provide any support for iteratively improving the model’s ability to generate accurate recommendations, or for any adjusting of model parameters.
Because the original disclosure does not support the identified limitations, one of ordinary skill in the art would not recognize the Applicant as in possession of the claimed invention at the time of filing. Therefore, Claim 21 is rejected under 35 U.S.C. 112(a). Because Claims 22-29 depend upon Claim 21, these claims are also rejected under 35 U.S.C. 112(a).
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 21-29 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Claims 21-29 are directed to a machine. As such, each claim is directed to a statutory category of invention.
Step 2A Prong 1
Independent Claim 21 recites the following abstract ideas: “generate a networking recommendation based on loss transmitted from a client, comprising: a connection identifier ; the connection identifier having access to a plurality of user accounts, each of the plurality of user accounts comprising: a user profile; a network strategy comprising at least one of current connections, candidate connections, network objectives, networking tasks, networking events, networking introductions, candidate introductions, and an ability to invite users ; and a plurality of primary connections, each primary connection of the plurality of primary connections defining a primary relationship between a given user account and a different user account of the plurality of user accounts; determine whether user accounts of the plurality of user accounts would be a good networking match by: the user profile, the network strategy, and the plurality of primary connections of a target user account of the plurality of user accounts, and generate a target output comprising a semantic representation corresponding to text of the target user account; the user profile, the network strategy, and the plurality of primary connections of a second user account of the plurality of user accounts, and generate a candidate output comprising a semantic representation corresponding to text of the second user account; determining a matching score based on comparing the semantic representation of the target output and the semantic representation of the candidate output, the matching score representing a likelihood that a user associated with the target user account and a user associated with the second user account would benefit from establishing a primary relationship based on a network objective of the target user account and skills and information of the second user account; and generating a networking recommendation based on determining that the matching score satisfies a matching score threshold; improve accuracy of the networking recommendation by optimizing the matching score to identify second user accounts having skills and information aligned with the network objective of the target user account, the improvement based on comparing the target output to the candidate output; transmit the networking recommendation to the target user account; receive, from the client, a loss indicating a value of the networking recommendation, the value rating how beneficial the networking recommendation was for the target user account on a numerical scale
The limitations, as drafted, are a process that, under its broadest reasonable interpretation, relates to managing relationships or interactions between people including social activities (i.e., generate a networking recommendation based on loss transmitted from a client, comprising: a connection identifier; the connection identifier having access to a plurality of user accounts, each of the plurality of user accounts comprising: a user profile; a network strategy comprising at least one of current connections, candidate connections, network objectives, networking tasks, networking events, networking introductions, candidate introductions, and an ability to invite users; and a plurality of primary connections, each primary connection of the plurality of primary connections defining a primary relationship between a given user account and a different user account of the plurality of user accounts; determine whether user accounts of the plurality of user accounts would be a good networking match by: the user profile, the network strategy, and the plurality of primary connections of a target user account of the plurality of user accounts, and generate a target output comprising a semantic representation corresponding to text of the target user account; the user profile, the network strategy, and the plurality of primary connections of a second user account of the plurality of user accounts, and generate a candidate output comprising a semantic representation corresponding to text of the second user account; determining a matching score based on comparing the semantic representation of the target output and the semantic representation of the candidate output, the matching score representing a likelihood that a user associated with the target user account and a user associated with the second user account would benefit from establishing a primary relationship based on a network objective of the target user account and skills and information of the second user account; and generating a networking recommendation based on determining that the matching score satisfies a matching score threshold; improve accuracy of the networking recommendation by optimizing the matching score to identify second user accounts having skills and information aligned with the network objective of the target user account, the improvement based on comparing the target output to the candidate output; transmit the networking recommendation to the target user account; receive, from the client, a loss indicating a value of the networking recommendation, the value rating how beneficial the networking recommendation was for the target user account on a numerical scale), but for the recitation of generic computer components (i.e., a system comprising one or more processors and one or more non-transitory computer-readable media storing instructions, a networking application for training a neural network model, executing at a cloud computing environment, and in communication with respective client-side user devices, a network, applying a natural language processing (NLP) module, generate NLP output, and train the neural network model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss). If a claim limitation, under its broadest reasonable interpretation, relates to managing relationships or interactions between people including social activities, but for the recitation of generic computer components, then it falls within the “Certain Methods of Organizing Human Activity” grouping of abstract ideas.
Accordingly, the claim recites an abstract idea.
Step 2A Prong 2
This judicial exception is not integrated into a practical application. Limitations that are not indicative of integration into a practical application include: (1) Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (MPEP 2106.05(f)), (2) Adding insignificant extra-solution activity to the judicial exception (MPEP 2106.05(g)), (3) Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)). In particular, the claim recites the additional elements of a system comprising one or more processors and one or more non-transitory computer-readable media storing instructions, a networking application for training a neural network model, executing at a cloud computing environment, and in communication with respective client-side user devices, a network, applying a natural language processing (NLP) module, generate NLP output, and train the neural network model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss. The computer hardware is recited at a high level of generality (i.e., generic computer, cloud-based computer application, and user devices receiving and transmitting information, and generic NLP module and neural network trained, based on user input, to determine and output information) such that it amounts to no more than mere instructions to apply the exception using generic computer components. Accordingly, these additional elements, when considered separately and as an ordered combination, do not integrate the abstract idea into a practical application, since they do not involve improvements to the functioning of a computer or to any other technology or technical field (MPEP 2106.05(a)), they do not apply the abstract idea with, or by use of, a particular machine (MPEP 2106.05(b)), they do not effect a transformation or reduction of a particular article to a different state or thing (MPEP 2106.05(c)), and they do not apply or use the abstract idea in some other meaningful way beyond generally linking its use to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception (MPEP 2106.05(e)). Therefore, the claim is directed to an abstract idea without a practical application.
Step 2B
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception because, when considered separately and as an ordered combination, they do not add significantly more (also known as an “inventive concept”) to the exception. The additional elements of using computer hardware (a system comprising one or more processors and one or more non-transitory computer-readable media storing instructions, a networking application for training a neural network model, executing at a cloud computing environment, and in communication with respective client-side user devices, a network, applying a natural language processing (NLP) module, generate NLP output, and train the neural network model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss) amounts to no more than mere instructions to apply the exception using generic computer components. Mere instructions to apply an exception using generic computer components cannot provide an inventive concept. Therefore, the claim is not patent-eligible.
Dependent claims 22-29 do not include any additional elements beyond those identified above. They further define the abstract idea that is present in their respective independent claims and hence are abstract for at least the reasons presented above. As such, they do not integrate the abstract idea into a practical application, nor are they sufficient to amount to significantly more than the abstract idea when considered both individually and as an ordered combination. Thus, the aforementioned claims are not patent-eligible.
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 21-27 and 29 are rejected under 35 U.S.C. 103 as being unpatentable over Han et al. (US-20160283500) in view of Poslavsky (US-20210182976), Yang et al. (US-20180107665), and Chinen et al. (US-20230099526).
Claim 21
Han teaches the following limitations:
A system comprising one or more processors and one or more non-transitory computer-readable media storing instructions that, when executed by the one or more processors, cause the one or more processors to implement a networking application for training a neural network model to generate a networking recommendation… the system comprising: the networking application executing at a cloud computing environment and comprising a connection identifier and the neural network model, the networking application at the cloud computing environment being in communication with respective client-side user devices of respective users of the networking application via a network ([0011] Social network application 122 is a software application providing a platform to a user to build social networks and social relationships among people who share interests, activities, backgrounds, or real-life connections. Social network application 122 can be a web-based service that allows a user to create a public profile, create a list of other users of a social network with whom to share connections, and to interact with the other users. A social network connection is a relationship between two users of a social network system, the connection allowing the users to share ideas, interests, and other items. The user created public profile may contain profile information such as identifying information, current activities, background information, and interests. Social network application 122 is a client-side application operating on user computing device 120, and allowing a user of user computing device 120 access to other users of a social network system via network 110; [0013] recommendation module 132 determines a mining engine with which to evaluate the user… Based on the mining engine evaluation, recommendation module 132 identifies one or more social network connections to the user. In an embodiment, recommendation module 132 is a plugin or an add-on to social network application 122; [0024] The text mining engine uses the data with supervised learning methods to train a model, for example, a decision tree, a deep neural network (DNN), etc.);
the connection identifier having access to a plurality of user accounts, each of the plurality of user accounts comprising: a user profile ([0003] a method, a computer program product, and a computer system for recommending one or more connections in a social network system. In the method, a computer retrieves user profile information for a user of a social network system; [0011] Social network application 122 is a software application providing a platform to a user to build social networks and social relationships among people who share interests, activities, backgrounds, or real-life connections. Social network application 122 can be a web-based service that allows a user to create a public profile, create a list of other users of a social network with whom to share connections, and to interact with the other users. A social network connection is a relationship between two users of a social network system, the connection allowing the users to share ideas, interests, and other items. The user created public profile may contain profile information such as identifying information, current activities, background information, and interests);
a network strategy comprising at least one of current connections, candidate connections, network objectives, networking tasks, networking events, networking introductions, candidate introductions, and an ability to invite users to the networking application ([0011] Social network application 122 is a software application providing a platform to a user to build social networks and social relationships among people who share interests, activities, backgrounds, or real-life connections. Social network application 122 can be a web-based service that allows a user to create a public profile, create a list of other users of a social network with whom to share connections, and to interact with the other users. A social network connection is a relationship between two users of a social network system, the connection allowing the users to share ideas, interests, and other items. The user created public profile may contain profile information such as identifying information, current activities, background information, and interests; [0016] When initialized, recommendation module 132 retrieves user profile information from the user's profile or a user's personal page. In various embodiments, user profile information may include, for example, user identifying information, such as name, location, career position, and company or business unit, user status updates, user comments, including either comments on user shared items or comments on items shared by other users, and user shared items or photos. User shared items may include, for example, news articles, blog posts, website links, restaurant or theater reviews, and other such items. User profile information may include user interests or activities, or information on the user's connections, for example, who the user is connected to and to how many other users the user is connected); and
a plurality of primary connections, each primary connection of the plurality of primary connections defining a primary relationship between a given user account and a different user account of the plurality of user accounts ([0011] Social network application 122 can be a web-based service that allows a user to create a public profile, create a list of other users of a social network with whom to share connections, and to interact with the other users. A social network connection is a relationship between two users of a social network system, the connection allowing the users to share ideas, interests, and other items; [0016] User profile information may include user interests or activities, or information on the user's connections, for example, who the user is connected to and to how many other users the user is connected);
determine whether user accounts of the plurality of user accounts would be a good networking match by: applying a… module to the user profile, the network strategy, and the plurality of primary connections of a target user account of the plurality of user accounts, and generate a target… output ([0013] Recommendation module 132 evaluates a user's profile and other social network information to determine a social maturity level, or stage, of the user, the social maturity stage indicating whether the user is a new user, for example, one with few, if any, connections, an intermediate user, for example, one with many connections but each in the same business unit or location, or an experienced user, for example, one with many connections across business unit, country, age range, etc. Based on the determined social maturity value and user stage, recommendation module 132 determines a mining engine with which to evaluate the user. Each mining engine utilized by recommendation module 132 retrieves a plurality of information, for example, a profile mining engine retrieves structured, basic, profile information of the user. Based on the mining engine evaluation, recommendation module 132 identifies one or more social network connections to the user; [0016] Recommendation module 132 retrieves user profile information (202); [0017] Recommendation module 132 determines a user stage (204). Recommendation module 132 evaluates the retrieved user profile information to determine to which stage the user belongs, where the user stage may also be referred to as the user's social maturity. In an embodiment, recommendation module 132 groups the user into one of three stages based on a number of current connections, and a location of each current connection, the location either a physical location or a business organization location, or status)
applying the… module to the user profile, the network strategy, and the plurality of primary connections of a second user account of the plurality of user accounts, and generate a candidate… output ([0022] Recommendation module 132 determines a mining engine (206), for each stage, for evaluating a user. In embodiments of the present invention, recommendation module 132 identifies a profile mining engine for users in the first stage, such profile mining engine evaluating structured user profile information from the user's profile, and determining other users with matching, or similar, profile information, for example, business unit, team, organization, etc. For users in the second stage, recommendation module 132 identifies a network mining engine, the network mining engine retrieving one or more connections with connections in common with the user, using, for example, a contact list of the user. Recommendation module 132 identifies a text mining engine for users in the third stage, which retrieves text from other users and determines other users with similar interests and activities as the user; [0024] Recommendation module 132 performs operations according to the determined mining engine (208). In various embodiments, each mining engine identified is used to extract information from the retrieved user profile information, in order to determine one or more connections for the user. In an embodiment, the profile mining engine evaluates structured user profile information from the user's profile, and determines other users with matching, or similar, profile information. The structured profile information can be stored in database 134. In an embodiment, the network mining engine retrieves potential connections via the user's contact list, using one of a plurality of network mining methods, such as collaborative filtering, to find a second user with a maximum connections in common with the user. In an embodiment, the text mining engine identified for users in the third stage retrieves and collects a corpus of data from other users of the social network system, including, for example, status updates, user comments, user shared items, and communities in which the other user may be involved)
determining a matching… based on comparing the target… output and the candidate… output, the matching… representing a likelihood that a user associated with the target user account and a user associated with the second user account would benefit from establishing a primary relationship based on a network objective of the target user account and skills and information of the second user account ([0025] Recommendation module 132 determines whether at least one social network connection is identified (decision step 210) … In some embodiments, recommendation module 132 may identify one or more social network connections for the user, and may determine to send one, or several, of the identified connections. Recommendation module 132 may rank the one or more connections, based on various criteria, including, for example, a closeness in location, a number of connections in common over a threshold number, or a strong similarity in interests versus a lower similarity in interests); and
generating a networking recommendation based on determining that the matching… satisfies a… threshold ([0025] Recommendation module 132 determines whether at least one social network connection is identified (decision step 210). If at least one social network connection is identified (decision step 210, “yes” branch), recommendation module 132 sends the at least one recommended social network connection to the user (212). Recommendation module 132, when the recommended connection is identified, sends the recommendation to the user, for example, as a message or alert in social network application 122. The recommendation may include a name of another user, or some other identifying information. In various embodiments, recommendation module 132 includes a list of reasons why the connection is recommended, for example, similar interests, connections in common, or similar location. In some embodiments, recommendation module 132 may identify one or more social network connections for the user, and may determine to send one, or several, of the identified connections. Recommendation module 132 may rank the one or more connections, based on various criteria, including, for example, a closeness in location, a number of connections in common over a threshold number, or a strong similarity in interests versus a lower similarity in interests);
the neural network model further trained to… identify second user accounts having skills and information aligned with the network objective of the target user account, the improvement based on comparing the target… output to the candidate… output ([0024] Recommendation module 132 performs operations according to the determined mining engine (208). In various embodiments, each mining engine identified is used to extract information from the retrieved user profile information, in order to determine one or more connections for the user. In an embodiment, the profile mining engine evaluates structured user profile information from the user's profile, and determines other users with matching, or similar, profile information. The structured profile information can be stored in database 134. In an embodiment, the network mining engine retrieves potential connections via the user's contact list, using one of a plurality of network mining methods, such as collaborative filtering, to find a second user with a maximum connections in common with the user. In an embodiment, the text mining engine identified for users in the third stage retrieves and collects a corpus of data from other users of the social network system, including, for example, status updates, user comments, user shared items, and communities in which the other user may be involved. The text mining engine uses the data with supervised learning methods to train a model, for example, a decision tree, a deep neural network (DNN), etc. Recommendation module 132, via the text mining engine, uses the model to predict a user's interests, given the user's information, where the model is based on a plurality of other users' data);
the networking application at the cloud computing environment configured to transmit the networking recommendation to a client-side user device associated with the target user account ([0025] If at least one social network connection is identified (decision step 210, “yes” branch), recommendation module 132 sends the at least one recommended social network connection to the user (212));
the networking recommendation ([0025] If at least one social network connection is identified (decision step 210, “yes” branch), recommendation module 132 sends the at least one recommended social network connection to the user (212)); and
the networking application at the cloud computing environment configured to train the neural network model ([0024] The text mining engine uses the data with supervised learning methods to train a model, for example, a decision tree, a deep neural network (DNN), etc.)
However, Han does not explicitly teach the following limitations:
application for… training a… model… based on loss transmitted from a client-side user device,
the neural network model trained to determine whether user accounts of the plurality of user accounts would be a good networking match by: applying a natural language processing (NLP) module to… a target user account of the plurality of user accounts, and generate a target NLP output comprising a semantic representation corresponding to text of the target user account;
applying the NLP module to… a second user account of the plurality of user accounts, and generate a candidate NLP output comprising a semantic representation corresponding to text of the second user account;
determining a matching score based on comparing the semantic representation of the target NLP output and the semantic representation of the candidate NLP output, the matching score representing a likelihood that a user associated with the target user account and a user associated with the second user account would benefit from establishing a primary relationship
generating a networking recommendation based on determining that the matching score satisfies a matching score threshold;
the neural network model further trained to improve accuracy of the networking recommendation by optimizing the matching score… based on comparing the target NLP output to the candidate NLP output;
the networking application at the cloud computing environment configured to receive, from the client-side user device, a loss indicating a value of the… recommendation, the value rating how beneficial the networking recommendation was for the target user account on a numerical scale; and
the networking application at the cloud computing environment configured to train the… model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss.
Poslavsky, in the same field of endeavor, teaches the following limitations:
the neural network model trained to determine whether user accounts of the plurality of user accounts would be a good networking match by: applying a natural language processing (NLP) module to… a target user account of the plurality of user accounts, and generate a target NLP output comprising a semantic representation corresponding to text of the target user account ([0077] A user Z of a social media network is planning an upcoming trip to Paris. Z searches for things to do in Paris, as well as airfare to get there. An artificial intelligence algorithm determines from these indications that Z is planning an upcoming trip and searches Z's friends on the social media network for someone who lives in Paris. When no friends were found to live in Paris, the artificial intelligence algorithm searches the friends of Z's friends, and locates two people who live in Paris, X and Y. The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network. Further, the artificial intelligence algorithm uses a computer vision algorithm to process photographs posted by Z, X, and Y to the social media network, where the processing determines a similarity of the activities Z, X, and Y are performing in the photographs. The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking; [0022] The term “artificial intelligence,” as used herein, generally refers to machine intelligence that includes a computer model or algorithm that may be used to make a recommendation or prediction, classify data, or otherwise take an action that maximizes the chance of achieving of one or more goals of the artificial intelligence. Artificial intelligence may be or include a machine learning algorithm. The machine learning algorithm may be a trained machine learning algorithm, e.g. a machine learning algorithm trained on historical data. Such a trained machine learning algorithm may be trained using supervised, semi-supervised, or unsupervised learning process. Examples of machine learning algorithms include neural networks, support vector machines, and reinforcement learning algorithms);
applying the NLP module to… a second user account of the plurality of user accounts, and generate a candidate NLP output comprising a semantic representation corresponding to text of the second user account ([0077] A user Z of a social media network is planning an upcoming trip to Paris. Z searches for things to do in Paris, as well as airfare to get there. An artificial intelligence algorithm determines from these indications that Z is planning an upcoming trip and searches Z's friends on the social media network for someone who lives in Paris. When no friends were found to live in Paris, the artificial intelligence algorithm searches the friends of Z's friends, and locates two people who live in Paris, X and Y. The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network. Further, the artificial intelligence algorithm uses a computer vision algorithm to process photographs posted by Z, X, and Y to the social media network, where the processing determines a similarity of the activities Z, X, and Y are performing in the photographs. The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking; [0022] The term “artificial intelligence,” as used herein, generally refers to machine intelligence that includes a computer model or algorithm that may be used to make a recommendation or prediction, classify data, or otherwise take an action that maximizes the chance of achieving of one or more goals of the artificial intelligence. Artificial intelligence may be or include a machine learning algorithm. The machine learning algorithm may be a trained machine learning algorithm, e.g. a machine learning algorithm trained on historical data. Such a trained machine learning algorithm may be trained using supervised, semi-supervised, or unsupervised learning process. Examples of machine learning algorithms include neural networks, support vector machines, and reinforcement learning algorithms);
determining a matching score based on comparing the semantic representation of the target NLP output and the semantic representation of the candidate NLP output, the matching score representing a likelihood that a user associated with the target user account and a user associated with the second user account would benefit from establishing a primary relationship ([0077] Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking. Based on the weights the artificial intelligence algorithm has given the various data, the artificial intelligence algorithm determines that Z and Y are more compatible than Z and X); and
generating a networking recommendation based on determining that the matching score satisfies a matching score threshold ([0047] The connecting may comprise using an artificial intelligence algorithm to determine a compatibility of the member and the user. The artificial intelligence algorithm may connect the user and the member if the compatibility is above a threshold value);
the neural network model further trained to improve accuracy of the networking recommendation by optimizing the matching score… based on comparing the target NLP output to the candidate NLP output ([0077] A user Z of a social media network is planning an upcoming trip to Paris. Z searches for things to do in Paris, as well as airfare to get there. An artificial intelligence algorithm determines from these indications that Z is planning an upcoming trip and searches Z's friends on the social media network for someone who lives in Paris. When no friends were found to live in Paris, the artificial intelligence algorithm searches the friends of Z's friends, and locates two people who live in Paris, X and Y. The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network. Further, the artificial intelligence algorithm uses a computer vision algorithm to process photographs posted by Z, X, and Y to the social media network, where the processing determines a similarity of the activities Z, X, and Y are performing in the photographs. The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking; [0022] The term “artificial intelligence,” as used herein, generally refers to machine intelligence that includes a computer model or algorithm that may be used to make a recommendation or prediction, classify data, or otherwise take an action that maximizes the chance of achieving of one or more goals of the artificial intelligence. Artificial intelligence may be or include a machine learning algorithm. The machine learning algorithm may be a trained machine learning algorithm, e.g. a machine learning algorithm trained on historical data. Such a trained machine learning algorithm may be trained using supervised, semi-supervised, or unsupervised learning process. Examples of machine learning algorithms include neural networks, support vector machines, and reinforcement learning algorithms);
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., using NLP to process information, such as users’ information to match users together, and performing an action, such as connecting users together, based on a score being above a threshold) into similar systems.
However, Han, in combination with Poslavsky, does not explicitly teach the following limitations:
application for… training a… model… based on loss transmitted from a client-side user device,
the networking application at the cloud computing environment configured to receive, from the client-side user device, a loss indicating a value of the… recommendation, the value rating how beneficial the networking recommendation was for the target user account on a numerical scale; and
the networking application at the cloud computing environment configured to train the… model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss.
Yang, in the same field of endeavor, teaches the following limitations:
application for… training a… model… based on loss transmitted from a client-side user device ([0036] The potential recommendation identification module 204 can apply the trained machine learning model to determine potential recommendations for a page... an administrator can provide feedback relating to a recommendation presented to the administrator. Feedback by administrators can be used to train or retrain the machine learning model for determining potential recommendations),
the networking application at the cloud computing environment configured to receive, from the client-side user device, a loss indicating a value of the… recommendation, the value rating how beneficial the networking recommendation was for the target user account ([0043] The section 320 can include a mechanism for an administrator to provide feedback regarding recommendations. For example, the section 320 can display a question 327 “Is this helpful” next to a recommendation, and the administrator can click “Yes” or “No.” As explained above, feedback from administrators regarding recommendations can be used to train and retrain machine learning models); and
the networking application at the cloud computing environment configured to train the… model based on the loss received from the client-side user device, thereby iteratively improving the neural network model’s ability to generate accurate networking recommendations by adjusting model parameters based on the loss ([0036] an administrator can provide feedback relating to a recommendation presented to the administrator. Feedback by administrators can be used to train or retrain the machine learning model for determining potential recommendations).
This known technique is applicable to the system of Han, in combination with Poslavsky, as they both share characteristics and capabilities, namely, they are directed to social networking systems that provide recommendations using machine learning. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Yang would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Yang to the teachings of Han, in combination with Poslavsky, would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., training a ML model based on user feedback) into similar systems.
However, Han, in combination with Poslavsky and Yang, does not explicitly teach the following limitations:
receive, from the client-side user device, a loss indicating a value… on a numerical scale; and
Chinen, in the same field of endeavor, teaches the following limitations:
receive, from the client-side user device, a loss indicating a value… on a numerical scale ([0044] a set of training data for training one or more machine-learned models according to example aspects of the present disclosure can include data associating video content and/or image content with viewer feedback (e.g., a score for content input by a viewer while or after viewing the content) regarding the perceptual quality of the content. The data can include one or more scores or values associated with the content, including, for example, an MOS or other numerical rating, natural language feedback, a preference indicator (e.g., a value indicating a selection of one content item as preferred over another content item), combinations thereof, or other indicators of an opinion regarding the perceptual quality of the content); and
Chinen shows that a numerical rating was known in the prior art before the effective filing date of the claimed invention. Since each individual element and its function are shown in the prior art, albeit shown in separate references, the difference between the claimed subject matter and the prior art rests not on any individual element or function, but in the very combination itself; that is, in the substitution of the numerical rating of Chinen for the yes/no rating of Yang. Thus, the simple substitution of one known element for another producing a predictable result renders the claim obvious.
Claim 22
Yang further teaches the following limitations:
wherein the value rates how beneficial the networking recommendation was for the target user account on a scale ([0043] The section 320 can include a mechanism for an administrator to provide feedback regarding recommendations. For example, the section 320 can display a question 327 “Is this helpful” next to a recommendation, and the administrator can click “Yes” or “No.” As explained above, feedback from administrators regarding recommendations can be used to train and retrain machine learning models).
This known technique is applicable to the system of Han, in combination with Poslavsky, as they both share characteristics and capabilities, namely, they are directed to social networking systems that provide recommendations using machine learning. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Yang would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Yang to the teachings of Han, in combination with Poslavsky, would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., providing user feedback on a recommendation in a certain form, such as a rating scale) into similar systems.
Claim 23
Han further teaches the following limitations:
wherein the neural network model is trained to generate the networking recommendation based on text corresponding to the target user account and text corresponding to the second user account ([0024] Recommendation module 132 performs operations according to the determined mining engine (208). In various embodiments, each mining engine identified is used to extract information from the retrieved user profile information, in order to determine one or more connections for the user. In an embodiment, the profile mining engine evaluates structured user profile information from the user's profile, and determines other users with matching, or similar, profile information. The structured profile information can be stored in database 134. In an embodiment, the network mining engine retrieves potential connections via the user's contact list, using one of a plurality of network mining methods, such as collaborative filtering, to find a second user with a maximum connections in common with the user. In an embodiment, the text mining engine identified for users in the third stage retrieves and collects a corpus of data from other users of the social network system, including, for example, status updates, user comments, user shared items, and communities in which the other user may be involved. The text mining engine uses the data with supervised learning methods to train a model, for example, a decision tree, a deep neural network (DNN), etc. Recommendation module 132, via the text mining engine, uses the model to predict a user's interests, given the user's information, where the model is based on a plurality of other users' data; [0025] If at least one social network connection is identified (decision step 210, “yes” branch), recommendation module 132 sends the at least one recommended social network connection to the user (212)).
Claim 24
Poslavsky further teaches the following limitations:
wherein the target NLP output provides an understanding corresponding to the text corresponding to the target user account, wherein the candidate NLP output provides an understanding corresponding to the text corresponding to the second user account ([0077] The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network… The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking).
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., NLP output providing an understanding corresponding to text) into similar systems.
Claim 25
Poslavsky further teaches the following limitations:
wherein the understanding is a semantic representation ([0077] The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network… The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking).
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., NLP output providing an understanding corresponding to text, the understanding being a semantic representation) into similar systems.
Claim 26
Poslavsky further teaches the following limitations:
wherein the neural network model is trained to apply the NLP module to respective portions of the text corresponding to the target user account to generate multiple target NLP outputs, wherein the neural network model is trained to apply the NLP module to respective portions of the text corresponding to the second user account to generate multiple candidate NLP outputs ([0077] The artificial intelligence algorithm next compares Z's interests with those of X and Y by using a natural language processing (NLP) algorithm to read Z, X, and Y's posts on the social media network… The artificial intelligence algorithm then searches for similarities between Z, X, and Y's answers to various surveys and quizzes each have taken on the social media network. Aggregating all of this data, the artificial intelligence algorithm determines that Z and X both enjoy hiking and classical music but differ fundamentally on politics, while Z and Y have similar pollical leanings, as well as both enjoying cooking).
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., generating multiple NLP outputs) into similar systems.
Claim 27
Poslavsky further teaches the following limitations:
wherein the neural network model is trained to generate the networking recommendation to initiate an introduction between a user associated with the target user account and a user associated with the second user account ([0047] The connecting may comprise using an artificial intelligence algorithm to determine a compatibility of the member and the user. The artificial intelligence algorithm may connect the user and the member if the compatibility is above a threshold value; [0022] Artificial intelligence may be or include a machine learning algorithm. The machine learning algorithm may be a trained machine learning algorithm… Examples of machine learning algorithms include neural networks).
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., generating a networking recommendation to initiate an introduction) into similar systems.
Claim 29
Poslavsky further teaches the following limitations:
wherein the matching score represents a similarity from a network objective of the target user account and skills and information of the second user account ([0009] a method for connecting users of a social network, comprising: (a) receiving, from the social network, an indication that a first user intends to travel to a location; (b) determining that a second user of the social network lives in the location, wherein the first user and the second user are associated with each other on the social network (i) directly or (ii) through one or more other users of the social network, and wherein the second user has offered a service on the social network; and (c) connecting the first user and the second user on the social network to enable the second user to perform the service for the first user; [0047] The connecting may comprise using an artificial intelligence algorithm to determine a compatibility of the member and the user. The artificial intelligence algorithm may connect the user and the member if the compatibility is above a threshold value).
This known technique is applicable to the system of Han as they both share characteristics and capabilities, namely, they are directed to matching / connecting users, through machine learning, based on comparing their profile data. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Poslavsky would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Poslavsky to the teachings of Han would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., a matching score that represents similarity between matched users) into similar systems.
Claim 28 is rejected under 35 U.S.C. 103 as being unpatentable over Han et al. (US-20160283500) in view of Poslavsky (US-20210182976), Yang et al. (US-20180107665), and Chinen et al. (US-20230099526), and further in view of Nigam et al. (US-20220101159).
Claim 28
Han, in combination with Poslavsky, Yang, and Chinen, does not explicitly teach the following limitations:
wherein the matching score represents a likelihood that the user associated with the target user account and the user associated with the second user account would like to establish a primary relationship with one another.
Nigam, in the same field of endeavor, teaches the following limitations:
wherein the matching score represents a likelihood that the user associated with the target user account and the user associated with the second user account would like to establish a primary relationship with one another ([0051] FIG. 4 illustrates the architecture for identifying possible new connections for a target entity; [0053] The training of the additional machine-learned models includes determining one or more additional sets of features relevant to determining second-pass ranking scores for each of the candidate connections. For example, the additional machine-learned models 413 can be trained to provide adjustments to each of the first-pass ranking scores based on various factors, such as a probability that a connectee will attempt to establish the recommended connection with a connector, a probability that a connector will accept a request to establish the connection from the connectee).
This known technique is applicable to the system of Han, in combination with Poslavsky, Yang, and Chinen, as they both share characteristics and capabilities, namely, they are directed to providing recommendations for matching / connecting users. One of ordinary skill in the art, before the effective filing date of the claimed invention, would have recognized that applying the known technique of Nigam would have yielded predictable results and resulted in an improved system. It would have been recognized that applying the technique of Nigam to the teachings of Han, in combination with Poslavsky, Yang, and Chinen, would have yielded predictable results because the level of one of ordinary skill in the art would have known to incorporate such features (i.e., a matching score that represents probability that users will match) into similar systems.
Response to Arguments
Applicant’s Argument Regarding 35 USC 112(a) Rejections of Claims 21-29: Applicant has amended Claim 21.
Examiner’s Response: Applicant’s amendments have been fully considered and they resolve the identified issue. However, the amended claim has raised new 112(a) issues. Refer to the rejection above.
Applicant’s Argument Regarding 35 USC 101 Rejection of Claims 21-29:
Step 2A, Prong One:
Properly viewed as an ordered combination, as required by McRO, Inc. V. Bandai Namco Games Am. Inc., 837 F.3d 1299, 1313 (Fed. Cir. 2016), the amended claims recite a specific technical implementation of neural network training with NLP-based semantic analysis. The claims do not merely recite the concept of introducing people. Rather, they recite: (i) a neural network model that applies an NLP module to user profiles, network strategies, and primary connections to generate semantic representations; (ii) a comparer that determines a matching score by comparing those semantic representations; (iii) a feedback loop in which loss values (i.e., numerical ratings of recommendation quality transmitted from client-side user devices) are used to retrain the neural network model; and (iv) iterative improvement of the neural network's matching accuracy through parameter adjustments based on the received loss.
This is analogous to the claims in PEG Example 39, which were found not to recite a judicial exception because they were directed to training a neural network, which is a fundamentally technical process. The present claims similarly recite the technical process of training and iteratively improving a neural network's ability to generate semantically informed matching scores.
Step 2A, Prong Two:
The precedential decision in Ex Parte Desjardins, Appeal No. 2024-000567 (PTAB Sept. 26, 2025) (ARP Decision), and the corresponding USPTO Memorandum dated December 5, 2025, are directly instructive. In Desjardins, the Appeals Review Panel found that claims directed to training a machine learning model integrated an abstract idea (a mathematical concept) into a practical application because the specification identified improvements to how the machine learning model itself operates. Specifically, the ARP credited benefits including reduced storage, reduced system complexity, and preservation of performance attributes as technological improvements. The December 5, 2025 Memorandum from Deputy Commissioner Charles Kim instructs that the MPEP § 2106.05(a), subsection I, now includes as examples of improvements to computer functionality: (xiii) "An improved way of training a machine learning model that protected the model's knowledge about previous tasks while allowing it to effectively learn new tasks;" and (xiv) "Improvements to computer component or system performance based upon adjustments to parameters of a machine learning model."
The present claims fall squarely within this guidance. Like Desjardins, the amended claims recite a specific improvement to how a neural network model operates:
First, the neural network model is trained using NLP-derived semantic representations, not raw text or generic features. The NLP module applies natural language processing to user profiles, network strategies, and primary connections to generate semantic representations that enable the neural network to understand the substantive meaning of user data, not merely its surface-level text. This is a specific technique that improves the neural network's ability to determine whether user accounts would be a good match, as demonstrated by the specification's examples of correctly matching a user who wants to "learn how to fly a plane" with an "airline pilot" while rejecting a "fly fisherman." See [0046]-[0047].
Second, the claims recite a closed-loop training mechanism in which loss values (quantitative ratings on a numerical scale) are transmitted from client-side user devices back to the cloud computing environment, where they are used to retrain the neural network model. This is not binary feedback (e.g., "yes" or "no"); it is a quantitative loss signal that enables gradient- based or analogous parameter optimization. The specification at [0027] discloses that the loss 118 is a numerical value (e.g., "4" on a scale of 1-5) and that the networking application "may use the loss 118 to train the neural network 400." This feedback loop constitutes a specific, concrete improvement to the neural network training process.
Third, the claims recite that this training results in iterative improvement to the neural network model's accuracy in generating networking recommendations "by adjusting model parameters based on the loss." This is a technological improvement to the functioning of the neural network itself, not merely an improvement to the abstract idea of matching people.
The Examiner previously stated that "Desjardins recites an improvement to ML technology while the present claims only improve the abstract idea itself." Office Action at 40. However, the amended claims now explicitly recite the technical mechanism by which the neural network model is improved: semantic representations generated by NLP, quantitative loss-based feedback from user devices, and parameter adjustment to iteratively improve model accuracy. These are improvements to the machine learning model's training process itself. This is precisely the type of improvement recognized in Desjardins.
Moreover, the Desjardins Memorandum instructs examiners that "[w]hen evaluating a claim as a whole, examiners should not dismiss additional elements as mere ‘generic computer components’ without considering whether such elements confer a technological improvement to a technical problem." The Memorandum further instructs that "[e]xaminers and panels should not evaluate claims at such a high level of generality" that potentially meaningful technical limitations are dismissed without adequate explanation. The specific combination of NLP-based semantic analysis, matching score computation, and loss-based iterative retraining recited in the amended claims constitutes a meaningful technical improvement that should not be dismissed as generic.
Additionally, the Federal Circuit's reasoning in DDR Holdings, LLC v. Hotels.com, L.P., 773 F.3d 1245 (Fed. Cir. 2014) supports eligibility here. The claimed system addresses a problem particular to computer networking technology: generating accurate networking recommendations in a system with potentially millions of users, using NLP-based semantic analysis to understand the meaning behind user profiles and objectives, and iteratively improving those recommendations through a distributed training architecture where loss values flow from client devices to a cloud-based neural network. This is a solution rooted in computer technology.
Step 2B:
Even under Step 2B, the ordered combination of limitations, including a neural network model trained using NLP-generated semantic representations, comparing those representations to produce a matching score that accounts for network objectives and skills alignment, transmitting recommendations to client-side devices, receiving quantitative loss values from those devices, and retraining the model based on those loss values, amounts to significantly more than any alleged abstract idea. No single reference in the record, alone or in combination, describes this specific technical architecture, and the specification demonstrates that this architecture produces concrete technical benefits in the form of improved recommendation accuracy. See BASCOM Global Internet Servs. v. AT&T Mobility, LLC, 827 F.3d 1341, 1350 (Fed. Cir. 2016) (finding an inventive concept in the particular arrangement of claim limitations even when individual elements were known).
Examiner’s Response: Applicant’s arguments have been fully considered but they are not persuasive.
Step 2A, Prong One:
Though the claims recite additional elements, such as the neural network, the claims are still directed to an abstract idea. The claims recite steps directed to generating networking recommendations by determining a matching score between two user accounts based on comparing data from the profiles of the users, and generating and providing the recommendation if the score satisfies a threshold. The neural network and the training of the neural network is used as a tool to implement the abstract idea. Regarding PEG Example 39, this is not analogous, because though both claims recite training a neural network, Example 39 does not recite any of the judicial exceptions, and is actually directed to the training of the neural network.
Step 2A, Prong Two:
The present claims are not analogous or similar to Desjardins. In Desjardins, the specification identified the technological improvements, including how the machine learning model itself operates, and training a model to learn new tasks while protecting knowledge about previous tasks to overcome the problem of “catastrophic forgetting” encountered in continual learning systems. Further, the claims reflected the improvement disclosed in the specification. On the contrary, the specification of the present application does not disclose any technological problem or technological solution. The NLP module and neural network are utilized in their ordinary capacity. Paragraphs [0046]-[0047] describe using an NLP module to generate NLP outputs, and generating a matching score. The paragraphs do not provide any technical improvement to NLP technology. Regarding the loss values, regardless of whether they are binary inputs or numerical inputs, these are human user inputs that are used to train the neural network. The loss values being inputted by a human user moves the invention further into the abstract idea. Further, the specification does not provide any technical improvement to how neural networks are trained. Regarding Applicant’s argument that “the claims recite that this training results in iterative improvement to the neural network model's accuracy in generating networking recommendations ‘by adjusting model parameters based on the loss,’” the specification does not disclose any details about any iterative improvement in the model’s accuracy or about adjusting model parameters. It simply states that the user inputted loss is used to train the neural network, without any further details. Further, the inclusion of this limitation in the claims is a new matter issue; refer to the 112(a) rejection above.
Step 2B:
The steps of comparing data to produce a matching score that accounts for network objectives and skills alignment, and transmitting recommendations to users are all part of the abstract idea. The training of a neural network using NLP-generated semantic representations, and receiving quantitative loss values from users to train the model does not amount to significantly more than the abstract idea.
Applicant’s Argument Regarding 35 USC 103 Rejections of Claims 21-29:
Claim 21 has been amended, and none of the cited references, individually or in combination, teaches or suggests the amended features.
Regarding Claim 28: Claim 28 recites that "the matching score represents a likelihood that the user associated with the target user account and the user associated with the second user account would like to establish a primary relationship with one another." While Nigam teaches probability-based scoring for predicting connection acceptance, Nigam does not teach this scoring in the context of the specific NLP semantic representation-based matching architecture recited in amended Claim 21. Nigam's probability scores are based on features distinct from the NLP-generated semantic representations of user profiles, network strategies, and primary connections recited in the claims. Accordingly, even if the remaining references were combinable (which Applicant does not concede), adding Nigam does not bridge the gap.
Examiner’s Response: Applicant’s arguments have been considered but are moot in light of the new ground of rejection above.
Regarding Claim 28: Nigam was not relied upon to teach NLP semantic representation-based matching. Poslavsky was used to teach this, and the combination of references, including Nigam, is a reasonable combination.
Conclusion
The prior art made of record and not relied upon, considered pertinent to applicant’s disclosure or directed to the state of art, is listed on the enclosed PTO-892.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KARMA EL-CHANTI whose telephone number is (571)272-3404. The examiner can normally be reached T-Sa 10am-6pm ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Sarah Monfeldt can be reached at (571)270-1833. 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.
/KARMA A EL-CHANTI/Examiner, Art Unit 3629 /SARAH M MONFELDT/Supervisory Patent Examiner, Art Unit 3629
Prosecution Insights