DETAILED ACTION
Acknowledgement
This final office action is in response to the amendment filed on 01/23/2026.
Status of Claims
Claims 1, 10, 12, 13, 14, and 18 have been amended.
Claims 1-20 are now pending.
Response to Arguments
Applicant's arguments filed on 01/23/2026 regarding the 35 U.S.C. 101, 102, and 103 rejections of claims 1-20 have been fully considered. The Applicant argues the following.
(1) As per the 101 rejection, the Applicant argues, in summary, that (i) the claims are not directed to an abstract idea. The amended claims recite a machine-implemented workflow that cannot practically be performed mentally; (ii) the claims as a whole integrate any such concept into a practical application by reciting a specific technological solution implemented through a concrete, technical pipeline for cross-silo enterprise indexing, model training, and inference with objective scoring and explainability; and (iii) the additional elements, considered in ordered combination, provide significantly more than any alleged exception. The two-stage training pipeline, the specific table fields (occurrence count and latest usage), the rarity/commonality and recency weighting, the grounded explanation output are not routine or conventional "generic computer implementation" features when claimed in this matter. They materially constrain how the system operates and are not a mere instruction to "use ML".
The Examiner respectfully disagrees. As per argument (i), the Examiner submits that the amended claims are directed to the abstract groupings of both Mental Processes and Certain Methods of Organizing Human Activity because the abstract claim limitations highlighted in Step 2A(1) describe a process of collecting, organizing, and analyzing data in order to fulfill user requests to identify subject matter experts within an organization. Collecting, organizing, and analyzing data can be practically performed mentally via use of mathematical models and with pen and paper. Fulfilling and facilitating user requests to identify and/or connect with other members of an organization reflects managing personal behavior and interactions between people. Per MPEP 2106.04(a), a claim recites a judicial exception when the judicial exception is “set forth” or “described” in the claim. Just because the referenced claim limitations are performed by a computer/machine does not negate the fact that they are still directed to an abstract idea (see MPEP 2106.04(a)(2)).
As per arguments (ii) and (iii), the Examiner submits that the additional elements recited in the claims and highlighted in Steps 2A(2) and 2B when considered as a whole do not integrate the abstract idea into a practical application nor provide significantly more because the additional elements to do improve the functioning of a computer, improve another technology, nor provide a technical solution to a technical problem. The claims recite specific uses of technology (e.g. ML, LLM, NLP, etc.) to perform an abstract process. However, these specific technologies’ capabilities or functions are not improved or enhanced beyond its original capabilities or functions. For example, the claims do not reflect a new and improved process of training ML models. Therefore, the additional elements are viewed as mere instructions to implement an abstract idea on a computer and indicates a field of use or technological environment in which to apply a judicial exception. The improvement as argued by the Applicant is in enterprise information retrieval and expertise location by reducing noisy or ungrounded results and enabling consistent scoring and explainability based on structured evidence signals, which is considered abstract. An improvement in the judicial exception itself is not an improvement in technology (MPEP 2106.05(a)(II). Therefore, the 35 U.S.C. 101 rejection is maintained.
(2) As per the 102 and 103 rejections, the Applicant argues that Yin does not disclose the amended two-stage training pipeline with significance/commonality analysis. Yin does not disclose a latest usage column and recency-based weighting as claimed. Yin does not disclose an output explanation grounded in occurrence count and recency for each identified member.
The Examiner finds the Applicant’s argument somewhat persuasive. Yin teaches multiple stages of training a ML models (e.g. [0043]-[0044]) and using recency-based weighting to determine a member’s expertise on a topic (Fig. 5B). However, Yin does not explicitly teach a latest usage date. Therefore, the previous 102 and 103 rejections are withdrawn. However, upon further search and consideration, a new ground of 103 is made. See details below.
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 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, “Machine Learning Model for Identifying Expertise Within an Organization”, is directed to an abstract idea, specifically Mental Processes and Certain Methods of Organizing Human Activity, without significantly more. The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception because the additional elements individually or in combination provide mere instructions to implement the abstract idea on a computer.
Step 1: Claims 1-20 are directed to a statutory category, namely a manufacture (claims 1-13), a system (claims 14-17), and a process (claims 18-20).
Step 2A (1): Independent claims 1, 14, and 18 are directed to an abstract idea of Mental Processes and Certain Methods of Organizing Human Activity, based on the following claim limitations: “gathering digital data from multiple sources within an organization; indexing the digital data in a table that includes at least a first column including names of a plurality of members of the organization and a second column including keywords attributed to the plurality of members; and … assigning weights to the keywords, and storing the weights in a third column in the table, … includes an initial crawling procedure that (i) adds the names of the plurality of members into the table and (ii) attributes the keywords used throughout digital records to respective members … determines whether the keywords attributed to each member are used in a significant or meaningful way and that determines a commonality of keywords such that keywords used by a smaller number of people are weighted more heavily than keywords used by a larger number of people and wherein the table further includes a latest usage column storing a last time that a respective member used a respective keyword and the weights are based at least in part on the latest usage; wherein, in response to receiving an inquiry from a user, … use information in the table to provide an output to the user identifying a member in the organization who demonstrates expertise on a specific topic, wherein the output identifies a plurality of members who demonstrate expertise on the specific topic and provides for each identified member an explanation describing (i) a number of times the member has used one or more keywords corresponding do the specific topic and (ii) a recency of the member’s usage of the one or more keywords.”. These claims describe a process collecting, organizing, and analyzing data in order to fulfill user requests to identify subject matter experts within an organization. Dependent claims 2-13, 15-17, and 19-20 further describe the process of collecting, organizing, and analyzing the data to identify subject matter experts. Collecting, organizing, and analyzing data can be practically performed mentally with pen and paper. Fulfilling and facilitating user requests to identify and/or connect with other members of an organization reflects managing personal behavior and interactions between people. Therefore, these limitations, under the broadest reasonable interpretation, fall within the abstract groupings of Mental Processes which include concepts performed in the human mind such as observations, evaluations, judgments, and opinions and Certain Methods of Organizing Human Activity which encompasses managing personal behavior or relationships or interactions between people including social activities, teaching, and following rules or instructions. Mental Processes include claims directed to collecting information, analyzing it, and displaying certain results of the collection and analysis even if they are claimed as being performed on a computer. The courts have found claims requiring a generic computer or nominally reciting a generic computer may still recite a mental process even though the claim limitations are not performed entirely in the human mind. Certain Methods of Organizing Human Activity can encompass the activity of a single person (e.g. a person following a set of instructions), activity that involve multiple people (e.g. a commercial interaction), and certain activity between a person and a computer (e.g. a method of anonymous loan shopping). Therefore, claims 1-20 are directed to an abstract idea and are not patent eligible.
Step 2A (2): This judicial exception is not integrated into a practical application. In particular, claims 1-4, 6-7, and 9-20 recite additional elements of “a non-transitory computer-readable medium for storing computer logic having instructions that enable a processing device to perform steps; training a Machine Learning (ML) model by data crawling through the digital data… wherein training further includes an in-depth training procedure; the ML model is configured during inference (claims 1, 14, and 18); wherein training the ML model includes a deep learning neural network process (claim 2); wherein the deep learning neural network process involves a Large Language Model (LLM) (claim 3); wherein the ML model includes a chatbot…and wherein the chatbot uses Natural Language Processing (NLP) (claims 4 and 16); a plurality of databases (claim 6); wherein the databases include at least a plurality of data silos each configured to store data in conjunction with use of one or more of collaboration tools, wiki tools, file sharing tools, messaging or chat tools, project management tools, and issue tracking tools associated with multiple different internal groups within the organization (claims 7 and 17); wherein a back-end training component is configured to perform a rudimentary crawling or scraping process to store the digital data in the table, the back-end training component further configured to train the ML model (claim 9); a front-end inference component (claim 10); a user device (claims 10 and 19); utilize the ML model (claims 10 and 19) ; a data warehouse (claims 10 and 19); wherein the back-end training component is configured to receive feedback for retraining the ML model (claim 11); system comprising: a processing device; and memory configured to store computing code having instructions that enable the processing device to (claim 14); and wherein training the ML model includes a Large Language Model (LLM) (claim 15); applying the ML model (claim 20) ”. These additional elements do not integrate the abstract idea into a practical application because the claims do not recite (a) an improvement to another technology or technical field and (b) an improvement to the functioning of the computer itself and (c) implementing the abstract idea with or by use of a particular machine, (d) effecting a particular transformation or reduction of an article, or (e) applying the judicial exception in some other meaningful way beyond generally linking the use of an abstract idea to a particular technological environment. These additional elements evaluated individually and in combination are viewed as computing components and devices that are used to perform the abstract process of collecting, organizing, and analyzing data in order to fulfill user requests to identify subject matter experts within an organization. Limitations that recite mere instructions to implement an abstract idea on a computer or merely uses a computer as a tool to perform an abstract idea are not indicative of integration into a practical application (see MPEP 2106.05(f)). Also, limitations that amount to merely indicating a field of use or technological environment (e.g. machine learning) in which to apply a judicial exception do not amount to significantly more than the exception itself, and cannot integrate a judicial exception into a practical application (see MPEP 2106.05(h)). Therefore, claims 1-20 do not include individual or a combination of additional elements that integrate the judicial exception into a practical application and thus are not patent eligible.
Step 2B: The claims do not include additional elements that are sufficient to amount to significantly more than the judicial exception. Claims 1-4, 6-7, and 9-20 recite additional elements of ““a non-transitory computer-readable medium for storing computer logic having instructions that enable a processing device to perform steps; training a Machine Learning (ML) model by data crawling through the digital data… wherein training further includes an in-depth training procedure; the ML model is configured during inference (claims 1, 14, and 18); wherein training the ML model includes a deep learning neural network process (claim 2); wherein the deep learning neural network process involves a Large Language Model (LLM) (claim 3); wherein the ML model includes a chatbot…and wherein the chatbot uses Natural Language Processing (NLP) (claims 4 and 16); a plurality of databases (claim 6); wherein the databases include at least a plurality of data silos each configured to store data in conjunction with use of one or more of collaboration tools, wiki tools, file sharing tools, messaging or chat tools, project management tools, and issue tracking tools associated with multiple different internal groups within the organization (claims 7 and 17); wherein a back-end training component is configured to perform a rudimentary crawling or scraping process to store the digital data in the table, the back-end training component further configured to train the ML model (claim 9); a front-end inference component (claim 10); a user device (claims 10 and 19); utilize the ML model (claims 10 and 19) ; a data warehouse (claims 10 and 19); wherein the back-end training component is configured to receive feedback for retraining the ML model (claim 11); system comprising: a processing device; and memory configured to store computing code having instructions that enable the processing device to (claim 14); and wherein training the ML model includes a Large Language Model (LLM) (claim 15); applying the ML model (claim 20)”. These additional elements evaluated individually and in combination are viewed as mere instructions to implement an abstract idea on a computer and merely indicates a field of use or technological environment in which to apply a judicial exception. The use of machine learning and trained models are considered instructions to apply or implement a model on a computer. Applying an abstract idea on a computer does not integrate a judicial exception into a practical application or provide an inventive concept (see MPEP 2106.05(f)). Therefore, claims 1-20 do not include individual or a combination of additional elements that are sufficient to amount to significantly more than the judicial exception and thus 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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-2, 5-14, and 17-20 are rejected under 3 35 U.S.C. 103 as being unpatentable over Yin et al. (US 2023/0289732 A1) in view of Brisebois et al. (US 9,317,574 B1).
As per claims 1, 14, and 18 (Currently Amended), Yin teaches a non-transitory computer-readable medium for storing computer logic having instructions that enable a processing device to perform steps of (Yin e.g. In one or more embodiments, a non-transitory computer readable medium may store instructions, that when executed by a hardware processor, cause execution of one or more operations of method 210. According to one or more embodiments, a system may include the non-transitory computer readable medium and the hardware processor [0102].); Yin teaches a system comprising: a processing device; and memory configured to store computing code having instructions that enable the processing device to perform steps of (Yin e.g. FIG. 1 illustrates a system for proposing connections between members of an organization using a machine learning (ML) model [0006]. FIG. 6 shows a block diagram of an example computing system that may implement the features and processes of FIGS. 1-5C [0012].); and Yin teaches a method comprising steps of (Yin e.g. FIG. 2A illustrates an example method for training a ML model to identify interactions between members of an organization [0007]. FIG. 2B illustrates an example method for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0008]. FIG. 3 illustrates an example method for proposing connections between members of an organization using a ML model, in accordance with one or more embodiments [0009].):
Yin teaches gathering digital data from multiple sources within an organization; (Yin e.g. FIG. 2A illustrates an example method 200 for training a ML model to identify interactions between members of an organization, in accordance with one or more embodiments [0074]. In operation 202, the computing device provides a ML engine with a set of interactions. Each interaction is tagged with an interaction type (e.g., email, phone call, social media communication, direct message, etc.) [0076]. In operation 204, the computing device provides the ML engine with a plurality of sets of topic-specific interactions. Each set of topic-specific interactions is tagged with the particular topic that is discussed by interactions in the set. Moreover, each topic that the ML engine is trained on is relevant to the organization. These relevant topics may be stored in a data repository [0077].)
Yin teaches indexing the digital data in a table that includes at least a first column including names of a plurality of members of the organization and a second column including keywords attributed to the plurality of members; and (Yin e.g. Information may be provided in training data 120 about each topic relevant to the organization to allow ML engine 102 to learn about these specific topics to be able to identify these topics from interactions 126 in observed data 124 after generating the ML model 104 [0050]. Training data 120 may include members 118 of the organization, such as specific persons at the organization for which interactions are most relevant for specific topics, and/or a type, role, or category of persons at the organization [0051]. An expert is a category specific to each topic 116 for individuals at the organization for which interactions are deemed more important for the specific topic than interactions for a regular member of the organization [0051]. In operation 206, the computing device provides the ML engine with a set of interactions tagged individually with one or more members of the organization that are involved in the interactions [0080]. The ML model is trained to recognize specific members of the organization, position or title of the members, and importance of interactions for specific members from observed interactions [0080]. In operation 208, the computing device (e.g., ML engine) trains the ML model based on the various sets of interactions provided in operations 202, 204, and 206, to recognize interaction types, topics relevant to the organization, and members specific to the organization [0083]. FIGS. 4A-4B are example tables showing correlations between members and topics, in accordance with one or more embodiments ([0010] and [0150]). A ML engine, or some other suitable component of a computing device, may learn different members of an organization and topics which are related to each member. These correlations may be stored in a data repository, with example tables being provided in FIGS. 4A-4B for Cutting Edge Technologies [0150]. )
Yin teaches training a Machine Learning (ML) model by data crawling through the digital data, assigning weights to the keywords, and storing the weights in a third column in the table; (Yin e.g. Initially, the system trains a machine-learning model to recognize a member interactions with a topic based on a set of historical data tagged with interactions. Once trained, the machine-learning model can identify members' interactions by continuously or periodically monitoring documents/activity records corresponding to the members of an organization [0034]. The ML engine may be trained with topic metadata both internal to the organization and external to the organization to identify relevant topics. For example, internal metadata may be gleaned from emails, direct messages, social media posts, meeting invitations, calendared events, documentation libraries, blogs, etc. External metadata, for example, may include trends, popularity, movements, local/regional/global influences, sector and industry news, market data and news, company reports, etc. [0079]. FIGS. 5A-5C are example tables showing connection factor weights for example connection factors; and ([0011] and [0161]). In some embodiments, ML engine 102 trains one or more ML models 104 to perform one or more operations. Training a ML model 104 involves using training data to generate a function that, given one or more inputs to the ML model 104, computes a corresponding output [0043]. Based on the training data 120, ML engine 102 generates the ML model 104, which is configured to provide suggested connections 132 between members 118 of the organization based on observed interactions 126 [0052]. ML engine 102 may calculate connection scores 108 based on a set of connection factors 106. The connection factors 106 may include any parameter, aspect, or characteristic that indicates a connection between a member of the organization and a particular topic [0056]. Each connection factor 106 may be assigned a connection factor weight 110 by ML engine 102 according to a ML model 104 for determining weights. A connection factor weight 110 allows for some connection factors 106 that are more indicative of a connection to a topic to be weighted more heavily in a calculation of an overall connection score 112 between the member and the topic. [0060]. FIG. 2B illustrates an example method 210 for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0085]. In operation 220, the computing device provides the ML engine with a plurality of experience levels. Each experience level is tagged with information that aids in determining a weight to assign to the different experience levels. Each member of the organization may have a certain experience level assigned for one or more topics that the member has been associated with, which may be stored in the member's profile [0098]. In operation 222, the computing device trains a ML model to generate weights for various different connection factors specific to the organization [0101]. Example connection factors that are described in FIG. 2B include interaction type, topic(s) included in the interaction, member(s) involved in the interaction, recency of the interaction, experience level of the member involved in the interaction [0101].) Yin teaches wherein the data crawling includes an initial crawling procedure that (i) adds the names of the plurality of members into the table and (ii) attributes the keywords used throughout digital records to respective members and wherein training further includes an in-depth training procedure that determines whether the keywords attributed to each member are used in a significant or meaningful way and that determines a commonality of keywords such that keywords used by a smaller number of people are weighted more heavily than keywords used by a larger number of people and wherein the table further includes a latest usage column storing a last time that a respective member used a respective keyword and the weights are based at least in part on the latest usage; (Yin e.g. Fig. 4A, FIGS. 4A-4B are example tables showing correlations between members and topics, in accordance with one or more embodiments [0150]. A ML engine, or some other suitable component of a computing device, may learn different members of an organization and topics which are related to each member. These correlations may be stored in a data repository, with example tables being provided in FIGS. 4A-4B for Cutting Edge Technologies [0150]. FIG. 4A shows an example table that lists each member of Cutting Edge Technologies that the ML engine learned in correlation with topics that the ML engine learned through interactions of the members, member profiles, etc. [0151]. FIGS. 5A-5C are example tables showing connection factor weights for example connection factors; and ([0011] and [0161]). In the example used for FIG. 5A, connection factor weights are used to describe the relative likelihood that a certain type of interaction correlates to a connection between the member and the particular topic [0167]. FIG. 5B is a table showing example connection factor weights for different ranges of relevancy of interactions. This table lists example ranges of relevancy and a connection factor weight assigned to each relevancy range. In other words, the weights in this table are used to enhance the effect that more recent interactions have on the overall connection score. Conversely, the weights in this table are also used to decrease the effect that less recent interactions have on the overall connection score [0168]. FIG. 5C is a table showing example connection factor weights for different connection factors. This table lists example connection factors and a connection factor weight assigned to each connection factor [0171]. In this example, the connection factors used to calculate the overall connection score include expertise, relevancy, experience, group participation, and frequency of interactions [0172]. ML engine 102 may calculate connection scores 108 based on a set of connection factors 106. The connection factors 106 may include any parameter, aspect, or characteristic that indicates a connection between a member of the organization and a particular topic [0056]. Each connection factor 106 may be assigned a connection factor weight 110 by ML engine 102 according to a ML model 104 for determining weights. A connection factor weight 110 allows for some connection factors 106 that are more indicative of a connection to a topic to be weighted more heavily in a calculation of an overall connection score 112 between the member and the topic [0060]. FIG. 2B illustrates an example method for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0085]. In operation 218, the computing device provides the ML engine with a plurality of time periods [0096]. Each time period is tagged with information that aids in determining a weight to assign to the different time periods as they relate to recency of interactions between members of the organization. More recent interactions may be considered more relevant than older interactions, and therefore may be assigned greater weight when determining a connection between a member and a topic [0096].)
Yin teaches wherein, in response to receiving an inquiry from a user, the ML model is configured during inference to use information in the table to provide an output to the user identifying a member in the organization who demonstrates expertise on a specific topic (Yin e.g. ML engine 102 is configured to analyze data 124 and propose connections 132 between members 118 of an organization based on observed interactions 126 between various members 118 in the organization [0040]. FIG. 3 illustrates an example method for proposing connections between members of an organization using a ML model, in accordance with one or more embodiments [0103]. According to an embodiment, the computing device may identify a need for the second member to receive expert help on the particular topic. This determination may be based on words or phrases included in interactions of the second member, such as "help," "expert," "aid," "teach," etc., in relation to the particular topic. In this embodiment, the communication to generate the connection is transmitted responsive to identifying the need for the second member to receive expert help on the particular topic. In one approach, the first member may be an expert on the particular topic, and this status is what dictates connection of the first member with the second member [0144]. FIG. 4B shows the learned topics correlated to each member associated with the topic. Using this information, the ML engine may propose connections between members who have common interest in a topic [0155].) Yin teaches wherein the output identifies a plurality of members who demonstrate expertise on the specific topic and (Yin e.g. FIG. 1 illustrates a system 100 for intelligently proposing connections between members of an organization using a ML model, in accordance with one or more embodiments. As illustrated in FIG. 1, system 100 includes a ML engine 102 and a data repository 134 [0039]. ML engine 102 is configured to analyze data 124 and propose connections 132 between members 118 of an organization based on observed interactions 126 between various members 118 in the organization [0040]. One or more embodiments apply a set of rules to generate transmissions that connect members of an organization. As an example, the system may generate an introduction email to introduce (a) a member that has been determined to be an expert on a particular topic based on the member's connection with the topic to (b) a member who started on a new project corresponding to the topic [0033]. The communication may include a user interface (UI) element that is selectable by the receiving member to indicate that a member would like to proceed with the connection to the other member(s) [0128]. The computing device may detect when such a UI element is selected in order to formally establish the connection with a message between the members introducing one another, and proposing a meeting time populated on each member's calendar when they are free to meet [0128].) Yin does not explicitly teach, however, Brisebois teaches provides for each identified member an explanation describing (i) a number of times the member has used one or more keywords corresponding do the specific topic and (ii) a recency of the member’s usage of the one or more keywords. (Brisebois e.g. The present invention relates generally to data aggregation and analysis and more particularly, but not by way of limitation, to systems and methods for identifying subject matter experts (col. 1 lines 22-25). The method includes, for each topic, for each user: the computer system measuring a proportion of the identified conversations that contain content suggestive of the topic, the measuring yielding at least one topical metric; the computer system analyzing timing of the identified conversations, the analyzing yielding at least one timing metric; and the computer system examining relationships among data attributes of the identified conversations, the examining yielding at least one expertise-scope metric (col. 2 lines 39-51). Further, the method includes the computer system generating multidimensional expertise data from the at least one topical metric, the at least one timing metric, and the at least one expertise-scope metric. The multidimensional expertise data is representative of the user's expertise on the topic. The multidimensional expertise data includes a topical dimension, an expertise-scope dimension, and a timeline dimension. (col. 2 lines 48-55). the method includes the computer system providing a searchable interface. The multidimensional expertise data is searchable by one or more topical parameters mapped to the topical dimension, by one or more scope parameters mapped to the expertise-scope dimension, and by one or more timeline parameters mapped to the timeline dimension (col. 2 lines 58-64). FIG. 11 illustrates an embodiment of an implementation of a SME system 1130 (col. 39 lines 62-63). The a posteriori classification engine 1128 can analyze and examine timing and data attributes of the communications to determine, for example, each user's scope of expertise and an expertise timeline (col. 41 lines 61-65). The timeline dimension generally includes data indicative of a recency and/or depth of the user's expertise on the topic. For example, the timeline dimension can include a timeline classification such as, for example, long-time expert, deep-domain expert, cutting-edge expert, and strategic expert (col. 43 lines 1-6). For example, the specific criteria for the role of long-time expert could require that the user's topic-relevant conversations begin prior to a certain date, that a certain number, percentage, or statistical distribution of the user's topic-relevant conversations occur prior to a certain date, etc. (col. 48 lines 5-9). The role of deep-domain expert is typically used for experts who exhibit continuous expertise over time. For example, the specific criteria for the role of deep-domain expert could require that the user have at least a minimum number of topic-relevant conversations within each of a plurality of periods of time (col. 48 lines 15-20). The role of cutting-edge expert is typically used for experts who exhibit extensive recent expertise. For example, the specific criteria for the role of cutting-edge expert could require that the user have at least a minimum number of topic-relevant conversations within a certain recent period of time (e.g., last month, last year, etc.) (col. 48 lines 26-31). The specific criteria for the role of strategic expert could require that the user have at least a minimum number of topic-relevant conversations within a certain recent period of time (e.g., last month, last year, etc.) along with having a certain position within the organization (e.g., manager, vice president, etc.) (col. 48 lines 37-44). Examples of the topical dimension, the scope dimension, and the timeline dimension will be described in greater detail with respect to FIG. 12 (col. 43 lines 6-8). FIG. 12 presents a flowchart of an example of a process 1200 for classifying users as SMEs (col. 44 lines 19-20). At block 1210, for each topic and user, the a posteriori classification engine 1128 analyzes a timing of the user's topic-relevant conversations. In a typical embodiment, the analysis results in one or more timeline metrics that indicate, at least in part, when the user developed expertise on the topic. For example, the one or more timeline metrics can include a date of a first topic-relevant conversation, a date of a most recent topic-relevant conversation, a number of topic-relevant conversations within certain periods of time (e.g., within the last month, within each month between the first date and the most recent date, etc.), a statistical distribution over time of conversations concerning the topic, etc. (col. 45 lines 12-23).)
The Examiner submits that before the effective filing date, it would have been obvious to one of ordinary skill in the art to modify Yin’s method and system for connecting member of an organization to providing an explanation of a number of times the member has used one or more keywords corresponding do the specific topic and a recency of the member’s usage of the one or more keywords as taught by Brisebois in order to identify SMEs and manage workflow related to request for SMEs (Brisebois e.g. cols. 39-40 lines 66-1).
As per claim 2 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein training the ML model includes a deep learning neural network process (Yin e.g. n an embodiment, ML engine 102 includes an artificial neural network. An artificial neural network includes multiple nodes (also referred to as artificial neurons) and edges between nodes. Edges may be associated with corresponding weights that represent the strengths of connections between nodes, which the ML engine 102 adjusts as machine learning proceeds [0045]. FIG. 2A illustrates an example method 200 for training a ML model to identify interactions between members of an organization, in accordance with one or more embodiments [0074]. In operation 208, the computing device (e.g., ML engine) trains the ML model based on the various sets of interactions provided in operations 202, 204, and 206, to recognize interaction types, topics relevant to the organization, and members specific to the organization [0083]. FIG. 2B illustrates an example method 210 for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0085].).
As per claim 5 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein the inquiry includes the specific topic or one or more of the keywords (Yin e.g. One or more embodiments apply a set of rules to generate transmissions that connect members of an organization [0033]. The rules may behave similar to those described for previous interactions, such as performing keyword searches for a topic, examining a degree of connection between a member and a topic, etc. [0142]. According to an embodiment, the computing device may identify a need for the second member to receive expert help on the particular topic. This determination may be based on words or phrases included in interactions of the second member, such as "help," "expert," "aid," "teach," etc., in relation to the particular topic [0144].).
As per claim 6 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein gathering the digital data includes obtaining data from a plurality of databases (Yin e.g. FIG. 1 illustrates a system 100 for intelligently proposing connections between members of an organization using a ML model, in accordance with one or more embodiments. As illustrated in FIG. 1, system 100 includes a ML engine 102 and a data repository 134 [0039]. In one or more embodiments, data repository 134 may be any type of storage unit and/or device (e.g., a file system, database, collection of tables, or any other storage mechanism) for storing data. Further, data repository 134 may include multiple different storage units and/or devices. The multiple different storage units and/or devices may or may not be of the same type or located at the same physical site. Further, data repository 134 may be implemented or may execute on the same computing system as ML engine 102 [0070]. In operation 204, the computing device provides the ML engine with a plurality of sets of topic-specific interactions. Each set of topic-specific interactions is tagged with the particular topic that is discussed by interactions in the set. Moreover, each topic that the ML engine is trained on is relevant to the organization. These relevant topics may be stored in a data repository [0077]. In an alternate approach, the ML engine may be trained on specific relevant topics (e.g., provided in a list or database) instead of from interactions [0078]. The ML engine may be trained with topic metadata both internal to the organization and external to the organization to identify relevant topics. For example, internal metadata may be gleaned from emails, direct messages, social media posts, meeting invitations, calendared events, documentation libraries, blogs, etc. External metadata, for example, may include trends, popularity, movements, local/regional/global influences, sector and industry news, market data and news, company reports, etc. [0079].).
As per claim 7 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 6, Yin teaches wherein the databases include at least a plurality of data silos each configured to store data in conjunction with use of one or more of collaboration tools, wiki tools, file sharing tools, messaging or chat tools, project management tools, and issue tracking tools associated with multiple different internal groups within the organization (Yin e.g. FIG. 1 illustrates a system 100 for intelligently proposing connections between members of an organization using a ML model, in accordance with one or more embodiments. As illustrated in FIG. 1, system 100 includes a ML engine 102 and a data repository 134 [0039]. In one or more embodiments, data repository 134 may be any type of storage unit and/or device (e.g., a file system, database, collection of tables, or any other storage mechanism) for storing data. Further, data repository 134 may include multiple different storage units and/or devices. The multiple different storage units and/or devices may or may not be of the same type or located at the same physical site. Further, data repository 134 may be implemented or may execute on the same computing system as ML engine 102 [0070]. FIG. 3 illustrates an example method for proposing connections between members of an organization using a ML model, in accordance with one or more embodiments [0103]. In operation 304, the computing device detects a plurality of interactions, of a second member of the organization, associated with the particular topic [0109]. Any type of interactions may be detected and used in method 300. Some example interactions include, but are not limited to, projects, emails, meetings, networking events, telephone conversations, smart speaker inquiries, direct messaging, social media posts and reactions, public speaking on a topic, trainings and/or sessions on a topic, etc. [0109]. The interactions may be detected from communication channels provided by, related to, and/or managed by the organization, such as an email server, local area network (LAN), virtual private network (VPN), organization-provided computing device (laptop, smartphone, tablet, etc.), internal direct messaging application, etc. [0111]. In operation 312, the computing device transmits a communication to generate a connection between the first member and the second member for collaboration about the particular topic [0126]. The communication may include a user interface (UI) element that is selectable by the receiving member to indicate that a member would like to proceed with the connection to the other member(s) [0128]. The computing device may determine what type of communication should be used to transmit the communication (e.g., email, phone call, direct message, etc.) [0131]. The communication, in one or more embodiments, may automatically suggest meeting time(s), initial agenda items, meeting objectives, and meeting take-aways [0132]. According to one approach, a virtual meeting room may be established with multiple interfaces for each member determined to be associated with a particular topic [0133]. The ML model may also provide additional services for the virtual meeting, such as produce a task list and predicted time to resolve the various tasks, assign members to perform individual tasks, list critical paths to accomplish tasks, provide further meeting(s) outlook, schedule a follow-up or regular meeting, etc. [0133].).
As per claim 8 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein the user associated with the inquiry is a fellow member of the organization (Yin e.g. One or more embodiments describe techniques for proactively connecting members of an organization together based on detected interest in a particular topic (Abstract). The organization may be a business, corporation association, educational institution, governmental agency, a collection of people, or any portion thereof [0041]. According to an embodiment, the computing device may identify a need for the second member to receive expert help on the particular topic. This determination may be based on words or phrases included in interactions of the second member, such as "help," "expert," "aid," "teach," etc., in relation to the particular topic [0144]. In this embodiment, the communication to generate the connection is transmitted responsive to identifying the need for the second member to receive expert help on the particular topic. In one approach, the first member may be an expert on the particular topic, and this status is what dictates connection of the first member with the second member [0144].)
As per claim 9 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein a back-end training component is configured to perform a rudimentary crawling or scraping process to store the digital data in the table, the back-end training component further configured to train the ML model. (Yin e.g. FIG. 1 illustrates a system 100 for intelligently proposing connections between members of an organization using a ML model, in accordance with one or more embodiments. As illustrated in FIG. 1, system 100 includes a ML engine 102 and a data repository 134 [0039]. FIG. 2A illustrates an example method 200 for training a ML model to identify interactions between members of an organization, in accordance with one or more embodiments [0074]. In operation 202, the computing device provides a ML engine with a set of interactions. Each interaction is tagged with an interaction type (e.g., email, phone call, social media communication, direct message, etc.) [0076]. Each interaction type that may be used by members of the organization are input to the ML engine to train the ML model, in one embodiment. In another embodiment, the ML engine is trained on interaction types that are most commonly used by members of the organization to discuss relevant topics to the organization, and based on these most common interaction types, the ML model may learn other less common interaction types [0076]. In operation 204, the computing device provides the ML engine with a plurality of sets of topic-specific interactions. Each set of topic-specific interactions is tagged with the particular topic that is discussed by interactions in the set. Moreover, each topic that the ML engine is trained on is relevant to the organization. These relevant topics may be stored in a data repository [0077]. In an alternate approach, the ML engine may be trained on specific relevant topics (e.g., provided in a list or database) instead of from interactions [0078]. The ML engine may be trained with topic metadata both internal to the organization and external to the organization to identify relevant topics. For example, internal metadata may be gleaned from emails, direct messages, social media posts, meeting invitations, calendared events, documentation libraries, blogs, etc. External metadata, for example, may include trends, popularity, movements, local/regional/global influences, sector and industry news, market data and news, company reports, etc. [0079]. FIGS. 4A-4B are example tables showing correlations between members and topics, in accordance with one or more embodiments; [0150]. A ML engine, or some other suitable component of a computing device, may learn different members of an organization and topics which are related to each member. These correlations may be stored in a data repository, with example tables being provided in FIGS. 4A-4B for Cutting Edge Technologies [0150].)
As per claim 10 (Currently Amended), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 9, wherein a front-end inference component is configured to:
Yin teaches receive the inquiry from a user device associated with the user; (Yin e.g. FIG. 1 illustrates a system 100 for intelligently proposing connections between members of an organization using a ML model, in accordance with one or more embodiments. As illustrated in FIG. 1, system 100 includes a ML engine 102 and a data repository 134 [0039]. ML engine 102 is configured to analyze data 124 and propose connections 132 between members 118 of an organization based on observed interactions 126 between various members 118 in the organization [0040]. The communication may include a user interface (UI) element that is selectable by the receiving member to indicate that a member would like to proceed with the connection to the other member(s) [0128]. The computing device may detect when such a UI element is selected in order to formally establish the connection with a message between the members introducing one another, and proposing a meeting time populated on each member's calendar when they are free to meet [0128]. According to an embodiment, the computing device may identify a need for the second member to receive expert help on the particular topic. This determination may be based on words or phrases included in interactions of the second member, such as "help," "expert," "aid," "teach," etc., in relation to the particular topic [0144].)
Yin teaches utilize the ML model and a data warehouse associated with the table to identify one or more members in the organization who demonstrate expertise on the specific topic; and (Yin e.g. FIGS. 4A-4B are example tables showing correlations between members and topics, in accordance with one or more embodiments; [0150]. A ML engine, or some other suitable component of a computing device, may learn different members of an organization and topics which are related to each member. These correlations may be stored in a data repository, with example tables being provided in FIGS. 4A-4B for Cutting Edge Technologies [0150]. FIG. 4B shows the learned topics correlated to each member associated with the topic [0155]. Using this information, the ML engine may propose connections between members who have common interest in a topic [0155]. Other forms of data retention than those shown in FIGS. 4A-4B may be used to store information collected from interactions and used by the ML engine and the ML model for proposed connections between members of the organization based on shared interest and/or expertise in a topic, in one or more embodiments [0159].)
Yin teaches provide the output to the user with an explanation describing how expertise is evaluated. (Yin e.g. In this embodiment, the communication to generate the connection is transmitted responsive to identifying the need for the second member to receive expert help on the particular topic. In one approach, the first member may be an expert on the particular topic, and this status is what dictates connection of the first member with the second member [0144]. Once these connections are proposed and/or attempted, by sending a message to each member that includes details about why the members may benefit from connecting together for a particular topic, additional interactions between the members may be detected [0156].) Yin does not explicitly teach, however, Brisebois teaches wherein the explanation includes at least (i) an occurrence count representing a number of times a keyword corresponding to the specific topic is attributed to a respective identified member and (ii) a latest usage time indicating a last time that the respective identified member used the keyword. (Brisebois e.g. The present invention relates generally to data aggregation and analysis and more particularly, but not by way of limitation, to systems and methods for identifying subject matter experts (col. 1 lines 22-25). The method includes, for each topic, for each user: the computer system measuring a proportion of the identified conversations that contain content suggestive of the topic, the measuring yielding at least one topical metric; the computer system analyzing timing of the identified conversations, the analyzing yielding at least one timing metric; and the computer system examining relationships among data attributes of the identified conversations, the examining yielding at least one expertise-scope metric (col. 2 lines 39-51). Further, the method includes the computer system generating multidimensional expertise data from the at least one topical metric, the at least one timing metric, and the at least one expertise-scope metric. The multidimensional expertise data is representative of the user's expertise on the topic. The multidimensional expertise data includes a topical dimension, an expertise-scope dimension, and a timeline dimension. (col. 2 lines 48-55). the method includes the computer system providing a searchable interface. The multidimensional expertise data is searchable by one or more topical parameters mapped to the topical dimension, by one or more scope parameters mapped to the expertise-scope dimension, and by one or more timeline parameters mapped to the timeline dimension (col. 2 lines 58-64). FIG. 11 illustrates an embodiment of an implementation of a SME system 1130 (col. 39 lines 62-63). The a posteriori classification engine 1128 can analyze and examine timing and data attributes of the communications to determine, for example, each user's scope of expertise and an expertise timeline (col. 41 lines 61-65). The timeline dimension generally includes data indicative of a recency and/or depth of the user's expertise on the topic. For example, the timeline dimension can include a timeline classification such as, for example, long-time expert, deep-domain expert, cutting-edge expert, and strategic expert (col. 48 lines 5-52). Examples of the topical dimension, the scope dimension, and the timeline dimension will be described in greater detail with respect to FIG. 12 (col. 43 lines 1-8). FIG. 12 presents a flowchart of an example of a process 1200 for classifying users as SMEs (col. 44 lines 19-20). At block 1210, for each topic and user, the a posteriori classification engine 1128 analyzes a timing of the user's topic-relevant conversations. In a typical embodiment, the analysis results in one or more timeline metrics that indicate, at least in part, when the user developed expertise on the topic. For example, the one or more timeline metrics can include a date of a first topic-relevant conversation, a date of a most recent topic-relevant conversation, a number of topic-relevant conversations within certain periods of time (e.g., within the last month, within each month between the first date and the most recent date, etc.), a statistical distribution over time of conversations concerning the topic, etc. (col. 45 lines 12-23).)
The Examiner submits that before the effective filing date, it would have been obvious to one of ordinary skill in the art to modify Yin’s method and system for connecting member of an organization to providing an explanation of an occurrence count representing a number of times a keyword corresponding to the specific topic is attributed to a respective identified member and a latest usage time indicating a last time that the respective identified member used the keyword as taught by Brisebois in order to identify SMEs and manage workflow related to request for SMEs (Brisebois e.g. cols. 39-40 lines 66-1).
As per claim 11 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 9, Yin teaches wherein the back-end training component is configured to receive feedback for retraining the ML model (Yin e.g. The system applies weights to different types of interactions to compute the connection score. A machine learning model may compute/adjust the weights for different types of interactions using a feedback loop. Specifically, the system obtains feedback on members determined to be connected to a topic. The feedback may indicate whether the members, classified by the system as meeting the threshold level of connection to a topic, were found to knowledgeable, helpful, or otherwise useful to other members on matters related to the topic [0037]. The machine-learning model identifies a correlation between interaction type and positive feedback to compute/adjust the weights assigned to interaction types for computation of the connection score. A weight assigned to a member's interaction, such as giving a presentation on a topic, may be increased if members with that interaction generally receive positive feedback [0037]. The ML model may measure effectiveness of the proposed connection based on this initial meeting once the meeting is complete [0133]. Once the computing device transmits a communication to generate a connection between the first member and the second member, additional interactions by one or more of the members who received the communication may be analyzed to determine the effectiveness of the connection [0134]. In one approach, based on these additional interactions, the computing device may determine an outcome score to rate effectiveness of the connection between the members. The outcome score may be used to tune the ML model and provide better connections in the future [0135].)
As per claim 12 (Currently Amended), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin teaches wherein the instructions further enable the processing device to perform steps of counting a number of times each keyword is attributed to the plurality of members and storing an occurrence count in a fourth column in the table. (Yin e.g. FIG. 3 illustrates an example method for proposing connections between members of an organization using a ML model, in accordance with one or more embodiments [0103]. In operation 302, the computing device analyzes a first profile, corresponding to a first member of an organization, to detect a particular topic associated with the first member. An associated topic is any topic that has been discussed, researched, or taught by the first member [0105]. In operation 304, the computing device detects a plurality of interactions, of a second member of the organization, associated with the particular topic [0109]. Any type of interactions may be detected and used in method 300. Some example interactions include, but are not limited to, projects, emails, meetings, networking events, telephone conversations, smart speaker inquiries, direct messaging, social media posts and reactions, public speaking on a topic, trainings and/or sessions on a topic, etc. [0109]. The particular topic may be detected by application of rules stored in an ML model trained to identify interaction types, members, and topics relevant to the organization. For example, a rule may perform a word search through all interactions to find each instance of the words "flying car" to determine which interactions are related to the topic "flying car." [0109]. In one approach, the computing device applies rules to the additional interaction to make one or more determinations. Some example determinations that the rule may make include, but are not limited to,..., determining a number and frequency of interactions associated with the particular topic between the first member and the second member subsequent to transmitting the message or the communication, and determining a length of each subsequent interaction associated with the particular topic [0141]. FIG. 5C is a table showing example connection factor weights for different connection factors. This table lists example connection factors and a connection factor weight assigned to each connection factor [0171]. In this example, the connection factors used to calculate the overall connection score include expertise, relevancy, experience, group participation, and frequency of interactions [0172].) Yin teaches wherein the counting includes detecting occurrences of a keyword and one or more variants of the keyword within internal documentation or records associated with use of one or more collaboration tools and increasing the occurrence count based on the detected occurrences. (Yin e.g. Initially, the system trains a machine-learning model to recognize a member interactions with a topic based on a set of historical data tagged with interactions. The historical data may include, for example, emails, calendar events, project descriptions, white papers, resumes, search history and conference descriptions with attendees. An interaction tag identify the member and specify the member's interaction type such as working on projects, publishing papers, discussing the topic over email, researching the topic, attending training, and giving trainings. Once trained, the machine-learning model can identify members' interactions by continuously or periodically monitoring documents/activity records corresponding to the members of an organization [0034]. In one or more approaches, training data 120 may include topics 116 that are relevant, important, and/or provided by the organization for monitoring. Topics 116 included in training data 120 may describe different ways of treating specific topics, such as topics to be tracked or monitored, topics to disregard (not track), topics that trigger an alert or message, special topics for which additional information should be obtained, etc. Information may be provided in training data 120 about each topic relevant to the organization to allow ML engine 102 to learn about these specific topics to be able to identify these topics from interactions 126 in observed data 124 after generating the ML model 104 [0050]. Example interactions 122 include meetings held virtually or in person by a member, messages sent and/or received by a member, events attended virtually or in person by a member, etc. [0049]. Based on observing interactions 126 and details of the profile 128, one or more particular topics 116, relevant to the organization, may be detected for this member [0055]. In one approach, the computing device applies rules to the additional interaction to make one or more determinations. Some example determinations that the rule may make include, but are not limited to,..., determining a number and frequency of interactions associated with the particular topic between the first member and the second member subsequent to transmitting the message or the communication, and determining a length of each subsequent interaction associated with the particular topic [0141]. The rules may behave similar to those described for previous interactions, such as performing keyword searches for a topic, examining a degree of connection between a member and a topic, etc. [0142].)
As per claim 13 (Currently Amended), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 12, Yin teaches wherein the ML model is configured to utilize the names, keywords, weights, occurrence counts, and/or proximity to time of use of the keywords to identify the member who demonstrates expertise in the specific topic (Yin e.g. FIG. 2B illustrates an example method 210 for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0085]. In operation 216, the computing device provides the ML engine with a plurality of members of the organization. Each member's identity is tagged with information that aids in determining a weight to assign to the different members when identified as a party to an interaction. These weights may be used to determine a connection of another member with a topic. Interactions of more important members may carry greater weight when discussing a topic than interactions with less important members [0094]. In operation 218, the computing device provides the ML engine with a plurality of time periods. Each time period is tagged with information that aids in determining a weight to assign to the different time periods as they relate to recency of interactions between members of the organization. More recent interactions may be considered more relevant than older interactions, and therefore may be assigned greater weight when determining a connection between a member and a topic [0096]. Some example connection factors include, but are not limited to, a demonstrated expertise in the topic, communications with one or more members specifically regarding the topic, how recent any communications regarding the topic have taken place (e.g., recency), an amount of experience or training that a member has in a particular topic, participation or inclusion in a network, committee, or group that relates to the topic, how frequently that a member interacts regarding the particular topic, etc. [0057].) Yin teaches wherein proximity to time of use includes determining from a latest usage column of the table a recency of a respective member’s usage of a respective keyword and weighting the expertise identification based at least in part on the recency. (Yin e.g. FIG. 2B illustrates an example method 210 for training a ML model to determine weights to assign connection factors, in accordance with one or more embodiments [0085]. In operation 218, the computing device provides the ML engine with a plurality of time periods. Each time period is tagged with information that aids in determining a weight to assign to the different time periods as they relate to recency of interactions between members of the organization. More recent interactions may be considered more relevant than older interactions, and therefore may be assigned greater weight when determining a connection between a member and a topic [0096]. Example connection factors that are described in FIG. 2B include interaction type, topic(s) included in the interaction, member(s) involved in the interaction, recency of the interaction, experience level of the member involved in the interaction [0101]. Some example connection factors include, but are not limited to, a demonstrated expertise in the topic, communications with one or more members specifically regarding the topic, how recent any communications regarding the topic have taken place (e.g., recency), an amount of experience or training that a member has in a particular topic, participation or inclusion in a network, committee, or group that relates to the topic, how frequently that a member interacts regarding the particular topic, etc. ([0057] and [0115]). FIG. 5B is a table showing example connection factor weights for different ranges of relevancy of interactions. This table lists example ranges of relevancy and a connection factor weight assigned to each relevancy range. In other words, the weights in this table are used to enhance the effect that more recent interactions have on the overall connection score. Conversely, the weights in this table are also used to decrease the effect that less recent interactions have on the overall connection score [0168]. In this example, one of the connection factors used to calculate the overall connection score describes how recent an interaction regarding the particular topic is, regardless of what type of interaction is being considered. If the member has an interaction regarding the particular topic in the last day, the connection factor weight for this interaction will be 1.0. If the member has interacted to discuss the particular topic more than a week ago, but less than a month ago, the connection factor weight will equal 0.6. Likewise, if the member last interacted on the particular topic more than a year ago, the connection factor score will equal 0.2. These example connection factor weights indicate a decaying relationship between length of time for the interaction and how strong the correlation is to establishing a connection between the member the particular topic [0169]. In this example, connection factor weights are used to describe the relative likelihood that an age of the interaction indicates a connection between the member and the particular topic. More, less, and/or different relevancy ranges may be considered in such a scheme, along with different weights associated with each relevancy range, in various approaches [0170]. FIG. 5C is a table showing example connection factor weights for different connection factors. This table lists example connection factors and a connection factor weight assigned to each connection factor [0171]. In this example, the connection factors used to calculate the overall connection score include expertise, relevancy, experience, group participation, and frequency of interactions [0172].)
As per claim 17 (Original), Yin in view of Brisebois teach the system of claim 14, Yin teaches wherein gathering the digital data includes obtaining data from a plurality of databases including one or more data silos each configured to store data in conjunction with use of one or more of collaboration tools, wiki tools, file sharing tools, messaging or chat tools, project management tools, and issue tracking tools associated with multiple different internal groups within the organization. (See claim 7 response.)
As per claim 19 (Original), Yin in view of Brisebois teach the method of claim 18, further comprising steps of: Yin teaches receiving the inquiry from a user device associated with the user; utilizing the ML model and a data warehouse associated with the table to identify one or more members in the organization who demonstrate expertise on the specific topic; and providing the output to the user with an explanation describing how expertise is evaluated. (See claim 10 response.)
As per claim 20 (Original), Yin in view of Brisebois teach the method of claim 18, further comprising steps of: Yin teaches counting a number of times each keyword is attributed to the plurality of members and storing an occurrence count in a fourth column in the table and (See claim 12 response.); Yin teaches applying the ML model by utilizing the names, keywords, weights, occurrence counts, and/or proximity to time of use of the keywords to identify the member who demonstrates expertise in the specific topic. (See claim 13 response.)
Claims 3-4 and 15-16 are rejected under 35 U.S.C. 103 as being unpatentable over Yin et al. (US 2023/0289732 A1) in view of Brisebois et al. (US 9,317,574 B1) and in further view of Belkin et al. (US 2025/0021919 A1).
As per claim 3 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 2, Yin in view of Brisebois and Belkin teach wherein the deep learning neural network process involves a Large Language Model (LLM).
Yin teaches a deep learning neural network process as shown in claim 2. Yin nor Brisebois explicitly teach the deep learning neural network process involves a Large Language Model (LLM).
However, Belkin teaches a Large Language Model (LLM) (Belkin e.g. Belkin teaches an enterprise knowledge retention and access system is disclosed (Abstract). Techniques are disclosed to create, maintain, and use Artificial Intelligence (AI)-based digital twins of company employees. Artificial Intelligence in this context includes, without limitation, the use of large language models (LLMs) and other machine learning techniques to mimic human interaction behavior [0013].)
The Examiner submits that before the effective filing date, it would have been obvious to one of ordinary skill in the art to modify Yin in view of Brisebois’s AI/ML neural network process to include a large language model as taught by Belkin in order to provide efficient data retrieval and human-like reasoning (Belkin e.g. [0016]).
As per claim 4 (Original), Yin in view of Brisebois teach the non-transitory computer-readable medium of claim 1, Yin nor Brisebois explicitly teach, however, Belkin teaches wherein the ML model includes a chatbot for receiving the inquiry and providing the output, and wherein the chatbot uses Natural Language Processing (NLP) (Belkin e.g. In various embodiments, a user can discover available digital twins (including for former users that are no longer with the company) already present in the system (created by other users) and interact with them (either individually or with several-by asking the same question of all of them), e.g., in a natural language via a chat interface [0066]. FIG. 5 is a flow diagram illustrating an embodiment of an interactive process to provide a response to a query using a digital twin service [0091]. In the example shown, at 502 a communication (e.g., email, chat or instant message, post, etc.) is determined to contain a query requiring or inviting a response. For example, an LLM and/or other techniques (regular expression, rule, filter, heuristic, etc.) may be used to determine that a message is from and/or to a specific user and poses an explicit or implied question on a subject [0091].).
The Examiner submits that before the effective filing date, it would have been obvious to one of ordinary skill in the art to modify Yin in view of Brisebois’s AI/ML neural network process to include a large language model as taught by Belkin in order to provide efficient data retrieval and human-like reasoning (Belkin e.g. [0016]).
As per claim 15 (Original), Yin in view of Brisebois teach the system of claim 14, Yin in view of Brisebois and Belkin teach wherein training the ML model includes a Large Language Model (LLM). (See claim 3 response.)
As per claim 16 (Original), Yin in view of Brisebois teach the system of claim 14, Yin in view of Brisebois and Belkin teach wherein the ML model includes a chatbot for receiving the inquiry and providing the output, and wherein the chatbot uses Natural Language Processing (NLP). (See claim 4 response.)
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Ayanna Minor whose telephone number is (571)272-3605. The examiner can normally be reached M-F 9am-5 pm.
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, Jerry O'Connor can be reached at 571-272-6787. 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.
/A.M./Examiner, Art Unit 3624
/Jerry O'Connor/Supervisory Patent Examiner,Group Art Unit 3624
Prosecution Insights