Patents by Inventor Divyakumar Menghani
Divyakumar Menghani has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).
-
Patent number: 11334928Abstract: The technical problem of predicting the level of activity driven by a web page representing a company in an on-line social network system is addressed by providing a company page assessment system. A company page assessment system is configured to use a machine learning model to determine features and respective thresholds that signal the likelihood of a web page representing a company (a company page) to drive member activity in the on-line social network system. The company page assessment system also provides the ability to assess quality of a company page. Different quality levels of a company page can be indicated by respective labels “Complete Page,” “Quality Page,” and “Engaged Quality Page.Type: GrantFiled: April 23, 2018Date of Patent: May 17, 2022Assignee: Microsoft Technology Licensing, LLCInventors: Rajan Ramesh Chaudhari, Sparsh Agarwal, Tomer Cohen, Divyakumar Menghani, Aaron Tyler Rucker
-
Publication number: 20210097424Abstract: Apparatuses, computer readable medium, and methods are disclosed for dynamic selection of features for training a machine learning model. A method includes using predetermined rules to determine first members that are hiring managers and second members that are not hiring managers, where the members are part of an online connection system. The method further includes selecting first features based on the predetermined rules, member activity data, member profile data, and member activity and usage data. The method further includes training a machine learning model to determine values for coefficients, where the machine learning model determines a score that indicates whether a member is a hiring manager. The method may further includes selecting a second features from the first features, where a feature of the first features is dropped if a value of a corresponding coefficient is lower than a threshold, and randomly selecting new features for the second features.Type: ApplicationFiled: September 26, 2019Publication date: April 1, 2021Inventors: Luxin Kang, Divyakumar Menghani
-
Publication number: 20200174984Abstract: Disclosed are systems, methods, and non-transitory computer-readable media for generating a diversity index. A diversity index system determines distribution values indicating a distribution of a set of users among demographic groups defined by two or more diversity dimensions. The diversity index system generates a distribution vector based on the distribution values. The diversity index system determines, based on the distribution vector and a similarity matrix, a set of diversity index values forming a diversity index vector. The similarity index includes a set of similarity scores for the demographic groups. The diversity index system determines a diversity index score for the set of users based on the diversity index vector and the distribution vector. The diversity index score indicates a level of diversity amongst the users from the set of users.Type: ApplicationFiled: November 29, 2018Publication date: June 4, 2020Inventors: Beibei Wang, Patrick Driscoll, Divyakumar Menghani, Luxin Kang
-
Patent number: 10496721Abstract: Techniques for searching for generating and using an online activity index for users of an online service are disclosed herein. In some example embodiments, for each one of a plurality of dimensions of online activity for an online service, the system determines a corresponding value for each one of a set of one or more activities corresponding to the dimension for a user of the online service, with the value representing a level of engagement by the user in the corresponding activity via the online service. The system generates a corresponding activity index for each one of the dimensions based on the corresponding value for each one of the set of one or more activities corresponding to the dimension. The system generates content based on the corresponding activity index of at least one of the dimensions, and causes the content to be displayed on a device associated with the user.Type: GrantFiled: October 31, 2016Date of Patent: December 3, 2019Assignee: Microsoft Technology Licensing, LLCInventors: Vibhu Prakash Saxena, Yiran Pang, Kuo-Ning Huang, Yue Li, Divyakumar Menghani, Ming Chen
-
Publication number: 20190325487Abstract: The technical problem of predicting the level of activity driven by a web page representing a company in an on-line social network system is addressed by providing a company page assessment system. A company page assessment system is configured to use a machine learning model to determine features and respective thresholds that signal the likelihood of a web page representing a company (a company page) to drive member activity in the on-line social network system. The company page assessment system also provides the ability to assess quality of a company page. Different quality levels of a company page can be indicated by respective labels “Complete Page,” “Quality Page,” and “Engaged Quality Page.Type: ApplicationFiled: April 23, 2018Publication date: October 24, 2019Inventors: Rajan Ramesh Chaudhari, Sparsh Agarwal, Tomer Cohen, Divyakumar Menghani, Aaron Tyler Rucker
-
Publication number: 20190197486Abstract: Techniques for calculating probability of hire scores in a streaming environment are described. In an embodiment, a system accesses a candidate stream definition comprising a hiring organization and a role. Additionally, the system accesses, based on the candidate stream definition, one or more stream-related information sources, and extracts attributes from the stream-related information sources. Moreover, the system identifies similar organizations based on the attributes, the role, and features of the hiring organization. Then, the system accesses historical data regarding hiring patterns at the similar organizations for the role. The system calculates, based on the historical data and interactions by the hiring organization with member profiles of an online system (e.g., hosting a social networking service), a probability of hire for the role. Next, the system presents the probability of hire.Type: ApplicationFiled: March 30, 2018Publication date: June 27, 2019Inventors: John Robert Jersin, Alexis Blevins Baird, Divyakumar Menghani, Benjamin John McCann, Huanyu Zhao
-
Publication number: 20180121479Abstract: Techniques for searching for generating and using an online activity index for users of an online service are disclosed herein. In some example embodiments, for each one of a plurality of dimensions of online activity for an online service, the system determines a corresponding value for each one of a set of one or more activities corresponding to the dimension for a user of the online service, with the value representing a level of engagement by the user in the corresponding activity via the online service. The system generates a corresponding activity index for each one of the dimensions based on the corresponding value for each one of the set of one or more activities corresponding to the dimension. The system generates content based on the corresponding activity index of at least one of the dimensions, and causes the content to be displayed on a device associated with the user.Type: ApplicationFiled: October 31, 2016Publication date: May 3, 2018Inventors: Vibhu Prakash Saxena, Yiran Pang, Kuo-Ning Huang, Yue Li, Divyakumar Menghani, Ming Chen
-
Publication number: 20170323397Abstract: Techniques for publishing a profile timeline of a member are described.. publisher receives, from a device of a viewer, a request to view a profile page of a member of an online social network. The publisher accesses employment data of the member having a first job associated with a first job title at a first employer with a specific duration. Additionally, the publisher can select a plurality of other members, each member in the plurality of other members having a similar job title related to the first job title. Furthermore, the publisher can determine an average duration for the first job title based on information accessed from the plurality of other members, and generate a first timeframe using the accessed employment data of the member. Subsequently, the publisher causes a presentation, on a display of the device, of the profile page of the member having the generated first timeframe.Type: ApplicationFiled: May 3, 2016Publication date: November 9, 2017Inventors: Divyakumar Menghani, Matthew Spencer Rendely, Vibhu Prakash Saxena