Microsoft Technology Licensing, LLC

Parul Jain

Krishnaram Kenthapadi

Stuart MacDonald Ambler

Fedor Vladimirovich Borisyuk

Anthony Duane Duerr

Kevin Kao

Caleb Timothy Johnson

Zheng Li

Songtao Guo

Bo Zhao

Aman Grover

Vijay K Dialani

Siyu You

Benjamin Hoan Le

David J Stein

Christopher Matthew Degiere

Kumaresh Pattabiraman

Kevin Chuang

Xie Lu

Jeffrey Warren Lee

Mohammad Aleagha

Dezhen Li

Minhtu Nguyen

Shidong Yan

Tao Zheng

Brian Hung-Chung Leung

Fenglin Li

Methods, systems, and computer programs are presented for providing a unified search tool for searching data in multiple domains. One method includes an operation for causing presentation of a user interface (UI) for entering one or more search parameters to search items in two or more domains. Each domain is owned by a different entity and data for each domain is acquired by the corresponding entity. Further, the method includes operations for receiving one or more search parameters values, searching, in response to the received search parameter values, the two or more domains based on the search parameters values, and ranking separately search results for each domain. Further, the method includes causing presentation in the UI of the search results for the two or more domains based on the ranking, the search results for each domain being presented in a separate results list.

Unified search for dual domains

Techniques for enhancing usability and electronic resource efficiency using job relevance are disclosed herein. In some embodiments, a list of a predetermined number of top job openings for a member of the social networking system is generated. The list is communicated to a device of the user for presentation in a user interface on the device. The user interface allows the member to browse through and provide an indication of a lack of relevancy of each of the list of the predetermined number of top job openings to the member and an indication of a reason for the lack of relevancy. Based on the user providing the indication of the lack of relevancy and the reason, a modified relevancy assessment of each of the plurality of job openings is generated. The list of the predetermined number of top job openings for the member is regenerated and communicated to the device.

Specialized user interfaces and processes for increasing user interactions with job postings in a social network/top jobs

A member profile including a vector containing a field for each of a plurality of skills and a rating of one or more of the skills in the vector for a member of a social networking service is obtained. A first distance indicating a vector distance between the vector of the member profile and a vector of a hypothetical member profile representing the perfect job candidate is obtained. A hypothetical member profile for the member is created by combining the vector of the member profile with the indication of how each of the one or more skills is improved through taking the course from course information. A second distance between the member and the hypothetical perfect candidate for the job is obtained, and the difference between the first distance and the second distance is calculated to determine an estimate of how much the course will increase the member's job chances.

Estimating effects of courses

The disclosed embodiments provide a system for processing data. During operation, the system obtains events reflecting responses by a user to job recommendations outputted to the user. Next, the system updates a set of features for the user from the events. The system then includes the updated set of features in a feature repository for use by a statistical model in generating a ranking of jobs for the user. Finally, the system retrains the statistical model using the events prior to using the statistical model to update the outputted job recommendations using the ranking.

Nearline updates to personalized models and features

In an example, a method includes requesting cohort data for the cohort, computing a plurality of cohort data first accuracy metrics, adding a threshold percentage of noise data points to the cohort data, computing a plurality of cohort data second accuracy metrics, repeating the adding and computing the second accuracy metrics until a mathematical difference between one or more of the first accuracy metrics and the second accuracy metrics exceed a threshold value, and suppressing displaying the cohort data in response to the mathematical difference exceeding the threshold value.

Mapped dataset error tolerance

In an example, a method includes determining an external quality metric for cohort data by comparing the cohort data with externally available data for persons who are members of the cohort; determining a confidence metric for the cohort data according to a variability between the cohort data and ground truth data; determining a member accuracy metric by requesting evaluation of the cohort data by members of the online social networking service; normalizing each of the metrics to a predetermined numerical range; retrieving a weight for each of the external quality metric, the confidence metric, and the member accuracy metric; filtering the cohort data according to a convex combination of the external quality metric, the confidence metric, and the member accuracy metric with their respective weights; and suppressing display of the cohort data in response to the filtered information for the cohort indicating that the convex combination traverses a threshold value.

Filtering mapped datasets

Methods and systems for generating queries from a behavioral index model are described. According to various embodiments, a system receives a query input including one or more search terms from a client device associated with a member profile. The system generates a query from the one or more search terms and one or more profile terms. The system causes one or more query machines to search one or more inverted indices using the query and receives one or more result of documents distributed across the inverted indices. The system generates an ordered list of the one or more result sets and causes presentation of the ordered list at a client device associated with the member profile.

Machine learned query generation on inverted indices

In an example, a method includes generating a targeted communication to respective computing devices of one or more members of an online social networking service, electronically collecting the responses to the targeted communication, mapping the responses to a cohort by updating a record in a database of members, the record identifying the cohort for the respective member, filtering information corresponding to a cohort to provide filtered cohort, and suppressing displaying of information corresponding to the cohort in response to the filtered information for the cohort indicating one or more of the biases being above a bias threshold value.

A machine may be configured to generate a universal concept graph. For example, the machine generates an internal set of concept phrases based on an internal dataset including content from internal documents associated with a social networking service (SNS). The machine generates an external set of concept phrases based on an external dataset including content from external documents that are external to the SNS. The machine generates a set of nodes for a universal concept graph based on performing a union operation of the internal set of concept phrases and the external set of concept phrases. Each node corresponds to a particular concept phrase. The machine generates a set of edges among a plurality of nodes of the set of nodes based on one or more relationship indicators for pairs of nodes. The machine generates the universal concept graph based on the set of nodes and the set of edges.

Universal concept graph for a social networking service

A machine may be configured to determining key concepts in documents. For example, the machine accesses a universal concept graph that includes a first set of nodes that represent concept phrases derived from internal documents associated with a social networking service (SNS) and external documents that are external to the SNS, and a first set of edges that connect a plurality of nodes of the first set of nodes. The machine accesses a content object associated with the SNS. The machine generates an induced concept graph associated with the content object based on an analysis of the content object and the universal concept graph. The machine identifies one or more key concept phrases in the content object based on applying one or more key concept selection algorithms to the induced concept graph. The machine stores the one or more key concept phrases in a record of a database.

Determining key concepts in documents based on a universal concept graph

In an example, a plurality of previously submitted confidential data values of a first confidential data type retrieved for a slice having one or more attributes. For a confidential data type, one or more submitted confidential data values of the confidential data type from the slice that are considered outliers based on an external data set or internal data set. A confidence score is calculated by multiplying a support score for the confidential data type in the slice by a non-outlier score for the confidential data type in the slice, the support score being equal to n'/(n&#x2032;+c), where c is a smoothing constant and n&#x2032; is the number of non-excluded submitted confidential data values of the confidential data type in the slice and the non-outlier score being equal to n&#x2032;/n, where n is the total number of non-null submitted confidential data value of the confidential data type in the slice.

Aggregation and generation of confidential data insights with confidence values

In an example, a plurality of segments of percentile information indicating submitted confidential data values from users matching at least one attribute of a submitting user are retrieved. Then, for each of the segments, an interquartile range is calculated for a first and a second of a plurality of percentiles in the segment, an initial lower limit is computed for the segment by taking a maximum of zero or the difference between the value for the first of the plurality of percentiles and a product of a preset alpha parameter and the interquartile range, and interpolation is performed on values for the plurality of percentiles for the segment to obtain values for a third percentile. The initial lower limits and the interpolated values for the third percentiles are aggregated across the segments. A merged lower limit is determined by applying a function to the aggregated initial lower limits and aggregated interpolated values.

External dataset-based outlier detection for confidential data in a computer system

In an example, a submission of a confidential data value of a first confidential data type is received from a first user with one or more attributes. A plurality of previously submitted confidential data values of a first confidential data type for a cohort matching the one or more attributes of the first user are retrieved. A plurality of percentiles for the confidential data values are calculated. Then, an interquartile range is calculated for a first and a second of the plurality of percentiles. A lower limit for the first confidential data type and the cohort is computed by taking a maximum of zero or the difference between the value for the first of the plurality of percentiles and a product of a preset alpha parameter and the interquartile range. Then it is determined if the confidential data value submitted by the user is lower than the lower limit.

Growing community of inventors

Cupertino, CA, United States of America

Parul Jain