Microsoft Technology Licensing, LLC

Cagri Ozcaglar

Sahin Cem Geyik

Krishnaram Kenthapadi

Qi Guo

Gungor Polatkan

Rohan Ramanath

Anish Ramdas Nair

Abhishek Gupta

Jaewon Yang

Xianren Wu

Erik Eugene Buchanan

Vijay K Dialani

Meng Meng

Sara Smoot Gerrard

Swanand Wakankar

Vitaly Abdrashitov

Gurwinder S Gulati

Brian Schmitz

Gio Carlo C Borje

Ketan Thakkar

Prakhar Sharma

The disclosed embodiments provide a system for processing data. During operation, the system obtains a first embedding generated by a topic model from parameters of searches by a first recruiting entity and obtains a set of additional embeddings generated by the topic model from attributes of a set of candidates. Next, the system determines match features that include measures of similarity between the first embedding and each embedding in the set of additional embeddings. The system then applies a machine learning model to the match features and additional features for the candidates to produce a set of scores for the candidates. Finally, the system generates a ranking of the candidates according to the scores and outputs at least a portion of the ranking as search results of a current search by the first recruiting entity.

Contextual search ranking using entity topic representations

Methods, systems, and computer programs are presented for improved search methods based on career transition embeddings. One method includes an operation for generating career transition vectors for members of an online service, each career transition vector comprising identifiers associated with the career transitions of each member. The method further includes operations for performing a similarity analysis of the career transition vectors to generate an embedding vector for each identifier, detecting access of a first member to a job search user interface, and selecting one or more top embedding vectors based on one or more embedding vectors of the first member. One or more search starters associated with the one or more top embedding vectors are generated, and the one or more search starters are presented on the job search user interface.

Data selection based on career transition embeddings

Techniques for implementing a learning semantic representations of sparse entities using unsupervised embeddings are disclosed herein. In some embodiments, a computer system accesses corresponding profile data of users indicating at least one entity of a first facet type associated with the user, and generating a graph data structure comprising nodes and edges based on the accessed profile data, with each node corresponding to a different entity indicated by the accessed profile data, and each edge directly connecting a different pair of nodes and indicating a number of users whose profile data indicates both entities of the pair of nodes. The computer system generating a corresponding embedding vector for the entities based on the graph data structure using an unsupervised machine learning algorithm.

Unsupervised learning of entity representations using graphs

The disclosed embodiments provide a system for processing data. During operation, the system obtains a set of feature additions and an evaluation metric for assessing the performance of a statistical model. Next, the system automatically builds treatment versions of the statistical model using a set of baseline features for the statistical model and feature combinations generated using the feature additions. The system then uses a hypothesis test and a fixed set of feature values to compare a baseline value of the evaluation metric for a baseline version of the statistical model that is built using the set of baseline features with additional values of the evaluation metric for the treatment versions. Finally, the system outputs a result of the hypothesis test for use in assessing an impact of the feature combinations on a performance of the statistical model.

Feature selection impact analysis for statistical models

In an example embodiment, gradient boosted decision trees are used to generate tree interaction features, which encode a set of decision rules for features of search results and hence allow feature interactions. These tree interaction features may then be used as features of a GLMix model, essentially injecting non-linearity into the GLMix model.

Entity-level search models with tree interaction features

Techniques for processing search queries are described. Consistent with some embodiments, a computer system generates a profile vector representation for each of several user profiles based on the user profile data of the user profiles, and then stores the vector representations. A subsequent query is processed to generate a query vector representation for the query. A neural network is used to generate a similarity score for each pairing of the query vector representation and a profile vector representation. Finally, some number of user profiles having the highest similarity scores are provided as search results.

Techniques for querying user profiles using neural networks

Techniques for improving search using generalized linear mixed models are disclosed herein. In some embodiments, a computer-implemented method comprises: receiving a search query comprising at least one search term and being associated with a user; extracting features from corresponding profiles of a plurality of candidates; for each one of the candidates, generating a corresponding score based on a generalized linear mixed model comprising a generalized linear query-based model and a random effects user-based model; selecting a subset of candidates from the plurality of candidates based on the corresponding scores; and causing the selected subset of candidates to be displayed to the user in a search results page for the search query.

Generalized linear mixed models for improving search

Growing community of inventors

Sunnyvale, CA, United States of America

Cagri Ozcaglar