Microsoft Technology Licensing, LLC

Erik Eugene Buchanan

Benjamin John McCann

John Robert Jersin

Qi Guo

Xianren Wu

Shan Zhou

Bo Hu

Lei Ni

Sahin Cem Geyik

Cagri Ozcaglar

Nadeem Anjum

Alexis Blevins Baird

Skylar Payne

Gio Borje

Jerry Lin

Jeffrey Michael Wright

Travis Nelson Hummel

Prakhar Sharma

Kevin Keck

Brian Schmitz

An indication of a plurality of different entities in a social networking service is received, including at least two entities having a different entity type. A plurality of user profiles in the social networking service is accessed. A first machine-learned model is used to learn embeddings for the plurality of different entities in a d-dimensional space. A second machine-learned model is used to learn an embedding for each of one or more query terms that are not contained in the indication of the plurality of different entities in the social networking service, using the embeddings for the plurality of different entities learned using the first machine-learned model, the second-machine learned model being a deep structured semantic model (DSSM). A similarity score between a query term and an entity is calculated by computing distance between the embedding for the query term and the embedding for the entity in the d-dimensional space.

Joint representation learning of standardized entities and queries

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

In an example, an indication of a plurality of different entities in a social networking service is received, including at least two entities having a different entity type. Then a plurality of user profiles in the social networking service are accessed. A machine-learned model is then used to calculate, based on co-occurrence counts reflecting a number of user profiles in the plurality of user profiles in which corresponding nodes co-occurred, a similarity score between a first node and second node by computing distance between the first node and the second node in a d-dimensional space on which a plurality of entities are mapped, the similarity score generated using a generalized linear mixed model having a global coefficient vector applied to global function pertaining to the co-occurrence counts and a first random effects coefficient vector applied to a random effects per-country function.

Smart suggestions personalization with GLMix

Techniques for dynamically altering weights to re-weight attributes of an ordering model based on feedback in a streaming environment are described. In an embodiment, a system accesses, based on a candidate stream definition comprising a role including a title, one or more stream-related information sources, and extracts attributes from the sources. Additionally, the system inputs the attributes to a combined ordering model that is trained by a machine learning algorithm to output ordering scores for member profiles of an online system (e.g., hosting a social networking service). The combined ordering model includes weights assigned to the attributes. Moreover, the system orders, based on the ordering scores, the profiles. Furthermore, the system presents, on a display, of top-ordered profiles. Then, the system accesses feedback regarding the top-ordered profiles and dynamically trains the weights assigned to each of the attributes to alter the weights assigned to the attributes based on the feedback.

Using feedback to create and modify candidate streams

Systems and methods for query term weighting are disclosed. A server receives a search query for employment candidates, the search query comprising a set of parameters, each parameter having a weight. The server generates, from a data repository storing records associated with professionals, a first set of search results based on the set of parameters and the weights of the parameters in the set. The server transmits the first set of search results. The server receives a response to search result(s) from the first set of search results, the search result(s) being associated with a set of factors, the response indicating a level of interest in the search result(s). The server adjusts the parameters in the set of parameters or adjusts the weights of the parameters based on the response to the search result(s). The server provides an output based on the adjusted parameters or the adjusted weights.

Query term weighting

In an example, a plurality of user profiles in a social networking service are accessed. A heterogeneous graph structure having a plurality of nodes connected by edges is generated, each node corresponding to a different entity in the social networking service, each edge representing a co-occurrence of entities represented by nodes on each side of the edge in at least one of the user profiles. Weights are calculated for each edge of the heterogeneous graph structure, the weights being based on co-occurrence counts reflecting a number of user profiles in the plurality of user profiles in which corresponding nodes co-occurred. The heterogeneous graph structure is embedded into a d-dimensional space. A machine-learned model is then used to calculate a similarity score between a first node and second node by computing distance between the first node and the second node in the d&#xac;-dimensional space.

Standardized entity representation learning for smart suggestions

In an example, a deep learning network is used to calculate a similarity score between a first query in a social networking service and each of one or more suggestable entities in the social networking service. The suggestable entities are determined via a first machine learned model. The deep learning network takes as input the suggestable entities as well as a history of interactions with a graphical user interface of a social networking service by a first member of the social networking service, a history of queries performed via the graphical user interface by the first member, and the first query itself.

Personalized deep models for smart suggestions ranking

Systems and methods for applicant tracking system (ATS) integration with a deduplicator are disclosed. A recruiting company computer system accesses a first entity record external to an ATS. The recruiting company computer system determines that the first entity record corresponds to a second entity record within the ATS based on at least first information of the first entity record and second information of the second entity record. The first information is different from the second information. The recruiting company computer system imports, into the second entity record within the ATS, information from the first entity record external to the ATS in response to the first entity record corresponding to the second entity record. The recruiting company computer system provides, in response a request to access information about an entity associated with the second entity record, the information from the first entity record.

Applicant tracking system integration

Systems and methods for messaging service integration with a deduplicator are disclosed. In example embodiments, a server determines that a first user from among a set of users is composing a message to an entity using a first contact information of the entity. The server determines that a second user from among the set of users previously contacted the entity using a second contact information of the entity, the second contact information being different from the first contact information. The server provides for display, within a graphical user interface, an indication that the second user has previously contacted the entity.

Messaging service integration with deduplicator

Systems and methods for identifying user information from a set of pages are disclosed. In example embodiments, a server determines that a first set of pages is associated with a specific user based on addresses of the first set of pages having a common portion of a uniform resource locator (URL). The server determines that at least a threshold number of pages from the first set of pages include common information, the common information comprising contact information or social networking information. The server associates the contact information or the social networking information with a user profile of the specific user. The server provides, as a digital transmission, the contact information or the social networking information.

Identifying user information from a set of pages

Systems and methods for deduplication and disambiguation are disclosed. In example embodiments, a server accesses stored information about a first entity and stored information about a second entity. The server determines, based on the accessed stored information about the first entity and the accessed stored information about the second entity, a set of information items known about both the first entity and the second entity. The server computes, based on the set of information items, a probability that the first entity corresponds to the second entity by computing one or more expressiveness scores corresponding to a value of a first information item and a value of a second information item from the set of information items. The server provides, as a digital transmission, an output representing the computed probability.

Growing community of inventors

Mountain View, CA, United States of America

Erik Eugene Buchanan