Microsoft Technology Licensing, LLC

Albert Lingfeng Cui

Yan Wang

Yang Zhou

Aastha Jain

Songtao Guo

Ankan Saha

Eric M Lawrence

Parag Agrawal

Ashwin Narasimha Murthy

Joonhyung Lim

Lizabeth Li

Aastha Nigam

Yafei Wang

Yi Wang

Bohong Zhao

Carrie Peng

Abdulla Mohammad Al-Qawasmeh

David Sung

Qiannan Yin

Qunzeng Liu

In an example embodiment, a single machine learned model that allows for ranking of entities across all of the different combinations of node types and edge types is provided. The solution calibrates the scores from Edge-FPR models to a single scale. Additionally, the solution may utilize a per-edge type multiplicative factor dictated by the true importance of an edge type, which is learned through a counterfactual experimentation process. The solution may additionally optimize on a single, common downstream metric, specifically downstream interactions that can be compared against each other across all combinations of node types and edge types.

Recommending edges via importance aware machine learned model

In an embodiment, the disclosed technologies include receiving a set of digital inputs; where the set of digital inputs includes a candidate entity-member entity pair that includes candidate entity data and member entity data; where the member entity data has been extracted from a node of an online service; where an exact match has not been found between the candidate entity data and the member entity data; in response to the set of digital inputs, outputting models of the candidate entity data and the member entity data, respectively; where the models indicate weight values assigned to text in the candidate entity data and weight values assigned to text in the member entity data, respectively; calculating a similarity score using the models; in response to the similarity score matching a threshold, inputting the candidate entity-member entity pair to a classifier to produce a classification; where the classifier uses a machine learning model that has been trained using features derived from previously-analyzed candidate entity-member entity pairs; where the classification indicates a mathematical likelihood that a match has been found between the candidate entity data and the member entity data.

Hybrid entity matching to drive program execution

Growing community of inventors

San Francisco, CA, United States of America

Albert Lingfeng Cui