Microsoft Technology Licensing, LLC

Daniel Sairom Krishnan Hewlett

Qi Guo

Dan Liu

Wenxiang Chen

Yi Wu

Wei Guang Lu

Songtao Guo

Sara Smoot Gerrard

Alexandre Patry

Runfang Zhou

Meng Meng

Baolei Li

Jeremy Lwanga

Xuebin Yan

Mingzhou Zhou

Anthony Hsu

Vitaly Abdrashitov

William Tang

Xuhong Zhang

Xiaoyi Zhang

Lester Gilbert Cottle, Iii

Yu Gong

Haitong Tian

Siyao Sun

Keqiu Hu

Pei-Lun Liao

Chenya Zhang

Tanvi Sudarshan Motwani

Sai Krishna Bollam

Wensheng Sun

Sriram Vasudevan

Techniques for incorporating sequence encoders into machine-learned models where the sequence encoders operate on bag of words (BOW) input are provided. Tokens that are associated with online activities of an entity are identified. Machine-learned embeddings that correspond to the tokens are identified. Based on one or more ordering criteria that are independent of the temporal occurrence of the online activities of the entity, an order of the machine-learned embeddings is determined. Based on the order, the machine-learned embeddings are inputted to a sequence encoder that generates output. Based on the output, a machine learned model that includes the sequence encoder generates a score. A content item is selected based on the score. The content item is transmitted over a computer network to a computing device.

Machine-learned models incorporating sequence encoders that operate on bag of words input

Methods are presented for providing dynamic search filter suggestions that are updated and ranked based on the user filter selections. One method includes detecting a query received in a user interface (UI), calculating, by a search-candidate model, first search results, and calculating, by a suggestions model, first filter suggestions for filter categories to filter responses to the query. The suggestions model is obtained by training a machine-learning algorithm utilizing pairwise learning-to-rank modeling. The first search results and the first filter suggestions are presented in the UI. When a selection in the UI of a filter suggestion is detected, the search-candidate model calculates second search results for the filter categories based on the query and the selected filter suggestion, and the suggestions model calculates second first filter suggestions based on the query and the selected filter suggestion. The second search results and the second filter suggestions are presented in the UI.

Context-aware query suggestions

In an example embodiment, a platform is provided that utilizes information available to a computer system to feed a neural network. The neural network is trained to determine both the probability that a searcher would select a given potential search result if it was presented to him or her and the probability that a subject of the potential search result would respond to a communication from the searcher. These probabilities are combined to produce a single score that can be used to determine whether to present the searcher with the potential search result and, if so, how high to rank the potential search result among other search results. During the training process, a rescaling transformation for each input feature is learned and applied to the values for the input features.

Rescaling layer in neural network

In an example embodiment, a platform is provided that utilizes information available to a computer system to feed a neural network. The neural network is trained to determine both the probability that a searcher would select a given potential search result if it was presented to him or her and the probability that a subject of the potential search result would respond to a communication from the searcher. These probabilities are essentially combined to produce a single score that can be used to determine whether to present the searcher with the potential search result and, if so, how high to rank the potential search result among other search results. In a further example embodiment, embeddings used for the input features are modified during training to maximize an objective.

Embedding layer in neural network for ranking candidates

In an example embodiment, position bias and other types of bias may be compensated for by using two-phase training of a machine-learned model. In a first phase, the machine-learned model is trained using non-randomized training data. Since certain types of machine-learned models, such as those involving deep learning (e.g., neural networks) require a lot of training data, this allows the bulk of the training to be devoted to training using non-randomized training data. However, since this non-randomized training data may be biased, a second training phase is then used to revise the machine-learned model based on randomized training data to remove the bias from the machine-learned model. Since this randomized training data may be less plentiful, this allows the deep learning machine-learned model to be trained to operate in an unbiased manner without the need to generate additional randomized training data.

Two-stage training with non-randomized and randomized data

In an example embodiment, position bias is addressed by introducing an inverse propensity weight into a loss function used to train a machine-learned model. This inverse propensity weight essentially increases the weight of candidates in the training data that were presented lower in a list of candidates. This achieves the benefit of counteracting the position bias and increases the effectiveness of the machine-learned model in generating scores for future candidates. In a further example embodiment, a function is generated for the inverse propensity weight based on responses to contact requests from recruiters. In other words, while the machine learned-model may factor in both the likelihood that a recruiter will want to contact a candidate and the likelihood that a candidate will respond to such a contact, the function generated for the inverse propensity weight will be based only on training data where the candidate actually responded to a contact.

Position debiasing using inverse propensity weight in machine-learned model

Growing community of inventors

Sunnyvale, CA, United States of America

Daniel Sairom Krishnan Hewlett