Query categorization is an essential part of query intent understanding in e-commerce search. A common query categorization task is to select the relevant fine-grained product categories in a product taxonomy. For frequent queries, rich customer behavior (e.g., click-through data) can be used to infer the relevant product categories. However, for more rare queries, which cover a large volume of search traffic, relying solely on customer behavior may not suffice due to the lack of this signal. To improve categorization of rare queries, we adapt the Pseudo-Relevance Feedback (PRF) approach to utilize the latent knowledge embedded in semantically or lexically similar product documents to enrich the representation of the more rare queries. To this end, we propose a novel deep neural model named Attentive Pseudo Relevance Feedback Network (APRF-Net) to enhance the representation of rare queries for query categorization. To demonstrate the effectiveness of our approach, we collect search queries from a large commercial search engine, and compare APRF-Net to state-of-the-art deep learning models for text classification. Our results show that the APRF-Net significantly improves query categorization by 5.9% on [email protected] score over the baselines, which increases to 8.2% improvement for the rare (tail) queries. The findings of this paper can be leveraged for further improvements in search query representation and understanding.

Users' clicks on Web search results are one of the key signals for evaluating and improving web search quality and have been widely used as part of current state-of-the-art Learning-To-Rank(LTR) models. With a large volume of search logs available for major search engines, effective models of searcher click behavior have emerged to evaluate and train LTR models. However, when modeling the users' click behavior, considering the bias of the behavior is imperative. In particular, when a search result is not clicked, it is not necessarily chosen as not relevant by the user, but instead could have been simply missed, especially for lower-ranked results. These kinds of biases in the click log data can be incorporated into the click models, propagating the errors to the resulting LTR ranking models or evaluation metrics. In this paper, we propose the De-biased Reinforcement Learning Click model (DRLC). The DRLC model relaxes previously made assumptions about the users' examination behavior and resulting latent states. To implement the DRLC model, convolutional neural networks are used as the value networks for reinforcement learning, trained to learn a policy to reduce bias in the click logs. To demonstrate the effectiveness of the DRLC model, we first compare performance with the previous state-of-art approaches using established click prediction metrics, including log-likelihood and perplexity. We further show that DRLC also leads to improvements in ranking performance. Our experiments demonstrate the effectiveness of the DRLC model in learning to reduce bias in click logs, leading to improved modeling performance and showing the potential for using DRLC for improving Web search quality.

We address the problem of learning how compatible two products are. Assessing compatibility is challenging because the meaning of compatibility changes depending on product categories. In this study, we leverage domain experts' knowledge to generate labels and datasets. Next, we engineer 58 different features from product titles and product descriptions. We experiment with both tree-based and deep learning classifiers using different sets of features to capture compatibility patterns across four product categories. Although the performance does not show consistent pattern across all categories, the precision and recall of the best algorithm from most categories are above 90%. In addition, we find that the performance of classifiers are in general satisfactory. Based on human validation, few best-performing classifiers demonstrate better performance than labels generated from domain experts.

Functional languages suffer from problems associated with inefficient use of store. In this paper we present a compile-time garbage collection optimization for a strict higher-order functional language to mitigate these problems. Compile-time garbage collection involves the determination, at compile-time, of points in a program's execution at which garbage collection will take place and of the parts of store to be garbage collected. Actual collection still takes place at run-time. The compile-time garbage collection optimization is validated by two static (compile-time) analyses. Generation analysis gives us sharing information about a given list while inheritance analysis essentially tells us which parts of lists evaluated during the application of a function are part of the result of the application. Correctness of the analyses and optimizations is considered, using a denotational store semantics as a reference. We apply the analyses and optimizations to an example quicksort program and quantify the improvement. We find that the program is significantly improved to the extent that its consumption of store is optimal after optimization.