Incorporating Behavioral Hypotheses for Query Generation

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 3105–3110, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 3105 Incorporating Behavioral Hypotheses for Query Generation Ruey-Cheng Chen SEEK Ltd rcchen@seek.com.au Chia-Jung Lee Microsoft cjlee@microsoft.com Abstract Generative neural networks have been shown effective on query suggestion. Commonly posed as a conditional generation problem, the task aims to leverage earlier inputs from users in a search session to predict queries that they will likely issue at a later time. User inputs come in various forms such as querying and clicking, each of which can imply different semantic signals channeled through the corre- sponding behavioral patterns. This paper in- duces these behavioral biases as hypotheses for query generation, where a generic encoder- decoder Transformer framework is presented to aggregate arbitrary hypotheses of choice. Our experimental results show that the pro- posed approach leads to significant improve- ments on top-k word error rate and Bert F1 Score compared to a recent BART model. 1 Introduction Query suggestion is key to the usability of a search engine in the way it helps users formulating more effective queries or exploring related search needs. Prior work tackles this problem by employing pri- marily two strategies. The first one is based on a discriminative characterization, with candidate queries drawn from production logs ranked to align with what users may most likely issue next. Al- though effective (Ahmad et al., 2018), this strategy is inherently restricted by what is available in the logs, which in turn can penalize tail queries (De- hghani et al., 2017). In this work, we pursue the second strategy where query suggestion is cast as a natural language generation problem, aiming at pro- ducing effective continuations of the user’s intent by using generative modeling. For query generation, prior research has focused mostly on extending standard Seq2Seq models where the input is a concatenation of earlier queries a user has submitted in a session (Sordoni et al., Figure 1: An example search session where a user is- sues queries and optionally performs clicking at times- tamps 1 to n. At time n+1, the user issues qn+1 follow- ing the previous search context of length n. 2015; Dehghani et al., 2017). However, literature often leaves out the influence of clickthrough ac- tions (i.e., red blocks in Figure 1), which we argue should be taken into account in the generative pro- cess as they could be surrogates of the user’s im- plicit search intent (Yin et al., 2016). Users may ex- hibit diverse behaviors such as consecutively issu- ing queries without further engagement, or follow- ing up a single query with extensive clickthrough actions. These vastly different patterns are indica- tive of information pieces that the users find most relevant, which we conjecture can help producing suggestions better aligned to the user needs. We present an encoder-decoder Transformer model for the generative task that includes these patterns that we called behavioral hypotheses. One challenge that arises with Transformers is that they make minimal assumptions …

Content truncated to 500 words