2017/08/10 by Mostafa Dehghani, Sascha Rothe, Dehghani, Mostafa +5 · 4 citations
Computer Science · #Advanced Image and Video Retrieval Techniques #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Information Retrieval and Search Behavior #Topic Modeling #Web Data Mining and Analysis
paper · pdf · doi:10.48550/arxiv.1708.03418
openalex publication_date 2017/08/10 · openalex created_date 2022/10/05 · openalex updated_date 2026/07/28
Users try to articulate their complex information needs during search\nsessions by reformulating their queries. To make this process more effective,\nsearch engines provide related queries to help users in specifying the\ninformation need in their search process. In this paper, we propose a\ncustomized sequence-to-sequence model for session-based query suggestion. In\nour model, we employ a query-aware attention mechanism to capture the structure\nof the session context. is enables us to control the scope of the session from\nwhich we infer the suggested next query, which helps not only handle the noisy\ndata but also automatically detect session boundaries. Furthermore, we observe\nthat, based on the user query reformulation behavior, within a single session a\nlarge portion of query terms is retained from the previously submitted queries\nand consists of mostly infrequent or unseen terms that are usually not included\nin the vocabulary. We therefore empower the decoder of our model to access the\nsource words from the session context during decoding by incorporating a copy\nmechanism. Moreover, we propose evaluation metrics to assess the quality of the\ngenerative models for query suggestion. We conduct an extensive set of\nexperiments and analysis. e results suggest that our model outperforms the\nbaselines both in terms of the generating queries and scoring candidate queries\nfor the task of query suggestion.\n