How can search engines better understand user intent? This paper introduces a novel query clustering method that leverages user logs to identify frequently asked questions or popular topics, essential for question-answering search engines. Unlike keyword-based approaches, this method analyzes user logs to determine which documents users select for a given query. The similarity between queries is then inferred from the common documents selected by users. Experiments demonstrate that a combination of keywords and user logs outperforms either method alone. The proposed query clustering method offers a significant improvement in understanding user intent and organizing search results. By incorporating user behavior data, search engines can provide more relevant and efficient responses to user queries.
This paper contributes to the journal's focus on information systems and search technologies. The research aligns with the journal's interest in innovative methods for improving information retrieval and organization, particularly in the context of question-answering systems.