User Engagement Prediction for Clarification in Search

Clarification is increasingly becoming a vital factor in various topics of information retrieval, such as conversational search and modern Web engines. Prompting the user for clarification session can be very beneficial to system user’s explicit feedback helps improve retrieval massively. However, it comes with high risk frustrating case fails asking decent clarifying questions. Therefore, great importance determine when how ask clarification. To this aim, work, we model prediction engagement problem. We assume that better is, higher would be. propose Transformer-based tackle task. The comparison competitive baselines on large-scale real-life data proves effectiveness our model. Also, analyse effect all result page elements performance find that, among others, ranked list engine leads considerable improvements. Our extensive analysis task-specific features guides future research.