From search session detection to search mission detection

摘要: Search mission detection aims at identifying those queries a user submits for the same information need. Such knowledge offers interesting insights into behavioral usage patterns and often can help to better support user. However, most existing query log studies focus on search sessions only (consecutive need) ignore multitasking behavior (interleaved needs) as well hierarchies of short-term goals in multiple that form long-term task such vacation planning. To understand dialog between engine we distinguish (1) physical sessions, characterized by time gap queries, (2) logical consecutive need within session, (3) missions, behavior, hierarchical goals.Our contributions are threefold. First, present new algorithm session detection, which follows state-of-the-art cascading method's rationale combining effectiveness with efficiency. Our approach is applicable time-critical online scenario, where tries users incorporating about their history fly, offline objective accurately partition collected log. We improve several steps method, among others exploiting Linked Open Data information. Second, demonstrate our algorithm's applicability detect missions. Third, introduce publicly available corpus 8800 labeled