Vertical Search Relevance.
As the web information exponentially grows and the needs of users become more specific, traditional general web search engines are not able to perfectly satisfy the nowadays user requirement. Vertical search engines have emerged in various domains, which more focus on specific segments of online content, including local, shopping, medical information, travel search, etc. For instance, a user may want to find a restaurant near his/her current location, search professionals on LinkedIn, and look for products on EBay. Recent studies show that Web queries have various vertical intents (e.g., 20% of them are local intent), and even more in mobile queries. As a response, vertical search engines start attracting more attention while relevance ranking in different vertical search engines is becoming the key technology. In addition, vertical search results are often slotted into general Web search results. Hence, designing effective ranking functions for vertical search has become practically important to improve users' experience in both web search and vertical search.
A natural way to build a vertical search engine is to apply the existing ranking techniques on a vertical (e.g., Local). However, we observed that in many verticals, the meaning of relevance is domain-specific. Blindly applying the conventional learning to rank approaches by ignoring vertical-specific domain knowledge (e.g., distances, ratings, and prices) may not be effective. We have identified a list of challenging research issues in the field of relevance for vertical search, which mainly fall into two categories. The first category includes how to learn an effective ranking model considering multi-facet relevance (e.g., text relevance, distance, ratings, and prices): (1) integrating multiple aspect relevance: vertical search engines need to consider and tradeoff the relevance from different aspects before making the overall relevance judgment; (2) query-dependent multiple aspects: such a tradeoff can vary for different queries or in different contexts (e.g., searching a chain-store like CVS is more location-sensitive than a specific restaurant); (3) the involved aspects and the tradeoff among them are vertical-dependent. Collecting training data with overall relevance for a new vertical requires human editors learn how to appropriately tradeoff different aspects. The second category focuses on building effective business model in the context of specific vertical search systems. Comparing with the business model in web search, some vertical search engines (e.g., shopping and travel) are naturally business-oriented, which enables flexible business models into ranking with awareness of relevance of original queries.
The workshop will bring together researchers from IR, ML, NLP, and other areas of computer and information science, who are working on or interested in this area. It provides a forum for the researchers to identify the issues and the challenges, to share their latest research results, to express a diverse range of opinions about this topic, and to discuss future directions.