Host Organisation: Faculty of Informatics, TU Wien, Vienna, Austria

Professionals working in law, healthcare, and intellectual property have specific requirements for how the search engines that they use should support their work, and specific patterns of searching that differ significantly from commonly studied web search patterns. In particular, the search sessions (sequences of queries related to the same information need) are usually longer in domain-specific search, often with more detailed queries. Currently, search solutions tend to be developed in isolation for each domain. This project will investigate a more integrative approach to developing solutions for multiple domains. The objectives of the project are:

• model information and search patterns in legal, intellectual property, and healthcare search.
• provide an understanding of the differences between domain-agnostic and domain-aware information flows, as well as of information flows in the three domains under investigation.
• develop information processing and search methods (including deep neural networks) that automatically take domain and context into account.
• develop evaluation benchmarks that combine legal, intellectual property, and healthcare search, to demonstrate the adaptability of the developed methods
• develop a methodology for domain-adaptation in further domains.

This project will produce the specification for a suite of tools needed to assist people in extracting information from and making use of search results, and additionally augment existing information interaction models of the search process.