Understanding comparative questions and retrieving argumentative answers

Making decisions is an integral part of everyday life, yet it can be a difficult and complex process. While peoples’ wants and needs are unlimited, resources are often scarce, making it necessary to research the possible alternatives and weigh the pros and cons before making a decision. Nowadays, the Internet has become the main source of information when it comes to comparing alternatives, making search engines the primary means for collecting new information. However, relying only on term matching is not sufficient to adequately address requests for comparisons. Therefore, search systems should go beyond this approach to effectively address comparative information needs. In this dissertation, I explore from different perspectives how search systems can respond to comparative questions. First, I examine approaches to identifying comparative questions and study their underlying information needs. Second, I investigate a methodology to identify important constituents of comparative questions like the to-be-compared options and to detect the stance of answers towards these comparison options. Then, I address ambiguous comparative search queries by studying an interactive clarification search interface. And finally, addressing answering comparative questions, I investigate retrieval approaches that consider not only the topical relevance of potential answers but also account for the presence of arguments towards the comparison options mentioned in the questions. By addressing these facets, I aim to provide a comprehensive understanding of how to effectively satisfy the information needs of searchers seeking to compare different alternatives.