Context of the Workshop
User-generated content and social media are increasingly exploited to support content mining, recommendation, social structure analysis, etc. However, for some kinds of online community sites such as question-and-answer sites and forums there is no explicit social network structures but instead specific discussion structures (e.g. threads) and contribution types (e.g. questions, answers, edits) that are traces of the activity of often implicit communities of interest. Community Question Answering sites (CQA) and question-answering services (Q&A) are now an important part of the information sources searched and accessed by users today to solve problems and take decisions. A famous example is the StackOverFlow website. Question-answer sites and services serve millions of users and generate an important and popular part of the Web content. In this context, being able to efficiently analyze, structure, search, automate and reuse question-answering data is an important challenge.
Goals and motivations
This workshop intends to bring together researchers and practitioners of Question Answering sites and services on the Web to present and discuss latest advances in analyzing, supporting and automating tasks of the life-cycle of such applications. The goal is to cover and bring together the different approaches existing in managing and answering natural language questions of users on the Web. This includes methods, models and algorithms from automated question-answering to question-answering forums mining, monitoring and management automation.
Topic and themes
- question routing, question answering
- question and answer recommendation
- question and need analysis, question modeling
- expert finding, expertise categorization, expertise labelling
- debate analysis, argument mining, argument schemes
- moderating support and automation
- animation fostering, targeted solicitation and notification
- answer generation, multiple and/or heterogeneous answer sources
- answer detection and ranking, best answer identification
- answer building and answer improving
- questions and expert topic labelling
- role detection (questioner, answerer, editor, etc.), user modelling
- social aspects of question answering
- fact checking, cross-validation, supporting evidence
- spam or abuse prevention in questions and answers
- answer personalization
- challenges, datasets, benchmarks for question-answering evaluation