An Enterprise Answering System for Collaborative Question Answering Platforms

点击翻译按钮获取中文摘要

Community Question-Answering (CQA) platforms, such as Stackoverflow, Quora, and Yahoo! Answers, have become popular sources of information over the past several years. As the number of questions increases, it becomes harder for users to track questions that are most relevant to them. As a result the quality of attention that different questions receive on the a platform becomes highly dependent on many contextual and social processes. Our industrial partner, IBM aims to become more responsive to communities of inquiry and addresses concerns and questions raised by customers regarding different IBM products effectively. IBM strives to automatically identify these threads of questions and addresses them effectively. While such questions might have an answer in internal enterprise resources or there is an IBM expert who is able to address the question effectively, such questions will be remained unanswered, thus increasing customer dissatisfaction and disengagement. Therefore, appropriate techniques are required that would look into questions posted by customers in the public domain and search internal resources (e.g., knowledge bases, internal expert knowledge) in order to efficiently address the customer's information needs and work towards their satisfaction.Our work will concentrate on formulating a multi-faceted framework that will collectively consider semantics of contents, analysis on available resources and effective allocation, and monitoring and predicting the quality of the provided answers in order to build a system that can extract the best answer effectively and efficiently. We believe that developing a coherent and effective QA system increases satisfaction for IBM's customers, and increases IBM's competitive advantage. As a result, it will be directly beneficial to Canada as more revenue for IBM Canada means more investment from IBM in Canada and hiring more Canadian talent.