On the role of information retrieval and information extraction in question answering systems

Question Answering, the process of extracting answers to natural language questions is profoundly different from Information Retrieval (IR) or Information Extraction (IE). IR systems allow us to locate relevant documents that relate to a query, but do not specify exactly where the answers are. In IR, the documents of interest are fetched by matching query keywords to the index of the document collection. By contrast, IE systems extrat the information of interest provided the domain of extraction is well defined. In IE systems, the information of interest is in the form of slot fillers of some predefined templates. The QA technology takes both IR and IE a step further, and provides specific and brief answers to open domain questions formulated naturally. This paper presents the major modules used to build IR, IE and QA systems and Shows similarities, differences and possible trade-offs between the three technologies.