|
Abstract : |
We describe the IBM Statistical Question Answering for TREC-9 system in detail and look at several examples and errors. The system is an application of maximum entropy classification for question/answer type prediction and named entity marking. We describe our system for information retrieval which in the first step did document retrieval from a local encyclopedia, and in the second step performed an expansion of the query words and finally did passage retrieval from the TREC collection. We will also discuss the answer selection algorithm which determines the best sentence given both the question and the occurrence of a phrase belonging to the answer class desired by the question. Results at the 250 byte and 50 byte levels for the overall system as well as results on each subcomponent are presented. 1 System Description Systems that perform question answering automatically by computer have been around for some time as described by (Green et al., 1963). Only recently though have systems been developed to handle huge databases and a slightly richer set of questions. The types of questions that can be dealt with today are restricted to be short answer fact based questions. In TREC-8, a number of sites participated in the first question-answering evaluation (Voorhees and Tice, 1999) and the best systems identified four major sub-components:, |
