Question Answering

This project will involve the implementation of artificial intelligence in order to answer questions posed to a computer which are written in natural language. Once a certain question is asked, the computer will answer it according to a pre-recorded database. If, however, the computer is incorrect in its response, or is not thorough enough, the user would be able to have the option to edit the program to include new or updated information that the user themselves are knowledgable of – thereby allowing the program to become more “intelligent” and to be able to answer more questions. In this way, the user would be able to teach to the program how to better itself.

I will probably be doing a very simple implementation of this myself. The database would just include  simple answers to simple questions which could then be edited/added to by the user.

One response to “Question Answering”

  1. Lekan

    This is interesting. However, there’s a few things to flesh out. How are you going to construct your database? This is difficult. You obviously do not want to store each questions as a string, because that doesn’t really capture the idea behind a question, so if someone asks “What’s 2+2,” you still want to be able to answer “What is 2 plus 2?” And once you store the knowledge, you do you find the right piece of knowledge? Finally, what would happen if you have conflicting answers to a question? Or conflicting matches to a question?

    It might be good to look at knowledge management, data ontologies, and also Bayes Nets for possible ways to think about this problem.

Leave a Reply

You must be logged in to post a comment.