Question answering (QA) is a task which has as the main purpose to answer automatically to questions done in a natural language.
In order to do this, a pre-structured database may be used, or in contrast, a collection of natural language documents such as the World Wide Web. The questions asked, which may go from why questions to definitions, can be of two different types: either close-domain questions or open-domain questions. With regard to the first ones, the are considered easier to answer since, they belong to a topic which can be easily found on the Internet. However, the open-domain questions are harder to reply as the answers rely on world knowledge.
In spite of the amount of money invested, question answering systems such as “Powerset” or “Assimov the chatbot” still show that much work has to be done as most of the questions, can not be answered accurately yet.
References:
- Question answering. (2009, March 20). In Wikipedia, The Free Encyclopedia. Retrieved 09:29, April 1, 2009, from http://en.wikipedia.org/w/index.php?title=Question_answering&oldid=278611780
- Giuseppe Attardi, Antonio Cisternino, Francesco Formica, Maria Simi, Alessandro Tommasi. PiQaSso: Pisa Questions Answering System. Retrieved 4th April, 2009 16.30 from http://trec.nist.gov/pubs/trec10/papers/piqasso.pdf
- Matthias Rauterberg, Mario Menozzi, Janet Wesson, International Federation for Information Processing. “Human-computer interaction”. ISBN: 1586033638, 9781586033637. Published by IOS Press, 2003. Retrieved 4th April 2009, 16.52 from http://books.google.es/books?id=PTg0fVYqgCcC&printsec=frontcover#PPR5,M1Matthias
