Do neural-network question answering systems have a role to play in the deployment of information systems?

Submitted by Elaine Almeida (elaine.almeida@nce.ufrj.br) on 2017-05-09T13:54:50Z No. of bitstreams: 1 04_06_000668091.pdf: 555812 bytes, checksum: 4a299e92692f879bb54b8f6c9b45b7c7 (MD5)Made available in DSpace on 2017-05-09T13:54:50Z (GMT). No. of bitstreams: 1 04_06_000668091.pdf: 555812 bytes, checksum: 4a299e92692f879bb54b8f6c9b45b7c7 (MD5) Previous issue date: 2006-12-31As Internet users become more numerous, experienced and skillful, and the number of companies doing ecommerce increases worldwide, so does the demand for online information about products and services. To satisfy this increasing demand for on-line information many companies have resorted to providing customer support services over a variety of on-line means of communication such as e.mail, chat services, voice on internet protocol (VOIP), etc. This article presents a stepwise approach to the construction of hybrid question answering systems based upon neural network technologies and natural language processing. These special kind of information systems not only provides high speed answers to questions posed by customers, but they also allow customers to receive answers to their questions on a 24/7 basis, provide well conceived standard answers to those questions, allow for a precise recording of customer communication, and make the management of customer support services easier. All of this is made clear by a case study about the development of an automatic question answering systems to the “SEBRAE Challenge”, a business game involving university students in seven different countries in South America