The latest of the University of Siena spin-off companies, QuestIT srl, develops question answering technologies and, more generally, technologies for the automatic analysis of natural language.
The Liaison Office (the University reference point for academic entrepreneurship) has recently granted the necessary authorizations to establish the company. We asked Marco Ernandes, one of the people in charge of QuestIT srl, what motivated them to set up this spin-off company and which developments may arise from the study and implementation of these technologies.

Q. How was the idea of creating a spin-off company to study and analyse question answering technologies born?

A. We set up QuestIT srl only a few months ago, but we had the idea of forming such a company in 2006, in response to a series of incentives from both academia and industry. In the academic context, Web Crow, an automatic crossword solving system, had been set up on the basis of an Engineering Department research project and had twice received financing from Google. At the same time, we encountered growing interest in this technology by some leading companies in this field. The wish to engineer the project to create a product for public use was born by these external incentives and was made real by our collaboration with Expert System http://www.expertsystem.it (a leader in the computational linguistics sector and specialised in the technology of meanings) to realize question answering software, which is currently under development in the spin-off company.

Q. Question answering is born from the marriage of information technology and semantics, the study of the meaning of words. In practical terms, how does it work and what are the prospective uses of this technology?

A. Our study aims to set up software that accurately and automatically answers questions asked in natural language, with single or multiple answers to the topic. The software could substitute or, more probably, integrate search engines (which do not currently carry out any kind of linguistic analyses); when it answers a question, the new technology applies semantics to the data and information in its possession, providing not only the sole meaning of the words, but also the relations between words and their meanings. By interpreting the search queries, the question answering software could help an internet search engine formulate an answer, or act as a local filter within a series of documents, to provide a more accurate data retrieval. This would improve the quality of the information found and reduce search times, making it possible to acquire more precise information and reducing the amount of unnecessary information retrieved. Our current priority is not actually the use of question answering in the context of search engines, but the application of the software within businesses.

Q. How can this technology be applied in businesses? What are the prospective national and international applications?

A. Nationally the software could be used in businesses, such as public administrations, to reduce the problem of searching through large and complex documentary archives. By using the question answering software, for example, a public administration could provide a brief and practical guide to its staff so that they can quickly and effectively find the information they need among the huge amount of information available. This software could also substitute FAQ (Frequently Asked Questions), which are very costly to produce and always risk being incomplete. The main burden to question answering is language. In the USA large projects have received funding to study this technology in the last couple of years, while in Italy it is restricted to small experiments in academic contexts. As it would be impossible for a single company to provide such a service for all the languages of the world, each country will develop its knowledge and areas of competence independently. The USA will probably first produce an English version, then a Spanish and a German one. This will allow us to develop software for the Italian language without interference from the international market. The question answering software will use a combination of human and artificial intelligence. A basic version of the software will be released next winter and in a couple of years we aim to obtain quality certificate.

Consult the list of the University of Siena spin-off companies.

For further information please contact: the University of Siena’s Liaison office; tel.: +39 0577 232187; mail: liaison@unisi.it.

Torna alla prima pagina