Scientists from the Institute of Computer Science of the Polish Academy of Sciences (IPI PAN) and from the Institute of Informatics of the Wroclaw University of Technology work on a new web browser called Nekst. The IT specialists cooperate with linguists specialising both in Polish and English.
The project execution was assessed at almost PLN 15 million, including PLN 4.3 million in the budget of the Wroclaw portion.
- We do not intend to compete with the most globally popular commercial browsers; we do not have such a financial potential or technical possibilities, says dr Maciej Piasecki from the Wroclaw University of Technology. - Popular browsers treat words as sequences of characters; they do not analyse in any way the context or the content of the documents they find. They do not take into account the properties of the Polish language, for example the fact that words in Polish undergo inflection. In Polish relations between words are due to inflection, while in English they result from the position of a word in a sentence.
The Nekst system will employ several tools developed by the scientists. The first one is a Polish-language-oriented browser, called BEATKA (Polish for Beth) by the originators from IPI PAN, which takes into account proper names. It is capable of recognising connections between words and the context.
It is possible due to the construction of a number of tools and data bases for the Polish language (some are available at: www.clarin-pl.eu), e.g. a semantic dictionary Słowosieć (www.plwordnet.pwr.wroc.pl). The scientists have already elaborated 134 thousand words and 190 thousand meanings; the objective is to include approximately 200 thousand words. Słowosieć is a basis for automatic recognition of the word's meaning.
- When we enter a query with the word "zamek" (Polish for castle, lock and zip), On the basis of nearby words, Nekst is able to determine the context in which it is used, says dr Maciej Piasecki.
The remaining tools in the Nekst system are: a program answering questions in Polish, a browser of photos similar to the one given and a program for retrieving specific types of information automatically from text documents.
- Nekst will enable one to obtain quickly and more accurately answers to concrete questions in documents written in Polish and there are approximately half a billion of those registered and the number is growing on a daily basis. This will facilitate knowledge management and make it easier to solve specific problems, says the scientist from the Wroclaw University of Technology.
Dr Maciej Piasecki: - Scientists from the Meta-Share language research project warn in their report that within 30 years many European languages, including Polish, will shrink to the level of a local dialect. It is due to the expansion of English, also in IT systems. A tool which may slow down the process could be an IT system employing rules specific for the Polish language, the so-called basic linguistic technology. This is yet another asset of the Nekst system.
The new browser should be ready by mid 2014 r.
wto
