JĘZYK W POZNANIU Ogólnopolska Studencka Językoznawcza Konferencja Naukowa
kng-jwp.home.amu.edu.pl

Karolina Jankowska (Uniwersytet im. Adama Mickiewicza w Poznaniu)

Trial use of the WordNet and NLTK as tools for keywords' synonyms based information search

As with the current information overload it is a great challenge to find the information that a user actually needs. There are numerous ideas how to overcome the problem with getting right to the useful sources without spending time reading huge volumes of data i.e. on the Internet. Among the most popular and relatively easy to conduct is context and synonyms-based searching. However, in order to perform such task a very good quality synonyms, collocation corpus is necessary. In fact there are many such tool available online, however not all of them are free to use or can be downloaded or synchronised with any software. A good example of ready to use and integrate English corpus is WordNet. On the other hand, there are many Natural Language Processing (NLP) tools available for linguistics in order to make their job easier and more effective such as Natural Language Toolkit (NLTK).
The purpose of this paper is to present the trial use of WordNet as a source linguistic data and the use of NTLK as a tool for effective NLP tasks. A simple code in Python programming language has been developed that performs searching text by a keyword and the keyword's synonym from WordNet. The lists of synonyms found via WordNet have been compared to results from Synonyms.com. Furthermore, for this very trial, a text corpus has been created. It contains 100 articles in the English language copied from available online sources, normalised and stored in a set of .txt files containing crucial metadata. The general conclusion from the experiment is that the results are not satisfying and searching based on synonyms only is of low value. 



Gdzie i kiedy: