Architecture of a concept-based information retrieval system for educational resources

Resumo

In the literature, the bag-of-concepts representation of textual documents is regarded as a convenient alternative to the bag-of-words representation, since the words that users choose as search terms may differ from the ones that the author of a particular document chose for referring to the same concept¿thus reducing recall. Besides, the bag-of-words representation does not detect the differences of context in ambiguous terms, what reduces precision in search results. The objective of our research is to evaluate the applicability of the bag-of-concepts paradigm to information retrieval of educational resources. We built an information retrieval system that follows that approach and evaluated it with final users. The main contribution of this paper is the description of the architecture of the information retrieval system. First evaluation results show that the information retrieval system based on bag-of-concepts works well for retrieving educational resources. The practical implications of this research are that: it demonstrates that it is workable to build information retrieval systems based on bag-of-concepts and that they are efficient for retrieving educational resources. This makes them an a priori interesting alternative to be applied in other domains.