Carlos Tobar

This chapter addresses a question regarding relevant information in a social media such as a wiki that can contain huge amount of text, written in slang or in natural language, without necessarily observing a fixed terminology set. This text could not always be adherent to the discussed subject. The main motivation leads to the need of developing methods that would allow the extraction of relevant information in such scenario. A result system was designed upon ideas from the semantic Web combined with adaptive mechanisms and a modification of the classic vector model for information retrieval. The semantic information is not embedded in the media but within a structurally independent ontology. It was implemented using Java and a MySQL database. The objective was the achievement of, at least, 80% for recall and precision on the system results. The system was considered successful by achieving rates of 100% of recall and approximately 93% of precision.