Automatic Extraction of Knowledge from Greek Web Documents

Extracting textual data from Greek corpuses poses additional difficulties than in English texts as inclinations and intonation differentiate terms of equal information weight. Pre-processing and normalization of text is an important step before the extraction procedure as it leads to fewer rules and lexicon entries, thus to less execution time and greater success of the mining process. This paper presents a system accessible via the Web which automatically extracts data from Greek texts. The domain of conference announcements is utilized for experimentation purposes. The success of the extraction procedure is discussed on the basis of an evaluative study. The conclusions and the techniques discussed are applicable to other domains as well.