Automatic document summarizer

The need for automatic summarization becomes crucial with the exponential increase of data available on www and digital libraries, which makes the search for relevant pieces of information a difficult task. Therefore, needless to say, an automatic summarizer would save a lot to user and operators. However, the development of such systems is also shown to be very challenging due to inherent difficulty in dealing with natural language processing, the subjectivity of the evaluation process and the limitation of the mathematical models. This paper puts forward a proposal for an automatic summarizer system, which explicitly makes use of the semantic relatedness of document sentences using WordNet taxonomy.On the other hand, several other attributes are taken into account in the design stage, which include similarity to user's query, if any, which allows us to integrate some personalization to the outcome, similarity to the document title, location and frequency of co-occurrence. The developed summarizer also enables some search abilities, where three distinct search platforms were integrated (Lucene, GTP and Xapian). The obtained results were compared to MEAD summarizer.