Ontology-Based Automatic Text Summarization Using FarsNet

To summarize a text means to compress the text source into a shorter text in a way that the informational content is kept the same. With regard to the irregular volume of information available on the internet, manual summarization of huge volume of information by humans will be very arduous and difficult. There have been many activities in the field of automatic summarization so far. However, a lack of having methods capable of achieving a semantic hierarchy available in the documents is still felt. In this article we will propose a method for summarizing Persian documents which uses ontology for recognizing the semantic relationship between different parts of a text and extracting important sentences. For this purpose, mapping an input document with ontology brings about a graph whose vertices are the concepts available in ontology and its edges are the relationships among these concepts. This graph which is a graph-based representation of the input text comprises the necessary computational base for recognizing the important sentences in the input document and the production of a summary. The achieved results indicate the acceptable capability of the proposed method in obtaining semantic relationships available in documents and automatic text summarization. The FarsNet ontology is the base for this article.