Email Ontology Learning System Based on Fuzzy Logic

Natural languages all over the world are contextbased. There is inherent vagueness present in the common usage of language, but it is easily understood by people because they can decipher the context of the communication easily. Emails are an extension of face-to-face communication and many elements of ambiguity are present in the language used within the body of emails. In a computer based application to extract the ontologies for multilingual emails, it is important to disambiguate the sense of the verb, noun and adjective in a sentence develop correct hierarchies of relations and concepts. We propose to use a fuzzy logic based system as it is quite suitable for identifying and addressing the vagueness and ambiguity in the natural languages in the context of email ontologies. For various domains of emails like ‘Meeting & Schedules’, ‘Comments &Discussions’ etc., we have developed fuzzy rules and corresponding associations. These are used to develop an efficient learning model for email ontologies and incorporate in the JAPE pattern rules for extraction of appropriate and context aware concepts and relations. Implementation of fuzzy logic improves the efficiency and simplicity of the design process. In this research work we have tried to tackle the ambiguity of pronouns which act as important agents in ontology development. This results in accurate representation of ontologies in various domains of emails.