Semantic Web Mining by Using Clustering and Pattern Discovery

Abstract

The World Wide Web consists of different types of data present on websites and is in different formats. Different types of data include structured, unstructured and semi- structured data present in various formats. The aim of this research is to extract relational clusters from unstructured data based on sentiment by making use of natural language processing and semantic web technologies which include RDF format, FOAF ontology and OLIA ontology. Semantic web mining technologies help in converting data present online into machine readable form w.r.t ontological stand point or frameworks [45]. We use tweets in the unstructured form consisting of two columns such as person/account column and the tweet column. We convert data present into machine readable form by using natural language processing methods. The verbs extracted from data by using NLP methods are treated as predicates and the nouns/pronouns are treated as subject and object in the finalized table of person, subject, predicate and object resulting in triples. We acquire an RDF file with respective ontologies incorporated for creation of relations among triples. RDF grapher is used to visualize these relations. This study provides an in-depth analysis and implementation of how to discover meaningful patterns based on sentiment or feature the data present in unstructured form needs to be processed in-terms of machine readable form for the creation of relational clusters using ontological frameworks. The results of this study consist of ontological framework based relational data visualized in the form of clusters within clusters.