چكيده به لاتين
Abstract:
Today text clustering is an important task in varies fields of science such as opinion mining, latent semantic analysis, social networks and etc. All of these text information are unstructured format, therefore development of methods for automatically clustering is very important, although there are lots of researches about clustering technique, but this is still a new concept and new area of study so on this thesis we introduce a new novel approach in text clustering.
In recent years ago clustering approach use Bag_of_Words model, instead of this traditional model we want to convert all text corpus and their documents into Graph_of_Words model. Unlike other approach we focus on related co-occurrence words in documents and convert those terms into graph with nodes as term and their relationships as edge.
After calculate centrality measure for all nodes, each node get a new weight. To evaluate our approach we use vector space model as input parameter for K_Means clustering algorithm, therefore after generating results we will see our approach will cluster all text document better than traditional model.
Keywords: Text Clustering, Graph, Text Mining, Centrality Measure
