Fuzzy Conceptualization Model for Document Representation

The Fuzzy Conceptualization Model (FCM) performs a fuzzy mapping of query set to the given word corpus to obtain fuzzy membership degree for a document based on semantic correlation among words quantified by cosine similarity measures between word embeddings of queries and word corpus. A fuzzy membership function is defined for the given search query with word attributes of the given corpus. The generated numerical vector representation is the measure of similarity between word attributes of document and base terms set. The ground truth fuzzy set obtained for the popular terms in the given data corpus is used for creating semantic table for the base term cluster and hence formulates a matrix representation for both accurate and semantically related terms to be getting processed and counted together with out much over head to typical Fuzzy Bag of Words (FBoW) models and its variants. The proposed FCM outperforms other comparative models with high confidence level under paired t-test. FCM performs well when the mapping bound λ is zero since the increased λ values increases the sparsity of the vector representation of the documents.