A Fuzzy Based Approach for Empirical Analysis of Unstructured Data

Abstract

Computational intelligence and soft computing has many promising technologies such as Text Mining. Document Classification using soft computing techniques like fuzzy logic helps to find a more practical solution due to ambiguity and uncertainty present in the text data. Uncertainty and information may be reflected as the part and parcel of any industrial or engineering problem to be solved. Information refers to the facts required to solve it and uncertainty refers to the non-random lack of certainty (‘non-random uncertainty’), ambiguity, haziness in the system. It is very important to ponder on the nature of uncertainty involved in a problem. Father of fuzzy logic, Lofti Zadeh (1965) suggested that decision-making using set membership is the key when it is required to deal with uncertainty. Fuzzy clustering helps to identify patterns which are difficult to be discovered using crisp clustering. Natural languages contain non-random uncertainty. To deal with non-random uncertainty or different degrees of truth or partial truth Fuzzy logic may be used. This work focuses on fuzzy logic based approaches being utilized for identification of coherent patterns. Empirical Analysis are conducted to realize and evaluate the effect of the methodology proposed.