Using Fuzzy Logic for Product Matching

Product matching is a special type of entity matching, and it is used to identify similar products and merging products based on their attributes. Product attributes are not always crisp values and may take values from a fuzzy domain. The attributes with fuzzy data values are mapped to fuzzy sets by associating appropriate membership degree to the attribute values. The crisp data values are fuzzified to fuzzy sets based on the linguistic terms associated with the attribute domain. Recently, matching dependencies (MDs) are used to define matching rules for entity matching. In this study, MDs defined with fuzzy attributes are extracted from product offers and are used as matching rules. Matching rules can aid product matching techniques in identifying the key attributes for matching. The proposed solution is applied on a specific problem of product matching, and the results show that the matching rules improve matching accuracy.