Association Analysis of Semi-structured Data for Discrimination Discovery in Business

Data mining techniques have taken a critical role in life in numerous domains such as consumer analytics, finance, banking, medicine, biology, and astronomy... Recently, data mining techniques have found their application also in discovering illegal discriminatory treatment on the bases of sensitive attributes such as race, color, religion, nationality, gender, age... In this paper, we propose a framework to solve the discrimination matter in the context of semi-structured business data, and in particular in the calculation of taxes for imported goods. This framework is able to discover possibly discriminatory relations among data by finding discriminatory association rules with the support of a common sense knowledge base and text mining techniques. The framework has applied to the problem of HTS (US Harmonized Tariff Schedule) showing some satisfactory results.