Extracting Business Rules from Web Product Descriptions

摘要

Products sold in e-commerce sites are usually attached with conditions, policies and provisions, which are sometimes important criteria for buyers to compare products. Although logical languages dealing with business rules have been proposed, most of business rules on the Web are still given in natural language texts only. In this paper, we discuss technologies for locating those rules in product descriptions, categorizing them, and translating them into a formal constraint language. We take airline tickets as an example, and discuss what processes in this translation can be automated. By translating into the constraint language, we can search, compare and verify complex conditions.