Learning Regular Expressions for the Extraction of Product Attributes from E-commerce

Abstract. A large number of e-commerce websites have started to markup their products using standards such as Microdata, Microfor-mats, and RDFa. However, the markup is mostly not as fine-grained as desirable for applications and mostly consists of free text properties. This paper discusses the challenges that arise in the task of matching descriptions of electronic products from several thousand e-shops that o↵er Microdata markup. Specifically, our goal is to extract product at-tributes from product o↵ers, by means of regular expressions, in order to build well structured product specifications. For this purpose we present a technique for learning regular expressions. We evaluate our attribute extraction approach using 1.9 million product o↵ers from 9,240 e-shops which we extracted from the Common Crawl 2012, a large public Web corpus. Our results show that with our approach we are able to reach a similar matching quality as with manually defined regular expressions