A Vector Space Model Approach for Searching and Matching Product E-Catalogues

In e-procurement, companies use e-catalogues to exchange product information with business partners. The large variety of e-catalogue formats which are used by various companies make it difficult to match a product request from a buyer (buyer e-catalogue) with products e-catalogues. While, there are too many different standards for e-catalogues in use, often companies do not follow standard formats. Hence we often encounter a plethora of catalogue formats ranging from unstructured text to well-structured XML documents. One traditional approach to solve this problem is to convert different formats to a general common structure. But within this heterogeneous set of known or even unknown structures achieving a global structure is impractical. In this paper, vector space model has been used to measure the similarity ratio of providers’ e-catalogues with a buyer’s e-catalogue. Attributes of known structures and their values have been used as terms and their weights in the vectors to find the correlation of e-catalogues based on relationship of common tags. In order to associate the structures in calculating similarity, levels of attributes in xml documents are also included in the terms. Natural language processing is used to extract the same attributes from unstructured or unknown structured documents.