A Product Classification Algorithm Based On Vector Space Model

In the recent yeas, the electronic commerce becomes more and more important along with the development of the Internet. Suppliers and consumers are all satisfied with the convenience brought by the electronic commerce. The electronic commerce is different with the traditional trading mode in many styles. Classifying the products effectively in the electronic commerce environment is an important technical skill, which could bring more convenience to the suppliers and consumers.The definition of the Product Automatic Classification is that finding out the most proper candidate category from a set of pre-defined categories by an unknown product's information. The product classification algorithm is an important research point in the electronic commerce. It helps the electronic commerce web site and consumers to analyze the information of the products. Recently, the product classification technical has been combined with the search engine, machine learning and information retrieval. The product classification algorithm based on the SVM could classify products by the terms in their information, so the algorithm's precision was improved greatly. The product automatic classification has experienced three steps from its appearance: based on rules, based on statistics, based on both rules and statistics.The following topics are discussed in the paper: firstly, we introduced the present researches of the product classification algorithm. Then we brought the technical skills about the product classification, which includes the classification model, terms filter and evaluation method. After that, the AutoCat algorithm, tree structure and DAP product category models are analyzed in detail. Further more, this paper brought an improved classification algorithm based on the similarity. Finally, we set up an experiment to analyze the efficiency of this algorithm.