Study And Implementation Of Product Platform System

With the rapid development of e-commerce, people will be faced the "information overload" problem. In order to improve the user experience, product search engine and personalized recommendation services need to carry out "information dimension reduction": mapping merchandise to its corresponding product. At the same time, integrating the information of the merchandise which corresponds to the same product can help data mining and knowledge discovery.We designed a product platform: it can automatically process massive amounts of online merchandise, identify products; automatic integration the product information, including product specifications and user comments; as a foundation platform for other e-commerce services.As we know, it may be the first automate product platform. The product warehouse is automatically built, updated and maintained. We leveraged the IR (information retrieval), NLP (natural language processing) technology to design key features of merchandise extraction algorithm and products identify algorithm. According to the characteristics of the online commodities' data, we studied and improved the property entity of the commodity title recognition and the attributes of the commodity reducing problems which belong to the feature extraction part, and used the experiments to prove the improvement. This paper designed and implemented a complete product identification algorithm and verified its effectiveness with the experimental results.