A Study On The Consumption Intent Identification Approach For Cross-domain Chinese Microblogs

With the rapid development of web and social media platforms, the roles we are playing in the Internet are more divers ified than before. The platforms such as microblogging service have become the important part of the daily life. At the same time, the scale of electronic commercial enterprise is also increasing quickly. The comments and suggestions of the products from the microbloggers are of great value for marketing strategy and product design. As a result, indentifying the intention in the microblogs is critical.The microblogs are from different domains, we propose an approach that use transfer learning method based on semantic distance to resolve the cross-domain features, the method is used to indentify the intention of the mircoblogs. The semantic distance is about the calculation of the consumption, and the transfer learning method is designed for the cross-domain. This text classification method can improve the precision of the identification.Moreover, we use the dependent grammar structure and Pin Yin to analyze the Chinese microblogs and the typos cases, thus we can get more information from the feature.Finally, we conduct the experiment of identification of consumption of Chinese mircoblogs. We analyze the results and compare them with the related research results, which shows that the novel transfer learning method based on semantic distance is practical and the result is ideal.