Chinese Futures Market Information Extraction and Classification Model Combined with HMM and SVM

Zhiming Yang, Heguo Yang, Guoping Zhao, Lingling Mu

Abstract: The Chinese futures information extraction and classification model can be used to analyze internet futures text, extracting text about the quotes and classifying them for analysis. At present, there is relatively little research on futures market analysis on network public opinion. Less attention is paid to the events that cause market changes. To this end, this paper proposes a Chinese futures market information extraction and classification model combined with HMM and SVM to help investors quickly understand the reasons for market changes. Firstly, feature extraction is performed on the original text through certain rules to highlight the key differences between the market description sentences and the other sentences. Secondly, the HMM model is used to process the feature sequence to obtain the output sequence about whether it is the market description. Then the SVM model is used to classify the sentence features that are judged as “market” in the output sequence. Finally, in the optimization module, the rules are used to optimize each market text to make the sentences more independent and readable. This model has a good performance on the dataset built by ourselves.