基于结构对应学习的跨语言情感分类研究

Research of cross-language sentiment classification based on structural correspondence learning

情感分类的主要目的是预测用户在互联网中发布情绪数据的极性(积极的或者消极的),各种语言的情感分析已经成为诸多应用的研究热点,然而由于不同语言的情感资源在质量和数量上的不平衡,通常使用源语言来改善目标语言的跨语言情感分类方法,来提高目标语言情感分类的准确性.传统的跨语言情感分类主要是通过机器翻译将目标语言映射到源语言中,但是分类的准确性严重受到机器翻译质量的影响.通过对跨领域文本分类的结构学习算法(SCL)的讨论和拉普拉斯映射对两种语言之间词对的影响,对跨语言结构对应学习算法(CLSCL)的改进,进而提出M-CLSCL算法,借助选出来的轴心词对来进行目标语言的情感分类,通过M-CLSCL方法与前述相关方法的实验结果进行比较,可以发现M-CLSCL提高了情感分类的准确性.

The main purpose of sentiment classification is to predict the sentiment polarity（positive or negative）of the data that the user publishes the emotional data on the Internet.The sentiment analysis of various languages has become a hotspot of many applications,but because of the different quality of sentiment resources and quantity on the imbalance,the researchers usually use the source language to improve the target language cross-language sentiment classification method to improve the accuracy of the target language sentiment classification.The traditional crosslanguage sentiment classification mainly maps the target language to the source language through machine translation,but the accuracy of the classification is seriously influenced by the quality of the machine translation.This paper discusses the structure of the cross-language（CLSCL）,and then put forward the M-CLSCL algorithm,by using the selected axis of the word to the target language of the language of the target language sentiment classification.Com-paring the experimental results of M-CLSCL method and the relevant methods,we find that M-CLSCL has improved the accuracy of emotional classification.