Sentiment Classification for Chinese Micro-blog Based on the Extension of Network Terms Feature
Sentiment analysis is widely used in product reviews, movie reviews, and micro-blog reviews. Micro-blog review is different from general commodity or movie reviews, which often contains the user’s randomness and lots of network terms. So the micro-blog re
- PDF / 196,961 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 91 Downloads / 165 Views
Abstract Sentiment analysis is widely used in product reviews, movie reviews, and micro-blog reviews. Micro-blog review is different from general commodity or movie reviews, which often contains the user’s randomness and lots of network terms. So the micro-blog reviews emotional analysis is not a small challenge. Network terms generally express strong emotions or the user’s point of view, the traditional bag words model and machine learning method do not use the network terms features. In the face of ever-changing micro-blog reviews manifestations, forecast accuracy may be affected. Therefore, in this paper our study focuses on the micro-blog emotional analysis through the extended network terms features and integration with other features. We are taking experiments to compare prediction performance under the different feature fusions, to find out which feature fusion can get the best results. Our results show that by the extended network term feature integration with other features ways to improve the accuracy of predictions, especially some of the most popular micro-blog reviews. Keywords Text sentiment classification learning
⋅
Support vector machine
⋅
Machine
1 Introduction With the development of Internet, network online user reviews have become an important way people express opinions and disseminate information. The reviews are shared, real-time, and interactive. The dissemination of diversification reflects the profound degree of public preference for a certain type of theme or a certain commodity. Reviews of one kind of topic have become an important reference on the choosing of purchase goods. Therefore, emotion analyses of reviews gradually become a hot research direction. Micro-blog is a relationship based on user F. Ye (✉) Southwest Jiaotong University, Chengdu, China e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 S.K. Bhatia et al. (eds.), Advances in Computer and Computational Sciences, Advances in Intelligent Systems and Computing 554, https://doi.org/10.1007/978-981-10-3773-3_22
231
232
F. Ye
information sharing, dissemination, and access platforms. Analyzing the sentiment tendency of micro-blog is very useful to company. You can get the user’s attention on a topic by studying these reviews, as well as the degree of preference for a movie or a commodity. Companies can use them to study consumer satisfaction with the products, so as an important reason for product improvement [1]. The sentiment classification of the micro-blog reviews can also predict the future fluctuations in stock prices [2]. Governments can use them to survey public opinion. In this article, the main focus of our study is the importance of networking terms in micro-blog reviews to judge the emotional tendencies, and analysis on the relationship between network terms and other emotional words. The contributions of this article are as follows: • Propose the optimization method of word segmentation and use word2vec tool to extend the original lexicon, and to construct expansion lexicon. • By collecting popul
Data Loading...
