A localization toolkit for sentic net

Xia, Yunqing; Li, Xiaoyu; Cambria, Erik; Hussain, Amir

Abstract

SenticNet is a popular resource for concept-level sentiment analysis. Because SenticNet was created specifically for opinion mining in English language, however, its localization can be very laborious. In this work, a toolkit for creating non-English versions of SenticNet in a time- and cost-effective way is proposed. This is achieved by exploiting online facilities such as Web dictionaries and translation engines. The challenging issues are three: firstly, when a Web lexicon is used, one sentiment concept in English can usually be mapped to multiple concepts in the local language. In this work, we develop a concept disambiguation algorithm to discover context within texts in the target language. Secondly, the polarity of some concepts in the local language may be different from the counterpart in English, which is referred to as language-dependent sentiment concepts. An algorithm is developed to detect sentiment conflict using sentiment annotation corpora in the two languages. Lastly, some sentiment concepts are not included in the local language after dictionary consulting and online translation. In this work, we develop a tool to extract these concepts from sentiment dictionary in the local language. Our practice and evaluation in constructing the Chinese version of SenticNet indicate that the proposed algorithms represent an effective toolkit for localizing SenticNet.

Citation

Xia, Y., Li, X., Cambria, E., & Hussain, A. (2014, December). A localization toolkit for sentic net. Presented at 2014 IEEE International Conference on Data Mining Workshop, Shenzhen, China

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 2014 IEEE International Conference on Data Mining Workshop
• Start Date: Dec 14, 2014
• End Date: Dec 14, 2014
• Online Publication Date: Jan 29, 2015
• Publication Date: 2015
• Deposit Date: Oct 10, 2019
• Publisher: Institute of Electrical and Electronics Engineers
• Pages: 403-408
• Series ISSN: 2375-9232
• Book Title: 2014 IEEE International Conference on Data Mining Workshop
• DOI: https://doi.org/10.1109/ICDMW.2014.179
• Keywords: Sentiment analysis, Sentic Net, common sense, localization
• Public URL: http://researchrepository.napier.ac.uk/Output/1792808