Unified Extensible Framework For Sentiment Analysis Based On Opinion Disambiguation

Information Technology

Sentiment Disambiguation is a vital component of any sentiment analysis or opinion analysis of the Data Mining domain. It plays a decisive feature to distinguish sentiment polarity of different text granularity. Another important point is to process the fine-grained large data sentiment analysis based on sentiment disambiguation lexicon to improve the accuracy of sentiment prediction. This system uses the Statistical Natural Language Processing computational techniques for the sentiment analysis. The existing system is based on a customized fuzzy method approach with two-stage training to deal with text ambiguity. The existing system classifies three types of cyberhate speeches: race, disability, and sexual orientation. In the existing system, the prediction accuracy has decreased when large data set contains huge diversity in the text. The existing system cannot achieve the efficiency of removing the ambiguity when there is intersectionality among a wide variety of cyberhate speech. The proposed system implements the process to determine the equivocal text into various categories toward Sentimental Disambiguation which narrows down the meaning of the words. The proposed system combines the sentiment label data and sentiment contrast data between focus keywords and context words into words vector representation learning model. The accuracy of the proposed prediction model will be very high even though the large diversity in the input text has fuzziness, ambiguity, and vagueness.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.