Improving the performance of lexicon-based review sentiment analysis method by reducing additional introduced...

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Machine Learning

Background:

• Sentiment analysis methods often suffer from inherent sentiment bias, particularly lexicon-based approaches.
• This bias leads to skewed classification results, predominantly positive or negative, hindering accurate opinion extraction from user-generated content (UGC).

Purpose of the Study:

• To propose and evaluate a novel sentiment bias processing strategy for lexicon-based sentiment analysis.
• To enhance the performance and reduce the polarity bias rate (PBR) of sentiment classification frameworks.

Main Methods:

• Developed a sentiment bias processing strategy incorporating weight and threshold parameters learned from a small training set.
• Integrated these learned parameters into the lexicon-based sentiment scoring formula for review classification.
• Utilized SentiWordNet (SWN) as the sentiment lexicon and Amazon product reviews as experimental datasets.

Main Results:

• The proposed bias processing strategy significantly reduced the polarity bias rate (PBR).
• Experimental results demonstrated an improvement in the overall performance of the lexicon-based sentiment analysis method.
• The framework effectively addressed sentiment bias in classifying user-generated content.

Conclusions:

• The novel sentiment bias processing strategy is effective in mitigating bias in lexicon-based sentiment analysis.
• This approach offers a promising solution for improving the accuracy of opinion extraction from UGC.
• The framework provides a robust method for more reliable sentiment classification.