Haar transformation in text analysis Using wavelet in sentiment analysis Text visualization using waveletSentiment and semantic analysis of a text are very important issues of today because of increasing text data.Our study proposes a new method to reveal the hypernym relations (generic term) of the words in the text and to enhance the accuracy of sentiment classification result of the texts. We used wavelet transform method that has been rarely used in text analysis.
Figure A. General Flow Diagram of Wavelet Sentiment ClassificationPurpose: The aim of the study is finding new approaches that uses wavelet transformation technique for text semantic representation and text sentiment analysis. We investigate the contribution of wavelet on the sentiment analysis classification problem. We used classical algorithms and hybrid wavelet algorithm for sentiment analysis problem.
Theory and Methods:The wavelet transformation, which is generally utilized for signal processing, is used for text semantic representation and text sentiment analysis. The frequency vectors that include term frequency of special words in the reviews are transformed to D matrix using a discrete wavelet transformation. The flow of the wavelet sentiment classification is seen on Figure A. We used 5 polarity classes, which are "high positive", "low positive", "neutral", "high negative", and "low negative" in our classification problem.
Results:Even wavelet is generally used for signal processing, it is seen that it is also useful for text semantic representation and sentiment analysis. In our study, to use wavelet transformation with classical classification algorithm provided % 3-5 increases in accuracy.
Conclusion:As a result, wavelet transform can be used in the semantic representation of the text. When the wavelet method is added to the classical sentiment classification problem, it helps to improve accuracy. It is expected that our method will contribute to text analysis problems and applications such as decision making and text applications containing hierarchical data.