ENHANCED LDA BASED INPUT SELECTION TOWARDS THE SENTIMENTAL ANALYSIS
Big data creates considerable challenges for businesses due to its complexity. The fundamental challenge that firms face is processing and storing large amounts of data. In addition, strategies for dealing with a befuddling amount of raw data in various forms must be enhanced. It is also vital to develop scalable data storage in order to efficiently acquire and retrieve critical information. The importance of feature selection in today's society cannot be overstated. Feature selection is one of the most important factors that can affect classification accuracy. If the dataset has a large number of characteristics, the space will be huge and congested, reducing the classification accuracy rate. It is possible to employ a method that is both efficient and reliable for removing noisy, irrelevant, and redundant data. Then it's just a matter of determining what sentimental analysis entails. Single words from a text document can be utilized as features, or more complex pairings can be retrieved using a variety of approaches that add more information to the feature-document matrix representation. The huge number of properties and relationships that diverse feature types hold, however, causes the high dimensionality problem. As a result, feature selection helps to build effective and efficient sentiment analysis applications by selecting relevant and informative features to enhance classifier performance while reducing processing load. In this work, Enhanced LDA-based feature selection has been applied. LDA is one of the generative statistical models. In basic, the Latent semantic analysis has been the most widely used distributive model with the singular value decomposition. LDA is utilized for removing points from text that empower effective preparing, particularly for huge information analysis.