The Establishment of Commodity Success Model Based on Text Sentiment Analysis

The paper provides a solution to the problem raised by Sunshine. For the review mode of the Amazon online market, the article innovatively proposes the Level of Success model (LOS) to help the company determine whether the product is successful after sales. The model based on the review quantification model, the help level model, and the star model. The three sub-models are analyzed by AHP to obtain the weight vector, and the LOS model is obtained by linear combination. The review quantization model is through natural language processing (NLP) Sentiment analysis and quantification of the review text. The help level model combines quantifies helpful votes and total votes based on the Wilson score algorithm. The star model normalizes the review stars of customers.For the time-based processing of data that Sunshine Company pays special attention to the article divides the data and visualizes it according to the time period of the data. Through the analysis of images and correlations, the law of data and time correlation is obtained.Finally, the article builds an LDA theme model to extract the four themes of use time, product attributes, after-sales service, and product using, helping companies to get more information from the text. With the development of e-commerce platforms, customers are more and more likely to use e-commerce platforms. Through the LOS model and customer reviews, we can effectively evaluate the market effect of the company's products and these models are powerful tools to assist companies in making sales decisions.