Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 1

In the online sales market, the feedback information within product reviews is one of the most

important sources that helps enterprises to acquire users’ demands accurately. Considering the
exploitation and application of product review information in the Amazon market, a product
review model was proposed based on massive feedback data and utilized semantic mining
techniques, the Bayesian average algorithm, and the idea of AR, where the rules of the sales
market were investigated under three application scenes.

When solving the quantitative problem of product feedback data indicators, the comment text data
were first preprocessed. The TF-IDF index was then used to extract review features, and the
Word2Vec model technology was combined to analyze the emotional tendency of reviews. After
the previous steps, the sentiment scores were obtained. The Wilson interval method was
innovatively exploited for the helpful voting data to solve the problem of sample imbalance, and
the quantifying model was constructed to analyze marketing data of three new Sunshine Company.

In order to study the quality evaluation of products in the market, the deeper length of review and
the review credibility related to the reviewer's identity were taken into consideration, and the
comprehensive score of a review was obtained by using the principal component analysis (PCA)
method. After then, the product quality evaluation model was constructed based on the Bayesian
average algorithm, whose scores can become more and more precise with the increasing number
of reviews. This can provide convenience for the company tracking the sales of products.

Aiming at the evaluation problem of product real-time competitiveness, the time variable was
introduced and the monthly cumulative clout of each product was counted. With the help of the
AR (2) model, the product clout scores were fitted, based on which the product competitiveness
evaluation model was established. This model was utilized to predict the future competitiveness
trend of the product in the market, formulate different marketing plans for possible future
situations. It turned out that our outcomes were verified to be at a high level of accuracy.

To help the company explore the rules of the sales market, three application scenarios were
designed for the product evaluation model. The monthly competitiveness scores were cumulated,
which was set as the standard of evaluating product potential success. The temporal causality
between star ratings and reviews was then explored. It was discovered that the low Star-rated
reviews were more likely to trigger an increase in the number of reviews. Finally, the strong
correlation between specific descriptors of review texts and rating levels was investigated.

In the end, the research results were summarized into a two-page report, giving Sunshine
Company our most honest and credible recommendations. Our model has strong practical
significance and wide range of application, which can help e-commerce company operators to
control the market trend, seek a business service model that is close to the real haineeds of users,
and achieve a "win-win" between customers and e-commerce companies.

You might also like