Research On Agricultural Products Recommendation System Based On Sentimental Analysis

With the rapid development of the Internet,more and more goods can be purchased without leaving home,and there exist numerous sellers and all types of goods as well.As for the various goods,product recommendation is playing an increasingly important role.In terms of the diversified agricultural products,the appropriate recommendation algorithm can not only make customers be satisfied with agricultural products,but also ease the unsalable situation of them in the e-commerce field,which in turn enables merchants and customers to achieve a "win-win" goal.This paper mainly studies the recommendation of agricultural products from the following three factors: shelf life,sentiment analysis and shop rating.First,while ensuring the "high quality" of goods,priority is given to the selection of short-term agricultural products,thereby reducing the problem of slow sales caused by short sales periods of agricultural products.What's more,the sentiment analysis is a way to calculate the emotional value of the emotional words by SO-PMI algorithm evolved from PMI-VL,and then put the emotional value into the comment to calculate the emotional value of comment.After standardizing the data of shelf life,as an another factor,the result can be concluded as the fact that the shorter the life is,the larger the value is.These three unstructured association recommendation factors are the weights calculated by the analytic hierarchy process.The main work of this paper is as follows:1.Crawl the information about agricultural products on T-mall and clean the data through the crawler--Python2.Set apart the sentences of the comment into single words and mark the part-of-speech of the them;extract evaluation objects and emotional words related to agricultural products.3.Use the SO-PMI algorithm evolved from PMI-VL to calculate the emotional value of the emotional words in the field.4.The three recommended indicators of unstructured association are analyzed by AHP.5.The recommended ranking calculation is performed using the constructed agricultural product recommendation factor matrix and weight matrix.Promote the introduction of agricultural products with emotional values and shelf life.After using the Python crawler software to capture more than 300 agricultural product data and cleaning the review data,a total of 11258 valid comment data were obtained.After extracting emotional words with word dependence and part of speech,PMI-Vl algorithm is used to analyze sentiment words.The positive and negative emotional word recall rates are 80.7% and 83.4%,respectively.It is concluded that when the recommended factors of the shelf life are added to the recommendation,the recommended agricultural products can not only ensure the quality of the products,but also give priority to the products with relatively short shelf life,and reach the conclusion of the purpose of pushing the products to shorten the shelf life.