A Judgment Method of Network News Value Orientation Based on Sentiment Analysis

Abstract

Nowadays, recommendation algorithms are playing an increasingly important role in online news platforms. Current personalized recommendation algorithms aim to find connections between user characteristics and news to be recommended, so as to achieve accurate recommendations. The goal of the personalized recommendation algorithm is to increase the click-through rate, which faces the problem of excessively catering to user interests, and may even recommend content that does not conform to mainstream values in order to satisfy users' curiosity. Considering that the emotional tendency of news reflects a certain value orientation, we propose a judgment method of network news value orientation based on sentiment analysis, calculate the objective value representation and subjective value representation of news through the sentiment analysis model, and analyze these two representations by Kano model to judge the value orientation. The judgment of value orientation with the help of sentiment analysis model can effectively reduce the audit of online news.