网红营销基于情感分析的消费评论数据的挖掘

Influencer Marketing Based on Consumer Comment Data Mining of Sentiment Analysis

随着2020年疫情爆发对经济的影响,常规的商业营销模式受到了冲击。以网红营销为代表的电子商务迅速崛起。电商平台作为网红营销的主要渠道,成为了消费者挑选、评价电商产品的媒介。因此,在电商平台的消费者评论具有文本挖掘的价值和潜力。本文使用网络爬虫技术,爬取到49,700条天猫商城评论数据其中有效数据为40,138条,采用TF-IDF算法提取评论中的关键词,通过Basic LDA模型对商品评论进行属性抽取,并运用变分贝叶斯推断对模型进行求解,最后通过朴素贝叶斯对评论进行情感极性分析,最终得到结论:消费者对电商产品的评论从功能性、消费理性两个角度出发,主题1、3、4表现出消费者重视电商产品使用的直观感受的倾向,主题2表现出消费者重视电商产品实际解决需求的功能性的倾向,消费者对电商产品的情感得分里积极、中性、消极分别为36%、63%、1%,总体上对电商产品持认可态度。

With the economic impact of COVID-19 since 2020, conventional business marketing models have been hit. E-commerce, represented by influencer marketing, has risen rapidly. As the main channel of influencer marketing, e-commerce platform has become a medium for consumers to select and evaluate e-commerce products. Therefore, consumer evaluation in e-commerce platform has the value and potential of text mining. In this paper, the web crawler technology is used to crawl 49,700 tmall review data, 40,138 of which are valid data. TF-IDF algorithm is used to extract key words in the review, Basic LDA model is used to extract attributes of the product review, and variational Bayesian inference is used to solve the model. Finally, naive Bayes was used to analyze the emo-tional polarity of the comments, and the final conclusion was drawn: Consumers’ comments on e-commerce products are from the perspectives of functionality and consumption rationality. Theme 1, 3 and 4 show consumers’ tendency to attach importance to the intuitive feeling of using e-commerce products, while theme 2 shows consumers’ tendency to attach importance to the func-tionality of e-commerce products to solve actual needs. The positive, neutral and negative senti-ment scores of consumers to e-commerce products are 36%, 63% and 1%, respectively. In general, consumers approve of e-commerce products.