Because of the popularity of information technology and internet development, the news media in the report more immediate and pluralistic, but after the release of news on the authenticity of text content and reference standards are uncertain, and through different words to influence investors' investment decisions, so when investors receive network news, easy to receive the text content of media news and produce information asymmetry problems, and thus derive investment adverse selection risk. Therefore, based on the concepts of artificial intelligence (AI) and big data, this study constructs an investment decision evaluation model for an auxiliary investor after receiving media news release. The research sample used the keywords " Cathay Holdings (2882)" published by China Times as an example, using the historical stock price data of the Taiwan Stock Exchange to categorize news text, and then to break news text through Jieba word breaker technology. Based on the weighted technology TF-IDF to obtain the weight of each word of news text to build a "key vocabulary-text matrix", and then by Cathay Holdings historical closing price to establish the market reaction after the news release. Finally, apply the Support Vector Machine (SVM) model of machine learning to analyze and predict. The results show that the keywords analyzed in this study have no influence on the market reaction as a whole.
