Sentiment analysis of travel reviews is actually to judge the emotional polarity of travel review texts, which can provide important references for potential tourists to choose tourist attractions. At present, there are few researches on the analysis of sentiment orientation of tourism reviews at home and abroad. Some sentences in the review text have no contribution to the sentiment orientation of the text, and in the sentiment tendency analysis studies，distributed word representation is mostly used, which ignore the sentiment information of the word. Therefore, this paper improves the feature word extraction according to the dictionary and semantic rules, and proposes a representation method of weighted word vector integrated with sentiment information and improved TF-IDF algorithm, which is applied to the task of sentiment analysis of tourism review text. Through the experiment on the collected data of tourism reviews in Hebei Province, the proposed method is compared with the four groups of control experiments set in this paper, the experimental results show that the proposed method has higher accuracy, recall and F value.