The development of artificial intelligence provides a new tool for the optimization and improvement of traditional bid evaluation methods. To tackle the problems of low manual bid evaluation efficiency and difficulty in identifying bidding collusion, this study proposes a comprehensive evaluation model for bid documents based on text analysis, including a text evaluation model and a text rating model. The proposed model can provide support for a more objective, scientific, and intelligent bid evaluation of construction projects and the prevention of bidding collusion. First, a text evaluation model is built. The repetition rate of bid documents calculated by the Shingling algorithm is added to traditional bid evaluation indicators. Then, the template catalog required by the bid documents is compared with the real catalog for the calculation of the response level of the bid documents. The analytic hierarchy process is employed to determine the weight of the text evaluation index. Next, a text rating model is built. The similarity between bid documents is calculated by the weight-improved Simhash algorithm. The comprehensive rating is performed with corporate relevance, consistency between qualifications and quotations, price floating, abnormal behavior, etc. as rating indicators. The rating results of bid documents are helpful for bid evaluation experts to identify bidding collusion. Finally, a quantitative calculation is conducted with the text evaluation model to yield the bid score ranking, and a qualitative analysis is made with the text rating model to present the results of identifying bidding collusion. Taken together, the two realize the comprehensive evaluation of bid documents.