Online reviews are essential to consumers' decision-making when purchasing products on e-commerce platforms. Most of the existing research conducts sentiment analysis on online reviews, yet the disclosure characteristics of these reviews have not received sufficient attention. Therefore, this paper evaluated the information characteristics of online reviews using review length, readability, redundancy, and specificity indicators. By collecting 18,131 online clothing reviews, we applied Latent Dirichlet allocation to divide the review texts into nine topics. We also investigate the relationship between review text characteristics and review sentiment and verify the robustness of the results using different regression models. We conclude that a review with more words, higher redundancy, lower fog index, and lower specificity tends to express a more positive emotion of the review. Our research will help e-commerce platforms construct general review writing guidelines to improve consumer satisfaction.