Abstract: Thesis quality is a key index to evaluate researchers’ work. Currently, thesis quality evaluation is mainly determined on their citations and published journals. However, the citation statistics lag and the one-sideness of judging only by the journal not by the content have affected the accuracy and scientific nature of individual thesis quality evaluation. This article took 6019 papers that have been included in F5000 as samples, and explored the keyword feature of excellent papers by using knowledge mapping and co-word analysis to analyze expert review comments. Finding out and summarizing the outstanding paper features in innovation, value, content and writing, not only provides a support method to identify excellent papers, but also gives new ideas and reference for thesis quality evaluation and researchers’ academic evaluation.