The invention relates to a software recommendation method capable of effectively combining software projects, which are currently developed by a developer, based on group comments. The software recommendation method comprises the following steps: obtaining open source data, and pre-processing the projects and comment document tags; extracting local project characteristics of a user, and matching the projects with comment documents by utilizing the comment document tags and titles and the project names and tags; calculating the similarity and the correlation between a local project and an acquired open source project; and performing linear weighted summation of the similarity and the correlation, and generating a project recommendation list for the user. The software recommendation method disclosed by the invention can be integrated with a software development environment of the developer; the retrieval efficiency of the developer on related software in an internet range can be increased; and technical support can be obtained or software reuse can be realized from the recommended software projects.