领域科研项目评审专家推荐算法

Algorithm recommended by experts for evaluation of scientific research projects in the field

为提高领域科研项目评审质量,提出一种基于领域标签体系的推荐算法。进行数据采集与处理,构建专家库;采用命名实体识别模型识别专业领域名词,用于构建领域标签体系;设计基于关键词抽取算法与专业实体识别模型相结合的项目申请文档标签标注算法,提出基于三元组的专家标签标注模型,构建推荐专家候选集;基于专家社会关系网络进行专家回避筛选,实现推荐。通过实际应用情况反馈,该算法以领域标签体系的形式,有效解决了技术名词命名不统一等问题,融入专家社会关系规避的推荐模型,为项目评审公平性提供了保障。

To improve the evaluation quality of field research projects,a recommendation algorithm based on field label system was proposed.Data collection and processing was carried out to build expert database.The named entity recognition model was used to identify domain nouns and construct domain label system.A label labeling algorithm for project application documents based on the combination of keyword extraction algorithm and professional entity recognition model was designed.A ternary-based expert label labeling model was proposed,and a set of recommended expert candidates was constructed.Expert avoidance screening based on expert social network was carried out to achieve recommendation.Through the feedback of practical application,the proposed algorithm solves the problem of non-uniform naming of technical terms effectively in the form of domain label system,and incorporates the recommendation model of social relationship evasion among experts,which guarantees the fairness of project review.