Recommendation method and system based on multi-domain semantic fusion

The invention discloses a recommendation method and system based on multi-domain semantic fusion. The recommendation method includes the steps: constructing semantic ontology, realizing fusion of ontology types in different domains by utilizing a skip-gram model, so as to realize cross-domain fusion of projects and obtain a cross-domain project similarity; iterating data of multiple domains according to domain relevancy sorting to generate a final user-project scoring matrix; and finally realizing cross-domain recommendation in combination with a collaborative filtering method to recommend a cross-domain project to a user. According to the recommendation method, the problems of data sparseness and cold start existing in single-field recommendation can be effectively relieved, meanwhile, across-domain recommendation method based on semantics is optimized, and a large amount of manual work in the forms of manual labels and the like is avoided. 本发明公开了一种基于多领域语义融合的推荐方法与系统，通过构建语义本体，利用skip-gram模型实现不同领域中本体类型的融合，从而实现项