多属性推荐算法在企业研发服务系统中的应用

Application of Multi-attribute Recommendation Algorithm in Enterprise Research and Development Service System

为满足企业的技术服务及研发需求,各地陆续建设线上供需服务平台.线上技术供需匹配难的主要原因在于文本相似度计算的准确性以及多元数据对于成交结果的影响.为解决上述问题,本文提出一种融合多属性的供需推荐模型,针对“陕西省中小企业研发服务平台”数据进行供需推荐,其中包括论文、专利、成果、项目等多属性特征,采用基于Bert(Bidirectional Encoder Representation from Transformers)的句向量表示方法计算供需文本的相似得分,并结合熵值法确定各属性权重,对各属性数据得分加权变换后实现推荐.实验表明,所提出的相似度计算模型在真实数据集上,各项评价指标优于词向量的表示方法.结合权值矩阵的多属性推荐模型,可较好的实现企业科技研发的供需推荐.

In order to meet the technical service and R&D needs of enterprises,online technology supply and demand service platforms have been successively built in various places.The main reasons for the difficulty in matching online supply and demand are the accuracy of text similarity calculations and the impact of multiple data on transaction results.In order to solve the above problems,this paper proposes a model,which fuses multi-attribute supply and demand recommendations model.The model can make a supply and demand recommendation,according to the data from"Shaanxi Province SME R&D Service Platform".Note that,the data includes multi-attribute features such as papers,patents,achievements,and projects.Take advantage of the sentence vector representation method based on Bert(Bidirectional Encoder Representation from Transformers)to calculate the similarity score of supply and demand text,and the entropy method is combined to determine the weight of each attribute,and the data score of each attribute is weighted and transformed to achieve recommendation.Experiments show that our similarity calculation model outperforms the word vector representation methods on the real data set.The multi-attribute recommendation model combined with the weight matrix can better realize the supply and demand recommendation of enterprise technology research and development.