Data Mining for Scientific Projects Recommendations Based on Knowledge Graph and Deep Learning

With the rapid development and fierce competition, new researches that based on scientific papers emerge in an endless stream. As the scientific research projects issuers, how to publish the great demand scientific projects become a difficult problem. This paper, firstly, drawn knowledge graph using co-occurrence matrix of the key words that based on the key words of papers in China National Knowledge Infrastructure (CNKI), taking Applied Economy as an example. Secondly, it used k-NN algorithm to get the best recommendation effect, which find using the split=0.67 between training set and test set getting the best recommendation effect. This finding can help the scientific research projects issuers, to find the implicit relationship features between papers published in last year, and then output an ordered list of keywords as its recommendation.