An API Recommendation Method Based On Mashup Service Semantic Expression Clustering

In recent years,a lightweight service composition model called Mashup technology has become popular on the Internet.Its principle is to mix and match two or more API services to build a brand new Web application.However,with the rapid development of service computing,more and more enterprises and departments publish their own developed services,data or resources to the Internet through APIs,which has led to an increasing number of API services.From such a large-scale API service collection,how to quickly,accurately and diversely recommend API services that meet the needs of developers and users' Mashup has become a challenging problem.There is a calling relationship between the Mashup service and the API service,so the functional clustering of the Mashup service can provide diversified suggestions for API recommendations.Therefore,this article focuses on the two core issues of Mashup service clustering and API recommendation.The main contents are as follows.(1)The paper proposed a semantic text representation model of Mashup service based on word vector weights.In order to solve the problem of insufficient vocabulary of the description text of the Mashup service,we augment the description text of the Mashup service with the label and description text of the API service called by the Mashup service;The Mashup service description text semantic items are given weights based on the semantic correlation between the Mashup service description text and the service tag,and a Mashup service description text semantic representation model is constructed.Through experimental simulation,the results show that the model proposed in this paper can improve the problem of poor focus of Mashup service text in the clustering model.(2)This paper proposes a Mashup service clustering method based on the similarity of Mashup service text semantic representation model.This method first calculates the similarity between Mashup service text semantic representation models.This method first calculates the similarity between the semantic representation models of the Mashup service text;Aiming at the problem of K-means clustering algorithm's difficulty in selecting the center point,an improved K-means and Agnes fusion clustering algorithm(CFFC)based on density peaks is proposed;finally,the CFFC algorithm clusters the Mashup service by model similarity.Through experimental simulation,the results show that the proposed clustering algorithm can improve the clustering effect.(3)This paper proposes an API recommendation method based on Mashup service.This method first builds a historical call relationship matrix between Mashups clusters and API services on the basis of Mashup service clustering;then explores the implicit factors in the call relationship matrix through matrix decomposition,and performs missing values on the historical call relationship matrix Prediction;Finally,based on the completed historical call relationship matrix,the Mashups class cluster is matched and recommended based on the user's demand text.Through experimental simulation,the results show that the recommendation method proposed in this paper can improve the accuracy and diversity of API service recommendation.