| In the applications development for mobile multimedia computing,improving the efficiency of applications development has always been a core problem.The use of API has played an obvious role in improving the efficiency of applications development,but it takes a lot of time to find a suitable Web API.Web APIs recommendation can solve this problem,and API will be directly recommended to you.We will use the matrix factorization(MF)model to predict the relationship between user and Web API.In this model,the relationship between user and Web API is represented by matrix.But different from the MF model,we find that the relationship between Web APIs will have a certain impact on the prediction results.Therefore,this paper uses embedding technology to calculate the vector of API,and then obtains the similarity between Web APIs through the calculation between vectors,and we develop a novel MF model that leverages the latent relationship between Web APIs.In this way,we generate a matrix factorization model,which is named as MF-AS.The similarity calculation results are input into the MF-AS model to realize the relationship prediction between API and users.By comparing with the results of other recommended algorithms,it can be found that the effect of MF-AS model is better in all cases,and there are still good prediction results in the case of cold start,which shows the effectiveness of the model and can also play a good role in special cases.At the same time,we have also designed and implemented the knowledge graph of Web APIs.With the development of information technology,the mode of software development is also improving.In the face of the current needs of intelligent software development,the knowledge graph can effectively integrate and utilize a large number of Web APIs and related data to meet the needs of intelligent software development.Based on Web APIs and other data,combined with knowledge graph construction technology,this paper establishes API and mashup knowledge graph respectively.Through the knowledge fusion of the two kind knowledge graph,we can finally get a more complete web API knowledge graph.The complete web API knowledge map obtained by knowledge map construction technology provides a new tool for the development of intelligent software.At the same time,we extract part of the data from the web API knowledge graph.We use the Trans E and Trans H methods to represent the API in the knowledge map.After obtaining the results,we cluster and calculate the similarity of the results to analyze the whole representation results.The experimental results show that it makes a contribution to the field of intelligent software development. |
