A Fusion of Java Domain Knowledge Base and Siamese Network for Java API Recommendation

APIs play an important role in modern software development. Programmers need to frequently search for the appropriate APIs according to different tasks. With the development of the information industry, API reference documents have become larger and larger. Due to redundant and erroneous information on the Internet, traditional search methods can also cause inconvenience to programmers' queries. At the same time, there is a gap in terms of vocabulary and knowledge between the natural language description of the programming task and the description in the API documentation, so it is difficult to find a suitable API. To solve these problems, this paper proposes a Java API recommendation model by fusing the Java domain knowledge base and the Siamese Network to improve the accuracy of API recommendation. Experiments on the BIKER data set show that our method has better recommendation results than the state-of-art DeepAPI and BIKER model.