Publication Type

Conference Proceeding Article

Version

publishedVersion

Publication Date

8-2016

Abstract

Nowadays, software developers often utilize existing third party libraries and make use of Application Programming Interface (API) to develop a software. However, it is not always obvious which library to use or whether the chosen library will play well with other libraries in the system. Furthermore, developers need to spend some time to understand the API to the point that they can freely use the API methods and putting the right parameters inside them. In this work, I plan to automatically recommend relevant APIs to developers. This API recommendation can be divided into multiple stages. First, we can recommend relevant libraries provided a given task to complete. Second, we can recommend relevant API methods that developer can use to program the required task. Third, we can recommend correct parameters for a given method according to its context. Last but not least, we can recommend how different API methods can be combined to achieve a given task. In effort to realize this API recommendation system, I have published two related papers. The first one deals with recommending additional relevant API libraries given known useful API libraries for the target program. This system can achieve recall rate@5 of 0.852 and recall rate@10 of 0.894 in recommending additional relevant libraries. The second one deals with recommending relevant API methods a given target API and a textual description of the task. This system can achieve recall-rate@5 of 0.690 and recallrate@10 of 0.779. The results for both system indicate that the systems are useful and capable in recommending the right API/library reasonably well. Currently, I am working on another system which can recommend web APIs (i.e., libraries) given a description of the task. I am also working on a system that recommends correct parameters given an API method. In the future, I also plan to realize API composition recommendation for the given task.

Keywords

API, Library, Recommendation System

Discipline

Computer Sciences | Software Engineering

Research Areas

Software and Cyber-Physical Systems

Publication

ASE 2016: Proceedings of the IEEE/ACM International Conference on Automated Software Engineering: Singapore, September 3-7, 2016

First Page

896

Last Page

899

ISBN

9781450338455

Identifier

10.1145/2970276.2975940

Publisher

ACM

City or Country

New York

Additional URL

http://doi.org/10.1145/2970276.2975940

Download

Share

COinS