Context-Aware API Recommendation for Mashup Composition Based on Path Ranking

In recent years, more software modules have been increasingly published onto the Internet as web APIs, making the activity of manually composing mashups time-consuming and error-prone for mashup developers. In both academia and industry, discovering and recommending suitable web APIs has become a co...

Full description

Saved in:
Bibliographic Details
Published in:2023 6th International Conference on Software Engineering and Computer Science (CSECS) pp. 01 - 07
Main Authors: Hu, Beichen, Xie, Xihao, Shen, Junhao
Format: Conference Proceeding
Language:English
Published: IEEE 22.12.2023
Subjects:
API recommendation
Mashup composition
Mashups
Path ranking algorithm
Predictive models
Software
Software algorithms
Software engineering
Task analysis
Trajectory
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In recent years, more software modules have been increasingly published onto the Internet as web APIs, making the activity of manually composing mashups time-consuming and error-prone for mashup developers. In both academia and industry, discovering and recommending suitable web APIs has become a core method to increase the working efficiency for mashup developers. In this paper, we propose a novel context-aware approach based on path ranking algorithm (PRA) to recommending next potential APIs in a recommend-as-you-go manner. At every step throughout the process of composing a mashup, the APIs that have been selected and the collection of words in the textually described requirement are regarded as contextual input of a prediction model which is comprised of weights of various relation paths being built from a relation schema. Given the input query, candidate APIs are ranked in descending order according to its predicted score of being the API that should be selected at next step. Finally, top K of these candidate APIs are recommended to the mashup developer. Extensive experiments on a real-world dataset demonstrate that our approach is effective.
DOI:10.1109/CSECS60003.2023.10428653