Mohammad Masudur Rahman, Chanchal K. Roy and David Lo
Year of Publication
2017
Summary
In this paper, the authors proposed a new API recommendation technique–RACK. When software developers use natural language language to search questions. In RACK, the questions can be text token from natural language processing. The text tokens is used to find recommended API tokens by Keyword–API Co-occurrence (KAC), Keyword–Keyword Coherence (KKC). The API is used to find related code snippets from GitHub. RACK can recommend relevant APIs with about 79% Top-10 accuracy. In addition, RACK can solve the vocabulary mismatch problem and use Java multi-threading to reduce computation and response time. The RACK compares with the state-of-the-art–Thung et al that proves RACK has better performance on API recommendation.
Limitations:
The database may need to be updated periodically to include new APIs or changes in API names.
RACK relies on the availability and quality of code on GitHub.
RACK may not support other programming languages.
Contributions of The Paper
1.RACK accepts queries from unstructured natural language and returns relevant code snippets from open source sites.
2.RACK can reduce vocabulary mismatches faced by traditional code search engines.
3.RACK can provide relevant insights for search queries and search results.
Comments
The paper uses questions and figures to illustrate the main points that makes it easier for readers to understand. And Key word use bold type to let readers find and understand easily.
Publisher
Shihui (Maxine) Gao
Link to The Paper
https://web.cs.dal.ca/~masud/papers/masud-ICSE2017.pdf
Name of The Authors
Mohammad Masudur Rahman, Chanchal K. Roy and David Lo
Year of Publication
2017
Summary
In this paper, the authors proposed a new API recommendation technique–RACK. When software developers use natural language language to search questions. In RACK, the questions can be text token from natural language processing. The text tokens is used to find recommended API tokens by Keyword–API Co-occurrence (KAC), Keyword–Keyword Coherence (KKC). The API is used to find related code snippets from GitHub. RACK can recommend relevant APIs with about 79% Top-10 accuracy. In addition, RACK can solve the vocabulary mismatch problem and use Java multi-threading to reduce computation and response time. The RACK compares with the state-of-the-art–Thung et al that proves RACK has better performance on API recommendation.
Limitations:
The database may need to be updated periodically to include new APIs or changes in API names.
RACK relies on the availability and quality of code on GitHub.
RACK may not support other programming languages.
Contributions of The Paper
1.RACK accepts queries from unstructured natural language and returns relevant code snippets from open source sites.
2.RACK can reduce vocabulary mismatches faced by traditional code search engines.
3.RACK can provide relevant insights for search queries and search results.
Comments
The paper uses questions and figures to illustrate the main points that makes it easier for readers to understand. And Key word use bold type to let readers find and understand easily.