In this paper, the author proposes a new technology–NLP2API which uses relevant API classes to reformulate natural language for code search by using Stack Overflow Q & A threads, PageRank, Borda count and extra-large data analytics. The author used 310 code search queries randomly collect websites from four popular (KodeJava, Java2s, CodeJava and JavaDB) programming tutorials and evaluate our query refactoring technique. Then the author selected five appropriate performance metrics from the literature (Top-K Accuracy, Mean Reciprocal Rank@K, Mean Average Precision@K, Mean Recall@K, Query Effectiveness) and evaluate related API class suggestions and query reformulation. The experiment provided NLP2API to perform 50% true of suggested classes and 82% Top-10 accuracy in recommending relevant API classes. And NLP2API is outperform the state-of-the-art technique on relevant API class suggestions, natural language queries and query reformulation. In addition, it can significantly improve the result in precision and NDCG. So NLP2API is the best one.
Contributions of The Paper
Proposed a new technology–NLP2API which uses relevant API classes to reformulate natural language for code search.
Prove NLP2API is better than the current state-of-the-art technique on Top-K Accuracy, Mean Reciprocal Rank, Mean Average Precision, Mean Recall, Query Effectiveness
NLP2API can be easily adapted for Java and other API recommendation in different programming areas.
Comments
The paper uses questions and data for a lot of comparisons to effectively prove that the technology is the best one and make readers understand.
Publisher
Shihui Gao
Link to The Paper
https://web.cs.dal.ca/~masud/papers/masud-ICSME2018.pdf
Name of The Authors
Mohammad Masudur Rahman, Chanchal K. Roy
Year of Publication
2018
Summary
In this paper, the author proposes a new technology–NLP2API which uses relevant API classes to reformulate natural language for code search by using Stack Overflow Q & A threads, PageRank, Borda count and extra-large data analytics. The author used 310 code search queries randomly collect websites from four popular (KodeJava, Java2s, CodeJava and JavaDB) programming tutorials and evaluate our query refactoring technique. Then the author selected five appropriate performance metrics from the literature (Top-K Accuracy, Mean Reciprocal Rank@K, Mean Average Precision@K, Mean Recall@K, Query Effectiveness) and evaluate related API class suggestions and query reformulation. The experiment provided NLP2API to perform 50% true of suggested classes and 82% Top-10 accuracy in recommending relevant API classes. And NLP2API is outperform the state-of-the-art technique on relevant API class suggestions, natural language queries and query reformulation. In addition, it can significantly improve the result in precision and NDCG. So NLP2API is the best one.
Contributions of The Paper
Proposed a new technology–NLP2API which uses relevant API classes to reformulate natural language for code search.
Prove NLP2API is better than the current state-of-the-art technique on Top-K Accuracy, Mean Reciprocal Rank, Mean Average Precision, Mean Recall, Query Effectiveness
NLP2API can be easily adapted for Java and other API recommendation in different programming areas.
Comments
The paper uses questions and data for a lot of comparisons to effectively prove that the technology is the best one and make readers understand.