Javawock: A java class recommender system based on collaborative filtering

Masateru Tsunoda, Takeshi Kakimoto, Naoki Ohsugi, Akito Monden, Ken Ichi Matsumoto

Research output: Chapter in Book/Report/Conference proceedingConference contribution

10 Citations (Scopus)

Abstract

Many software development platforms provide a large number of library components to make it easy to build high quality software. On the other hand, it became more and more difficult for developers to find useful components in each development context because the amount of components provided became too large today. This paper proposes a recommender system that provides useful Java components (library class files) to a developer based on collaborative filtering (CF). When a developer gives an unfinished Java program to the system, it investigates Java library class files used in the given program and finds Java programs that arc similar to the given program from a program repository. Then, the system recommends to the developer Java library class files that were used in the similar programs but were not used in the developer's program. An experimental evaluation showed that the recommendation accuracy of the proposed system was much higher than that of a naive (non-CF) method in all four evaluation criteria (recall, precision, F1 value, and half-life utility).

Original languageEnglish
Title of host publication17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005
Pages491-497
Number of pages7
Publication statusPublished - Dec 1 2005
Externally publishedYes
Event17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005 - Taipei, Taiwan, Province of China
Duration: Jul 14 2005Jul 16 2005

Publication series

Name17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005

Other

Other17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005
CountryTaiwan, Province of China
CityTaipei
Period7/14/057/16/05

Keywords

  • Cosine similarity
  • Information retrieval
  • J2SE
  • Recommender system
  • Software component

ASJC Scopus subject areas

  • Software

Fingerprint Dive into the research topics of 'Javawock: A java class recommender system based on collaborative filtering'. Together they form a unique fingerprint.

  • Cite this

    Tsunoda, M., Kakimoto, T., Ohsugi, N., Monden, A., & Matsumoto, K. I. (2005). Javawock: A java class recommender system based on collaborative filtering. In 17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005 (pp. 491-497). (17th International Conference on Software Engineering and Knowledge Engineering, SEKE 2005).