skip to main content
Idioma:
Primo Search
Clear Search Box
Search in: Busca Geral
Busca Avançada
Busca por Índices

CrossRec: Supporting software developers by recommending third-party libraries

Nguyen, Phuong T. ; Di Rocco, Juri ; Di Ruscio, Davide ; Di Penta, Massimiliano

The Journal of systems and software, 2020-03, Vol.161, p.110460, Article 110460 [Periódico revisado por pares]

Elsevier Inc

Texto completo disponível

Citações Citado por
  • Título:
    CrossRec: Supporting software developers by recommending third-party libraries
  • Autor: Nguyen, Phuong T. ; Di Rocco, Juri ; Di Ruscio, Davide ; Di Penta, Massimiliano
  • Assuntos: Mining software repositories ; Open Source software ; Recommender systems
  • É parte de: The Journal of systems and software, 2020-03, Vol.161, p.110460, Article 110460
  • Descrição: •A system assists developers in selecting highly relevant third-party libraries.•It outperforms three state-of-the-art approaches on different dimensions.•It accounts for different quality metrics, including library novelty.•Differently from existing approaches, our tool recommends libraries with version. When creating a new software system, or when evolving an existing one, developers do not reinvent the wheel but, rather, seek available libraries that suit their purpose. In such a context, open source software repositories contain rich resources that can provide developers with helpful advice to support their tasks. However, the heterogeneity of resources and the dependencies among them are the main obstacles to the effective mining and exploitation of the available data. In this sense, advanced techniques and tools are needed to mine the metadata to bring in meaningful recommendations. In this paper, we present CrossRec, a recommender system to assist open source software developers in selecting suitable third-party libraries. CrossRec exploits a collaborative filtering technique to recommend libraries to developers by relying on the set of dependencies, which are currently included in the project being developed. We perform an empirical evaluation to compare the proposed approach with three state-of-the-art baselines, i.e., LibRec, LibFinder, and LibCUP on three considerably large datasets. The experimental results show that CrossRec overcomes the limitation of the baselines by recommending also libraries with a specific version. More importantly, it outperforms LibRec and LibCUP with respect to various quality metrics.
  • Editor: Elsevier Inc
  • Idioma: Inglês
Voltar para lista de resultados

  • Realização: ABCD-USP USP   Logos de Redes Sociais: Freepik

Buscando em bases de dados remotas. Favor aguardar.