Hits:

Indexed by:会议论文

Date of Publication:2016-12-12

Included Journals:EI

Volume:10086 LNAI

Page Number:361-373

Abstract:Features recommendation is an important technique for getting the requirements to develop and update mobile Apps and it has been one of the frontier study in requirements engineering. However, the mobile Apps ?descriptions are always free-format and noisy, the classical features recommendation methods cannot be effectively applied to mobile Apps ?features recommendation. In addition, most mobile Apps ?source codes that contain API calling information can be obtained by software tools, which can accurately indicate the functional features. Therefore, this paper proposes a hybrid feature recommendation method of mobile Apps, which is based on both explicit description and implicit code information. A self-adaptive similarity measure and KNN is used to find relevant Apps, and functional features are extracted from the Apps and recommended for developers. Experimental results on four categories Apps show that the proposed features recommendation method with hybrid information is more effective than the classical method. © Springer International Publishing AG 2016.