一种面向移动应用开发的第三方库混合推荐方法

Hybrid Recommendation Method of Third-party Library for Mobile Application Development

随着智能手机的普及,移动应用的数量和使用人群都在飞速的增长.日益增长的用户群为移动应用的开发提出了更高的要求,如何帮助开发者更好更快的完成移动应用的开发成为一个关键问题.在移动应用开发过程中,重用已有的第三方库可以有效的帮助开发者减低开发时间与开发成本.然而,可用的第三方库数量繁多且功能各异,这为开发者选择合适的第三方库提出了挑战.本文从第三方库重用这个角度出发,提出了一种第三方库的混合推荐技术,旨在帮助开发者在移动应用开发过程中快速推荐合适的第三方库,从而改进其开发效率.在推荐过程中,使用贝叶斯定理将基于用户的协同过滤方法与基于内容的TF-IDF方法融合来实现推荐任务.在实验环节,我们爬取了5000多个移动应用的描述信息及其第三方库的调用信息.基于这些数据开展的实验结果证明了所提方法的有效性.

With the popularity of smartphones,the number of mobile applications and the number of users are growing rapidly. The growing user base puts higher demands on the development of mobile applications. How to help developers to complete mobile application development faster and faster becomes a key issue. In the mobile application development process,reusing existing third-party libraries can effectively help developers reduce development time and development costs. However,the number of available third-party libraries is varied and the functionality is different,which poses a challenge for developers to choose the right third-party library. This paper proposes a hybrid recommendation technology for third-party libraries from the perspective of third-party library reuse. It aims to help developers quickly recommend suitable third-party libraries in the process of mobile application development,thus improving their development efficiency. In the recommendation process,the Bayesian theorem is used to combine the user-based collaborative filtering method with the content-based TF-IDF method to implement the recommendation task. In the experimental session,we craw led more than 5,000 mobile application descriptions and call information from third-party libraries. Experimental results based on these data demonstrate the effectiveness of the proposed method.