Method for code failure prediction in smartphone application and development

Abstract

The invention discloses a method for code failure prediction in smartphone application and development. The method comprises a step 1) of tidying code units, distinguishing a label set and a target set, defining characteristic indexes, and extracting characteristic vectors; a step 2) of endowing weights to the code units of the label set based on the characteristic vectors; a step 3) of training a weighted classifier, and calculating the defectiveness probability and no-defect probability of the code units of the target set; a step 4) of adopting a self-learning strategy to correct predicted results, and repeating the steps until prediction results calculated in a preceding time and a subsequent time are similar to or reach given iteration times; and a step 5) of outputting the code units with a high suspiciousness ratio in the target set. According to the method, data of historical versions of target smartphone applications and other smartphone applications is used, the characteristic indexes are defined in the code unit level, a machining learning technology is applied, the code units which possibly contain defects are predicted, and defective code units in the smartphone applications are positioned quickly and accurately. The method has simplicity in calculation, defective codes can be positioned quickly and effectively, and the method is suitable for development and maintenance of the smartphone applications.

Description

technical field [0001] The present invention relates to the field of smart phone application program development, specifically a method for predicting code defects in smart phone application development. The method predicts whether smart phone application code units contain defects based on massive open source codes and defect information, thereby quickly locating defects The code unit can effectively improve the efficiency of defect repair and code quality of smart phone applications. Background technique [0002] With the development of science and technology, smartphones have become an important tool for people to connect with each other, entertain and understand current affairs, and are indispensable for modern people. As a result, the number of smart phone applications has also increased rapidly, and the functions have gradually diversified. As of 2016, the number of smartphone applications has exceeded 3 million, and there are still a large number of smartphone applic...

AI Technical Summary

Problems solved by technology

At present, defect code search is mainly based on empirical judgment, and interactive debugging tools are mostly used, which requires a lot of manual intervention and manual positioning, which consumes a lot of time and cost, relies on personal experience and cannot be repeated

Images