Alarm Correlation Research And Implementation Based On Similar Data Sources

The results of static defect detection are subject to false positives and false negatives, which need to be confirmed manually one by one. A large number of manual validation work reduces the detection efficiency,increase the cost of using static detection software, and even make part of the software company refuse to use static testing tools.In this paper, an alarm correlation technology based on similar data sources is proposed. This technique associates alarms with the same validation result and classifies them into one class and labels them in the result file. When performing manual validation, the reviewer can obtain the results of a class of alerts only by confirming one alert, which will greatly improve the efficiency of manual validation.The alarm correlation method proposed in this paper is based on the define information of the alarm-related variables. In order to extract the define information in a comprehensive and accurate way, we propose a symbolic model, which integrates the structure, variables, constants and position characteristics of all define point related to this alert. According to this model, we extract each alert's define information and associate them with these info. The experimental results show that the alarm correlation method based on similar data sources can correlate nearly 71%of alarms in the detection results.