The Research On The Evaluation Index Of The Rationality Of Logs

With the gradual development of information technology and internet technology, the amount of software developers and codes in the fields of software engineering keeps increasing. The logs by developers in the codes are the main basis of fault diagnosis and defect reparation. The quality of the logs is the key of the software's maintenance, too many logs will make it hard to analyze while too few logs will make it hard to analyze too because of the lack of information. Currently, there isn't an automatically method to evaluate the rationality of the log and there is few evaluation standards on the rationality of the log. In this circumstance, we raise the research on the evaluation index of the rationality of logs based on models.By investigating and researching the current situation of research at home and abroad as well as the current log analyzing standards, we researched and analyzed several log analyzing tools which are commonly used after deciding the method. Then we raised a multi-dimension and multi-index log evaluation standard based on the method of software quality management. We refined several dimensions for further research, mainly includes efficiency dimension, correction dimension, maintenance dimension, scalability dimension and normative dimension. Through analyzing these five dimensions, we chose the log density, correct mark rate, omission rate and false alarm rate and other indexes to research the preprogressing algorithm of the log document, log mining algorithm, Apriori improved algorithm for association rules at the same time. Then we use C++ language to realize some of these indexes under the Ubuntu environment and analyzed the result.After the realization of index extraction, we conducted the simulation test in the experimental environment in the Intel multi-core Laboratory in the school of software of Huazhong University of Science and Technology, that is, providing the method and tool for the evaluation of log's rationality, obtaining the evaluation model of log's rationality and raising improving methods. The results show that compared with other methods, its correction rate, efficiency and many other aspects are better, its evaluation dimensions and indexes are effective that can help engineers finishing Fault diagnosis and positioning quickly and conducting the monitoring and maintenance of the software quality conveniently in the late period of product delivery.