Abstract:

The use of software failure mode can help developers and testers carry out the work of software failure prevention and discovery, and thus improve the quality of software. The usability of the existing software failure mode makes it difficult to play a greater role in the actual work, and the failure mode of the target software is difficult to manually extract. Therefore, the software failure mode is presented in the form of failure cause and failure effect combination, and the software failure mode automatic generation method for software failure text description is proposed by using text mining method. This method uses the software anomaly classification dictionary to construct the classifier to classify the description of the history software failure text. After the pretreatment of the invalid text, the k-means clustering algorithm is used to carry out the text clustering. And then the text matching method is used to achieve the automatic generation of failure mode. Finally, the textual extraction of the failure cause and the failure effect is carried out and the automatic generation of the failure mode is realized by the text matching method.

Key words: software failure mode, text mining, clustering analysis, text similarity, automatic extraction

摘要：

软件失效模式的使用可以帮助开发人员与测试人员高效地开展失效的预防和发现等工作，进而提高软件质量。现有软件失效模式的通用性使得在实际工作中难以发挥更大作用，且目标软件的失效模式难以人工提炼，因此提出利用失效原因和失效影响组合的形式表示软件失效模式，并利用文本挖掘方法，提出面向失效文本描述的软件失效模式自动生成方法。该方法利用软件异常分类字典构建分类器，用以对历史软件失效文本的描述进行分类；之后对经过预处理后的失效文本，开展文本聚类工作，进而从聚类后的类簇中选取有代表性的失效文本作为类簇标签；最后开展失效原因和失效影响的文本抽取并通过文本匹配方法实现失效模式的自动生成。

关键词: 软件失效模式, 文本挖掘, 聚类分析, 文本相似度, 自动抽取