基于知识提取的测试文档自动生成  

Automatic Generation of Test Documents Based on Knowledge Extraction

作　　者：孙晓韩 李宁[1] 于杨 杨子仪 张林 SUN Xiao-han;LI Ning;YU Yang;YANG Zi-yi;ZHANG Lin(Beijing Key Laboratory of Internet Culture and Digital Dissemination Research,Beijing Information Science and Technology University,Beijing 100101,China)

机构地区：[1]北京信息科技大学网络文化与数字传播北京市重点实验室,北京100101

出　　处：《计算机技术与发展》2022年第12期110-116,共7页Computer Technology and Development

基　　金：国家重点研发计划(2018YFB1004100);国家自然科学基金项面上项目(61672105)。

摘　　要：测试文档是产品的重要组成部分,与测试数据紧密相关,且有严格的编制要求。针对以往测试文档编写中存在的重复繁琐、灵活性不高、效率低下等问题,运用数据到文本的生成理论,该文提出了一种基于知识提取的测试文档生成方法。该方法首先对原始的测试记录数据进行分析和理解,重点进行电子表格的表头识别和单元格关联关系识别,抽取出表格数据的逻辑关系;再根据规范化的测试本体,转换成规范的知识表达,形成测试知识的结构化表示并记录于测试知识库中;最后通过局部文档模板的填充,逐步形成整体的测试文档。该方法解决了灵活多变的测试记录导致的软件重复开发的一系列问题,有利于测试结果的有效利用。此外,自底向上的文档构造方式有助于按需展现内容,提高测试文档的可读性。The test document is an important part of the product,which is closely related to the test data and has strict compilation requirements.In order to solve the problems of repetition and triviality,low flexibility and low efficiency in the preparation of test documents in the past,we propose a test document generation method based on knowledge extraction by using the theory of data-to-text generation.This method firstly analyzes the original test data,focusing on the spreadsheet header recognition and cell relationship recognition,extracting the logical data.Then,according to the normalized ontology,the data is transformed into the standardized knowledge expression,forming the structured representation of test knowledge and recording it in the test knowledge base.Finally,the whole test document is gradually formed by filling each basic document template.This method solves the problems of repeated software development caused by flexible and changeable test records and is beneficial to the effective reuse of test results.In addition,the bottom-up document construction helps to present content on demand and improve the readability of test documents.

关 键 词：文档生成 知识提取 表格识别 计算机辅助写作 软件测试 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]