TiQi: answering unstructured natural language trace queries
TiQi: answering unstructured natural language trace queries
复制标题
TiQi：回答非结构化自然语言跟踪查询
DOI：
10.1007/s00766-015-0224-4
复制
发表时间：
2015
影响因子：
2.8
通讯作者：
J. Cleland
中科院分区：
文献类型：
--
作者：
Piotr Pruski;Sugandha Lohar;Will Goss;A. Rasin;J. Cleland
Software traceability is a required element in the development and certification of safety-critical software systems. However, trace links, which are created at significant cost and effort, are often underutilized in practice due primarily to the fact that project stakeholders often lack the skills needed to formulate complex trace queries. To mitigate this problem, we present a solution which transforms spoken or written natural language queries into structured query language (SQL). TiQi includes a general database query mechanism and a domain-specific model populated with trace query concepts, project-specific terminology, token disambiguators, and query transformation rules. We report results from four different experiments exploring user preferences for natural language queries, accuracy of the generated trace queries, efficacy of the underlying disambiguators, and stability of the trace query concepts. Experiments are conducted against two different datasets and show that users have a preference for written NL queries. Queries were transformed at accuracy rates ranging from 47 to 93 %.
安全关键项目的战略可追溯性
影响因子：
3.3
作者：
Mäder;Patrick;Cleland-Huang Jane
通讯作者：
Cleland-Huang Jane