Modelling timed reactive systems from natural-language requirements
Modelling timed reactive systems from natural-language requirements
复制标题
根据自然语言需求对定时反应系统进行建模
DOI:
10.1007/s00165-016-0387-x
复制
发表时间:
2016
影响因子:
1
通讯作者:
Carvalho G
中科院分区:
文献类型:
--
作者:
Carvalho G
At the very beginning of system development, typically only natural-language requirements are documented. As an informal source of information, however, natural-language specifications may be ambiguous and incomplete; this can be hard to detect by means of manual inspection. In this work, we present a formal model, named data-flow reactive system (DFRS), which can be automatically obtained from natural-language requirements that describe functional, reactive and temporal properties. A DFRS can also be used to assess whether the requirements are consistent and complete. We define two variations of DFRS: a symbolic and an expanded version. A symbolic DFRS (s-DFRS) is a concise representation that inherently avoids an explicit representation of (possibly infinite) sets of states and, thus, the state space-explosion problem. We use s-DFRS as part of a technique for test-case generation from natural-language requirements. In our approach, an expanded DFRS (e-DFRS) is built dynamically from a symbolic one, possibly limited to some bound; in this way, bounded analysis (e.g., reachability, determinism, completeness) can be performed. We adopt the s-DFRS as an intermediary representation from which models, for instance, SCR and CSP, are obtained for the purpose of test generation. An e-DFRS can also be viewed as the semantics of the s-DFRS from which it is generated. In order to connect such a semantic representation to established ones in the literature, we show that an e-DFRS can be encoded as a TIOTS: an alternative timed model based on the widely used IOLTS and ioco. To validate our overall approach, we consider two toy examples and two examples from the aerospace and automotive industry. Test cases are independently created and we verify that they are all compatible with the corresponding e-DFRS models generated from symbolic ones. This verification is performed mechanically with the aid of the NAT2TEST tool, which supports the manipulation of such models.
登录
查看更多内容
DOI:
--
发表时间:
2014
期刊:
International Workshop on Artificial Intelligence for Requirements Engineering
影响因子:
--
作者:
Alessio Ferrari;Giuseppe Lipari;S. Gnesi;G. Spagnolo
通讯作者:
G. Spagnolo
DOI:
--
发表时间:
2009
期刊:
2009 Third IEEE International Conference on Secure Software Integration and Reliability Improvement
影响因子:
--
作者:
Matthias Schnelte
通讯作者:
Matthias Schnelte
DOI:
--
发表时间:
2013
期刊:
IEEE International Conference on Formal Engineering Methods
影响因子:
--
作者:
Gustavo Carvalho;A. Sampaio;A. Mota
通讯作者:
A. Mota
影响因子:
3.4
作者:
V. Ambriola;V. Gervasi
通讯作者:
V. Gervasi
DOI:
--
发表时间:
2010
期刊:
ACM Symposium on Applied Computing
影响因子:
--
作者:
A. Bucchiarone;S. Gnesi;A. Fantechi;G. Trentanni
通讯作者:
G. Trentanni