Natural Language Specifications

Johannisson, Kristofer

doi:10.1007/978-3-540-69061-0_7

Cited by 4 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There have also been notable attempts to translate formal specifications into English, such as the support in the KeY theorem prover [38], and a predecessor system [13] that directly uses categorial grammar in reverse for natural language generation [81] (another established use for categorial grammars beyond the semantic parsing we focus on).…”

Section: Related Workmentioning

confidence: 99%

Natural Language Specifications in Proof Assistants

Gordon¹,

Sergey²

2022

Preprint

View full text Add to dashboard Cite

Interactive proof assistants are computer programs carefully constructed to check a human-designed proof of a mathematical claim with high confidence in the implementation. However, this only validates truth of a formal claim, which may have been mistranslated from a claim made in natural language. This is especially problematic when using proof assistants to formally verify the correctness of software with respect to a natural language specification. The translation from informal to formal remains a challenging, time-consuming process that is difficult to audit for correctness. This paper argues that it is possible to build support for natural language specifications within existing proof assistants, in a way that complements the principles used to establish trust and auditability in proof assistants themselves.

show abstract

Section: Related Workmentioning

confidence: 99%

Natural Language Specifications in Proof Assistants

Gordon¹,

Sergey²

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…As long as most of the abstract syntax is re-used, the translation into Datalog should also be readily adaptable. Related Work Johannisson [7] describes a CNL targeting the Object Constraint Language (OCL) for use in reasoning about Java program correctness in the KeY system [3]. The language features dynamic vocabulary based on input UML diagrams where vocabulary updates are achieved by re-compiling the grammar using the GF compiler when needed.…”

Section: Evaluation and Conclusionmentioning

confidence: 99%

“…In our collaboration with Norwegian railway engineers, we have focused on regulations in Norwegian language 7 , but our general approach (Section 2) is language-independent. In Section 3 we present RailCNL, a user-friendly verification front-end language for static railway infrastructure analysis.…”

Section: Introductionmentioning

confidence: 99%

Participatory Verification of Railway Infrastructure by Representing Regulations in RailCNL

Luteberget

Camilleri

Johansen

et al. 2017

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Designs of railway infrastructure (tracks, signalling and control systems, etc.) need to comply with comprehensive sets of regulations describing safety requirements, engineering conventions, and design heuristics. We have previously worked on automating the verification of railway designs against such regulations, and integrated a verification tool based on Datalog reasoning into the CAD tools of railway engineers. This was used in a pilot project at Norconsult AS (formerly Anacon AS). In order to allow railway engineers with limited logic programming experience to participate in the verification process, in this work we introduce a controlled natural language, RailCNL, which is designed as a middle ground between informal regulations and Datalog code. Phrases in RailCNL correspond closely to those in the regulation texts, and can be translated automatically into the input language of the verifier. We demonstrate a prototype system which, upon detecting regulation violations, traces back from errors in the design through the CNL to the marked-up original text, allowing domain experts to examine the correctness of each translation step and better identify sources of errors. We also describe our design methodology, based on CNL best practices and previous experience with creating verification front-end languages.

show abstract