A CNL for Contract-Oriented Diagrams

Abstract

Abstract. We present a first step towards a framework for defining and manipulating normative documents or contracts described as ContractOriented (C-O) Diagrams. These diagrams provide a visual representation for such texts, giving the possibility to express a signatory's obligations, permissions and prohibitions, with or without timing constraints, as well as the penalties resulting from the non-fulfilment of a contract. This work presents a CNL for verbalising C-O Diagrams, a web-based tool allowing editing in this CNL, and another for visualising and manipulating the diagrams interactively. We then show how these proof-ofconcept tools can be used by applying them to a small example.

A web-based tool for analysing normative documents in english

“…CNLs are often used as interfaces for formal languages which are human-friendly, yet still unambiguous and well-defined. The CNL designed for our contract formalism is decribed in previous work [9]. We use the Grammatical Framework (GF) [21] for defining the grammar for our CNL and converting to and from our internal formal representation.…”

Participatory Verification of Railway Infrastructure by Representing Regulations in RailCNL

“…Camilleri et al [5] take a CNL approach to manipulating contract-oriented diagrams using a visual diagram editor, a CNL with text editor support, and a spreadsheet representation as interfaces to a common model, which can be translated into timed automata for reasoning about system properties. Calafato et al [4] describe queries for tax fraud detection, where a GF CNL Vocabulary Matching. The Norwegian regulations are written in Norwegian and use other terms for class names, properties and relations than the railML representation does.…”

“…The structure of the CNL is modelled after the CL language. Camilleri et al [4] take a CNL approach to manipulating contract-oriented diagrams using a visual diagram editor, a CNL with text editor support, and a spreadsheet representation as interfaces to a common model, which can be translated into timed automata for reasoning about system properties. Other efforts to define domain specific languages for railway verification have typically focused on the implementation of control systems, such as Vu et al [21], while also considering the verification to be an activity which is separate from design and implementation.…”

“…Different approaches to CNL editors have been explored (see e.g. [4,10,14]). We plan to investigate these further and integrate one such editor for RailCNL in the RailCOMPLETE CAD environment, and carry out a usability study on its efficacy.…”

“…to change and adjust them when needed. 4 Thirdly, we want that the domain experts are able to create their own specifications and feed these into the verification engine, e.g. define specific expert knowledge as verification conditions.…”