A requirements capture method and its use in an air traffic control application

Int. J. Artif. Intell. Tools

2001

Self Cite

In this paper we describe a project (IMPRESS) in which machine learning (ML) tools were created and utilised for the validation of an Air Traffic Control domain theory written in first order logic. During the project, novel techniques were devised for the automated revision of general clause form theories using training examples. These techniques were combined in an algorithm which focused in on the parts of a theory which involve ordinal sorts, and applied geometrical revision operators to repair faulty component parts. While we illustrate the feasibility of applying ML to this area, we conclude that to be effective it must be focused to the application at hand, and used in mixed-initiative mode within a tools environment. The method is illustrated with experimental results obtained during the project.

Section: Introductionmentioning

confidence: 99%

The Application of Machine Learning Tools to the Validation of an Air Traffic Control Domain Theory

West

Int. J. Artif. Intell. Tools

2001

Self Cite

“…This allows the model's notation, or an equivalent 'pseudo-natural language' form, to be understandable to non-computing professionals (overcoming at least to some extent Leveson et al's criticism of the use of formal systems in requirements analysis [10]). We chose Many Sorted First Order logic (here abbreviated to 'msl') to encode the model for a number of reasons, detailed in [14]. As msl is a very general language we customised it chiefly through the imaginative and precise use of syntactic constructs.…”

Section: Producing a Customised Modelmentioning

confidence: 99%

Towards the automated debugging and maintenance of logic-based requirements models

Proceedings 13th IEEE International Conference on Automated Software Engineering (Cat. No.98EX239)

West

In this paper we describe a tools environment which automates the validation and maintenance of a requirements model written in many-sorted first order logic. We focus on: a translator, that produces an executable form of the model; blame assignment functions, which input batches of mis-classified tests (i.e. training examples) and output likely faulty parts of the model; and a theory reviser, which inputs the faulty parts and examples and outputs suggested revisions to the model. In particular, we concentrate on the problems encountered when applying these tools to a real application: a requirements model containing air traffic control separation standards, operating methods and airspace information.

“…It draws on ideas from other areas of computing, and is inspired particularly from our work in requirements capture 0-8186-7686496 $05.00 0 1996 IEEE [17] and formal methods [21]. The method's benefits include (i) it can be used to systematically create models for classical planners, (ii) it produces a tight model (the property of a valid state is well defined, and operators can be checked for their operational consistency), (iii) it forces attention on the semantic level of object rather than the literal (iv) we have empirical evidence that the model compilation tools associated with the method produce a huge speed-up in plan generation.…”

Section: Introductionmentioning

confidence: 99%

Object-centred planning: lifting classical planning from the literal level to the object level

Proceedings Eighth IEEE International Conference on Tools With Artificial Intelligence

Kitchin

Porteous

A great deal of emphasis in classical AI planning research has been placed on search-control issues in plan generation, while the issue of knowledge representation and acquisition of models for use with classical planning engines has been largely ignored. Work in knowledgebased planning, on the other hand, is open associated with 'scrum' AI, there being no standard representation languages with associated fomuzl semantics for encoding domain models. In this paper we describe a method to create a planning domain model which preserves the domain independence, generality and 'clean' properties of generative planners to which the model can be attached. Our method is based on lifling the level of domain representation from the literal-centred, to the object-centred. This object-centred method has the advantage that it naturally allows for the creation of a supporting tools environment to help in (i) the creation and validation of a precise planning model, and (ii) the speed-up of plan generation.