[ Bristol CS | Index | Research ]

Declarative Languages

A declarative language is a higher level one where you specify what you want, not how to get it. Such languages are particularly useful to provide higher level interfaces to underlying complex systems. A simple example is HTML where, for example, you specify what is to appear on a page, but not how it is to be laid out. Another example is SQL where you specify what you want out of a database query, but do not give code for the looping and testing needed to produce it (see the Databases unit). There are modelling languages such as CCS (see the Concurrency unit), or UML (see the Design Methodology unit) and formal specification or proof languages.

These are special purpose languages, but there are also general purpose declarative languages, in which any program can be written in a declarative style. These particularly include the functional language Haskell (see the Intro Software Engineering unit), and the logic programming language Prolog and its derivatives (see the Artificial Intelligence unit).

This subject overlaps with the interests of other research groups, e.g. formal methods, so people such as Kerstin Eder may also be possible supervisors.

Animation (Ian Holyer)
Produce a graphical tool for checking and animating an object-oriented system modelling language (such as UML/OML/SOMA).
Modelling (Ian Holyer, Kerstin Eder)
Create a graphical tool for creating models of state transition systems, such as Petri nets, and a translator from the graphical form into a formal language suitable for checking properties of a model.
A deductive interface to a relational database. (Kerstin Eder)
Logical inference rules could be applied to deduce information from the data, and check sophisticated integrity constraints. A logical language can be implemented in a conventional host language (such as Java) and interfaced to an SQL database (using JDBC for example).
Constraint programming library for imperative languages. (Steve Gregory)
Constraint solving provides algorithms for solving certain kinds of search problems efficiently. Design a Java API/library for constraint programming.
A Concurrent File System for Haskell (Ian Holyer)
Currently, file I/O in Haskell is forced to interact sequentially with the file system, in order to match the sequential design of the operating system. It is possible to provide a more natural, direct, and functional, design for file access by keeping track of file versions.
Expert system and knowledge base systems (Kerstin Eder)
Identify an interesting application area. Build an expert or knowledge base system. Investigate algorithms for knowledge acquisition, data manipulation, data consistency checks etc.
Theorem proving and formal verification (Kerstin Eder)
Model an application (e.g. an integrated circuit or a controller) in a logic. Test, analyse and extend (existing) proof procedures for this logic. Define properties of the application and prove them.