Projects available for honours

Here is a list of projects available for honours. These are just proposals, do not expect detailed specification, the first part of doing honours is typically doing research what can and should be done, so you should not consider any aspect of the proposals to be fixed -- feel free to ask us if you would like to do a project slightly differently and feel free to propose your own project if you do think it will fit into the context of our work. Click on the project titles to get more detailed information.

Software Refactoring Using Formal Concept Analysis

Modern software development approaches often rely on a technology called refactoring: enhancing the quality of software by changing the code into better designs without actually adding functionality, typically in small steps. One of the main questions that arise in this context are finding the parts of the software that have to be refactored and how. This project shall try to use a data analysis technique called Formal Concept Analysis to answer these questions.

Designing an XML processing model

XML is used not only for storing information, but also for processing it using technologies like e.g. XSLT. Though available technologies are already quite powerful, a higher level model for modelling information flows is still lacking.

Generic Framework for Information Visualisation

This project involves creating a front end for presenting a visual representation of information stored within a "triple store". The presentation is conditioned by (i) meta-data attached to the information, (ii) the context of the user, and (iii) the logical structure of the data. Presentation forms include indented lists and tables, and line diagrams of concept lattices.

Data Mining with Formal Concept Analysis

Formal Concept Analysis has a tradition of being used as analysis tool for relational database systems. It gives a quite different view on the data compared to classical, numerical and statistical analysis techniques. Usually it is used independent of those classical techniques, but more modern developments like ToscanaJ allow combining both, which should be investigated in this project.

A Dynamic Hyper-linking system based on Text Compression Algorithms

Experience with text compression algorithms has shown how string encodings can be used to address text without the use of a keyword index database. Text retrieval systems that use this idea have been engineered and these show that a dictionary (or ontology) can be used to drive text search to precise (and repeated) hits in the text. The trade-off is that more space is used that with a conventional keyword database indexing solutions. This project involves the development and study of a dynamic hyper-linking system based on this idea that allows a user to select a number of keywords from a dictionary that will then dynamically thread a text collection without recourse to a index database. Performance would need to be measured against a traditional keyword index solution. Almost any programming language or environment can be used to engineer the idea. The student should however be interested in algorithm design and complexity analysis for this project.