| id |
diss_cole |
| authors |
Cole, R.J. |
| year |
2000 |
| title |
The Management and Visualisation of Document Collections Using Formal Concept Analysis |
| source |
Griffith University, Australia |
| summary |
This thesis proposes a methodology, notation/theory, and software framework for organising documents using formal concept analysis. The documents are organised for the purposes of retrieval and analysis using background information in the form of a taxonomy of terms. An emphasis is placed on the development of a methodology that employs scalable computer programs to assist humans in the process of organisation, retrieval and analysis of document collections.The text retrieval community has also been concerned with the organisation of documents. The work outlined in this thesis makes use of the results of the text retrieval community at its lowest layer. Above this layer formal concept analysis is used as a mechanism to allow users to organise document collections using views determined by small numbers of attributes. These views, also known as scales, can make a mixture of coarse and speci c distinctions between documents, and are either selected or created by the users to make precisely the distinctions between documents that are important to their current tasks.The primary tool for the presentation of the results of formal concept analysis is a line diagram. The e ectiveness of the presentation of information contained in a line diagram is heavily dependent on the quality of the diagram. To support users in arriving at a quality diagram for a newly created view, graph drawing algorithms are adapted to the special case of determining a good layout for a concept lattice. This task is di erent from traditional graph layout problems because lattices exhibit a high degree of structure which should be exploited and made evident in the nal diagram. A new layout algorithm is proposed that combines a layered diagram approach and an additive diagram methodology. This new hybrid algorithm is shown to produce better diagrams than other adapted graph drawing algorithms. |
| series |
thesis:PhD |
| more |
http://www.kvocentral.com/ |
| full text |
 file.pdf (1,322,478 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2003/11/28 07:36 |
|
