|
Describing Time-Varying Data: A
Corpus Analysis
Sarah Boyd
MRI Language Technology Group
When: Tuesday, 15th April 1997
Time: 11:30am
Where: Room E6A357, Macquarie University
Abstract:
Automatic text generation is the creation of linguistic content from some underlying source such as a knowledge base or database. Automatically generating textual descriptions of numerical data is a particularly useful task especially with the explosion of accessible online information. A textual description may replace a graphical representation of the data or may better explain key features. A great deal of data is time-varying in nature: e.g. stockmarket prices, government figures, patient medical records, computer network statistics and weather data. The added dimension of time in the data means that besides the standard measures of maximum, minimum and mean there are significant patterns and trends to describe.
This research is concerned with automatically generating descriptions of time-varying data. The first step in this process is an examination of existing descriptions of time-varying data. In this talk, I describe a corpus analysis of Australian daily financial newspapers and how this corpus analysis fits in with the broader aims of my research.
Enquiries: sals@mri.mq.edu.au
| Last modified: July 1997 |