 |  |
 |
Abstract
The Semantic Map is an interface for visualizing and exploring of information spaces based on semantic relationships. Its interface model is based on information clustering. Semantically correlated information items are positioned near to each other, while items having little similarity are positioned further away. This is based on a combination of methods for semantic text analysis with a self-organising neural network (Kohonen Map). The text analysis encodes patterns of word occurences in individual texts as numeric vectors, while the neural network produces a map which positions similar vectors close to each other. Since the vectors encode semantic properties of texts, correlated texts will be grouped close to each other.The system-generated clustering provides a point of departure for discovering possible relationships that normally might not be perceived by the human user. The Semantic Map establishes a context for information coming from different thematic fields or archives with different categorisation structures.
Authors
- Jasminko Novak, Fraunhofer Institute for Media Communication, MARS Exploratory Media Lab › Biography
- Boris Müller, Fraunhofer Institute for Media Communication, MARS Exploratory Media Lab › Biography
- Martin Schneider, Fraunhofer Institute for Media Communication, MARS Exploratory Media Lab
Origination
Germany, 2001
Submission
jasminko novak, Sep 12, 2001
Category
- Software
Keywords
- Topics:
- archive |
- human-machine-interaction (HCI) |
- communication |
- information architecture |
- perception |
- datamining |
- knowledge management |
- hypermedia |
- artificial intelligence |
- self organization |
- knowledge spaces |
- data space
- Formats:
- hypermedia |
- internet |
- networked
- Technology:
- shockwave |
- java |
- XML |
- databank |
- neural networks
Additions to Keyword List
- information visualisation |
- Knowledge discovery