Semantic mapping (SM) is a statistical method for dimensionality reduction — the transformation of data from a high-dimensional space into a low-dimensional space. SM can be applied to a set of multidimensional vectors of features to extract a few new features that preserve the main data characteristics.
Overview
SM performs dimensionality reduction by clustering the original features into semantic clusters and then combining the features mapped to the same cluster to generate a single extracted feature. Given a data set, this method constructs a projection matrix that can be used to map a data element from a high-dimensional space into a reduced-dimensional space.
Applications
Semantic mapping can be applied in the construction of text mining and information retrieval systems, as well as in any system that manages vectors of high dimensionality. It is an alternative to other dimensionality reduction techniques such as:
- Random mapping
- Principal component analysis (PCA)
- Latent semantic indexing (LSI)
Key Publications
- Corrêa, R. F.; Ludermir, T. B. (2006). "Improving Self Organization of Document Collections by Semantic Mapping." Neurocomputing, 70, 62–69. doi:10.1016/j.neucom.2006.07.007
- Corrêa, R. F.; Ludermir, T. B. (2007). "Dimensionality Reduction of very large document collections by Semantic Mapping." Proceedings of 6th Int. Workshop on Self-Organizing Maps (WSOM). ISBN 978-3-00-022473-7.
Distinction from Other Uses
The term "semantic mapping" is also used in other fields, including literacy education (graphical models for vocabulary learning) and the Semantic Web (transformation of data elements between namespaces). In statistics, however, it refers specifically to the dimensionality reduction method described above.
