KFCA demo allows the analysis of a confusion matrix (also called contingency tables) using a generalization of Formal Concept Analysis to produce a sequence of concept lattices that allows an improved observation of the groupings and hierarchies of errors.

In Formal Concept Analysis parlance the emitted stimuli or classes are called objects whereas received stimuli or classes are attributes, and confusion matrices, incidences.

Users can upload their own confusion matrices as a MAT file generated by Matlab (plain text files with labels for objects and attributes are optional) or as a CSV (Comma Separated Values) file with labels included. Download examples of these files: MAT confusion matrix, text labels or CSV confusion matrix with labels.

The description of the generalization of Formal Concept Analysis can be found in:

Applications of the method on phonetic and articulatory confusion matrices can be found in:

C. Peláez-Moreno, A.I. García-Moral and F. J. Valverde-Albacete. Analizing phonetic confusions using formal concept analysis. Journal of the Acoustical Society of America,Volume 128, Issue 3, pp. 1377-1390 (September 2010).

C. Peláez-Moreno, A. I. García-Moral, F. J. Valverde-Albacete. Eliciting a hierarchical structure of human consonant perception task errors using Formal Concept Analysis. In Interspeech 2009: 10th Annual Conference of the International Speech Communication Association 2009, Vols 1-5, 10th INTERSPEECH 2009 Conference SEP 06-10, 2009 Brighton, ENGLAND, Pages 824-827, 2009.

For any doubts, comments or suggestions please write to carmen or fva at tsc.uc3m.es

Acknowledgements: the preliminary version of this demo is due to Víctor Esteban Alonso as his Final Year Project (documentation in Spanish).