Description
In this talk we present an additive fuzzy clustering method for similarity data as oriented towards representation of activities of research organizations in a hierarchical taxonomy of its knowledge domain. We developed a one-by-one cluster extracting strategy which leads to an algorithm of spectral clustering for similarity data. The method, FADDIS, has been experimentally verified on simulated and real-world data. Specifically, we developed two synthetic data generators simulating affinity data of Gaussian clusters and genuine additive similarity data, with a controlled level of noise, and we discuss FADDIS results on these data in comparison with two state-of-the art fuzzy clustering methods. Then, we extend our analysis to “difficult” data structures taken from literature. Finally, we present FADDIS results when applied to in-house data of similarity between research topics according to the work of a research centre.| Period | 18 Mar 2015 |
|---|---|
| Held at | DI - Departamento de Informática |