It has been announced that Viscovery Software GmbH, a long-established provider of data mining systems based on self-organizing maps (SOMs) since 1994, will be closing down. SOM is one of the well-known data mining techniques, and although open source and commercial software exist, most of them are incompletely implemented and are insufficient for practical data mining. For at least the past 25 years, Viscovery SOMine has been the world’s leading SOM data mining system.
For example, one of the world’s leading automobile manufacturers has been analyzing the global automobile market with Viscovery SOMine for many years, and SOM data mining has been one of the company’s top secrets that gives it a competitive advantage. Viscovery SOMine was developed in the 1990s and was somewhat at odds with the recent data science boom, but to some experts it was the secret weapon that made truly practical data mining possible. If Viscovery SOMine becomes unavailable, it is expected that this will cause inconvenience to users around the world.
Mindware Research Institute has been the exclusive distributor of Viscovery SOMine in Japan for the past 25 years and is a professional in SOM data mining. ConceptMiner is a concept structure analysis system developed by Mindware Research Institute. Until now, ConceptMiner has implemented only GNG+MST instead of SOM to avoid competition with Viscovery SOMine. Of course, GNG+MST has the advantage of being able to represent more complex topologies than SOM. On the other hand, SOM offers the advantage of readability.
With Viscovery SOMine now defunct, there is no longer any reason not to implement SOM in ConceptMiner. Mindware Research Institute repositioned ConceptMiner as the enterprise text and data mining system to replace Viscovery SOMine. Currently, ConceptMiner is deployed as a SaaS solution. However, to release the enterprise version more quickly, we plan to launch a version that includes SOM and is designed to run locally on PCs at an early stage.
