Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Extraction of ideas from microsystems technology

: Thorleuchter, Dirk; Poel, Dirk van den

Preprint urn:nbn:de:0011-n-2114657 (89 KByte PDF)
MD5 Fingerprint: 15a3eead59f01e48b6e095699d475c8e
The original publication is available at
Created on: 16.8.2012

Jin, David (Ed.):
Advances in Computer Science and Information Engineering. Vol.1 : Proceedings of the CSIE 2012, May 19-20, Zhengzhou, China
Berlin: Springer, 2012 (Advances in intelligent and soft computing 168)
ISBN: 978-3-642-30125-4 (Print)
ISBN: 978-3-642-30126-1 (Online)
International Conference on Computer Science and Information Engineering (CSIE) <2, 2012, Zhengzhou/China>
Conference Paper, Electronic Publication
Fraunhofer INT ()
idea mining; microsystem; technology; textmining; knowledge discovery

In literature, idea mining is introduced as an approach that extracts interesting ideas from textual information. Idea mining research shows that the quality of the results strongly depends on the domain. This is because ideas from different domains consist of different properties. Related research has identified the idea properties for the medical domain and the social behavior domain. Based on these results, idea mining has been applied successfully in these two domains. In contrast to previous research, this work identifies the idea properties from a general technological domain to show that this domain differs from the two above mentioned domains and to show that idea mining also can applied successfully in a technological domain. Further, idea properties are identified by use of backward selection as main approach in stepwise regression, which is in contrast to previous research. Predictive variables are selected considering their statistical significance and a grid search is used to adapt the parameters of the idea mining algorithm. Microsystems technology is selected for a case study. It covers a wide range of different technologies because it is widely used in many technological areas. The case study shows that idea mining is successful in extracting new ideas from that domain.