FeGeLOD
Please note: FeGeLOD is now developed further as the RapidMiner Linked Open Data Extension! The tools available from this website are no longer updated.
Basic Idea
Machine Learning and Data Mining tools are used to make predictions and generate explanations on data. For example, the sales of a new book can be predicted based on some features of the book (its genre, its author, its length, its cover design...), using previous sales figures of other books.
Sometimes, it is desireable to have more features for making more precise predictions. In the book sales example, more features about the city where individual book stores are located may lead to more precise predictions, such as "crime novels are sold well in larger cities". FeGeLOD is a tool that aims at automatically creating such features from Linked Open Data (for the current version, only DBPedia is used, but we work on supporting more endpoints).
Fig. 1: Principle workflow of FeGeLOD
Downloads
Note: Software is provided as is, without any warranty, including any damage done to hardware and/or software.
FeGeLOD is provided as open source and may be used and altered freely. For external software or data that may be included in the distributables like libraries or datasets, please contact the original authors for their terms of use.
fegelod-0.1-bin.zip
fegelod-0.1-src.zip
Preliminary Evaluation
This page shows the quality of the data mining features generated by FeGeLOD, using different data sets and different quality measures.
Acknowledgements
FeGeLOD makes use of Jena, Weka, and the data contained in dbpedia.
Contact
For any questions and/or suggestions concerning FeGeLOD, please contact Heiko Paulheim.
Further Developers (past and present): Raad Bahmani, Alexander Gabriel