Exposé and Workflow Construction

[William RAYNAUT]

Hello everyone,

I am currently interested in the topic of Data Mining Workflow Construction, and though it received quite some interest few years ago, there seems to be few ongoing research on that topic. The most interesting approaches to be found make use of data mining ontologies to support different kinds of workflow planning (see excellent review by Serban et al. in 2013 : A survey of intelligent assistants for data analysis). However, none seems to have reached an actual production phase and have been kept up to date since then...

Following the further developments of Data Mining ontologies led me to Exposé, which appears to aggregate some of the most interesting ontologie of the domain (OntoDM and DMOP), and have received the most recent attention (as it seems to be the foundation of OpenML database).

The only ressource I could find regarding Exposé, is a publication describing an early version of OpenML experiment database : 

- Vanschoren, J., Blockeel, H., Pfahringer, B., & Holmes, G. (2012). Experiment databases. Machine Learning, 87(2), 127-158.

Is Exposé still under active development ? If so, in which direction ? I would greatly appreciate directions to any ressource concerning Exposé most recent developments.

In particular, is there (or has there been) any work related to  Data Mining Workflow Construction using Exposé ?

Best regards,

William Raynaut