Experimentation is an integral part of any engineering or science-based knowledge creation activity. Moreover, the data generated by a systematic set of experiments are the foundation for the generalizations which constitute the derived knowledge. For the purpose of completeness, validation, reproducibility and future use, it is critical that all experimental data be stored systematically, be sharable and be machine interpretable. Research organizations engaged in experimental work typically do provide "Data Management" plans which define the process for access to data. However, very often the interpretation of data is very difficult because the complete provenance of the data is not recorded. Over the past 5 years sophisticated tools such as Electronic Lab Notebooks (ELN) and Laboratory Information Management Systems (LIMS) have become available. These tools are very expensive and/or do not provide the complete information for easy access to data.
We have developed an experimental knowledge management system, which is based on the HUBzero platform and uses a unique workflow model to fully capture the provenance of the data. In addition, the workflow model assists the experimenter in adhering to the standard operating procedure agreed-upon prior to the initiation of the experimental campaign and in ensuring that the stipulated information is collected. A graphical tool that works on HUBzero was developed to build workflow models of any underlying experimental procedure. A database schema that uses the entity-attribute-value (EAV) model was designed to define the tables and attributes associated with the building blocks of the workflow. The workflow template provides a unique signature for an experiment which can be used in sharing information as well as accessing information. In this paper, we will illustrate the various aspects of this knowledge management system using a lab-scale pharmaceutical manufacturing line at Purdue University.
Girish S. Joglekar is a Senior Research Scientist with the Engineering Research Center for Structured Organic Particulate Systems (ERC-SOPS) at Purdue University. He has a Ph.D. in Chemical Engineering and M.S. in Computer Science from Syracuse University. Before coming to Purdue he was the President of Batch Process Technologies, Inc., a software development company specializing in discrete and dynamic simulation of recipe driven chemical processes. While at BPTech, he was also responsible for providing custom solutions to clients. Since joining Purdue in 2009, he has worked on developing ontology based and HUB based knowledge management technologies which support the research groups within the ERC. He has more than 25 publications.
Cite this work
Researchers should cite this work as follows: