Our research group at the Université de Montréal has launched the UML Labeling Initiative, aiming to build a data set of UML class diagrams and their associated human-language descriptions.
The utility of such data cannot be emphasized enough. At the very least, it will vastly improve the quality of many features of modeling tools, for example diagram search, diagram generation and documentation.
Upon reaching sufficient volume and maturity, the data set will be publicly available for future research.
This is a very short and modular exercise: participants are asked to briefly summarize fragments of UML diagrams in English. There are no requirements on number of diagrams to be described, or any time constraints. You are, of course, invited to label as many diagrams as your time allows.
Please refer to https://umllabels.herokuapp.com to help this great initiative.
For more information, please contact Song Yang at firstname.lastname@example.org.