Abstract
Data Modelers document their understanding of the users' work domain via conceptual models. Once a model has been developed, they ought to check that it has no defects. The literature has little guidance about strategies and tactics to improve the effectiveness of model validation. In this light, we propose a theory arguing that two factors have a major impact on the effectiveness of validation-namely, the (a) ontological clarity of the models prepared, and the (b) extent to which a validation method engages users more with the semantics of the domain represented by a model. We experimentally tested the theory in which we systematically varied the levels of these two factors. Forty-eight expert data-modelers participated in our experiment. Their task was to find defects in the model they were given. Our results showed that those who received the model that had greater ontological clarity on average detected more defects. We obtained no effect for the validation method that we predicted would engage participants more with the semantics of the domain represented by the model they had been given.
Original language | English |
---|---|
Pages | 31-36 |
Number of pages | 6 |
Publication status | Published - 1 Jan 2009 |
Event | Workshop on Information Technologies and Systems (WITS 2009) - Phoenix, United States of America Duration: 14 Dec 2009 → 15 Dec 2009 Conference number: 19th |
Workshop
Workshop | Workshop on Information Technologies and Systems (WITS 2009) |
---|---|
Abbreviated title | WITS 2009 |
Country/Territory | United States of America |
City | Phoenix |
Period | 14/12/09 → 15/12/09 |
Keywords
- Data model validation
- Experimental design
- Ontological clarity