Discussion

In the following lines, wehat are relevant for aesthetics validation and th discuss both the results and the validation of the preceding results. First of all, the fact that results indicate 4 representative metrics out of 12 do not imply that the other metrics have to be rejected when considering potential elements for aesthetics evaluation. Indeed, it just points out that the results were not significant for the remaining 8 metrics. Another possible interpretation is relative to the metric itself that may be reconsidered in order to obtain relevant results.

Regarding the validation of the whole experiment, a critic has to be made. Indeed, we used a set of only 4 UIs and that implies a lack of significance in the results. Therefore, we cannot assess a sufficient margin error in the constitution of previous rankings to be assured that they are indeed well representative. The results obtained gives thus an indication of a possible GUI evaluation by metrics but need to be treated with caution and request further empirical support.

Finally, another critic that can be made is the subjective nature of the UI grid transformation process that does not follow specific rules and is let to the judgement of the researchers. Indeed, another grid proposition for a UI could change considerably results obtained by the metrics. It should be appropriate for a metrics-based evaluation to rely on a consistent process for defining UI regions of interests.