Solving Guesstimation Problems Using the Semantic Web: Four Lessons from an Application

The authors have addressed all important points and significantly improved the explanation of their approach. I think the paper is ready for publication.

Here are some comments which may or may not be addressed by the authors:

Regarding the "navigation service to improve distance calculations", I meant something like this:
https://developers.google.com/maps/documentation/directions/
http://msdn.microsoft.com/en-us/library/ff701717.aspx
http://tasks.arcgisonline.com/ArcGIS/rest/services/NetworkAnalysis

Better reference one of these services in order to make clear what you mean (whether you should reference me is up to you, but I am unsure whether my name as such is very informative ;)).

Regarding evaluation, I understand that quantitative evaluation can only be done if systems are comparable to some extent.

Regarding the units ontology, I understand that this is only helpful once it is used. However, your system could offer the possibility of using such an ontology, since measuring units are such an essential part of guesstimation tasks. Once it is used, it would immediately improve the guesstimation, which would again feedback on the use of the ontology, and so on. In my view, your "enforce" argument is not convincing. None of the heavily used ontologies on the Web, such as dublin core, e.g., were enforced. Note that offering the possibility of using an ontology is not the same thing as forcing people to use an ontology. Rather, people start realizing that using ontologies solves problems. Maybe one could put this to the outlook.

I believe the paper is now ready to be published. I have two small suggestions:

1) The abstract is a little thin and could be expanded.
2) I would at least provide an explanation for what GORT means when it is first introduced. It would be enough to spell out the acronym. That is, simply add GORT (Guesstimation with Ontologies and ReasoningTechniques) on p. 1