RILEGGERE E CORREGGERE:

// certainly non-optimal presentation and structure. There are also quite many
   typos
/ English: lots of errors in the third-person conjugation of verbs (missing s),
  please correct them - I will not list all of them. Try to have a native
  English speaker read the paper.
- Sessions 2 and 4 are too long, there are many repetitions.


ROBA SOSTANZIALE:

// The work presented in this paper, however, is still in a preliminary stage
   and convincing, novel scientific aspects are hardly identifiable
   (vedi sotto "Possibili spunti")

Possibili spunti:

NO:   H-Bugs is an improved realization of the O-Ants approach; the
      architecture of H-Bugs is superior to that of O-Ants and H-Bugs
      provides more or at least some improved functionalities.
YES:  H-Bugs is simply an adaptation of a subset of the functionalities
      of O-Ants to the MONET web-services/brokers approach.
HHMM: The paper presents a case study with H-Bugs that gives novel insights or
NO    provides new evidence for the usefulness of systems like H-Bugs and
      O-Ants. (Because the title says  ...: a case study  this is what the
      reader first expects).

(ii)  In the conclusion the authors admit that their work is rather
      preliminary and that a real integration into the
      MONET architecture is still missing. In  addition the following critique
      should be addressed: The representations for  proof status  and  hints
      in the brokers and tutors are client specific. What level of abstraction
      is thus reached by the approach over the peculiarities and
      representation formats of the client systems? Is it a feature that for
      each client-tutor
      combination new representation formats have to be agreed upon and
      provided? I was actually hoping to find some more client independent
      modelings of tutor services. Since this seems to be not an issue it
      raises the question what the benefits of a modeling of H-Bugs within the
      MONET web-services / brokers approach generally can be? Where is the
      abstraction that is typically achieved in service/broker modelings (see
      second paragraph in the Introduction)? If there is no abstraction
      achieved over the representation format for proofs used by HELM what
      again is the difference to O-Ants: the main application of O-Ants
      operates on the OMEGA proof datastructure; however it has been claimed
      that the O-Ants architecture is independent from it and this has
      recently been illustrated by applications of the system in other
      settings (e.g. paper at MKM-1001, Festschrift-Siekmann).
(iii) A proper evaluation of the case study is missing and the authors
      themselves mention proper assessment as future work. The claim that the
      tutor embodying the HELM-Search-Engine has effectively increased the
      productivity  of users is not at all justified in the paper.