stigLD: Stigmergic Coordination in Linked Systems^†

Round 1

Reviewer 1 Report

The paper suggests using a value-passing fragment of Milner’s Calculus for specifying hypermedia-driven behaviors of agents. The work developed stigLD for facilitating the design and declarative implementation of sematectonic, persistent marker-based and transient marker-based stigmergic mechanisms within a hypermedia MAS. Two evaluation use-cases are employed to demonstrate the effectiveness and general value of the proposed framework.

The work is interesting and has merit of theoretical and practical nature. The presentation is in general accessible, even though some parts require additional improvement (noted in the following).

The content is consistent with the provided results and remarks made along the text.

There are some minor typos/grammatical errors across the manuscript – they can be easily spotted and corrected.

Section 1 needs to be modified and become more specific, especially lines 35-49. In the current form, the introduction does not indicate the specific problem addressed in the work and the original contributions offered compared to the existing literature. This section should be progressive and rather specific with respect to these aspects.

Similarly, section 2 should provide better the motivation for the proposed framework and the need to introduce it given the existing literature. Lines 103-106 should be modified towards this direction.

Lastly, it is necessary especially in the early parts, namely sections 1, 2, to make the manuscript more progressive. The employed process calculus is not particularly widespread in the corresponding community, and thus, to enable the layman audience, the introduction of the corresponding concepts should be more gradual.

Author Response

"There are some minor typos/grammatical errors across the manuscript – they can be easily spotted and corrected."

Typos and grammar errors have been fixed.

"Section 1 needs to be modified and become more specific, especially lines 35-49. In the current form, the introduction does not indicate the specific problem addressed in the work and the original contributions offered compared to the existing literature. This section should be progressive and rather specific with respect to these aspects."

We added a stronger motivation and more specific problem formulation in lines 34 to 48.

"Similarly, section 2 should provide better the motivation for the proposed framework and the need to introduce it given the existing literature. Lines 103-106 should be modified towards this direction."

We added a improved demarcation against previous works in lines 114-123.

"Lastly, it is necessary especially in the early parts, namely sections 1, 2, to make the manuscript more progressive. The employed process calculus is not particularly widespread in the corresponding community, and thus, to enable the layman audience, the introduction of the corresponding concepts should be more gradual."

We added a high-level description of CCS and point the interested reader to introductory material in lines 132-137.

Reviewer 2 Report

The motivation behind the paper must be better presented. 

The paper's main contributions must be stated at the end of the Introduction section. Preferably in bullet points.

Is the code somewhere published?

The whole pseudocode must be added to the paper.

Otherwise, the proposed model and algorithm are well defined, and a reader can easily follow the paper. The parameters used in the experiment are discussed, and so are the experiment's steps. The results are provided and well discussed. Used references are relevant and up-to-date.

Author Response

"The paper's main contributions must be stated at the end of the Introduction section. Preferably in bullet points."

We added our contributions as bullet points in lines 41-48.

"Is the code somewhere published?"

All source codes and materials are available online as indicated in the paper and via various footnotes.

"The whole pseudocode must be added to the paper."

Could you please clarify this point of review? All system components are formally specified using the presented CCS fragment. All system components are implemented in source code and available online (see above) .

Reviewer 3 Report

The paper presents the usage of a value-passing fragment of Milner’s Calculus to formally specify the generic  hypermedia-driven behaviour of Linked Data agents and the Web as their embedding environment. The Authors presented a domain model and a SPARQL function library facilitating the design and implementation of stigmergic coordination between Linked Data agents on the Web. The efficiency of the elaborated approach was tested in a Make-to-Order fulfilment scenario involving transient stigmergy and negative feedback as well as by solving problem instance from the TrucksWorld domain. The topic is interesting and the paper is well corresponding to the journal aim and scope.

The paper is well structured.

However, there are some minor shortcomings in this paper. All of the formulas are unnumbered. Equation numbering only appears in section 5.

The Authors omitted the limitations of the presented approach in the conclusions

There aren't many new references in the article.

Minor typos:

The description of P set presented in section 3.1. should be displayed separately – not directly in the text (e.g. P is a part of the text on the left)

The table header should be above the table (e.g. Tables 1-5)

In general, the article is complete; several elements need to be refined, which will improve the reception of the article as a whole.

Author Response

"All of the formulas are unnumbered. Equation numbering only appears in section 5."

• We used equation numberings only for equations we refer to in text.
• However, we enabled numbering for all equations.

"The Authors omitted the limitations of the presented approach in the conclusions."

• In lines 576-579, we added non-executability as a limitation of system specification using CCS.
• In addition, we point out ongoing works for auto-translation from CCS specs to executable behaviour trees.

"There aren't many new references in the article."

• The article is referencing 48 relevant pieces of previous works.
• 23% of all references are from 2018 or newer
• 29% all references are from 2016 or newer
• 60% of all reference are from 2006 or newer
• Could you please be more specific about this point of review?

"The description of P set presented in section 3.1. should be displayed separately – not directly in the text (e.g. P is a part of the text on the left)"

• The abstract syntax is now presented in a figure of its own.

"The table header should be above the table (e.g. Tables 1-5)"

• All table headers are now above the respective tables.

Round 2

Reviewer 1 Report

The paper has been revised and my comments have been almost all addressed. There are still some remaining typos, e.g., in the new version, all references are broken, and in the first three bullets in lines 41-48, no full-stop has been employed. A final careful proofreading can tackle such issues.

Author Response

"The paper has been revised and my comments have been almost all addressed. There are still some remaining typos, e.g., in the new version, all references are broken, and in the first three bullets in lines 41-48, no full-stop has been employed. A final careful proofreading can tackle such issues."

-Typos that we found have been fixed. 

-Sorry for the display error in the last review version, the references are all correct now.