• Unknown User
    Unknown User, 05/07/2014 01:36

    Ms. Ref. No.: PMC-D-14-00059 Title: Developing Pervasive Multi-Agent Systems with Nature-Inspired Coordination Pervasive and Mobile Computing

    Reviewers' comments:

    Dear Authors,

    Thank you for your patience! The review process took a bit longer than intended. Please take a look at the reviews. We encourage you to address the reviews and revise your paper. Please submit the revised paper along with a rebuttal to reviewer comments, by July 31. We will try to speed up the process to get the special issue published soon.

    If you have any questions don't hesitate to contact us.

    Best regards,

    Christian and Mohan


    Reviewer #1: Contents: This is an article that elaborates on the results of the EU project SAPERE. The results can be summarized as a concept and implementation (including tests) of tools for the creation of pervasive multi-agent systems. This is an extension of the autonomous agent concept by borrowing structures and processes from nature, in particular spatial environments. These environments may contain agents and other items of interest. Agents are equipped with a certain attribute called LSA, short for Live Semantic Annotation. These attributes are the actual objects to be deposited in the spatial environments. By their presence, an agent declares to be reachable for specific forms of interaction that can be expressed by some formal language designed for these LSAs. LSAs as described in this article resemble tupels from the LINDA concept, although they are much more expressive. The reason for this approach, and its unique feature, is that the coordination of a set of agents is controlled by matching LSA with each other at the local environment. That is, instead of implementing the concrete activity and finally the complete plan of the agent, the authors state that the overall activity of the agent group emerges from the processing of matching LSA. Eventually, by the emergence of some kind of informational force field, information flows towards an intended direction. There are similar approaches in the direction, like Tschudin's Fraglets or, as cited, various chemical computing models. The authors add a description of a test scenario in the Vienna City Marathon where the SAPERE system proved to support spectators and security personnel in the dynamics of an urban event.

    Evaluation: Overall, this is an interesting concept, and there is a notable tradition in the concept coming from diverse research endeavors around methodologies for creating agent systems, self properties, emergence engineering, and so on - which have been in constant, warm debate for the last 10 years. Still, although looking promising, this approach feels like another bold move to mastering self organization and emergence, but without a really clear perspective how this can be achieved for possibly complex scenarios. Maybe there is still some way to go after all, and - as mentioned - there are still a lot of open research questions, some of which actually actually concern the targeted application of this approach.

    Accordingly, the paper appears a bit too long and verbose for making its central points. I would argue that the same points could be described more concisely in some fewer pages. While reading, quite some interesting information is provided, but as with a long movie, it has some lengths.

    I am not sure whether it is possible to find more evidence for a general applicability and viability of this approach at this point; perhaps adding some discussion about limitations, or more sample scenarios to get a better impression what can be done. A bit less past project, a bit more future and potential.

    Thus I would recommend this paper to be published with minor modifications as described before, concerning the overall contents. I will now add some specific points for the respective chapters. Detailed evaluation Language: The paper is well written and clearly understandable. Some minor flaws are notable.

    Section 1. "of complex pervasive service system is recognizing that" : replace "recognizing" with "to recognize" (to avoid misunderstandings in this long sentence)

    Section 1. "Eventually, the paper summarizes" : replace "Eventually" by "Finally". Although both should mean the same, in my non-native understanding, "eventually" carries some notion of "after all", "glad that it happened".

    Section 2. "pervasive MAS intrinsically feature" : I am not sure whether this is a too strong claim. This still depends on the degree of autonomy of agents; is there an agreed measure for autonomy? I'm a bit cautious about such kinds of implications. Should there be some advise for system developers?

    Section 3. "when possibly brought to computational systems" : "possibly" is a bit weird at that point; better remove it or make clear what you want to imply by this "possibility" 

    Section 3.2. "suitably expressive" : is there some agreement what needs to be expressed?

    Section 3.2. "science of managing the space of interaction" : I would say that "art" is a better word here for "science"

    Section 3.4.4. "More generally, biochemical tuple space" : I expected to read "more specifically"

    Section 4.1. "sort shared coordination media" : add "of"

    Section 4.1. There are a bit too many "(...)" in this section. Please decide whether to include text or not, or to use footnotes. For the reader it is not always clear how to interpret text in parentheses.

    Section 4.3. (and other sections): The "eco-laws" and "LSA" should be explained at this point in more detail, with examples. They are treated as black boxes a bit too long.

    Section 4.3. The actual methods for manipulating the LSA space are less interesting and may be removed if space should be saved.

    Section 4.3.1. "unless bond to one": bound or bonded?

    Section 4.4. "The eco-laws form a necessary and complete core ...". While you try to present arguments in the following text, many points sound more like claims, and one need not believe them to full extent. Maybe there should be some support from external references. Although, on the other hand, this may become a bit too philosophical.

    Section 4.4.1. I suggest to add some more conceptual illustrations. It will help to digest this parts better.

    Section 4.4.2. Figure 5 should be explained in a bit more detail in the text. The cancel numbers are hop counts, right?

    Section 4.4.2. "However, the contextual information" : suggest to replace "However" by "In addition" since there is no real constraint of facts

    Section 4.4.2. "As an example, of particular relevance" : Sentence structure is somewhat broken. Please rephrase.

    Section 4.5: "form a smart phone" : replace "form" by "from"

    Section 4.5: "SAPERE naturally suits application scenarios … to to get a richer": remove one "to"

    Section 4.5: "and spectators and making the turning the phone" : remove "making the"

    Section 5.1: "Top vs. Bottom Up" : should read "Top-Down vs. Bottom-Up"

    Reviewer #2: The paper is concerned with nature-inspired approaches for coordinating agents in pervasive systems. It introduces a number of nature-inspired coordination models (stigmergy, chemical coordination, field-based coordination and biochemical coordination), describes the development of a middleware (SAPERE) aimed at supporting these types of coordination, and ends with a discussion of open challenges and research directions.

    The proposed middleware is potentially an elegant solution to developing self-adapting and self-organising pervasive systems. I enjoyed reading the paper; however, it is long and unfocused. The paper is excessively broad at the expense of sufficient depth and technical detail about SAPERE. The introductory matter (pages 1-10) is long and rambling, and should be considerably trimmed to get to the meat of the paper much faster. The paper is 40 pages long, with only ~16 devoted to SAPERE. There are quite a few minor grammatical errors that should be corrected prior to publication; however, these don't detract from the meaning of the paper. Some sentences in the introduction are overly long, hindering understanding (e.g., the second bullet point is a single sentence nearly 7 lines long).

    Section 2 could be removed or, at the very least, considerably shortened. The discussions later in the paper don't rely heavily on this material.

    The paragraph on page 7 beginning "Most of the complexity in computational systems..." requires clarification - specifically, what it is meant in terms of expressive power and interaction? Interaction is a term with many meanings in computer science and associated fields. Is it limited here to agent-agent interaction, or does it also include human-agent (software) or human-computer interaction?

    What is meant by associatively in the third bullet point on page 8?

    What is meant by unsupervised in the last paragraph, p. 11?

    The figures on page 13 are inadequately labeled (e.g., it is unclear what the small rectangles and arrows denote). There should be a legend, or at least a more detailed explanation within the caption.

    The section on SAPERE (Section 4) is arguably the strongest part of the paper. However, depth and analysis are missing. In particular, the section lacks a thoughtful analysis of the LSA matching approach, covering expressiveness, scalability, etc. How does this approach compare to other solutions, such as other component coordination languages (including top-down coordination models) and context query languages? What are the specific limitations of the current implementation supporting only "common order and duplicate insensitive (ODI) aggregation functions", and what are the implications for application developers? The propagation model seems to rely heavily on the concept of hops within the network: what are the implications with regard to the network topology and the type of pervasive system topology / application domain this model suits? What is the complexity for the application programmer in using this type of nature-inspired coordination model, which, while elegant, is radically different from most conventional programming approaches? The paper would be stronger with a discussion of all of these issues; arguably, this could displace the current Section 5 (Challenges and Research Directions), which is overly theoretical and broad.

    The Vienna City Marathon application example is very nice, but again the discussion is quite superficial. What were the challenges in developing the application, what was the network environment and topology, and what was the experience of the application developers in using SAPERE and its coordination model? Was it significantly easier to develop the application with SAPERE than without? The concluding sentence of the section on the VCM application needs to be adequately supported.

    Although the paper contains a large number of references (92), it doesn't have a related work section providing direct comparison with other middleware approaches for pervasive computing.

    There are many interesting aspects to this research, and, if the paper is reworked to provide deeper and more concrete analysis of the proposed middleware approach, I believe it would be of interest to PMC readers.

  • Unknown User
    Unknown User, 03/12/2014 17:34

    ========================================================================

    Ms. Ref. No.:  PMC-D-14-00059R2 Title: Developing Pervasive Multi-Agent Systems with Nature-Inspired Coordination Pervasive and Mobile Computing

    Dear Prof. Franco Zambonelli,

    I am pleased to confirm that your paper "Developing Pervasive Multi-Agent Systems with Nature-Inspired Coordination" has been accepted for publication in Pervasive and Mobile Computing.

    Comments from the Editor and Reviewers can be found below.

    When your paper is published on ScienceDirect, you want to make sure it gets the attention it deserves. To help you get your message across, Elsevier has developed a new, free service called AudioSlides: brief, webcast-style presentations that are shown (publicly available) next to your published article. This format gives you the opportunity to explain your research in your own words and attract interest. You will receive an invitation email to create an AudioSlides presentation shortly. For more information and examples, please visit http:www.elsevier.com/audioslides.

    Thank you for submitting your work to this journal.

    With kind regards,

    Behrooz A Shirazi, PhD Editor-in-Chief Special Issues Pervasive and Mobile Computing

Partita IVA: 01131710376 - Copyright © 2008-2021 APICe@DISI Research Group - PRIVACY