• Andrea Omicini
    Andrea Omicini, 17/08/2021 11:04

    Dear Andrea,

    we are glad to inform you that your paper

    A mechanism for reasoning over defeasible preferences in Arg-tuProlog

    has been accepted for presentation at CILC 2021.

    Please find below the detailed reviews for your paper.

    For the preparation of the camera-ready copy of your paper, please consider the following requirements.

    For original papers:
    1) The camera-ready copy must be uploaded in PDF format on EasyChair by August 25th.
    2) A ZIP file containing the sources of the camera-ready copy must be sent in reply to this message by August 25th.
    3) The camera-ready copy must be formatted using the Springer LNCS style with suppressed headings and page numbers (i.e., \documentclass{llncs}, with no options).
    4) The camera-ready copy must not exceed 15 pages (including references) for full papers and 8 pages (including references) for short papers.
    5) The camera-ready copy must have the title capitalized in title case.
    6) The camera-ready copy must have the following title footnote "Copyright \copyright\ 2021 for this paper by its authors. Use permitted under Creative
    Commons License Attribution 4.0 International (CC BY 4.0)."

    For non-original papers:
    1) The camera-ready copy must be uploaded in PDF format on EasyChair by August 25th.
    2) A ZIP file containing the sources of the camera-ready copy must be sent in reply to this message by August 25th.
    3) The camera-ready copy must be formatted using the Springer LNCS style with suppressed headings and page numbers (i.e., \documentclass{llncs}, with no options).
    4) The camera-ready copy must not exceed 15 pages (including references).
    5) The camera-ready copy must have the title capitalized in title case.
    6) The camera-ready copy must have a title footnote including the complete reference of the published paper from which the non-original paper is derived.
    7) If the paper was shortened with respect to the published version to fit 15 pages, please also add "(an abridged report)" at the end of the title.

    Note that at least one author for each accepted paper is required to register to the conference and present the paper. Only presented original papers will be included in the CEUR volume.

    Additional information about the participation to the event, in person or remotely, will be provided with a dedicated message as soon as possible.

    Please do not hesitate to contact us for further requirements.

    Best regards,
    Stefania and Federico

    SUBMISSION: 10
    TITLE: A mechanism for reasoning over defeasible preferences in Arg-tuProlog


     REVIEW 1 -
    SUBMISSION: 10
    TITLE: A mechanism for reasoning over defeasible preferences in Arg-tuProlog
    AUTHORS: Giuseppe Pisano, Roberta Calegari, Andrea Omicini and Giovanni Sartor


     Overall evaluation -
    SCORE: 2 (accept)


     TEXT:
    The paper addresses the problem of considering preference relations to decide the acceptability of arguments in structured argumentation frameworks (AFs).
    Conditional preferences can be expressed over defeasible rules to determine how to construct the arguments. Enabling defeasible reasoning on preferences allows determining the acceptance status of arguments that otherwise would be undecidable.
    Two are the main contributions: a generalisation of preference-based frameworks (previously proposed in the literature), and a working implementation of such a mechanism.

    The described implementation makes use of the core functionalities of Arg-tuProlog (a tool for AFs, based on prolog and previously introduced by the same authors).
    Preferences can be included in AFs by applying a two-step transformation that replaces particular attacks with arguments. AFs obtained as a result of the transformation can be then evaluated in the classical manner, i.e., semantics of preference-based AFs can be computed through Arg-tuProlog functions for standard argumentation frameworks.
    Therefore, the implementation part boils down to providing a prolog procedure realising the transformation.
    The authors also provided filtering strategies to improve the performance of the tool by keeping track of the attacks to be transformed.
    The whole transformation process and its implementation are clearly described and the many examples further help to understand implementation details.

    The proposed generalisation, then, consists of an extension of the framework in which also group preference attacks are allowed between arguments, in addition to single preference attacks. This extension is also implemented in Arg-tuProlog.

    As final considerations, the paper is well written, easy to follow, and presents novel enough contributions; the implementation is sound and the details are carefully explained. I, therefore, lean towards acceptance.

    Minor comments
    - Definition 1: identifier, antecedent and consequent of a defeasible rule are not defined
    - Definition 4: "argument B at 'B" -> "argument B at B'"
    - Definition 4: undermining attacks are mentioned, but neither defined nor used in the paper
    - Definition 4: the parenthesis in the second element seem wrong
    - Notation 5: "argument and an attack" -> "argument to an attack"
    - page 7, second line: "being is" -> "being it"
    - page 12, third line below the definition: ").If" -> "). If"
    - pag 12, first paragraph, last line: incorrect reference to the figure


     REVIEW 2 -
    SUBMISSION: 10
    TITLE: A mechanism for reasoning over defeasible preferences in Arg-tuProlog
    AUTHORS: Giuseppe Pisano, Roberta Calegari, Andrea Omicini and Giovanni Sartor


     Overall evaluation -
    SCORE: 2 (accept)


     TEXT:
    The paper presents two main contributions: (1) the implementation in Arg-tuProlog of the defeasible preference model introduced by Dung, and (2) an improvement of the computational mechanism for supporting arbitrary preference relations over arguments.

    I think that the contribution is significant, the topic is interesting and very relevant to CILC.
    Overall, the presentation of the main ideas is good, but both the writing and the technical details should be improved. Below I list some detailed comments.

    p.1 and elsewhere: "In [5]"
    As a rule of good writing style, one should not use references as nouns. In this case,
    Dung et al. [5] cope with the problem ...

    Def.1 (and also elsewhere)
    commas missing after dots 

    Def.2
    Why a Defeasible Theory is a tuple, and not just a set, of rules?
    Is the order of the rules relevant? Just before the definition it is said that
    "A theory is a set of rules"

    Def.3
    It is unclear what the basis of the induction is.
    I guess when n=0 and \phi is a literal (not an argument).

    The notions "closed under transposition" and "directly consistent" are not defined.

    Def.4
    'B should probably be B'
    "undermines" is not defined (and not used?)
    "\neg N(r)" is not a literal (unless one can use identifiers as atoms  this should be said)
    Why not use the \overbar notation for \neg\phi?
    There are some extra parentheses in the definition of "rebuts"

    Def.6
    "are \in" > "are in"

    Def.7
    What is an "abstract argumentation framework"?
    The only notion that has been defined is an "argumentation graph".
    You also use "structured argumentation framework" and "argumentation framework" with no
    adjective (and without an explicit definition). This is a bit confusing. 

    Subsec.2.3
    "premises that entails the conclusion." > "premises that entail the conclusion."

    "- proposition" should be "- connective" (or operator)
    Also, normally one does not negate terms, but formulas.

    p.7
    "being is" > "being it"

    Listing 1.1
    "circustances" > "circumstances"

    Example 3.
    Can parts of this example be shown ealier to illustrate the definitions of sec.2?
    In the text, the format of some identifiers (p1, p2) should be uniformed.

    Listing 1.4
    "thoery" > "theory"


     REVIEW 3 -
    SUBMISSION: 10
    TITLE: A mechanism for reasoning over defeasible preferences in Arg-tuProlog
    AUTHORS: Giuseppe Pisano, Roberta Calegari, Andrea Omicini and Giovanni Sartor


     Overall evaluation -
    SCORE: 2 (accept)


     TEXT:
    The paper shows how a conditional-preference mechanism taken from cited literature can be effectively implemented in Arg-tuProlog, an extension of tuProlog that provides a modular framework for structured argumentation based on logic programming. First, the considered (structured) argumentation framework is defined formally. Then, the adopted preference-based argumentation framework is briefly summarized using the adopted notation. As anticipated, the adopted preference-based argumentation framework is suitable for direct implementation in Arg-tuProlog, and the underlying transformation algorithm coded in Arg-tuProlog is shown and discussed together with a few examples. Such an algorithm is the basis for a second algorithm, which is a generalization of the first algorithm intended to tackle group preference attacks. The second algorithm is also discussed together with a few dedicated examples.
    The paper is well structured and it provides a self-contained description of the topic. It can even accommodate the needs of casual readers because the subject is split into small parts discussed incrementally.
    Unfortunately, the two proposed algorithms are studied only superficially, and the reader is left alone with open questions regarding the characteristics of the algorithms with respect to, e.g., termination, computational complexity, and relations with available alternatives (if any). I suggest that, at least, some of such characteristics of the proposed algorithms are briefly mentioned, e.g., as work-in-progress research to be discussed in future papers.

Partita IVA: 01131710376 - Copyright © 2008-2022 APICe@DISI Research Group - PRIVACY