• Andrea Omicini
    Andrea Omicini, 24/05/2016 17:10

    Dear Andrea,

    siamo felici di comunicarti che  il tuo articolo

     Labelled Variables in Logic Programming: Foundations

    e' stato accettato per la presentazione a CILC 2016.

    Sotto trovi i commenti dei revisori, da prendere in considerazione per la
    preparazione della versione finale che verra' pubblicata su CEUR.

    Il lavoro va consegnata entro martedi' 14 giugno,
    e' possibile estendere l'articolo di una pagina. 

    Invieremo al piu' presto le istruzioni dettagliate per la preparazione
    della versione finale e per l'iscrizione alla conferenza.

    Arrivederci a Milano!

     Alberto, Camillo, Ugo

     REVIEW 1 -
    PAPER: 7
    TITLE: Labelled Variables in Logic Programming: Foundations
    AUTHORS: Roberta Calegari, Enrico Denti, Agostino Dovier and Andrea Omicini

    OVERALL EVALUATION: 0 (borderline paper)

     Review -
    The paper introduces an extension of Logic Programming (LP) based on "Labeled Variables" in order to support novel application scenarios that the authors refer to as "distributed situated intelligence".
    The paper introduces first (Section 2) the notion of rules, programs and domain, essentially extending those of LP with a domain of labels and function describing the coherence of (Herbrand)-terms and labels. Then the authors  introduce (Section 3.1) the denotational semantics as the least fix point of the "one-step" consequence operator (as it is usually done in LP) that is proved to be monotone, continous and thus having a least fix-point (which is a model). Operational semantics (Section 3.2) is defined by means of derivations and the main result (Proposition 2.2) states the correspondence between denotational and operational semantics. The paper concludes (Section 4) with two nice case studies 
    numeric intervals and WordNet network  showing the importance of labeled variables.

    Overall, I think that the Logic Programming with Labeled Variable extends LP to many interesting application-domains. However, the presentation of the paper is in many places sloppy (see below), but hopefully this could be improved in the camera ready version. What really prevent me to give an higher score is that the presented results are neither deep nor surprising: all the proofs go through almost exactly as in the standard LP. This makes me think that they should be instances of some more abstract approach, like the one pursued in
    Gianluca Amato, James Lipton, Robert McGrail: On the algebraic structure of declarative programming languages. Theor. Comput. Sci. 410(46): 4626-4671 (2009)

    Minor comments:
    I have found Sections 1 and 2 a bit hard to follow. Maybe having a concrete example could help the reader to understand what is going on. Maybe you can already think to move the Case Studies in the Introduction as motivating examples.

    At the end of Section 3, Table 1 is described: "lo lighten the notation, undefined elements in the tuple are omitted", but i do not see any undefined elements in Table 1.

    Page 4, Line 6: "via by" > Either "via" or "by" not both...
    The following sentences "For this purpose, ... the basic model above" are a bit confusing in my opinion

    Page 6, Line 6: "Truthness of f_C is based on the LVLP domain X". What does it mean? X is just a pair. Maybe X should be a triple further containing f_C?

    In the definition of T_P, "r" occurs with two different characters. This is really confusing because makes the reader thinks that they are different rules...

    After page 7, the order of body and labelings in the rule changes! Before page 6, a rule was head:-body, labeling and after is head:-labeling,body. This is really confusing. Please try to be as more uniform as possible!!!!

    Page 7. What is CLP(X)? Constraint Logic Programming? If so, please spell it in full.

    In Page 8, in Example 3, in the clauses for Animal, it occurs the notation " X^[...] " but this is introduced only in Section 4.

    Proof of the Corollary 1. Why do you need both Knaster-Tarski and Kleene? One of the two is not enough?

    Page 9, Section 3.2: What is "Colmerauer for Prolog III"? Please, at least put a reference!!!

    I am a bit confused about Proposition 2. As I said, I guess that this is totally standard, but what puzzle me is the fact that the clause of the body are resolved following the order of the list. Does not this make us loose some solutions?

     REVIEW 2 -
    PAPER: 7
    TITLE: Labelled Variables in Logic Programming: Foundations
    AUTHORS: Roberta Calegari, Enrico Denti, Agostino Dovier and Andrea Omicini

    OVERALL EVALUATION: 1 (weak accept)

     Review -
    In this paper, a new notion of truth for logic programs extended with labelled variables, interpreted in non-Herbrand domains, is presented.

    In addition, both the denotational and operational semantics are discussed, and some testbed case studies are presented in order to show the effectiveness of the proposed approach.

    The paper has some good ideas in it, and the topic is interesting. However, it seems to be unclear along some lines making the technical and presentation aspects appear weak. Some comments and suggestions, which hopefully can help the authors improve the study, are listed below.

    Section 3 and Section 4 should be explained more clearly. No significant limitations are discussed in Section 5.
    It may be worthwhile to mention the tradeoffs involved in choosing the presented approach.

    Relevance of contribution.
    Some additional references could be useful to describe some related work in order to emphasise the novelty of the presented work w.r.t. the existing one. 

    Reference to the literature.
    I think the motivation for this study need to be made clearer. A more detailed overview of the state of art and context problems would be a welcome addition in order to to emphasise the motivation of the work.

    Section 1, Page 1.  
    The authors could add some references to justify the statement "[..] a pure LP approach seems not to fit well the needs of situated systems".

    The autor should check the grammatical correctness and avoid mistakes, in order to improve the quality of the paper. For instance:

    Section 2, Subsection  2.2, Page 4:
    "While this choice clearly confines the impact of labelling, keeping the label computational model well separate from the LP one, it may also result to be too restrictive in practice: in fact, application scenarios often need to express some relevant properties via by suitable terms, so that they can in uence the label computation."

    The author should pay attention to some expressions ("separete", "may result to be", "via by").

    Section 2, Subsection  2.3, Page 4:
    The expression "This is as to say that" should be substituted with "This means that"

    The author should correct the following
    "there exists at least an element of the domain which the interpretations of t and l both refer to the sameventity"

    Section 3, Subsection  3.1, Page 8:
    Example 3: The author should use quotation marks correctly.

    Definition 1: The author should improve the item list.

    Section 3, Subsection  3.2, Page 11:
    Typing mistake: Appendixfor

    Section 4, Subsection  4.1, Page 11:
    The author should improve the following "By suitably exploiting labelled variables, however, the query can be rened by specifying the desired interval in the goal, and exploiting the combining function to get only the admissible solutions (coherently to the fL function)."

     REVIEW 3 -
    PAPER: 7
    TITLE: Labelled Variables in Logic Programming: Foundations
    AUTHORS: Roberta Calegari, Enrico Denti, Agostino Dovier and Andrea Omicini

    OVERALL EVALUATION: 2 (accept)

     Review -
    This work presents a foundation for logic programming extended with
    labelled variables. The paper is well-written, however, it needs
    some more concrete scenario and example. The authors create a lot of
    expectations in the abstract and in the introduction concerning
    "distributed situated intelligence" that however is not satisfied in
    the remaining of the paper. This does not means that the paper is
    not sound or it is nor relevant, just that the hype created is not
    satisfied. It would be better to introduce some practical scenario
    about distributed situated intelligence and showing how the proposed
    solution may help more than current solutions. A comparison also
    with CLP and other constraint languages may help as well to understand
    the proposal. 

    The proposal is sound to the best of my knowledge. It deserves to be
    presented and discussed at the venue.

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