Over the last few months, there has been an intense discussion within the Executive Committee of the Association for Logic Programming concerning the publication structure adopted in the International Conference for Logic Programming (ICLP) – the flagship conference of the Association.
Because of the extensive discussions, the diversity of the perspectives proposed, and the recurring disagreements, a number of initiatives have been planned to expand the discussion to the broad LP community. In particular, a panel on this issue will take place during the upcoming ICLP 2015 (in Cork, Ireland).
We, your beloved ALP Newsletter, would like to collect perspectives and opinions ahead of this panel, to enable a more comprehensive discussion and to ensure that all different points of view are present.
In the rest of this posting, you will find:
- A thorough review (PDF version) of the history of ICLP publications and a summary of the main perspectives, nicely prepared by Tomas Eiter and Francesca Toni (PC Chairs of ICLP 2015)
- A report PDF Version by Manuel Hermenegildo (former ALP president and PC Chair of ICLP 2010), describing some statistics and his view of the current format. The report refers to a previous document by Manuel where the motivations are deeply discussed. We have linked it here for reader’s convenience (2012 Report)
- A space (below) where YOUR opinion will be posted (anonymous if you so desire)
We would like to strongly solicit your opinions, perspectives, comments, criticisms, etc. on this issue. Please email them to
[email protected]
(with a note of whether you would like your posting to be anonymous or not) and we will post them in this discussion thread.
This discussion is crucial for the future of our association.
Agostino and Enrico
UPDATE:
- ACM has recently published an interesting article on this matter with some followups.
- Make sure to join the discussion at the meeting of the ALP during ICLP 2015!
OPINIONS
- A personal view of Agostino Dovier (ALP Newsletter co-editor, general chair of ICLP 2008 and program Chair of ICLP 2012) and Enrico Pontelli (ALP Newsletter co-editor, general chair of ICLP 1999 and program Chair of ICLP 2008):
The current model of publication was launched in 2010. As discussed in Manuel’s manuscript, the change was introduced to address a concrete and recurring problem – the deep difference between the model of publication in Computer Science (CS) and the other disciplines. In CS, conference publications are often the preferred venue – they are quick, but at the time they are highly selective and competitive, and regarded by the CS community as more reputed than many journals. This model is at odds with most other scientific disciplines, where conference publications are not refereed (or lightly reviewed), while journal publications are the norm. Journal publications in other disciplines are often short (e.g., 6 pages) and thus comparable in content and complexity to our typical conference publication. This divergence of publication models puts CS at a disadvantage in scenarios where CS researchers are “compared” to researchers in other disciplines – since, in absence of proper instructions, conference publications (composing the majority of CS resumes) would be easily discarded.Having the paper immediately published on a Journal is surely a positive effect of the current option.
But we believe this is the unique advantage. ICLP reviewers are suggested to judge the papers as “journal” papers.
The immediate effect is that consolidated works are in a big advantage towards new results or tools that are evidently interesting but not yet deeply described or extensively tested as required by a journal paper. Comments such as “The paper is an interesting contribution and opens new directions. However, it requires still some sets of experimental results to be accepted as journal paper” is not what we want to hear when we submit a new promising work to a conference. These papers might be accepted as TC. But, often, TC presentations have very small audience. This creates a delay in the spreading of new results within the community and a loss of interest in the participation to the meeting. Personally, I’d prefer the “old” schema with LNCS proceedings and subsequent TPLP special issue (with revised versions of the papers, only then reviewed as journal paper). Sure, we should ensure a fast track then, but as far as we know (have a look here https://logicprogramming.org/tplp/tplp-accepted-papers/) the number of TPLP “extra-ICLP” accepted papers is not too big to cause a delay of the publication of a ICLP special issue (with some efforts, it might be ready by december). We also add a graph on the number of submissions to the recent editions of ICLP proving that Udine is the best place where organizing ICLP … 🙂
Thus, the following are some points of concern:
* Even though PC chairs have done an outstanding job in coordinating the now significantly more complex reviewing process – there is still the perception and the anecdotal evidence of reviewers feeling the need of treating submissions as journal submissions, not as conference submissions. We have always viewed a conference as a venue for the community to exchange the newest, brightest, most exciting ideas. The conference is a place of discussion, debate, where ideas are presented, validated, refuted, refined. This goes completely against a model where reviewers seek a journal quality paper, with ideas that are polished, fully compared with any existing even vaguely related existing work. This perspective is not supporting innovation, promotes incremental advances.
* Even though we (LPers) know well how to “read” the list of papers appearing in TPLP, being able to distinguish regular publications from ICLP publications, this may not be the case for outsiders. We are running the risk of hurting the reputation of TPLP, as people from other communities may look at it as a journal that effectively publishes conference-quality papers.
* While the TPLP publishers have been exceptional in assisting ALP in these experiments, we are still facing the difficulties of a fixed
number of pages published per year – thus, ICLP may endanger TPLP by removing availability of pages for regular TPLP submissions, while in turn ICLP may have to limit the number of accepted papers to conform to publisher restrictions. The number of papers accepted in ICLP should be dictated exclusively by scientific quality, not by other technical criteria.So what model of publication we would like to see?
1. We were actually happy with the traditional model (e.g., Springer/MIT Press); this does not prevent the opportunity to organize special issues of journals, with invited extended versions of the top papers in the conference, processed using some type of fast-track. This would still improve the transition to journal publications but without impacting the conference.
2. Alternatively, we could follow the model used by certain conferences (e.g., some of the local ACM regional college conferences), which established a dedicated journal exclusively for the publication of the proceedings. This does not create confusion with regular journal submissions and relaxes the restrictions on number of pages (having the opportunity of spreading over several journal issues the proceedings).
One final thought – ICLP launched into this new model 6 years ago, and to date we can count very few other similar attempts by other computing communities. Why is this the case? Why even our closest relatives (e.g., CP, AAAI) have not jumped in and followed our lead? - A personal view by Francesca Toni and Thomas Eiter (PC chairs of ICLP2015): We strongly believe that the current ICLP proceedings situation is unsatisfactory: ‘- since 2013, not only technical communications get no visibility and no DOI or ISSN/ISBN, but they are in the supplementary material of the editorial/introduction on TPLP, so they are appendices to a publication with its own DOI and by other authors (the editors of the proceedings).- One may wonder whether they could be re-published at all by their authors, since in principle they are already published (authors need to sign a copyright form to grant TPLP permission to publish them).- Limited Visibility includes that the ICLP proceedings and the TCs are not indexed as such in DBLP — the entries of ICLP in fact stop in 2012.Technical communications play a big part in the ICLP programme, as
shown by the numbers (TCs are 50% or more):ICLP15 TPLP:21(conditionally accepted) TCs:25(accepted)
ICLP14 TPLP:25 TCs:22
ICLP13 TPLP:27 TCs:26
ICLP12 TPLP:20 TCs:37
ICLP11 TPLP:23 TCs:23TCs are thus important to actually render ICLP viable, in terms of participation.We discussed the possibility to increase the number of papers accepted as TPLP but understandably the TPLP editors-in-chief (Illka first and Mirek now) have been reluctant to set such a high target while at the same time guaranteeing the journal quality that TPLP requires. As ICLP15 PC chairs, we would indeed have had problems to meet that target this year, considering that authors of TPLP accepted papers only have about a month to revise and resubmit their papers, and reviewers only one week to check the revised papers. In addition to the quality issue, there is also a page budget issue for TPLP (the number of pages devoted to ICLP special issues).We considered various options to mitigate what we believe are the problems (and, considering the ICLP15 PC votes on alternatives for publishing TCs, the community also perceives as a problem). These options included in particular standard proceedings followed by journal post-proceedings. We felt that we could not take a drastically different approach without broader consultation with the ALP community. Therefore, in agreement with the ALP executive committee, we decided to consider solely variants of the current model, as discussed in the summary.For the decision about the future of ICLP proceedings, the following aspects are especially important, among other aspects:- long term strategy: whatever serious scientific publisher we will engage with, long term liaison will be important. ICLP made (perhaps too) many changes in the past already, sometimes rather quickly and after a short term liaison.
- authors’ needs wrt publication status: whether a paper has DOI (and as such considered to be published) or an ISSN number etc. Different countries do have different policies for recognizing papers. Furhermore, this also regards the issue of republication, especially if the TCs do not count as formal publication.
- visibility of the proceedings and the TCs in databases like DBLP and accessibility of the papers to a broad audience.
We think that this should be brought to the attention of ALP/ICLP community.
