Why are you doing this?
Many researchers in computer science are unhappy with the current conference publication system. This is reflected in several papers by well-known computer scientists, e.g., in ACM Communications, as well as recent changes in reputable conferences. There is criticism on the highly variable quality of conference reviews; proliferation of conferences (such that the main purpose of conferences, bringing the community together, is lost); high reviewing workloads; slower reviewing for journals; the decrease of information content of conferences and journals alike; lengthy journal articles; confusion among scientists in interdisciplinary contexts; unfair evaluations of academics. With the modified system, ECMLPKDD intends to support the movement towards new publication models.
Why would one submit an article to the journal track and not to the journal itself?
There are several benefits to the ECMLPKDD journal track:
- The ECMLPKDD journal track intends to have faster turn-around times.
- For articles in the special issue an abstract will be included in the LNAI proceedings as well.
- The authors will be granted the opportunity to present their work at ECMLPKDD.
- Authors of promising work that is not of sufficient maturity or depth (yet) for the journal, will be offered the option to present their work at the conference, and have a shortened version in the conference proceedings.
Why would one submit an article to the journal track and not to the proceedings track?
There are several benefits to the ECMLPKDD journal track:
- Articles can be submitted and accepted to the journal track all year long, and get published online well before the proceedings.
- Papers with minor flaws can be revised and resubmitted within the same track, providing more opportunities to get the paper accepted for the conference.
- Reviews are likely to be more informative.
- Restrictions on the paper’s length are less stringent.
Doesn’t this have a negative impact on the ECMLPKDD conference proceedings?
First, all submissions will be represented in the ECMLPKDD LNAI proceedings, as it is currently also the case. The journal articles will be included as an abstract, while for other papers the full text will be included. Springer has agreed that all abstracts will include a link to the extended article in the journal; furthermore, Springer proposed that any institution with access to LNAI proceedings will have access to these journal articles as well. (Negotiations pending.) The conference proceedings will hence provide a complete picture of all work presented at ECMLPKDD.
Second, we believe that it is a misunderstanding that the most impactful articles will appear in the journal, or that the conference proceedings will only contain second rate papers. Articles in the journal will need to be novel and inspiring, but also sufficiently mature, in the sense that they should present a sufficiently complete analysis (theoretical or empirical). Promising, yet not fully evaluated ideas, will only be published in the conference proceedings. It is very likely that some of the more interesting ideas will first be published in the conference proceedings. We believe that the overall perceived quality and interestingness of ECMLPKDD will only increase as a result.
Doesn’t this affect the ranking of ECML/PKDD in conference rankings?
This is only a risk for rankings that are based entirely on citation indexing and that count only citations to full papers in the proceedings of the conference. To the best of our knowledge, there currently is no commonly accepted conference ranking system of this kind. Popular existing conference rankings are based on the conference’s reputation in the field, which we expect to be positively affected by the new model. By clearly linking journal articles and conference abstracts to each other, journal articles will be recognized as part of the ECMLPKDD conference.
For potential Guest Editorial Board (GEB) members:
What do we expect from GEB members?
We expect that GEB members:
- read our regular mails concerning submitted articles;
- bid on the articles of interest to them;
- provide timely reviews of a paper, if they made a bid on it and it was assigned to them;
- produce high-quality reviews: well-motivated, in-depth, and constructive in nature;
- agree that their performance will be monitored: in the worst case, they will not be invited to an ECMLPKDD editorial board in the future.
If GEB members agree to be a member of the ECMLPKDD programme committee as well, their reviewing load for the proceedings track will be reduced proportional to the number of articles reviewed in the journal track. Our aim is that GEB members are a member of the programme committee without reviewing papers submitted to the proceedings track.
Why would I participate as a GEB member?
There are several reasons why participating in the GEB is interesting:
- reviews performed for ECMLPKDD count for your performance as ML or DMKD reviewer;
- you can reduce your reviewing load at the time of the conference deadline;
- you get regular updates about state-of-the-art research submitted to ML and DMKD;
- you have the freedom to choose when you review;
- you participate in and contribute to the success of an interesting experiment.
How many articles do GEB members have to review?
Our aim is that no GEB member reviews more than 5 articles. We will invite additional GEB members if needed. GEB members reviewing articles over the page limit of 20 pages will have a reduced workload (and less strict reviewing deadlines).
How do you intend to allocate articles? Is the expertise of reviewers taken into account?
Reviewers will be allocated manually by the program chairs in the Springer system. In principle the chairs will allocate an article only to GEB members that made a bid on the paper. However, if a paper did not receive a sufficient number of bids, we will follow the traditional procedure and invite reviewers we believe to be knowledgeable. These reviewers are free to reject our request (although we hope that this will not happen very often).
- detailed description of the new publication model