First Publication Servers
The publication and review system of ETAI (the Electronic Transactions on Artificial Intelligence) is based on the idea that articles are made available to the research community in electronic form already at the beginning of the public review process, and with a guarantee of persistence so that the article can not be `improved' retroactively. It is intended that the originally published version of the article shall count for priority purposes, and therefore it must be permanently available without change.
The main mechanism for achieving this is the First Publication Archives (FPA's), that is, servers where articles are made public, and where considerable care is taken to guarantee the permanence and unchangeability of the works. It is foreseen that many major universities and research institutes will set up such FPA's, but this will take some time, in particular since each FPA must have a strong guarantee that the published works will remain on-line for an extended period of time (at least 20 years). Such commitments can only be made at a high level in the university or organization.
There is therefore a need for an intermediate solution. It was originally foreseen that the first participating FPA's would help out by accepting to publish works from other universities than their own for a limited period of time. However, some newly arrived copyright restrictions have made it necessary to add another intermediate solution to the first one. The present memo describes the problem and the solution.
It is common practice that work that has been published at conferences is later on resubmitted to a journal. Often, the author takes the chance to extend and modify the article in-between, but sometimes the conference article is ready for journal submission as is. The journal publication is often perceived as more valuable because the journal reviewing process is more careful than the one at conferences, due to the possibility of dialogue and feedback in the journal, and the haste that is intrinsic to the reviewing schedule of all conferences.
The ETAI is going to maintain at least the same reviewing requirements as the best conventional journals in its area, and add the qualitative advantage of public debate and feedback. We therefore wish to invite authors of accepted IJCAI and AAAI papers in ETAI speciality areas to resubmit their works to the ETAI.
The standard copyright agreement of IJCAI stipulates that the author is authorized to put his or her article on-line on a www or ftp server operated by herself or by her employer, but not elsewhere. The AAAI has a similar requirement. These rules are compatible with publication in an FPA provided that the author's university or institute already operates such an FPA, but not otherwise.
If the author makes nontrivial changes to the paper before resubmission, there is not a problem anyway. However, many authors may feel that their paper is ready for journal resubmission as is, or with so small changes that it does not qualify for the formula "the author may include parts of the present article in his own later works". We need to have a reasonable procedure for these cases as well.
Also, the problem does not only concern the IJCAI. The AAAI already has a similar rule, and one can foresee that additional conferences may impose similar restrictions as well. It would be undesirable to have a special rule for one or a few conferences; any ETAI rule should apply in equal ways to all conferences with properly published proceedings.
Basing ETAI submissions on paper copies of articles, that is, on the conference proceedings themselves would force us out of the electronic world and create a whole array of well-known practical problems. It is not a viable solution. From the ETAI point of view, there is therefore a need to complement the electronic posting of the article which can be done by the author's home institution, with the guarantee of persistence that is necessary for the ETAI review and publication scheme.
For these reasons, I propose the following addition to the previously established rules regarding FPA's and submission of papers to the ETAI.
An entity within a university or research institute that posts research articles on the Internet in an organized way may be recognized as a First Publication Server (FPS) for the ETAI if it satisfies the following requirements:
The ETAI will receive articles which have been accepted for publication at conferences in the same way as FPA published articles, provided that the following conditions are met:
If these conditions are met, the article will be treated in the ETAI system in the same way as if it has been published in an ETAI recognized FPA. Its date of appearance in an ETAI newsletter will count as date of first appearance for the purpose of the ETAI.
ETAI will provide a standard form of the agreement between the FPS and the FPA that provides verification.
In order to meet the requirement for a URL that can remain valid over a long period of time, it is recommended to use a "cover page" system, where for each article there is a short HTML page that contains links to the full file for the article and the basic bibliographic information. The URL of the cover page is used as an identifier for the article. In other words, there is one level of indirection when you follow a link to the article. By this scheme, it is sufficient that the cover page retains the same URL over time, but the physical location of the article itself may change.
For example, if the copyright agreement with the publisher of the conference proceedings should only allow the article to be maintained at the server of the author's current employer, after he or she has moved to another employer, then the cover page can easily be modified to point to the new location of the article. In this way, the URL serving as the key to finding the article retains its validity.
From the point of view of participating universities or research institutes, the scheme described here can serve as a convenient entry point: while waiting for the administrative measures which are needed in order to qualify as a First Publication Archive, the department or institute in question can get started by making a First Publication Server agreement.
In most cases, it will be quite easy to make the arrangements required for being a First Publication Server. Most major computer science departments (or equivalent) already have an organized way of putting their preprints online. Basically, what is needed in addition is to set up a cover page system, and to do the paperwork of the verification agreement.
In particular, in order to be able to comply with the copyright consistency requirement, the participating department may consider setting up a register of the copyright releases that are signed by its employees, if it does not have such a register already.
An FPS arrangement should be seen as a provisional one, which normally is only to be used for a limited time, after which the FPS is upgraded to FPA status. This has several advantages for them, in particular, that articles can be published without having to wait for acceptance at some conference, and that an earlier date of appearance is obtained.
In addition, it may be appropriate to have FPS's on a more or less permanent basis in small universities, or universities with little electronic publishing activity, so that it is impossible to set up a full FPA there.
The arrangements described above are quite straight-forward, and should work without problems in all normal cases. The following are some observations for the unlikely case that someone tries to challenge the arrangement on formal grounds.
It has to be verified that the conference's copyright agreement does not preclude obtaining the electronic and paper copies for archive purposes. What we have here certainly falls under "academic purposes", but it does not qualify as "personal use". If a fee has to be paid for the release of those copies, it has to be paid by the author or the FPS; if even that is not sufficient, then the ETAI system will not be able to receive the article.
We assume that the creation of a link to an electronic document counts as "mention" rather than "inclusion", so that it is not an infringement into the copyright of that document. We take that as completely obvious, but since the reverse position has been taken in at least one instance, it must be declared that if a university or institute would ever attempt to impose restrictions on links to their on-line publications, then we would not merely abstain from making links to the restricted publications, we would also avoid any mention at all of any publication from that university.