Reidar Conradi, IDI, NTNU, Trondheim,
17 Dec. 2003
To: alle-idi@idi.ntnu.no  
CC: einar.aas@fysel.ntnu.no, Tore.R.Jorgensen@ime.ntnu.no,
    Solfrid.Bergsmyr@ime.ntnu.no
Subject: Oppfordring til å registrere seg i FORSKDOK-databasen osv.

Hi, dear IDI folks!

The end of the year is approaching. Since a considerable part of our university budget gradually will come from documented research publications, it is important that each of us document our own research production in FORSKDOK. This also provides a convenient way of documenting all of IDI's research in one place.
See also major problems with incomplete PERSON register in S7 below.

Mini-guide to FORSKDOK, NTNU's official system for registration of scientific publications

Table of contents

S1. General access to FORSKDOK.
S2. Registration of user and password in FORSKDOK.
S2a. The first sub-step is to be registered as a valid user.
S2b. After being a registered user, request a password.
S3. Registration of your publications etc., possibly in several sessions.
S4. General searches.
S5. Comments on problems and usability.
S6. User support.
S7. Problems in the PERSON register for IDI's 22 full professors.
S8. Appendix: proposal for a hybrid registration system.

Try the steps S1-S4 below.

S1. General access to FORSKDOK.

FORSKDOK is available through the BIBSYS main page: http://www.bibsys.no.

Note, that the portal at The University in Oslo seems much better for normal FORSKDOK search access; try http://uio-forsk.uio.no/.

1. To register new/changed information, there are three identical ways to start:

1a. Click at the Registrer i ForskDok button to the right (this button does not exist in the English version!).
1b. Look under the Andre tjenester top menu and click on the ForskDok registrering item.
1c. Go directly to: http://www.bibsys.no/forskdok/innloggreg.htm.

In either case, you will now be asked to login, by giving your username (= email-address) and a 6-char passweord (e.g. "conradi@idi.ntnu.no" and "mypswd"), see in S2 below how to get these.
You will then have access to the FORSKDOK main page: either http://wgate.bibsys.no/forskdok/servlets/registrering? or http://wgate.bibsys.no/forskdok/servlets/registrering#forskdok.

To search for information, there are two services, numbered 2 and 3:

2. Interactive searches:
From the BIBSYS main page, look under the Søketjenester top menu and click on the ForskDok søk item to search for info interactively.
Without username/password clearance you are now presented with a web page called: http://wgate.bibsys.no/search/pub?base=FORSKPU2.
You can now search either for publications (FORSKPUB) or projects (FORSKPRO), see S4 below.
In the English version: from the BIBSYS main page, click on "Bibsys at the turn of the century - the road to information", and click on the FORSKDOK entry in the general page that comes up.

3. Batchwise searches:
Or: from the BIBSYS main page, look under the Andre tjenester top menu and click on the ForskDok lister/statistikk item.
This service is seldom used, but many query/formatting options exist. It is a batch service, as the requested list is generated over night, and you "pick it up" the next day.

Most of the FORSKDOK web pages are available also in English. Please click on the in English button on the main BIBSYS page, or on an English flag button on many local pages.

The BibSys/FORSKDOK responsible person at IDI is ellen.solberg@idi.ntnu.no at phone 73.593440.

S2. Registration of user and password in FORSKDOK.

You cannot use your normal NTNU userid/password, sorry.

S2a. The first sub-step is to be registered as a valid user in the FORSKDOK database.

To achieve this, send the following email:

To: bibdrift@bibsys.no
cc: 
Subject: Registration of a new, valid FORSKDOK user
  Start explanation -- this text is not part of actual email:
    See below format with tagged fields, starting with "Fornavn:".
    This format must be exactly as below, due to automatic scanning.
    Use one text line per tag (e.g. "Fornavn:") and
    its contents (e.g. "Reidar"), with no blanks in the end of a line.
    Only the contents of Fornavn, Etternavn and E-Mail is specific for you;
    the other fields can be reused "as-is" from the below set-up.
  End explanation -- this text is not part of actual email.
Fornavn:
Reidar
Etternavn:
Conradi
Institusjon:
Norges teknisk-naturvitenskapelige universitet
Avdeling/Institutt:
Institutt for datateknikk og informasjonsvitenskap
Postnr:
7491
Poststed:
Trondheim
E-mail:
conradi@idi.ntnu.no

You will then get a response to the provided email address. It tells you that you now are registered user (username = your provided email address) in the FORSKDOK database. It also instructs you how to ask for a FORSKDOK password (see S2b below).

S2b. After being a registered user, request a password.

Go to the page: http://www.bibsys.no/forskdok/innloggreg.htm and click on the last her in command "Passord bestiller du her". A new web page comes up, and you should enter your email address (e.g. "conradi@idi.ntnu.no") and click on the Sjekk elpostadresse button just below. Another web page comes up, and click on the Bestill passord button.

Your FORSKDOK password will now be sent to your above email address in a few seconds.

S3. Registration of your publications etc., possibly in several sessions.

Go first to: http://www.bibsys.no/forskdok/innloggreg.htm (same as S1-point1 and S2b), and login as indicated above.
Your user name is your full email address as given above, and your password is the one provided by FORSKDOK in step S2b. As in S1 under point1, you will now have the web page: either http://wgate.bibsys.no/forskdok/servlets/registrering? or http://wgate.bibsys.no/forskdok/servlets/registrering#forskdok as your "base".

Then start registration of your research publications: use button for Registrer ny publikasjon, or possibly first search for your recent publications by Søk i FORSKPUB (publikasjoner) button to get the FORSKPUB submenu/subpage.

You can also query any other information in the database, e.g. registered persons (button Søke/oppdatere i FORSKDOKs personregister) or projects (button Søk i FORSKPRO (prosjekter) to get FORSKPRO submenu/subpage), see S4 below. The "Institusjons-ID" of IDI (at the IME-faculty) is 03140100 - but is only needed to constrain searches, see again S4 below.

You can at any time correct the stored record ("post" in Norwegian) for a given publication or project. First, search for the relevant record and click to have it displayed (needed if there are more that one search result). Then click on the Klikk her button at the bottom of the displayed page to obtain a field-by-field list of the actual record (this page is the same as when you first registered the publication or project). Now perform the neccessary updates, and finally click on, respectively, the Lagre publikasjon or Lagre prosjekt button to store your changes. This button is both at the top and bottom of the shown field-by-field page.
Digression: For updates, this button should have been marked Oppdater ..., not Lagre ...!

All co-authors can normally change such records (?).

Some hints:

Hint 0: Have publication meta-data readily available before FORSKDOK start-up.

Such meta-data typically deals with author(s), title, conference, publisher, year, volume/number, page numbers etc. Please check that the actual (co-)authors or projects already have been registered -- see below. Also provide the URL-address, where the full contents of the publication is available.

Problem: in the author list, you can only delete an author (anywhere in the list) or append an author (in the end of the list). So if the author list gets incorrect, e.g. stands in the wrong sequence, you may at worst have to delete all authors, and build up the entire author list from scratch. It is anyhow a small job, as authors are easily searchable by their (last) names. But this procedure seems unnecessary, as most other information is fairly easy to change.

Hint 1: Have also the electronic text of the actual publication available.

Much text (title, abstract, keywords etc.) can then be copy-pasted from this and save you from tedious re-typing. If you only have a .pdf-file, let Adobe Acrobat generate a .rtf-version, that you can read with Word and from which you can copy-paste relevant text fragments. If only a .ps-file is available, use Acrobat Destiller to first generate a (hopefully readable) .pdf-file, or possibly use tools like GhostScript or NotePad to generate a plain ASCII-file. At worst, "cannibalizing" a .ps file is also possible, as we deal with small amounts of text.

Hint 2: Finding correct paper category.

When you register a paper, the first question is the actual paper category and a menu for filling in this info (- after this, you get a long list of fields to fill in).
The most common paper categories are (just look up in the menus):

- Conference paper:
    In the menu you then click on item number four: Konferansebidrag eller faglig foredrag.
    In the submenu you then also click on item number four: Foredrag ved int.vit. konferanse med publisert artikkel.

- Journal paper:
    In the menu you then click on item number one: Faglig artikkel, kapittel, kronikk, anmeldelse eller ....
    In the submenu you then also click on item number one: Artikkel i internasjonalt vitenskapelig tidsskrift med referee.

- Book chapter:
    In the menu you then click on item number one: Faglig artikkel, kapittel, kronikk, anmeldelse eller ....
    In the submenu you then click on item number four: Kapittel i faglig bok utgitt på forlag.

- (Edited) book:
    In the menu you then click on item number two: Faglig bok utgitt på forlag.
    In the submenu you then click on item number three: Faglig bok utgitt på forlag. (same item text!)

- PhD thesis:
    In the menu you then click on item number three: Faglig publikasjon utgitt av institusjon, forening o.a..
    In the submenu you then click on item number two: Doktoravhandling.

- Poster without accompanying position paper, abstract etc.:
    In the menu you then click on item number four: Konferansebidrag eller faglig foredrag.
    In the submenu you then click on item number nine: Poster ved int. vit. konferanse uten publisering.

- Popular article:
    In the menu you then click on item number one: Faglig artikkel, kapittel, kronikk, anmeldelse eller ....
    In the submenu you then click on item number 13: Faglig artikkel eller kronikk i dags- eller ukepresse.

Hint 3: Registration of new persons (potential co-authors) in the PERSON register.

If you have non-registered co-authors, they must first be registered in a special PERSON register. You can do this directly through a special menu, luckily available as part of the publication registration web page. All persons must be registered by last name, first name, gender, position, and academic degree. In addition comes the person's affiliation ("institusjon"). Here, local co-authors must be registered through a system of menus, so click for IME first and IDI next. External co-authors instead require one line of free text, e.g. "Universita di Pisa, Pisa, Italia".

If you later want to correct some PERSON data, you must have the FORSKDOK "contact person" at NTNU to do this (S6 below).

Hint 4: Referring to projects as part of publication registration.

Please first search for the name(s) of the actual project(s) (via FORSKPRO submenu) and "remember" the(ir) project-id(s), before you start registration of a new publication. Then these project(s) can be referred to by their project-id(s) -- to be typed in fully by you -- in the end of the registration process.

NB: it is not possible to register new projects or even search for such, during registration of a publication.
The easiest way is then to complete the registration of the actual publication, and either note its cryptic publication-ID or just search for it later by normal queries. Then provide the corresponding project-ID (old or new) by separate commands (in FORSKPRO submenu). Lastly, go back and correct the publication record. This round-about is actually not much extra work.

Remedy: Before you start registration of a paper, click on a search command that will start a new and parallel web browser application. This allows easy switch between entry and search mode during the registration process, and you can "cut-paste" the project-ID for use in the publication record.

As mentioned, you can at any time correct or supplement the stored publication record.

Hint 5: Registration of a new project via FORSKPRO submenu.

As said in Hint 4 above, it is advisable to register your projects, before an associated publication is registered. Again, have relevant electronic texts available for copy-paste. Most researchers do not have many new projects per year, so this is not much work.
Upon registration of a project, let its name/title be written like "BUCS, BUsiness-Critical Software". It is then easy to search for a project by just typing "BUCS" in the "Tittel" field (NB: not in the "Navn" field).
As mentioned, you can at any time correct or supplement the stored project record.

Summing up:
You will use 15-20 minutes to register a new publication, and the same for a project.
Registering a new person takes about one minute.

S4. General searches.

Start as above in S1.
NB1: Below apostrophs (") are never written in the query fields.

NB2: When you initiate a search command for projects or publications, as described below, a new and parallel web browser application is started and will normally be reused in later search commands.

To get all publications for IDI, there is a slight complication, as IDI is both the old "FIM"-IDI (Institusjons-ID 03170200) and the new "IME"-IDI (Institusjons-ID 03140100). And we cannot rewrite history. All IDI persons should therefore ensure that their PERSON records from now (also) are associated with the new "IME"-IDI. Many IDI persons have therefore two or more "identities", but this poses small problems.

When you search for a person (first click on the Søke/Oppdatere i FORSKDOKs personregister button in the FORSKDOK main page), often the last name "conradi" (meaning "conradi,?") in the single "Navn" field is enough. You can possibly expand to "conradi,reidar" or just "conradi,r?".

When you search for a project in submenu/page FORSKPRO (first click on the Søk i FORSKPRO (prosjekter) button in the FORSKDOK main page), just use the project acronym (NB in the "Tittel" field), e.g. "BUCS".

When you search for a publication in submenu/page FORSKPUB (first click on the Søk i FORSKPUB (publikasjoner) button in the FORSKDOK main page), v use "Forfatter" = "conradi,r?" or "conradi,reidar" to get all of Reidar Conradi's publications (NB: there is another "conradi", namely Hallgerd Conradi at Høgskolen i Volda, whose first name does not start with "r"). Note, that an author name either must be complete, such as "conradi,reidar" or open-ended such as "conradi,r?" - just "conradi" won't do.
To combine several authors in searches, you must use normal parentheses and "*" (AND-operator) to connect the alternatives. An example is "Forfatter" = "(conradi,r? * jaccheri,?)" to get these two author's joint publications. Here, "jaccheri" is a unique last name, but "conradi" also must here have an appended ",r?" or ",reidar".
Use "Institusjons-ID" = "03140100" to get the new IDI's registered publications. To get all publications from both the old and new IDI, use "Fritekst" = "(03170200 + 03140100)/pi" (not "Institusjons-ID" = "..."!). For year 2003 this query gives back 72 records per 1 Dec. 2003. Regrettably, there is no easy way to get even older publications from IDT (Institutt for datateknikk og telematikk in 1987-96 at the previous NTH) out as a group. Use "År" = "2003" to get the publications for that year, or "År" = "2002-2003" to specify these two years.
Use "Andre" = "a11" ("a-one-one") and also (NB) click on the Publ.kategori item in the Publikasjons-ID menu to the right, in order to get e.g. only journal publications.
Etc. etc.

As shown, you may add blanks around "*" (AND-operator) and "+" (OR-operator).
In general: click on the Veiledning button to get the full range of search options.

S5. Comments on problems and usability.

       General comments:
  1. Make a better over-all system, where each department/research group (dynamically) could generate their special views on this, e.g. using a PHP-script. Now, there is a (partially unusable) central FORSKDOK register, local publication lists at the department/group level, and private publication lists in the CVs/home pages of most employees -- i.e. 2-3 variants of the same information.
  2. Or just drop FORSKDOK entirely and either 1) rely on local publication lists on the web, and/or 2) use public citation indices or other search machines to get general surveys of NTNU's total publications, e.g. try http://citeseer.nj.nec.com/ or http://dblp.uni-trier.de/ for informatics. Especially if the national UFD department only wants a few key figures from NTNU, all FORSKDOK registrations seems unnecessary.
  3. The two-stage approach to get user and password is a nuisance, and should be merged into one step.
    Rather: all NTNU employees should be automatically registered in FORSKDOK at the start of their employment, i.e. zero steps!
    One easily looses 2/3 of all intended users by such obstacles. As shown in S7, only 4 out of 22 IDI professors (less than 20%) stand with correct and updated information in the PERSON register.
    Such erroneously registered PERSON data prevents generation a complete picture of e.g. IDI's publications for a given year, cf. comment 21.
  4. To copy-paste between fields, you must first "copy", then "paste" - you cannot just "move" by drawing the cursor. There is also no "undo" functionality.
  5. The buttons are sometimes verbose and partly confusing, e.g. there is no need to have a separate Velg button before an explanatory item text.
  6. No easy way to close a single field, during fill-in. The alternatives are Åpne (open) and Slett (delete) - not an individual "Lukk" (close) command, but a group "Lukk" command for all fields.
    Also some annoying "flickering" when fields are opened and closed.
  7. Sometimes confusing to go back one menu level: use the Back button in the web browser, or the Tilbake button in most FORSKDOK pages. E.g. to get back to the FORSKDOK main page from FORSKPUB or FORSKPRO searches, you must manually click on the previous web page or kill the current web-browser incarnation.
  8. "ForskDok" stands intermixingly also with capital letters ("FORSKDOK"), e.g. in menus.
  9. The Registrer i ForskDok button on the main BIBSYS page does not exist in the English version, rather a more general button.
  10. For wrong-spelled queries: as mentioned, there are sometimes strange error messages, see also point 31 below.
  11. Availability should be improved: at least two cases of partial service denial has been experienced in three weeks (20 Nov. - 10 Dec. 2003), a period characterized with much use.
  12. Problems in correcting records: I was allowed to correct an old publication record put in by a co-author, but after some changes, I was disallowed to make further changes.
    In general, you cannot correct records entered by other persons e.g. by co-authors (?).
  13. All in all: too much wrong and incomplete information is registered (see S7 below), plus lots of missing registrations.
    No clear or active procedures for Quality Assurance of FORSKDOK data exist. The responsibility lies officially at NTNU centrally, i.e. at the BibSys group.
  14. No comprehensive user guide exists on how to use FORSKDOK, e.g. to explain many of the issues commented upon here. The English version is especially incomplete.

    Person/author data:

  15. There are at least three "person" registers:
    - local user/password registers,
    - central FORSKDOK's user/password register, and
    - central PERSON register inside FORSKDOK being used for publication authors.
  16. At least one co-author must (via the PERSON register) be affiliated with one of the ca. 60 "official" FORSKDOK user institutions -- i.e. NTNU, UiO, UiB, HIST etc.. So just an Italian and a Chinese co-author will not do.
    Hint: Perhaps one of the foreign co-authors also has had a temporary affiliation with NTNU?
  17. Cannot easily reshuffle/correct the author list of a publication.
  18. Anybody can register a new person in the PERSON register, but only the FORSKDOK "contact person" at NTNU can change such person data (see S6 below).
  19. Many persons seems to be missing in the PERSON register, but are yet found by "Forfatter" = "...", e.g. try last names "Blake" and "Bratbergsengen" at IDI!
    Reason: The PERSON register was first introduced in 2001. Many papers are implicitly registered on the institution of the person that registered the publication in the first place.
  20. No check when registering a new (NTNU) person, whether the affiliation ("institusjon") is supplied by the formalized faculty/department menus or by arbitrary text.
    Similarly, there is no check that gender data (male/female) has been filled in.
    Publication records before 2001 seems to be registered at the "home" institution of the person that creates the record. This means that the institutions of co-authors may not be associated with that record.
  21. Since the author affiliation often is missing, outdated or erroneous, the generated publication list for a given institution may be grossly incomplete.

    Registration of and search for publications:

  22. Cannot easily register/search projects during registration of a publication.
  23. Cannot easily go back and correct information from the 1990's, e.g. to insert publication year which was not compulsory before.
  24. When categorizing a new publication, the main category's item selection is not displayed when selecting a sub-item.
    Suggestion: Have the ten most common categories as direct choices.
  25. There is no easy way to click on a recorded URL address to get the contents of that publication.
  26. The last name (e.g. "conradi") suffices when looking up in the PERSON register, but has to be expanded to "conradi,?" when searching for publications.
  27. Most search alternatives (with rather cryptic codes) are not available as pull-down menus, but stand in a separate Veiledning guide (- NB: in Norwegian!).
  28. When searching for projects, you must give the project name in the "Tittel" (not "Navn" field) -- not so obvious.
  29. Composite queries are allowed in the "Forfatter" field, but not in the "Institusjons-ID" field (see point 31 below).

    Reorganized institutions - Old IDI (at FIM-faculty) vs. new IDI (at IME-faculty):

  30. Be observant on old and new IDI institution codes:
    E.g. when searching for publications with "Forfatter" = "conradi,r?" and the old "Institusjons-ID" = "03170200", then one shared publication with Guttorm Sindre et al. is found, because Guttorm Sindre is (only) affiliated at the old IDI.
    Repeating the above query now with IDI's new Institusjons-ID" = "03140100", gives many more publications - including the above publication - since Reidar Conradi happens to be affiliated with the new IDI in this publication record. (Reidar Conradi is registered as "two" persons, one for both the old and new IDI!)
  31. Use "Fritekst" = "(03170200 + 03140100)/pi" to search for publications from both the old and new IDI, i.e. not "Institusjons-ID" = "...". In fact, the latter gives a dubious error message "Systemfeil ..."!!
    However, "Fritekst" = "(03170200)/pi" with only one search item does work, so you may choose to use this field all the time.
Advice: Have patience.

FYI: A new FRIDA system will be taken in use from mid-2004, being a cooperation project with the other Norwegian universities, and developed by the University of Oslo. All old Forskdok data will be automatically transferred to the new system. Oyvin.sather@adm.ntnu.no, sits as the NTNU representative in the project group, lead by Tina Lingjærde, USIT, g.c.lingjarde@usit.uio.no. This system will be based on import of external, quality-assured data e.g. from Science Citation Index ( http://www.isinet.com/products/citation/sci/). See also comments in S8 on how a new system should be.

S6. User support.

Try:
BIBSYS brukerstøtte attn/ Hege Johannesen,
BIBSYS - Library Automation System, bibdrift@bibsys.no, http://www.bibsys.no, N-7491 Trondheim.
Tlf.: +47 73.592096/97, Fax: +47 73.596848

Also: Arild.Sontvedt@ub.ntnu.no (main FORSKDOK responsible).

Good luck! - reidar

S7. Problems in the PERSON register for IDI's 22 full professors.

Note: No checks for førsteamanuensis, PhD students, or other scientific personnel has been done.

4 cases of correct PERSON data has been observed per 30 Nov. 2003:

9 cases of partially correct PERSON data has been observed: 2 cases of wrong PERSON data has been observed: 7 cases of missing PERSON data has been observed: However, several of the non-registered persons do have registered publications with their own names as (co-)authors! This is because the PERSON data register first was put into operation in 2001.

S8. Appendix: proposal for a hybrid registration system.

Main idea: take the FORSKDOK/FRIDA information from existing publication lists, thus improving the format and quality of the latter.

1. Rationale:
Most researchers meticulously maintain their own publication lists on the web, usually available through their home pages. These lists can be part of their CVs or part of aggregated publication lists of their research groups or local departments ("institutter").

This information is usually of high quality, because each researcher through this can document and possibly advance his/her own scientific career -- see e.g. the publication lists of my own research group http://www.idi.ntnu.no/grupper/su/publ/INT-PUBL.php3.

I cannot see that a central FORSKDOK/FRIDA system (called a FR-system) easily can replace these local lists (web-pages). Both are needed for different purposes, and we therefore face double work. In all library systems, a major cost is information entry. Given the circumstances documented in S5 and S7, we also have very poor data quality, and the motivation of the researchers is alarmingly low. So, please work with the reseachers, not against them.

My proposal is therefore to pre-structure the local publications lists (mostly .html files) a bit according to some central template, make a hyperlink-aggregations of these on the department/faculty etc. level, and let a dedicated web-crawler parse these lists regularly. The crawler could mostly do counting and syntax checking, and be run at any organizational level at any time. There must also be a department-level quality control of the lists, e.g. of use of classification categories. In addition, at a certain date (15 Jan.?) all this information is put into the FR-system automatically - possibly requiring an overnight run.

2. Preliminary proposal for set-up of local publication list:

Example of parts of a standardized publication list, e.g. in .php3 or .html format:
-- using [] for ".html-parentheses"

  ...
  [H3]Publications for SU-group, IDI-department, IME-faculty, NTNU in 2003[/H3]
  
  [ol]
  [li][b]Marco Torchiano and Giorgio Bruno:[/b]
      "Enterprise Modeling by Means of UML Instance Models",
      In [I]ACM SIGSOFT Software Engineering Notes[/i] 28(2):12-12,
      March 2003, as a summary (6 p. full version).
  [!--Project: INCO--]
  [!--Keywords: Enterprise modeling, UML, software engineering.--]
  [!--File: ...--]
  [!--Type: JOURNAL-REFEREED--]
  [BR]
  [a href="http://www.idi.ntnu.no/grupper/su/publ/marco/SEN28-2-2003-UMLinstances.pdf">
  download paper (.pdf -- 63 kB)[/a]
  [a href="http://www.idi.ntnu.no/grupper/su/publ/marco/SEN28-2-2003-abstract.html">
  download abstract (html -- 0.5 kB)[/a]
  [/li]
  ...
  [/ol]
  ...
The hidden comment fields ("[!-- ... ---"]) can of course be elaborated more, but a full-fledged XML-format seems too much. In spite of many cryptic details, most researchers can manage to annually maintain such web-entries for their own publications. The format is mostly self-explanatory, and simplified wrt. details of publication place and time. The researchers also "own" their local publication lists, and can reuse the exact same information and partly its format for traditional purposes, such as writing research applications, being visible on the web etc. Thus trustful participation and local quality procedures (no mostly absent) can be established.

3. Search engines:
Standard web search engines (Google-type) should be used, either to search in parts of such publication lists, in the FR-database, or connected to existing search engines and databases belonging to the specific discipline, cf. the two above for informatics. So, please ensure that the FR-system gives some extra functionality to the "rank-and-file" also on this point. Now it appears to be a "black hole".

This file: http://www.idi.ntnu.no/grupper/su/publ/forskdok.html
Last changed 17 Dec. 2003 by Reidar Conradi