How many conferences are there? #241
Replies: 5 comments 1 reply
-
|
Hi Wolfgang, knowing/estimating the ratio between proceedings and digital events is great, however as a number, it is limited to what it represents. It would be interesting to see the distribution. I am assuming (until debunked) that there is a long tail of event which took place only once. Having a distribution allows us also to see how many conferences are successful (i.e. they run/are running over a sustained period of time). What do you think? |
Beta Was this translation helpful? Give feedback.
-
|
@angelosalatino my ratio interpretation was a bit different then yours - i'll get back to this futher down. As for your assumed long tail distribution this is the result for wikidata scientific papers with "Proceedings of" in the name and the CEUR-WS volumes - in total 8471 events are covered by this statistic: For these two proceedings title parser sources I am already extracting the "ordinal" information. Most other sources do not have that attribute as an explicit field. Still your long tail assumption is correct and most events are 1st or 2nd events while only a few event series make it to ten ore more installments. My own ratio intepretation was about how many volumes a proceedings publication may have. In the GND / TIBKAT dataset each volume is considered to be its own "Tagungsband". My favorite query on TIBKAT shows a diagram |
Beta Was this translation helpful? Give feedback.
-
|
So i assume your interpretation of "How many conferences are there" is "How many scientific event series are there?". In may I did a delta comparison of how many new events and series showed up in the ptp sources in the last few month: The highest number of series is 9564 in WikiCfP with an increase of 13 series per week. All sources overlap and i didn't consolidate yet by doing proper entity matching yet. Note that the big increase in wikidata series entries is due to @Simon-Cobb 's work on #127 |
Beta Was this translation helpful? Give feedback.
-
|
@WolfgangFahl , I was wondering if you could also easily generate ordinal distribution from Confref data (where conference number in the series is stored in a dedicated field, so it could be considered gold standard) or from the DBLP (who would have bigger coverage than confref, i.e. would index first events published in CEUR-WS which later came to Springer).
|
Beta Was this translation helpful? Give feedback.
-
|
@WolfgangFahl , I read the draft with interest. For the Gold standard, I'd recommend mentioning DBLP, which was de-facto standard since 1990s and also the LOD pilot from Springer - as it was nicely received by the community. I can dig up more facts (and put them on Arxiv since we never formally published the extended version). The preliminary version is at http://ceur-ws.org/Vol-1155/paper-01.pdf and an evolved (but short) at http://ceur-ws.org/Vol-1963/paper566.pdf |
Beta Was this translation helpful? Give feedback.
-
|
Also, we worked on the dataset of peer review descriptions to make the peer review process more transparent - see https://www.slideshare.net/birukou/creating-a-dataset-of-peer-review-in-computer-science-conferences-published-by-springer, also unpublished |
Beta Was this translation helpful? Give feedback.
-
http://ptp.bitplan.com/settings shows records from a few sources that might overlap.
There are at least 86000 WikiCFP call for papers in the meantime. The TIBKAT / GND data contains at least 630.000 proceedings - depending on the proceedings / event ratio i'd estimate that there are some 500.000 scientific events with digital traces. Getting a more precise answer to this question is part of my research.
Beta Was this translation helpful? Give feedback.
All reactions