Dieser confluence Space wird nur dazu genutzt, um News bezüglich dblp in Form von Blogeinträge zu verfassen. Diese Blogeinträge werden täglich exportiert und auf der Startseite eingebunden.
With the latest release of the dblp XML data dump, we made some extensions to the document type definition (DTD) of the XML format. The goal of these changes is to lay the groundwork for a number of planned extensions of the dblp data model that will be realized at a later date.
These changes will only affect you if you use:
- the daily XML data dump, available at http://dblp.org/xml/,
- any future snapshot releases of the XML dump, available at http://dblp.org/xml/release/, or
- the live XML web API, e.g. http://dblp.org/rec/xml/journals/pvldb/Ley09.xml
Regular users of the web site or other export formats – like the BibTeX export – are not affected.
We encourage you to update any local copy of the
dblp.dtd file on your system and adjust data import code accordingly. The new DTD should be compatible with all earlier dblp XML dumps. First changes to the actual dblp data records that make use of these changes are expected to be made not before mid of October.
As always, the latest version of the DTD can be found at:
All upcoming as well as earlier versions of the DTD file will be preserved along with the persistent snapshot releases at:
A brief summary of the recent changes is available at:
Thanks to the search API provided by our friends at the Allen Institute for Artificial Intelligence (AI2), we have added a new full-text search option to dblp. By choosing "Semantic Scholar search" as your default search action in the drop down menu next to the search bar, searching will now retrieve documents from dblp based on matches within the full texts as indexed by the Semantic Scholar service. Since dblp usually does not have access to the full texts of research articles, this is something that has not been possible before. Please try it for yourself.
The new search option is currently an experimental feature (which may still break once in a while) and we will be happy to hear your thoughts and comments about it.
In the past weeks, we imported bibliographic metadata for more than 9.000 PhD theses from the French archive Hyper Articles en Ligne (HAL) into dblp. These theses cover a significant part of the French computer science community, with some theses reaching back as far as the 1950s. The HAL archive is the second source of PhD metadata to be continuously imported into dblp in addition to the more than 10.000 data records already added from the Deutsche Nationalbibliothek (DNB), the German National Library. In a next step, we plan to include data from the EThOS Repository of the British Library.
We understand the PhD thesis as a very important and central publication in the professional life of a computer scientist which should have its place in a computer science bibliography. Hence, we aim to further increase the coverage of PhD theses in dblp. However, open bibliographic metadata of PhD theses is often surprisingly difficult to obtain. For many countries we are still unaware of any central repository collecting and aggregating information on national PhD theses. In some countries, bibliographic metadata of PhD theses even seems to be hidden behind a paywall.
If you are aware of an open metadata repositories for CS theses in your country or international level, we will be grateful for your advise. Please feel free to contact us.
For almost ten years now, the CompleteSearch interface has provided fast and convenient access to all of the metadata collected by dblp. Features like faceted search and fast search as you type made it a valuable asset to the computer science community. Since 2011, the service has been available under the easily memorizable domain dblp.org. With today's update of the dblp web system, the final stage of the integration of the CompleteSearch capabilities and its convenient search interface has concluded, and CompleteSearch is now an integral part of the dblp web system. At the same time, the domain dblp.org will now point directly to the dblp main site. This domain will play a more prominent role in the future URL and ID schemes of dblp.
The CompleteSearch engine and search interface has been developed and maintained by Hannah Bast (formerly at the Max Planck Institute for Informatics in Saarbrücken, now at the University of Freiburg, Germany). Although operated independently from the main dblp web site, the development of the search interface has always been in consultation with the dblp team. The layout of both web sites had been unified, and a preview of CompleteSearch's facets had been part of the classic dblp author pages. The interface provided a lot of search features that the rather basic search of the dblp site in Trier had been lacking (like an extremely fast and responsive search interface, searching by facets, search as you type, Boolean operators for complex queries, and more), and quickly became the community's first choice when searching publications within dblp.
With the overhaul of the dblp web system (which started about two years ago), the features of the old CompleteSearch interface have been integrated step by step into the dblp web systems hosted in Trier and Dagstuhl. This has been made possible by local back-end instances of the CompleteSearch engine that are now powering all of the search, filter, and browsing capabilities of the dblp web pages. With today's update, all major features of the interface have been integrated, as well as the former URL.
This does – of course – not mean that the search interface won't see further improvements and updates in the future. After all, dblp is a project that is continuously adapting and evolving. The improvement of the search engine's front end and back end will remain a collaboration between the dblp team in Trier and Hannah Bast's research group in Freiburg.
At the same time, today's update also means that the "modern-style" dblp web page layout is now established as standard for all aspects of the dblp web site, and that the old "classic-style" layout which is currently still available at some of the dblp pages will fade out with future updates.
For dblp, it is of utmost importance that every update improves the user experience and the utility of dblp for you (the computer scientists) in your daily work. Although any major update runs the risk of breaking existing features or disrupting established work habits, please be assured that this is never made intentionally. Hence, if you feel that an important feature of the old interface is missing, or if you encounter any problems when using the new search pages, please contact us under email@example.com.
Today, dblp reached the wonderful "Schnapszahl" of 3,333,333 publications.
With the next upcoming XML dump, you might notice that we added some additional attributes to a (very small) number of dblp records. These changes have become necessary to realize extensions of the dblp data model in our back-end. Most of our users won't notice any difference at all, but if you are working on our XML raw data dump, you may need to update any local copy of the dblp.dtd file in order to validate future versions of the dblp.xml file. This is also true for all future stable releases of the dblp data set. The new DTD is compatible with all earlier XML dumps.
We are happy to announce that on Thursday, June 18, 2015, the dblp computer science bibliography indexed its 3-millionth publication. Thanks to the joint effort of the University of Trier and Schloss Dagstuhl, as well as the commitment and help of our dedicated users and data partners, dblp has grown to be the world's most comprehensive open bibliographic data service in computer science. And the service is still growing strong: In each of the past three years, more than 325.000 new publications have been added to the database.
Every month, more than 450.000 distinct users visit the dblp web pages. On average, five dblp pages are viewed per second, and every three seconds a new user session is started. The dblp team understands this ever-growing demand as an incentive to keep consolidating and improving our service.
dblp was built for the computer science research community. It is your tool, and we want it to meet your specific needs, e.g., when you are searching for articles, looking for bibliographic data, or browsing authors and publication venues of computer science. So if you have any ideas or suggestions about how we can improve the utility of dblp for your daily work, please let us know!
At Schloss Dagstuhl – Leibniz Center for Informatics seeks to hire a Research Scientist (Computer Science or Mathematics). Position start date: July 1, 2015.
Job description: The dblp computer science bibliography, jointly operated by Schloss Dagstuhl and the University of Trier, is the world’s most comprehensive open data collection on bibliographic meta data in computer science, hosting about three million publication records. In the research project Scalable Author Disambiguation for Bibliographic Databases, the Heidelberg Institute for Theoretical Studies (HITS), FIZ Karlsruhe – Leibniz Institute for Information Infrastructure, and Schloss Dagstuhl have joined forces to develop algorithmic methods for determining the unambiguous attribution of research articles to their authors in scholarly databases such as dblp or zbMATH. The person hired for this position will be responsible for developing and implementing such methods in the live production environment of dblp, and participating in the ongoing technological and content-related development of the dblp service. Job duties will be performed in close cooperation with the existing dblp team of Schloss Dagstuhl and our project partners.
For more information, please visit the website of Schloss Dagstuhl.
With great sadness we must report the death of Klaus Tschira, a great patron and sponsor of science and strong supporter of dblp and Schloss Dagstuhl.
The dblp data set (found at dblp.dagstuhl.de/xml/) is studied or used as a benchmark in many scientific publications. However, due to the daily updates to dblp, it can be sometimes difficult to reproduce experimental results made on earlier versions. To remedy this problem, starting from today, will provide and archive stable monthly snapshot releases of the dblp data set. We encourage you to use these snapshot releases for your experiments and to cite their persistent URLs in published articles to. You can find the archived dblp data snapshots at dblp.dagstuhl.de/xml/release/.
Since the last weekend, the dblp computer science bibliography lists more than 2.8 million publications . See also more statistics about dblp.
We often receive questions like "How many users are using dblp?" or "How many hits do you get per day?" While we do not have a live statistic for these questions like the one we have for our data stock (yet) we can of course share the usage statistics we learn from our web server's log files.
There are three "official" dblp servers which are updated and synchronized on a daily basis:
- server Trier 1: http://informatik.uni-trier.de/~ley/db
- server Trier 2: http://dblp.uni-trier.de/
- server Dagstuhl: http://dblp.dagstuhl.de/
These servers do show a very different rate of usage, with Trier 1 being the by far most widely known URL. This is of course due to the fact that server Trier 1 is ranked so highly by Google. Server Trier 1 is currently still hosting the old dblp web system, while servers Trier 2 and Dagstuhl are already hosting a beta version of the brand new web system. (A full update to the new web system is scheduled for early 2015.)
The following figures have been collected during October 2014. These figures ignore the traffic caused by known bots and crawlers.
Average number of user sessions (visits) per day on server ...
- Trier 1: 27,194 sessions
- Trier 2: 4,653 sessions
- Dagstuhl: 602 sessions
Average number of page views per day on server ...
- Trier 1: 192,972 pages
- Trier 2: 53,701 pages
- Dagstuhl: 16,638 pages
Average number of page views per user session on server ...
- Trier 1: 7.1 pages per session
- Trier 2: 11.5 pages per session
- Dagstuhl: 27.6 pages per session
Total number of distinct users (i.e., IPs) in October 2014 on server ...
- Trier 1: 430,890 IPs
- Trier 2: 74,818 IPs
- Dagstuhl: 6,782 IPs
Total amount of data served in October 2014 (in parenthesis: including bots) on server ...
- Trier 1: 999.1 GB (4,166.2 GB)
- Trier 2: 220.7 GB ( 397.5 GB)
- Dagstuhl: 26.8 GB ( 36.3 GB)
Browse by date
- No labels