Dr. rer. nat. Robert Jäschke

  • Appelstraße 4
  • 30167 Hannover
  • Email: (no Word attachments!)
  • For Spammers:
  • PGP-Key: 0x7762DDED (E034 140F BBDF D647 BDA7 192D F17D 65F5 7762 DDED)
  • ORCID: 0000-0003-3271-9653
Robert Jäschke

Research

My research interest is Web Science, “the emergent science of the people, organizations, applications, and of policies that shape and are shaped by the Web, the largest informational artifact constructed by humans in history” (from the call for papers of the ACM Web Science conference). Thereby, my research is situated in computer science, with multi-disciplinary connections to psychology, sociology, economics, and the digital humanities.

More specifically, together with colleagues from the Leibniz Research Alliance Science 2.0, I am investigating, how the (social) web is changing the research landscape and how it can improve communication, collaboration, participation, and open discourse.

I am also leading the development of the collaborative tagging system BibSonomy, which is both a valuable tool for researchers to organize their literature as well as a test-bed for our methods and results. In that context, I am interested in the development and integration of recommendation methods for tags and scientific publications for social bookmarking systems. Further topics of interest include citation and link analysis, entity matching and resolution, and social network analysis.

I extensively leverage big data technologies like Hadoop, HBase, Drill, or Elasticsearch for my research, e.g., to analyze crawled web pages of universities in the context of Science 2.0. Therefore, I have designed a dedicated cluster system for L3S Research Center, consisting of 40 nodes with an overall disk space of 1 Petabyte and 400 CPU cores. Since 2013 the first stages are installed and I am managing the operation and further extension of the cluster.

Publications

My publications

Top Publications

Books

Recommender Systems for Social Tagging Systems
Formal Concept Analysis and Tag Recommendations in Collaborative Tagging Systems

Theses

Events

We are organizing the 8th International ACM Web Science Conference 2016 in Hannover. As a local chair I welcome you to submit your paper to the conference and visit Hannover from May 22 to May 25, 2016.

PC/Workshop Chair

PC Member

past events

Talks

Projects

BibSonomy and PUMA

Much of our research is focused around our social bookmark and publication sharing system BibSonomy. The system is online since 2006 with me being the main developer from 2005 to 2012. Since 2009 I am leading the development and operation of the system together with Andreas Hotho. If you are interested in a cooperation, just let me know.

Together with the University Library Kassel we have extended the BibSonomy platform in the DFG-funded PUMA project for academic publication management. If you are interested in using PUMA, please contact us.

World Literature

Together with Frank Fischer and Mathias Göbel we are writing about Digital Humanities in general and our research on world literature in particular on weltliteratur.net.

Science 2.0

As a member of the Leibniz Research Alliance Science 2.0 we are partners in the following projects:

ALEXANDRIA

I am leading the development of the ALEXANDRIA testbed which includes a dedicated Hadoop cluster environment which I have designed. The current configuration of the cluster is 25 nodes with 268 CPU cores, 2688 GB main memory and 1008 TB disk space. We regularly crawl the German Academic Web and curate a large collection of tweets (currently 5 billion tweets spanning more than three years).

L3S Internal Research Projects

I am project leader of the #science, FizzStream, and Hadoop@L3S projects.

Fun

Source Code

Some of my source code is available on GitHub, other code is linked here.

FolkRank
FolkRank is an algorithm for search and ranking in collaborative tagging systems. It has been integrated into the community support architecture of the social semantic desktop developed by the NEPOMUK project. The source code is available from the project’s SVN repository.
Trias
Trias is an algorithm for computing triadic concepts which fulfill minimal support constraints. The source code is available on the project page.
BibSonomy
Some of the modules which BibSonomy is based on are available in a Maven repository. The complete source code of BibSonomy is available on Bitbucket, there you can also find an issue tracker and exemplary code snippets in the tools and BibSonomy Python projects.
ConExp-NG
ConExp-NG is a re-implementation of ConExp - the Concept Explorer - a well-known tool for Formal Concept Analysis.

Data Sets

We publish regular snapshots of the BibSonomy database. There you can also find the datasets of the ECML PKDD Discovery Challenges 2008 and 2009.

Datasets from publications:

All datasets from L3S Research Center can also be found on data.l3s.de.

As part of the ALEXANDRIA project we are creating a longitudinal collection of academic web pages from Germany. Since 2012 we crawl every six months the web sites of all German universities. This currently (2016) comprises seven crawls, each comprising around 6TB and 100 million URLs.