Home - juergenlerner/eventnet GitHub Wiki

Event network analyzer (eventnet) is a software for the statistical analysis of networks of relational events.

A relational event represents time-stamped interaction ( who does when what to whom). Examples include a person sending an message to another person, a Wikipedia user editing a Wikipedia article, a customer buying a certain product, or a country signing an agreement with another country. A relational hyperevent represents time-stamped interaction among varying and potentially large numbers of "nodes". Examples include meeting events, team work (e.g., coauthoring of papers), multi-cast or "one-to-many" communication such as email, or papers citing lists of other papers in their references.

Relational (hyper) event models R(H)EM seek to model sequences of such interaction events to uncover or test factors explaining that some nodes interact at a higher rate than others. While the core statistical models are well-established in survival or event-history analysis (e.g., Cox regression) and are implemented in numerous free and commercial statistical software packages, typical explanatory variables of REM and RHEM are novel, network-specific, and sometimes conceptually and computationally demanding to understand and implement. This gap is filled by the eventnet software.

To use eventnet (for REM and for RHEM), download the file eventnet-1.3.jar and start the program by either of:

  • double-clicking on the JAR file opens the eventnet graphical user interface (GUI)
  • typing the command java -jar eventnet-x.y.jar opens the eventnet GUI
  • typing the command java -jar eventnet-x.y.jar <configuration_filename.xml>, where <configuration_filename.xml> is the name of a file containing a fully specified eventnet configuration, executes this configuration without opening the GUI.

Eventnet is written in java and needs the java runtime environment (JRE), Java 8 (JRE version 1.8) or higher. An alternative to the JRE by Oracle is the Eclipse Temurin JRE.

Eventnet one (Version 1.0 or later) comes with three important changes.

  • The functionality for dyadic relational event models (REM) and for relational hyperevent models (RHEM) is now provided in a single JAR file (eventnet-1.0.jar or later).
  • RHEM can now also be specified purely in the graphical user interface (GUI).
  • RHEM effects have been completely reorganized. The number of different core types of RHEM statistics could be reduced but a more efficent use of the arguments of statistics actually provides a much larger variation of possible RHEM effects than in prior versions. Note that because of this reorganization, configuration files from versions prior to 1.0 will most likely no longer work with eventnet one. (Note, however, that the JAR files of prior versions are still available at https://github.com/juergenlerner/eventnet/tree/master/jars/old_versions-0.x.)

RHEM effects available in Version 1.0 or later are exhaustively listed and discussed in the Reference guide on RHEM effects. However, users new to eventnet are recommended to first have a look at the more basic tutorials linked below.

First steps and tutorials.

A first introduction to eventnet is provided in the first-steps tutorial, which briefly discusses the main parts of the eventnet GUI and then goes step by step through an exemplary analysis with a tiny, made-up dataset.

More details can be found in the basic tutorial. This tutorial goes step by step through an empirical case-study seeking to explain allocation of attention of contributing Wikipedia users to Wikipedia articles.

The tutorial on structural balance specifically demonstrates how to test the predictions of balance theory in networks of positive and negative interaction events. Adding to more basic tutorials, it illustrates two main issues: (1) how to assess the effect of exogenously given covariates (that is, externally defined properties of nodes or dyads) on dyadic interaction and (2) the different possibilities to model interaction events that are typed and/or weighted.

We further illustrate how to analyze large event networks, comprising millions of nodes and hundreds of millions of events.

Relational hyperevent models (RHEM) are for events involving any number of actors or nodes, such as meetings, co-authoring, multi-cast communication (e.g., email), or citation networks. A first introduction to specifying RHEM with eventnet is provided in the RHEM first steps tutorial, which illustrates the analysis of meeting events in the "Southern Women Data" from Davis, Gardner, and Gardner.

RHEM for directed hyperevents are explained in the tutorial on RHEM for multicast interaction networks. That second RHEM tutorial also explains how to specify effects based on actor-level (node-level) attributes. It uses example data from the Enron email corpus.

Extending the above tutorial, we provide a tutorial on analyzing the coevolution of collaboration and references to prior work, which illustrates how to specify and estimate RHEM for events in mixed two-mode networks. The illustrative empirical example are scientific networks, where events are generated by the publication of scientific papers, written by a team of authors and citing a list of prior papers in their references. Nodes in this network are authors or papers, where authors are connected to the papers they write and papers are connected to the references they cite. Building on the previously described RHEM, this tutorial explains how to specify effects in these interdependent collaboration and citation networks.

RHEM effects available in Version 1.0 or later are exhaustively listed and discussed in the Reference guide on RHEM effects. We recommend that new users should first have a look at least at some of the more basic tutorials linked above and only then read (selected parts of) this more formal reference guide.

A different type of information is provided in the Troubleshooting help page. You would look into this page if you have read one or few tutorials but you could not reproduce the examples there - or you could not adapt them to your own project. The help page goes through a list of common or less common problems and tries to suggest solutions.

References

There is a commented literature list of own work on relational event models (REM) or relational hyperevent models (RHEM), using the eventnet software or (published or unpublished) predecessor software in its empirical analysis. The listed papers are expected to complement the eventnet tutorials in providing more formal details, a better embedding into related work, and a more thorough discussion of objectives and contributions. They may also point to further illustrating use cases of REM and RHEM.

Forum

Use the eventnet-users forum to discuss all questions, issues, or topics concerning the eventnet software, relational event models (REM), and relational hyperevent models (RHEM).

Training

Moreover, training workshops or courses introducing eventnet are offered at the following conferences or summerschools.

Next training events

  • The workshop Extending the relational event model by Ernst Wit, Alessandro Lomi, Jürgen Lerner, Martina Boschi, and Melania Lembo at the INSNA Sunbelt 2025, 23-29 June 2025 in Paris, France will teach several basic and advanced concepts on REM and RHEM, including non-linear effects, time-varying effects, polyadic interaction events (hyperevents), and methods to asses the goodness-of-fit.

Past training events

  • RHEM course at the International "Winter Course," Theory, Methods and Applications of Personal Networks February 3-7, 2025. UAB Egolab, Barcelona, Spain.
  • REM/RHEM course at the POLNET Winter School on Social Network Analysis 13-17 January 2025. IBEI, Barcelona, Spain.
  • REM beyond dyads: relational hyperevent modeling with eventnet (beginners and advanced), within the INSNA Sunbelt Conference 2024, Edinburgh. 24-30 June 2024.
  • International "Winter Course," Theory, Methods and Applications of Personal Networks, UAB Egolab, Barcelona, Spain. February 5-9, 2024.
  • REM beyond dyads: relational hyperevent modeling with eventnet (beginners and advanced), within the European Conference on Social Networks (EUSN 2023), Ljubljana, Slovenia. 4-8 September 2023.
  • Relational Event Models (REMs) for the Analysis of Social Networks: A Hands-on Introduction, part of the NUSC Summer School in Network and Data Science, Greenwich, UK. 19th - 23rd June 2023.
  • International "Winter Course," Theory, Methods and Applications of Personal Networks, Barcelona, Spain. February 6-10, 2023.
  • REM beyond dyads: relational hyperevent modeling with eventnet, within the European Conference on Social Networks (EUSN 2022), Greenwich, UK. 12-16 September 2022 (the RHEM workshop is on September 12, afternoon).
  • Relational Event Models (REMs) for the Analysis of Social Networks: A Hands-on Introduction, part of the NUSC Summer School in Network and Data Science, Greenwich, UK. Mon 20th - Sat 25th June 2022 (the REM course is on June 24, 25).
  • International "Winter Course," Theory, Methods and Applications of Personal Networks, Barcelona, Spain. February 7-11, 2022.
  • European Conference on Social Networks, Zurich, Switzerland. September 12, 2019.
  • Sunbelt Social Networks Conference, Montreal, Canada. June 19, 2019.
  • Relational event models for the analysis of social networks, University of Exeter, Business School. June 3-4, 2019.

Citation

For eventnet in general please cite: Lerner and Lomi (2020). Reliability of relational event model estimates under sampling: how to fit a relational event model to 360 million dyadic events. Network Science, 8(1):97-135. (DOI: https://doi.org/10.1017/nws.2019.57)

For relational hyperevent models please cite: Lerner and Lomi (2023). Relational hyperevent models for polyadic interaction networks. Journal of the Royal Statistical Society: Series A.

Contact

Use the eventnet-users forum to discuss all questions, issues, or topics concerning the eventnet software, relational event models (REM), and relational hyperevent models (RHEM).

Funding

We acknowledge financial support from Deutsche Forschungsgemeinschaft (DFG) under Project No. 321869138. From 2017 to 2020: "Statistical analysis of structural balance in signed social networks" (LE 2237/2-1) and from 2021 to 2025: "Statistical analysis of time-stamped multi-actor events in social networks" (LE 2237/2-2).

We are happy to announce that further development of eventnet will continue, funded by Deutsche Forschungsgemeinschaft (DFG) under Project No. 555455503. From 2026 to 2028: "Statistical network analysis of time-ordered social interaction events with internal structure" (LE 2237/4-1).

Third-party material

Below is a list of potentially useful resources related with eventnet, REM, and/or RHEM created by third parties.