---

FairMon is a graphical interface for fairness monitoring with RTLola. It provides an intuitive configuration interface for the monitor and enables visualization of its results at runtime.

Running FairMon

We provide prebuild binaries for running FairMon. We recommend downloading and running the fairmon_*.AppImage image:

chmod +x fairmon_*.AppImage
./fairmon_*.AppImage

Building from Source

FairMon is implemented using Rust and Tauri. It requires an installation of Rust and pnpm.

To run the graphical interface of FairMon, execute

pnpm tauri dev

inside the gui subdirectory.

Further, a command line interface is available as

cargo run --bin rtlola-dsa

EU API

To access the EU Transparency API, an API key is required. We refer to the EU research API for more information.

Examples

Here we collect examples to try out in FairMon. Each section references a specification file, which should be copied into the FairMon IDE, and the configuration of the data source.

1. COMPAS

• Specification: specs/compas.lola
• Data Source: CSV
  • Trace File: Select traces/compas.csv
  • Delay: No Delay
  • Time Format: Relative Float

2. EU API

If you obtain an API key, you can run the examples with arbitary configuration. However, we also provide a trace file that you can use without an API key. The following two configurations should produce the same results:

• Data Source: DSA API
• • From: TODO!
• • To: TODO!
• • Page Size: 50
• • Require Decision: CHECKED
• • Filter Language: English, German
• • Filter Platform: TikTok, Facebook, Instagram
• • API Key: Insert you API key here

---

• Data Source: CSV
  • Trace File: Select traces/moderation.csv
  • Delay: No Delay
  • Time Format: Absolute Rfc

---

We provide the following specifications:

1. specs/automation_analysis.lola: Analyse the ratio of automatic vs manual decisions per platform.
2. specs/automation_per_language.lola: Analyse the ratio of automatic vs manual decisions per language.
3. specs/communication_delay.lola: Analyse the time delay for the moderation decision per platform.

Copyright

Copyright (C) CISPA - Helmholtz Center for Information Security 2025-2026.
Authors: Frederik Scheerer, Tobias Wagenpfeil