Real-time process execution logging program for the Cloud Process Execution Engine (CPEE) (https://cpee.org/). The server subscribes to CPEE's Data Stream Interface and produces XES-compatible multi-document YAML trace files without modifying the engine itself.

Developed as part of the Bachelor Thesis "Using Instance-Based Events During Process Execution to Create Trace-Based XES Logs" at the Technical University of Munich.

---

Core Features:

• Event Exclusion
• Loop Unrolling
• Subprocess Integration

---

Repository Structure:

log_server.py              # Logging server
run_evaluation.py          # Evaluation tool
requirements_eval.txt      # Python dependencies
alternative_traces/        # Traces generated by the logging server (5 for each process model)
expost_traces/             # Traces generated by CPEE's internal logging server (5 for each process model) 
testsets/                  # CPEE process model XML files
  case_01_plain.xml
  case_02_parallel.xml
  case_03_exclusive_choice.xml
  case_04_subprocess_default.xml
  case_05_subprocess_include.xml
  case_06_subprocess_exclude.xml
  case_07_subprocess_exclude_include.xml
  case_08_loop_pretest.xml
  case_09_loop_posttest.xml
  case_10_nested_loop.xml

---

Requirements:

Python 3.9 or higher. Install dependencies with:

pip install flask PyYAML fpdf2

Or using the requirements file:

pip install flask -r requirements_eval.txt

flask is required for the logging server but not listed in requirements_eval.txt (which covers the evaluation tool only).

---

Running the Logging Server:

Start the server on the default port (9379):

python3 log_server.py

The server binds to all interfaces on port 9379. Two directories are created automatically on startup:

• raw_events/ — per-instance event journals (phase 1 output)
• traces/ — final XES YAML trace files (phase 2 output)

---

Local Setup:

The testset XML files currently contain subscription and endpoint URLs that point to the example university server "lehreserver".

To run the logging server locally, these URLs must be updated to point to the local machine.

1. Subscription URL

Each testset contains a <subscription> element at the bottom, with an url attribute. This specifies where CPEE sends its events. By default this points to the university server:

<subscription xmlns="http://riddl.org/ns/common-patterns/notifications-producer/2.0" id="_multitracelog" url="https://lehre.bpm.in.tum.de/ports/9379">
      <topic id="state"><event>change</event></topic>
      <topic id="activity"><event>calling</event><event>receiving</event><event>done</event><event>annotation</event></topic>
      <topic id="description"><event>change</event><event>exposition</event></topic>
      <topic id="endpoints"><event>change</event></topic>
      <topic id="attributes"><event>change</event></topic>
      <topic id="task"><event>instantiation</event></topic>
      <topic id="gateway"><event>annotation</event><event>decide</event></topic>
</subscription>

The url needs to be changed to the address the local logging server, for example:

url="http://localhost:9379"

2. Subprocess Endpoint URLs

Testsets with subprocesses involved, reference a subprocess model URL inside the <arguments> block:

<url>https://lehre.bpm.in.tum.de/~ge42toh/logging_sub_process.xml</url>

This URL must also be updated to point to the location where the subprocess model file is hosted, so that the CPEE instance can fetch and execute it.

Alternatively, it can be done by uploading the testset into CPEE, and chaning it directly in the UI.

---

Configuring Logging Behavior:

Each task and loop element can be annotated in the CPEE graphical interface under Annotations → Logging Behavior:

Option	Applies to	Effect
Exclude from log	Tasks, subprocess calls	Removes all lifecycle events of this activity from the trace
Include nested events	Subprocess calls	Merges child subprocess events into the parent trace
Log per iteration	Loop elements	Generates one trace file per loop iteration

---

Output Format:

Traces are written to the traces/ directory as multi-document YAML files.

Running the Evaluation:

The evaluation tool compares the alternative logger's output against reference traces.

Run:

python3 run_evaluation.py \
  --xml path/to/reference_processmodel.xml \
  --original-logs path/to/reference_traces/ \
  --alternative-logs path/to/traces/ \
  --output-dir eval_results/

Results are written to eval_results/ as:

• evaluation_report.csv
• evaluation_report.pdf