OOPS+

Authors:

• Carlos Gaspar Santiago,
• María Poveda Villalón
• Daniel Garijo

Description

OOPS+ is an extension of the original OOPS! ontology pitfall scanner. The original OOPS! application provides the immediate ontology evaluation report, while OOPS+ adds a second report pipeline for checks that require an LLM.

The current setup contains two web applications:

• OOPs: the regular OOPS! application. It generates the normal report immediately.
• OOPsPlus: the extended application. It processes OOPS+ jobs and generates the LLM-based report.

When the user enables OOPS+ in the OOPs web interface, OOPs stores the submitted ontology in a shared analysis directory, creates a job, and publishes it to RabbitMQ. An OOPS+ worker consumes that job, runs the OOPsPlus analysis, writes the generated HTML report, and updates the job status. The OOPS+ report page reads that status and displays the report when it is ready.

Architecture

The root Docker Compose file starts:

• web-app-oops: normal OOPS! web app, available on http://localhost:8080.
• web-app-oops-plus: OOPS+ report viewer, available on http://localhost:8081.
• oops-plus-worker: background worker that consumes ontology jobs from RabbitMQ.
• rabbitmq: job queue used to communicate between OOPs and OOPsPlus.
• ollama: local LLM runtime used by the OOPS+ checkers.
• ollama-pull-model: helper container that pulls the configured LLM model.

Analysis files are shared through:

./data/oops_analyses

This directory contains each OOPS+ analysis folder, including the ontology, status file, and generated report.

Deployment

Create the environment file from the example:

cp .env.example .env

On Windows PowerShell:

Copy-Item .env.example .env

Check the values in .env :

LLM_IP=http://ollama:11434
LLM_MODEL=gemma3:4b

RABBITMQ_HOST=rabbitmq
RABBITMQ_PORT=5672
RABBITMQ_USER=oops
RABBITMQ_PASSWORD=oops
RABBITMQ_OOPS_PLUS_JOBS_QUEUE=oops.plus.analysis

Start the full system:

docker compose up -d --build

The first run may take longer because the Ollama model has to be downloaded.

Downloading and changing Ollama models

The model used by OOPS+ is configured in .env:

LLM_MODEL=gemma3:4b

Changing LLM_MODEL only tells the OOPS+ worker which model to use. The model must also exist in the shared Ollama volume. If you choose a model that has not been used before, pull it once with the helper service:

docker compose pull ollama ollama-pull-model
docker compose run --rm ollama-pull-model

Then check that the model is available:

docker compose exec ollama ollama list

For example, after pulling both gemma3:4b and gemma4:12b, the list should contain entries like:

NAME          ID              SIZE
gemma3:4b     ...             3.3 GB
gemma4:12b    ...             7.6 GB

If you change .env to a model that is already listed by ollama list, you do not need to download it again. Recreate the services that read the environment variable so they pick up the new value:

docker compose up -d --build web-app-oops web-app-oops-plus oops-plus-worker

If the OOPS+ worker logs show an error like:

model 'gemma4:12b' not found

then the configured model is not available in the Ollama volume. Run the pull helper again and check its output:

docker compose run --rm ollama-pull-model

If Ollama says the model requires a newer version, update the Ollama image first:

docker compose pull ollama ollama-pull-model

and then retry the pull helper.

Useful URLs:

• OOPS!: http://localhost:8080/oops-0.3.0-SNAPSHOT/
• OOPS+ report viewer: http://localhost:8081/oops-0.3.0-SNAPSHOT/
• RabbitMQ management UI: http://localhost:15672

Default RabbitMQ credentials are defined in .env.

Useful Commands

View worker logs:

docker compose logs -f oops-plus-worker

Stop the system:

docker compose down

Rebuild after code changes:

docker compose up -d --build

Optionally run multiple OOPS+ workers to process several ontology jobs in parallel:

docker compose up -d --scale oops-plus-worker=3

Keep in mind that the LLM runtime can become the bottleneck when several workers run at the same time.