QualiLens

Research applicability

Local web application documenting researcher-checkpointed LLM assistance for qualitative coding and literature synthesis.

Getting started

Review the manual and participant-data handling, launch via run.sh, select a method and provider, upload a small consented corpus and inspect checkpoints and located quotes. These are documentation-based editorial adoption suggestions, not an exercised workflow.

Maintenance

Maintenance and runtime behavior not independently audited.

License

README identifies Apache License 2.0; obligations and dependency licenses not audited.

Workflow & human oversight

Defined wizard/checkpoint workflow: researchers set questions and methods and edit codes/themes before continuation; these are documented design features, not verified execution.

Workflow level is not an evidence score; higher is not better evidence.

Approach

README describes five methods, editable checkpoints, excerpt/source-location tracking and audit/report exports; literature synthesis uses located extraction quotes.

Limitations

Local storage does not mean offline inference: data goes directly to the selected LLM provider and synced folders expose it to the sync service. README says no multiple-human intercoder reliability modeling, no Whisper speaker diarization and no Intel Mac support. Tolerant/partial quote matching is not full semantic validation; security and spend protections were not audited.

Reproducibility

Primary repository documentation checked live. No installation, software execution, security audit, or outcome replication performed.

Adoption requirements

Python 3.11+ on Apple-silicon Mac or Linux/WSL, supported provider API credentials, and ffmpeg for video; audio transcription needs an OpenAI key.