-
Hello Team,
Thank you for the comprehensive survey. Can you please include the work [PII-Compass: Guiding LLM training data extraction prompts towards the target PII via grounding](https://aclanth…
-
Recently as new [paper](https://arxiv.org/abs/2202.00443) released a new -
> Text Anonymization Benchmark (TAB), a brand-new dataset containing 1268 court cases from the European Court of Human Rig…
-
From the README (https://github.com/DigitalPhonetics/VoicePAT?tab=readme-ov-file#anonymization) it says that to anonymize my own data I should modify the following fields in the config file:
```
dat…
-
Thanks for this open source project!
I'm currently writing my thesis on text anonymization models and their performance. I would like to assess different models on different datasets and I'd also lik…
-
Comparison with [pydicom](https://github.com/pydicom/pydicom) and other DICOM handling packages should be added at some point.
Some helpful benchmark(s):
- Anonymization (for text/other well-formatt…
-
Hi,
The TAB dataset and evaluation approach is amazing! It would be very useful for those interested to train models on this dataset, to have it as a [HuggingFace dataset](https://huggingface.co/da…
-
Follow-up on #59
In anonymization mode the whole text is alawys displayed in the graph.
-
- [ ] Default to using the [much faster `System.Text.Json`-based FHIR parser](https://github.com/miracum/fhir-pseudonymizer#usesystemtextjsonfhirserializer) and remove the option to unset it
- [ ] Fo…
chgl updated
11 months ago
-
### Feature request
When calling scrub.scrub_text, instead of just replacing with `***`, we would like to retain semantic meaning. e.g. given:
```
Alice said hi to Bob
```
We want:
```
s…
-
has to be refined
![grafik](https://github.com/dzhw/metadatamanagement/assets/17981185/5a861fba-ab11-4dd7-a6e2-25d63c7a7710)