← Search

Data · dataset · 2025

PresidioMCP: Restricted FastMCP Server for Detecting and Pseudonymizing PII in Tabular Data

Listed in International Potato Center

This dataset contains restricted research software implementing PresidioMCP, a FastMCP server that supports privacy-preserving processing of tabular datasets by detecting potential personally identifiable information (PII) and pseudonymizing selected columns.

Description

The server provides tools to (1) list columns from local CSV/Excel files, (2) detect PII presence per column by sampling rows and running Presidio-based structured analysis, and (3) pseudonymize selected columns using consistent per-column pseudonyms written to a new output file.

The implementation includes a helper Pseudonymizer that ensures stable replacements within each entity/column type. The deposit also includes pytest-based automated tests that validate the end-to-end workflow (column listing, PII detection, and anonymization) and error handling. The deposited files are restricted because they currently contain non-public software.

Links

Where it is published

Catalogue records · 1

Topics

Inferred from text
Tabular 75%
Provenance · 1 source records, 8 field assertions
SourceKeyLast seenRaw
International Potato Centerdoi:10.21223/P3/DX5MWO13 d agoJSON v1
FieldAssertionExtractorEvidence
concepts[field].dataverse_subject:computer-and-information-sciencesource · data cipotato orgconnector:data_cipotato_org@1.0.0/subjects
concepts[modality].local:modality:tabularenrichment · data cipotato orgkeyword-concept-rules@1.0.0title+description (75%)
created_datesource · data cipotato orgconnector:data_cipotato_org@1.0.0
descriptionsource · data cipotato orgconnector:data_cipotato_org@1.0.0/description
publication_datesource · data cipotato orgconnector:data_cipotato_org@1.0.0
titlesource · data cipotato orgconnector:data_cipotato_org@1.0.0/name
updated_datesource · data cipotato orgconnector:data_cipotato_org@1.0.0
version_labelsource · data cipotato orgconnector:data_cipotato_org@1.0.0