The article discusses a new approach for identifying personally identifiable information (PII) spans during text sanitization. Traditional methods often rely on Named Entity Recognition (NER) models, which may overlook non-named entity identifiers. This study introduces the use of knowledge graphs, particularly Wikidata, to develop specialized gazetteers for various types of PII. By combining these resources with NER, the framework aims to improve the detection of text spans that can expose personal information, ultimately enhancing privacy measures in data handling.
Our priority is establishing a robust mechanism for identifying personally identifiable information (PII) by leveraging advanced techniques that integrate Named Entity Recognition within diverse contexts.
The combination of knowledge graphs and NER allows for the effective identification of PII spans that may not be labeled as named entities but still carry identifiable information.
Collection
[
|
...
]