The Empirical Impact of Data Sanitization on Language Models

Data sanitization in the context of language modeling involves identifying sensitive information, such as personally identifiable information (PII), and redacting them from a dataset corpus. It is a common practice used in natural language processing (NLP) to maintain privacy. Nevertheless, the impact of data sanitization on the language understanding capability of a language model remains less studied. This paper empirically analyzes the effects of data sanitization across benchmark language-modeling tasks including comprehension question answering, entailment, sentiment analysis, and text classification. Our experiments cover a wide spectrum comprising finetuning small-scale language models, to prompting large language models (LLMs), on both original and sanitized datasets and comparing their performance across the tasks. Interestingly, our results suggest that, for some tasks such as sentiment analysis or entailment, the impact of redaction is quite low, typically around 1-5\%, while for tasks such as comprehension Q\&A there is a bigger drop of >25\% in performance observed in redacted queries. For tasks that have a higher impact, we perform a deeper dive to inspect the presence of task-critical entities. Finally, we investigate correlation between performance and number of redacted entities, and also suggest a strategy to repair an already redacted dataset by means of content-based subsampling.