A tokenization system receives a request for data anonymization, the request referencing unstructured/semi-structured content containing values of interest. The tokenization system performs a tokenization operation on the unstructured/semi-structured content, generates self-describing tokens for the values of interest, each self-describing token having a preconfigured pattern, an indication of a protection strategy, and a token value, and stores the values of interest in a secure data vault. The tokenization system may receive a request to reveal the self-describing tokens in the unstructured/semi-structured content. In response, the tokenization system searches the anonymized version of the unstructured or semi-structured content for the preconfigured pattern, identifies self-describing tokens, uses the self-describing tokens to retrieve the values of interest from the secure data vault, and produces a detokenized version of the unstructured/semi-structured content containing the values of interest.