Knowledge Sanitization of Large Language Models