As the metaverse is hot, nonhomogenized tokens (NFT) as digital artwork identifiers present different characteristics and application values from other homogenized tokens, and their use for copyright verification will suffer from the problems of storage space limitation and data verification reliance on database, this study designs an NFT digital copyright authentication model for textual works. To cater for the uncontrollability of conventional Hash algorithms in stream matching due to the high conflict rate, a new random matrix theory is applied to propose a new Hash algorithm, which is used on the block structure of NFT credential authentication, while extending the block structure so that the data within the work is completely stored in the blockchain with NFT as the credential, allowing the database to store the work data in a relatively safe manner. The verification of the work data has the immutability and unique cryptographic solution of NFT. The NFT-based digital model collects TR information and conducts 30 tests, and the average test time for generating blocks is 0.53 s. Through the block query for detection, 248,655 words have exceeded the number of words of an article, and the consumption time is only 0.23s, to meet the customer's real-time query requirements for the system. According to the overhead ratio, the record storage expenditure is about 2-3 times of the text storage expenditure, the work storage expenditure, and the storage expenditure for storing 240,000 words for authentication is about 3720 KB.