Identification method and system for intelligent quality inspection of service forbidden words

The invention relates to a natural language processing technology, in particular to an identification method and system for intelligent quality inspection of service forbidden words, and the method comprises: performing text transcription and text preprocessing, i.e., carrying out the letter capital and lower-case conversion and part-of-speech tagging of text data of customer service; performing keyword retrieval; performing text transcription error identification, i.e., judging whether text transcription is reasonable or wrong according to historical customer service session text data; performing text semantic analysis, i.e., analyzing semanteme of the text in the customer service session content hitting the service forbidden word keyword to obtain an identification result of whether thesupport staff uses forbidden words or not ; and filtering an identification result that the support staff does not use forbidden words to obtain a final inspection result. The context of the keywordsin the text and historical