Intention knowledge base construction method, identification method, device, equipment and medium

Abstract

The present application relates to the technical field of semantic recognition, and also relates to the technical field of artificial intelligence, in particular to a construction method and a recognition method of an intent knowledge base, a device, an electronic device and a storage medium, a knowledge graph is acquired, attribute strings are extracted from the knowledge graph to establish a matching knowledge base, an intent library is acquired, sentences in the intent library are classified into corresponding behavior categories respectively to obtain behavior labels of each sentence, each sentence in the intent library is matched with each attribute string in the matching knowledge base respectively to obtain attribute labels of the sentences, sentences that fail to match are analyzed to obtain fine-grained intent labels of the sentences that fail to match, and an intent knowledge base is established according to the sentences, the attribute labels or the fine-grained intent labels of the sentences, and the behavior labels of the sentences. Through the above-mentioned manner, the technical problem of low accuracy of intent recognition caused by fuzzy intent representation and intent inclusion in the prior art is solved.

Images