Corpus cleaning method, device and equipment and medium

The invention provides a corpus cleaning method, device and equipment and a medium, and the method comprises the steps: obtaining a sentence vector extraction model structure, taking the sentence vector extraction model structure from one part of a pre-trained question and answer pair model for assessing the matching condition of a question and an answer, and extracting a sentence vector of an input question or answer; extracting at least one part of corpora from all corpora serving as question and answer pairs to be cleaned; obtaining a labeling result of at least one part of corpus; a classification model is trained on the basis of a training set composed of at least one part of corpora and annotation results of the corpora, and the classification model evaluates whether the corpora is suitable for being used as a question and answer pair or not on the basis of sentence vectors extracted from question and answer pairs of the input corpora through a sentence vector extraction model structure; and screening out corpora suitable for being used as question and answer pairs from the unmarked corpora in all corpora by utilizing the trained classification model. Therefore, a large number of corpora with high quality can be obtained through a small number of manual annotations.