Multi-hop question and answer model construction method based on table-text mixed data

The invention discloses a multi-hop question and answer model construction method based on table-text mixed data. A constructed model is composed of a text content filtering module, a table line content screening module and a fusion data extraction question and answer module. The text content filtering module is used for screening text sentence segment content related to the question; the table line content screening module is used for constructing table line-text fusion data and further screening out content related to the question; and the fusion data extraction question and answer module is used for extracting answers from the table line content-text fusion data. According to the model constructed by the method, redundant information is greatly reduced by screening contents, so that text contents effectively supplement table contents, the model training speed is increased, and the accuracy of table-text mixed data questions and answers is improved. 本发明公开了一种基于表格-文本混合数据的多跳问答模型构建方法，所构建的模型由文本内容过滤模块、表格行内容筛选模块以及