Abstract: Research on question answering system of agricultural knowledge map is of great significance in the field of intelligent agriculture．A question answering system of potato pests and diseases knowledge map was designed, taking potato pests and diseases as an example．After labeling data by three models of BIO, BMES and BIOES, Bilstm-CRF, Word2vec Bilstm-CRF and Bert Bilstm-CRF, were designed for named entity recognition．Experimental results showed that BEMS annotation method was the best, Bert BiLSTM CRF model has the best performance in named entity recognition, and its F₁ value was 85.62%．Question and answer interaction was realized by matching corresponding cypher statements with the Neo4j diagram database, and the VUE front-end framework based on javascript language was applied to build a complete question and answer system page．