The invention discloses a knowledge graph construction method for a mathematical tutoring question-answering system, and a system thereof. The knowledge graph construction method comprises the following steps: acquiring a preliminary entity of mathematical knowledge by utilizing a web crawler program; electronizing mathematics teaching materials and tutoring books, employing character matching, aTF-IDF algorithm, a TextRank algorithm and word vector similarity calculation for screening, expanding and aligning of preliminary entities, and forming entities containing mathematical concepts, knowledge points and question types; carrying out the entity classification and entity relationship identification by combining Bi-LSTM with Attention; taking knowledge content, error-prone points, difficulties and tutorials as attributes, and adopting a rule template to extract entity attributes and attribute values; and storing, querying, reasoning and visualizing the mathematical knowledge graph byusing a graph database Neo4j. The invention has the beneficial effects that the construction efficiency and the knowledge coverage rate are improved, and the association between subject knowledge andquestion types, key and difficult points, error-prone points and tutorials is established; the constructed graph not only supports automatic reasoning of mathematical problems, but also supports automatic questioning and answering of mathematical tutoring.