The invention relates to a subject and predicate coding based text watermark embedding and extraction method. An embedding method comprises: 1) representing each character of watermarking information with Unicode codes to form a Unicode code string; 2) detecting out subjects and predicates of statements in a to-be-embedded text, and storing the subjects and predicates in a set; 3) according to the quantity of the subjects and predicates, dividing the Unicode code string into a plurality of segments, representing each of the subjects and predicates with one of the segments by coding, and giving a number; and 4) storing the subjects and predicates, the Unicode code segments corresponding to the subjects and predicates, and the numbers of the code segments in sequence to form a codebook, and finishing the coding to realize the watermark embedding. An extraction method comprises: finding out the detected subjects and predicates in the text; by reference to the codebook, extracting the Unicode code segments and the numbers of the code segments corresponding to the subjects and predicates; splicing the Unicode code segments according to a number sequence; and converting the obtained Unicode code string into the corresponding character to form the watermark information. The method makes no change for the format and content of the text, has good concealment and robustness, and is simple in algorithm construction and easy to realize.