The invention relates to an open entity relationship extracting method based on a sentence meaning structure model and belongs to the technical field of computer and information sciences. Firstly, texts of microblog data are extracted to segment sentences and words, remove stop words and label characteristics or properties of words, then a dependency parsing tool is utilized to obtain a dependencysyntax parsing tree; secondly, candidate arguments are determined according to basic noun identifying rules, an entity relation triad is obtained by combining with relative word extracting rules andargument extracting rules, confidence coefficient calculating rules are utilized to screen the entity relation triad, and a candidate entity relation pair is obtained; thirdly, Sim1 is obtained basedon CSM calculation sentence similarity, and accordingly a sentence similarity matrix is obtained; finally, similar sentence groups are divided through the generated sentence similarity matrix according to a similarity threshold value, and the entity relation pairs included by sentences in the groups and the confidence coefficients corresponding to the pairs are combined to merge the entity relation pairs in the groups. NLP&&CC microblog evaluating linguistic data is experimentized by adopting the method, a result shows that the accuracy rate and recall rate are improved by clamping the confidence coefficients of the entity relation pairs, dividing similar sentence groups and merging the entity relation pairs in the groups.