This paper discusses how to (semi-)automatically extract non-taxonomic relation of discipline terms from Chinese unstructured text so as to find feasible and effective extracting methods. First, papers of Digital Library are retrieved from CNKI; then terms and transitive verbs are extracted; third, vector space models are constructed; fourth, association rules are analyzed and evaluated; and last, the term pairs with stronger relation are acquired and the transitive verbs used as the labels of relation, thus the non-taxonomic relation of Chinese discipline terms is extracted. The above method is more feasible and effective than other methods, and it can improve the extracting accuracy by evaluating the effectiveness and practicality. This paper of course has limits. When evaluating the effectiveness and practicality of association rules, the indicators and thresholds are determined by manual intervention, so the method has subjectivity to some extent.