The invention discloses a JSON data analysis and modeling method oriented to Web 2.0. The method comprises the following steps that 1, a metadata tree MDT is defined and comprises a root node Troot, an index node INode and a data node DNode, meanwhile, the number of layers of the nodes in a tree structure is expressed through a depth level, and the root node is defaulted to be a first layer, namely the level is equal to 1; secondly, JSON data returned by the called REST API is read, a JSON data set ds is used as input of a modeling algorithm, and tree structure modeling is conducted on ds; and3, through a pruning method, basic data structure information is reserved, the same repeated data structure is cut off, the volume of the model is reduced, but the basic data structure is still reserved, and the query overhead caused by the JSON data structure is reduced. The method is beneficial to accelerating retrieval and positioning of the data, enhancing the accuracy of judging the subordination relation, and improving the data utilization rate and usability.