The purpose of this article is to identify the difficulty of learning materials in the network by using learner's portfolio in the asynchronous learning system. Asynchronous learning takes the advantage of information technology that records the learning portfolio of the learner. The data of the learning portfolio reflects the characteristics of the learning materials and the user. To explore useful information from the huge learning data warehouse and to build up a material characteristic model, the researchers use data mining technique. Several experiments were conducted, based on the built material characteristic model, to estimates the value of the difficulty of learning materials by a recursive method. In order to obtain an accurate difficulty value of the material, the researchers continuously conducted the experiments several times. After the data of each experiment was collected and integrated with the previous ones, the difficulty of the material was revised in the recursive way. Therefore, the degree of difficulty of learning material was dynamically adjusted according to the real learning data. It becomes more accurate and more objective. (Contains 10 figures and 8 tables.)