50 Q. Liu and Z. Sun of learning content depend on small granularity of learning content, which should be a meaningful and perfect learning content unit. The second is adaptation of learning object for different environments. Adaptive schedule of learning objects involves in some related specifications. The IMS Simple Sequencing Specification (SSS) provides a means to represent information needed to sequence learning activities in a variety of ways [3]. It is, however, a general purpose sequencing method and may be incorporated in other applications or environments. LOM Specification already provides a level of description and potential access to units of learning using its existing fields and vocabularies. However, it does not in- clude elements for either Learning Objectives or Prerequisites, which are explicitly included in the Learning Design model [4]. The ADL Sharable Content Object Refer- ence Model (SCORM) describes a Web-based learning content aggregation model and runtime environment for learning objects [5]. These specifications provides gen- eral methods for scheduling learning content, but lacks in the adaptive organization in learning content with the similar knowledge according to equipment, bandwidth, and file formats as well as other factors. In addition, many scholars provided some benefits in learning content’s organiza- tion. Mohammad etc. [6] showed adaptation models for personalization in web-based learning environment including learners’ styles, inferring engine, learning content library and result analysis, and so on. Nobuko [7] studied student model, in which it provided the appropriate learning content for the similar characteristic learners ac- cording to knowledge point’s LOC (mastering level), LOD (difficulty level) and DBC (the distance between knowledge points). Others researches focused on knowledge taxonomy, knowledge expression, inference mechanism and adaptive selection and presentation of learning content [8-11]. The paper studies the adaptive learning resources organization model based on learning objects by considering different granularity and different learning environ- ments, who takes knowledge point as units. The second analyses status of art of learn- ing content schedule. In the third, it defines the granularity of learning object, dis- cusses the relations such as prerequisite, successor, parallel and so on between knowl- edge points, designs learning content level framework, which takes knowledge point as the meaning unit, builds up the foundation of learning resources reuse and share through the granularity definition of learning resources. The forth is to provide the basic scheduling model for learning content organization. Lastly, the scheduling algo- rithm is proved effectiveness and stable. 2 Learning Object and Its Granularity The learning object is a component of learning content [2] which could be reused. It may be a knowledge point, also lesson content. According to the learner’s individual demand, the realization of learning content‘s dynamic construction have the special demand on learning objects. The learning content is formed by some relatively inde- pendent learning units, which are combined by some ways. And the learning unit is also formed by certain knowledge points by similar ways. Thus, a knowledge point possibly can appear in many different learning units. So, the paper takes knowledge point as learning object. Research on Learning Resources Organization Model 51 2.1 Granularity of Learning Object and Learning Content Hierarchy According to the granularity of learning content, a learning content may be divided into atomic knowledge, knowledge point, knowledge cluster, and knowledge tree. Atomic Knowledge: ∀ a,b is the learning content, and a ∩ b= ∅ , then a, b is the atomic knowledge. Atomic knowledge is the knowledge unit, who is cannot be di- vided again in knowledge system, also is the smallest unit, who can embody concept, rule, and theory and so on. Knowledge point: i a i 1,2,……,n composes the set K, i a is atomic knowl- edge, ∀ i a ∈ K, n aaa ∪∪ ⋅⋅⋅ 21 A i 1,2,……,n ,also A ≠ ∅ , then A is knowledge point. Knowledge point “A” is the relative integrated learning content. A knowledge point is composed of learning content, appraisal content and practice con- tent and so on. Knowledge cluster: ∀ i A is knowledge point, n AA ∪∪ ⋅⋅⋅ 1 α i 1,2,……,n , then α is the knowledge cluster. Knowledge cluster comprised with knowledge points is basic independent knowledge organization unit in the learning resources scheduling process. According to learner's memory characteristic in psy- chology, a knowledge cluster is composed of 7 2 knowledge points. Knowledge tree: In knowledge system, all knowledge clusters is in terms of certain strategy or sequence, for instance, instruction sequence is called the knowledge tree. In learning resources, knowledge point is not isolated, but has some correlations. The knowledge points’ relations mainly include inheritance, prerequisite, successor, parallel, connection, similar and so on. Fig. 1. Resource hierarchy model based on knowledge point 2.2 Learning Content Hierarchy Learning resources hierarchy is showed in fig 1. The figure shows that learning re- sources is on the top of the hierarchy. In their returns, there are subject, specialty, 52 Q. Liu and Z. Sun course, and knowledge cluster and knowledge point. The top conception is composed of the below, the relation between different levels is inheritance, but in identical level, possibly includes prerequisite and successor, parallel as well as relevancy and other relations. The knowledge point has many kinds of manifestations and different presentation formats. The same knowledge point may represent by some similar learning object, each learning object according to its own demand. In the scheduling process of learn- ing content, organization algorithm and the learner’s characteristic are the key factors to realize the optimization of learning process. Fig 2 shows the knowledge points’ sequence result. Knowledge point 1 has learning object LO1, LO2, and LO3 with high similar degree. One learner’s possibility learning sequence is: [KP1, LO1] → [KP2, LO3] → [KP3,L02] → [KPn,LO1] Fig. 2. The sequence representation of the similar knowledge points 3 Learning Resource Organization Model Learning content defines a knowledge cluster as the basic learning unit. Fig 3 is the model of learning content organization, which includes pre-test, content outline, 7±2 knowledge points, brief summary, knowledge tree, post-test and evaluation [12]. According to the results of testing and the sequence of knowledge trees, the system provides the learner with the appropriate knowledge point. 3.1 Pre-test It needs a pre-test to test whether the learner have the ability to learn the knowledge cluster after choosing it. There are more questionnaires and testing during the pre-test. If the learner does not pass the test, the system will tell him to carry on remediation learning or advise the learner to learn other knowledge clusters. otherwise, the learner begins to learn the knowledge cluster or the next knowledge cluster. 3.2 Scheduling Algorithm of Knowledge Points 3.2.1 Choosing the Knowledge Points To choose knowledge points is the key of personalized scheduling of learning content, which is mainly choosing the learning knowledge points on the basic of knowledge Research on Learning Resources Organization Model 53 Fig. 3. Learning content organization Model points’ relations. If K is the learning knowledge point, and it is included in knowledge cluster KC, we can choose the successor knowledge point according to the following rules. Rule 1: if K has only one successor, the learning knowledge point is the successor. Rule 2: if K has many successors, the learning knowledge point is the nearest suc- cessor. Rule 3: if K has no successor, the learning knowledge point is the knowledge point paralleling with K. Rule 4: if K is the last knowledge point of KC, and the learner studies all, the sys- tem will provide a test, according to the result, the system determines to whether learn the next knowledge cluster or not. 3.2.2 Knowledge Point Relations Graph The knowledge points’ relations are illustrated by using knowledge concepts graph. Knowledge concepts graph is a directed graph, and it has some characteristics. (1) Each node represents a knowledge point. (2) Node A points to Node B, which means knowledge point A is the prerequisite of knowledge point B. (3) Each node has a weight value, the value is bigger, the degree of knowledge master is more important. The knowledge concept graph comes into being on the foundation of analyzing learning objectives and the courses’ structure. In order to provide knowledge point for the learner, and the present knowledge point and the sequences of knowledge points’ relations are considered. Figure 4 shows five knowledge points’ relations in a KC (Knowledge Cluster). The relations of the five knowledge points can be illustrated by using prolog as follows. Nonpre (K1) K1 has no prerequisite knowledge point. Nonpre (K2) K2 has no prerequisite knowledge point. pre (K3,K1) K1 is one of the prerequisite knowledge points of K3. 54 Q. Liu and Z. Sun Fig. 4. The figure of knowledge sequence According to the rules of choosing knowledge points as above, if the learner wants to learn a knowledge point, he must learn all prerequisite knowledge points. For ex- ample, if he wants to learn K3, he must have learned K1 and K2. Also, if he wants to learn a knowledge cluster, he must have learned all knowledge points of the prerequi- site cluster and passed the test. 3.2.3 Personalized Presentation of Knowledge Choice of the knowledge point determines the learning contents be learned, but each knowledge point has several kinds of manifestations. How to choose appropriate learning content is the key of knowledge point’s schedule according to learning envi- ronments and personalized needs. Also, the same knowledge point has several similar contents. The system can find the best personalized presentation according to the knowledge attributes and personalized characteristics of the learner. The method is explained as the following. Supposing the learning style of the learner is that: S: field dependent (0.6), field independent (0.4). auditory (0.4), visual (0.3), kines- thetic (0.1) The selective learning object attributes of choosing knowledge is: LO1: field dependent (0.2), field independent (0.8), auditory (0.4), visual (0.3), kinesthetic (0.3) LO2: field dependent (0.7), field independent (0.3), auditory (0.2), visual (0.5), kinesthetic (0.3) LO3: field dependent (0.9), field independent (0.1), auditory (0.5), visual (0.2), kinesthetic (0.3) The system can choose the knowledge point in the similar learning object, like LO1, LO2, LO3, according to the relativity of learning objects’ attributes and learning style. The calculation is expression (1). ∑ ∑ = = −− = 1 22 1 i )()( i ii i i yyxx yx ρ (1) ρ=0 is irrelevant, ρ=1 is complete correlation. x , y means the average of stat about x i and y i . According to calculate, LO3 is the most appropriate learning object in the example. . Knowledge Points 3.2.1 Choosing the Knowledge Points To choose knowledge points is the key of personalized scheduling of learning content, which is mainly choosing the learning knowledge points on. personalization in web-based learning environment including learners’ styles, inferring engine, learning content library and result analysis, and so on. Nobuko [7] studied student model, in which. points. Knowledge tree: In knowledge system, all knowledge clusters is in terms of certain strategy or sequence, for instance, instruction sequence is called the knowledge tree. In learning