8. My Online Teacher 2.0 Third Prototype 142
8.2. MOT 2.0 Third Implementation 144
The scenario above reflects the goals set for the functionality desired for the third prototype of the MOT 2.0 system. The third prototype is based on the evaluation feedback from the first and second prototypes, as well as on the exploration of new research ideas. Concretely, in the third prototype,fine-grained content-basedandpeer-based adaptations
were introduced. Simplified versions of the procedures used are shown below, for the three strategies of recommended learning items, recommended expert peers, and adaptive user privileges.
Strategy 1. Recommended learning contents (items):
RI = ;
If (
Similarity (I, Ii) > Value_1 && Rating (Ii) > Value_2 &&
knowledge (User, I) == Value_3 )
RI = RI {Ii};
Where:
I:is the current item.
RI:is the set of Recommended Items for the current item, I.
Ii: refers to an arbitrary item in MOT 2.0 (third prototype); where the variable i is the item identifier.
Similarity (I, Ii): is the Euclidean Distance between the current item, and any other item in MOT 2.0 (third prototype). The similarity is calculated between the two strings, with each string representing the tag (keyword) set that describes the item.
Rating (Ii): represents the average rating of the item, as decided by previous users with rating rights.
knowledge (User, I):is the User’s knowledge level for the current item, I.
Value_1:is the threshold value of the similarity, which can be any value between 0 and 1. The MOT 2.0 (third prototype) system is flexible in that the value can be changed every time a module is taught. This value is normally set by the teacher, or course technician.
Value_2:is the value of the average rating threshold, which can be any value from 1 to 5. This value is also set by the teacher or course technician.
Value_3: is the user’s knowledge about a specific item, which can be either 1 or 0. For instance, a learner just starting to learn about the topic of ‘Implicit and Explicit rating’ can be sent back to read more about ‘Collaborative filtering’ (see Figure 38).
Thus, the content-based adaptation recommends another item to a learner, if his knowledge on the current item is insufficient, and there exists another item Ii with an acceptable level of similarity to the current item, and a good rating from peers. Therefore,
this content-based adaptation is fine-grained both in items of input (the current content item, instead of a whole module) and output (a recommended item, and not the whole module). In the following, we introduce a simple version of the peer-based adaptation.
Strategy 2. Recommended expert peers / learners:
RU = 0;
If (knowledge (User, I) == Value)
RU = RU {User};
Where:
I: is the current item.
RU: is the Recommended Users set for the current item, I.
Knowledge (User, I): is the User’s knowledge level for the current item, I.
Value: is the value of the knowledge level, which can be “beginner”, “intermediate”, or “expert”. For instance, an expert learner on the subject of “Uses of Collaborative Filtering” in the course on “Collaborative Filtering” would be recommended as a knowledgeable contact to another learner who is just starting to learn about that subject (see Figure 39).
The strategy simply states that if a user’s knowledge about an item is above a certain threshold, this user will be recommended to all other users. Moreover, the strategy above illustrates a fine-grained recommendation of peers in terms of input (the recommended user’s knowledge of the current item) and in terms of output (the recommended user of the current item). This strategy moves completely away from traditional Adaptive Hypermedia, as it does not focus on classical content recommendation, but on peer recommendation.Next, we describe a simplified version of the third strategy of adaptive user privileges.
Strategy 3. User’s privileges based on the knowledge:
If (knowledge (User, I) == V1) {
CanTag = V3; CanFeedback = V4; }
Where:
I: is the current item.
knowledge (User, I): is the User’s knowledge level of the current item,I.
Value: is the value of the knowledge level, which can be “beginner”, “intermediate”, or “expert”.
V1, V2, V3, V4: can be either 0 or 1 to determine if the user has a privilege or not.
The strategy states that privileges are granted, for each item, to each user depending on the user’s knowledge about that item. Therefore, an expert user can be given all privileges, of rating, tagging and giving feedback, whereas anintermediate user is only allowed to add comment, and a beginneruser is restricted to reading only. Similarly, the adaptation is fine-grained, at the level of the items and not the modules. This strategy moves away from both Adaptive Hypermedia systems and recommender systems, and introduces adaptivity to the Web 2.0 context.
Therefore, the main difference between the strategies presented in this chapter and the previous chapter is the granularity level. Previously, we have used low granularity recommendations, i.e., recommendations based on whole modules and recommendations of whole modules. Here, we concentrate on fine-grained recommendations, i.e., recommendations based on items within a module.
Moreover, whilst we have introduced these three axes of adaptation via their representative separate strategies, one of the main features of the third prototype is the merger of all these three types of adaptation. Therefore, all these types of adaptations are to be applied at the same time, as will further be detailed in the next section.