The first works on the field of CF were published in the early 1990s. Goldberg et al. (1992) presented the Tapestry system that used collaborative filtering to filter mails simultaneously from several mailing lists, based on the opinion of other users on the readings. Resnick et al. (1994) described the GroupLens system that was one of the pioneer applications of the field where users could rate articleson a 1–5 scale after having read them and were then offered suggestions. Breese et al. (1998) divided the underlying techniques of predicting user preferences into two main groups. Memory based approaches operate on the entire database of ratings collected by the vendor or service supplier. On the other hand, model-based approaches use the database to estimate or learn a model and then apply this model for prediction.
Collaborative filtering (CF), which makes use of only past user activities (for example, transaction history or user satisfaction expressed in ratings), is usually more feasible. CF approaches can be applied to recommender systems independently of the domain. CF algorithms identify relationships between users and items, and make associations using this information to predict user preferences.
Sherry (2000) claims that Computer-Supported Collaborative Learning (CSCL) is an emerging paradigm of education that emphasizes a melding between individual cognitions and socially shared representations, developed through ongoing discourse and joint activities that take place within a learning community.
Based on the work of Koschmann (2002), CSCL focuses on meaning and the practice of making meaning in the context of joint activity, “and the ways in which these practices are mediated through designed artifacts”
The community learns from its individual participants, and each individual learns from the community: powered by the engine of cognitively engaging discussion, a functioning, dynamic learning system is created. Typically, CSCL environments incorporate CMC tools to facilitate communication among all participants for the purposes of developing shared knowledge and understandings..
According to Sherry (2000), one of the contributing foundations of CSCL is based on the concept of situated learning (Lave & Wenger, 1991). This, and other related perspectives (e.g., communities of learners, cognitive apprenticeships) challenges traditional views on the nature of learning and cognition, and instructional design, to move “beyond the individual mind to include learning that is built up by mediated conversations among members of peer groups, local learning communities, and broader cultural systems” (Sherry, 2000, p. 21). CSCL environments often represent a process of combining particular pedagogical assumptions and then operationalizing these with the use of computer mediated communication tools. This approach demands a reconceptualization of learning from individual acquisition of knowledge and skills to learning as individual and collective engagement (Jones, Valdez, Nowakowski, & Rasmussen, 1995), with relevant activities and authentic tasks embedded within and contiguous with the larger online learning community or environment.
An analysis of the learning processes in a CSCL environment has been conducted using both qualitative and quantitative approaches; however, several studies suggest using a mixed methods approach (Henri, 1991; Hara, 2000; Lally & DeLatt, 2003; Martinez, de la Fuente & Dimitriadis, 2003; Daradoumis, Martinez & Xhafa, 2004; Pozzi, Manca, Persico & Sarti, 2007).
Utilizing a mixed methods approach allows the researcher to collect multiple forms of data. Garrison and Anderson (2003) use a three dimensional model to investigate learning processes in distance education.Their model focuses on social presence, cognitive presence and teaching presence. Text analysis in computer-mediated communication can track specific indicators of social, cognitive and teaching presence (Garrison & Anderson, 2003). Pozzi, Manca, Persico and Sarti (2007) propose a five dimension model for the study of learning processes in a collaborative environment. Their five dimensions include the following: Participative dimension, Interactive dimension, Social dimension, Cognitive and Meta-cognitive dimension, and the Teaching dimension (Pozzi et al, 2007). Within each of these dimensions, specific indicators are defined. Within the participative dimension the following categories are established:
Indicators of active participation: which include the number of messages sent by individual participants, the number of documents uploaded, the number of chat sessions attended, etc;
Indicators of passive participation: which include the number of messages read, the number of documents downloaded, etc; Indicators of continuity, that is the distribution of participation along time (Pozzi et al, 2007, p. 172-173).
The interactive dimension uses content analysis of messages and documents shared by students: Passive participation before posting, that is the number of relevant messages read by a student before posting his/her own, the number of documents downloaded before posting, etc. References to other students’ messages, that is the number of answers to other students’ messages, the number of implicit or explicit citations of other students’ messages, etc. Consideration of other students’ contributions in products, that is qualitative analysis of students’ messages and documents with the aim of finding references to others’ messages or documents (Pozzi et al, 2007, p. 173).
The social dimension is investigated through the identification of “cues that testify to affection and cohesiveness within communication acts (Pozzi et al, 2007, p. 173). The following are cited as indicators in the social dimension: “Thematic units characterized by Affection, that includes expression of emotions, expression of intimacy, presentation of personal anecdotes. Thematic units characterized by Cohesiveness, that include vocatives, references to the group using inclusive pronouns, phatics, salutations” (p. 173). Indicators for the cognitive and meta-cognitive dimension also include thematic units: “Revelation, that is recognizing a problem, showing a sense of puzzlement, explaining or presenting a point of view; Exploration, that is expressing agreement/disagreement, sharing ideas and information, brainstorming, negotiating, exploring; Integration, that is connecting ideas, making synthesis, creating solutions; Resolution, that is real-life applications, testing solutions” (Pozzi et al, 2007, p. 173).
Pozzi, Manca, Persico and Sarti (2007) include the following indicators for teaching presence in their five dimensions model:
Thematic units containing direct instruction, that is presenting contents, proposing activities, diagnosing misconceptions, confirming understanding through assessment and explanatory feedback; Thematic units aimed at facilitating discourse, that is identifying areas of agreement/disagreement in order to achieve consensus, encouraging, acknowledging or reinforcing participant contribution, setting the climate for learning; Thematic units addressing organizational matters, that is introducing topics, planning the course, explaining methods, reminding students of deadlines (p. 174).
Using all or several of the dimensions proposed by Pozzi, Manca, Persico, and Sarti (2007),qualitative and quantitative data can be collected to evaluate the learning processes in a computer-supported collaborative learning environment.