Learning Management Systems (LMSs) offer a lot of methods for the distribution of information and for the communication between the participants on a course. They allow instructors to deliver assignments to the students, produce and publish educational material, prepare assessments and tests, tutor distant classes and activate archive storage, news feeds and students' interaction with multimedia. They also enhance collaborative learning with discussion forums, chats and wikis (Romero et al., 2008a).
Some of the most well-known commercial LMS are Blackboard, Virtual-U, WebCT and TopClass, while Moodle, Ilias, Claroline and aTutor are open source, freely distributed LMSs (Romero et al., 2008a). In Greece, the Greek University Network (GUNet) uses the platform Open eClass (GUNet, 2009), which is an evolution of Claroline (Claroline, 2009). This system is an asynchronous distance education platform which uses Apache as a web server, MySQL as its database server and has been implemented in PHP. Open eClass is open source software under General Public Licence (GPL).
Due to the volume of data, one of the main problems of any LMS is the lack of exploitation of the acquired information. Most of the times, these systems produce reports with statistical data, which, however, don't help instructors to draw useful conclusions either about the course or about the students; they are useful only for administrative purposes of each platform. Moreover, the existing e-learning platforms do not offer concrete tools for the assessment of user actions and course educational content.
Data and web mining
Data mining is the search for relationships and patterns that exist in large databases, but are 'hidden' among the vast amounts of data. It is part of the whole Knowledge Data Discovery (KDD) process. KDD is the complete set of processes for knowledge discovery in databases that aims at the detection of valid information and pattern recognition in raw data (Kantardzic, 2003). The classical KDD process includes 5 phases: data pre-processing, data transformation, data mining, data visualization and data interpretation. The first two phases select and "clean" a given dataset. The next phase, data mining, is essential in the whole KDD process; through it non-trivial patterns in data are found with the use of algorithms. Data mining consists of such tasks as classification, clustering, time series discovery or prediction and association rules mining (Witten and Eibe, 2000).
Web mining (Srivastava et al., 2000) is a sub-category of data mining. Data mining techniques are applied to extract knowledge from web data. There are three main web mining categories from the used data viewpoint: Web content mining, Web structure mining and Web usage mining (Spiliopoulou, 1999; Kosala and Blockeel, 2000; Bing, 2007). Web content mining is the process used to discover useful information from text, image, audio or video data on the web. Web structure mining is the process of using graph theory to analyze the node and connection structure of a web site. Web Usage Mining (WUM) is the application that uses data mining to analyze and discover interesting patterns of user data on the web. The usage data records the user's behavior when he/she browses or makes transactions on the web site. The first web analysis tools simply provided mechanisms to report user activity as recorded in the servers. Using such tools, it is possible to determine such information as the number of accesses to the server, the times or time intervals of visits, as well as the domain names and the URLs of users of the Web server. However, in general, these tools provide little or no analysis of data relationships between the accessed files and directories within the Web space.
Data mining in e-learning
Data mining techniques have been used to discover the sequences patterns of students' web usage after the analysis of log files data (Romero et al, 2007). …