An Extraction, Transformation, and Loading Tool Applied to a Fuzzy Data Mining System

Introduction

A Data Warehouse (DW) is defined as “a subject-oriented, integrated, time-variant, non-volatile collection of data in support of management’s decision-making process” (Inmon, 2005). Data is extracted from the sources and then loaded into the DW using various data loaders and Extraction, Transformation and Loading (ETL) tools. We can define Data Mining (DM) as the process of extracting interesting information from the data stored in databases. According to (Frawley et al, 1991), a discovered knowledge is interesting when it is novel, potentially useful and non-trivial to compute. A series of new functionalities there exists in DM, which reaffirms that it is an independent area (Frawley et al, 1991): high-level language on the discovered knowledge and for showing the results of the user's information requests (e.g. queries); efficiency on large amounts of data; handling of different types of data; etc. There is a symbiotic relationship between the activity of DM and the DW. The DW sets the stage for effective DM. DM can be done where there is no DW, but the DW greatly improves the chances of success in DM (Wang, 2009; Inmon, 1996). The World Wide Web (WWW) has become an important resource of information for the DM process. Consequently, the integration of the WWW information into a DW is important in order to get a more effective DM.

One of the most complex issues about the integration and transformation interface is the case where there are multiple sources for a single element of data in the DW. Our proposal is to integrate semantically heterogeneous data from various websites with opinions about educational issues in order to obtain a more effective DM on this information. Similar integration problems have already been solved in various platforms of the so-called Web 2.0, where people are encouraged to post reviews or express their opinions on several subjects, such as: education (PlanetRate, 2010), tourism (Booking.com, 2010; eDreams.com, 2010; TripAdvisor.com, 2010), etc., using numerical values and/or natural language (forums, news groups, etc.). The general approach of these websites is to compute only the accurate numerical information given by users in order to provide a ranking value (e.g. see Figure 1). However, the opinions expressed by the users in natural language are an important source of information. Therefore, the overall problem is the integration of information collected in these questionnaires which are available on various websites and formats, including also linguistic information.

Figure 1.

Example of rating on education in http://www.planetrate.com/category/education

Many aspects of different activities in the real world cannot be assessed in a quantitative form, but rather in a qualitative one (i.e., with vague or imprecise knowledge). In these cases, a better approach may be to use linguistic assessments instead of numerical values. The fuzzy linguistic approach, which was introduced by (Zadeh, 1975), is a theory that facilitates the coding of human knowledge in the form of linguistic concepts, and proposes a tool for modelling qualitative information in a problem. Consequently, the fuzzy linguistic approach seems to be an appropriate framework for solving our problem.