Abstract:

Introduction:
Applying for higher education is a crucial decision that influences future career. Applicants have to deal with lots of information in a restricted time. When inquired, in previous works, 93% of students agreed that a specialized Intelligent Decision Support System (IDSS) can help to make better decisions [1]. Such IDSS should be able to adapt to the constant changing of courses, institutions and other variables. This work intends to design an IDSS based in a universal architecture totally independent of the decision models, i.e., the architecture can automatically adapt to the intrinsic and extrinsic properties of the model. The architecture was tested using Portuguese real data, however it can be adapted to other countries and can be accessed anywhere and anytime.

Objectives:
The main objective of this work is to study the viability and consequently develop and test a Pervasive Architecture which can be used by Higher Education applicants regardless of their country. The architecture should be the base of an IDSS able to support the choice of students in the Higher Education Areas / Courses.

Material & Methods:
To design the architecture it was applied knowledge from different areas: intelligent agents, decision support systems, systems interoperability, pervasiveness, education and psychology. To prove the architecture viability, it was used a decision model framed in the Portuguese reality [2] which combines variables from 10 logical blocks (High School, College, DGES Areas, Social, Academic, Cultural, Sports, Prestige, Vocational (RIASEC) and Personal).

Results:
In the first instance the system developed receives the decision model configurations and according to them, designs the IDSS automatically.
The Architecture is divided into four subsystems: Data Load (Load the decision models), Information Management (Manage the information loaded), Inference (Induce the decision rules) and Interface (Design and present the IDSS).
Intelligent agents are used to extract, transform and loading tasks automatically and an interface will be designed according to the data presented in database. The models used by this architecture are divided into Logical Blocks (LB).

The designed architecture is characterized by:
• Adaptive models;
• Optimized models;
• Online-learning;
• Flexible number of outputs;
• Customization made by the user;
• Acomodation of new variables or LBs.

Additionally, weights are associated to the variables selected by the student in order to adjust the model to the intrinsic characteristics of each student. As a result, this architecture allows for presenting anywhere and anytime a set of courses adequate to the student profile.

Conclusion:
This work proved the viability of having a multi-model platform prepared to accomodate different types of decision models.
This system is also universal because, regardless of the country and the decision model, it can be adopted. Students can access the platform anywhere and anytime.

References:
[1] Silva, João Pedro, Portela, Filipe, Santos, Manuel Filipe (2013) A Decision Support System for Portuguese Higher Education course selection – First Round. KMIS 2013.
[2] João Pedro Silva, Filipe Portela, Manuel Filipe Santos, Maria do Céu Taveira. Intelligent Decision Support for University application using RIASEC codes. INTED 2014. Valência, Spain. IATED. (2014)

Keywords:

Intelligent Decision Support System, Higher Education, University Application Process, RIASEC, Student Support, Multi-Model Architecture.