NBU ADVANCED e-LEARNING SYSTEM

Petar Atanasov

New Bulgarian University, Montevideo 21 Street, Sofia, Bulgaria

Keywords: e-Learning, databases, OLAP, data warehouse, grid architecture, web services.

Abstract: This paper introduces the design and implementation of information solution especially designed for the

purposes of e-Learning at New Bulgarian University. The described architecture combines the best of

modern technologies and theoretical foundations. In addition is examined the close future plans on

researching for creation of data repository with semantic file system and its relation with the system.

1 INTRODUCTION

Databases are commonly used for storing structured

data and for analysis of its content. They present the

most modern manner of persisting and manipulating

information. Most e-Learning systems (e.g. Moodle,

A-tutor, OLAT) use relational database systems

(RDBMS) when handling that process such as My

SQL, PostregSQL, Oracle, MS SQL Server. Each of

these systems allows in different stage the

organization and the systematical approach (ad hock

queries, stored procedures, views, materialized

views) in data manipulation as well as optimal disk

storage and security.

At the present moment this is the best way of

working with large amounts of data, especially in the

large fast growing field of e-Learning and especially

in assessment, where the amount of content and

active users is vast. The structure of the data itself is

with critical importance for the correct functioning,

ease collection and responsive interaction with the

system. In the process of e-Learning as a whole the

process of testing and assesment is using most

actively the database resources provided by the

particular RDBMS.

The modern need of handling large data amount

activities leads to cross various combinations and

techniques in definitely not so close fields like

OLAP database architectures and e-Learning.

There are several, reported as successful,

projects that combine OLAP technologies and use

them as powerful data analysis tool.

iSUS (University of Manchester, 2006) uses the

technology in the process of classifying items for

tests. Some metrics of the product:

1. manages over 8000 learning events for

more than 140 students

2. classified more than 24 000 qualitative and

more than 2 000 quantitative test elements

3. perspective direct and aggregated view over

data for objects like (again classified according

to the product specification), Student, Teacher,

Manager.

In others such as the system (Lay, 2006) from

the presentation of Mhairi McApline form Scottish

Qualifications Authority (SQA), who presents

complex set of programatical methods implemented

in services and business operations, the foundations

of the system are based in centralized data

repository.

On the other hand several e-Learning entreprise

companies had developed OLAP compatible

systems in addition to their product lines. For

example such a tool is Docent Analytics (Docent,

2006) which gives the opportunity to analyze trends

in education, certification, compatibility, partnership

channels, sales and etc.

All projects present independent approaches

toward solving the problem of manipulating and

analyzing large amounts of data for the purposes of

e-Learning. They influenced strongly the solution

designed and implemented at New Bulgarian

University, but still does not contain the key of

mastering the solution for the concrete goals and

further research activities, but rather are milestones

showing that the direction is right.

In addition a very interesting application of the

OLAP technology can be found in the process of

generating tests, classifying items and calibrating

251

Atanasov P. (2008).

NBU ADVANCED e-LEARNING SYSTEM.

In Proceedings of the Third International Conference on Software and Data Technologies - ISDM/ABF, pages 251-254

DOI: 10.5220/0001867202510254

 SciTePress

them as well as in presenting reports for test

activities and results.

2 OVERVIEW

The solution is composed of system framework and

four independend but integrated parts, grouped in

two major segments according to their purpose:

Figure 1: Segment A.

Segment A is covering the area of everyday

activities: it includes LMS installation integrated

with external database containing user accounts. The

additional server on the figure is serving backup role

and in case of failure can recover completely the

working process.

Figure 2: Segment B.

The foundation of Segment B is data analysis

(using OLAP technology). The main purpose is

processing assessment data for prognosis,

simulations and further testing of different learning

styles as well as observing student behaviour.

The framework purpose is synthesized in the

communication of both segments. It collects user

data from Segment A which is than aggregated in

the OLAP database in Segment B.

2.1 Background

At project initiation time NBU was using three

separate web platforms wich were partially serving

the e-Learning activities, both for students and

teachers. Yet no integration between the systems,

nor abilities of sharing data, resources and activities

were met. In addition there were no feature in any of

the systems to continuously monitor and analyze

data, user activities and etc.

The needs of the modern times and the new

challenges of e-Learning in nowadays had shaped

the requirements into two separated but completely

compatible systems and the framework beneath

them: LMS for handling activities in everyday work

and analytical system for collecting, analysing and

reporting:

2.2 Design Concepts

The design policy is based on the best practices and

theoretical basis in the software engineering and is

completely focused on the two main streams of the

solution and its supportive framework. The main

concepts are as follows:

 modularity

 maintenance

 security

 customizations

 use design patterns in development practice

 scalability

 framework settlement

 open source

2.3 Major Objectives

There are two main objectives that shape the

solution: First to provide solid base for everyday

activities and than after successful deployment of the

first stage, as second sequential phase, to develop an

advanced analysis and reporting system for the

resultant data of the first stage.

 provide LMS environment for completely

serving various activities

 provide homogeneous approach in

customizations and maintenance of the

working environment

 provide high availability and security

 provide homogeneous approach in managing

users

 provide service oriented framework

 provide analytical features

 provide ability for continuous monitoring in

acceptable way of the user activities and

testing results

2.4 Physical Structure of the Solution

Containing four physically independend servers

responsible for management of the e-Learning

process, the e-Learning content, management of

ICSOFT 2008 - International Conference on Software and Data Technologies

252

users, system backup and archive, and deployment

of particular customization and modules source

code. The servers are segregated according to its

major priorities in the following manner:

 Working server – used for everyday work

 Staging server – pre-release with

customizations after the successful

deployment of particular features

 Deployment server – responsible mainly for

the development and customization of the

working installation

 Archive server – contains backups and

archives of the working installation, as well as

complete versioning of particular modules and

services

2.5 Organizational Structure of the

Solution

 Separate installation of LMS Moodle (latest

stable version) used for everyday work

 Separate (legacy) system with users integrated

with the working Moodle installation

 Development area – customizations, bug fixes

and tests of new features with latest archived

working copy as playground

 Regular backup of the working installation –

with on demand feature as well as regularly

scheduled timed backups of the database and

the external data. The module is purposed as

instant replacement and recovery of the

working version.

2.6 Services Module

 User management- handle, secure,

import/export, etc.

 Searches

 Dynamic ranking of searched results and

categorization algorithms

 Data tier (for various data sources: relational

database including pilot module with semantic

file system)

 e-Learning content presentation

 OLAP Toolbox.

 Separate reporting module – user activity and

others (e.g. test results)

2.7 Comparison of Different OLAP

Strategies

The different strategies when designig OLAP

solutions shape behaviour not only for the various

products on the market but also sketch the different

purposes and goals of the respective information

systems. At present moment 'four main options

dominate' (White, 2003) in designing OLAP

solution. The most conservative approach is when

data is stored in a relational database and than

accessed with SQL statements. Data is structured in

star or snowflake table design and the database is

managed by client or server RDBMS. One of the

best features when using this way is that it avoids

the need of purchasing specialized multidimensional

database product while on the other hand its biggest

fault resides in pour performance and limitations of

traditional SQL. Vendors in the area like Microsoft,

Oracle and IBM, are constantly working on

improvements on SQL analysis power and

performance. This option is supported from any

product that can provide relational view of data.

The second approach is depending hardly on vendor

provided OLAP engine which retrieves data from

the working database and performs more complex

processing on it. This processing is achieved by

using vendor-specific visual tools or by applications

that execute OLAP language statements through a

provided API. There are couple of options for the

OLAP engine: it can reside in the same operating

platform as the RDBMS, it may be integrated with

the RDBMS or may be presented as middle-tier

server in a three-tier architecture. Solutions based on

this option are provided by Applix, IBM, Microsoft,

MicroStrategy and Oracle.

The third way offers the ability of storing data in a

multidimensional database (as arrays or cubes) and

manipulate it using queries and OLAP visual tools.

As in the first approach the database may be

managed by client- or server-based

multidimensional DBMS (MDBMS). The reason

why this approach is popular is because the

MDBMS can be optimized for OLAP, which leads

to good performance particularly in array processing

large amounts of memory. But this still not means

that the approach is scalable to processing large

amounts of data an large user activities which results

in using this approach for manipulation of

summarized data. Key vendors in this area include

Applix, Cognos, Comshare, Hyperion, Microsoft,

Oracle and SAS.

The last way in designing OLAP solutions is to store

small amounts of data in files in user computers and

than access it with the OLAP engine. Usually this

data is gathered from RDBMS or MDBMS. At the

present moment this option is really popular, further

more the growth in the use of Web-based thin clients

indicates that the business is exploring the

opportunity to move client OLAP processing

NBU ADVANCED e-LEARNING SYSTEM

253

(described in the third option) to Web-based servers.

Vendors working in the area of the fourth approach

are Business Objects and Cognos.

Each option has its own positive and negative sides,

so the successive OLAP solution will combine

different features from the various approaches.

Alternatve method would be the integration of the

OLAP solution with pilot module with semantic file

system. The succesive result would lead to variuos

possibilities when collecting data not only from the

current chosen LMS (Moodle) but actually from any

other giving the oportunity to process and analyse

different data sources.

3 RESULTS

The HP servers are operating on Gentoo Linux

(Gentoo 2007) with PHP 5 and MySQL 5 (standart

edition) as database server. We are expecting future

migration to My SQL Enterprise Edition as the need

of more data manipulation arises. Currently we

server about 5 000 users with almost 3GB user data

stored in the database and additional 10GB of data

stored in the file system and served by the LMS

Moodle.

The working server is operating with Moodle 1.7

with integrated module (both using the Moodle

coding style and another with completely external

method) for user import and verification.

Currently our plans are focused on establishing

framework for handle and manipulation of large

volumes of data and the process of reviewing,

reporting and describing the university processes

and activities as a whole. In order to sketch the

contours of the big picture we evolved to initiating

the development of OLAP module closely coupled

with the chosen working LMS (Moodle) and

implementation of data warehouse solution.

The OLAP toolbox will represent subsystem

with two main branches: data visualization module

and services for aggregating, collecting and feeding

the reporting module.

In addition we plan implementation of a data

repository with the particular goal of organizing and

manipulating bank of items. Again this module as

well all others in the system is designed in service

oriented manner.

Finally we are discussing and exploring the

possibilities of integration with other modern e-

Learning systems as A-tutor and OLAT, keeping in

mind that e-Learning still doesn't mean Moodle,

although so wide spread and well accepted around

the world

4 CONCLUSIONS

This paper has presented the conceptual basis and

the implementation status of the NBU's university e-

Learning system. The theory and practices behind

the solution are inspired form the most advanced

modern techniques, which are very new to the e-

Learning in Bulgaria.

The homogeneous approach implemented in grid

organized with operational LMS and the subsequent

analyzing system are in harmony with the latest

tendencies in modern technology and wich is more

importeant serve better its purposes.

REFERENCES

Docent Analytics, 2006

iSUS, University of Manchester, 2006

Lay S., (2006), QTI and Item Banking

White C., (2003), OLAP in the Database, Intelligent

Business Strategies, DM Review Magazine

ICSOFT 2008 - International Conference on Software and Data Technologies

254