Copyright © All rights reserved micansinfotech.com
The unprecedented growth of the Internet, its pervasive accessibility and ease of use has increased students’ dependencies on the Web for quick search and retrieval of learning resources. However, current search engines tend to rely on the correct keywords. This excludes other characteristics such as the individual’s learning capability and readiness for specific learning materials. As a result, the same set of search-keywords deliver the same search results. This situation hinders the optimisation of the Web search engines in supporting the heterogeneity of its users in their learning endeavours. This paper aims to address the issue. It attempts to augment Web search engines with personalised recommendations of search results which match students’ learning competencies and behaviours. The results drawn from our experiments suggest that our novel approach can provide a notable improvement in terms of performance and satisfaction for the students.
This module aims to construct and maintain the profile of each individual student. It allows the system to understand the different learning needs and capabilities of each individual student. It then uses this information to improve the relevancy of the returned Web search results by selecting the most relevant personalised links. This is achieved by prioritising the links according to each profile. To accomplish this, the module has two functional components: Academic Record Analyser and Behavioural Activity Analyser.
The academic record is used to measure the individual student’s past and present academic performance. This information is derived from the Student MIS. The system also observes the student’s learning behavioural activities through his/her browsing histories and session data.
The Academic Record Analyzer module is responsible for identifying the signed-in students by retrieving their profiles from the Student MIS. It mainly retrieves students’ previous and current academic information and then stores the information in the local server for further processing. Subsequently, the module calculates the standard T-Scores extracted from the raw scores of the academic records for each student using the grading policy used by the University of Texas at Austin .
The T-Score is one form of standardised test statistics that transform individual raw scores into standardised forms of scores for ease of comparison. It provides a constant of the mean and standard deviation on any set of data . Furthermore, the T- Score also reduces natural variations that occur within grade points thereby rendering a way to ascertain whether the scores are high or low. By averaging the calculated standard T-Scores, an average score is achieved for each student. This helps to classify the students’ profiles. In this study, we define it as Knowledge Point (KP).
The Behavioural Activity Analyser module is responsible for continuously monitoring and capturing the students’ learning behaviours. This is accomplished through their Web search activities while they use the proposed system. Each student’s learning activities are recorded and stored individually as extracted from their browsing histories and session logs.
All information is stored inside the local server. The data captured comprises the number of times they login, the number of search per login, the issued queries, the selected documents, page names, page sizes, average scrolls, links clicked, and time spent. Subsequent to that, the stored activities are analysed so as to classify the students’ level of interest towards learning while searching for the relevant learning materials from the Web search engine.
From the earlier sections, the student’s knowledge point (KP) and level of interest are obtained. These outputs require further processing so as to enable every student’s profiles to be classified. The students’ profiles are classified according to their academic performance (knowledge point) and learning behaviours (level of interest towards learning). In order to achieve this, the extended classification rule is applied
the decision tree is translated into the following rules by using the student’s Knowledge Point (KP) and interest level: 1) Students with KP higher than 80 will be assigned a Master Class. 2) Students with KP less than 63 will be assigned a Beginner Class. 3) Students with KP<=80, >=63 are classified
a Web search engine having personalised features can return different search results for dissimilar individuals. Alternatively, it can match the search outcomes in a different way according to the user’s intent based on the user’s information needs and preferences . However, it is very challenging to collect user’s data which should be rich enough that can understand the user’s precise needs and preferences.
One way to address this challenging task is by combining related data gathered from other individuals with similar profiles when building personalisation features. In the ‘Groupization’ technique, personalisation is established by assigning higher weights to pages that are apposite with the predominance of the members of the group. This is achieved by harmonising with every group member’s document term frequencies and Web histories
Keywords are required in making a web search and it appears that learners or students are given the same keywords when making their search through any major commercial search engines such as Google, Yahoo or Bing). Without doubt, this process is likely to return the same set of search results for all the students in the group regardless of their learning proficiencies.
Moreover, it was identified that when a dominant search engine was used to fetch educational materials, four results were found profoundly educative among the topmost 50 returned documents . However, the relevancy of the results mainly relies on the right combination of keywords used in the query. Many students, especially novice learners often struggle with finding the right keywords particularly when they are new to the learning topics
we present a personalised group-based recommendation approach for Web search in e-learning. The primary motivation of this study is to present an adaptive e-learning method for students of different learning capabilities when using the popular search engines.
To achieve this, a Web search recommender system was developed as a gateway between the Google search engine and the institutional e-learning portal so as to enable the search engine to deliver personalised search results as recommendations for students based on their individual needs.
Although search engines have been extensively used by the students for educational purposes, they deliver similar contents regardless of students’ profiles. This is not beneficial to the students because the same contents may not meet the requirements of every student. Web search results that are personalised to a student’s learning profile is therefore, necessary.
In this study, a personalised group- based recommendation approach for Web search in e- learning was proposed. We designed an adaptive system that comprises dynamic profiling and content re-ranking mechanisms that will cater to students in finding Web- based learning materials based on their academic records and learning behaviours. The proposed system augments the Google search engine with the ability to recommend and prioritise the top five most suitable links to students depending on their personal profiles.