Graduate School Recommender System: Assisting Admission Seekers to

Apply for Graduate Studies in Appropriate Graduate Schools

Department of Computer Science and Engineering 1University of Dhaka, 3Primeasia University, 2,4 Bangladesh University of Engineering and Technology

Mahamudul Hasan1, Shibbir Ahmed2, Deen Md. Abdullah3, and Md. Shamimur Rahman4

Introduction

To recommend accurately list of universities to apply for

graduate admission abroad considering applicant-applicant

similarity

Objective

To assist the graduate admission seekers in order to

apply for graduate admission to pursue higher study abroad

with funding

To develop a technique of using academic records of

successful applicants for making graduate school

recommender system

Methodologies

Flow Chart for Grad School

Recommender System

Design & Implementation

Experimental Evaluation & Analysis

Based on the experimental

analysis, if the number of

recommended universities

(N) is increased, the

accuracy is also increased.

Change of accuracy due to variation

of the training and test data

Top-N Recommended Universities vs. Accuracy

Conclusions

Our proposed recommender system will recommend list of

universities to applicants trying to pursue higher study abroad

Eventually assisting them to apply for graduate admission

in appropriate universities with best possible financial support

In the entire process of getting opportunity of graduate

studies university selection is the most crucial step for

applying to graduate admission

Knowledge discovery from the academic records of

successful graduate applicants is very important for the

graduate admission seekers in foreign institutions in respect

of choosing appropriate higher educational institute

5th International Conference on Informatics, Electronics & Vision (ICIEV)

13-14 May, 2016, Dhaka, Bangladesh

Compute a weighted

score from prior

information of successful

applicants such as

undergrad CGPA, GRE,

TOEFL Scores etc.

Calculate similarity

between weighted scores

using mean squared

deviation similarity metric

Test

Data Set

Recommend Top K

Universities to the test

users from N similar users

Training

Data Set

Compute a weighted

score from provided

information of current

applicants such as

undergrad CGPA, GRE,

TOEFL Scores etc.

Calculate top N similar users for

the Test users using K-nearest

neighbor algorithm

Recommend List of

Universities to users to apply

for graduate admission

Grad School

Recommender System

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Ac

cu

rac

y

Top-N Recommended Universities

Accuracy

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

1

Ac

cu

rac

y

Percentage of training data (%)

Accuracy Change due to Variationof Training & Test Data

Based on the experimental

analysis, applicants get

admission into universities from

our recommended universities

in 65-70% cases of several training-testing dataset

Every time for each test

applicant’s computed weighted

score is considered with the

similar weighted score of those

applicants from training set, thus

we have recommended Top-N

universities (for our

experiment, N=10 is considered)