WOLKITE UNIVERSITY WEB USERS’ NAVIGATIONAL PATTERN DISCOVERY AND ANALYSIS

Abstract

Wolkite University has recently experienced a decline in internet connection speed and inconsistency. The university is investing money on information communications technology infrastructure, services, and telecom services to assist in the achievement of the mission of the University. The objective of this research work is to discover Wolkite University internet users’ navigational patterns using statistical and association rules mining techniques by considering two-time situations. Three months of WKU proxy server access log is collected from Meskerem 28/2016 (October 9/2023) to Tahsas 30/2016 (January 9/2024). In the data preparation phase, the python program is used to parse and convert the plaintext to csv file and remove unnecessary entries that are irrelevant to this study’s objectives. Users are categorized as staffs and students based on the design of WKU networkdesign. Therefore, VLANs information is used to categorize the users. There are two-time situations in universities that is class time and non-class time for students and office and non-office time for staffs. Statistical data analysis and Frequent Pattern growth (FP-growth)algorithm are applied on each dataset to discover the most frequently accessed websites and websites that are visited together. The result of the data analytics showed that the entertainment, educational and social media websites are accessed most frequently by students during the class and non-class time. the educational, entertainment and social media websites are most frequently visited websites by staffs during the office and non-office time.The associations rules discovered from these datasets showed that (openai, paperpass and sci-hub) and (tutorials point, tiktok, YouTube and Facebook) are websites that are visited together by students during the class time with confidence values 86.46% and 84.11%respectively. (savefrom, youtube and tiktok), (facebook, crazygames and bgames) and (tactic, xnxx and pornhub) are discovered rules from non-class student dataset with confidence values 100%, 97.29% and 85.42% respectively. The result of (FP-growth)algorithm showed that (qwickbet, betika and flashscore), (scopus, elsevier and openai) and(linkedin, ethiojobs, youtube and facebook) are the websites accessed together by staffs during the office time with confidence values 96.97%, 95.18% and 94.18% respectively. And (savefrom, tiktok and youtube), (ethiopianrepoterjobs, etcareers and ethiojobs) and(facebook, tiktok and youtube) are the interesting patterns from the staff non-office time dataset with confidence values 98.53%, 97.51% and respectively