Dove e quando
2021-06-14
Web usage mining on e-commerce websites
Grażyna Suchacka – University of Opole
Duration: 8 hours
When: 14 – 16 June, 2021
Where: Microsoft Teams
PhD credits (DIBRIS metric): 2
PLEASE ENROLL HERE (attendance is FREE):
https://docs.google.com/forms/d/1ivAOxr5QpZroTrENtsjEKAg_t5OXfy4nboJLjg…
Abstract
--------
The course deals with the application of machine learning
methods to Web data, in particular in the context of two
research problems: detecting Web bots and predicting purchases
in online stores. The first problem is due to the presence of
artificial agents on the Web which pose a threat to the
website security, privacy, and performance. Continuous
development of artificial agents’ technology makes Web bot
detection, both in the offline and real-time settings, harder
and harder. The second problem is connected with discovering
various user profiles on e-commerce websites and identifying
user sessions with high probability of making a purchase. The
problems under consideration are key issues in the era of the
rapid development of e-commerce, advanced Web-based
technologies, and big data.
Program
-------
(Italian times, CEST)
MONDAY, 14 JUNE 2021 h 14.00 - 16.00
Lecture 1 "Introduction to Web usage mining"
- Web usage data
- Web usage mining in the context of online stores
- Characteristics and differences in bot and human Web traffic
- Data pre-processing for Web usage mining, reconstruction of user sessions
TUESDAY, 15 JUNE 2021 h 11.00 - 13.00
Lecture 2 "Web bot detection, part 1"
- Problems of offline and online bot detection
- Feature selection and feature extraction for bot detection
- Offline bot detection with supervised classification methods
TUESDAY, 15 JUNE 2021 h 14.00 - 16.00
Lecture 3 "Web bot detection, part 2"
- Offline bot detection with unsupervised classification methods
- Online bot detection
WEDNESDAY, 16 JUNE 2021 h 11.00 - 13.00
Lecture 4 "Online purchase prediction"
- Problem of predicting online purchases
- Feature selection for online purchase prediction
- Purchase prediction with machine learning methods