Mining Patterns With Durations from E-commerce Dataset

Mohamad Kanaan 1 Hamamache Kheddouci 1
1 GOAL - Graphes, AlgOrithmes et AppLications
LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
Abstract : Given a dataset of clickstream extracted from e-commerce logs, can we find a clear usage of the website? Are there hidden relationships between the purchased products? Are there any discriminatory behaviors leading to the purchase? To answer these questions, we propose in this paper a new Sequential Event Pattern Mining algorithm (SEPM). The endeavor is to mine clickstream data in order to extract and analyze useful sequential patterns of clicks. Also, in order to make these patterns clearer, the time spent on each page is taken into account. SEPM maintains the items durations during the mining process and extracts patterns with the average durations of these items without multiple scans of the dataset. Our experimental results on both real and synthetic datasets indicate that SEPM is efficient and scalable.
Document type :
Conference papers
Complete list of metadatas

Cited literature [12 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01960321
Contributor : Mohamad Kanaan <>
Submitted on : Wednesday, December 19, 2018 - 12:44:50 PM
Last modification on : Saturday, April 13, 2019 - 1:23:25 AM
Long-term archiving on : Wednesday, March 20, 2019 - 7:07:14 PM

File

Mining_Patterns_With_Durations...
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01960321, version 1

Citation

Mohamad Kanaan, Hamamache Kheddouci. Mining Patterns With Durations from E-commerce Dataset. Complex Network, Dec 2018, Cambridge, United Kingdom. pp.603-6015. ⟨hal-01960321⟩

Share

Metrics

Record views

35

Files downloads

127