Dances With Thy Users Automatic Personalization Through Web Usage Mining ShuiLung Chuang Jan' 20, 20

1
Dances With Thy UsersAutomatic Personalization
Through Web Usage MiningShui-Lung ChuangJan.
20, 2000

• A Case Study A Web Page Recommendation System

2
Tendency of Personalization

• Web Evolves
• Individual User ? Personal Agent
• Information Filtering
• Browsing Assistant
• Purchasing Agent
• you name it!
• Web Site
• Passive Brochure Model ? Active eBusiness Model
• Dynamic Content
• Usage Analysis
• Goal turn users to customers

3
Case A Page Recommendation System
4
Case A Page Recommendation System
5
Case A Page Recommendation System
Profile User
Classify User
Drive Service
User Model
6
Knowledge Discovery and Data Mining

• Knowledge Discovery from Data
• A non-trivial process of identifying valid,
  novel, potentially useful, and ultimately
  understandable patterns in data. (Fayyad, 1996)
• Data Mining
• Applying data analysis and discovery algorithms
  that, under acceptable computational efficiency
  limitations, produce patterns (or models)
  over the data. (Fayyad, 1996)
• Classification, Regression, Clustering,
  Summarization, Dependency Modeling, Change and
  Deviation Detection

7
Road to Knowledge Discovery

8
Example Detecting Credit-Card Fraud
9
Web Mining

• Data on the Web
• Content and Hyperlink Structure ? Web Content
  Mining
• User Behavior (Web Log Packages) ? Web Usage
  Mining
• Web Content Mining
• Thesaurus, Language Model, Knowledge Extraction,
  
• Web Usage Mining
• Discovery and Analysis of User Access Patterns
• 45 of visitors who accessed page A also purchase
  product 1.
• 60 of clients who purchased A also purchased B
  in page product.html within 15 days.

10
Case A Page Recommendation System (cont.)
Server Log
User Transactions
Usage Mining Stage
affafaksaflas Fjaslkfjsjhjha Ksalfasfkjash Faklf
afksdf fsdklfasklfajf
Data Preparation
-----------------

Clustering
Suggestion Engine
?
?
User Model
Web Server
Client Browser
11
Theme One Profile User

• Acquire Users Behavior on the Site
• Impediments
• Local Cache
• Proxy
• Approach
• Cache Busting
• Cookies
• Embedded Agent
• Dynamic Content
• Reexamine HTTP Protocol

12
Theme Two Data Preparation

• Prepare the Data for Mining Algorithm
• Approach
• Data Cleaning
• User Identification
• Session Identification
• Path Completion
• Transaction Identification

13
Theme Two Data Preparation

• Prepare the Data for Mining Algorithm
• Approach
• Data Cleaning
• User Identification
• Session Identification
• Path Completion
• Transaction Identification

14
Theme Two Data Preparation

• Prepare the Data for Mining Algorithm
• Approach
• Data Cleaning
• User Identification
• Session Identification
• Path Completion
• Transaction Identification

1. index, research, advisor, projects,
publication 2. index, bulletin, news 3. index,
people, advisor, student
15
Theme Three Mining Algorithm
Clustering
1. lt1,1,1,1,0,0,0,0gt 2. lt0,0,0,0,1,1,0,0gt 3.
lt0,1,0,0,0,0,1,1gt
1. index, research, advisor, projects,
publication 2. index, bulletin, news 3. index,
people, advisor, student
16
Theme Four Put to Praxis
?
Suggestion Engine
ltindex, researchgt
Web Server
?
Client Browser
17
Summary Conclusion

• A Toy Page Recommendation System
• A General Approach to Web Usage Mining
• Profile User server log, sniff,
• Preprocess Data heuristics on browsing
  behavior, site topology,
• Mine from Data basic statistic, sophisticated
  analysis,
• Put to Praxis find commercial opportunity,
  refine web design,
• Other Arguables
• Designers View ? Users View
• User Privacy ? Site Owners Hunger for User
  Information

18
References

• Cooley, R., Mobasher, B., Srivastava, J.,
  Automatic Personalization Based on Web Usage
  Mining.
• Cooley, R., Mobasher, B., and Srivastava, J.,
  1999, Data Preparation for Mining World Wide Web
  Browsing Patterns, in Journal of Knowledge and
  Information Systems.
• Cooley, R., Mobasher, B., and Srivastava, J.,
  1997, Web Mining Information and Pattern
  Discovery on the World Wide Web, in International
  Conference on Tools with Artificial Intelligence,
  p. 558-567, Newport Beach, CA.
• Fayyad, U., Piatetsky-Shapiro, G. Smyth, P.,
  1996, Knowledge Discovery and Data Mining
  Towards a Unifying Framework, in Proceedings of
  2nd International Conference on Knowledge
  Discovery and Data Mining (KDD-96).
• Pitkow, J., 1997, In Search of Reliable Usage
  Data on the WWW, in 6th International World Wide
  Web Conference, p. 451-463. (http//decweb.ethz.ch
  /WWW6/Technical/Paper126/Paper126.html).
• Spiliopoulou, M., 1999, Tutorial Data Mining for
  the Web, in PKDD99. (http//www.wiwi.hu-berlin.de
  /myra).