Transcript A survey of Context-Aware Mobile Computing Research

Using Data Mining Methods
to Build Customer Profiles
Gediminas Adomavicius, Alexander Tuzhilin
New York University, USA
2008.11.10
Summarized & Presented by Jungyeon Yang
IDS Lab., Seoul National University
Contents
 Introduction
 Building Customer Profiles
 Rule Discovery
 Rule Validation
 Validation Operators
 Case Study – The 1:1 Pro System
 Discussion
Copyright  2008 by CEBT
2
Introduction
 Personalization community must deal with

Who customers are, How they behave, How similar they are to
others, How to extract this knowledge
 Customer Profile contains

Facts about a customer

Rules describing that customer’s behavior
 This research is focused on

Rule validation

Implement a validation system
Copyright  2008 by CEBT
3
Building Customer Profiles
 Data model

Two basic types of the data
–
Factual : who the customer is
–
Transactional : what the customer does
 Profile model

A complete customer profile has two parts
–
a factual profile : gender, age, etc.
–
a behavioral profile : customer’s actions, is derived from user’s
transactional data
Rule discovery
Rule validation
Copyright  2008 by CEBT
4
Rule Discovery
 In order to discover rules that describe the behavior

Apriori algorithm for association rule

CART(Classification and Regression Trees) for classification rule
–
Classification Tree : in case of categorical values
–
Regression Tree : in case of continuous values
Copyright  2008 by CEBT
July 18, 2015, Page 5
Rule Validation
 One way to Validate rules is to let a domain expert
inspect rules

There is scalability problem
 Solution of this approach

Uses validation operators that let a expert validate large numbers
of rules at a time with relatively little input from the expert.
Copyright  2008 by CEBT
July 18, 2015, Page 6
Rule Validation (Cont.)
 Collective rule validation lets the expert deal with such
common rules just once.
 The expert choose various validation operators and
applies them successively to the set of rules
 The set of all discovered rules is split into three mutually
disjoint sets

accepted rules(Rall)

rejected rules(Rrej)

possibly some
- until some predefined % of rules
is validated
- until validation operators validate
only a few rules at a time
remaining unvalidated
rules(Runv)
Copyright  2008 by CEBT
July 18, 2015, Page 7
Validation Operators
 Similarity-based rule grouping

This operator puts similar rules into groups according to expertspecified similarity criteria

Ex) according to the attribute structure similarity condition, all rules
that have the same attribute structure are similar
Copyright  2008 by CEBT
8
Validation Operators (Cont.)
 Template-based rule filtering

This operator filters rules that match expert-specified rule templates

The expert specifies accepting and rejecting templates
 Examples


REJECT HEAD = {Store = RiteAid}
–
“Reject all rules that have Store = RiteAid in their heads.”
–
Rule 1 would be reject
ACCEPT BODY ⊇ {Product} AND HEAD {DayOfWeek, Quantity}.
–
“Accept all rules that have the attribute Product (possibly among other
attributes) in their bodies, that also have heads restricted to the
attributes DayOfWeek or Quantity.”
–
Rule 5 & 7 match
Copyright  2008 by CEBT
9
Validation Operators (Cont.)
 Redundant-rule elimination

It eliminates rules that by themselves carry no new information
about a customer’s behavior

Example
–
Product = AppleJuice => Store = GrandUnion (2%, 100%)
–
Assume that the fact “The customer shops only at Grand Union” in
one’s factual profile
–
AppleJuice rule would be eliminated
Copyright  2008 by CEBT
10
Case study – The 1:1 PRO SYSTEM
 Short for “One-to-One Profiling System”
 Profiling and validation system
 Input

The factual and transactional data stored in a DB or files
 architecture
Copyright  2008 by CEBT
11
Case study – The 1:1 PRO SYSTEM
Copyright  2008 by CEBT
12
Disscussion
 “Quality” of generated rules

Different expert has different validation
 Scalability

Attributes ↑
Apriori has bottleneck

User ↑ validation operators should scale up
 Constraint-based rule generation vs. post-analysis
 Examination of groups of rules

Expert can apply validation operators just to particular group of
rules and examine its subgroups
Copyright  2008 by CEBT
13
Opinions
 Pros

Can handle large numbers of rules

Easy to validate rules using GUI

Provide several ways to validate rules using operators
 Cons

Other mining algorithms should be applied

Constraints are needed when rules are generated. (too many)
 In order to use rules for context-aware services, a system which
has intuitive UI and many functionality is necessary.
 A system should have flexibilities to be applied in many domains
Copyright  2008 by CEBT
14