ECommerce - supporting lehigh cse

Case-Based Reasoning in
E-Commerce
Joe Souto
CSE 435
What is E-Commerce?

“The exchange of information,
goods, or services through
electronic networks”1
How can CBR help?

How many times have you seen this?
How can CBR help?

Or this?
What’s wrong?

Demand is either over-specified or
under-specified

It is up to the user to find what they want

There is no intelligent sales support
We have a problem


Buyer has limited
knowledge of product
base
Seller has limited
knowledge of buyer’s
requirements
”Knowledge Gap”
We have a problem



Knowledge gap is solved in real-life by a
human sales agent as a mediator. We don’t
have this luxury online.
Solution: CBR approach  product
knowledge is stored as experience in a case
base.
Sales agent makes recommendations based
on the stored experience.
Some Preliminary Info


We need a way to define user requirements
Customers buy items in order to satisfy their
desires
Define a customer’s desire as a “Wish”

Wishes have various properties
Individual Wish Properties

Importance


Hard: MUST be met (ie: “vacation for <$2000”)
Soft: not essential, but helpful (ie: “red” car)


Agent must satisfy ALL hard req’s and as many soft as
possible
Precision


Precisely Determined (specific, ie: “>3GHz P4”)
Undetermined (vague, ie: “fast processor”)
Individual Wish Properties

Certainty


Certain
Uncertain

Sales agent must try to increase certainty of wishes
and make recommendations based on them
Overall Wish Properties

Redundancy

Wishes can be redundant



Ex: Computer that’s “fast” and can play Half-Life 2
Agent must recognize and avoid redundant inquiries
Consistency

Wishes can be contradictory


Ex: new Ferrari, and under $1000
Agent must either ask user to clarify, or suggest
products that satisfy one of the two wishes
Product Classifications
How Do These Properties Help?
1.
Customers want a product to satisfy a wish
2.
Products have various properties
3.
Therefore, product properties can be
mapped to the satisfaction of a customer’s
wish
With all that in mind, now we can
look at the transaction process
Transaction Model

Single transaction can be modeled with
three phases
Pre-Sales

Buyer wants a product, Seller provides information

3 Phases
Supplier Search


Product Search


Client determines which supplier can satisfy their wishes
Mapping of customer criteria to products
Negotiation
1.
2.
3.
Price and way of payment
Details of delivery
Regulations about cost and delivery
Pre-Sales



Recall the Google
Example
No “intelligent sales
support”
Burden of knowledge
is in hands of the
customers
Example


Due to Knowledge Gap, Analog Devices
added a CBR system to assist Pre-Sales
Analog Devices:
http://www.analog.com
How Does It Work?


Similarity Metrics!
Similarity function for
single attribute


OK to be under, less similar
if over desired value
The overall similarity
is computed weighted
average of local
similarities.

Remember the “priority”
box
Sales



Product has been chosen, must be configured
and paid for
Customer and Sales Agent negotiate about
product attributes and costs
Intelligent Support is needed for negotiation
Negotiation



“A process where two parties bargain
resources for an intended gain”1
In Sales phase, customers navigate through
products to satisfy their wish.
Some wishes known, others discovered in the
process. Hard wishes must be fulfilled, soft
wishes can be negotiated. Agent finds out
these demands with the customer and finds a
product which fulfills them.

Agent can be “Active” or “Passive”
Sales


CBR Model must be modified
Standard Model:
Case
Library
1. Retrieve
4. Retain
Background
Knowledge
3. Revise
2. Reuse
Sales



New Model
No Retain phase: sale
does not add another
product to the product
base
Add Refine phase: user
demands refined based
on the evaluations given
by the customer.
Example



CBR approach
to negotiating a
BMW sale
Agent here is
passive
Buttons for
“sportier”,
“more
comfortable”,
“cheaper”, etc.
After-Sales



Customer has already bought a product and needs
support during its usage
To assist the customer, they are supported with a
case base of possible product problems, a query
interface, and similarity measures which should help
to find a similar problem and solution
Many companies have online CBR customer-support
websites (Dell, 3Com, etc)  Help Desk Systems
Example

Dell Support site:
http://support.dell.com
Summary




E-commerce is a growing field with lots of
potential revenue
Standard search technology is too limited
CBR can be applied in all 3 transaction
phases
Key is to provide intelligent sales support 
agent guides customer through each phase
of transaction
References
1.
2.
3.
“Intelligent Sales Support with CBR”
Wilke, Lenz, Wess
“Experience Management for Electronic
Commerce”
Bergmann
Wikipedia: http://en.wikipedia.org/wiki/Ecommerce