Transcript Hai_ecommerce - Computer Science & Engineering

E-Commerce
Hai Hoang
4/25/2005
Outline
► Definition
► Why use E-commerce?
► Current State of E-Commerce
► Barriers in Adopting E-Commerce
► Motivations for CBR
► Using CBR to Provide Intelligent Support
► Why database search is not good enough?
► Phases of Sales in E-commerce
► Demo
► Conclusion
Definition
► E-Commerce
exchange of:
typically refers to the
 1) information
 2) goods
 3) and services
► Through
electronic networks
Why use E-commerce?
► Can
be summarized as providing “incentives”
► Incentives:
 Browse catalogs without leaving house or office
(convenience)
 Shop anytime, anywhere
 Compare prices from different vendors (variety)
 Cheaper prices
 Direct from supplier without 3rd parties
 No Tax if store is out of state
 Oops, all from the customer’s perspective, - sellers also
benefit from e-commerce too
Why use E-Commerce?
► Seller:






World-wide customers
No need for a physical store
Cheaper to setup a store
Maintenance cost?
(packaging cost? Example software)
Hire less workers
► And
a lot more – depending on the business
Current State of E-Commerce
► Online
sales represent 5.4% of all retail
sales in 2004
► By 2008, it has been estimated to only
reach about 10%.
► With all the incentives listed on the previous
two slides:
 Why is it only a small percentage of all retail
sales?
Barriers in Adopting E-Commerce
► What
kind of barriers?
► Tannenbaum summarizes into 3 words:
 Confidence – security, privacy, liability
 Convenience – easy of access, use, purchase,
and get support
 Content – variety of products and services
► Which
one can CBR help to overcome?
 Make the it more convenient
Motivations for CBR
•“On the Internet, companies only have computers to represent them.
They better be intelligent computers.”
Chuck William, CEO Brightware
•The answer CBR???
Motivations for CBR (cont.)
► Dell:
I want a gaming computer that could
support the latest game, ie Half-Life 2
► Preferred System Configuration
2.4 GHz Processor
512MB RAM
DirectX 9 capable graphics card 256mb
Windows 2000/XP/ME/98
Internet Connection
► Example:
Dell fails to provide the sale
support (show example at dell.com)
Using CBR to Provide Intelligent
Support
► How
can we use CBR?
 search, select, and support.
► Refine
searching capabilities
 Traditional database queries is not powerful
enough
► Reduce
required domain knowledge
► Therefore, reduce customer support staff
Why database search is not good
enough?
► What’s
wrong with a huge database with a
search database search engine on top of it?
 Too many results
► Provided
that the search engine is good. It
could narrow down the search for you very
well. What else could be wrong?
 It fails to provide sales support (tailored
recommendations)
► Remember
the Dell example???
Phases of Sales in E-commerce
►3
phases of sales
 Pre-sales
 Sales
 After-sales
► All
3 could use CBR to provide support
Pre-sales
► Customer
has an idea of what he wants to
buy
► Customer attains information about
products and services
► Traditional way is electronic catalogs and
databases
Pre-sales: presenting products
► Index
& Ontologies
 Amazon.com
► Requirements
 Expedia.com
based
Pre-sales: Misses
► With
index & onthologies
 Too many results
 Specific speaker might be listed in a different category
► With




requirements based
Typos
What if there is a date close by the one specified?
What about close by airports?
Special deal for the specific destination?
Pre-sales: Misses example
► Example
at expedia.com
 (Flight search)
 (Last minute packages)
Pre-sales: CBR example
► Analog
► Travel
Devices
Agency
 Last Minute Deals (no implementation yet)
 (DietoRecs)
 (Nut King)
Analog Devices Example
► Analog
Devices makes electronic parts
► Most of the calls to their support number is
because they can’t find the desired parts by
looking in the catalog.
► Solution: create an online store that allows
engineers to find desired parts based on
functionality using CBR
Analog Devices Example
►A
case is represented with 40 parameters (String,
integer, symbols)
 Type, input/output voltage, size
► Parameters
are divided into a few categories
► System creates a query based on these
parameters, using a few or all
► www.analog.com
for example of UI
► (Note the priority checkbox)
AD: Similarities (discrete)
► similarities
are calculated by applying local
similarity measures.
► Note:
table is asymmetric
AD: Similarities (continuous)
► Represented
► Anything
as a function instead of a table
greater than query, 1, otherwise, it’s
from 0 to .5
Travel Agency
► Interactive
 Expedia, Orbitz, Hotwire
 Nut King, DieToRecs (CBR)
► Interactive
Last Minute Deals
 Somewhat available at those sites mentioned
above
 CBR to recommend close matches
Travel Agency (cont.)
► Last
minute deals are sold as-is
► Used to be distributed with flyers
► Up to 200,000 deals sold in peak season
► could
be represents as cases
► When user search for a destination,
recommend similar deals
Travel Agency (CBR impl.)
► Treat
deals as cases with attributes
► Recommend deals with intelligence using
CBR
Search:
1.) January
Response:
No match at
Blue Mountain
2.) Blue Mountain
3.) Skiing, Hotel
1.) January
2.) Bear’s Creek
Suggest
closest case
3.) Skiing, Hotel
4.) 1 day
4.) 1 day
5.) $50 - $100
5.) $89
Sales Phase
Sales (CBR Cycle)
• Might have to go
through the cycle a
few times before
user accepts the
proposed product.
•Each time user’s
evaluation refines
the demand
•Retain phase must
not take place
because a
successful sale does
not lead to a new
product in the
product base
Negotiation During Sales
► During
the negotiation process, the buyer
and seller refine the problem description
and the product description to solve the
customer’s problem
► The seller should help the customer to make
a decision not tell them when their
requirements are fullfilled
CBR in Negotiation
► Active
or Passive Sales Agent
 Active – suggest a particular modification of a
demand to the user
 Passive – Offer possible modifications to the
user, the user pick what to modify
► Modification
Products
of the Demands vs the
 Ex free gifts or configurable products
Sales
► Over
or Under specification of customer’s
demands
 Need to relax the demands
 Can see this at Nut King and DieToRecs
After the Sales
► Support
► CBR
in online help systems
 HP, Gateway, Dell, and 3com have been using
CBR in their online support system
► The
reason? Call-avoidance
Demo
► DieToRecs
 http://eu-project.hgb.tiscover.at/dialogservlet
► Nut
King
 http://itr.itc.it:8080/dev/jsp/language.do?action
=english
Conclusions
► E-Commerce
is still only a small portion of the
retail market – there is a lot of potential for
growth
► CBR can be used with all 3 phases of the sales.
► Help to narrow down the search result
► Make intelligent suggestions
► Make online system behave more like human sales
agents
► Provide intelligent support
References
►
►
►
►
►
►
►
►
Intelligent Sales Support with CBR. Wolfgang Wilke, Mario Lenz, Stefan
Wess
Experience Management for Electronic Ecommerce.
Francesco Ricci http://ectrl.itc.it/home/home_people/ricci/ (last
visited: April 20, 2005)
(online e-commerce stat)
http://retailindustry.about.com/od/seg_internet/a/bl_nrf052504.htm
(online shopping benefits)
http://wiki.media-culture.org.au/index.php/Online_Shopping__Benefits_for_Buyers
(half-life 2 spec)
http://www.ultimate-gamer.com/halflife2/hl2.htm
(DieToRecs)
http://eu-project.hgb.tiscover.at/dialogservlet
(Nut King) http://itr.itc.it:8080/dev/jsp/language.do?action=english