Virtual Sales Agent for Personalized Internet Shopping

1
Virtual Sales Agent for Personalized Internet
Shopping
13th Japan-Germany Forum on Information Technology
Wolfgang Wahlster
German Research Center for Artificial
Intelligence, DFKI GmbH Stuhlsatzenhausweg
3 66123 Saarbruecken, Germany phone (49 681)
302-5252/4162 fax [phone] e-mail
wahlster_at_dfki.de WWWhttp//www.dfki.de/wahlster
2
Three Generations of Web Sites
First Generation
Second Generation
Third Generation
Virtual Webpages
Interactive Web Sites
Static Web Sites
Netbots, Information Extraction, Presentation
Planners
JavaScripts and Applets
User Modeling, Machine Learning, Online Layout
Database Access and Template-based Generation
Fossils cast in HTML
Dynamic Web Sites
Adaptive Web Sites
3
What are Virtual Sales Agents?
l appear as life-like characters l plan
interactive behavior autonomously l can
initiate interaction
ACTIVE
l understand the users requests l answer
clarification questions l allow mixed
initiative dialogs
l respond immediately to interruptions l
criticism and clarification questions l
direct manipulation
RE- ACTIVE
INTERFACE AGENTS
INTER- ACTIVE
PROACTIVE
l anticipate the user's needs l adopt the user's
goals l provide unsolicited comments
4
The Idea of Personalized Netbots
Personal Assistants
e.g. MetaCrawler
Softbots
Indices, Directories, Search Engines
Mass Services
WWW
Travellers Netbot Tries to achieve travellers
goals (finding and executing plans) ? checks
availability ? finds best price ? uses personal
preferences (e.g. frequent flyer programme,
seating preferences) ? lets the traveller know,
when seats become available (active help)
5
Intelligent Web Services
Provider
Netbot
Consumer
sells ? Information ? Goods ? Services
buys ? Information ? Goods ? Services
? Intelligent Parallel Retrieval ? Information
Extraction and Summarization ? Personalized
Presentation ? Matchmaking ? Teleshopping
Assistance ? Telemarketing Assistance ?
Translation Services ? Data Mining Services
Knowledge about ? Usage Patterns ? User
Models ? Consumer Profiles
Web Sites
6
Intelligent Agent Technology is a
Prerequisitefor Advanced WebCommerce
Shopbots for Automated Comparison Shopping
Virtual Web Pages
Text Analysis and Generation
Multimedia Presentation Planning
Information Extraction from HTML/XML Documents
Advanced WebCommerce
User Modeling and Language Generation Coordinated
Text Graphics Planning
Machine Translation
One-to-One Marketing
Robust Dialogue Understanding Advanced
Speech Synthesis
Intuitive, Multilingual Access
Multimodal Interfaces
Dialogue with Virtual Sales Agents
7
Virtual Market Places with Human and Machine
Agents
8
Five Generations of Internet Applications
2000
Embedded Internet Agents
Mobile Internet Services
WWW
EMail
Research Net
1
2
3
4
5
Every Car has a homepage, Agents are
main Internet users, Ubiquitious Computing
t
Internet Access via WAP and UMTS devices
9
Towards Mobile and Speech-based E-Commerce Using
WAP Phones
l WAP phones (Wireless Application Protocol
for Cellular Phones)
l WML as a markup language for
interactive content
l Mobile access to virtual shops allows price
comparisons during real shopping l Multimodal
dialog Voice In (Speech) - Web Out (Graphics,
Hypertext) l Voice input using advanced speech
understanding technology l Easy to use
customers simply say what they want
10
What is a Virtual Web Page?
Virtual Memory, Virtual Relation, Virtual
Reality...
A Virtual Web Page l is generated on the fly as
a combination of various media objects
from multiple web sites or as a transformation
of a real web page. l looks like a real web
page, but is not persistently stored. l
integrates generated and retrieved material in a
coordinated way. l can be tailored to a
particular user profile and adapted to a
particular interaction context. l has an
underlying representation of the presentation
context so that an Interface Agent can comment,
point to and explain its components.
11
Virtual Webpage Retrieved from 5 Different Servers
12
AiA Information Integration for Virtual Webpages
13
The Generation of Virtual Webpages with PAN and
AiA
Hotel Agent
Map Agent
Address
AiA
Presentation Planner
Pictures and Graphics
Netbot PAN
Pieces of Text
Components of virtual Webpages
Virtual Web Presentation
Coordinates for Pointing Gestures
Trip Data
Input for Speech Synthesis
Persona Server
Icons for Hyperlinks
Constraint- based Online Layout
Weather Agent
Train Flight Scheduling Agent
Major Event Agent
14
Persona as a Personal Travel Consultant
15
A Natural Language Agent for Finding Pre-Owned
Porsche Cars
Boxter, not red, must have AC, less than 20k
16
Enhanced ECommerce through Personalization
System is able to flexibly tailor product
presentations to the individual user and the
current situation.
An animated character serves as Alter Ego of
the presentation system.
Personalized Presenters at DFKI
17
PPPs Persona Server implements a generic
Presentation Agent that can be easily adapted to
various applications

Visual Appearances
Behaviors
Hand-drawn
l
Presentation Gestures
Cartoon
l
Reactive Behaviors
Bitmaps
l
Idle-time actions
l
Navigation actions
Persona
Server
Auditory Characteristics
Video Bitmaps
l
Sound effects, auditory icons
l
Voice male, female
Generated
Bitmaps

from

3D-Models
18
Classification of Persona Gestures
Gesture Catalogue

• Talking Posture 1
• cautious, hesitant
• appeal for compliance
• avoids body-gestures

• Talking Posture 2
• active, attentive
• self-confident
• uses body-gestures

19
Context-Sensitive Decomposition of Persona Actions
High-Level
take-position (t
t
)
point-to (t
t
)
4
1
2
3
Persona Actions
Context-Sensitive
move-to (t
t
)
r-stick-pointing (t
t
)
Expansion
1
2
3
4
(including Navigation Actions)
Decomposition

r-turn (t
t
)
1
21
r-hand-lift (t
t
)
into

3
31
r-step (t
t
)
Uninterruptable

21
22
Basic Postures
f-turn (t
t
)
r-stick-expose (t
t
)
22
2
31
4
Bitmaps
...
...
...
...
20
Extensions of the Representation Formalism
Distinction between production and presentation
acts
(i.e. Persona- or display acts)
Explicit representation of qualitative and
quantitative constraints
Production Act
Presentation Act
Introduce
S-Show
S-Position
Elaborate-Parts
Create-
Graphics
S-Wait
Label
Label
S-Create-
S-Depict
Window
S-Speak
S-Point
S-Speak
S-Point
Qualitative constraints
Create-Graphics meets S-Show, ...
Metric constraints
1 lt Duration S-Wait lt 1, ...
21
PET Persona-Enabling Toolkit
Objective l Enable non-professional computer
users to populate their web pages with
lifelike characters PET comes with l a set of
characters and basic gestures l an
easy-to-learn Persona markup language Developers
PET will include l a character design tool
which enables users to build their own
characters Technical Realization l Based on
XML and Java
22
The Persona Markup Language
lthtmlgt ltheadgt lttitlegt Persona Test
lt/titlegt lt/headgt ltbodygt ltpersona bitmapcartoon
...gt ltuselib url .../gt ltdo namegreet/gt ltdo
name speak argshello/gt lt/personagt lt/bodygt lt/
htmlgt

• Features
• XML-based
• easy to learn

23
Functional View of PET
Bitmaps
Webpage with Reference to Java Applet
URL of Webpage with Persona Tag
lthtmlgt ... ltAPPLET archivepersonaplayer.jar...lt
/APPLETgt ...lt/htmlgt
PET Application Server
lthtmlgt ltheadgt lttitlegt Persona Test
lt/titlegt lt/headgt ltbodygt ltpersona bitmapcartoon
...gt ltuselib url .../gt ltdo namegreet/gt ltdo
namestandard/gt ltdo name speak
argshello/gt lt/personagt lt/bodygt lt/htmlgt
Persona Scripts
waitscreen 4 gesture greet 0 0 null gesture laugh
0 0 null ...
Persona Engine
Audio Data
24
The Bidirectional Control Flow onPersona-Enabled
Webpages
l Mouse Clicks l Mouse Movements
Triggers actions of the Persona
l Text Input l Speech Input l Menu Input l Direct
Manipulation Input
Triggers operations on elements of the webpage
Web Persona
25
Porsche 9 11 Boxter
26
Persona Active Elements (PAE)
l Active Images An active image starts a
persona action when clicked. l Addressable
Objects An addressable object is an object which
can be addressed and manipulated by Persona via
its name and its position.
27
A Virtual Sales Agent for OTTO Worlds Largest
Tele-Ordering Company
28
DFKIs Ecommerce Presentation Planner has been
extended to accommodate for various target
platforms through the introduction of a mark-up
language layer
Presentation Planner
Agent Script
PET- PML
WML
SMIL
PET Persona Player
WML-Browser
MS-Agent Controller
SMIL Player
29
Information Extraction Agents

• Information Filtering
• Information Retrieval
• Information Integration

• ...
• identify and extract relevant
• pieces of information
• transform them into canonical form

• wrappers
• operational descriptions of a target concept
• abstract from concrete occurrence within
  document
• robust against modifications

30
Use of a Life-like Character for Electronic
Commerce
Digital Assistant Selector
31
Simulated Dialogues as a Novel Presentation
Technique

• Presentation teams convey certain rhetorical
  relationships in a more canonical way
• Provide pros and cons
• The single presenters can serve as indices which
  help the user to classify information.
• Provide information from different points of
  view, e.g. businessman versus tourist
• Presentation teams can serve as rhetorical
  devices that allow for a continuous reinforcement
  of beliefs
• involve pseudo-experts to increase evidence

32
Presentation Teams for Advanced ECommerce
I recommend you this SLX limousine.
33
Underlying Knowledge Base

• Representation of domain
• FACT attribute car_1 consumption_car_1
• Value dimensions for cars adopted from a study of
  the German car market
• safety, economy, comfort, sportiness, prestige,
  family and environmental friendliness
• FACT polarity consumption_car_1 economy negative
• Difficulty to infer implication of dimension on
  attribute
• FACT difficulty consumption_car_1 economy low

34
Example of a Dialogue Strategy

• Header
• (dampening_counter ?agent ?prop ?dim)
• Constraints
• (and
• (positive ?agent)
• (pol ?prop ?other_dim positive))
• Inferiors
• (Speak ?agent
• (Forget about the ?dim !))
• (Speak ?agent
• (Think of the ?other_dim !))

• Question
• How much gas does it consume?
• Answer
• It consumes 8l per 100 km.
• Negative Response
• Im worrying about the running costs.
• Dampening Counter
• Forget about the costs.
• Think of the prestige!

35
Current and Future Work Multiple Interface
Agents for User-adaptive Decision Support
weighted propositions
...
...
Multiple Decision Support Agents
Spare parts for this car are rather expensive!
But, its fast!
36
Personified Agents Increase the User's Trust in
the System's Presentation
Experimental evidence for effects of modality on
the user's trust (van Mulken, 1999) The system
gives recommendations, which turn out to be wrong
in some cases. How much does a user trust the
system's advice depending on the modality of a
presentation?
1.0
0.8
0.7
0.6
0.5
Self-animated Persona, Speech, Gesture,
Facial Expression, Pointing
Speech, Graphical Highlighting
Text, Graphical Highlighting
37
Impact of the modality of a Presentation on the
User's Trustfulness
Result Persona gt Speech gt Text
Conclusion If the presentation is more
human-like, recommendations are more
readily followed
For l decision support systems l tutoring
systems l recommendation systems l virtual
sales agents personified interface agents have a
clear advantage They increase the user's trust
in the information presented by the system
38
Sending Interface Agents to Clients Plug-Ins or
Applets?
Plug-Ins
Applets
l Add features (character players) to browser l
Download triggered by user l Requires disk
space on client l Unrestricted access to
client l Less appropriate for WebCommerce,
Guides l Agents integrated in 3D
environments l Appropriate for
Entertainment Examples l Extempo's Jennifer
James (Hayes-Roth et al. 98) l PFMagic's
virtual petz
l Java animation code sent over the net l
Automatic loading l Requires no disk space on
client l Restricted access to client l
Appropriate for WebCommerce, Guides l Agents
integrated in 2D environments l Less appropriate
for Entertainment Examples l DFKI's Web
Persona (Müller et al. 98) l ISI's Adele
(Johnson et al 98)
New in AiA/PAN Balanced combination of Applets
and Servelets Efficient distribution of
client-side Java and server-side Java for driving
the Interface Agent
39
Research on Personalized Interface Agents brings
disparate subfields in the area of intelligent
systems together
User Modeling
Planning
Knowledge Representation
Image Understanding
Personalized Interface Agents
Intelligent Web Services
Natural Language Understanding
Plan Recognition
Machine Learning
Information Retrieval
Multimodal User Interfaces
40
Conclusion
ECommerce projects of DFKI have shown that
research on personalized interface agents can be
transferred to real world applications Dekra
(largest European organization of used car
dealers) FairCar as an ECommerce platform with
NL access and a comparison shopping agent for
used cars DaimlerChrysler online user modelling
in a one-to-one marketing system for Mercedes
cars Otto/Shopping24/Eddie Bauer (largest
European tele-order company) virtual sales
agents for one-to-one marketing of fashion and
computer hardware Porsche Virtual Market for
Pre-owned Porsche Cars
41
Conclusion
The generation of virtual webpages with
personalized interface agents leads to innovative
applications in Electronic Commerce, Electronic
TV Guides (EPG) Telelearning environments, Call
Centers and Help Desks Two Research
Challenges Making the Interface agents
sensitive to temporary limitations of the users
time and working memory capacity (cf. our READY
project, Jameson et al., p. 79-85 in IUI99
Proceedings) Making the agents instructible, so
that they can learn from the user in a dialog,
how to extract information in a new domain (cf.
Mathias Bauer, Dietmar Dengler TrIAs Trainable
Information Assistants for Cooperative Problem
Solving in Agents'99, on Tuesday)