INFSCI 2480���Adaptive Information Systems ���

Adaptive Mobile Guides

Peter Brusilovsky School of Information Sciences University of Pittsburgh, USA

http://www.sis.pitt.edu/~peterb/2480-012/

The Vision

Introduction

Limited Resources

Technical resources

Cognitive resources

Screen size

Bandwidth

Ergonomics

Connectivity

Attention span

Cognitive load

Unfamiliarity

Display resolution

Time pressure

Adaptation on a desktop is a luxury. Adaptation on a mobile device is a necessity. Barry Smyth

User-Adaptive vs. Context-Based

1. Context-based, not adapted to the user as a person 2. User-adaptive but not context-based 3. Context-based and user-adaptive

User-adaptive Context-based

2 3

1

What is the Context? Environment User Task

1.  Technical (devices and services in use and available)

2.  Physical (place, location coordinates)

3.  Ambient (e.g. temperature, noise, lightness)

4.  Social (e.g. cultural environment, people near by)

1.  Abilities, skills, knowledge, habits, feelings

2.  What the user has in mind, where his/her attention is

What the user 1.  is doing 2.  has been doing 3.  plans to do

A prospect of context-based and ubiquitous computing

Four Dimensions for Mobile Guides

User-adaptive

Platform-adaptive Social

Location and context-based

How to Model

•  Modeling user/community characteristics – Traditional UM approach – observing user/

community, asking the users •  Modeling device characteristics

– Device profiles •  Modeling location and context

– Sensing! – Location, acceleration, light, time…

Ontology-based Modeling  Mobile applications require a variety of spatial, temporal,

physical, and activity  Using Ontology model

o  To enable exchange of knowledge o  To share common understanding of the structure of

information among people or websites o  To enable reuse of domain knowledge o  To separate domain knowledge from the operational

knowledge

Users, context and situation modeling: an ontology-based approach

  UbisWorld: the modeling and representation of users, context, and situations

An Ontology for Location Modeling

UserML forms the syntactic description in the knowledge exchange process Semantic definition of each concept is defined in GUMO

GPS Where am I?

Cell Tower

Bluetooth

Detecting Location

Outdoor Indoor

City/Navigation Guides Museum Guides

Shopping Assistants

City Guides: Guide o  A context-aware tourist guide

o  User Interests o  Location o  Time of the day

•  Provide a tailored-tour of the city •  Example of adaptation-Nearby

attractions page –  system performs both filtering and

information visualization –  uses visitors current location,

attractions already visited •  Uses two models

–  user model (wizard type interface) –  environment model (information

regarding attractions within the city)

Cheverst, K., Mitchell, K., and Davies, N. 2002. The role of adaptive hypermedia in a context-aware tourist GUIDE. Community. ACM 45, 5 (May. 2002)

•  Network cells defined by strategically placed wireless access points determine current position of the user (does not rely on GPS)

Guide: The Sensing Part

•  MoG, the mobile guide, providing multimedia content about the city of Cesena, using both WLAN and GPRS

•  MoG devivered multimedia content MoG uses W3 standard SMIL (Synchronized Multimedia Integration language)

MOG (2003)

A Multimedia Mobile City Guide for Outdoor Learning Paola Salomoni (2003)

•  A context-aware tourist guide that assists roaming tourists

•  Approach used: the deployment of intelligent agents which collectively determine the user context and retrieve and assemble multi-media presentations that are wirelessly transmitted and displayed on a PDA •  As tourists near attractions, Genie displays an

electronic map with their position and orientation highlighted on it.

•  As they approach a tourist attraction, Genie will start pre-caching a presentation.

•  When user is at attraction, attraction is presented as a small image with a list of available topics.

•  Each topic has a sound file that introduces the attraction.

•  List of topics are based on the user’s interests.

Gulliver’s Genie (2003)

Gulliver’s Genie: A Multi-Agent System for Ubiquitous Intelligent Content Delivery. G.M.P. O’Hare, M.J. O’Grady, Computer Communications, Vol.26, Issue 11, pp.1177-1187, 2003 Elsevier

BPN Navigation System (2004) o  Type: Resource adapting o  An entirely system that

combines a desktop route planning, a car navigation system, and a multi-modal in-outdoor pedestrian navigation

o  Provide a seamless service for travelers in different situations

Kruger, A., et al (2004), The connected User Interface: Realizing a Personal Situated Navigation System. In IUI2004: International Conference on Intelligent User Interfaces, New York, 161-168

Just-For-Us (2005)

Just-for-Us: A Context-Aware Mobile Information System Facilitating Sociality���Jasper Kjeldskov and Jeni Paay ���

Proceedings of the Mobile Human Computer Interaction, 2005���

•  Making the Invisible Visible: Augmenting the user’s physical surroundings

•  Supporting Ad-Hoc Communication about Places, Activities, and Time

•  Indexing Recommendations and Content to History and Context

•  Representing Activities within Proximity and Indexing to Familiar Places

Just-For-Us: Functionality

The System Architecture

NaviTime (2007)   Type: Resource adapting   Provide users with total navigation support

  All mode of transportation such as walking, driving, and riding trains, buses, taxis, and airplanes

  Encompasses the entire traveling activity

  The best route is calculated based on user’s preference (e.g., route with minimum expenses)

Arikawa, M., Konomi, S., Ohnishi, K. (2007), Navitime: Supporting Pedestrian Navigation in the real world, Pervasive Computing, IEEE,6(3), 21-29

iCity Project (2008)   Type: resource adapting   An adaptive, social, multi-device

recommender guide   The combination of the principles of

adaptation and web 2.0   Allowing users to post new events, rate

existing items, provide comments and judgments to a resource

  Provide the adaptive presentation based on the device between desktop and mobile

  The recommended items are tailored to individual according to the user model and contextual elements

F. Carmagnola, F. Cena, L.C.O. Cortassa, C. Gena, A. Goy, I. Torre, A. Toso, and F. Vernero, Tag-based user modeling for social multi-device adaptive guides. User Modeling and User-Adapted Interaction (2008)

Sotto Voce Museum Guide   Type: Resource adaptive   A handheld electronic guidebooks   Take into account the special needs of

groups, not individual user   By having pictures of the walls presented

on the PDA’s screen and asking from the user to select the exhibit he/she is interested in, by pressing it

  Encourage communications among group members by using audio sharing mechanism called “eavesdropping”

Aoki, P. M., Grinter, R. E., Hurst, A., Szymanski, M. H., Thornton, J. D., and Woodruff, A. 2002. Sotto voce: exploring the interplay of conversation and mobile audio spaces. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems: Changing Our World, Changing Ourselves

Agent Salon Project o  A life-like character-based

system for personal guidance o  Require explicit personal

interests and rate each exhibit during the tour

o  Support face-to-face discussions and exchange of knowledge by tempting users to chat with each other

Sumi, Y., Mase, K. In:Interface Agents that facilitate knowledge interactions between community members. Springer-Verlag (Cognitive Technologies series)(2004) 405-427

Information Kiosk

PEACH   Using the virtual character acts as a

tour guide   Using “like-o-meter” to collect user

feedback   Provide the seamless ubiquitous

presentation between a handheld device and a stationary device

  Provide automatically generated, adaptive video documentaries on mobile device

  Provide automatically generated post-visit summaries that reflect individual interests of visitors

Stock o., et al (2007), Adaptive, intelligent presentation of information for the museum visitor in PEACH, User Modeling and User-Adapted Interaction , 17, 257-304

Like-o-meter

PhoneGuide   Using mobile phones for on-device

object recognition   Image classification techniques with

user feedback are used to automatically recognize objects

  Provide a visitor with multimedia information

Lawrence Rosenblum and Simon Juliar (2008), Mobile Phone-Enabled Museum Guidance with adaptive classification, IEEE Computer Society, pp. 88-102

Video link

Smart Shopping Assistant (2003)   Utilizing plan recognition techniques

(inferring a user’s plan by observing his actions) to aid a user while shopping o  Provide product information o  Provide product recommendation o  Provide product comparison o  Provide recipe hints

Tablet PC RFID antenna

Schneider, M., (2003), Towards a transparent proactive user interface for a shopping assistant, In: Workshop on Multi-user and Ubiquitous User Interfaces, pp. 10-15

Metro Future Store (2006)   Type: Resource adapted   Integrate multiple emerging

technologies into an existing store   Electronic shopping lists   Provide product information and

navigate customer within a store   Recommend promotional

information using video and animation

  Smart check out system

Video Links

Kirthi Kalyanam, Rajiv Lal, and Gerd Wolfram (2006), Future Store Technologies and their impact on grocery retailing, In Retailing in the 21st Century, pp.95-112

Magitti •  Started as Japanese project, aiming to develop a service

to replace printed city-guides, sponsored by Dai Nippon Printing Co., Ltd.

•  Magitti –  Activity-centered mobile leisure-time guide. –  For young city-travelers (20-30) interested in all kinds of

leisure activities, emphasizing spontaneity rather than fixed and planified sightseeing.

•  What does it do? –  It first predicts ongoing and future activity based on the

user’s context and models of past behavior, and then –  Predicts (and recommends) what information will be

most useful within the predicted activity.

Victoria Bellotti, Bo Begole, Ed H. Chi, Nicolas Ducheneaut, Ji Fang, Ellen Isaacs, Tracy King, Mark W. Newman, Kurt Partridge, Bob Price, Paul Rasmussen, Michael Roberts, Diane J. Schiano, Alan Walendowski: Palo Alto Research Center

Design Requirements

•  Context: Field study in Tokyo. •  After a study for understanding leisure time

priorities (interviews and mockups, online survey, focus groups, mobile phone diaries, street activity sampling, expert interviews, and informal observation) they obtained the following design requirements: –  Relaxation, Serendipity and Spontaneity –  Avoidance of Information Overload –  Minimal Size (keitai – handhelds) –  One-handed operation (commuters on subway)

Magitti Features •  Context Awareness:

–  It knows about current time, location, weather, store hours, and user patterns.

–  It also lets people specify a future context for planning. •  Activity Awareness:

–  It filters items to recommend based on its user’s inferred or explicitly specified activity modes.

–  Five modes were derived from observations in field work: Eating, Shopping, Seeing, Doing, or Reading.

–  Each item in the Magitti database is explicitly tagged as being associated with these modes.

•  Serendipitous, relaxing experience: –  Users need not enter profile, preferences, or queries. –  They can rely on context and activity inferencing for Magitti to

continually and automatically update recommendations.

Magitti System Design

•  System shows up to 20 recommendations and updates while the user moves Allows

the user to rate

Magitti System Design - II

•  Mode 2