personalization and web design monica bonett, ukoln, university of bath, uk [email protected]
TRANSCRIPT
Overview
• Motivation for providing personalization
• Terms and Techniques
• Issues: Privacy, Usability
• Use of Frameworks and Standards
• Examples in Learning and Teaching
• The IMesh Toolkit Project
Why Tailor Content?
• Build personal relationships– treat the user as an individual– increase user loyalty
• Control information overload
• Improve accessibility– cater for variation in physical capabilities– adapt to different devices or connection modes
Desired Outcomes
• Increase user satisfaction
• Repeat visits– e.g. saved information
• Increase sales or popularity
…. in general, to meet the user’s needs or preferences
Kinds of Preferences
• Look and feel
• Channels of information
• Customise parameters e.g. search
• Methods of delivery
• Tastes/interests (recommendations)
How preferences are stated
• Explicitly– Form filling– Ratings
• Inferred– click-throughs– purchases– can be implicit
Using the Preferences of Others
• Commonly used for recommendations
• Collaborative Filtering– recommendation seeker expresses preference by
rating an item/s– matching people determined by comparing tastes– recommendation/s generated
• Various algorithms
Issues: Usability
• Personalization is not an excuse for poor usability
• Cater for users who want to do sophisticated customisation, and those who will do none
• Provide adequate defaults to meet basic needs• Monitor usage patterns
Issues: Privacy
• Potentially large amounts of information are collected, sometimes implicitly
• P3P: a W3C proposal
• Privacy Statements– exactly what information is collected– how it is used (why is it needed ?)– how widely shared
Standards and Frameworks
The Argus Framework
Users ContentUser Interface Layer
Personalization
Profile Layer
Specific ValuesUser Profile
Personalization rules
Personalization rules
ContentProfile
Vocabulary Layer
Set of Attributes Content Attributes
User Attributes
Framework Components
• Users have profiles that represent their interests and behaviours
• Content is profiled, based on a set of attributes that are assigned specific values
• The business context has certain rules that govern how personalization happens. – match attributes of the content with attributes
captured in the user profile to determine which content to display.
Describing Users: Metadata
• IMS– describes characteristics of a learner to enable
exchange of learner information– structured information model (XML binding)
• eduPerson– EDUCAUSE/Internet2 task force– LDAP (directory building)
Example 1: IMS
• example attributes – identification (names, addresses, demographics)– accessibility (cognitive, technical, physical,
language)– interest (information describing hobbies and
recreational activities)
Example 2: eduPerson
• Example attributes– eduPersonAffiliation (person’s relationship to
the institution student, staff etc.)– eduPersonNickname (informal name)– preferredLanguage
Accessing profiles: SOAP
• W3C working draft (Version 1.2)
• Supports communication in a distributed environment
• Exchange of structured information based on XML
• User preferences could be exchanged in this way
SOAP Example
• Get price of books from ISBN number#!/usr/bin/perl use SOAP::Lite;
$s = SOAP::Lite -> uri('urn:xmethods-BNPriceCheck') -> proxy('http://services.xmethods.net/soap/servlet/rpcrouter');
my $isbn = '0201704471';
print $s->getPrice(SOAP::Data->type(string =>$isbn))->result;
Contact SOAP server
Contact SOAP server
Data: ISBNData: ISBN
Send data and get result
Send data and get result
The IMesh Toolkit
Gateway service(s)Gateway service(s)
Gateway service(s)Gateway service(s)
Profiledatabase
Authdatabase
AccessProfile AccessAuth
Register UpdateProfile
CGI/HTTP
Athens
SOAP
ODBC
Portability: the vision
"For the end user it would be a much better world if he or she could simply have a program pass a collection of history and opinion data to each system he or she wishes to interact with and instantly obtain personalized behaviour and where appropriate recommendations from it” (Cliff Lynch, June 2001)
Sharing User Descriptions
• Example: P3P - a W3C proposed recommendation– users can describe their privacy preferences– websites disclose how they handle information– make information available in machine-readable
format
• Identify commonality
• Allow for variation
Acknowledgments
The IMesh Toolkit Project
• Funded by JISC/NSF
• Based at UKOLN
• Working with Pete Cliff, Rachel Heery, Andy Powell and Richard Waller, ILRT (Bristol), ISP (University of Wisconsin, USA)
• In collaboration with Resource Discovery Network (RDN) and Subject Portals Project (SPP)
• Thanks also to Keith Instone and Argus Associates