Personalization

1
Personalization

• User Attitudes Regarding a User-Adaptive
  eCommerce Web Site
• Personalizing the User Experience on ibm.com
• Impacts of User Privacy Preferences on
  Personalized Systems a Comparative Study

Frans Faizal ffaizal_at_ics.uci.edu
ICS 206 Spring 2003
2
Personalization
User Attitudes Regarding a User-Adaptive
eCommerce Web Site Personalizing the User
Experience on ibm.com Impacts of User Privacy
Preferences on Personalized Systems a
Comparative Study
3
Overview

• Describes user studies that focused on the
  perceived value of a variety of personalization
  features for an eCommerce Web site for computing
  machinery sales and support.
• Describes how the results of the studies affect
  the design of user-adaptive applications.

4
Definitions

• Personalization
• The use of information about a particular user to
  provide tailored (personalized) user experiences
  for that user.
• A personalized Web site
• A system that adapts the content structure,
  and/or presentation of the networked hypermedia
  objects to each individual users
  characteristics, usage behavior, and/or usage
  environment.

5
Overview of User Studies (1)

• Purpose
• To determine which specific personalization
  features would be judged the most usable,
  valuable, and attractive to users of an eCommerce
  Web sites.
• Gathered a large amount of quantitative and
  qualitative data.
• Written and spoken opinions, written
  questionnaires, think aloud protocols, free-form
  group and one-on-one discussions, as well as
  direct observations.

6
Overview of User Studies (2)

• Obtained clear attitudes of users toward adaptive
  techniques that were intrinsic to the
  implementation and design of the personalization
  features being tested.
• Conducted three studies, carried out in multiple
  laboratory settings.
• Each has different participants and different
  methodologies (group vs. individual study).

7
Personalization Feature Space

• Started with 75 techniques (clustered based on
  similarities).
• Wanted to refine the list based on measure of
  effectiveness, usefulness, and user attitudes
  derived from successive user studies.

8
(No Transcript)
9
Prototype Adaptive Web Site (1)

• Two prototype systems low-fidelity (Study 1 2)
  and interactive versions (Study 3).
• Implemented in Microsoft PowerPoint and presented
  on an IBM ThinkPad computer.
• Low-fidelity prototype consisted of screen shots.
• Lead experimenter clicked on a widget and the
  response was displayed on the screen.

10
Prototype Adaptive Web Site (2)

• Designed to demonstrate specific personalization
  and adaptive features.
• Exemplified a Web site (a system) that maintains
  a profile of the users personal information and
  tailors the sites content to that user based on
  the profile and navigational context.
• PersonalBook
• Central personalization tool that is closely tied
  to user profile.

11
(No Transcript)
12
(No Transcript)
13
Written Questionnaires (1)

• Used to capture both quantitative and qualitative
  data.
• Also asked subjects to rate the personalization
  features demonstrated in each study.
• Stated as assertions.
• E.g., you control all the data kept in your
  profile and can review and edit it at any time.

14
Written Questionnaires (2)

• In Study 1, participants were asked to rank the
  features shown based on their value to the
  participants.
• In Study 2 3, they were asked to rate the
  features using a 7-point scale (1 is Highly
  Valuable, 7 is Not at all Valuable).
• Questionnaires also asked marketing and business
  case issues (whether subjects thought they would
  be more likely to come back and buy more).

15
User Task Scenario

• You and your department have made various
  server, laptop, and desktop purchases. You now
  think you may have to purchase additional memory
  to enhance the capabilities of the laptops used
  by your department members. Starting from your
  PersonalBook, find 128MB add-on memory chipsets
  compatible with those laptops. Then also find
  memory compatible with the desktop machines your
  department owns

16
(No Transcript)
17
(No Transcript)
18
Results and Conclusion (1)

• Users want to be in control of their personal
  information.
• Able to review, modify, and delete personal
  information in their profile.
• Able to control over who sees and uses the
  information.
• Do not want their information gathered
  implicitly.
• Able to decide which information to be stored in
  their profile.

19
Results and Conclusion (2)

• Users want to be in control of the content shown
  on a site.
• Seems to defeat the purpose of an adaptive site.
• They are happy as long as the content is
  generated based on the information they provide
  explicitly to the system.
• E.g. content-filtering and content-refinement in
  the PersonalBook.

20
Results and Conclusion (3)

• Adapting content based on implicit information is
  undesirable.
• E.g. Compatible Memory scenario.
• Adapting content based on past navigation is also
  undesirable.
• You cant do it well, so dont do it at all.
• Users want to be invisible during exploratory
  sessions.
• I.e. multiple user roles or persona.

21
Results and Conclusion (4)

• Adapting content based on transient information
  is good as long as it is clear what is going on.
• Collaborative filtering was not supported fully.
• I am not like other people. I have different
  needs.
• Inappropriate products or services?

22
Questions/Comments?
23
Personalization
User Attitudes Regarding a User-Adaptive
eCommerce Web Site Personalizing the User
Experience on ibm.com Impacts of User Privacy
Preferences on Personalized Systems a
Comparative Study
24
Overview

• Describes a strategy for bringing personalization
  to the ibm.com public Web site.

25
Definitions (1)

• Personalizing interaction
• The use of information about a user to alter the
  content and functionality of the user experience.
• Personalizing a Web site
• Using personal information about an individual to
  tailor the experience for that individual on the
  site.

26
Definitions (2)

• Personalization policy
• A decision made by an eCommerce company involving
  the handling of personal data on the companys
  Web site.
• Personalization feature
• A method for collecting and using personal
  information in order to tailor a Web site
  visitors experience on the Web site.
• A personalization policy applies to the whole Web
  site, while a feature provides functionality for
  a particular task on the site.

27
Personalization for eCommerce

• Involving customer and provider (producer) roles
  that interacts with each other.
• The goal is to provide increased interaction
  value to both parties using their personal
  information.
• Value of customer
• F(cost of providing info, perceived benefits)
• Value of company (provider)
• F(cost of gathering info, perceived value)

28
Personalization Value Space

• A range of information type and possible values
  to customers and businesses.
• The value of techniques to any customer will vary
  with the role of the customer at any time.
• The value of a technique to a business will
  depend on the kind of business objective they
  have.
• There are likely to be interactions between
  techniques resulting in a package of techniques
  that would be optimally effective.

29
Project Goals

• To understand the value of personalization to
  customers and IBM.
• To develop the strategy for bringing
  personalization to the ibm.com public Web site
  which ensures that the top-priority goals of
  customers and the business are met.

30
Project Approach (1)

• Completing a literature review of the published
  research in the area of personalization.
• Identify possible personalization features and
  understand state of the art.
• Completing a set of heuristic evaluations of the
  ibm.com site and key competitors to understand
  current best practices.
• Dell, HP, Compaq, IBM, Sun, and Amazon

31
Project Approach (2)

• Identify business requirements
• Done primarily by ibm.com stakeholders.
• Gathering information about personalization
  features that might be used.
• Came up with 75 features (as described in the
  previous paper) and three policies (described
  next).
• Executing iterative user studies.

32
(No Transcript)
33
(No Transcript)
34
Three Policies (1)

• Giving Web site visitors control of the data in
  their profiles.
• They can review, edit, or delete information
  about themselves, their purchase, etc.
• Asking visitors for the minimal amount of
  personal information necessary and providing
  immediate value to the customer based on use of
  it (Permission Marketing).
• The customers profile is built slowly over time
  as the individual develops trust in the eCommerce
  company.

35
Three Policies (2)

• Enabling Web site visitors to adopt different
  level of identities as appropriate to their tasks
  on the Web site.
• Level of identity is based on degree of personal
  information provided.
• If no information is given, the visitor is
  invisible.

36
(No Transcript)
37
Questions/Comments?
38
Personalization
User Attitudes Regarding a User-Adaptive
eCommerce Web Site Personalizing the User
Experience on ibm.com Impacts of User Privacy
Preferences on Personalized Systems a
Comparative Study
39
Overview

• Compares 30 opinion surveys on Internet privacy,
  categorizes the responses, and matches them with
  possible impacts on personalized systems.
• A first contribution towards the identification
  of requirements for privacy-preserving
  personalization, to improve users trust when
  interacting with personalized systems.

40
What is Personalization?

• Personalization is predictive analysis of
  consumer data used to adapt targeted media,
  advertising, or merchandising to consumer needs.
• A personalized hypermedia application is a
  hypermedia system which adapts the content,
  structure and/or presentation of the networked
  hypermedia objects to each individual users
  characteristics, usage behavior and/or usage
  environment.

41
User-Adaptable vs. User-Adaptive Systems

• User-adaptable systems
• User is in control of the initiation, proposal,
  selection, and production of the adaptation.
• User-adaptive systems
• Performs all steps autonomously.
• E.g., Amazon.com.
• Generates purchase recommendations based one a
  users purchase and interaction history.

42
Advantages of Personalization

• Enables online sites to offer more relevant
  content and to recall user preferences and
  interests.
• Improves the learning progress in educational
  software.

43
Privacy-Critical Personalization Processes (1)

• Personalization
• Recurring processes of data collection,
  profiling, and matching.
• From the collected data, user profiles are
  created and used to personalized contents.
• Then, new data are collected, and profiles are
  updated.

44
Privacy-Critical Personalization Processes (2)

• Data collection
• The most privacy-critical in the personalization
  process.
• Could provoke privacy fears that limit consumers
  willingness to share information.

45
Data Types (1)

• User data
• Information about personal characteristics of the
  user.
• E.g., demographic data and user knowledge,
  skills, capabilities, interests, preferences,
  goals, and plans.

46
Data Types (2)

• Usage data
• Related to users interactive behavior.
• E.g., selective actions, temporal viewing
  behavior, ratings, purchases and purchase-related
  actions, and other confirmatory and
  disconfirmatory actions.

47
Data Types (3)

• Usage regularities
• Based on frequently re-occurring interactions of
  users.
• E.g., usage frequency, situation-action
  correlation, and action sequences.
• Environment data
• Focuses on the users software and hardware and
  the characteristics of the users current locale.

48
Privacy Surveys

• Looked at 30 surveys (or summary of survey) from
  2001-2002.
• Eleven included all questions (full reports).
• Six provided an extensive discussion of survey
  results (elaborate executive summaries).
• Ten gave factual executive summaries.
• Three were only available in a form of press
  releases.

49
Different Aspects of Privacy (1)

• Privacy of personal information in general
• User statements addressing this aspect have a
  direct impact on personalized systems requiring
  personal information.
• E.g., statements regarding security of providing
  personal and sensitive information and sharing of
  such information.

50
Different Aspects of Privacy (2)

• Privacy of personal information in a commercial
  context.
• User statements addressing this aspect primarily
  affect eCommerce in general and specifically
  personalized systems in an eCommerce environment.
• E.g., statements regarding security and sharing
  of personal information given during an online
  transaction.

51
Different Aspects of Privacy (3)

• Tracking of user sessions and the use of cookies
• User statements addressing this aspect influence
  user-adaptive systems requiring usage data.
• E.g., statements regarding accepting, rejecting,
  or deleting cookies and tracking visited Web
  sites.

52
Different Aspects of Privacy (4)

• Email privacy
• User statements addressing this aspect could have
  an impact on user-adaptive systems dealing with
  emails.
• E.g., statements regarding irrelevant,
  unsolicited, or offensive emails.

53
Discussion of the Results (1)

• Concern over the use of personal information
• A few users supplied false information to a Web
  site when asked to do so.
• A significant percentage of Internet users never
  consider providing personal information to a Web
  site.
• This severely affects personalized systems that
  require user to submit user data.

54
Discussion of the Results (2)

• Concern over the sharing of personal information
• Almost half of the Internet users think that
  sharing personal information with other sites
  invade privacy, unless sharing can be controlled
  by the users.
• This has a severe impact on central user modeling
  servers that collect and share data with
  different user-adaptive applications.

55
Discussion of the Results (2)

• Concern over the tracking and cookies
• More than 50 of Internet users concern about
  Internet tracking.
• A significant number claimed they would set their
  browser to reject cookies.
• More than half of the users stated they would
  delete cookies periodically.
• This affects machine-learning methods dealing
  with log data.

56
Discussion of the Results (3)

• Concern over email privacy
• 62 complains about irrelevant emails.
• Almost every Internet user has received
  unsolicited emails.
• This especially affects personalization systems
  that deal with personalized emails.

57
Discussion of the Results (4)

• Most users willing to give personal information
  in exchange for personalized user experience, but
  not sensitive information.
• Users demonstrate less commitment in providing
  information to a Web site that shares the
  information to other sites.

58
Discussion of the Methodology (1)

• Lack of comparability of studies
• Small differences in wording of questions,
  context of questionnaires, sample size,
  recruiting method and demographic characteristic
  will influence the result.
• Using imprecise terminology
• The term privacy is often used as a synonym of
  security against identity fraud or spam.

59
Discussion of the Methodology (1)

• Users stated privacy preferences and the actual
  behavior may diverge.
• 76 of users states that privacy policies are
  important, but barely view such pages when they
  visit Web pages.
• Users willingness to share information depends
  on other factors, such as usability of the sites,
  level of trust, and to whom the sites belong to.

60
Future Directions

• Giving a guarantee that users personal data will
  only be used for the intended purposes.
• Such guarantee is forced by privacy laws.
• Allowing anonymous interaction.
• Users will be more open.
• Relieves the provider from the restriction of
  privacy laws.

61
Questions/Comments?