Comp3150Comp4700

1
Comp3150/Comp4700

• Personalization and Privacy Technology

2
Lecture Outline

• What is personalization
• What is it for
• Components of a Personalization System
• Example Personalization Techniques
• Cookies
• Collaborative Filtering
• Collaborative Filtering Algorithms
• Examples of Personalization Technique in Practice
• Personalization and Privacy
• Platform for Privacy Preferences Project (P3P)
• References/Reading
• Expected Learning Outcomes

3
Personalization What is it? (I)

• Personalization is the ability to provide
  content and services tailored to individuals
  based on knowledge about their preferences and
  behavior.
• Paul Hagen, Forrester Research, 1999

4
Personalization What is it? (II)

• Personalization is the use of technology and
  customer information to tailor electronic
  commerce interactions between a business and each
  individual customer.
• Using information either previously obtained or
  provided in real time about the customer, the
  exchange between the parties is altered to fit
  that customers stated needs, as well as needs
  perceived by the business based on the available
  customer information.
• Personalization Consortium, 2003

5
Personalization What is it? (III)

• Personalization is the capability to customize
  customer communication based on knowledge
  preferences and behaviors at the time of
  interaction with the customer.
• Jill Dyche, Baseline Consulting, 2002

6
Personalization What is it? (IV)

• Personalization is about building customer
  loyalty by building a meaningful one-to-one
  relationship by understanding the needs of each
  individual and helping satisfy a goal that
  efficiently and knowledgeably addresses each
  individuals need in a given context.
• Doug Riecken, IBM, 2000.

7
Personalization What is it?

• Personalization is the ability to provide
  content and services tailored to individuals
  based on knowledge about their preferences and
  behavior.
• Personalization is the use of technology and
  customer information to tailor electronic
  commerce interactions between a business and each
  individual customer.
• Using information either previously obtained or
  provided in real time about the customer, the
  exchange between the parties is altered to fit
  that customers stated needs, as well as needs
  perceived by the business based on the available
  customer information.
• Personalization is the capability to customize
  customer communication based on knowledge
  preferences and behaviors at the time of
  interaction with the customer.
• Personalization is about building customer
  loyalty by building a meaningful one-to-one
  relationship by understanding the needs of each
  individual and helping satisfy a goal that
  efficiently and knowledgeably addresses each
  individuals need in a given context.

What are these people talking about anyway?
8
Personalization What is it?

• Personalization is the ability to provide
  content and services tailored to individuals
  based on knowledge about their preferences and
  behavior.
• Personalization is the use of technology and
  customer information to tailor electronic
  commerce interactions between a business and each
  individual customer.
• Using information either previously obtained or
  provided in real time about the customer, the
  exchange between the parties is altered to fit
  that customers stated needs, as well as needs
  perceived by the business based on the available
  customer information.
• Personalization is the capability to customize
  customer communication based on knowledge
  preferences and behaviors at the time of
  interaction with the customer.
• Personalization is about building customer
  loyalty by building a meaningful one-to-one
  relationship by understanding the needs of each
  individual and helping satisfy a goal that
  efficiently and knowledgeably addresses each
  individuals need in a given context.

Using user information to better design products
and services tailored to the user.
9
Personalization What is it?

• Personalization is the ability to provide
  content and services tailored to individuals
  based on knowledge about their preferences and
  behavior.
• Personalization is the use of technology and
  customer information to tailor electronic
  commerce interactions between a business and each
  individual customer.
• Using information either previously obtained or
  provided in real time about the customer, the
  exchange between the parties is altered to fit
  that customers stated needs, as well as needs
  perceived by the business based on the available
  customer information.
• Personalization is the capability to customize
  customer communication based on knowledge
  preferences and behaviors at the time of
  interaction with the customer.
• Personalization is about building customer
  loyalty by building a meaningful one-to-one
  relationship by understanding the needs of each
  individual and helping satisfy a goal that
  efficiently and knowledgeably addresses each
  individuals need in a given context.

On a Web site, personalization is the process of
tailoring pages to individual users'
characteristics or preferences
10
Personalization What is it?

• Still too abstract?

11
Personalization What is it?
User A
12
Personalization What is it?
User B
13
Personalization What is it?
User A
14
Personalization What is it?
User B
15
Personalization What is it?

• Can even be more intelligent

16
Personalization What is it?
Returns based on what the users search history
Returns without any knowledge of user preference
17
Personalization What is it?
18
Personalization What is it for?

• Commonly used to enhance customer service or
  e-commerce sales
• Personalization is sometimes referred to as
  one-to-one marketing, because the enterprise's
  Web page is tailored to specifically target each
  individual consumer.
• Personalization is a means of meeting the
  customer's needs more effectively and
  efficiently, making interactions faster and
  easier and, consequently, increasing customer
  satisfaction and the likelihood of repeat visits.
  

19
Personalization What is it for?

• Commonly used to enhance customer service or
  e-commerce sales
• Personalization is sometimes referred to as
  one-to-one marketing, because the enterprise's
  Web page is tailored to specifically target each
  individual consumer.
• Personalization is a means of meeting the
  customer's needs more effectively and
  efficiently, making interactions faster and
  easier and, consequently, increasing customer
  satisfaction and the likelihood of repeat visits.
  

"If we have 4.5 million customers, we shouldn't
have one store, we should have 4.5 million
stores." Jeff Bezos, CEO, Amazon.com
20
Personalization What is it for?

• The goals of personalization technology are
• It must deliver relevant, precise recommendations
  based on each individuals tastes and
  preferences.
• It must determine these preferences with minimal
  involvement from consumers.
• And it must deliver recommendations in real time,
  enabling consumers to act on them immediately.

21
Personalization Techniques

• Cookies
• Collaborative Filtering
• User Profiling

22
Personalization Techniques Cookies

• Cookies
• Although not designed specifically for
  personalization, cookies can be used to implement
  Personalization functions.
• A cookie is a piece of text that a Web server can
  store on a user's hard disk.
• Cookies allow a Web site to store information on
  a user's machine and later retrieve it.
• The pieces of information are stored as
  name-value pairs.

23
Personalization Techniques Cookies

• Cookies
• A Web site might generate a unique ID number for
  each visitor and store the ID number on each
  user's machine using a cookie file.

24
Personalization Techniques Cookies

• Cookies
• For example, some one visited goto.com, and the
  site has placed a cookie on his machine. The
  cookie file for goto.com contains the following
  information
• UserID A9A3BECE0563982D www.goto.com/
• Goto.com has stored on this machine a single
  name-value pair. The name of the pair is UserID,
  and the value is A9A3BECE0563982D. The first time
  the user visited goto.com, the site assigned the
  users browser a unique ID value and stored it on
  his machine.

25
Personalization Techniques Cookies

• Cookies
• A name-value pair is simply a named piece of
  data.
• It is not a program, and it cannot "do" anything.
  
• A Web site can retrieve only the information that
  it has been placed on your machine.
• It cannot retrieve information from other cookie
  files, nor any other information from your
  machine.

26
Personalization Techniques Cookies

• Cookies How Does Cookie Data Move?
• If you type the URL of a Web site into your
  browser, your browser sends a request to the Web
  site for the page. For example,
• If you type the URL http//www.amazon.com into
  your browser, your browser will contact Amazon's
  server and request its home page.
• When the browser does this, it will look on your
  machine for a cookie file that Amazon has set. If
  it finds an Amazon cookie file, your browser will
  send all of the name-value pairs in the file to
  Amazon's server along with the URL. If it finds
  no cookie file, it will send no cookie data.
• Amazon's Web server receives the cookie data and
  the request for a page. If name-value pairs are
  received, Amazon can use them.

27
Personalization Techniques Cookies

• Cookies How Does Cookie Data Move?
• If no name-value pairs are received, Amazon knows
  that you have not visited before. The server
  creates a new ID for you in Amazon's database and
  then sends name-value pairs to your machine in
  the header for the Web page it sends. Your
  machine stores the name-value pairs on your hard
  disk.
• The Web server can change name-value pairs or add
  new pairs whenever you visit the site and request
  a page.
• There are other pieces of information that the
  server can send with the name-value pair. One of
  these is an expiration date. Another is a path
  (so that the site can associate different cookie
  values with different parts of the site).
• You have control over this process. You can set
  an option in your browser so that the browser
  informs you every time a site sends name-value
  pairs to you. You can then accept or deny the
  values.

28
Personalization Techniques Cookies

• Cookies How Do Web Sites Use Cookies?
• In the broadest sense, a cookie allows a site to
  store state information on your machine. This
  information lets a Web site remember if you have
  visited before.
• Web sites use cookies in many different ways.
  Here are some of the most common examples
• Sites can accurately determine how many people
  actually visit the site. It turns out that
  because of proxy servers, caching, concentrators
  and so on, the only way for a site to accurately
  count visitors is to set a cookie with a unique
  ID for each visitor. Using cookies, sites can
  determine
• o How many visitors arrive
• o How many are new versus repeat
  visitors
• o How often a visitor has visited

29
Personalization Techniques Cookies

• Cookies Summary
• A message given to a Web browser by a Web server.
  
• The browser stores the message in a text file (in
  your local disk).
• The message is then sent back to the server each
  time the browser requests a page from the server.
  
• The main purpose of cookies is to Identify users
  and possibly prepare customized/personalized Web
  pages for them.

30
Personalization Techniques Cookies

• Do Cookies Compromise Security?
• Cookies do not act maliciously on computer
  systems. They are merely text files that can be
  deleted at any time - they are not plug ins nor
  are they programs.
• Cookies cannot be used to spread viruses and they
  cannot access your hard drive.
• However, this does not mean that cookies are not
  relevant to a user's privacy and anonymity on the
  Internet.
• Cookies cannot read your hard drive to find out
  information about you however, any personal
  information that you give to a Web site,
  including credit card information, will most
  likely be stored in a cookie unless you have
  turned off the cookie feature in your browser.
• In only this way are cookies a threat to privacy.
  The cookie will only contain information that you
  freely provide to a Web site.

31
Personalization Techniques Cookies

• Do Cookies Compromise Security?
• Both Netscape and Microsoft Internet Explorer
  (IE) can be set to reject cookies if the user
  prefers to use the Internet without enabling
  cookies to be stored.
• In Netscape, follow the Edit/Preferences/Advanced
  menu
• In IE, follow the Tools/Internet Options/Security
  menu to set cookie preferences.

32
Personalization Techniques Cookies

• Cookies Used for Personalization
• Sites can store user preferences so that the site
  can look different for each visitor (often
  referred to as customization or personalization).
  
• For example, if you visit msn.com, it offers you
  the ability to "change content/layout/color." It
  also allows you to enter your zip code and get
  customized weather information.

33
Personalization Techniques Cookies

• Cookies Used for Personalization
• For example, if you visit msn.com, it offers you
  the ability to "change content/layout/color." It
  also allows you to enter your zip code and get
  customized weather information.

34
Personalization Techniques Cookies

• Cookies Used for Personalization
• E-commerce sites can implement things like
  shopping carts and "quick checkout" options.
• The cookie contains an ID and lets the site keep
  track of you as you add different things to your
  cart. Each item you add to your shopping cart is
  stored in the site's database along with your ID
  value. When you check out, the site knows what is
  in your cart by retrieving all of your selections
  from the database.
• It would be impossible to implement a convenient
  shopping mechanism without cookies or something
  like them.
• Note that what the database is able to store is
  things you have selected from the site, pages you
  have viewed from the site, information you have
  given to the site in online forms, etc. All of
  the information is stored in the site's database,
  and in most cases, a cookie containing your
  unique ID is all that is stored on your computer

35
Personalization Techniques Cookies

• Cookies Used for Personalization
• Problems
• People often share machines (On something like a
  Windows NT machine or a UNIX machine that uses
  accounts properly, this is not a problem. The
  accounts separate all of the users' cookies)
• Cookies get erased.
• Multiple machines - People often use more than
  one machine.

36
Personalization Techniques Cookies

• Cookies Used for Personalization
• Problems - Solutions
• There are probably not any easy solutions to
  these problems, except asking users to register
  and storing everything in a central database.
• When you register with a sites registration
  system, the problem is solved in the following
  way The site remembers your cookie value and
  stores it with your registration information. If
  you take the time to log in from any other
  machine (or a machine that has lost its cookie
  files), then the server will modify the cookie
  file on that machine to contain the ID associated
  with your registration information. You can
  therefore have multiple machines with the same ID
  value.

37
Personalization Techniques Cookies

• Cookies and Privacy
• Cookies are benign text files, they provide lots
  of useful capabilities on the Web.
• What is the problem?

38
Personalization Techniques Cookies

• Cookies and Privacy
• Web sites can sell your personal information.
• They can track not only your purchases, but also
  the pages that you read, the ads that you click
  on, etc. If you then purchase something and enter
  your name and address, the site potentially knows
  a lot about you. This makes targeting much more
  precise, and that makes a lot of people
  uncomfortable.

39
Personalization Techniques Cookies

• Cookies and Privacy
• There are certain infrastructure providers that
  can actually create cookies that are visible on
  multiple sites.
• DoubleClick is the most famous example of this

40
Personalization Techniques Cookies

• Personalization and privacy conflict how to
  resolve it?
• More later

41
Components of a Personalization System

• A
• Choice Set. The choice set represents the
  universe of content, products, media, etc. that
  are available to be recommended to users.
• Depending on the type of personalization
  technology employed, information gathered about a
  choice set can include anything from basic item
  ID or stock keeping unit (SKU) numbers to
  detailed lists of attributes concerning each item.

42
Components of a Personalization System

• B
• Preference Capture. User preferences for content
  can be captured in a number of ways.
• Users can rate content, indicating their level of
  interest in products or content that are
  recommended to them.
• Users can fill out questionnaires, providing
  general preference information that can be
  analyzed and applied across a content domain(s).
• And, where privacy policies allow, a
  personalization system can observe a users
  choices and/or purchases and infer preferences
  from those choices.

43
Components of a Personalization System

• C
• Preference Profile. The user preference profile
  contains all the information that a
  personalization system knows about a user.
• The profile can be as simple as a list of
  choices, or ratings, made by each user.
• A more sophisticated profile might provide a
  summary of each users tastes and preferences for
  various attributes of the content in the choice
  set.

44
Components of a Personalization System

• D
• Recommender. The recommender algorithm uses the
  information regarding the items in a choice set
  and a users preferences for those items to
  generate personalized recommendations.
• The quality of recommendations depends on how
  accurately the system captures a users
  preferences as well as its ability to accurately
  match those preferences with content in the
  choice set.

45
Personalization Techniques CF

• Collaborative Filtering (CF)
• Collaborative filtering (CF) is the method of
  making automatic predictions (filtering) about
  the interests of a user by collecting taste
  information from many users (collaborating).
• The underlying assumption of CF approach is that
  
• Those who agreed in the past tend to agree again
  in the future. For example, a collaborative
  filtering or recommendation system for music
  tastes could make predictions about which music a
  user should like given a partial list of that
  user's tastes (likes or dislikes).

46
Personalization Techniques CF

• User-based Collaborative Filtering 1st
  generation personalization technology

Target user selects item A
CF finds all users who have selected A
B C are the most frequently selected items by
all users who have selected A. CF recommends B
C to the target user
47
Personalization Techniques CF

• Item-based Collaborative Filtering 2nd
  generation, more scalable personalization
  technology
• Item based filtering is another method of
  collaborative filtering in which items are rated
  and used as parameters instead of users.

48
Personalization Techniques CF
User selected item A
CF finds all users who have rated item A
CF finds items having the most similar ratings by
all users
49
Personalization Techniques CF

• Collaborative Filtering Algorithms
• Item-based collaborative filtering proceeds in an
  item-centric manner
• Build an item-user matrix

v(i, j) is the vote/rating user i has put on item
j
50
Personalization Techniques CF

• Collaborative Filtering Algorithms
• Item-based collaborative filtering proceeds in an
  item-centric manner
• Using the matrix, and the data on the current
  user, infer his taste

v(i, j) is the vote/rating user i has put on item
j
51
Personalization Techniques CF

• Collaborative Filtering Algorithms Memory based
  Algorithms
• If user i has voted/rated on Ni items, then the
  mean vote for user i is

52
Personalization Techniques CF

• Collaborative Filtering Algorithms Memory based
  Algorithms
• The predicted vote of the active (target) user a
  for item j, p(a, j), is a weighted sum of the
  votes of other users (an active user is the one
  we want to predict his/her taste)
• where M is the number of users in the
  collaborative filtering database with nonzero
  weights. The weights w(a, i) can reflect
  distance, correlation, or similarity between each
  user i and the active user. k is a normalizing
  factor such that the absolute values of the
  weights sum to unity

53
Personalization Techniques CF

• Collaborative Filtering Algorithms Memory based
  Algorithms
• Computing the weight Correlation based Approach
• where the summations over j are over the items
  for which both users a and i have recorded votes

54
Personalization Techniques CF

• Collaborative Filtering Algorithms Memory based
  Algorithms
• k is a normalizing factor such that the absolute
  values of the weights sum to unity

55
Personalization Techniques CF

• Collaborative Filtering Algorithms A worked
  example

56
Personalization Techniques CF

• Collaborative Filtering Algorithms A worked
  example

m(1) (0.511.5)/3 1 m(2) (0.51.5)/2 1
- m(1)
- m(2)
Step 1
57
Personalization Techniques CF

• Collaborative Filtering Algorithms A worked
  example

W(2,1)(-0.5)(-0.5)0(0.5)/sqrt(-0.5)2(-0.
5)2(0.5)202 0.25/sqrt(0.125) 0.707
- m(1)
- m(2)
Step 2
58
Personalization Techniques CF

• Collaborative Filtering Algorithms A worked
  example

W(2,1)(-0.5)(-0.5)0(0.5)/sqrt(-0.5)2(-0.
5)2(0.5)202 0.25/sqrt(0.125) 0.707
Step 3
59
Personalization Techniques CF

• Collaborative Filtering Algorithms A worked
  example

W(2,1)(-0.5)(-0.5)0(-0.5)/sqrt(-0.5)2(-0
.5)2(-0.5)202 0.25/sqrt(0.125) 0.707
Step 4
p(2,3) m(2) kw(2,1)v(1,3)-m(1)) 1
(1/0.707)0.707(1.5-1) 1.5
60
Personalization Techniques CF

• Collaborative Filtering Algorithms Another
  example

61
Personalization Techniques CF

• The Slope one collaborative filtering algorithm

The slope one schemes take into account both
information from other users who rated the same
item and from the other items rated by the same
user. However, the schemes also rely on data
points that fall neither in the user array nor in
the item array (e.g. user As rating of item I),
but are nevertheless important information for
rating prediction. Much of the strength of the
approach comes from data that is not factored in.
Specifically, only those ratings by users who
have rated some common item with the predictee
user and only those ratings of items that the
predictee user has also rated enter into the
prediction of ratings under slope one schemes.
Basis of SLOPE ONE schemes User As ratings of
two items and User Bs rating of a common item is
used to predict User Bs unknown rating
62
Personalization Techniques CF

• Slope one collaborative filtering algorithm

We try to use the ratings of item k to predict
the rating a user may put on item l.
63
Personalization TechniquesCF

• Slope one collaborative filtering algorithm

Formally, given two evaluation arrays, v (i,
k) and v(i, l), with i 1, . . . , n, are the
ratings user i gives to item k and l. We search
for the best predictor of the form f (x) x b
to predict v(i, l) from v(i, k) by minimizing
..
..
..
..
We try to use the ratings of item k to predict
the rating a user may put on item l.
64
Personalization Techniques CF

• Slope one collaborative filtering algorithm

Deriving with respect to b and setting the
derivative to zero, we get Summation is over
all is who have voted both k and l. In other
words, the constant b must be chosen to be the
average difference between the two arrays.
..
..
..
..
We try to use the ratings of item k to predict
the rating a user may put on item l.
65
Personalization Techniques CF

• Slope one collaborative filtering algorithm

..
..
The prediction of v(j, l) from v(j, k)
..
..
We try to use the ratings of item k to predict
the rating a user may put on item l.
66
Personalization Techniques CF

• Slope one collaborative filtering algorithm

The prediction of v(j, l) from v(j, k)
User 1
User i
v(i,k)
v(i,l)
This was the prediction from ratings of item
k What should we do if we have k gt 1?
v(j,k)
User j
v(j,l) ?
User M
Item k
Item l
Item 1
Item N
We try to use the ratings of item k to predict
the rating a user may put on item l.
67
Personalization Techniques CF

• Slope one collaborative filtering algorithm A
  worked example

• Prediction of v(2,3) from item 1
• b(1, 3) v(1,3)-v(1,1) 1.5 0.5 1
• V(2,3) v(2,1) b(1,3) 0.5 1 1.5
• Prediction of v(2,3) from item 2
• b(2, 3) v(1,3)-v(1,2) 1.5 1 0.5
• V(2,3) v(2,2) b(2,3) 0.5 0.5 1
• The final prediction score should be the average
  of the two prediction scores
• ? (1.5 1)/2 1.25

68
Personalization Techniques CF

• Slope one collaborative filtering algorithm A
  worked example

From item 1 b(1, 3) v(1, 3)-v(1,1)v(3,3)-
v(3,1)/2 (1.5-0.5)(0.9-0.3)/2 0.8 ?
(0.5 0.8) 1.3
69
Personalization Techniques CF

• Slope one collaborative filtering algorithm A
  worked example

From item 2 b(2, 3) v(1, 3)-v(1,2)v(3,3)-
v(3,2)/2 (1.5-1)(0.9-0.3)/2 0.55 ?
(1.3 1.05)/2 1.175
70
Personalization Techniques CF

• Slope one collaborative filtering algorithm A
  worked example

From both items ? (0.5 0.55) 1.05
71
Personalization Techniques CF

• Model-based Collaborative Filtering algorithm
• Cluster Models
• Bayesian Network Model
• Attributed Bayesian Choice Modelling
• Others

72
Personalization Examples

• My Yahoo! (my.yahoo.com)
• is a customized personal copy of Yahoo!
• Users can select from hundreds of modules, such
  as news, stock prices, weather, and sports
  scores, and place them on one or more Web pages.
• The actual content for each module is then
  updated automatically, so users can see what they
  want to see in the order they want to see it.
• This provides users with the latest information
  on every subject, but with only the specific
  items they want to know about.

73
Personalization Examples
74
Personalization Examples

• My Yahoo!
• Personalization often occurs inside the modules.
  For example, users can choose which TV channels
  they want to include in their TV guide in
  addition to which local cable system they use.
  Other modules are more general, for example, top
  health news.
• Not only is the content customized, but the
  layout can be customized, too.
• Some content is personalized automatically.
  Although this may seem like an oxymoron, it does
  work (according to Yahoo!). An example of such
  content is a sports module that lists the teams
  in the users area after obtaining that
  information from the user profile.
• A My Yahoo! option enables the My Yahoo! page to
  automatically update at any user-specified
  interval from 15 minutes to several hours. The
  page is always being built on-the-fly by matching
  the users preferences with the available
  content. The architecture is efficient enough to
  be able to provide this service to millions of
  people from thousands of sources changing
  thousands of times a day, using a relatively
  small number of off-the-shelf computers. The
  architecture is completely scalable. As our user
  base grows, they simply add more (similarly
  configured low-cost) hardware, eliminating the
  need for expensive hardware solutions.
• Modules can be selected from a (long) list, but
  can also be added by clicking on a button at the
  original content page. For example, every weather
  page (weather.yahoo.com) contains an add to My
  Yahoo! button, which adds that page directly to
  the users My Yahoo! page. Also, each module on a
  My Yahoo! page has an edit and a remove button,
  allowing users to manipulate their pages
  directly, without ever needing to visit an
  edit/layout screen.

75
Personalization Examples

• Yahoo! Companion
• A browsers embedded toolbar from which a user
  can directly access most of Yahoo! features from
  anywhere on the Web. In a sense, it is like a
  mini My Yahoo! That takes a small space at the
  top of the page, and is always with you. One can
  customize the look and makeup of that toolbar at
  any time, and changes stay with users even if
  they switch to a different computer.

76
Personalization Examples

• iGoogle

77
Personalization Examples

• Amazon

78
Personalization and Privacy

• Personalized interaction and user modeling have
  significant privacy implications, due to the fact
  that personal information about users needs to be
  collected to perform personalization.
• The privacy of an individual's personal data on
  the Internet is a top concern for business,
  government, media and the public.
• Opinion surveys consistently show that privacy
  concerns are a leading impediment to the further
  growth of Web-based commerce.

79
Personalization and Privacy

• Initial efforts by Web sites to publicly disclose
  their privacy policies have had some impact.
• But these policies are often difficult for users
  to locate and understand, too lengthy for users
  to read, and change frequently without notice.

80
Platform for Privacy Preferences Project (P3P)

• http//www.w3.org/P3P/
• P3P 1.0, developed by the World Wide Web
  Consortium, is emerging as an industry standard
  providing a simple, automated way for users to
  gain more control over the use of personal
  information on Web sites they visit.

81
Platform for Privacy Preferences Project (P3P)

• At its most basic level, P3P is a standardized
  set of multiple-choice questions covering all the
  major aspects of a Web site's privacy policies.
• Taken together, they present a clear snapshot of
  how a site handles personal information about its
  users.
• P3P-enabled Web sites make this information
  available in a standard, machine-readable format.
  
• P3P-enabled browsers can "read" this snapshot
  automatically and compare it to the consumer's
  own set of privacy preferences.

82
Platform for Privacy Preferences Project (P3P)

• P3P enhances user control by putting privacy
  policies where users can find them, in a form
  users can understand, and, most importantly,
  enables users to act on what they see.
• In short, the P3P specification brings ease and
  regularity to Web users wishing to decide whether
  and under what circumstances to disclose personal
  information.
• User confidence in online transactions increases
  as they are presented with meaningful information
  and choices about Web site privacy practices.

83
Platform for Privacy Preferences Project (P3P)

• The P3P Vocabulary
• Nine aspects of online privacy are covered by
  P3P. Five topics detail the data being tracked by
  the site.
• Who is collecting this data?
• Exactly what information is being collected?
• For what purposes?
• Which information is being shared with others?
• And who are these data recipients?

84
Platform for Privacy Preferences Project (P3P)

• The P3P Vocabulary
• The remaining four topics explain the site's
  internal privacy policies.
• Can users make changes in how their data is used?
  
• How are disputes resolved?
• What is the policy for retaining data?
• And finally, where can the detailed policies be
  found in "human readable" form?

85
Platform for Privacy Preferences Project (P3P)

• How It Works
• P3P enables Web sites to translate their privacy
  practices into a standardized, machine-readable
  format (Extensible Markup Language XML) that can
  be retrieved automatically and easily interpreted
  by a user's browser. Translation can be performed
  manually or with automated tools. Once completed,
  simple server configurations enable the Web site
  to automatically inform visitors that it supports
  P3P.

86
Platform for Privacy Preferences Project (P3P)

• How It Works
• On the user side, P3P clients automatically fetch
  and read P3P privacy policies on Web sites. A
  user's browser equipped for P3P can check a Web
  site's privacy policy and inform the user of that
  site's information practices. The browser could
  then automatically compare the statement to the
  privacy preferences of the user, self-regulatory
  guidelines, or a variety of legal standards from
  around the world. P3P client software can be
  built into a Web browser, plug-ins, or other
  software

87
Platform for Privacy Preferences Project (P3P)

• Participants, Supporters, Developers

88
References/Further Reading

• S. Braynov, Personalization and customization
  technologies, Dept. of Computer Science and
  Eng., State University of New York at Buffalo
• Review of Personalization Technologies, Technical
  Brief, ChoiceStream, Inc. http//www.choicestream.
  com/
• D. Lemire and A. Maclachlan, Slope One
  Predictors for Online Rating-Based Collaborative
  Filtering, SIAM Data Mining (SDM05), Newport
  Beach, California, April 21-23, 2005.
• J. S. Breese, D. Heckerman and C. Kadie,
  Empirical Analysis of Predictive Algorithms for
  Collaborative Filtering, Technical Report,
  MSR-TR-98-12, Microsoft Corporation
• G. Linden, B. Smith and J York, Amazon.com
  Recommendations item to item collaborative
  filtering, IEEE Internet computing,
  January/February 2003
• http//www.w3.org/P3P/
• Further Reading
• Bamshad Mobasher, Robert Cooley, and Jaideep
  Srivastava, Automatic Personalization Based on
  Web Usage Mining, Communications of ACM, vol. 43,
  no. 8, August 2000
• S. Stewart and J. Davies, User Profiling
  Techniques A Critical Review, Proceedings of the
  19th Annual BCS-IRSG Colloquium on IR Research,
  Aberdeen, Scotland, 8-9 April 1997

89
Expected Learning Outcomes

• You should have a basic understanding of
  personalization in the context of the Internet.
• You should have good knowledge of the various
  components of a personalization system and their
  roles
• You should be aware of techniques such as Cookie
  can implement personalization and be able to
  discuss how it works
• You should have a good understanding of
  Collaborative Filtering techniques and be able to
  implement the algorithms explained in the
  lectures
• You should be aware of how personalization
  techniques are used in various real world
  applications
• You should be able to discuss the strengths and
  limitations of collobrative filtering
• You should be able to discuss why implementing
  personalization techniques will cause privacy
  problems
• You should be aware of the P3P project and how it
  proposes to resolve the conflict between
  personalization and privacy