1
Personalized Search
By James Pitkow

• ..a contextual computing approach may prove a
  breakthrough in personalized search efficiency..

Ashley Twichell Emily Lang
2
Contextual Computing

• Refers to the enhancement of a users
  interactions by understanding the user, the
  context, and the applications and information
  being used
• Its about actively adapting the computational
  environment - for each and every user - at each
  point of computation
• Focuses on understanding the information
  consumption patterns of each user, the various
  information foraging strategies and applications
  they employ, and the nature of the information
  itself
• A shift from consensus relevancy (relevancy for
  entire population used for every person) to
  personal relevancy (relevancy is determined for
  each individual)
• This shift to personal relevancy decreases the
  time it takes people to find information

3
Review of Information Retrieval

• Content-based approaches - using language to
  match a query with results - this approach
  doesnt help users determine which results are
  actually worth reading
• Author-relevancy techniques - using citation and
  hyperlinks - sometimes presents the problem of
  authoring bias and/or ranking bias (results
  that are valued by authors are not necessarily
  those valued by the entire population)
• Usage rank - this leverages the actions of users
  to compute relevancy - the usage rank is
  computed from the frequency, recency, and/or
  duration of interaction by users - usage ranks
  allow for changes in relevancy over time to be
  determined
• All of the above techniques measure relevance as
  a function of the entire population of users
• This does not acknowledge that relevance is
  relative for each user
• There needs to be a way to take into account
  that different people find different things
  relevant and that peoples interests and
  knowledge change over time - personal relevance

4
The Outride Approach

• In order to personalize search, we need to
  combine at least two different computational
  techniques - contextualization and
  individualization
• Contextualization - the interrelated conditions
  that occur within an activity..includes factors
  like the nature of information available, the
  information currently being examined, and the
  applications in use
• Individualization - the totality of
  characteristics that distinguishes an
  individual.. Uses the users goals, prior and
  tacit knowledge, past information-seeking
  behaviors

5

• Main ways to personalize a search are query
  augmentation and result processing
• Query augmentation - when a user enters a query,
  the query can be compared against the contextual
  information available to determine if the query
  can be refined to include other terms
• Query augmentation can also be done by computing
  the similarity between the query term and the
  user model - if the query is on a topic the user
  has previously seen, the system can reinforce the
  query with similar terms
• This more concise query is then shown to the user
  and submitted to a search engine for processing
• Once the query has been augmented and processed
  by the search engine, the results can be
  individualized
• The results being individualized - this means
  that the information is filtered based upon
  information in the users model and/or context
• The user model can re-rank search results based
  upon the similarity of the content of the pages
  in the results and the users profile

6

• Another processing method is to re-rank the
  results based upon the frequency, recency, or
  duration of usage..providing users with the
  ability to identify the most popular, faddish and
  time-consuming pages theyve seen
• Have Seen, Have Not Seen - this features allows
  new information to be identified and return to
  information already seen

7
The Outride Personalized Search System

• Designed to be a generalized architecture for
  the personalization of search across a variety of
  information ecologies
• The Outride client can be integrated into the
  sidebar of the Internet - it supports direct
  manipulation and has access to all user
  interactions
• Sidebar is split up into four separate
  information spaces - Personal (personal hierarchy
  of each users links), Directory (a catalog of
  links), History (users surf history), Web
  (search results from the entire Web)
• The user models are computed from the content in
  these information spaces in the sidebar

8
Testing Methodology and Results

• Outride used eTesting Lbs to design a series of
  test to measure if the Outride system actually
  succeeded in making searches faster and easier to
  complete
• The elapsed time to successfully complete a
  search and the number of interface actions (mouse
  clicks/number of entries entered) were used as
  the measurements
• Participants performed 12 search tasks with
  Outride and a different search engine
• Default user model was used for all participants
• Participants found the answers more quickly with
  Outride than with any other search engine - on
  average, participants took 39 seconds to complete
  the tasks using Outride and 75 seconds using
  Google
• Participants also needed fewer interface actions
  when using Outride - 11 when using Outride and 21
  using the other search engine

9
Issues/problems with the experiments

• Some of the scenarios contained tasks directly
  supported by the functionality provided by the
  Outride system, creating an advantage against the
  other search engines
• Default profiles were used, instead of
  individualized profiles - therefore, it did not
  represent the test participant actual surfing
  patterns, nor were the participants intimately
  familiar with the content of the profiles
• Despite these issues, the magnitude of the
  difference between the Outride system and the
  other engines is compelling

10
Future Directions

• One problem is modeling a users changing
  interests over time
• However, carefully designed interfaces can help
  alleviate inaccurate personalization and allow
  users to control the extent of the
  personalization
• Privacy issues are a problem since it is a system
  that stores models based upon users interactions
  with information

11
Experience with Personalization on Yahoo!

• When designing Web personalization products,
  make sure you address all your users
• By Udi Manber, Ash Patel, and John Robison

12
Overview

• This article discusses three different examples
  of personalization on Yahoo! Including
• My Yahoo!
• Yahoo! Companion
• Inside Yahoo! Search

13
My Yahoo!

• My Yahoo! Is a customized personal copy of Yahoo!
• Users select from various models such as news,
  stock prices, weather, and sports scores to put
  on their Web page.
• Provides users with the latest information on
  every subject, but with only the specific items
  they want to know about.

14
My Yahoo! Features

• Personalization
• Users can do such things as chose certain TV
  channels to put in their TV Guide
• Customized Content
• Example of this is a sports module that lists the
  teams in the users area after obtaining that
  information from the users profile.
• Automatic Updates
• A My Yahoo! Option allows this page to
  automatically update at any user-specified
  interval from 15 minutes to several hours
• Original Module Ability
• Modules can be selected from a long list, but can
  also be added by clicking on a button at the
  original content page.
• Each module on a My Yahoo! Page also has an edit
  and remove button, allowing users to manipulate
  their pages directly, without ever needing to
  visit an edit/layout page.

15
Yahoo! Companion

• A browsers embedded toolbar from which a user
  can directly access most of Yahoo! features from
  anywhere on the Web.
• Like a mini My Yahoo! that takes a small space at
  the top of the page is always with you.

16
Yahoo! Companion Features

• The user interface is similar to any other
  bookmark feature, but the difference is the
  bookmarks are kept on the server (not simply on
  the specific computer)
• Therefore changes that users make to their
  toolbar will stay with users even if they switch
  to a different computer
• Users have the ability to chose from several
  toolbars (such as a regular one a stock market
  one) and change them at any time

17
Inside Yahoo!

• Yahoo! like many other search engines tries to
  personalize searches using information it is able
  to obtain from the user
• It would be impossible for Yahoo! to customize
  every search.

18
Yahoo! Search Example

• If a user searches for the name of current movie,
  Yahoo will show results for Yahoo! Movies, show
  an image for the movie, the cast, and a pointer
  to a page with current show times
• If the user had looked at showtimes on a page
  previously and entered a zip code, Yahoo! can now
  use that information to show the user movie times
  in his or her own are

19
Solving Privacy Problems

• Any company that collects private information
  must guard that information with its life.
• Personal information about Yahoo! Is maintained
  in a specially designed User Database (UDB) which
  was built on Yahoo!s own customized software.
• Yahoo! has data replication and distribution
  capabilities allowing them to replicate and
  distribute the UDB over secure links to remote
  locations in Asia and Europe
• Yahoo! has enlisted a security-audit company to
  evaluate our procedures periodically and suggest
  necessary changes, as well as employ several
  internal people devoted solely to privacy and
  security issues.

20
User Interface

• The issue of usability focuses mostly on the
  issue of predictability
• Personalization features that learn what user
  want and attempt to satisfy them are hotly
  debated
• A weakness in these personalization features is
  unpredictability
• Example A lot of people do not want customized
  news, they want just the same news as everyone
  else
• Also getting news about cancer because a user
  some medical journal on cancer in the past can
  confuse the user and even jeopardize user trust
  and raise serious privacy concerns in the users
  mind
• Any effective personalization feature should
  encourage experimentation.

21
Observations/Lessons Learned

• Most users take what is given to them and never
  customize.
• Even though companies like Yahoo! offer
  customized pages for users, a great deal of
  effort must still go into the default page.
• Companies should never underestimate power users
• Customization should follow you as much as
  possible
• People generally dont understand the concept of
  customization
• Make sure you address all your users
• Learn from users

22
Conclusion

• Too many attempts have been made without
  sufficient regard to what people really want,
  what they can use, and how best it should fit
  their needs.
• A major challenge to large-scale personalization
  is to lower the entry bar, making it easier for
  less-experienced users to customize their pages,
  and making it clear to novices that customization
  is possible.