Research Update on WebPlaces: Application of Implicit Networks

1
Research Update on WebPlaces Application of
Implicit Networks

• Danyel Fisher
• Human-Centered Computing Retreat
• Summer, 1999

2
Overview

• WebPlaces
• Motivation
• Places on the Web
• Implicit Networks
• Ambiguities in Places
• Technical Details
• Questions?

3
Motivation

• People spend lots of time on web
• More generally, doing stuff with documents
• Alone and non-interactive
• A few sites haveadded local discussions, chat
  rooms, and similarno universals.

4
Places vs Spaces

• Erickson, Dourish and others
• Spaces (Latitude 37.77003, Longitude
  -122.446882)
• URLs. Locations. Bits.Dry. Empty. (Cold?
  Hard?)
• as opposed to
• Places (Haight and Ashbury)
• Awareness. People. Connections. Ideas.
  Stuff.Interactive. (Warm? Soft?)

5
From This, An Answer

• Add Placeness to the web.
• Leverage community experience
• Get other peoples ideas, too!
• Use Technology from
• Recommendation Systems
• Who do I want to talk to?
• Groupware
• How can I share ideas?

6
WebPlaces

• IBM Research USER group, summer 98. Intern
  project
• Who else is in the same place? How can you get in
  touch with them?
• What can you do with them?
• More than just URL want a cluster of ideas

7
Affordances in a Place

• See other users
• Chat
• Follow
• Hide

8
WebPlaces
Three users in nearby places...
9
Selection and Commands
10
Uses

• Internet Find other people with the same
  interests
• Corporate Intranet Help employees share
  knowledge.
• Academic nets and databases link users to learn
  from each other

11
Working Issues and Details

• Mappings from URLs and Users to Places
• Ambiguities in Distance Measures
• Technical Details
• Web Proxies
• Architecture of WebPlaces with WBI

12
Mapping from URL to Places

• Straight URL?
• Pure Net Topology?
• For example, Cannys web page is one link away
  from mine.
• Sophisticated Topology a site is a 2-clan of
  pages on the same server
• Information Retrieval
• Search engine words convey similarity
• Tie similarity to adjacency

13
(URL, User) ? Place

• User Paths
• Related paths through information space implies
  related interest.
• User-Defined
• Even if the system doesnt know that two pages
  are the same place, the user does. Can we
  accommodate explicit preferences?

14
Implicit Structure

• Social Networking Notion
• Graph of information and users
• Changes as users travel, communicate
• Reflects users shared interests
• The map changes as you travel and interact with
  it.
• Is this confusing?

15
Place Ambiguity Distance Measures
If the internet looks like the map on top, then
it is hard to decide which web sites are close
together and how to cluster them. If the internet
looks like the map below, than the clustering
is more obvious.
16
More Ambiguities
These illustrations show two possible
arrangements of places over a single network.
In each case, the algorithm is that everything
at a distance of 1 is in the same
place. Clearly, we need a less simple-minded
algorithm.
17
Web Proxies

• Modify web content on-the-fly
• Change HTTP requests going out
• and bits coming back in.
• Typical Uses
• Transcoding formats (Shrink image size?)
• Tracking User Interests
• Filters (e.g. Kidproofing)

18
Architecture
19
References

• Tom Ericksons Babble, CHI 99 Paper
• Judith Donath, Crowds at Web Sites
• MSRs Data Mountains, CHI 99
• Commercial Product
• ThirdVoice www.thirdvoice.com
• Paul Maglio and Rob Barrett, IBM Research
• WWW8 Conference Poster

20
Questions?

• Project Links
• WebPlaces
• Talk of the Net Melissa Virus
• Social Networks and Mailing Lists
• http//www.cs.berkeley.edu/danyelfSearch
  engine Danyel Fisher