Nocturnal: Automated Messenger-based Information Sharing and Collaboration

1
Nocturnal Automated Messenger-based Information
Sharing and Collaboration

• Coby Fernandess
• (Intern) the Hebrew University
• Dahlia Malkhi, Udi Wieder, Lidong Zhou
• MSR, Silicon Valley

2
Nocturnal
4. Personalized, collaborative search

• A collaborative personalized search tool
• Messenger-based sharing of URL-ratings

1. Exchange URL Ratings
2. Store in local DB
3. Web search results compared with local DB
Collaborative Personalized Search Results

5. Feed back satisfaction rating to local DB
3
Nocturnal
4. Personalized, collaborative search

• A collaborative personalized search tool
• Messenger-based sharing of URL-rating

1. Exchange URL Ratings
2. Store in local DB
3. Web search results compared with local DB
Collaborative Personalized Search Results

5. Feed back satisfaction rating to local DB
4
Nocturnal Vision

• Messenger as a Social Infrastructure for
  Communities.
• Decentralized social information sharing.
• Data propagates to 1-hop a users
  contacts2-hop contacts contacts3-hop
  contacts contacts contactsand so on
• Collaborative and personalized web search.
• Each user stores information locally in a
  database

5
Messenger as a Social Infrastructure for
Communities.

• Social communities are a powerful tool but social
  connections are hard to elicit from users.
• MSN messenger is an existing large social network
• Solves the bootstrapping problem.
• Yields effective communities (see poster).

6
Decentralized social information sharing.

• No central repository P2P paradigm
• Personalization Each user has distinct data
• Privacy User information propagates only to
  trusted peers
• Trust Information is pulled only from trusted
  peers

7
Collaborative and personalized web search.

• Things you can do with Nocturnal
• Share URL ratings among Messenger peers
• Collaborative web spam filtering
• Exchange reviews and recommendations
• A browser enhanced with our toolbar uses the
  local database to filter search results and to
  give peer-recommended URLs higher importance.

8
Research Challenges

• Properties of Messenger-Based Communities and
  Implications on Information Sharing and
  Dissemination.
• Trust and Reputation on a Social Network.
• Collaborative and Personalized Search
  information gathering, filtering, aggregation and
  ranking.

9
Thanks!