Searching the Social Web The Challenges of  Socially-Connected Search IR Leadership Seminar 2008 / Ofer Egozi
The problem… What to choose? Whom to trust??...
… The solution? What to choose? Whom to trust??...
<ul><li>Leveraging the Social Graph in web search </li></ul><ul><ul><li>Focused crawling </li></ul></ul><ul><ul><li>Person...
<ul><li>Trusted results </li></ul><ul><ul><li>Friends qualify content/sources </li></ul></ul><ul><ul><li>Potential contact...
Outline <ul><li>Approaches to Social Search </li></ul><ul><li>The Social Graph </li></ul><ul><li>Graph-Related Challenges ...
Humans in the loop <ul><li>Search =  crawl  +  index  +  rank  +  query </li></ul><ul><li>Crawling (Dmoz, Mahalo) </li></u...
A Taxonomy of Social Search Aggregated Personalized Network-based Behavior- based ? ?
The Social Graph <ul><li>A directed, cyclic graph </li></ul><ul><ul><li>Nodes are people (identities) </li></ul></ul><ul><...
Social Graph in Research <ul><li>Extraction from interactions </li></ul><ul><ul><li>Email (Van Alstyne et al. 2003), Chat ...
So first we need to draw the graph…
Social Graph - challenges <ul><li>Social graph nodes </li></ul><ul><ul><li>Identities/relations across networks </li></ul>...
 
Social Graph - challenges <ul><li>Social graph nodes </li></ul><ul><ul><li>Identities/relations across networks </li></ul>...
So now we’ve mapped the social graph… … and attached each node with its content…
… can we finally go fetch?
S-C Search - challenges <ul><li>Must  build  a search engine… </li></ul><ul><ul><li>Store graph, attach content to nodes <...
Not in Google’s / Yahoo!’s top-1000!… (dominated by authorities)
S-C Search - challenges <ul><li>Must  build  a search engine… </li></ul><ul><ul><li>Store graph, attach content to nodes <...
Socially-Connected Search <ul><li>What are the enablers?  </li></ul><ul><ul><li>Social networks </li></ul></ul><ul><ul><li...
Upcoming SlideShare
Loading in …5
×

Searching The Social Web

3,441
-1

Published on

A talk on social search as implemented in Delver, presented in IBM-HRL IR technologies/social search seminar, 16/12/2008.

Published in: Technology
0 Comments
4 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
3,441
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
39
Comments
0
Likes
4
Embeds 0
No embeds

No notes for slide
  • Talk in IBM-HRL Information Retrieval seminar 16/12/2008
  • Searching The Social Web

    1. 1. Searching the Social Web The Challenges of Socially-Connected Search IR Leadership Seminar 2008 / Ofer Egozi
    2. 2. The problem… What to choose? Whom to trust??...
    3. 3. … The solution? What to choose? Whom to trust??...
    4. 4. <ul><li>Leveraging the Social Graph in web search </li></ul><ul><ul><li>Focused crawling </li></ul></ul><ul><ul><li>Personalized ranking </li></ul></ul><ul><li>Delver is a first-mover </li></ul>The solution? Socially- connected search: What to choose? Who to trust??...
    5. 5. <ul><li>Trusted results </li></ul><ul><ul><li>Friends qualify content/sources </li></ul></ul><ul><ul><li>Potential contact in reach </li></ul></ul><ul><ul><li>Spam is inherently low </li></ul></ul><ul><li>Reasoning over results </li></ul><ul><ul><li>Ranking is transparent </li></ul></ul><ul><ul><li>Easier to assess relevance </li></ul></ul><ul><li>Network discovery </li></ul><ul><ul><li>Experts in my network </li></ul></ul><ul><ul><li>Serendipity </li></ul></ul>The solution? Socially- connected search: What to choose? Who to trust??...
    6. 6. Outline <ul><li>Approaches to Social Search </li></ul><ul><li>The Social Graph </li></ul><ul><li>Graph-Related Challenges </li></ul><ul><li>Search-Related Challenges </li></ul>
    7. 7. Humans in the loop <ul><li>Search = crawl + index + rank + query </li></ul><ul><li>Crawling (Dmoz, Mahalo) </li></ul><ul><li>Indexing (del.icio.us, Flickr) </li></ul><ul><li>Querying (ChaCha) </li></ul><ul><li>Ranking – that’s what we’ll discuss… </li></ul>
    8. 8. A Taxonomy of Social Search Aggregated Personalized Network-based Behavior- based ? ?
    9. 9. The Social Graph <ul><li>A directed, cyclic graph </li></ul><ul><ul><li>Nodes are people (identities) </li></ul></ul><ul><ul><li>Edges are relations between them </li></ul></ul><ul><li>Large portion is public on social networks </li></ul><ul><li>A lot isn’t – cellular, email, non-digital </li></ul><ul><li>Emerging web standards </li></ul><ul><ul><li>OpenID/hCard – identifier/identity </li></ul></ul><ul><ul><li>Contact APIs/PoCo/XFN – private/public contact lists </li></ul></ul><ul><ul><li>FB Connect – a full proprietary framework </li></ul></ul>
    10. 10. Social Graph in Research <ul><li>Extraction from interactions </li></ul><ul><ul><li>Email (Van Alstyne et al. 2003), Chat (Tuulos & Tirri 2004), IM (Lang 2004) </li></ul></ul><ul><li>Correlation with “physical” </li></ul><ul><ul><li>Bluetooth contact (Mtibaa et al. 2008) </li></ul></ul><ul><li>Security and Privacy </li></ul><ul><ul><li>Shared knowledge authentication (Toomim et al. 2008) </li></ul></ul><ul><ul><li>Graph link privacy (Xu 2008), (Korolova et al. 2008) </li></ul></ul><ul><li>Enhance IR ranking </li></ul><ul><ul><li>Index friends browse history (Mislove et al. 2006) </li></ul></ul><ul><ul><li>Rank by author centrality (Kirchhoff et al. 2008) </li></ul></ul><ul><ul><li>Rerank by sampling network click-log (Das et al. 2008) </li></ul></ul>
    11. 11. So first we need to draw the graph…
    12. 12. Social Graph - challenges <ul><li>Social graph nodes </li></ul><ul><ul><li>Identities/relations across networks </li></ul></ul>Joe friend-of follows friend-of follows Joe JJ123 Joey
    13. 14. Social Graph - challenges <ul><li>Social graph nodes </li></ul><ul><ul><li>Identities/relations across networks </li></ul></ul><ul><ul><li>Identity impersonation </li></ul></ul><ul><ul><li>Non-individual identities (groups, shared authorship…) </li></ul></ul><ul><ul><li>Privacy is an issue, even with public data </li></ul></ul><ul><li>Social graph edges </li></ul><ul><ul><li>Relation “strength” not exposed </li></ul></ul><ul><ul><li>Super nodes may dominate results </li></ul></ul><ul><ul><li>“ Politeness” relations are not filtered out </li></ul></ul><ul><ul><li>Automatic generation – double-edged sword </li></ul></ul>Joe friend-of follows friend-of follows Joe JJ123 Joey
    14. 15. So now we’ve mapped the social graph… … and attached each node with its content…
    15. 16. … can we finally go fetch?
    16. 17. S-C Search - challenges <ul><li>Must build a search engine… </li></ul><ul><ul><li>Store graph, attach content to nodes </li></ul></ul><ul><ul><li>Reranking will not do, this is the long tail </li></ul></ul>
    17. 18. Not in Google’s / Yahoo!’s top-1000!… (dominated by authorities)
    18. 19. S-C Search - challenges <ul><li>Must build a search engine… </li></ul><ul><ul><li>Store graph, attach content to nodes </li></ul></ul><ul><ul><li>Reranking will not do, this is the long tail </li></ul></ul><ul><ul><li>Scale well, including graph functions </li></ul></ul><ul><li>Personalized graph-based rank </li></ul><ul><ul><li>Integrate content-based with static ranking </li></ul></ul><ul><ul><li>Use web graph structure, like PageRank etc. </li></ul></ul><ul><ul><li>Network is egocentric , unlike PageRank </li></ul></ul>
    19. 20. Socially-Connected Search <ul><li>What are the enablers? </li></ul><ul><ul><li>Social networks </li></ul></ul><ul><ul><li>Users’ content boom </li></ul></ul><ul><li>What can be achieved? </li></ul><ul><ul><li>Search-based access to network content </li></ul></ul><ul><ul><li>Trusted and transparent social ranking </li></ul></ul><ul><li>What are the challenges? </li></ul><ul><ul><li>Fragmented social graph </li></ul></ul><ul><ul><li>Personal-network ranking </li></ul></ul>Thank you! http://www.delver.com
    1. A particular slide catching your eye?

      Clipping is a handy way to collect important slides you want to go back to later.

    ×