personalization and search jaime teevan microsoft research
TRANSCRIPT
![Page 1: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/1.jpg)
Personalization and Search
Jaime TeevanMicrosoft Research
![Page 2: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/2.jpg)
Information Retrieval
User Context
Domain Context
Task/Use Context
Query Words
Ranked List
Query Words
Ranked List
> Query
![Page 3: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/3.jpg)
Personalization and Search
• Measuring the value of personalization– Do people’s notions of relevance vary?
• Understanding the individual– How can we model a person’s interests?
• Calculating personal relevance– How can we use the model to measure relevance?
• Other ways to personalize search– What other aspects can we personalize?
![Page 4: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/4.jpg)
• Measuring the value of personalization– An example– Lots of relevant results ranked low– Best group ranking v. individual ranking
• Understanding the individual• Calculating personal relevance• Other ways to personalize search
Personalization and Search
![Page 5: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/5.jpg)
![Page 6: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/6.jpg)
Relevant Content Ranked Low
1 2 3 4 5 6 7 8 9 10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
Rank
Highly Relevant
Relevant
Irrelevant
![Page 7: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/7.jpg)
Best Rankings
…
…
…
![Page 8: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/8.jpg)
Potential for Personalization
1 2 3 4 5 60.600000000000001
0.700000000000001
0.800000000000001
0.900000000000001
1
Individual
Number of People
DC
G
![Page 9: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/9.jpg)
Potential for Personalization
1 2 3 4 5 60.600000000000001
0.700000000000001
0.800000000000001
0.900000000000001
1
Individual Group
Number of People
DC
G
Potential for personalization
![Page 10: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/10.jpg)
Potential for Personalization
1 2 3 4 5 60.600000000000001
0.700000000000001
0.800000000000001
0.900000000000001
1
Individual Group Web
Number of People
DC
G
Potential for personalization
![Page 11: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/11.jpg)
Overview
• Measuring the value of personalization• Understanding the individual– Gather information beyond the query– Explicit v. implicit– Client-side v. server-side
• Calculating personal relevance• Other ways to personalize search
![Page 12: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/12.jpg)
Learning More Explicitly v. Implicitly
• Explicit– User shares more about query intent– User shares more about interests– Hard to express interests explicitly
berkeley
Query Words
restaurants
California or Illinois?Fancy or low key?
![Page 13: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/13.jpg)
Learning More Explicitly v. Implicitly
• Explicit– User shares more about query intent– User shares more about interests– Hard to express interests explicitly
Arts Business Computers
Games Health Home
Kids and Teens News Recreation
Reference Regional Science
Shopping Society Sports
Rock climbing?Tobacco and guns
Intellectual property?
![Page 14: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/14.jpg)
Learning More Explicitly v. Implicitly
• Explicit– User shares more about query intent– User shares more about interests– Hard to express interests explicitly
• Implicit– Query context inferred– Profile inferred about the user– Less accurate, needs lots of data
![Page 15: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/15.jpg)
Profile Information
• Behavior-based– Click-through– Personal PageRank
• Content-based– Categories– Term vector
[topic: computers]
[computers: 2, microsoft: 1, click: 4, what: 3, tablet: 1]
![Page 16: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/16.jpg)
Profile Information
• Behavior-based– Click-through– Personal PageRank
• Content-based– Categories– Term vector
Server information
• Web page index• Link graph• Group behavior
![Page 17: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/17.jpg)
Server-Side v. Client-Side Profile
• Server-side– Pros: Access to rich Web/group information– Cons: Personal data stored by someone else
• Client-side– Pros: Privacy– Cons: Need to approximate Web statistics
• Hybrid solutions– Server sends necessary Web statistics– Client sends some profile information to server
![Page 18: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/18.jpg)
Match Individual to Group
• Can use groups of people to get more data• Back off from individual group all• Collaborative filtering
![Page 19: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/19.jpg)
Overview
• Measuring the value of personalization• Understanding the individual• Calculating personal relevance– Behavior-based example– Content-based example
• Other ways to personalize search
![Page 20: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/20.jpg)
Behavior-Based Relevance
• People often want to re-find• People have trusted sites• Boost previously viewed URLs
![Page 21: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/21.jpg)
Content-Based Relevance
• Explicit relevance feedback– Mark documents relevant– Used to re-weight term frequencies
![Page 22: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/22.jpg)
Content-Based Relevance
N
ni
ri
R(ri+0.5)(N-ni-R+ri+0.5)
(ni-ri+0.5)(R-ri+0.5)wi = log
Score = Σ tfi * wi
(N)
(ni)wi = log
World
![Page 23: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/23.jpg)
Content-Based Relevance
• Explicit relevance feedback– Mark documents relevant– Used to re-weight term frequencies
• Lots of information about the user– Consider read documents relevant– Use to re-weight term frequencies
![Page 24: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/24.jpg)
Content-Based Relevance
ri R
N
ni
ri
R(ri+0.5)(N-ni-R+ri+0.5)
(ni-ri+0.5)(R-ri+0.5)wi = log
Score = Σ tfi * wi
(N)
(ni)wi = log
(ri+0.5)(N’-n’i-R+ri+0.5)
(n’i-ri+0.5)(R-ri+0.5)wi = log
Where: N’ = N+R, ni’ = ni+ri
World
Client
![Page 25: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/25.jpg)
Personalization Performance
• Personalized search hard to evaluate• Mostly small improvements despite big gap• Identify ambiguous queries– Personalize: “berkeley”– Don’t personalize: “uc berkeley homepage”
• Identify easily personalized queries– Re-finding queries
![Page 26: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/26.jpg)
Other Ways to Personalize
• Measuring the value of personalization• Understanding the individual• Calculating personal relevance• Other ways to personalize search– Match expectation for re-finding queries– Personalized snippets
![Page 27: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/27.jpg)
Ranking Results for Re-Finding
![Page 28: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/28.jpg)
Ranking Results for Re-Finding
![Page 29: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/29.jpg)
People Don’t Notice Change
![Page 30: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/30.jpg)
People Don’t Notice Change
![Page 31: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/31.jpg)
People Don’t Notice Change
![Page 32: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/32.jpg)
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company. Some wine companies own many wineries. Besides wine making equipment ...
en.wikipedia.org/wiki/Winery
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company. Some wine companies own many wineries. Besides wine making equipment ...
en.wikipedia.org/wiki/Winery
Snippets to Support Re-FindingQuery: “winery”
If the person has visited the page before:
Last visit: November 14, 2007
![Page 33: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/33.jpg)
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company. Some wine companies own many wineries. Besides wine making equipment ...
en.wikipedia.org/wiki/Winery
Winery - Wikipedia, the free encyclopediaNew content: It has been suggested that Winery wastewater be merged into this article or section.
en.wikipedia.org/wiki/Winery
Snippets to Support Re-FindingQuery: “winery”
If the person has visited the page before:
Last visit: November 14, 2007
![Page 34: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/34.jpg)
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company. Some wine companies own many wineries. Besides wine making equipment ...
en.wikipedia.org/wiki/Winery
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company… For example, in Maui there is a pineapple winery. …
en.wikipedia.org/wiki/Winery
Interest-Based SnippetsQuery: “winery”
If the person is interested in Maui:
![Page 35: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/35.jpg)
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company. Some wine companies own many wineries. Besides wine making equipment ...
en.wikipedia.org/wiki/Winery
Winery - Wikipedia, the free encyclopediaA winery is a building or property that produces wine, or a business involved in the production of wine, such as a wine company… For example, in Maui there is a pineapple winery. …
en.wikipedia.org/wiki/Winery
Interest-Based SnippetsQuery: “winery”
If the person is interested in Maui:
![Page 36: Personalization and Search Jaime Teevan Microsoft Research](https://reader036.vdocuments.site/reader036/viewer/2022062421/56649ce15503460f949ac84c/html5/thumbnails/36.jpg)
Summary
• Measuring the value of personalization– There’s a big gap between group and individual
• Understanding the individual– Building a profile, explicit v. implicit
• Calculating personal relevance– Relevance feedback, boost click through
• Other ways to personalize search– Rank based on expectation, personalized snippets