Welcome! Graz, 2. September, 2008 Markus Strohmaier 2008 1 - - PowerPoint PPT Presentation

▶
welcome
SMART_READER_LITE
LIVE PREVIEW

Welcome! Graz, 2. September, 2008 Markus Strohmaier 2008 1 - - PowerPoint PPT Presentation

Sep 16, 2023 460 likes •767 views

Knowledge Management Institute Informal Research Meeting RWTH Aachen / TU Graz Welcome! Graz, 2. September, 2008 Markus Strohmaier 2008 1 Knowledge Management Institute Morning Agenda 09:00 - 09:30 Opening 09:30 - 10:10 Overview of

slide-1
SLIDE 1

Knowledge Management Institute 1

Markus Strohmaier 2008

Informal Research Meeting

RWTH Aachen / TU Graz

Graz, 2. September, 2008

Welcome!

slide-2
SLIDE 2

Knowledge Management Institute 2

Markus Strohmaier 2008

Morning Agenda

09:00 - 09:30 Opening 09:30 - 10:10 Overview of research activities R. Klamma (RWTH Aachen) Ralf Klamma 10:10 - 10:40 Student presentations 1 (15min Pres + 15min Disc) Yiwei Cao “Dimensions of tagging on the Web 2.0″ 10:40 - 11:00 Break 11:00 - 11:30 Student presentations 2 (15min Pres + 15min Disc) Anna Glukhova “Traceable cooperative requirements engineering for communities of practices” 11:30 - 12:00 Student presentations 3 (15min Pres + 15min Disc) Zina Petrushyna “Web emotional intelligence” 12:00 - 12:45 Overview of research activities M. Strohmaier’s group

  • M. Strohmaier

13:00 - 15:00 Lunch Break

slide-3
SLIDE 3

Knowledge Management Institute 3

Markus Strohmaier 2008

Afternoon Agenda

15:00 - 15:30 Student Presentations 1 (15min Pres + 15min Disc) Mark Kröll “Human Goal Classification of Natural Language Text” 15:30 - 16:00 Student Presentations 2 (15min Pres + 15min Disc) Christian Körner “Constructing Large Scale Goal Graphs from Search Query Logs” 16:00 - 16:15 Student Presentations 3 (7,5min Pres + 7,5min Disc) Maida Osmic “Problem Statement: A Social Goal-Recommender System” 16:15 - 16:45 Student Presentations 4 (15min Pres + 15min Disc) Monika Schubert “Network Analysis of Software Repositories: The Eclipse Bugzilla Case” 16:45 - 17:05 Break 17:05 - 18:30 Discussion and closing 19:30 - Informal social event

slide-4
SLIDE 4

Knowledge Management Institute 4

Markus Strohmaier 2008

From Content to Intent

Aspects of Goal-Oriented Social Computing

Markus Strohmaier

Assistant Professor (Univ. Ass.) Knowledge Management Institute and Know-Center Graz Graz University of Technology, Austria web: http://www.kmi.tugraz.at/staff/markus

Mark Kröll (PhD student) Monika Schubert (PhD student) Christian Körner (MSc student) Maida Osmic (MSc student) Markus Strohmaier

Agents and Social Computation Group

Peter Prettenhofer (Collaborator)

slide-5
SLIDE 5

Knowledge Management Institute 5

Markus Strohmaier 2008

Outline

  • Motivation: Content vs. Intent
  • Goal Modeling: How can goals be modeled?
  • Goal Mining: How can goals be acquired from text?
  • Goal Representation: How can goals be related with each
  • ther collaboratively?
  • Goal Prediction: How can user goals be predicted?

In the Context of Search Query Logs and Tagging Systems

slide-6
SLIDE 6

Knowledge Management Institute 6

Markus Strohmaier 2008

Motivation

slide-7
SLIDE 7

Knowledge Management Institute 9

Markus Strohmaier 2008

slide-8
SLIDE 8

Knowledge Management Institute 10

Markus Strohmaier 2008

“How culture made your modern mind”

17 May 2008, New Scientist Print Edition, Andy Coghlan Markus Strohmaier, http://pipes.yahoo.com/mstrohm/43thingsgeosearch

Latent Intentional Communities

Agent

Resource Resource Resource

Agent Agent Agent Agent

slide-9
SLIDE 9

Knowledge Management Institute 11

Markus Strohmaier 2008

Content vs. Intent

Content (What it is) Intent (What goals it aims at / helps to achieve)

  • find a physician
  • organize a high-school

reunion

  • contact an old friend
  • organize a marketing

campaign

  • find others who share

the same family name

  • find my way to an

address

  • …

Blog Posts, Speeches, Web Services, … Gulf of Execution [Norman 1988] <meta name="Keywords" content=

  • „yellow pages,
  • directory, local,
  • search,
  • business listings,
  • phone numbers,
  • maps,
  • driving directions,
  • white pages,
  • user reviews,
  • ratings,
  • internet yellow pages,
  • yellowpages,
  • telephone numbers" />

Tags are present in the pagetext of 50% of the pages they annotate and in the titles of 16% of the pages they annotate [Heymann 2008].

slide-10
SLIDE 10

Knowledge Management Institute 12

Markus Strohmaier 2008

Intent

Intent is:

  • Mostly latent [He 2007]
  • Not constrained by corresponding resources but by agents
  • Massively diverse [Chulef et al 2001, Strohmaier et al 2008]

[Anderson 2004]

Intentions Common human goals Highly indivi- dual goals Culture- specific goals

slide-11
SLIDE 11

Knowledge Management Institute 13

Markus Strohmaier 2008

Goal Modeling

slide-12
SLIDE 12

Knowledge Management Institute 14

Markus Strohmaier 2008

Search Query Logs as a Source of User Intent

  • M. Strohmaier, M. Lux, M. Granitzer, P. Scheir, S. Liaskos, E. Yu, How Do Users Express Goals on the Web? -

An Exploration of Intentional Structures in Web Search, We Know'07 International Workshop on Collaborative Knowledge Management for Web Information Systems in conjunction with WISE'07, Nancy, France, 2007.

[further non-intentional queries and further more complex intentional queries related to “wine crops”] Re-formulation 2006-04-07 06:40:45 How to have good [item wine crop] How to have good wine crop #6 [non-intentional query “wine crop”] Different Goal Formulation 2006-04-07 06:29:19 How to get rich [item wine crop] How to get rich wine crop #5 Generalization 2006-03-30 19:48:25 increase [item wine crop] Increase wine crop #4 [further non-intentional queries, not related to wine crop] Refinement 2006-03-30 19:46:11 [cause Fertilizer] to increase [item wine crop] Fertilizer to increase wine crop #3 Refinement 2006-03-30 19:45:28 [cause Fertilizer] or [cause insecticide] to increase [itemwine crop] Fertilizer or insecticide to increase wine crop #2 Formulation 2006-03-30 19:29:59 How to get more [itemwine crop] How to get more wine crop #1 Goal Time Stamp Frame Annotation Query Nr.

based on the i* framework [Yu 1995]

slide-13
SLIDE 13

Knowledge Management Institute 15

Markus Strohmaier 2008

Goal Mining

slide-14
SLIDE 14

Knowledge Management Institute 17

Markus Strohmaier 2008

Different Degrees of Explicitness in Search Queries

  • M. Strohmaier, P. Prettenhofer, M. Lux, Different Degrees of Explicitness in Intentional Artifacts - Studying User Goals in

a Large Search Query Log, CSKGOI'08 International Workshop on Commonsense Knowledge and Goal Oriented Interfaces, in conjunction with IUI'08, Canary Islands, Spain, 2008.

  • Search queries exhibit considerable variety with respect

to intentional degree of explicitness

  • Explicit vs. Implicit intentional queries

car, car Miami, car Miami dealer, buy a car in Miami, buy a used car in Miami, get loan to buy a used car in Miami car, car Miami, car Miami dealer, buy a car in Miami, buy a used car in Miami, get loan to buy a used car in Miami

implicit explicit

Example:

slide-15
SLIDE 15

Knowledge Management Institute 18

Markus Strohmaier 2008

Search Query Log Analysis - Results

  • M. Strohmaier, P. Prettenhofer, M. Kroell, Goal Acquisition from Search Query Logs (under review)

[adapted from Anderson 2004]

Common human goals Highly indivi- dual goals Culture- specific goals Intentions Data: Based on ~ 20 million search queries collected from 657,426 unique user ID’s between March 1, 2006 and May 31, 2006 by AOL [Pass 2006].

slide-16
SLIDE 16

Knowledge Management Institute 19

Markus Strohmaier 2008

Search Query Log Analysis - Results

  • M. Strohmaier, P. Prettenhofer, M. Kroell, Goal Acquisition from Search Query Logs (under review)
slide-17
SLIDE 17

Knowledge Management Institute 20

Markus Strohmaier 2008 Goals marked with (*) are also included in ConceptNet Commonsense Knowledge Base v2.1 [H. Liu and P. Singh 2004]

Search Query Log Analysis - Results

  • M. Strohmaier, P. Prettenhofer, M. Kroell, Goal Acquisition from Search Query Logs (under review)
slide-18
SLIDE 18

Knowledge Management Institute 21

Markus Strohmaier 2008

Goal Representation

slide-19
SLIDE 19

Knowledge Management Institute 22

Markus Strohmaier 2008

slide-20
SLIDE 20

Knowledge Management Institute 23

Markus Strohmaier 2008

An Extended Model of Folksonomies

  • M. Strohmaier, Purpose Tagging - Capturing User Intent to Assist Goal-Oriented Social Search, SSM'08

Workshop on Search in Social Media SSM'08, in conjunction with CIKM'08, Napa Valley, USA, 2008.

O T U F × × ⊆

s r q

O T U F × × ⊆ O T U F

p ×

× ⊆

q r s

Extended Model of Folksonomies

q...types of users r...types of tags s...types of objects

p

Purpose Tags

p...purpose

Traditional Model of Folksonomies

U...users T...tags O...objects

For example, types of tags include: [Golder und Hubermann 2005] 1) Identifying what a resource is about 2) Identifying what it is 3) Identifying who owns it 4) Refining categories 5) Identifying qualities or characteristics 6) Self reference 7) Task organizing

slide-21
SLIDE 21

Knowledge Management Institute 24

Markus Strohmaier 2008

Intentional Social Bookmarking –

  • M. Strohmaier, Purpose Tagging - Capturing User Intent to Assist Goal-Oriented Social Search, SSM'08

Workshop on Search in Social Media SSM'08, in conjunction with CIKM'08, Napa Valley, USA, 2008.

O T U F

p ×

× ⊆

w p c

O T U F × × ⊆

with students Andreas Haselsberger and Christoph Ruggenthaler

slide-22
SLIDE 22

Knowledge Management Institute 26

Markus Strohmaier 2008

Purpose Tagging

  • M. Strohmaier, Purpose Tagging - Capturing User Intent to Assist Goal-Oriented Social Search, SSM'08

Workshop on Search in Social Media SSM'08, in conjunction with CIKM'08, Napa Valley, USA, 2008.

  • 1. Can Purpose Tags Expand the Vocabulary of Existing

Tags?

slide-23
SLIDE 23

Knowledge Management Institute 27

Markus Strohmaier 2008

Purpose Tagging

  • M. Strohmaier, Purpose Tagging - Capturing User Intent to Assist Goal-Oriented Social Search, SSM'08

Workshop on Search in Social Media SSM'08, in conjunction with CIKM'08, Napa Valley, USA, 2008.

  • 2. Are Purpose Tag Graphs Meaningful?
slide-24
SLIDE 24

Knowledge Management Institute 28

Markus Strohmaier 2008

Goal Prediction

slide-25
SLIDE 25

Knowledge Management Institute 29

Markus Strohmaier 2008

We must do better!

Search today

Example: A query that probably every prospective college student performs once in his life.

slide-26
SLIDE 26

Knowledge Management Institute 30

Markus Strohmaier 2008

Goal Prediction – Work in Progress

AOL Search Query Log based Intent Prediction

Christian Körner

(yet untested) hypothesis: The shorter the query, the better our algorithms work

slide-27
SLIDE 27

Knowledge Management Institute 32

Markus Strohmaier 2008

Conclusions

slide-28
SLIDE 28

Knowledge Management Institute 33

Markus Strohmaier 2008

Conclusions

Goals on the web …

  • 1. are latent

– To make user intent explicit, new algorithms and techniques are necessary

  • 2. Are dynamic

– List of user goals which can not be enumerated

  • 3. vary dramatically

– Dependent on agent, situation and context

[adapted from Anderson 2004]

Intentions

slide-29
SLIDE 29

Knowledge Management Institute 34

Markus Strohmaier 2008

End of presentation

Markus Strohmaier

Graz University of Technology, Austria

http://www.kmi.tugraz.at/staff/markus/