The Multimedia Articulation of Answers in a Natural Language ...

Stas Kolenikov

Description

• cours - matière potentielle : structure - centering
• cours magistral
• exposé - matière potentielle : to the user

The Multimedia Articulation of Answers in a Natural Language Database Query System Susan E. Brennan Stanford Un ivers i ty and Hewlet t Packard Labs 1501 Page Mi l l Road Pa lo A l to , CA 94304 Abstract This paper describes a domain independent strategy for the multimedia rticulation of answers elicited by a natural language interface to database query ap- plications. Multimedia answers include videodisc im- ages and heuristically-produced complete sentences in text or text-to-speech form.

• videodisc images
• appropriate nouns
• h.p. grice
• b.j. grosz
• query
• database
• answers
• user
• answer
• language

Sujets

• Cours magistral
• Chiang Kai-shek
• Brennan
• Browne
• Pollack

Informations

• Publié par: irra0
• Nombre de lectures: 18
• Langue: English

Extrait

Whither Social
Science
Statistics?
Stas
Kolenikov
U of Missouri
Introduction
Aim of the talk Whither Social Science Statistics?
Statistics timeline
Structural
equation
models Stas Kolenikov
Formulation
Identification
Estimation
Department of Statistics
Multilevel University of Missouri Columbia
models
Item response
theory September 27, 2006
ReferencesWhither Social
Science
Statistics? Outline
Stas
Kolenikov
U of Missouri
1 Introduction
Introduction Aim of the talk
Aim of the talk
Statistics timeline Statistics timeline
Structural
equation
models 2 Structural equation models
Formulation
Identification Formulation
Estimation
Identification
Multilevel
models Estimation
Item response
theory
References 3 Multilevel models
4 Item response theory
5 ReferencesAttractors of nonlinear dynamic systems!⇐= physics
Whither Social
Science
Statistics? Is this mathematics?
Stas
Kolenikov
U of Missouri
Introduction
Aim of the talk
Statistics timeline
Structural
equation
models
Formulation
Identification
Estimation
Multilevel
models
Item response
theory
ReferencesWhither Social
Science
Statistics? Is this mathematics?
Stas
Kolenikov
U of Missouri
Introduction
Aim of the talk
Statistics timeline
Structural
equation
models
Formulation
Identification
Estimation
Multilevel
models
Item response
theory
References
Attractors of nonlinear dynamic systems!⇐= physicsWhither Social
Science
Statistics? Is this statistics?
Stas
Kolenikov
U of Missouri
1.195
Introduction
Aim of the talk
Statistics timeline ξ1
Structural
equation
models
Formulation 1 0.464
Identification
Estimation 0.345 0.633
Multilevel
models
y y y yItem response 1 2 3 4
theory
References
δ δ δ δ1 2 3 4
− 0.195 1.018 1.109 1.066
.
1Whither Social
Science
Statistics? Aim of the talk
Stas
Kolenikov Winemiller conference topics
U of Missouri
Introduction Some of the developments in social sciences are essentially
Aim of the talk
Statistics timeline statistics, and that is why we wanted to present them under
Structural
the umbrella of developments in social science statistics.equation
models
Formulation
Identification This presentation will overview some of those areas with
Estimation
which I am relatively more familiar.Multilevel
models
Item response
theory
ReferencesWhither Social
Science
Statistics? History of...
Stas
Mainstream Econometrics Quant soc.Kolenikov
U of Missouri statistics sci. stat.
1900s Correlation, PCA: Pearson, Spearman
Introduction
Aim of the talk 1910s
Statistics timeline
1920s MLE: Fisher
Structural 1930s Multivariate: Econometric Psychometrics:
equation
models Hotelling society Thurstone
Formulation 1940s Sampling: Factor
Identification
Estimation Deming; HHM analysis
Multilevel Prob. theory:
models
Kolmogorov
Item response 1950s Kendall & Stuart Simultaneous MDS
theory
Cox & Cochran equations Cronbach’sα
References
Le Cam IV regressionWhither Social
Science
Statistics? History of...
Stas
Mainstream Econometrics Quant soc.Kolenikov
U of Missouri statistics sci. stat.
1960s 1st Nobel prize: Measurement
Introduction
Aim of the talk Frisch, Tinbergen theory: Rasch
Statistics timeline
1970s Jackknife Panel data LISREL:
Structural Missngdata Limited dep. vars, Joresk¨ og
equation
models Robust methods (Co)integration
Formulation 1980s Bootstrap GMM Multilevel
Identification
Estimation (G)ARCH models
Multilevel 1990s EmpirclLKHD Simulated MM IRT
models
MCMC Programevaluation
Item response 2000s Statistical Structural games Synthesis
theory
learning Quantile regrssn of models
References
2006 Winemiller conferenceWhither Social
Science 1
Statistics? Something∈/ C
Stas
Kolenikov • Lack of communication between fields of social
U of Missouri
sciences
Introduction
Aim of the talk • Lack of communication back and forth with statistics
Statistics timeline
•Structural Dominance of (what statisticians would see as) ad hoc
equation
solutionsmodels
Formulation
Identification • Software reliance/dominance
Estimation
• Some ideas are still difficult to pinpoint mathematicallyMultilevel
models
• Lack of mathematically strong students in social
Item response
theory science programs
References
• Lack of interest among mathematical statisticians (or is
there?)Whither Social
Science
Statistics? Structural equation models
Stas
Kolenikov (SEM)
U of Missouri
• Standard multivariate technique in social sciences
Introduction
Aim of the talk
•Statistics timeline Models explicitly the constructs that cannot be directly
Structural observed (level of stress; political attitudes; quality of
equation
models democratic institutions; genotype and environment)
Formulation
Identification • Special cases: linear regression, factor analysis,
Estimation
simultaneous equationsMultilevel
models
• Bridges to: multilevel models, IRT
Item response
theory
References