Representation of Actions as an Interlingua

Slides:



Advertisements
Similar presentations
Dr. David A Ferrucci -- Logic Programming and AI Lecture Notes Knowledge Structures Building the Perfect Object.
Advertisements

Computational language: week 10 Lexical Knowledge Representation concluded Syntax-based computational language Sentence structure: syntax Context free.
CODE/ CODE SWITCHING.
 Christel Kemke 2007/08 COMP 4060 Natural Language Processing Feature Structures and Unification.
Introducing Formal Methods, Module 1, Version 1.1, Oct., Formal Specification and Analytical Verification L 5.
Linking to the ICF One of the aims of this work is to link this representation scheme to the WHO ICF.
PM—Propositional Model A Computational Psycholinguistic Model of Language Comprehension Based on a Relational Analysis of Written English Jerry T. Ball,
Syntax-Semantics Mapping Rajat Kumar Mohanty CFILT.
Knowledge Representation
Statistical NLP: Lecture 3
Steven Schoonover.  What is VerbNet?  Levin Classification  In-depth look at VerbNet  Evolution of VerbNet  What is FrameNet?  Applications.
Semantic analysis Parsing only verifies that the program consists of tokens arranged in a syntactically-valid combination, we now move on to semantic analysis,
LTAG Semantics on the Derivation Tree Presented by Maria I. Tchalakova.
Copyright © Cengage Learning. All rights reserved. CHAPTER 2 THE LOGIC OF COMPOUND STATEMENTS THE LOGIC OF COMPOUND STATEMENTS.
NLP and Speech Course Review. Morphological Analyzer Lexicon Part-of-Speech (POS) Tagging Grammar Rules Parser thethe – determiner Det NP → Det.
KAIST CS780 Topics in Interactive Computer Graphics : Crowd Simulation A Task Definition Language for Virtual Agents WSCG’03 Spyros Vosinakis, Themis Panayiotopoulos.
C SC 620 Advanced Topics in Natural Language Processing Lecture 20 4/8.
NaLIX: A Generic Natural Language Search Environment for XML Data Presented by: Erik Mathisen 02/12/2008.
1 Introduction to Computational Linguistics Eleni Miltsakaki AUTH Fall 2005-Lecture 2.
Let remember from the previous lesson what is Knowledge representation
Translation Divergence LING 580MT Fei Xia 1/10/06.
Stochastic POS tagging Stochastic taggers choose tags that result in the highest probability: P(word | tag) * P(tag | previous n tags) Stochastic taggers.
Meaning and Language Part 1.
Business Domain Modelling Principles Theory and Practice HYPERCUBE Ltd 7 CURTAIN RD, LONDON EC2A 3LT Mike Bennett, Hypercube Ltd.
CSC 8310 Programming Languages Meeting 2 September 2/3, 2014.
Embedded Clauses in TAG
Copyright © Cengage Learning. All rights reserved.
Frames and semantic networks, page 1 CSI 4106, Winter 2005 A brief look at semantic networks A semantic network is an irregular graph that has concepts.
QUALITATIVE MODELING IN EDUCATION Bert Bredweg and Ken Forbus Yeşim İmamoğlu.
1 The Interaction Between Verbs And Constructions Lucas Champollion Oct 18 th, 2004 Goldberg, Adele E. (1995): Constructions. Ch. 2.
INF 384 C, Spring 2009 Ontologies Knowledge representation to support computer reasoning.
Tree-adjoining grammar (TAG) is a grammar formalism defined by Aravind Joshi and introduced in Tree-adjoining grammars are somewhat similar to context-free.
1 LIN 1310B Introduction to Linguistics Prof: Nikolay Slavkov TA: Qinghua Tang CLASS 22, March 27, 2007.
Introduction to Embodied Construction Grammar March 4, 2003 Ben Bergen
Introduction to Linguistics Ms. Suha Jawabreh Lecture # 2.
1 Introduction to Computational Linguistics Eleni Miltsakaki AUTH Fall 2005-Lecture 4.
What you have learned and how you can use it : Grammars and Lexicons Parts I-III.
1 1. Representing and Parameterizing Agent Behaviors Jan Allbeck and Norm Badler 연세대학교 컴퓨터과학과 로봇 공학 특강 학기 유 지 오.
Programming Languages and Design Lecture 3 Semantic Specifications of Programming Languages Instructor: Li Ma Department of Computer Science Texas Southern.
The Minimalist Program
Towards Linguistically Grounded Ontologies Paul Buitelaar, Philipp Cimiano, Peter Haase, and Michael Sintek Proceedings of the 6 th European Semantic Web.
Knowledge Representation
PRACTICAL KNOWLEDGE REPRESENTATION FOR THE WEB Frank van Harmelen Dieter Fensel AIFB Kim Kangil Structural Complexity Laboratory.
LING 6520: Comparative Topics in Linguistics (from a computational perspective) Martha Palmer Jan 15,
Levels of Linguistic Analysis
Machine Translation Divergences: A Formal Description and Proposed Solution Bonnie J. Dorr University of Maryland Presented by: Soobia Afroz.
An Introduction to Semantic Parts of Speech Rajat Kumar Mohanty rkm[AT]cse[DOT]iitb[DOT]ac[DOT]in Centre for Indian Language Technology Department of Computer.
Lec. 10.  In this section we explain which constituents of a sentence are minimally required, and why. We first provide an informal discussion and then.
SYNTAX.
What is a semantic role?. A semantic role is the underlying relationship that a participant has with the main verb in a clause.verbclause Also known as:
XML Databases Presented By: Pardeep MT15042 Anurag Goel MT15006.
Natural Language Processing Vasile Rus
EXTRACTING COMPLEX PREDICATES IN HINDI ACROSS PARALLEL CORPORA
Advanced Computer Systems
Knowledge Representation Techniques
Conditional Statements
Chapter 6 : Transition and Transfer predicates
Lecture – VIII Monojit Choudhury RS, CSE, IIT Kharagpur
Statistical NLP: Lecture 3
Knowledge Representation
Introduction to Computational Linguistics
Chapter 2: A Simple One Pass Compiler
Communicative Competence (Canale and Swain, 1980)
Levels of Linguistic Analysis
SYNTAX DIRECTED DEFINITION
Ling 566 Oct 14, 2008 How the Grammar Works.
Semantic Nets and Frames
Structure of a Lexicon Debasri Chakrabarti 13-May-19.
Syntax Lecture 12: Extended VP.
CIS Monthly Seminar – Software Engineering and Knowledge Management IS Enterprise Modeling Ontologies Presenter : Dr. S. Vasanthapriyan Senior Lecturer.
Presentation transcript:

Representation of Actions as an Interlingua Karin Kipper & Martha Palmer Presented in CIS630 by Sriram Venkatapathy

Framework Command Animation In Performed by Natural Language a Human Agent Command In Natural Language (“open a door”)

Primary Goal Command Animation In Performed by Natural Language a Human Agent Command In Natural Language (“open a door”) Bridge the Gap

Parametrized Action Representation (PAR) Used to bridge the gap between the command and the animation performed by the virtual human agent. PARs make explicit many details that are underspecified in the human language.

PARs as an Interlingua This detailed representation is well suited for an interlingua for MT applications because Animations of Actions – and therefore the PARs that control them – will be equivalent for the same actions described in different languages.

PARs as an interlingua – contd. Spanish instruction English instruction PARs Animation

PAR Action’s participants (agent and objects) Kinematic Properties like its path, manner and duration.

PAR – contd. Traditional state-space properties of actions such as applicability conditions and preparatory actions. Termination and Post assertions which determine when an action is concluded.

PAR ( for the action “contact”)

Hierarchy of actions Verbs can be represented in a lattice that allows semantically similar verbs, such as motion verbs or verbs of contact, to be closely associated. A common parent captures the properties that these verbs all share.

Verbs related to “contact”

Representation of “hammer” Inherits PAR of hit , and ultimately the PAR for contact. with “forceful” manner. and “hammer” as an instrument.

PAR – contd. PAR is intended to provide slots for information that is typically conveyed in modifiers or adjuncts. “John hit the ball” “John hit the ball with a bat” “John swung mightily and his bat hit the ball with a resounding sound”. All map to the same PAR schema.

Generating animations

Deriving the PAR schema Synchronous Tree-adjoining grammar is used for parsing natural language instructions and obtaining the predicate-argument dependencies. Assemble the corresponding schemas, and fill in the participants and modifiers to output the PAR schema.

Planner These schemas might be underspecified for actions like “enter”, “put” and hence don’t have enough information for the animation. So, a planner is used to Select the way (activity) “enter by walking/swimming” Determine preparatory actions “open the door” Decompose complex actions “put the glass on the table”

PAR as Interlingua PAR representation is a general template. It includes properties of the action that can occur linguistically either as the main verb or as adjuncts to the main verb phrase. Captures divergences such as, for verb-framed versus satellite-framed languages.

Verb-Framed vs Satellite-Framed Satellite-Framed (English) Manner -> Main verb. Motion (path or path + ground location) -> Satellite Ex. The bottle floated out. Verb-Framed (Spanish) Motion -> Main verb Manner -> Satellite Ex. La bottella salio flotanda. (the bottle exited floating)

Verb-Framed vs Satellite-Framed For the sentences, The bottle floated into the cave. (English) La bottella entro flotanda a la cueva (Spanish) (The bottled entered floating the cave)

Verb-Framed vs Satellite-Framed The PAR schemes don’t distinguish the representation for these sentences, because there is a single schema that includes both the manner and the path without specifying how they are realised linguistically. EN float/[par:motion,activity:float] into/[term:in(AG,OBJ] SP entrar/[par:motion,term:in(AG,OBJ) flotar/[activity:float]

Schema for the above sentences

Comparison with other work Considerably different from the approach outlined in Palmer et al. (1998) which discusses the use of Feature-based Tree-adjoining grammars. It was a transfer-based mechanism expressed in Synchronous TAG to capture divergences in VFL and SFL through the use of semantic features and links between the grammars.

Comparison with other work Similar to Lexical Conceptual Structures (LCS) approach. LCS allows the separation of the concepts of motion, direction, and manner of motion in the sentence “John swam across the lake”. (represented as GO,PATH,MANNER). This approach allows for a similar representation and the end result is the same.

Conclusions This work discusses a parameterized representation of actions grounded by needs of animations in a simulated environment. Generalizations based on action classes provide the basis for an interlingua approach that captures the semantics without committing to any language-dependent specification. The PAR schema incorporates in a single template both VFL and SFL languages.

Thank you.