Text Language Technology Natural Language Understanding Natural Language Generation Speech Recognition Speech Synthesis Text Meaning Speech.

Text

Language Technology

Natural Language Understanding

Natural Language Generation

Speech Recognition

Speech Synthesis

Text

Meaning

Speech Speech

Text

Language Technology

Natural Language Understanding

Natural Language Generation

Speech Recognition

Speech Synthesis

Text

Meaning

Speech Speech

What is NLG?

Natural language generation is the process of deliberately constructing a natural language text in order to meet specified communicative goals.

[McDonald 1992]

Example System: FoG• Function:

– Produces textual weather reports in English and French

• Input: – Graphical/numerical weather depiction

• User: – Environment Canada (Canadian Weather Service)

• Developer: – CoGenTex

• Status: – Fielded, in operational use since 1992

FoG: Input

FoG: Output

Example System: TEMSIS

• Function: – Summarises pollutant information for environmental officials

• Input: – Environmental data + a specific query

• User: – Regional environmental agencies in France and Germany

• Developer: – DFKI GmbH

• Status: – Prototype developed; requirements for fielded system being analysed

TEMSIShttp://www.dfki.de/service/nlg-demo/

TEMSIS: Output Summary

• Le 21/7/1998 à la station de mesure de Völklingen -City, la valeur moyenne maximale d'une demi-heure (Halbstundenmittelwert) pour l'ozone atteignait 104.0 µg/m³. Par conséquent, selon le decret MIK (MIK-Verordnung), la valeur limite autorisée de 120 µg/m³ n'a pas été dépassée.

• Der höchste Halbstundenmittelwert für Ozon an der Meßstation Völklingen -City erreichte am 21. 7. 1998 104.0 µg/m³, womit der gesetzlich zulässige Grenzwert nach MIK-Verordnung von 120 µg/m³ nicht überschritten wurde.

A further system

• ILEX– generation of virtual museum information

online– http://www.hcrc.ed.ac.uk/ilex/demos/museum.cgi

• SUMTIME– generation of weather reports– http://www.csd.abdn.ac.uk/~ssripada/cgi_bin/StartSMT.html

TEMSIS: Input Query

((LANGUAGE FRENCH)(GRENZWERTLAND GERMANY)(BESTAETIGE-MS T)(BESTAETIGE-SS T)(MESSSTATION \"Voelklingen City\")(DB-ID \"#2083\")(SCHADSTOFF \"#19\")(ART MAXIMUM)(ZEIT ((JAHR 1998) (MONAT 7) (TAG 21))))

Basic Generation Problem

• How to go from an abstract semantic input to a concrete linguistic form that is

– semantically correct– stylistically appropriate– textually appropriate

???

Standard Pipelined Architecture

Document Planning

Microplanning

Surface Realisation

Document Plan

Text Specification

KPMLlexicogrammar

semantics

sentence

Semantic specification

TACTICAL GENERATOR

KPMLlexicogrammar

semantics

sentence

Semantic specification

TACTICAL GENERATORKPML is a

Resources

Processgeneration

engine

lexicogrammar

semantics

sentence

Semantic specification

TACTICAL GENERATION

What is NLG?

Natural language generation is the process of deliberately constructing a natural language text in order to meet specified communicative goals.

NLG is a process of choice under specified constraints

[McDonald]

syntagmatic

Linguistic Description with system networks

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finiteparadigmatic

AXES

lexicogrammar

Resource Architecture in KPML:system networks

imperative

indicative

interrogative

declarative

Resource Architecture in KPML:system networks

imperative

indicative

interrogative

declarative

grammaticalsystems

Resource Architecture in KPML:system networks

imperative

indicative

interrogative

declarative

grammaticalfeatures

Resource Architecture in KPML:system networks

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Resource Architecture in KPML:system networks

imperative

indicative

interrogative

declarative

realizationstatements

+Finite

Finite^Subject

Subject^Finite

Generation Process:system networks

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:system networks

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

imperative

indicative

interrogative

declarative+Finite

Finite^Subject

Subject^Finite

Generation Process:traversal

indicative

interrogative

+Finite

Finite^Subject

Generation Process:structure

+Finite

Finite^Subjectinterrogative

Generation Process:structure

+Finite

Finite^Subject

interrogative

Immediate Dominance

Linear Precedence

Generation Process:realization statements

+Finite

Finite^Subject SubjectFinite

[clause]

Are you going?

[interrogative]

Types of Realization Statements

• Ordering (immediate, relative)

• Structure building

• Lexicalization

Functionally Motivated Grammatical

Choices

USER

Functionally Motivated Grammatical

Choices

USER

user = language engineer:developing and debugging the “grammatical competence”of a language resource

Functionally Motivated Grammatical

Choices

USER

SemanticSpecifications

Functionally Motivated Grammatical

Choices

USER

user = system builder:developing and debugging asystem that expects naturallanguage generationfunctionality

SemanticSpecifications