Prembul: presentaci perfil: llicenciada Filologia Espanyola, UAB, - - PowerPoint PPT Presentation

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Preàmbul: presentació

perfil:

llicenciada Filologia Espanyola, UAB, 2000 doctora, UPF (àrea: Lingüística Computacional), 2007

Doctorat Interuniversitari en Ciència Cognitiva i Llenguatge GLiCom: http://glicom.upf.es/

post-doc Juan de la Cierva, UPC, mitjan 2008 - mitjan 2011 (?)

pla de la xerrada:

tesi mica intenció futur

1 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Automatic acquisition of semantic classes for adjectives

Gemma Boleda Torrent

GLiCom Universitat Pompeu Fabra / Barcelona Media Centre d’Innovació

NLP Seminar, UPC, November 14th 2007

2 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Overview

automatic acquisition of semantic classes for Catalan adjectives two main hypotheses:

adjective meanings can be assigned to a set of classes semantic distinctions mirrored at different linguistic levels

Lexical Acquisition

infer properties of words from their linguistic behaviour in corpora

3 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Approach (I)

no general, well established semantic classification → propose and test classification iterative methodology

deductive phase: define a classification and apply it to a set

• f adjectives

→ manual annotation and machine learning experiments inductive phase: use the evidence gathered to refine the classification proposal

4 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Approach (II)

three iterations Experiment Technique Main goal A Unsupervised refine classification B validate refined classification C Supervised integrate polysemy

5 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Contents

1

Introduction

2

Initial classification

3

Experiments A and B: Testing the classification

4

Experiment C: Integrating polysemy

5

Conclusion

6

I a la UPC?

6 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Initial classification

insights from descriptive grammar and formal semantics Qualitative adjectives denote attributes or properties of objects. Examples: ample, autònom ‘wide’, ‘autonomous’ Intensional adjectives denote second order properties. Examples: presumpte, antic ‘alleged’, ‘former’ Relational adjectives denote a relationship to an object. Examples: pulmonar, botànic ‘pulmonary’, ‘botanical’ semantic classification supported by distinctions at other levels of description

7 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Initial classification

insights from descriptive grammar and formal semantics Qualitative adjectives denote attributes or properties of objects. Examples: ample, autònom ‘wide’, ‘autonomous’ Intensional adjectives denote second order properties. Examples: presumpte, antic ‘alleged’, ‘former’ Relational adjectives denote a relationship to an object. Examples: pulmonar, botànic ‘pulmonary’, ‘botanical’ semantic classification supported by distinctions at other levels of description

7 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Initial classification

insights from descriptive grammar and formal semantics Qualitative adjectives denote attributes or properties of objects. Examples: ample, autònom ‘wide’, ‘autonomous’ Intensional adjectives denote second order properties. Examples: presumpte, antic ‘alleged’, ‘former’ Relational adjectives denote a relationship to an object. Examples: pulmonar, botànic ‘pulmonary’, ‘botanical’ semantic classification supported by distinctions at other levels of description

7 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Initial classification

insights from descriptive grammar and formal semantics Qualitative adjectives denote attributes or properties of objects. Examples: ample, autònom ‘wide’, ‘autonomous’ Intensional adjectives denote second order properties. Examples: presumpte, antic ‘alleged’, ‘former’ Relational adjectives denote a relationship to an object. Examples: pulmonar, botànic ‘pulmonary’, ‘botanical’ semantic classification supported by distinctions at other levels of description

7 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Initial classification

insights from descriptive grammar and formal semantics Qualitative adjectives denote attributes or properties of objects. Examples: ample, autònom ‘wide’, ‘autonomous’ Intensional adjectives denote second order properties. Examples: presumpte, antic ‘alleged’, ‘former’ Relational adjectives denote a relationship to an object. Examples: pulmonar, botànic ‘pulmonary’, ‘botanical’ semantic classification supported by distinctions at other levels of description

7 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Criteria (I): position with respect to the head noun

qualitative (1) intensional (2) relational (3) pre- and post-nominal pre-nominal post-nominal

1

les avingudes amples / les amples avingudes ‘wide avenues’

2

#l’assassí presumpte / el presumpte assassí ‘the alleged murderer’

3

una malaltia pulmonar / #una pulmonar malaltia ‘a pulmonary disease’

8 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Criteria (II): predicativity

qualitative (1) intensional (2) relational (3) predicative non-predicative marginally predicative

1

les avingudes són amples ‘avenues are wide’

2

#l’assassí és presumpte ‘the murderer is alleged ’

3

?la malaltia és pulmonar ‘the disease is pulmonary’

9 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Polysemy

1

edifici antic (qualitative) / antic president (intensional) ‘ancient building / former president’

2

reunió familiar (relational) / cara familiar (qualitative) ‘family meeting / familiar face’ in each sense, the adjective’s behaviour corresponds to that of the relevant class

10 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Polysemy

1

edifici antic (qualitative) / antic president (intensional) ‘ancient building / former president’

2

reunió familiar (relational) / cara familiar (qualitative) ‘family meeting / familiar face’ in each sense, the adjective’s behaviour corresponds to that of the relevant class

10 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Motivation for unsupervised experiments

classification based primarily on literature review does it account for the semantics of a broad range of adjectives? empirical test: use information extracted from corpus in machine learning experiments exploratory experiments → clustering (unsupervised)

no bias by previous annotation insight into the actual structure of the data

two sets of experiments (Exp. A, Exp. B)

11 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Motivation for unsupervised experiments

classification based primarily on literature review does it account for the semantics of a broad range of adjectives? empirical test: use information extracted from corpus in machine learning experiments exploratory experiments → clustering (unsupervised)

no bias by previous annotation insight into the actual structure of the data

two sets of experiments (Exp. A, Exp. B)

11 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Material and method (I) – resources

resources also used in Experiments B and C CTILC corpus (Institut d’Estudis Catalans):

14.5 million words, written, formal texts manually lemmatised and POS-tagged automatically shallow-parsed (noise)

adjective database [Sanromà, 2003]:

almost 2,300 lemmata from CTILC corpus morphological information manually coded

12 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Material and method (II)

Gold Standard: 101 lemmata, random choice

classes: qualitative, relational, intensional, int-qual, qual-rel

technique: clustering, k-means (CLUTO) features: semantic, distributional semantic (6 features) pre-nominal position, predicativity, . . . distributional (36 features) POS unigrams (two words left, two words right of target)

13 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Material and method (II)

Gold Standard: 101 lemmata, random choice

classes: qualitative, relational, intensional, int-qual, qual-rel

technique: clustering, k-means (CLUTO) features: semantic, distributional semantic (6 features) pre-nominal position, predicativity, . . . distributional (36 features) POS unigrams (two words left, two words right of target)

13 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Feature example

I IQ Q QR R 1 2 3 4 5

Predicative

I = intensional IQ = int−qual Q = qualitative QR = qual−rel R = relational

14 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Summary of results

strong support for classes qualitative and relational intensional class not separated with this methodology group of problematic adjectives identified in error analysis: indicador, parlant, protector, salvador ‘indicating’, ‘speaking’, ‘protecting’, ‘saviour’ (they do not fit into the classification) classification is modified according to these results approach to polysemy is clearly wrong → Exp. C semantic and distributional features yield similar results

15 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Summary of results

strong support for classes qualitative and relational intensional class not separated with this methodology group of problematic adjectives identified in error analysis: indicador, parlant, protector, salvador ‘indicating’, ‘speaking’, ‘protecting’, ‘saviour’ (they do not fit into the classification) classification is modified according to these results approach to polysemy is clearly wrong → Exp. C semantic and distributional features yield similar results

15 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Summary of results

strong support for classes qualitative and relational intensional class not separated with this methodology group of problematic adjectives identified in error analysis: indicador, parlant, protector, salvador ‘indicating’, ‘speaking’, ‘protecting’, ‘saviour’ (they do not fit into the classification) classification is modified according to these results approach to polysemy is clearly wrong → Exp. C semantic and distributional features yield similar results

15 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment A: Summary of results

strong support for classes qualitative and relational intensional class not separated with this methodology group of problematic adjectives identified in error analysis: indicador, parlant, protector, salvador ‘indicating’, ‘speaking’, ‘protecting’, ‘saviour’ (they do not fit into the classification) classification is modified according to these results approach to polysemy is clearly wrong → Exp. C semantic and distributional features yield similar results

15 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Modified classification

Basic adjectives (formerly qualitative) Object-related adjectives (formerly relational) Event-related adjectives denote a relationship to an event. Examples: protector, variable ‘protecting’, ‘variable’ Syntactic/distributional properties? relationship with morphology basic event

• bject

non-derived deverbal denominal supported by Ontological Semantics [Raskin and Nirenburg, 1998]

16 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Modified classification

Basic adjectives (formerly qualitative) Object-related adjectives (formerly relational) Event-related adjectives denote a relationship to an event. Examples: protector, variable ‘protecting’, ‘variable’ Syntactic/distributional properties? relationship with morphology basic event

• bject

non-derived deverbal denominal supported by Ontological Semantics [Raskin and Nirenburg, 1998]

16 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Modified classification

Basic adjectives (formerly qualitative) Object-related adjectives (formerly relational) Event-related adjectives denote a relationship to an event. Examples: protector, variable ‘protecting’, ‘variable’ Syntactic/distributional properties? relationship with morphology basic event

• bject

non-derived deverbal denominal supported by Ontological Semantics [Raskin and Nirenburg, 1998]

16 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment B

characteristics

acquire predominant class of each adjective

ignore polysemy

focus on distributional features (empirical approach)

results

• ne-to-one mapping between clusters and manually

assigned classes. Accuracy:

baseline 49%

• distr. (clust.)

73% morph. 65% distributional information superior to morphological information difficulties with the event class arise

17 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment B

characteristics

acquire predominant class of each adjective

ignore polysemy

focus on distributional features (empirical approach)

results

• ne-to-one mapping between clusters and manually

assigned classes. Accuracy:

baseline 49%

• distr. (clust.)

73% morph. 65% distributional information superior to morphological information difficulties with the event class arise

17 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Experiment B

characteristics

acquire predominant class of each adjective

ignore polysemy

focus on distributional features (empirical approach)

results

• ne-to-one mapping between clusters and manually

assigned classes. Accuracy:

baseline 49%

• distr. (clust.)

73% morph. 65% distributional information superior to morphological information difficulties with the event class arise

17 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Polysemy, revisited

1

explicació embolicada (basic) / regal embolicat (event) ‘unclear explanation / wrapped present’

2

reunió familiar (object) / cara familiar (basic) ‘family meeting / familiar face’

3

tasca docent (event) / planificació/equip docent (object) ‘teaching task / teaching planning/team’ each adjective can be assigned to more than one class → polysemy acquisition as multi-label classification

18 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Polysemy, revisited

1

explicació embolicada (basic) / regal embolicat (event) ‘unclear explanation / wrapped present’

2

reunió familiar (object) / cara familiar (basic) ‘family meeting / familiar face’

3

tasca docent (event) / planificació/equip docent (object) ‘teaching task / teaching planning/team’ each adjective can be assigned to more than one class → polysemy acquisition as multi-label classification

18 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Goals of Experiment C

1

include polysemy in the acquisition experiment

2

assess the role of different levels of linguistic description for semantic classification

3

test ways to combine linguistic information

19 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Goals of Experiment C

1

include polysemy in the acquisition experiment

2

assess the role of different levels of linguistic description for semantic classification

3

test ways to combine linguistic information

19 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Material

210 lemmata stratified sampling approach

frequency, morphology (derivational type, suffix)

large-scale manual annotation experiment

administered via Web 322 subjects does not yield reliable classification (K 0.31-0.45)

Gold Standard classification: committee of 3 experts

20 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Method

algorithm: Decision Trees (C4.5 as implemented in Weka) features: Level Explanation # F. morph morphological (derivational) properties 2 func syntactic function 4 uni uni-gram distribution 24 bi bi-gram distribution 50 sem distributional cues of semantic properties 18 all combination of the 5 linguistic levels 10.3

Table: Linguistic levels as feature sets.

21 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Procedure

binary decision

basic/non-basic, event/non-event, object/non-object

for each adjective, merge classifications familiar: basic event

• bject

merged yes no yes basic-object (BO)

• btain 100 accuracy estimates for each class and level

10 run, 10-fold cross-validation

test differences between levels with a statistical test

corrected resampled t-test [Nadeau and Bengio, 2003]

baseline: most frequent class (basic)

22 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Procedure

binary decision

basic/non-basic, event/non-event, object/non-object

for each adjective, merge classifications familiar: basic event

• bject

merged yes no yes basic-object (BO)

• btain 100 accuracy estimates for each class and level

10 run, 10-fold cross-validation

test differences between levels with a statistical test

corrected resampled t-test [Nadeau and Bengio, 2003]

baseline: most frequent class (basic)

22 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Procedure

binary decision

basic/non-basic, event/non-event, object/non-object

for each adjective, merge classifications familiar: basic event

• bject

merged yes no yes basic-object (BO)

• btain 100 accuracy estimates for each class and level

10 run, 10-fold cross-validation

test differences between levels with a statistical test

corrected resampled t-test [Nadeau and Bengio, 2003]

baseline: most frequent class (basic)

22 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

accuracy results for merged classification

2 4 6 8 50 55 60 65 Runs Accuracy

A: Full accuracy

2 4 6 8 60 65 70 75 80 Runs Accuracy

B: Partial accuracy

bl morph func uni bi sem all 23 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

baseline (most frequent class – basic): 51% level morph (60.6%) is the best unique level of information

morphology is most useful for our task? BUT: morphology included in the sampling scheme. . .

level all (combination of information) improves upon morph: 62.3% error analysis shows that all and morph make very different mistakes → is there a better combination than all?

24 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

baseline (most frequent class – basic): 51% level morph (60.6%) is the best unique level of information

morphology is most useful for our task? BUT: morphology included in the sampling scheme. . .

level all (combination of information) improves upon morph: 62.3% error analysis shows that all and morph make very different mistakes → is there a better combination than all?

24 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

A better combination

ensemble classifier each level proposes one class, the majority class is chosen intuition: expert committee

morphologist, syntactician, engineer...

25 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

Levels Classifs. morph+func+uni+bi+sem+all 6 84.0±0.06 morph+func+uni+bi+sem 5 82.3±0.04 func+uni+bi+sem 4 81.5±0.04 morph+func+sem+all 4 72.4±0.03 morph+func+sem 3 76.2±0.03 bl

• 51.0±0.0

all

• 62.3±2.3

Table: Results for ensemble classifier.

average improvement over level all: 15.8

26 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

Levels Classifs. morph+func+uni+bi+sem+all 6 84.0±0.06 morph+func+uni+bi+sem 5 82.3±0.04 func+uni+bi+sem 4 81.5±0.04 morph+func+sem+all 4 72.4±0.03 morph+func+sem 3 76.2±0.03 bl

• 51.0±0.0

all

• 62.3±2.3

Table: Results for ensemble classifier.

average improvement over level all: 15.8

26 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

Levels Classifs. morph+func+uni+bi+sem+all 6 84.0±0.06 morph+func+uni+bi+sem 5 82.3±0.04 func+uni+bi+sem 4 81.5±0.04 morph+func+sem+all 4 72.4±0.03 morph+func+sem 3 76.2±0.03 bl

• 51.0±0.0

all

• 62.3±2.3

Table: Results for ensemble classifier.

combination > type of linguistic information used

27 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Results

Levels Classifs. morph+func+uni+bi+sem+all 6 84.0±0.06 morph+func+uni+bi+sem 5 82.3±0.04 func+uni+bi+sem 4 81.5±0.04 morph+func+sem+all 4 72.4±0.03 morph+func+sem 3 76.2±0.03 bl

• 51.0±0.0

all

• 62.3±2.3

Table: Results for ensemble classifier.

combination > type of linguistic information used

27 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

To sum up

use of computational techniques for linguistic research

[Merlo and Stevenson, 2001, Schulte im Walde, 2006]

• ne step further: re-shaping target classification based on

experimental results

combine insights from linguistic theory with evidence gathered from machine learning and human annotation experiments revise hypotheses wrt adjective classification according to analysis of results

28 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

To sum up

use of computational techniques for linguistic research

[Merlo and Stevenson, 2001, Schulte im Walde, 2006]

• ne step further: re-shaping target classification based on

experimental results

combine insights from linguistic theory with evidence gathered from machine learning and human annotation experiments revise hypotheses wrt adjective classification according to analysis of results

28 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Adjective classification

broad and consistent classification proposal

• btained through theoretical and empirical exploration

characterisation of the classes

morphological, syntactic, semantic properties

exploration of polysemy some difficulties in the classification

event-related adjectives non-prototypical basic adjectives nationality-denoting adjectives

29 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Human annotation experiments

3 manual annotation experiments large-scale, Web experiments can be fruitful to gather linguistic data

multiple analysis possibilities

but they are very difficult to design (non-expert subjects) bottleneck for our task: obtention of reliable linguistic data

30 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Automatic acquisition of semantic classes for adjectives

3 sets of machine learning experiments: existing techniques, new uses

use of unsupervised techniques to provide feedback wrt classification proposal multi-label classification architecture → polysemy systematic comparison of different linguistic levels of description combination of different types of linguistic evidence

semantic classification of Catalan adjectives using morphological and distributional information → feasible

no need of intensive resources methodology can be extended to other languages

31 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Future research

adjective classification: theoretical implication of results

definition of a semantic classification for adjectives characterisation of each class polysemy within our task (e.g., polysemy judgements)

manual annotation experiments:

design of adequate experiments to build reliable datasets

machine learning experiments:

type of information (selectional restrictions) datasets (other corpora) machine learning techniques (MBL, kernel methods, other types of ensemble classifiers) external evaluation (POS-tagging, Paraphrase Detection)

32 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Automatic acquisition of semantic classes for adjectives

Gemma Boleda Torrent

GLiCom Universitat Pompeu Fabra / Barcelona Media Centre d’Innovació

NLP Seminar, UPC, November 14th 2007

33 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (I)

adquisició lèxica

semàntica lèxica: restriccions selecció, classes semàntiques, . . .

altres aspectes semàntica lèxica

Multi-Word Expressions?

recerca en lingüística amb eines computacionals

Machine Learning, estadística, dades de corpus → Quantitative Linguistics

34 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (I)

adquisició lèxica

semàntica lèxica: restriccions selecció, classes semàntiques, . . .

altres aspectes semàntica lèxica

Multi-Word Expressions?

recerca en lingüística amb eines computacionals

Machine Learning, estadística, dades de corpus → Quantitative Linguistics

34 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (II)

creació de corpus i d’altres recursos

BancTrad [Badia et al., 2002] CUCWeb [Boleda et al., 2006] CatCG [Alsina et al., 2002]

experiments amb subjectes humans per a CL/PLN

[Boleda et al., 2008]

disseny i anàlisi metodologia d’avaluació: inter-annotator agreement

35 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (II)

creació de corpus i d’altres recursos

BancTrad [Badia et al., 2002] CUCWeb [Boleda et al., 2006] CatCG [Alsina et al., 2002]

experiments amb subjectes humans per a CL/PLN

[Boleda et al., 2008]

disseny i anàlisi metodologia d’avaluació: inter-annotator agreement

35 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (III)

models cognitivament plausibles de representació (i

processament?) de la informació semàntica

→ Ciència Cognitiva espais vectorials? psicolingüística: recerca en conceptes [Murphy, 2002]

36 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Interessos (IV)

. . . i les col·laboracions que puguin sorgir!

37 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Alsina, A., Badia, T., Boleda, G., Bott, S., Gil, A., Quixal, M., and Valentín, O. (2002). CATCG: a general purpose parsing tool applied. In Proceedings of Third International Conference on Language Resources and Evaluation, Las Palmas, Spain. Badia, T., Boleda, G., Colominas, C., Garmendia, M., González, A., and Quixal,

• M. (2002).

BancTrad: a web interface for integrated access to parallel annotated corpora. In Proceedings of the Workshop on Language Resources for Translation Work and Research held during the 3rd LREC Conference, Las Palmas. Boleda, G., Bott, S., Castillo, C., Meza, R., Badia, T., and López, V. (2006). CUCWeb: a Catalan corpus built from the Web. In Kilgarriff, A. and Baroni, M., editors, 2nd Web as Corpus Workshop at EACL ’06. Boleda, G., Schulte im Walde, S., and Badia, T. (2008). Modelling polysemy in adjective classes by multi-label classification. Accepted for publication in Research on Language and Computation. Special issue on “Ambiguity and semantic judgments”, edited by Massimo Poesio and Ron Artstein. Merlo, P . and Stevenson, S. (2001).

37 / 37

Introduction Initial classification Experiments A and B: Testing the classification Experiment C: Integrating polysemy Conclusion I a la UPC?

Automatic verb classification based on statistical distributions of argument structure. Computational Linguistics, 27(3):373–408. Murphy, G. L. (2002). The Big Book of Concepts. MIT Press, Cambridge (etc.). Nadeau, C. and Bengio, Y. (2003). Inference for the generalization error. Machine Learning, 52(3):239–281. Raskin, V. and Nirenburg, S. (1998). An applied ontological semantic microtheory of adjective meaning for natural language processing. Machine Translation, 13(2-3):135–227. Sanromà, R. (2003). Aspectes morfològics i sintàctics dels adjectius en català. Master’s thesis, Universitat Pompeu Fabra. Schulte im Walde, S. (2006). Experiments on the automatic induction of German semantic verb classes. Computational Linguistics, 32(2):159–194.

37 / 37