Acquisition of Semantic Classes for Adjectives from Distributional Evidence

1
Acquisition of Semantic Classes for Adjectives
from Distributional Evidence

• Gemma Boleda
• Universitat Pompeu Fabra
• Barcelona

2
general picture

• automatic classification of adjectives
• Catalan
• according to broad semantic characteristics
• clustering
• syntactic evidence

3
motivation

• Lexical Acquisition
• infer properties of words
• lexical bottleneck
• both symbolic and statistical approaches
• adjectives
• determining NP reference
• the French general
• establishing properties of entities
• this maimai is round and sweet

4
motivation

• initial motivation POS-tagging
• 55 remaining ambiguity involves adjectives
• general francès French general or general
  French?
• observations
• general tendencies in syntactic behaviour of
  adjectives
• ... which correspond to broad semantic properties
• generalisation best at semantic level
• low-level tasks (POS-tagging)
• initial schema for lexical semantic representation

5
approach

• no general, well established semantic
  classification
• have to build and test ours!
• clustering unsupervised technique
• groups objects according to feature distribution
• does not depend on pre-classification
• provides insight into the nature of the data
• shallow approach to syntax n-grams
• limited syntactic distribution
• local relationship to arguments
• gt test feasibility

rodó round 0.4 0.4 0.2
dolç sweet 0.5 0.4 0.1
francès French 0.1 0.6 0.3
italià Italian 0.05 0.5 0.45
6
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

7
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

8
adjective syntax

• default function noun modifier (92)
• right of the noun (default position 72)
• some to the left (epithets 28)
• predicative uses unfrequent (7), but significant

9
two-way classification

• number of arguments
• unary pilota vermella red ball
• binary professor gelós de la Maria teacher
  jealous of Maria
• ontological kind (Ontological Semantics)
• basic vermell red
• object malaltia pulmonar pulmonary disease (gt
  lung)
• event propietat constitutiva constitutive
  property (gt constitutes)

10
Ontological Semantics

• coverage (ordinary cases)
• machine tractability
• explicit model of world ontology
• vermell gt attributecolourred(x)
• pulmonar gt related-tolung(x)
• constitutiu gt eventbenefconstitute(x)
• however no commitment to particular framework

11
rationale

• observation syntactic preferences correspond to
  semantic properties
• hypothesis we can use syntactic features to
  infer semantic classes

12
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• conclusions and future work

13
data and procedure

• 2283 adjectives
• gt50 times in 16 million word Catalan corpus
• lemma and morphological info
• cluster the whole set
• perform different tasks on different subsets
• tuning subset choose features
• Gold Standard evaluation and analysis

14
features and feature selection

• features
• empirically chosen from blind distribution
• double bigram, simplified POS-representation

ella diu que la pilota vermella és seva
she says that the ball red is hers
-3ey -2dd -1cn 1ve

• tuning subset 100 adjectives
• choose features (distribution)

15
Fig. A Feature selection
16
analysis

• Gold Standard
• 80 adjectives
• annotated by 3 human judges, acceptable agreement
  (92 and 84, .72 and .74 kappa)

17
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

18
experiment 1 unary / binary

• final evaluation10 features, raw percentage
• clustering algorithm k-means (cosine)
• predictions
• binary adjectives cooccur with prepositions more
  frequently than unary ones
• unary adjectives are more flexible

19
unary / binary results

• agreement with Gold Standard
• 97, kappa 0.87
• comparable to humans
• features

cl high low
0 (un) -1cn 1prep
1 (bin) 1prep (-1cn)
20
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

21
experiment 2 basic / object / event

• final evaluation 32 features, normalisation
• clustering algorithm k-means (cosine)
• predictions
• basic adjectives are flexible, work as epithets,
  occur in predicative contexts, appear further
  from the noun
• object adjectives appear rigidly after the noun
• event adjectives tend to occur in predicative
  positions and do not act as epithets

22
basic / object / event results
object (yellow)

• agreement with Gold Standard
• 73, kappa 0.56
• lower than humans
• features

event (orange)
cl high low
0 (obj) -1cn -1ve
1 (ev) 1prep
2 (bas) -1co 1aj
basic (red)
23
basic/object/event error analysis

• something has gone wrong!
• characterisation of event adjectives

binary!
basic adjectives with an object reading (polysemy)
unary event adjectives
binary event adjectives
24
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

25
partial conclusions

• overall, results seem to back up
• use of syntax-semantics interface for adjectives
• linguistic predictions as to relevant features
  and differences across classes
• shallow approach
• unary / binary piece of cake
• few binary adjectives, but worth spotting (denote
  relationships)

26
partial conclusions

• basic / object / event need reworking
• object adjectives seem to be the most robust
  class
• variation in basic adjectives (default class),
  polysemy
• event adjectives seem to behave much like basic
  adjectives with respect to features chosen gt
  redefine class?

27
outline

• adjective syntax and semantic classification
• methodology
• experiment 1
• experiment 2
• partial conclusions
• outlook rest of the thesis

28
outlook rest of the thesis

• rethink classification
• redefine features in light of results
• integrate polysemy judgments into the experiment
  and analysis
• perform experiments with other corpora

29
classification

• what to do with event adjectives? cp.
• constitutiu constitutive (active)
• legible readable (passive)
• reproductor reproducing (active, habituality)
• yet another parameter gradability
• important for adjectives
• should be easy to induce

30
better blind distribution or self-defined
features?

• n-grams sparseness, selection

empirical accurate sparseness objective
blind ? X X ??
self X? (depends on method) ? ? X

• other features?
• account for different levels of description

31
polysemy

• crucial aspect, explains much of results
• difficult to integrate!
• meaningless kappa values
• alternatives?
• clearer definition of polysemy within task
• specific tests
• other resources dictionary?

32
other resources

• CUCWeb (208 million word)
• http//www.catedratelefonica.upf.es
• test whether more data is better data (Mercer
  and Church 1993 18-19)
• advantages and challenges of Web corpora
• current results for verb subcategorisation
  experiment, results 12 points lower than using
  smaller, balanced, controled corpus

33
Acquisition of Semantic Classes for Adjectives
from Distributional Evidence

• Gemma Boleda
• Universitat Pompeu Fabra
• Barcelona