PD3: Better Cross-Lingual Transfer By Combining Direct Transfer and - - PowerPoint PPT Presentation

▶
pd3 better cross lingual transfer by combining direct
SMART_READER_LITE
LIVE PREVIEW

PD3: Better Cross-Lingual Transfer By Combining Direct Transfer and - - PowerPoint PPT Presentation

Dec 23, 2022 141 likes •404 views

PD3: Better Cross-Lingual Transfer By Combining Direct Transfer and Annotation Projection Steffen Ege r* , Andreas Rckle, Iryna Gurevych 27.03.2018 | Fachbereich Informatik | UKP Lab 1 Argumentation Mining Fast-growing research field

slide-1
SLIDE 1

27.03.2018 | Fachbereich Informatik | UKP Lab 1

Steffen Eger*, Andreas Rückle, Iryna Gurevych

PD3: Better Cross-Lingual Transfer By Combining Direct Transfer and Annotation Projection

slide-2
SLIDE 2

20.02.2018 | Fachbereich Informatik | UKP Lab 2

Argumentation Mining

  • Fast-growing research field in NLP
  • Different sub-tasks:

1) segmenting arguments from non-arguments in text; 2) classifying them (claim, premise, ...); 3) finding relations between arguments (support, attack) 4) Ranking arguments

slide-3
SLIDE 3

27.03.2018 | Fachbereich Informatik | UKP Lab 3

Challenges for argumentation mining

  • Going cross-lingual

○ I.e. train system in a source language L1 (typically: English), then apply system to specific target language L2 of interest ○ Avoids having to redo (high) annotation costs

  • Recently, several works have addressed variants of this setup:

○ Aker and Zhang, 2017; Sliwa et al. 2018; Eger et al., 2018; Rocha et al. 2018

slide-4
SLIDE 4

27.03.2018 | Fachbereich Informatik | UKP Lab 4

Task considered in our work

  • We consider argumentation mining

○

On the sentence-level

○

Classifying each sentence into 4 classes:

■ Claim, MajorClaim, Premise, None

  • Dataset is derived from the Persuasive Essay (PE) dataset of Stab

and Gurevych (2017); Eger et al. (2018) (bi-lingual variant)

○ But token-level annotations are mapped to the sentence-level

slide-5
SLIDE 5

27.03.2018 | Fachbereich Informatik | UKP Lab 5

(Mono-lingual) Examples

  • Not cooking fresh food will lead to lack of nutrition Claim
  • To sum up, [...] the merits of animal experiments still outweigh the

demerits Major claim

  • For example, tourism makes up one third of Czech’s economy

Premise

  • I will mention some basic reasoning as follows O
slide-6
SLIDE 6

27.03.2018 | Fachbereich Informatik | UKP Lab 6

Our contribution

  • We explore cross-lingual argumentation mining in the low-resource

setting, i.e., having very little parallel data, …. ○ Which is likewise a hot topic concurrently (Zhang et al., 2016; Artetxe et al., 2017; Artetxe et al., 2018; Lample et al., 2018; Schulz et al. 2018)

  • … by combining two standard cross-lingual approaches --- direct

transfer and annotation projection

slide-7
SLIDE 7

7

Excursion - Cross-lingual transfer 1: Direct Transfer

L1 L2 Die Stube brennt Kinder sind doof ….. Bilingual word embeddings I/PRON love/V children/N Cats/N like/V me/PRON …..

slide-8
SLIDE 8

8

Direct Transfer

I/PRON love/V children/N Cats/N like/V me/PRON ….. L1 L2 Die Stube brennt Kinder sind doof ….. Bilingual word embeddings

slide-9
SLIDE 9

9

Excursion - Cross-lingual transfer 2: Annotation Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses eat carrots Pferde essen Möhren Soccer is football Fußball ist Fußball ….. I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-10
SLIDE 10

10

Projection

I/PRON love/V cats/N Cats/N like/V me/PRON ….. L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses eat carrots Pferde essen Möhren Soccer is football Fußball ist Fußball ….. Train

slide-11
SLIDE 11

11

Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde essen Möhren Soccer/N is/V football/N Fußball ist Fußball ….. Annotate I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-12
SLIDE 12

12

Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Project I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-13
SLIDE 13

13

Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Project I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-14
SLIDE 14

14

Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Project I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-15
SLIDE 15

15

Projection

L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Train/An notate I/PRON love/V cats/N Cats/N like/V me/PRON …..

slide-16
SLIDE 16

16

PD3

I/PRON love/V cats/N Cats/N like/V me/PRON ….. L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Train/An notate

Train on bilingual repres./Annotate

slide-17
SLIDE 17

17

PD3

I/PRON love/V cats/N Cats/N like/V me/PRON ….. L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Train/An notate

Train on bilingual repres./Annotate

slide-18
SLIDE 18

18

PD3

I/PRON love/V cats/N Cats/N like/V me/PRON ….. L1 L2 Die Stube brennt Das Wasser läuft ….. L1-L2 Horses/N eat/V carrots/N Pferde/N essen/V Möhren/N Soccer/N is/V football/N Fußball/N ist/V Fußball/N ….. Train/An notate

Train on bilingual repres./Annotate

slide-19
SLIDE 19

19

PD3: Combining Direct Transfer and Projection

  • One last issue:
  • Can either merge all 3 datasets
  • Or use multi-task learning, taking e.g., both L1 datasets as Task1

and the L2 dataset as Task2

slide-20
SLIDE 20

20

Experiments

  • Bilingual data:
  • en: To sum up [...], the merits of animal experiments still outweigh

the demerits MajorClaim

  • de: Zusammenfassend kann ich bestätigen [...], dass die Vorzüge

von Tierversuchen die Nachteile [...] überwiegen MajorClaim

  • About 7k parallel sentences, available here:

https://github.com/UKPLab/coling2018-xling_argument_mining

  • Setup:

○ 2k for train (en), 0.5k for dev (en), 1.5k for test (de) ○ 3k as parallel data (and further subsets thereof) ■ We also consider non-argumentative parallel data from TED ○ Evaluation Metric is Macro-F1

slide-21
SLIDE 21

21

Results - high quality bilingual embeddings

slide-22
SLIDE 22

22

Results - low quality bilingual embeddings

slide-23
SLIDE 23

23

Results - low quality bilingual embeddings

slide-24
SLIDE 24

24

Results - non-argumentative parallel data

slide-25
SLIDE 25

25

Conclusion

  • Considered low-resource language transfer for ArgMin

○ By combining direct transfer and annotation projection

  • There are benefits, but they’re small
  • Also, they diminish quickly
  • True low-resource language transfer still a big challenge

○ And an important avenue for the future

  • Doing annotation projection using machine translation without any

parallel data (Artexte et al. 2018, Lample et al. 2018) may be worthwhile to investigate prospectively

slide-26
SLIDE 26

27.03.2018 | Fachbereich Informatik | UKP Lab 26

THÁNK YÕU