티스토리 뷰
차이 정리한 slides: framenet,prombank,vernet
모두 lexical database 라고 보면 될까?
Framenet: The basic idea is straightforward: that the meanings of most words can best be understood on the basis of a semantic frame: a description of a type of event, relation, or entity and the participants in it. (ref: https://www.nltk.org/howto/framenet.html)
=> 그래서 비슷한 frame 을 가지는 단어들을 그룹화 하고, 그 그룹의 description, 그 그룹이 가지는 relation 등을 정의해놓은 듯.
예를 들어, Arrest 라는 프레임의 경우, Frame elements 로 아래 리스트를 가짐.
'Authorities',
'Charges',
'Co-participant',
'Manner',
'Means',
'Offense',
'Place',
'Purpose',
'Source_of_legal_authority',
'Suspect',
'Time',
'Type'
Propbank : predicate 이랑 arguments 들을 정리해 놓음. 장점은 very standardized (arg0, arg1.. ). Propbank 는 a set of frame files 로 이루어져 있고, 각 frame file 마다 하나 혹은 여러개의 propbank verb sense 가 정의되어 있는데 이걸 바로 frame sets 라고 한다.
Verbnet: English verb 만 annoate 한 건데, verb 마다 agent, theme, etc 등을 정의해 놓음. 그리고 비슷한 verb 끼리 묶어놓음.
발췌
FrameNet, VerbNet and PropBank. The FrameNet corpus contains the examples annotated with semantic roles whereas the VerbNet lexicon provides the knowledge about the syntactic behavior of the verbs. We connect VerbNet and FrameNet by mapping the FrameNet frames to the VerbNet Intersetive Levin classes. The PropBank corpus, which is tightly connected to the VerbNet lexicon, is used to increase the verb coverage and also to test the effectiveness of our approach.
정리: predicate 을 정리한 lexical dataset 임. 어떻게 정리했냐? 각 predicate 마다 description 과 함께 쓰이는 elements 들로 구조화하고, 공통적인 단어들끼리 묶었다는 점에서 비슷한데, role 을 어떻게 정의했느냐 (e.g. AMR: arg0, arg1 / Verbnet: agent, theme / Frameset: 이건 frame 마다 다름), 어떤 단어들끼리 어떻게 묶었냐 이런 것들이 다른 듯.
'NLP > 이것저것' 카테고리의 다른 글
Dependency parsing (0) | 2022.06.08 |
---|---|
Transformer / BERT / GPT2 / BART 의 차이 (0) | 2021.12.30 |
N-gram language model (0) | 2021.12.20 |
Naïve Bayes Classifier (0) | 2021.12.06 |
Accumulated gradients (0) | 2021.08.03 |
- Total
- Today
- Yesterday
- Elmo
- language model
- cs224n
- Bert
- Statistical Language Model
- neural network
- LM
- neurone
- transformer
- Pre-trained LM
- 벡터
- GPTZero
- Neural Language Model
- Contextual Embedding
- weight vector
- 뉴런
- word embedding
- 워터마킹
- nlp
- 언어모델
- Attention Mechanism
- 뉴럴넷
일 | 월 | 화 | 수 | 목 | 금 | 토 |
---|---|---|---|---|---|---|
1 | 2 | 3 | 4 | 5 | ||
6 | 7 | 8 | 9 | 10 | 11 | 12 |
13 | 14 | 15 | 16 | 17 | 18 | 19 |
20 | 21 | 22 | 23 | 24 | 25 | 26 |
27 | 28 | 29 | 30 |