What is a potential benefit of using transformer models like BERT or AraBERT for detecting hate speech?

B. They capture deeper semantic meaning in text

What is one reason why traditional approaches, like dictionaries, struggle to detect slang?

Slang changes over time and breaks grammatical rules

What helped models understand slang the most?

Training on real internet language plus slang examples

Why do traditional machine learning models struggle with detecting hate speech in Arabic slang?

Arabic slang is informal, regional, and fast-evolving

How do the proposed deep learning models understand slang words they have never seen before?

By reading the sentence both forward and backward to understand context.

What is the purpose of the Normalized Token Concatenator?

Reassemble processed tokens into normalized text

studbud

NLP Slang

8 termsStudBud Administration14 plays

Imported from external platform

Flashcards8 cardsFPractice quizPHost liveL

Questions 8

Answers

1
Which language model handled slang the best?
BERTweet
BERT
GPT-3
FastText
2
What is the first step of the authors’ slang normalization system?
Tokenizer
Spell correction
Slang dictionary lookup
POS tagging
3
What is a potential benefit of using transformer models like BERT or AraBERT for detecting hate speech?
They are faster than all traditional models
They are designed specifically for emoji detection
They require no training data at all
B. They capture deeper semantic meaning in text
4
What is one reason why traditional approaches, like dictionaries, struggle to detect slang?
In fact, slang is easy to detect using static dictionaries.
Slang is only used in spoken language
Slang words are always misspelled
Slang changes over time and breaks grammatical rules
5
What helped models understand slang the most?
Reducing the training data
Training on real internet language plus slang examples
Using only slang dictionaries
Adding more formal language
6
Why do traditional machine learning models struggle with detecting hate speech in Arabic slang?
They cannot process emojis properly
They are only trained on English text
Arabic slang is formal and consistent
Arabic slang is informal, regional, and fast-evolving
7
How do the proposed deep learning models understand slang words they have never seen before?
By ignoring the sentence context and focusing on individual words
By reading the sentence both forward and backward to understand context.
By relying solely on static slang dictionaries.
By translating slang into synonyms.
8
What is the purpose of the Normalized Token Concatenator?
Filter out non-standard words
Create new slang
Reassemble processed tokens into normalized text
Add punctuation

NLP Slang Flashcards | StudBud