Deep CNN-LSTM with Word Embeddings for News Headline Sarcasm Detection

Mandal, Paul K.; Mahto, Rakeshkumar

doi:10.1007/978-3-030-14070-0_69

Cited by 28 publications

(12 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Earlier methods employed linguistic rules [51] or classical machine learning models [49], [52]. More recent methods used neural networks [53], [54], [55], [56], [57], [58] or pre-trained language models [59], [60], [61], [62].…”

Section: Related Workmentioning

confidence: 99%

Zero-resource Multi-dialectal Arabic Natural Language Understanding

Khalifa¹,

Hassan²,

Fahmy

2021

IJACSA

View full text Add to dashboard Cite

A reasonable amount of annotated data is required for fine-tuning pre-trained language models (PLM) on downstream tasks. However, obtaining labeled examples for different language varieties can be costly. In this paper, we investigate the zero-shot performance on Dialectal Arabic (DA) when finetuning a PLM on modern standard Arabic (MSA) data only -identifying a significant performance drop when evaluating such models on DA. To remedy such performance drop, we propose self-training with unlabeled DA data and apply it in the context of named entity recognition (NER), part-of-speech (POS) tagging, and sarcasm detection (SRD) on several DA varieties. Our results demonstrate the effectiveness of self-training with unlabeled DA data: improving zero-shot MSA-to-DA transfer by as large as 10% F1 (NER), 2% accuracy (POS tagging), and 4.5% F1 (SRD). We conduct an ablation experiment and show that the performance boost observed directly results from the unlabeled DA examples used for self-training. Our work opens up opportunities for leveraging the relatively abundant labeled MSA datasets to develop DA models for zero and low-resource dialects. We also report new state-of-the-art performance on all three tasks and open-source our fine-tuned models for the research community.

show abstract

Section: Related Workmentioning

confidence: 99%

Zero-resource Multi-dialectal Arabic Natural Language Understanding

Khalifa¹,

Hassan²,

Fahmy

2021

IJACSA

View full text Add to dashboard Cite

show abstract

“…(Maynard & Greenwood, 2014) achieved 50% success with their work on sentiment analysis on tweets. (Mandal & Mahto, 2019) studied another detector built using Deep CNN-LSTM. They used a news headline dataset to achieve 86.16% performance.…”

Section: Related Workmentioning

confidence: 99%

Sarcastic Text Detection Using Keras

Akkaya¹

2021

Computer Science

View full text Add to dashboard Cite

Sarcastic text is a type of text that contains a kind of irony, in which negative expressions are expressed as positive by attributing meanings to words that contradict their real meanings during communication. During face-to-face communication, changes in tone of voice, body language, eye contact or word stress make it easier for the other person to detect the sarcastic expression. However, it is challenging to detect sarcastic expressions only through text in machine learning-based systems since human-centred qualities cannot be transferred. Newspapers often use sarcastic expressions in their headlines to attract people's attention. However, many people cannot fully understand whether these expressions are sarcastic or not without reading the content of the text. As a result, they can transmit false information to the people around them through direct communication or social media. In this study, to prevent such misinformation, newspaper headlines with and without sarcasm are tried to be classified on three different GPUs with deep learning methods. As a result, the developed model can successfully detect news headlines containing sarcasm.

show abstract

“…In [10] Mandal & Mahto proposed the CNN-LSTM model approach added with word embedding to improve the accuracy of sarcasm detection by using a data headline dataset that has been categorized manually as many as 26.709 data and divided into 11.725 sarcasm and 14,984 non-sarcasm. The model used consists of a preprocessing stage where the data is normalized by removing punctuation marks, special characters and regular expressions then the data is stemmed to return the word to its basic form after the data is tokenized using a word embedding dictionary that has been created based on the 10.000 words that appear most frequently in the data.…”

Section: Related Workmentioning

confidence: 99%

“…The concatenated feature map is input to the fully connected softmax layer, which calculates the probability of any output word and classifies the news headline as sarcastic or non-sarcastic as an output. The output vector of the softmax layer P i is as (10).…”

Section: Representation Layermentioning

confidence: 99%

Sarcasm Detection Using Multi-Channel Attention Based BLSTM On News Headline

Azwar

Suharjito

2020

Preprint

View full text Add to dashboard Cite

Sarcasm is often used to express a negative opinion using positive or intensified positive words in social media. This intentional ambiguity makes sarcasm detection, an important task of sentiment analysis. Detecting a sarcastic tone in natural language hinders the performance of sentiment analysis tasks. The majority of the studies on automatic sarcasm detection emphasize on the use of lexical, syntactic, or pragmatic features that are often unequivocally expressed through figurative literary devices such as words, emoticons, and exclamation marks. In this paper, we introduce a multi-channel attention-based bidirectional long-short memory (MCAB-BLSTM) network to detect sarcastic headline on the news. Multi-channel attention-based bidirectional long-short memory (MCAB-BLSTM) proposed model was evaluated on the news headline dataset, and the results-compared to the CNN-LSTM and Hybrid Neural Network were excellent.

show abstract

Deep CNN-LSTM with Word Embeddings for News Headline Sarcasm Detection

Cited by 28 publications

References 3 publications

Zero-resource Multi-dialectal Arabic Natural Language Understanding

Zero-resource Multi-dialectal Arabic Natural Language Understanding

Sarcastic Text Detection Using Keras

Sarcasm Detection Using Multi-Channel Attention Based BLSTM On News Headline

Contact Info

Product

Resources

About