Dark Green and White Modern Minimalist Thesis Defense Presentation

ROBERTA-LSTM: A HYBRID MODEL FOR SENTIMENT ANALYSIS WITH TRANSFORMER AND RECURRENT NEURAL NETWORK.

OVERVIEW. INTRODUCTION. RESULT AND ANALYSIS. LITERATURE PREVIEW.

LITERATURE REVIEW. RESEARCH GAP. Sentiment analysis is a process that studies the emotions, sentiments, and attitudes of people from their written language. Sentiment analysis is important in the field of Natural Processing Language (NLP) and has a significant influence on business and social media. The challenges of sentiment analysis include long-distance dependencies and lexical diversity of texts. [1] Sequence models are commonly used for sentiment analysis as they can encode long-distance dependencies, but they are less computationally efficient. [1] [2] Transformer models, on the other hand, improve computation through parallelized processing. [1] [2].

LITERATURE REVIEW. RESEARCH GAP. OBJECTIVE. INTRODUCTION.

Author Paper Name Approach Key Contribution Wongkar and Angdrese (2019) [5] Sentiment analysis using naive Bayes algorithm of the data crawler: Twitter Machine Learning The authors analyzed a Twitter dataset related to the 2019 Indonesian presidential candidates using three machine learning methods: Naïve Bayes (75.58% accuracy), KNN (73.34% accuracy), and SVM (63.99% accuracy) Jung et al. (2016) [7] Enhanced naive Bayes classifier for real-time sentiment analysis with SparkR Machine Learning The authors used multinomial Naïve Bayes for sentiment analysis on the sentiment140 dataset and achieved 85% accuracy D. K. Madhuri (2019) [8] A machine learning-based framework for sentiment classification: Indian railways case stud Machine Learning The study compared four machine learning methods on the Indian Railways Twitter dataset, with accuracies ranging from 89% to 91.5%.

Author Paper Name Approach Key Contribution Younas et al. (2020) [11] Sentiment analysis of code-mixed Roman Urdu-English social media text using deep learning approaches Deep Learning Authors introduced two deep learning methods for sentiment analysis on multilingual social media text using 2018 Pakistan election tweets. mBERT achieved 69% accuracy, while XLM-R achieved 71% Dholpuria et al. (2018) [13] Sentiment analysis approach through deep learning for a movie review Deep Learning The paper analyzed 3000 IMDb movie reviews, and the CNN model achieved a remarkable 99.33% accuracy Uddin et al. (2019) [18] Depression analysis from social media data in Bangla language using long short term mem- ory (LSTM) recurrent neural network technique Deep Learning The authors employed LSTM for Bangla Tweets sentiment analysis, obtaining an 86.3% accuracy with an 80% training, 10% validation, and 10% testing dataset split.

LITERATURE REVIEW. Limited research on using Transformer models for sentiment analysis Lack of exploration of hybrid models that combine the strengths of Transformer models and sequence models for sentiment analysis Insufficient investigation on the impact of data augmentation techniques.

LITERATURE REVIEW. To propose a hybrid deep learning model for sentiment analysis by combining the strengths of the RoBERTa approach and the Long Short-Term Memory (LSTM) model To address the challenges of sentiment analysis, such as lexical diversity, imbalanced datasets, and long-distance dependencies in texts To overcome the limitations of sequence models, which require longer execution time, by integrating the capabilities of Transformer models.

lowercase. special symbols are eliminated. stemming.

LITERATURE REVIEW. RESEARCH GAP. OBJECTIVE. INTRODUCTION.

LITERATURE REVIEW. RESEARCH GAP. OBJECTIVE. INTRODUCTION.

DATASET. Three sentiment analysis datasets used: IMDb, Sentiment140, and Twitter US Airline Sentiment. IMDb dataset: 50,000 movie reviews, evenly split between positive and negative. Sentiment140 dataset: 1.6 million reviews, balanced between positive and negative. Twitter US Airline Sentiment dataset: 14,640 tweets related to American airlines, with imbalanced distribution: 9,178 negative, 2,363 positive, and 3,099 neutral reviews..

THE EXPERIMENTAL RESULTS ON THE IMDB DATASET.

THE EXPERIMENTAL RESULTS ON THE TWITTER US AIRLINE SENTIMENT..

THE EXPERIMENTAL RESULTS ON THE SENTIMENT140 DATASET.

RoBERTa-LSTM, a hybrid model, excels in sentiment analysis by combining RoBERTa and LSTM, overcoming key challenges..

The paper lacks a comparison of RoBERTa-LSTM with other hybrid models or state-of-the-art approaches in sentiment analysis, missing a broader context..

The impact of hyperparameter choices and model architecture variations on RoBERTa-LSTM's performance remains unexplored..

There is no analysis of the model's interpretability or explainability, hindering understanding of the decision-making process..

Lorem ipsum dolor sit amet, consectetur adipiscing elit..

THANK YOU.

QUESTION?.