Browsing by Author "Mahmud, Md. Shihab"

Now showing 1 - 8 of 8

FinTech: Deep Learning-Based Sentiment Classification of User Reviews from Various Bangladeshi Mobile Financial Services
(Springer Nature Limited, 2023-07-24) Ryan, Abdullah Al; Mahmud, Md. Shihab; Mahi, Hasibul Hasan Chowdhury; Hossen, Md Shakil; Shimul, Nazmul Islam; Noori, Sheak Rashed Haider
Banking has become an integral part of our lives. Fintech (Financial Technology) skyrocketed the number of people willing to use Mobile Financial Services (MFS) for their daily financial transactions. The banks are providing their services via mobile applications, which can be found on the Google Play Store. These Mobile Financial Services (MFS) provide mobility and increase efficiency by 10-fold. With an astonishing number of users came an abundant number of reviews for these apps. User reviews are the backbone of an application’s success. They provide information about hands-on experience. This study mainly focuses on the reactions of the users of such apps. Sentiment analysis is being used to draw out emotions from the users based on their written reviews. The primary goal of this paper is to examine the points of view of such application users. A total of 5414 pieces of data were collected from the Google Play Store and classified as negative, neutral, or positive. The data model has been evaluated using CNN, LSTM, and BiLSTM algorithms. Compared to CNN and LSTM, the BiLSTM algorithm produced the best model with an accuracy of 97.07%.
Newspaper’s Editorial Opinion Prediction in Sentiment Analysis Using Deep Learning Methods
(Daffodil International University, 2022-01-04) Momtaj, Mst. Anika; Mahmud, Md. Shihab; Saha, Uchchhwas
Sentiment analysis is a computational method that uses preliminary emotion analysis to retrieve feelings and key phrases from various texts (e.g., Positive, Negative, Neutral). It’s necessary to extract useful information from big data, categorize it, and predict end-user behavior or emotions. Text classification is a research area of Natural Language Processing (NLP). Which is converted from unstructured data to meaningful categorical classes. All previous work is most likely based on traditional different classifiers such as KNN, SVM, and so on. In this study, we propose a method which is combined in two familiar deep learning models: Convolutional Neural Networks (CNN) and Bidirectional Long Short Term Memory (BiLSTM). CNN method retrieves greater characteristics by convolutional layers and max pooling layers and BiLSTM can capture long term dependencies by lexical items and it’s better for text classification. Our own built datasets collected from various Bengali newspapers, such as Bangla Tribune, The Daily Sun etc. generate massive amounts of data. Our dataset size is 5996. The accuracy of the proposed model differs optimizer wise. We use three different optimizers: Adam, Adamax, RMSProp. With these three optimizers, the highest outcomes come from Adam optimizer. Accuracy of our first proposed model (BiLSTM CNN) with word2vec and Glove word embedding is 91.59% and 99.40%. In (CNN-BiLSTM) methods obtained outcomes are 92.60% and 94% and the final model (BiLSTM-CNN-BiLSTM) got their results 82.79% and 98.47%. The experimental results represent how the deep learning models effectively work. We say that the amount and quality of training examples have a significant impact on models performance.
Sentiment Analysis from User-Generated Reviews of Ride-Sharing Mobile Applications
(Daffodil International University, 2022-04-15) Mahmud, Md. Shihab; Bonny, Afrin Jaman; Saha, Uchchhwas; Jahan, Mehrin; Tuna, Zannatul Ferdhoush; Marouf, Ahmed Al
Smartphone applications play an increasingly significant part in our everyday lives, and their use has skyrocketed. The Google Play Store is a well-known plat-form through which one may obtain various Android applications whereas ap-plication like Ridesharing play a significant role in delivering public services more efficiently and effectively, as seen by the widespread adoption of many different types of innovative applications. This study focuses on users' reactions to these ridesharing applications, and it employs sentiment analysis to extract emotions from text reviews posted on the Google App Store platform given by the users. The primary goal is to examine the perspectives of customers and users of these applications. A total of 1818 data was gathered from the Google Play Store and divided into three categories: positive, negative, and neutral. The model was evaluated using the CNN, LSTM, and DistilBERT algorithms, with DistilBERT outperforming the others and achieving the highest accuracy of 98.84 %.
Sentiment Analysis of Bengali Textual Comments in Field of Sports Using Deep Learning Approach
(Daffodil International University, 2022-10-16) Rabeya, Mosa; Tuly, Mumtahina Rahman; Mahmud, Md. Shihab; Sattar, Abdus
In recent days, people are expressing their emotions, feelings or opinions on various social platforms. In those opinions some are real and some are fake. There are a lot of discussions about sports. When their team wins a match, they celebrate this highly but when a match loses, they criticize, bullying them. And then they express them angrily to different sites, like Facebook pages, Facebook groups etc. This issue may be resolved by using natural language processing (NLP) to analyze the sentiment of the relevant comments. Here we analyze sentiment in various sports related Bangla comments. We collected almost 4061 data from various Facebook pages and groups. After collecting those data, we classified them into five different categories: neutral, happy, sad, positive and negative. We use some preprocessing techniques like removing punctuation, data cleaning, manual validation to prepare our data. In this study, we used three different familiar deep learning models to predict sentiment of our dataset. Here our models are CNN, LSTM and BiLSTM. In these three models CNN with the glove word embedding performed better than other two models, and it is 94.57%. Finally, the CNN model outperforms other models in a way that captures the sentiment of the fans' remarks.
Sentiment Analysis of User-Generated Reviews of Women Safety Mobile Applications
(Daffodil International University, 2022-04-15) Mahmud, Md. Shihab; Bonny, Afrin Jaman; Saha, Uchchhwas; Jahan, Mehrin; Tuna, Zannatul Ferdhoush; Marouf, Ahmed Al
Google play store is an application store from where people get various kinds of applications for android certified devices which makes life a lot easier and faster through the diverse functionalities the apps contain. Numerous users are using applications as per their needs and putting their experience, thoughts of using that application via reviews in form of ratings and texts. As the safety of women is threatened, whether applications like women's safety apps are appreciated, can be detected through text reviews and ratings by the users. This paper analyzes the positive, negative, neutral polarity of the sentences or text reviews that are given by the users of the women's safety app through the google play store. To detect the emotions of the users through the given text reviews and star ratings, the machine learning (ML) algorithms using natural language processing (NLP) are conducted to analyze the sentiments of the review given by the users. For this study, the data was collected from the app reviews and star ratings provided by the users of the women's safety related applications whose main purpose is to provide necessary functionality that can keep women safe in any dangerous and unwanted situation. The purpose of this paper is to mine the opinion of the users and get their viewpoint about those apps of specific polarity levels. As the current user's ratings, reviews, or their viewpoint helps the new user understand the performance of the applications and insights in advance, so the mining of their opinion is helpful for both parties - developers and general users. To detect the level of the sentiment, several machine learning algorithms were applied, namely Multinomial Naive Bayes (MNB), Logistic Regression (LR), Support Vector Machine (SVM), and k-nearest neighbor (k-NN). Among these algorithms, the MNB has outperformed all other algorithms in terms of accuracy (85.42%).
Sentiment Classification for IMDB Movie Reviews in Benchmark Dataset Using LR, MNB and SGD
(IEEE, 2023-11-23) Habib, Haifa Binte; Chowdhury, Md Kamruzzaman; Islam, Md. Tauhidul; Mahmud, Md. Shihab
A breakdown of all movie reviews can help viewers decide by saving them the time it would take to read all of the reviews. Critics commonly use movie-rating websites to submit comments and rate films, so guiding viewers in determining whether or not to see the film. Sentiment analysis was used to determine reviewers' attitudes based on their opinions. Sentiment analysis of a movie review can help assess how positive or negative a review is, and thus the film's overall rating. In this research, the sentiment classification methods LR, MNB, and SGD are suggested for a big movie review data set. We chose 50K IMDB movie review entries that are entirely written in English, with 25K positive and 25K negative ratings. The results show that the Multinomial Naive Bayes (MNB) algorithm outperforms other classification algorithms in terms of mistake rate and accuracy.
Sentiment Classification in Bengali News Comments Using a Hybrid Approach with Glove
(Daffodil International University, 2022-05-14) Saha, Uchchhwas; Mahmud, Md. Shihab; Chakrobortty, Aisharjo; Akter, Mst.Tuhin; Islam, MD Rakib; Marouf, Ahmed Al
Sentiment analysis (SA) is a greater part of Natural Language Processing (NLP) in the research field. Previously a lot of work was done by SA in various languages. Most of the work was done in the English language. A little work was done in the Bengali language and it's increasing day by day. Nowadays, Bangla News comments are very demanding for research work in the Bangla language. The procedures of text categorization, classifying, and different techniques for extracting features in textual information were discussed in this paper. Some Bangla newspaper dataset is currently available in online platforms. This research study has analyzed Bangla news comments sentiment using a hybrid approach and a pre-trained deep learning classifier. The proposed hybrid model utilizes an optimizer function “Adam” along with a word embedding “Glove”. The dataset utilized in the proposed model is collected from online platform Kaggle, the largest data science forum on the world. This dataset contains 13802 data. Using five different classes to our dataset. For building a model, this research study has applied preprocessing techniques, which makes a significant contribution in cleaning our dataset. This leads to get a well formed dataset. The proposed hybrid model has been combined with two familiar deep learning methods, namely BiLSTM and CNN. By comparing these two methods, the proposed hybrid model has gained better accuracy than the FastText model. The accuracy of hybrid model is about 89.89%.
The Corporeality of Infotainment on Fans Feedback Towards Sports Comment Employing Convolutional Long-Short Term Neural Network
(Daffodil International University, 2022-05-27) Saha, Uchchhwas; Mahmud, Md. Shihab; Shimu, Sumaia; Eva, Shabikun Naher; Khushbu, Sharun Akter; Asif, Imtiaz Ahmmed
In ODIs, World Cups and T-20 matches of various sports like Football, Cricket, Hockey, Basketball and Badminton, fans express their feelings and emotions towards the players by posting their status on social media like Facebook, Twitter etc. By collecting these opinions and feelings of the fans from different mediums, this research study has become more focused on a sentimental analysis of the sport with a total of 3759 comments related to Football (both national, international), Cricket, Hockey and Badminton. Since sports related opinions have been taken up in Bengali, global vector (glove) word embedding techniques are used for pre-processing which can retrieve word meanings and synthetic information. It also specializes in creating word vectors, including the structure of word embedding infrastructure, and provides a special advantage over statistics. Three models have been proposed in our study, one of which is a hybrid model of CNN-LSTM. In the proposed CNN-LSTM model, the CNN model is used to quote various features from word embedding that reflect short-term sentiment dependence while creating long-term sentimental relationships between LSTM words. In comparison to the hybrid model, two single models CNN and LSTM are proposed in five categories (i.e. Positive, Negative, Neutral, Happy, Sad). The sport's dataset integrates the CNN-LSTM hybrid model with the glove embedding layer, providing 97.45% accuracy. Lastly, the LSTM-CNN hybrid models perform comparatively better, realizing the feeling of the fans' comments.