Dynamic Sentiment Analysis: A Low‑Latency System for Social

Media Monitoring

Sanskar Kumar Agrahari

, Arjun Kumar Das

, Krishna Bhagat

, Vivek Kumar Shah

Nikita Sharma

and Gayathri Ramasamy

Department of Computer Science & Engineering, Amrita School of Computing, Amrita Vishwa Vidyapeetham, Bengaluru,

Karnataka, India

Department of Electronics & Electrical Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham,

Bengaluru, Karnataka, India

Keywords: Sentiment Analysis, Kafka, Django, Big Data, Hadoop, Machine Learning.

Abstract: Social media sentiment analysis needs real-time tracking of public opinion therefore requires fast processing

together with low latency and high accuracy. To achieve this, PySpark is used for the data preprocessing and

model training process. A web application that is developed through Django lets users submit tweets that

generate instant sentiment predictions whether the tweet is positive, negative, neutral, or irrelevant while

Kafka manages real-time streaming of processed results. MongoDB utilizes NoSQL architecture to effectively

store sentiment forecasts to- gather with their associated data. Among different trained models, Logistic

regression achieved maximum accuracy according to testing while the system showed successful operation

through real-time sentiment analysis with high- speed data processing and quick response times and

approachable user interface which proved its usefulness for sentiment trend analysis.

1 INTRODUCTION

The sudden explosion in social media platforms

results in gigantic amounts of user-generated content,

which makes in-the-moment tracking difficult for

public sentiment. This is where sentiment analysis

and opinion mining come into play, which according

to them would involve the extraction of all emotions,

opinions, and sentiments from a text, such as from a

social media site. However, unstructured social media

content processing for the bottom line remains an

arduous task. Traditional techniques of sentiment

analysis mostly lack real-time processing, scalability,

and accuracy, which makes action-happy insight

derivation impossible.

In this digital era, real-time sentiment analysis is

very important to understand public sentiment tables

and study emerging trends for better decisions (C.

Verma and R. Pandey, 2016). Real-time delays in

business processes result in both lost market

opportunities and ineffective marketing strategies

along with ineffective crisis response measures. The

formal nature of social media content along with its

rapid updates creates obstacles such as informal

language and abbreviated writing that make tweet

sentiment detection jobs more challenging. Phrases

must be classified in real-time through an accurate

and efficient platform for individuals and

organizations requiring upcoming insight analysis (S.

Sanjana et al., 2024).

Sentiment analysis automation occurs through the

implementation of machine learning models to tackle

the noted challenges. The current sentiment analysis

techniques need intensive data processing prior to

analysis because they create high processing costs

that restrict their ability to perform real-time

operations. From 2016 to 2018 big data platforms

together with distributed computing networks

facilitated the expansion of sentiment analysis

solutions on a large scale (A. Kc and R. Sumathi,

2018). System performance receives enhancement

through the combination of data preprocessing with

PySpark and sentiment classification with machine

learning algorithms as well as web applications and

real-time streaming platforms for data flow. The

proposed system brings together machine learning

algorithms with streaming and big data approaches to

Agrahari, S. K., Das, A. K., Bhagat, K., Shah, V. K., Sharma, N. and Ramasamy, G.

Dynamic Sentiment Analysis: A Low-Latency System for Social Media Monitor ing.

DOI: 10.5220/0013940500004919

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies (ICRDICCT‘25 2025) - Volume 5, pages

641-649

ISBN: 978-989-758-777-1

641

run real-time sentiment analysis. The preprocessing

tasks and model training process relieve PySpark as

the main engine while logistic regression emerged as

the optimal classifier method during assessment. The

developed Django web application allows users to

enter tweets for immediate sentiment analysis

prediction processing. The system implements Kafka

(R. Shree et al., 2017) as its real-time data streaming

platform to maintain system component connection

while using MongoDB as a NoSQL database to store

and retrieve sentiment prediction data efficiently. The

design features computation efficiency together with

quick processing time and easy usability to make

sentiment analysis more efficient and accessible for

users. The research presents a list of its main

achievements as follows.

• Real-time sentiment classification for instant

analysis of social media data.

• Integrates multiple technological systems to

build an efficient high-performance solution.

• Demonstrates in processing rapid and

unstructured data found in social media

platforms.

• Potential for scalability and adaptation to other

real-time text analysis applications.

The research supports the United Nations

Sustainable Development Goal on industry,

innovation, and infrastructure (UN SDG-9) for the

development of an innovative sentiment analysis

system through advanced technologies. The rest of the

paper is structured as follows: Section II covers a

review of existing methods regarding real-time

sentiment analysis. Section III describes the twitter

dataset which is used for the sentiment analysis. In

Section IV, the system architecture is presented along

with the description of how the machine learning

models work together with streaming frameworks and

database management. Section V includes

information about data pre-processing procedures and

model training and is evaluated through accurate

results along with computational efficiency and real-

time processing abilities. Finally, Section VI

concludes with an overview of essential discoveries

together with prospective enhancements for the work.

2 RELATED WORK

A huge volume of data is produced per second which

needs a system to analyze, and the opinions of people

need to be processed with high accuracy and the result

should be used for the improvement in the field.

Arokia et al. produced positive and negative labels on

tweets through the application of sentiment (Mary et

al., 2021) analysis on Twitter data from 2020-2021

using Linear SVC with high training precision and

Logarithmic Regression achieving maximum testing

accuracy to demonstrate sentiment analysis has

practical value for public emotion prediction. Madhu

et al., (2021) conducted real time sentiment analysis

on Twitter tweets through big data analysis alongside

tools including Hive and Machine Learning algorithm

to achieve fast and secure data processing. The K-

Means method and TF-IDF models identify tweets

among three clusters: fresh or deteriorating or stale

equilibrium through classification and the output

judgment determines positive or negative or neutral

sentiment based on total analysis. Alawad et al.,

(2021) examined Hadoop cluster performance

through an investigation of configuration

dependencies and Map Reduce model advantages for

big data analysis while reporting positive effects from

Hadoop cluster and Map Reduce model applications

in big data analysis. However, the paper notices

performance benefits from increasing dataset size in

large data analysis although significant cluster size

growth diminishes communication and lengthens

CPU times. Arafat et al., (2017) built VIM as a web-

based tool that allows users to visualize generic data

while also performing data pre-processing and mining

tasks and data drift analysis through statistics and

functions for feature-based association rule mining.

The VIM tool functions with Python Django Web

Framework and incorporates Graph Lab library for its

implementation.

Vatambeti et al., (2024) performed lexicon-based

sentiment analysis on Indian Twitter tweets about

Swiggy, Zomato and UberEats utilizing deep learning

methods to supply statistical feedback that supported

company recommendations. Zomato obtained the

highest positive ratings among all food delivery

services while showing the smallest proportion of

negative reviews. Ismail et al., (2025) featured a new

ETL framework that uses big data tech for Twitter-

based sentiment analysis while delivering efficient

processing of data streams together with bias

correction capabilities and sentiment analysis and

geographic visualization of Twitter data. Joloudari et

al., (2023) evaluated sentiment analysis systems for

COVID-19 tweets by studying BERT together with

deep CNN and demonstrated how these approaches

successfully extract tweet meaning while creating

embedding structures. The research has enhanced

sentiment detection abilities, and it provides guidance

for designing an efficient lightweight BERT model.

Singh et al., (2023) utilized data from WHO and CDC

along with social media sentiment data which is

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

642

analyzed using classification and regression models

with healthcare data between 2010 and 2020 for

predicting diseases and analyzing public health

sentiments.

Alqarni et al., (2023) examined COVID-19’s

influence on public emotions through both CNN and

BiLSTM model approaches of Arabic tweets. The

research proved that negative sentiments grew

significantly before and after the pandemic’s

outbreak and it demonstrated superior efficiency in

sentiment classification since it obtains 92.80%

accuracy with CNN and 91.99% accuracy with

BiLSTM. Yadranjiaghdam et al., (2017) discussed in-

memory processing for real- time Twitter data

analysis while studying current workflows and

developing a new framework using Apache Kafka for

data intake followed by Spark execution of real-time

processing and machine learning methods using

earthquakes in Japan as a case study for evaluating

origin analysis with timing and public response

evaluation. Fahd et al., (2021) proposed a real-time

sentiment analysis platform that combines multiple

social media data with Big Data technology through

Apache Kafka as a data collector and a lexicon-based

algorithm together with Spark analytics, YARN

resource scheduler and MongoDB for data storage

while evaluating multiple performance measures.

Mane et al., (2014) executed a sentiment analysis

strategy based on Hadoop infrastructure to handle the

extensive daily tweet volume and create time-

sensitive industrial and business insights with

improved solution speed from distributed computing

systems. Bikku et al., (2016) addressed the challenges

faced by the big data and highlighting Hadoop’s

architecture and efficiency with the help of Map

Reduced Framework. Ganesh et al., (2016) explored

biga data analytics for processing large image data

using Hadoop for handling the large dataset and for

efficient data. Saravanan et al., (2018) evaluated big

data analytics with Hadoop architecture and Spark

making a GUI for a user-friendly interaction. Singh et

al., (2015) evaluated and analyzed the presented

architecture which handles the huge data generating

per day with Hadoop explaining the challenges faced

by modern architecture. Radhika et al., (2017)

performed a sentimental analysis on Tamil news feed

based on POS Tagger based on characteristics and

entities of the various topics. Despite various analysis

and evaluations in tweet sentiment and feedback or

opinion of product, no paper has existed that

introduces a standardized system which processes

massive data inputs to evaluate product quality and

determine positive or negative ratings through

Docker and Kafka for comment-live streaming.

3 DATASET DESCRIPTION

The paper utilizes social media tweets primarily

sourced from Twitter into its dataset. The dataset

provides the necessary base for building and training

the sentiment analysis model so it can be applied in

real-world scenarios. Each data entry in the provided

dataset consists of textual information linked to

accompanying Metadata in its word file format. The

dataset contains 80,000 records which hold one

distinct entry for each record. The designed structure

enables detailed research of sentiment patterns within

multiple sources of input data. The dataset fulfills both

the requirements of being reliable and adequate for

developing the analysis model.

Figure 1: Sample dataset of the tweets.

The dataset demonstrates broad and extensive

content from a methodological standpoint. A

representative part of the dataset appears in the above

Figure 1. The tweets receive classification based on

their sentiment into four color-coded sections which

include Positive comments marked green and

Negative comments marked red while Neutral

comments have a yellow designation together with all

other irrelevant content receiving purple notification.

The dataset contains numerous tweets which ensure

an ample amount of data both for training and

validation procedures and model testing operations.

Each individual tweet represents a sample number that

reveals complete details about social media

interactions through hashtag usage and share mentions

and URLs and emoji presence. Timestamp and

language details within the data enable additional

dimensions in the dataset.

Dynamic Sentiment Analysis: A Low-Latency System for Social Media Monitoring

643

4 METHODOLOGY

The methodology of this work explains the

systematized procedure followed for the purpose of

attaining real-time sentiment prediction concerning

the tweets available in the data stream from twitter

platform. With the implementation of the system

involving several phases, the process helps in

achieving its success factors. Figure 2 is the

architecture diagram that depicts the overall flow and

interaction between various components. It includes

the integration of machine learning, big data analytics,

use of web technologies and real time streaming for

building comprehensive sentiment analysis systems.

Figure 2: Architecture diagram for sentiment analysis.

The approach is divided into five key stages: Data

pre-processing, model building, GUI creation, live

stream processing, and Data storage. All the stages

play the role of contributing to designing a scalable,

efficient, user-friendly system. Each stage is

connected to each other making it easy to pass data

from one stage to the other while keeping the analysis

in real time. The design is modular, which provides

the space for the further changes that are to be made

to the system later.

5 DATA PREPROCESSING WITH

PYSPARK

Preprocessing of data is important in text data

preparation for the learning process in the case of text

data. The tweets to be analyzed are extremely raw and

basic pre-processing is needed to modify and arrange

the data so that the model can correctly identify the

correct sentiment.

The preprocessing process included the following key

steps:

Data Cleaning: Data cleaning took place and

resulted in the removal of empty rows as well

as removal of special characters that have no

meaning to the algorithm. Also, to make the

input more manageable, the text is translated

to lowercase, thus creating a laid-back format.

Tokenization: Each sentence is split into

individual words (tokens) for more granular

analysis. This tokenization enabled the model

to focus on meaningful word-level patterns.

Spark Capabilities: For high processing

velocity and keeping the system scalable, the

distributed features of PySpark are used to

handle large datasets. This feature was

especially helpful because the content of real-

time tweet analysis changes frequently.

To overcome such issues as the issue of class

imbalance in the dataset an analysis was made on the

four classes of sentiment which are as follows. This

imbalance is illustrated in Figure 3, and requires

subsequent, but similar techniques such as

oversampling or under sampling of the dataset to

enhance the model.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

644

Figure 3: Class imbalance in dataset.

The insights from this visualization informed

specific preprocessing strategies, ensuring the data

was optimized for training and real-time predictions.

6 MODEL TRAINING

After preprocessing data, the preprocessed data was

used to train a sentiment analysis model. Several

machine learning algorithms were applied on the

dataset and evaluated. Finally, the logistic regression

model was chosen for this dataset because of its

accuracy and suitability for the dataset.

Algorithm Selection: Several classification

algorithms were tested from PySpark MLlib,

including logistic regression, decision trees,

and support vector machines.

Performance Evaluation: Cross-validation

techniques were applied to optimize

hyperparameters and model performance.

Models were evaluated using different metrics

such as accuracy, precision and recall.

Pipeline Creation: The PySpark pipeline was

constructed that includes preprocessing steps

and the logistic regression model for

streamlined predictions.

7 GUI CREATION WITH

DJANGO

A high-level Python web framework, Django, was

used to develop web applications, enabling users to

interact with the system and analyze tweets

dynamically. A form has also been designed in the

GUI that allows users to input tweets for sentiment

analysis. The system ensured real-time feedback by

displaying results instantly on the web interface.

8 REAL-TIME STREAMING

WITH KAFKA

Kafka was implemented in docker containers to

facilitate real-time data transmission between

components, ensuring seamless interaction and low-

latency processing. Kafka producer reads tweets

from a CSV file and sends them to a Kafka topic

(numtest) every three seconds to simulate real-time

data streaming. The consumer uses PySpark to load

a trained and stored logistic regression model. Then,

once Kafka consumer Retrieves tweets from the

numtest topic in real-time, it processes each tweet

through the PySpark pipeline for sentiment

prediction. Finally, Smooth communication was

established between the Kafka consumer and the web

interface to display the prediction result instantly.

9 DATA STORAGE WITH

MONGODB

A NoSQL database, MongoDB, was used to store

sentiment analysis results and other related data. Each

processed tweet was stored in MongoDB as a

document containing the original text and the

predicted sentiment. MongoDB’s flexibility was

utilized to handle semi-structured data and support

future data analysis tasks.

This methodology ensures a robust, scalable, and

user-friendly system for real-time sentiment analysis,

integrating cutting-edge tools and frameworks for

optimal performance.

10 RESULTS

The findings presented in this paper prove that the

proposed real-time sentiment analysis system

possesses high accuracy and scalability of sentiments.

It achieves preprocessing tweet data, building various

machine learning models, and real-time processing of

tweets stream to provide a picture of the distribution

of sentiments. The logistic regression classifier

achieved the most accurate results during the testing

and validation of the developed machine learning

Dynamic Sentiment Analysis: A Low-Latency System for Social Media Monitoring

645

model. Metrics of cross-validation showed good

performance on different datasets, and this meant that

the model was able to perform well on any different

inputs. For measuring the performance, the

parameters used are accuracy, precision, recall, and

F1 score. Other than the logistic regression, Random

Forest and Decision Tree models are also used during

the experimentation phase of the thesis as depicted in

Figure 4.

Figure 4: Models used for analysis.

Apache Kafka for real-time streaming integration

empowered the system to analyze tweets as they

arrived in the system. The free sentiment that is

predicted about each tweet that is analyzed is whether

it is positive, negative, neutral, or irrelevant on the

web interface in a matter of seconds. It kept low

latency for the system that keeps user interaction

smooth and without any interruption. Figure 5

illustrates the dashboard which shows how at the end

of the given comment it times it is categorized and

then placed under the sentiment’s column. Also, it

gives the user a qualitative opinion of the expected

result, which is easy to understand.

Figure 5: Dashboard for analysis of the tweet.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

646

Figure 6: MongoDB database for storing the tweets.

Figure 6 shows the usage of the MongoDB

database allowed for storing of the predicted

sentiment outcomes, as well as the tweets with their

references within a reasonable amount of time and

space. To further illustrate the potential of the stored

data, patterns over time are identified or recurring

tweet sentiments, which would enforce further

applications like market research or social sentiment

analysis.

The real-time update of the web interface made it

possible to visualize the trend of sentiment made by

the user. Figure 7 shows the interface to visualize the

distribution of the tweets along with its classification.

A table is presented with the list of tweets, and the

tweets are classified in a particular class in the target

column; thus, it is convenient for the users to compare

the results of the predictions with the expected results.

A pie chart is also introduced for the sentiment

distribution for different classes within the interface.

A bar graph below depicted the distribution

percentage of positive, negative, and neutral

sentiments depending on tweets by users. Through

this feature, a way of analyzing social media trends is

made available, by observing the dynamic features of

sentiment patterns.

These results provide evidence of the

effectiveness of planning and executing the presented

system to process large and constantly updating

datasets, as well as generate accurate sentiment

predictions. For this reason, the scalability and the

high resilience of the proposed system recommends

the developed model for real-world sentiment

analysis applications.

Figure 7: Web interface to visualize the classification and

distribution of a tweet.

11 CONCLUSIONS

The work done in this project is quite effective to

underscore how progressive technologies can be

Dynamic Sentiment Analysis: A Low-Latency System for Social Media Monitoring

647

applied to make sentimental analysis of tweets in real

time. The system uses PySpark for the data

preprocessing and machine learning algorithm,

Django for building an appealing web user interface,

Kafka for effective real time stream processing, and

MongoDB for storing big amounts of data in

optimized manners makes the present system a

perfect, efficient and effective solution for the

sentiment analysis.

As a result of the process of model selection

during model evaluation, the selected logistic

regression model offers accurate and timely

classification of tweet sentiments. The employment

of real-time streaming guarantees that users can

receive sentiment predictions with least delay,

making the application fast and efficient. Also, the

use of web interface is rather convenient to interact

with, as well as entering and analyzing tweets which

enable users to get the idea of sentiments’

distributions. Not only does the project prove the

possibility of carrying out real-time sentiment

analysis, but also the need for combining machine

learning with big data and web frameworks. The

system proposes complex factors of performance:

scalability, operational productivity and adaptability,

which allows considering it as a perspective for usage

in such fields as social media monitoring, market and

trend analysis.

Two suggestions for future work are the

expansion of the current system with features such as

multilingual sentiment analysis, topic extraction and

the addition of an emotion recognition feature. An

uplift in model performance, employing state-of-art

deep learning techniques, can act as a further

enhancement of the proposed system. In sum,

implementing this project means a major shift

towards utilizing real time sentiment analysis for real

life application which will prove as handy and helpful

for the purpose of meaningful decision making in a

constantly evolving digital landscape.

REFERENCES

A. Kc and R. Sumathi, “Analyzing twitter sentiments with

big data,” in 2018 2nd International Conference on

Trends in Electronics and Informatics (ICOEI), pp.

989–994, IEEE, 2018.

A. Alqarni and A. Rahman, “Arabic tweets-based sentiment

analysis to investigate the impact of covid-19 in ksa: a

deep learning approach,” Big Data and Cognitive

Computing, vol. 7, no. 1, p. 16, 2023.

A. Ismail, F. H. Sazali, S. N. A. Jawaddi, and S. Mutalib,

“Stream etl framework for twitter-based sentiment

analysis: Leveraging big data technologies,” Expert

Systems with Applications, vol. 261, p. 125523, 2025.

B. Yadranjiaghdam, S. Yasrobi, and N. Tabrizi,

“Developing a real-time data analytics framework for

twitter streaming data,” in 2017 IEEE International

Congress on Big Data (BigData Congress), pp. 329–

336, IEEE, 2017.

Bikku, T., Rao, N.S. and Akepogu, A.R., 2016. “Hadoop

based feature selection and decision-making models on

big data”. Indian Journal of Science and Technology,

9(10), pp.1-6.

C. Verma and R. Pandey, “Big data representation for grade

analysis through hadoop framework,” in 2016 6th

international conference-cloud system and big data

engineering (Confluence), pp. 312–315, IEEE, 2016.

Dr. Tripty Singh and S, D. V., “A modern data architecture

with apache Hadoop”, in 2015 International Conference

on Green Computing and Internet of Things

(ICGCIoT), Greater Noida on the beautiful campus of

Galgotias Institutions , 2015.

G. P. A. Mary, M. Hema, R. Maheshprabhu, and M. N.

Guptha, “Sentimental analysis of twitter data using

machine learning algorithms,” in 2021 International

Conference on Forensics, Analytics, Big Data, Security

(FABS), vol. 1, pp. 1–5, IEEE, 2021.

G.S. Ganesh, Yazhini Samyukkktha Ramanikaran,

M.Sindhuja, G. Rajesh Kannan, J. Jeyalakshmi, “Big

data in visual analytics to supplement digital image

processing using Hadoop”, Journal of Chemical and

Pharmaceutical Sciences,JCHPS., Special Issue

9,pp.54-56,Dec,2016.(SCOPUS) ISSN: 0974-2115

J. H. Joloudari, S. Hussain, M. A. Nematollahi, R. Bagheri,

F. Fazl, R. Alizadehsani, R. Lashgari, and A. Talukder,

“Bert-deep cnn: State of the art for sentiment analysis

of covid-19 tweets,” Social Network Analysis and

Mining, vol. 13, no. 1, p. 99, 2023.

K. S. Madhu, B. C. Reddy, C. Damarukanadhan, M.

Polireddy, and N. Ravinder, “Real time sentimental

analysis on twitter,” in 2021 6th International

Conference on Inventive Computation Technologies

(ICICT), pp. 1030–1034, IEEE, 2021.

K. Fahd, S. Parvin, and A. de Souza-Daw, “A framework

for real-time sentiment analysis of big data generated

by social media platforms,” in 2021 31st international

telecommunication networks and applications confere-

nce (ITNAC), pp. 30–33, IEEE, 2021.

R. Shree, T. Choudhury, S. C. Gupta, and P. Kumar,

“Kafka: The modern platform for data management and

analysis in big data domain,” in 2017 2nd international

conference on telecommunication and networks (TEL-

NET), pp. 1–5, IEEE, 2017.

R. G., K. Prasath, N., and Rajasundari, T., “Sentimental

Analysis on Tamil News Feed”, Journal of Advanced

Research in Dynamical and Control Systems, vol. 9, pp.

397-399, 2017.

R. Singh and R. Singh, “Applications of sentiment analysis

and machine learning techniques in disease outbreak

prediction–a review,” Materials Today: Proceedings,

vol. 81, pp. 1006–1011, 2023.

ICRDICCT‘25 2025 - INTERNATIONAL CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION,

COMMUNICATION, AND COMPUTING TECHNOLOGIES

648

R. Vatambeti, S. V. Mantena, K. Kiran, M. Manohar, and

C. Manjunath, “Twitter sentiment analysis on online

food services based on elephant herd optimization with

hybrid deep learning technique,” Cluster Computing,

vol. 27, no. 1, pp. 655–671, 2024.

S. B. Mane, Y. Sawant, S. Kazi, and V. Shinde, “Real time

sentiment analysis of twitter data using hadoop,”

IJCSIT) International Journal of Computer Science and

Information Technologies, vol. 5, no. 3, pp. 3098–3100,

2014.

S. S. I. Arafat, M. S. Hossain, M. M. Hasan, S. A.-H. Imam,

M. M. Islam, S. Saha, S. Shatabda, and T. I. Juthi,

“Vim: A big data analytics tool for data visualization

and knowledge mining,” in 2017 IEEE International

WIE Conference on Electrical and Computer

Engineering (WIECON-ECE), pp. 224–227, IEEE,

2017.

S. Saravanan, K.E., K., Balaji, A., and Sajith, A.,

“Performance Comparison of Apache Spark and

Hadoop Based Large Scale Content Based Recomme-

nder System”, Intelligent Systems Technologies and

Applications. Springer International Publishing, Cham,

2018.

S. Sanjana, N. J. Rozario, S. Dewan, et al., “Utilizing big

data analytics for social media insights,” in 2024

International Conference on IoT Based Control

Networks and Intelligent Systems (ICICNIS), pp. 45–

49, IEEE, 2024.

W. Alawad and A. Balobaid, “In big data era: analysis of

hadoop cluster performance,” in 2021 International

Conference of Women in Data Science at Taif

University (WiDSTaif), pp. 1–6, IEEE, 2021.

Dynamic Sentiment Analysis: A Low-Latency System for Social Media Monitoring

649