AI‑Enabled Academic Conference Management System

Ranjit Reddy Midde, Geetha Seri Lingala, Hemalatha Dubasi,

Gayathri Mangala and Prashanth Kumar Reddy Pullaiahgari

Department of CSE, Srinivasa Ramanujan Institute of Technology, Rotarypuram Village, B K Samudram Mandal,

Anantapuramu - 515701, Andhra Pradesh, India

Keywords: Support Vector Machines (SVM), K‑Nearest Neighbors (KNN), Random Forest Algorithm.

Abstract: Academic conferences play a vital role in fostering knowledge sharing and collaboration among researchers

and experts. However, managing these events involves complex tasks such as paper submission, peer review,

scheduling and communication which are often inefficiently handled. Existing methods and systems show

significant limitations such as inconsistent reviewer assignments, reliance on manual sorting of papers and

limited interaction between authors and reviewers. These issues make the conference process less efficient.

Addressing these issues, AI-enabled Academic Conference Management System aims to overcome these

challenges by using machine learning techniques such as Support Vector Machines, K-Nearest Neighbors and

Random Forest to automatically classify research papers by domain and assign reviewers based on domain

expertise. Additionally, it introduces organized feedback mechanisms via email and supports direct author-

reviewer collaboration through video conferencing to facilitate the review process. The system reduces

manual effort by automating paper classification, ensuring accurate matches, user-friendly interface enables

faster workflows and a more efficient review process.

1 INTRODUCTION

Academic conferences are essential platforms for

researchers and experts to share knowledge, foster

collaborations, and disseminate groundbreaking

research. However, managing these conferences is a

complex process that involves numerous

interdependent tasks, including paper submissions,

reviewer assignments, session scheduling, and

participant communication. Traditional conference

management systems often struggle to address these

challenges effectively. Issues such as manual sorting

of research papers, inconsistent reviewer

assignments, and inadequate communication

channels between authors and reviewers lead to

inefficiencies, delays, and potential errors. For

instance, manual paper review processes are prone to

human error, while poorly matched reviewers may

result in superficial evaluations, negatively impacting

the quality of feedback and the overall conference

experience.

To overcome these challenges, this research

introduces an AI-enabled Academic Conference

Management System designed to optimize and

streamline conference workflows. The system

leverages advanced machine learning techniques

such as Support Vector Machines (SVM), K- Nearest

Neighbors (KNN), and Random Forest for

automating classification of research papers into

specific domains, ensuring precise reviewer

assignments based on domain expertise.

Additionally, the system integrates innovative

features such as structured feedback mechanisms and

direct collaboration through video conferencing.

These enhancements facilitate seamless

communication between authors and reviewers,

accelerate the review process, and foster meaningful

interactions. By reducing manual effort, improving

collaboration and due to a user-friendly interface the

proposed system aims to transform academic

conference management, addressing inefficiencies

and promoting a more organized and effective review

process.

2 RELATED WORK

Several systems for managing conferences are

developed to streamline various processes like paper

submission, review, and feedback.

328

Midde, R. R., Lingala, G. S., Dubasi, H., Mangala, G. and Pullaiahgari, P. K. R.

AI-Enabled Academic Conference Management System.

DOI: 10.5220/0013927800004919

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

328-334

ISBN: 978-989-758-777-1

M. Spilker, F. Prinsen, and M. Kalz (2019)

analyzed the impact of technology-enhanced

academic conferences on professional development,

highlighting benefits such as improved accessibility,

interactive learning, and global reach, while

addressing challenges like equitable access and

interaction quality. Similarly, MyConfree, developed

by Metta Santiputri Nindy et al. (2018), leverages

PHP, CodeIgniter, and MySQL to streamline

workflows with features like paper submissions, call-

for-paper announcements, and traditional review

processes, demonstrating how technology simplifies

conference management.

Other systems further emphasize innovation in

conference workflows. Cheng Zheng et al. (2008)

introduced a collaborative platform with tools for

real-time updates, session scheduling, and participant

registration to improve coordination. K. Ahmad et al.

presented a system tailored for higher education,

offering features like author notifications and

deadline alerts. OpenConf, introduced by the Zakon

Group (2001), simplifies tasks such as peer reviews,

attendee registration, and scheduling with its user-

friendly interface, reducing administrative burdens

and improving the experience for organizers and

participants alike.

A. Malinowski and B. Wilamowski, in their work

titled “Paper Collection and Evaluation through the

Internet” (Proceedings of the 27th Annual

Conference of the IEEE Industrial Electronics

Society, Denver) , addressed limitations in standalone

web-based conference management systems. These

systems often lack fault tolerance, and the capacity to

support distributed users. To overcome these

challenges, they proposed a collaborative conference

management system that utilizes advanced

collaboration technologies to improve fault tolerance,

scalability, and user accessibility.

Rigaux Ph., in “An Iterative Rating Method:

Application to Web-based Conference Management”

(Proceedings of the 2004 ACM Symposium on

Applied Computing), proposed a method for

collecting user preferences or ratings on a large set of

items without asking each user to rate each one. The

approach relies on an iterative process, where each

step, or ballot, involves users rating a sample of items.

A collaborative filtering algorithm predicts the

missing ratings along with their confidence levels,

which are initially set to zero. Subsequent ballots

improve the prediction accuracy, and the system

administrator determines when to stop the iteration

upon reaching a satisfactory level. This method was

applied to assign reviewers to papers prior to the

review phase in conference management and was

implemented in the MYREVIEW web-based system.

A web-based academic conference management

system called ConfSys (Huang et al., 2008) is

introduced. ConfSys is intended to assist program

chairs, general chairs, and program committees in

overseeing the operations of scholarly conferences

and to offer conference-related services to authors

and attendees. It can post papers, assign them to

reviewers automatically, let the chairs change the

assignment, debate and rate papers, create the

program, register for conferences, gather presentation

slides, and more. These features facilitate rapid and

simple conference management.

Ware M., in “Online Submission and Peer‐

Review Systems” (Learned Publishing, Vol. 18, No.

4, pp. 245-250, 2005), explored the adoption of online

systems for managing submissions and peer reviews

in academic publishing. The study highlighted the

benefits of such systems, including streamlined

submission processes, efficient reviewer

assignments, and improved communication between

authors, reviewers, and editors. Organizing scientific

conferences involves managing paper submissions

and reviews, which are crucial yet often complicated

tasks. Existing systems are complex, with many

unused features, and typically rely on hosting

services, raising concerns about data security. This

study presents the Online Paper Submission System

(OPSS) (Rotikan, Reymon. 2016), an online

application created to streamline the process of

submitting and reviewing papers.

3 PROPOSED SYSTEM

3.1 Workflow of Proposed System

The workflow of AI-Enabled Academic Conference

Management System consists of the following major

steps: Step 1: Abstract Submission – Authors submit

abstracts through a user- friendly web interface,

ensuring ease of access. Step 2: Domain

Classification – These abstracts are classified into

specific domains using advanced machine learning

techniques like SVM, KNN, and Random Forest,

enabling accurate domain categorization. Step 3:

Reviewer Assignment – Domain-specific reviewers

are assigned to evaluate the abstracts and provide

feedback. Step 4: Full Paper Submission – Abstracts

that meet the required standards lead to the

submission of full papers for further review. Step 5:

Full Paper Evaluation – These full papers undergo a

thorough evaluation process, and feedback is sent to

the authors via email. Step 6: Collaboration – Video

AI-Enabled Academic Conference Management System

329

conferencing tools facilitate discussions between

authors and reviewers, fostering better

communication and understanding.

Figure 1 depicts the suggested system’s structured

process. It begins with the submission of an abstract,

which is classified into a relevant domain using

machine learning algorithms and ends with the paper

submission.

Figure 1: Proposed system workflow.

3.2 Architecture of the Proposed

System

Three main roles are included in the centralized

database and role-based access control architecture of

the AI-Enabled Academic Conference Management

System: Admin, Organizer, and User. Admin

manages critical functionalities such as creating

conferences, uploading research papers, and

assigning reviewers to specific domains. Organizers

act as intermediaries, handling the creation and

management of authors and reviewers while ensuring

smooth workflow execution.

Figure 2: Architecture of proposed system.

Users submit abstracts through a user-friendly

web interface, which are then classified into relevant

domains using machine learning algorithms like

SVM, KNN, and Random Forest. The system ensures

efficient handling of data through secure storage and

retrieval in the database, with seamless interaction

between roles and real-time collaboration enabled via

integrated video conferencing tools. This architecture

provides a streamlined and scalable framework for

managing academic conferences (figure 2).

The system architecture showcases a role-based

have distinct functionalities. The Admin manages

conferences, uploads papers, and assigns reviewers,

while the Organizer handles creating authors and

reviewers. Users can log in to predict domains, view,

and download papers, with all actions interacting with

a centralized database.

4 IMPLEMENTATIONS

4.1 Dataset

Abstracts were manually curated from research

papers available online. Each abstract was labeled

with its respective domain like cybersecurity,

blockchain, cloud computing etc. In the dataset

preparation stage, the dataset is loaded from an Excel

file, where the first column represents the textual

content (abstracts), and the second column contains

the corresponding labels. The dataset is examined

to identify the number of rows and columns for

a clearer understanding of its structure. Text data

undergoes preprocessing to ensure uniformity and

remove noise. This requires converting text to

lowercase, deleting digits and special characters,

eliminating stopwords, and performing

lemmatization to retrieve the base form of words.

After processing, the text input is vectorized to

provide numerical feature representations using the

TF- IDF (Term Frequency-Inverse Document

Frequency) technique. Simultaneously, the labels are

encoded into numerical values using a label encoder.

Finally, the data is then divided into two parts: 80%

for training the model and 20% for testing it. This

ensures a fair evaluation of the model's performance.

4.2 Algorithms Used

Three machine learning techniques were

implemented for classifying the abstracts into specific

domains as follows: Support Vector Machine (SVM)

was used to classify research paper abstracts by

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

COMMUNICATION, AND COMPUTING TECHNOLOGIES

330

creating the best possible boundary in a high-

dimensional space. This boundary helps separate the

abstracts into their respective domains. The main goal

of SVM is to ensure there is as much space as possible

between this boundary and the closest data points

from each category, which are called support vectors.

This approach improves classification accuracy by

clearly distinguishing different domains. The SVM

classifier was particularly effective for this project

because of its ability to handle high-dimensional data,

making it ideal for the TF-IDF feature representation

of abstracts. Random Forest was used as an ensemble

learning algorithm to improve classification

performance by constructing multiple decision trees

and aggregating their predictions through majority

voting. Each decision tree in the forest was trained on

different bootstrap sample of dataset, and feature

selection was performed randomly at each split,

ensuring diversity among the trees. K-Nearest

Neighbors (KNN) was employed as a simple,

instance-based learning algorithm to classify

abstracts by measuring their proximity to labeled

examples in the feature space. KNN will operate by

calculating the Euclidean distance in between a test

sample and all other training samples, selecting k

nearest neighbors and determining the class label

based on the category that appears most frequently

among them.

4.3 Model Training

The Support Vector Machine (SVM), Random

Forest, and K-Nearest Neighbors (KNN) models are

trained using data that has been split into 80% for

training and 20% for testing to ensure reliable

evaluation, where 80% of the 330 manually labeled

abstracts were used for training and 20% for testing.

SVM was tuned to find an optimal hyperplane for

separating the data based on their respective domains.

Random Forest utilized multiple decision trees,

aggregating results through majority voting to

improve classification accuracy. KNN classified

abstracts based on the proximity of test samples to

labeled training examples, using distance-based

classification.

4.4 Model Performance Evaluation

The evaluation focuses on accuracy, precision, recall,

and F1-score, providing a clear understanding of how

well each model performs in classifying abstracts,

assigning reviewers, and facilitating the overall

workflow. Through these metrics, we aim to validate

the efficiency and reliability of the implemented

algorithms in real-world conference management

scenarios.

4.5 Web Development

A relational database structure to store and manage

information, including author submissions, reviewer

assignments, paper classifications, review statuses,

and video conference schedules. The below Figure 3

depicts ER diagram that illustrates the structure and

relationships within the database. It showcases the

entities, their attributes, and the connections between

them, providing a clear overview of how data is

organized and interrelated.

Figure 3: Entities and their relations.

The AI-Enabled Academic Conference Management

System uses Django's built-in SQLite database to

manage all application tables efficiently. Django's

ORM automatically translates models into tables and

handles data operations like insertion, updates, and

queries. SQLite's lightweight nature and seamless

integration with Django make it an ideal choice. The

database consists of the following tables:

main_author, main_chair, main_conference,

main_reviewer, main_user, main_creates,

main_paper, main_schedules, main_submits,

main_assigns, and main_reviews. The database

designed to manage various entities and operations

within the system. to the system's functionality. The

AI-Enabled Academic Conference Management

System was built using Django's MVT (Model-

View-Template) architecture, where the Model

defines the application's data structure and handles

database operations, such as managing entities like

authors, reviewers, papers, conferences, and

schedules. The View serves as the intermediary,

processing user requests, manipulating data through

the Model, and passing it to the Template for

rendering. The Template is responsible for the user

interface, dynamically presenting data using

AI-Enabled Academic Conference Management System

331

Django's templating engine, with templates designed

for pages like the home page, domain prediction,

reviewer assignment, submission status, and

schedule, ensuring a user-friendly and visually

appealing interface.

4.6 Integration of Machine Learning

Model into the Web Page

Machine learning model, trained to classify abstracts

into specific domains using algorithms like Support

Vector Machines (SVM), Random Forest, and K-

Nearest Neighbors (KNN), was seamlessly integrated

into the Django-based MVT (Model-View-Template)

architecture. After training and evaluating the model

using metrics like recall, precision, and F1-score,

final model was serialized and saved using the joblib

library for efficient reuse in the application. The

integration process of the trained machine learning

model into the MVT architecture involved several

steps. First, the model was serialized and saved as a

.joblib file, enabling efficient loading during runtime

without the need for retraining. In the View

component of the Django framework, the saved

model was loaded and utilized for predicting the

domain of submitted abstracts. Upon submission, the

abstract data was preprocessed and passed to the

model for classification. The classification results

were then stored in the Model component of the

application, associating each abstract with its relevant

domain to ensure the appropriate assignment of

reviewers based on domain expertise. Finally, the

prediction results were dynamically displayed on the

Template pages, such as the abstract submission

status page and reviewer assignment page, providing

users with real-time feedback and facilitating a

streamlined workflow.

5 RESULTS

Table 1: Performance ranking and accuracy.

Ran

Algorith

Precision Recall F1-Score Accurac

Random

Forest

84% 85% 84% 89%

Support

Vector

Machine

(SVM)

87% 86% 86% 86%

K-Nearest

Neighbors

(

KNN

)

76% 82% 76% 82%

Performances of different Algorithms were checked

and contrast has been made to identify the most

efficient algorithm for the domain prediction of the

uploaded abstract.

Table 1 represents performance comparison of the

machine learning techniques used in the system based

on metrics like F1-score, accuracy, precision, and

recall. Random Forest algorithm achieved the highest

accuracy at 89%, with strong precision and recall

values, making it the top- performing model. The

Support Vector Machine (SVM) closely followed,

demonstrating balanced performance across all

metrics. K-Nearest Neighbors (KNN), while showing

reasonable recall, had lower precision and F1-scores,

ranking third. This ranking highlights the

effectiveness of Random Forest for domain

classification tasks in the system.

Finally, the Random Forest model is seamlessly

integrated with the web interface using Django,

making the system highly user-friendly. This

integration allows for easy and accurate domain

prediction of abstracts. Additionally, other

functionalities, such as creating conferences and

reviewing abstracts, are designed with intuitive and

user-friendly interfaces. The AI-enabled Academic

Conference Management System is designed to

support four distinct user roles: Admin,

Chair/Organizer, Author, and Reviewer. Each role is

provided with a tailored interface that allows them to

perform specific tasks relevant to their

responsibilities in the conference management

process. The following key web pages illustrate how

the interface accommodates the needs of each role.

Figure 4: Home page.

Figure 4 depicts homepage that serves as a central

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

COMMUNICATION, AND COMPUTING TECHNOLOGIES

332

hub, offering intuitive navigation tailored to authors,

reviewers, and administrators. It streamlines access to

role-specific features, enabling authors to submit

papers, reviewers to evaluate assignments, and

administrators to manage workflows efficiently,

ensuring a seamless user experience.

Figure 5: Page for domain prediction.

Figure 5 depicts the abstract classification page,

where page highlights the predicted domain and

provides transparency into the classification results.

The machine learning model is integrated into the

webpage using Django, enabling seamless interaction

between the backend and the user interface. All the

conference details are stored in the backend using

sqlite database which is a built in database in Django.

Figure 6: Page for reviewer assignment.

Figure 6 depicts the reviewer assignment page, where

administrators assign reviewers to abstracts based on

their classified domains. The interface displays the

abstract details, predicted domain, and a list of

available reviewers with their respective expertise.

This ensures that reviewers are matched with

abstracts relevant to their domain, improving the

quality of feedback and reducing manual workload.

The streamlined process enhances the efficiency of

the review workflow.

6 FUTURE SCOPE

The future scope of the AI-enabled Academic

Conference Management System includes integrating

multiple conferences on a unified platform, enabling

centralized management and resource sharing. The

system can be enhanced with scalable conference

management, dynamic reviewer matching across

events, and real-time collaboration tools for authors

and reviewers. Mobile application development can

improve global accessibility, while AI-driven

analytics can offer valuable insights for organizers.

Additionally, integrating blockchain technology can

enhance transparency, and automated session

scheduling can streamline event organization, making

the platform a comprehensive solution for academic

conferences.

7 CONCLUSIONS

The AI-enabled Academic Conference Management

System successfully achieves its objectives of

automating abstract classification, streamlining

reviewer assignment, and enhancing collaboration

between authors and reviewers. By employing

machine learning techniques like Random Forest, K-

Nearest Neighbors, and Support Vector Machines,

the system ensures precise domain classification of

abstracts, facilitating efficient reviewer allocation.

Integrated tools, including structured feedback

mechanisms and video conferencing, further promote

effective communication and collaboration. The

platform addresses inefficiencies in traditional

conference workflows by providing an organized,

transparent, and user-friendly interface for all

stakeholders. With these advancements, the system

establishes a strong foundation for modernizing

academic conferences while meeting the objectives

set forth in the project.

REFERENCES

Cheng Zheng, Weiming Shen, Qinghua Zheng, Feng Tian,

"Design and Implementation of a Collaborative

Conference Management System", 2008.

Huang, M., Y. Feng, B. C. Desai. CONFSYS: A Webbased

Academic Conference Management System.

Proceedings of 2008 C3S2E conference. ACM Press,

2008.

K. Ahmad, A. A. Abdullah, and A. M. Zeki, “Web-based

Conference Management System for Higher Learning

Institutions,” International Islamic University Malaysi

AI-Enabled Academic Conference Management System

333

a, Kulliyyah of Information and Communication

Technology.

M. Spilker, F. Prinsen and M. Kalz, "Valuing technology-

enhanced academic conferences for continuing

professional development", A systematic literature

review Professional Development in Education, 2019.

Malinowski and B. Wilamowski, "Paper Collection and

Evaluation through the Internet", Proceedings of the 27

Annual Conference of the IEEE Industrial Electronics

Society, Denver.

Metta Santiputri, Nindy, Surya Agustin, MeraKartika

Delimayanti, “MyConfree: A Web- based Conference

Management System” Published by IEEE, Proceedings

of IEEE 2018.

Rigaux Ph. An Iterative Rating Method: Application to

Web-based Conference Management. Proceedings of

the 2004 ACM symposium on applied computing.

Rotikan, Reymon. 2016. OPSS: Online Paper Subm-ission

System Untuk Kegiatan Konferensi Ilmiah Cogito

Smart Journal/VOL.2/No/2/Desember 2016.

Ware M. “Online submission and peer‐review systems” in

learned publishing.

Zakon Group, “OpenConf Conference Management

System,” Web Site: www.openconf.org, 2001.

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

COMMUNICATION, AND COMPUTING TECHNOLOGIES

334