International Journal of Scientific Research in Computer Science, Engineering

and Information Technology

ISSN : 2456-3307 Available Online at : www.ijsrcseit.com
doi : https://doi.org/10.32628/CSEIT2410275

Resume Screening using Machine Learning

Dr. Sandeep Tayal1, Taniya Sharma1, Shivansh Singhal3, Anurag Kumar Thakur4
1Assistant Professor, Department of Computer Science, Maharaja Agrasen Institute of Technology, Delhi, New Delhi, India
2,3,4 UG Student, Department of Computer Science, Maharaja Agrasen Institute of Technology, Delhi, New Delhi, India

ARTICLEINFO ABSTRACT

Article History:
This study explores the utilization of Machine Learning (ML) and Natural
Language Processing (NLP) in automating the resume screening process.
Accepted: 10 April 2024
Traditional methods, often manual and subjective, fail to efficiently manage the
Published: 19 April 2024
volume and variety of resumes. By employing NLP techniques like named entity
recognition and part-of-speech tagging, coupled with ML classifiers such as K-
Nearest Neighbors and Support Vector Machines, we propose a system that
Publication Issue
enhances the precision of candidate selection while significantly reducing time
Volume 10, Issue 2
and effort.
March-April-2024
Keywords : Machine Learning, Natural Language Processing, Resume Screening,
Page Number
NLTK, K-Nearest Neighbors, Support Vector Machines.
602-606

I. INTRODUCTION a single job posting, a task that is not only time-

consuming but also fraught with the potential for
The modern job market is characterized by a dynamic oversight and bias. The traditional approach to resume
and ever-evolving landscape where both employers screening involves a manual review of each application
and job seekers strive for the best possible matches. to assess the candidate's fit based on experience,
Employers seek to fill positions with candidates who education, skills, and other relevant criteria. This
not only possess the necessary skills and qualifications process is not only labour-intensive but also subjective,
but also align with the company's culture and values. leading to inconsistencies in candidate selection.
Conversely, job seekers aim to find roles that leverage Furthermore, the manual screening process struggles
their strengths and offer growth opportunities. This to cope with the volume and variety of resumes, often
dual objective sets the stage for a complex resulting in qualified candidates being overlooked or
matchmaking process. Traditional resume screening the recruitment cycle being unnecessarily prolonged.
methods, heavily reliant on human judgment and In response to these challenges, our research proposes
manual effort, present significant challenges in an automated system that harnesses the power of
meeting these objectives efficiently. Recruiters often Machine Learning (ML) and Natural Language
sift through hundreds, if not thousands, of resumes for Processing (NLP) to revolutionize the resume

Copyright © 2024 The Author(s): This is an open-access article distributed under the terms of the Creative 602
Commons Attribution 4.0 International License (CC BY-NC 4.0)
 Dr. Sandeep Tayal et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol., March-April-2024, 10 (2) : 602-606

screening process. By automating the extraction and qualifications. Data for model training comes from
interpretation of information from resumes, this Kaggle.
system aims to significantly reduce the time and effort
B. Resume Screening Classification using Artificial
involved in screening candidates while improving the
Intelligence and Natural Language Processing
accuracy and objectivity of the selection process.
The paper "Resume Screening Classification using
ML algorithms can analyse vast amounts of data to Artificial Intelligence and Natural Language
identify patterns and make predictions. In the context Processing" introduces the Prospect model, a
of resume screening, ML models are trained to machine learning-based system for automating
categorize candidates based on their suitability for a resume screening. This model achieves a
role, using historical hiring data and outcomes as a remarkable accuracy of 93.5%, significantly
learning basis. This approach enables the system to outperforming traditional convolutional neural
evaluate candidates more accurately and consistently network models by 19.5%. It employs a two-phase
than human screeners, potentially uncovering strong approach, starting with the pre-processing and
candidates who might otherwise have been overlooked feature extraction from a unique dataset called the
due to unconventional career paths or non-traditional Prospect dataset, which includes around 5,000
skill sets. NLP techniques are utilized to interpret the resumes. This setup ensures an unbiased
textual content of resumes, extracting valuable classification of resumes into "selected" or "rejected"
information such as skills, work experience, education, categories based on a sophisticated matching score
and achievements. This technology allows the system algorithm and custom logic. The integration of
to understand and process the natural language found artificial intelligence and machine learning
in resumes, transforming unstructured text into techniques in this model offers a promising
structured data that can be easily analysed and direction for enhancing the efficiency and fairness
compared across candidates. of the resume screening process.

C. A Machine Learning approach for automation of

II. LITERATURE
Resume Recommendation System

The paper evaluates four machine learning models

A. Resume Screening Using Machine Learning and
for classifying resumes to suggest job roles and
NLP: A Proposed System
improvements. The models and their accuracies are
The model described processes student resumes to as follows: Random Forest at 38.99%, Multinomial
determine job role suitability and areas for resume Naive Bayes at 44.39%, Logistic Regression at
improvement. It utilizes machine learning, with 62.40%, and Linear Support Vector Machine
input being the resume and optional LinkedIn and Classifier at 78.53%. The SVM classifier
GitHub links for additional insights. The outperforms others, highlighting its effectiveness in
architecture includes SQL databases for structured resume classification and job recommendation tasks.
data storage and employs either K-Nearest
D. Automatic Resume Filtering Using Machine
Neighbour or Support Vector Machine for job role
Learning
prediction. Cosine similarity is used in a
recommendation system to suggest resume The paper details a study on the automation of
enhancements based on desired roles and existing resume screening through the application of
Machine Learning (ML) and Natural Language

Volume 10, Issue 2, March-April-2024 | http://ijsrcseit.com 603

 Dr. Sandeep Tayal et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol., March-April-2024, 10 (2) : 602-606

Processing (NLP). The literature review highlights tokenization, and cleaning. Through the application
significant contributions to the field, including of advanced NLP techniques like named entity
works by Nandhini S, Gomathi S, Lavanya S, recognition and part-of-speech tagging, the system
Kondapalli Sai Pranay, Shweta Agrawal, and Sumit adeptly extracts critical data from resumes, such as
Gupta, among others. These studies collectively skills, education, and work experiences, while
explore various methodologies for extracting and accommodating various resume formats and
ranking data from resumes using NLP techniques languages. This foundational work facilitates the
and matching them with job descriptions through transition to feature extraction, where key data
ML algorithms. The emphasis across the research is points are transformed into a numerical format
on enhancing the efficiency and accuracy of the suitable for ML model training, employing
resume screening process, which is crucial for methodologies like TF-IDF and word embeddings.
streamlining recruitment and ensuring optimal job- Subsequent stages involve the deployment of ML
candidate matches. classifiers—K-Nearest Neighbours, Support Vector
Machines, and One v/s Rest among them—to
categorize resumes effectively. This process is
III.PROPOSED SYSTEM
refined through rigorous training and testing phases,
A. Problem Statement
employing metrics such as accuracy, precision,
recall, and F1 score for evaluation. The culmination
The current manual process of resume screening is of the system's development sees the integration of
labor-intensive, time-consuming, and susceptible to the NLP-based resume parser with ML classifiers,
bias, failing to efficiently handle the volume and resulting in a comprehensive automated screening
diversity of job applications. This necessitates an system. This system is meticulously tested and
innovative approach to automate and enhance the refined with industry feedback, ensuring it not only
screening process. Leveraging Natural Language meets but exceeds the requirements of modern
Processing (NLP) and Machine Learning (ML) recruitment processes by delivering a solution that
technologies, this study proposes a system aimed at is both scalable and unbiased, significantly
improving the accuracy, efficiency, and fairness of enhancing the efficiency, accuracy, and fairness of
candidate selection, addressing the pressing need candidate selection in the recruitment landscape.
for a scalable and unbiased recruitment solution in C. System Architecture
the digital age.
B. Solution
To address the inefficiencies and limitations of
traditional manual resume screening, this research
introduces a cutting-edge automated system,
leveraging the synergy of Natural Language
Processing (NLP) and Machine Learning (ML)
technologies. The solution encompasses a
comprehensive strategy beginning with the
collection and preprocessing of a diverse dataset of
Image source:
resumes, ensuring readiness for detailed analysis
through techniques such as text normalization,

Volume 10, Issue 2, March-April-2024 | http://ijsrcseit.com 604

 Dr. Sandeep Tayal et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol., March-April-2024, 10 (2) : 602-606

https://dev.mrdbourke.com/zero-to-mastery- with the ML classifiers, giving rise to a unified

ml/introduction-to-pandas/ automated resume screening system. This system
undergoes rigorous end-to-end testing with a distinct
Figure 1. System Architecture
set of resumes to validate its accuracy and operational
At its core, the system intends to handle the vast efficiency. Feedback from recruiters and industry
diversity of resumes across different job roles and professionals plays an instrumental role in refining the
industries by initially collecting a wide-ranging dataset system, ensuring it meets the practical demands of the
of resumes. This dataset undergoes a comprehensive recruitment process.
preprocessing phase where text normalization, D. Results and Accuracy Table
tokenization, and cleaning are performed to prepare Table 1: Accuracy for Different Model
the data for subsequent analysis. Crucial to enhancing Model Used Accuracy
the dataset's utility, resumes are annotated with labels
KNN 98.44%
that mark relevant categories such as skills and
experience levels, paving the way for a more structured One v/s Rest 92.23%
and informative dataset. The utilization of NLP SVM 94.76%
techniques, including named entity recognition and
part-of-speech tagging, forms the backbone of the
The table demonstrates the KNN model's superior
resume parsing process. This enables the system to
performance, achieving an accuracy of 98.44%, making
accurately extract and organize key information such
it the most effective model among those tested for the
as skills, educational background, and work experience
purpose of resume screening.
from the resumes. Special algorithms are developed to
accommodate the variations in resume formats and
IV. CONCLUSION
languages, ensuring a versatile and robust parsing
mechanism. With the foundation laid by NLP, the
The research embarked upon the ambitious goal of
system then transitions to feature extraction, where a
automating the resume screening process, harnessing
specific set of features is identified from the parsed data.
the power of Natural Language Processing (NLP) and
This step is crucial for feeding well-structured input
Machine Learning (ML) to introduce a solution capable
data into the ML classifiers.
of transforming the recruitment landscape. Through
The next phase involves the selection and diligent exploration and implementation of various ML
implementation of ML classifiers, including but not models, including K-Nearest Neighbors (KNN), One vs
limited to K-Nearest Neighbours, Support Vector Rest, and Support Vector Machine (SVM), the study
Machines, and One v/s Rest. The system employs a demonstrated not just the feasibility but the
strategic split of the dataset into training and testing remarkable efficiency and accuracy of automating
sets, allowing for the comprehensive training and resume classification. The results underscored the
subsequent evaluation of the models. Performance superior performance of the KNN model, achieving an
metrics such as accuracy, precision, recall, and F1 score impressive accuracy of 98.44%, thereby showcasing its
serve as benchmarks for evaluating the models' potential as a leading tool for resume screening. While
effectiveness in categorizing resumes into relevant job the One vs Rest classifier and the SVM also performed
roles or skill categories. The culmination of the commendably, with accuracies of 92.23% and 94.76%
system's development is the integration phase, where respectively, the KNN model stood out for its ability to
the NLP-based resume parser is seamlessly combined navigate the complexities of resume data, offering a

Volume 10, Issue 2, March-April-2024 | http://ijsrcseit.com 605

 Dr. Sandeep Tayal et al Int. J. Sci. Res. Comput. Sci. Eng. Inf. Technol., March-April-2024, 10 (2) : 602-606

scalable and highly reliable method for candidate

classification.

IV. REFERENCES

[1]. B. Kinge, S. Mandhare, P. Chavan, and S. M.

Chaware, "Resume Screening Using Machine
Learning and NLP: A Proposed System," in
International Journal of Scientific Research in
Computer Science, Engineering and Information
Technology, vol. 253, doi: 10.32628/CSEIT228240
[2]. A. K. Sinha, M. A. K. Akhtar, M. Kumar, and S.
Upadhyay, "Resume Screening Classification
using Artificial Intelligence and Natural Language
Processing," Name of Journal, doi:
10.48047/ecb/2023.12.si4.130
[3]. P. K. Roy, S. S. Chowdhary, and R. Bhatia, "A
Machine Learning approach for automation of
Resume Recommendation system," in Procedia
Computer Science, vol. 167, pp. 2318-2327, Jan.
2020, doi: 10.1016/j.procs.2020.03.284
[4]. D. L. Padmaja, Ch. Vishnuvardhan, G. Rajeev, and
K. N. S. Kumar, "Automated Resume Screening
Using Natural Language Processing," in Journal of
Emerging Technologies and Innovative Research
(JETIR), vol. 10, no. 3, Mar. 2023

Volume 10, Issue 2, March-April-2024 | http://ijsrcseit.com 606