Resume Based Scanner

Resume Based Scanner

Divya Bai B

Department of Computer Application St Joseph Engineering College Mangalore, India

Sumangala N

Department of Computer Application St Joseph Engineering College Mangalore, India

Abstract – In a rapid development environment across digital, automation and intelligence change traditional employment processes. This study represents an AI-based Smart hire and is based on role-play jobs designed to optimize staff using Curriculum vitae intellectual screening and personalized work recommendations. The system involves three key stakeholders. Administrators, companies, users, and each plays a variety of roles. Administrators control platform integrity, including business claims, user access, work categories, payments, and comments. Companies can register, create profiles, publish vacant seats, and manage applications. Users can register, enter profiles, download resumes, receive recommendations according to artificial intelligence, explore tasks, and follow application conditions. Essentially, Smart hire uses an automated learning model (ML), trained names, skills, and descriptions. After requesting a CV (NLP), similarity algorithms are used to extract key features such as skills, experience, titles and other corresponding features for the work publication, providing appropriate recommendations. Built with Backend and the ML/NLP user engine, Smart hire improves efficiency.

It reduces manual filtration and provides a more intelligent experience in recruiting staff. It shows the true impact of artificial intelligence on the creation of adaptive and scalable solutions for comparing jobs

Index TermsMachine Learning, Natural Language Processing, Resume Screening, Job Recommendation, Role-based Access, Flask, TF-IDF, Cosine Similarity

1. INTRODUCTION

   The recruitment industry is undergoing a shift in paradigm due to its achievements in the field of intellectual technology. Traditional employment processes that rely heavily on manual screening and subjective decision-making are often ineffective and are subject to human bias. These normal systems are difficult to deal with large amounts of applications, often leading to suboptimal agreements between candidates and practical roles, leading to an increase in staff fluidity and over- all dimensions. To solve these problems, modern employment systems incorporate artificial intelligence (AI) and natural language therapy (NLP) to automate and optimize candidates and work comparison processes [1]. In this context, we represent Smart hire based on AI based on role-play work that uses Auto-Learning (ML) and NLP to improve the accuracy and efficiency of all staff. Smart hire is developed with three levels of user model (looking for work) including administrators, companies and users. Administrators support platform security and control critical components such as user verification, work categories, comments, and payments. Companies can register,

   publish jobs, and manage applications. Users are permitted to load resumes, receive intellectual offers for work, and apply for the appropriate role. This platform uses the recommended multiliphas mechanism that begins with an analysis of CVs that extract text using PymUPDF, and then extracts semantic properties using the Nature of Space (NER) and Name Recognition Method [2]. The extracted data is converted into vector representations using the method of Document Frequency (TF-IDF)[3], allowing comparison of CVs and instructions. Cosines similarity is used to classify work on relevance and recommend the options that are most suitable for the user [4]. The system is implemented using BACAND based on balloons, reaction front-ends, and MongoDB to manage the database. AI BACAND includes a RESTFUL API layer for communication between customers and servers. Smart hire is trying to optimize selection conveyors by automating boring processes, increasing the relevance of candidate work, and reducing recruiter workloads through intellectual filtration. By applying real-time similarities to NLP-based algorithms, the platform demonstrates the possibility that AI can translate adoption into faster, fairer, and more efficient processes [5].

2. LITERATURE REVIEW

   The use of artificial intelligence (AI) has been very focused over the past decade due to its ability to increase the accuracy of decisions and the complex procedures automata. Many studies have focused on analysis of automated learning (ML) and natural language therapy (PNL) use, classification of candidates, and the generation of job recommendations. The authors [1] proposed a mechanism for recommendations that proposed relevant roles by studying candidate preferences and work history. Despite its efficiency in a static context, this model was not flexible enough to process unstructured data that is common in curriculum vitae. Similarly, [2] presented a semantic similarity model. This used language comparisons based on natural language treatments corresponding to CVs with instructions. However, the systems vocabulary was field- specific and limited ability to generalize across different sectors and workplaces. Additional studies on [3] introduced a classification method using vector support machine (SVM) and naive classifiers for candidate attribute-based classification CVS. This improved the accuracy of classification, but he demanded enormous engineering of functions and fought against the semantically complex curriculum of analytical analysis.

   Another approach to [4] used a detailed learning architecture, in particular the classification of dual short-term memory (LSTM) and profile profiles of extraction entities. Nevertheless, this occurred due to increased computational complexity and reduced output speed. From a system architecture perspective, certain existing platforms implement role (RBAC) control to isolate management, enterprise, and user-level functions [5]. Although this modular architecture provides secure access and segmentation of functionality, many of these systems still lack intellectually recommended mechanisms. Furthermore, most regular portals do not offer individual work suggestions based on real-time dynamic content extracted from CVs, limiting their usefulness to candidates.

3. PROPOSED SYSTEM

   1. System Overview

      Smart Hire is a technology platform that simplifies and optimizes the recruitment process by using artificial intelligence (AI) to effectively compare candidates with affected employers. This is accomplished through integration of natural language therapy (NLP), automated learning (ML), and access control base (RBAC). The application focuses on three different user roles: manager, employer and candidate, due to the supply of individual features. Architecturally, the system is developed using a modular approach using a service structure, using the FORT.JS interface[7]. Each user role interacts with the system via a special panel. Applicants can register, create their own profile, download curriculum vitae, explore recommended jobs, and submit requests. Employers may manage business profiles, create employment reports, and observe incoming requests. Meanwhile, administrators are responsible for monitoring platform actions, including user and employer permissions, employment classification, managing payment transactions, and maintaining system-wide integrity. Smart Rental is based on an engine of recommendations using AI to perform automated CV screening and task comparisons. When users provide the curriculum vitae in PDF format, the system uses PymUPDF to extract the content of the text. This raw text is then analyzed using NLP tools to recognize personalized objects (NERs) and determine appropriate details such as names, skills, and profesional experience. Processed information is converted into a digital form using the TF-IDF method, which converts unstructured data into high green. These vectors are compared to vectors obtained from the work description using the Cosinus algorithm. This allows the platform to classify and recommend the most important job possibilities. This intellectual work process significantly reduces manual interventions, increases comparison accuracy, and leads to a more effective employment process. The platform is de- signed to support future updates, including behavioral analysis, direct relationships between candidates and recruiters, and integration with external work councils. Intelligent hiring, a combination of automation and adaptability, demonstrates how

      II can provide intelligence and personalization to modern

      recruitment experiences. The IEEEtran class file is used to format your paper and style the text. All margins, column

      widths, line spaces, and text fonts are prescribed; please do not alter them. You may note peculiarities. For example, the head margin measures proportionately more than is customary. This measurement and others are deliberate, using specifications that anticipate your paper as one part of the entire proceedings, and not as an independent document. Please do not revise any of the current designations.

   2. Architecture Diagram

   Fig. 1. Architecture Diagram for smart hire

4. METHODOLOGY

   The proposed system, Smart hire, follows a multi-stage pipeline for the treatment of curriculum vitae, extracts relevant information and publication recommendations based on intellectual comparisons. This methodology combines natural language classification (NLP), automated learning algorithms, and provides accurate and personalized work recommendations based on similarity.

   • Resume Text Extraction : This process begins when the user loads the curriculum vitae in PDF format. The system uses PymupDF (also known as Fitz) to extract text content from loaded documents. This allows the system to process CVs regardless of design or format. Sending the extracted text to the next step to extract the indicator.

   • Job Enrichment and Recommendation Output: Finally, the recommended jobs are enriched with additional metadata such as job category names fetched from the database. The top- matched jobs, along with relevant de- tails, are returned to the user through a REST API end- point, completing the intelligent recommendation cycle.

   • Feature Extracting Using NLP: As soon as unconverted text is extracted, make the NLP library open source to analyze the text using spaces. Object (NER) recognition is used to extract personal information such as the candidates name (direct mark). Additionally, names are used to identify and extract important skills and sentences that reflect the candidates experience and experience.

   This semantic analysis guarantees that clear and implicit possibilities are captured.

   • Similarity Matching with Cosine Similarity: The vectorized resume is then compared against the vectors of active job postings using Cosine Similarity. This algorithm calculates the angle between the resume vector and each job vector, with a smaller angle indicating a higher degree of similarity. Based on the computed similarity scores, the system ranks the jobs and selects the top N most relevant job postings to recommend to the user.

5. RESULTS

   The Smart Hire system demonstrates its ability to optimize the hiring process through the use of automated learning and natural language therapy techniques. The platform successfully removed relevant information from the busy curriculum vitae and provided customized job recommendations based on skill sets and user experience. Employers were able to effectively publish and manage the open list while candidates received appropriate job recommendations and were able to apply gently through

   Cosine Similarity(A, B)= cos()= A · B

   Where:

   A · B is the dot product of the vectors,

   the portal. The smooth functioning of this role-based

   1. structure has benefited all DMIN users, businesses, and users. The results show that Smart Hire enhances the

      overall work experience by automating critical tasks, reducing manual interventions, and improving accidental

      • A is the Euclidean norm (magnitude) of vector A,

      • B is the Euclidean norm (magnitude) of vector B,

      • is the angle between the two vectors.

   • Resume Representation with TF-IDF: To compare the resume with job descriptions, both are converted into numerical vectors using the TF-IDF (Term FrequencyInverse Document Frequency) technique. TF- IDF helps highlight the most important terms in a document by reducing the weight of commonly used words and increasing the significance of rare, job-specific keywords. This step transforms textual data into a form suitable for similarity computation.

   TF-IDF(t, d) = TF(t, d) × IDF(t) (2)

   1. Term Frequency (TF)

      Term Frequency measures how often a term appears in a document. It is normalized by the total number of terms to avoid bias toward longer documents. Mathematically:

      freq(t, d)

      relevance.

      • Reduced manual resume screening time through AI- powered extraction and filtering.

      • personalized job recommendations tailored to user profiles and skills.

      • .Enabled smooth interaction between different user roles with secure access control

6. DISCUSSION

   The development and implementation of Smart Hire reveal the practical value of integrating machine learning (ML) and natural language processing (NLP) into the recruitment domain. By automating key functions such as resume parsing, skill extraction, and job matching, the system addresses several inefficiencies found in traditional hiring workflows. The use of TF-IDF vectorization and cosine similarity enabled the system to measure the semantic relevance between candidate profiles and job postings. This approach proved to be effective in delivering job recommendations that closely aligned with user- provided resumes. Additionally, the spaCy NLP model

   where:

   TF(t, d) = n

   i=1

   freq

   i

   (t , d)

   1. allowed for meaningful extraction of skills and personal

      information, which was essential for accurate job-role matching. Smart Hires role-based designfeaturing Ad-

      • freq(t, d) is the count of term t in document d,

      • n is the number of distinct terms in d.

   1. Inverse Document Frequency (IDF)

   IDF evaluates the importance of a term across the corpus. Common terms have lower IDF, while rare terms have higher values. The IDF is calculated as:

   IDF(t) = log N (4)

   count(t)

   where:

   • N is the total number of documents in the corpus,

   • count(t) is the number of documents in which term

   t appears.

   min, Company, and User modulesensured that each stakeholder had dedicated tools to interact with the plat- form. Admins could manage platform activity, companies could efficiently handle job postings and applications, and users could benefit from an intelligent, personalized experience. Moreover, the use of a Flask backend and MongoDB ensured a scalable and lightweight architecture, making the system adaptable for real-world deployment. However, further enhancement can be made by integrating deeper models like BERT for improved semantic understanding, and by refining skill-tagging to distinguish between soft and hard skills more accurately. In summary, Smart Hir demonstrates a viable model for next-generation recruitment systems, showing how AI can drive better candidate-job matching and enhance overall user satisfaction in digital hiring environments.

7. CONCLUSION

This research document is presented by Smart hire on AI based on work that has the role of improving the employment process by integrating automated learning and natural language therapy. The system effectively automates CV screening, skill extraction and personalized work recommendations, reduced manual effort, and increased recruitment efficiency. Smart hire uses TF- IDF vectorization, similarity between Cosines and NLP spaces to accurately correspond to user profiles using the corresponding task list. The administrators architecture constitutes the platform, and the company and user modules determine an optimized, secure experience for all stakeholders. Smart hire is built with the product and MongoDB database, simultaneously scalable and suitable for real-world applications. This system highlights the possibilities of artificial intelligence in the full modernization of staff and serves as the basis for new outcomes of intellectual systems for comparing jobs.

VIII REFERENCES

1. Nikethani Gangoda, Kavindu Piumal Yasantha, Chamina Sewwandi, Navindu Induvara, Samantha Thelijjagoda, Nishantha Giguruwa, Resume Ranker: AI-Based Skill Analysis and Skill Matching System, 2024 Sixth International Conference on Intelligent Computing in Data Sciences (ICDS), pp.1-8, 2024.

2. J. Liu, X. Li, and X. Tang, Personalized Job Recommendation System Using NLP and Machine Learning, 2021 IEEE 4th Inter- national Conference on Information Systems and Computer Aided Education (ICISCAE), Dalian, China, 2021, pp. 10221026.

3. Y. Zhang, G. Lai, M. Zhang, Y. Zhang, and S. Ma, Explicit factor models for explainable recommendation based on phrase-level sentiment analysis, Proceedings of the 37th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2014, pp. 83 92.

4. Minh-Luan Tran ; Anh-Tuye Nguyen ; Quoc-Dung Nguyen ; Tin Huynh, A comparison study for job recommendation, In 2017 International Conference on Information and Communications (ICIC), pp. 199-204, 2017.

5. S. Bhor, V. Gupta, V. Nair, H. Shinde and M. S. Kulkarni, Resume parser using natural language processing techniques, Int J Res Eng Sci (IJRES), vol. 9, no. 6, pp. 01-06, 2021.

6. A. Mankawade, V. Pungliya, R. Bhonsle, S. Pate, A. Puro- hit and

   A. Raut, Resume Analysis and Job Recommenda- tion, 2023 IEEE 8th International Conference for Convergence in Technology (I2CT), Lonavla, India, 2023, pp. 1-5, doi: 10.1109/I2CT57861.2023.10126171. keywords: Machine learning algorithms; Resumes; Companies; Machine learning; Predictive models; Prediction algorithms; Natural language processing; Re- sume; Cosine similarity; Na¨ve Bayes; Web scraping;

   Job Rec- ommendation

7. A. Aggarwal, S. Jain, S. Jha, and V. P. Singh, Resume Screen- ing, International Journal for Research in Applied Science and Engineering Technology 134 (May 2022): 6688.

8. K. Tejaswini, V. Umadevi, S. M. Kadiwal, and S. Revanna, Design and Development of Machine Learning Based Resume Ranking System, Global Transitions Proceedings 3 (October 2021): 371.

9. S. Lokesh, S. Mano Balaje, E. Prathish, and B. Bharathi, Resume Screening and Recommendation System Using Machine Learning Approaches, An International Journal (CSEIJ) 12, no. 1 (Febru- ary 2022): 17.

10. J. A. A. Zubeda, M. A. A. Shaheen, G. R. N. Godavari, and S.

Z. A. M. S. Naseem, Resume Ranking Using NLP and Machine Learning, Computer Science and Engineering: An International Journal (CSEIJ) 12, no. 1 (February 2022): 16.