Machine Learning Curriculum Classifier API

Curriculum Classifier -Natural Language Process- API designed to process resumes in PDF format, extract text, perform seniority classification, classify the curriculum job area with machine learning methods and provide a detailed analysis of the resume content.   
Built with the FastAPI framework, it incorporates various services for PDF manipulation and analysis.

         

Table of contents

• Featrures
• Getting Started
• API
• Machine Learning Graphics
• Contribute

Features

• Curriculum PDF Process

  • Pdf Text Extraction
    • library: PdfToText
  • PdF Text Pre-process
    • library: Spacy
    • pipeline:
      • lemmatizer
      • stopWord extraction
      • punct extraction
      • allows only alpha
      • to lower case
  • Background Processing of PDF Uploads
    • When a PDF is uploaded to the API, it undergoes a series of background processing steps to ensure efficient handling and analysis.
    • After uploading, the API processes the PDF text in the background, so when the text is requested, it is already ready.
  • Dataeset
    • https://www.kaggle.com/datasets/danicardeal/resume-occupation-and-seniority

• Machine Learning Curriculum Classifier

  • Vectorization
    • Count vectorizer
  • Machine Learning
    • Random Forest Classifier
    • Accuracy
      • Job model = 81%
      • Seniority model = 72%

Getting Started

• Envirement preparation

  1. Prepare Libraries:

     pip install -r ./configure/requirements.txt
     

  2. Git lfs:

     git lfs install
     
     git lfs pull     
     

  3. Run:

     uvicorn application:app --reload
     

  4. API references

• Use Instruction

  1. Open http://127.0.0.1:8000

  2. Upload Curriculum PDF

     • EndPoint 'POST /pdf'

  3. Requests

     • Processed Text

       • EndPoint 'GET /curriculium'
       • Retuns:
         • Processed pdf text

     • Curriculum Classification

       • EndPoint 'GET / curriculum/classification'
       • Returns:

         • Job Area

         • Probability

         • Seniority Classification

         • Probability

         • Response

            {
             "classification":
             "classification probabilities"
             "seniority":
             "seniority probabilities": 
             }
           

     • Model Classification Areas

         BUSINESS-DEVELOPMENT     
         INFORMATION-TECHNOLOGY   
         FINANCE                  
         CHEF                     
         ACCOUNTANT               
         ADVOCATE                 
         ENGINEERING              
         AVIATION                 
         FITNESS                  
         SALES                    
         HEALTHCARE               
         CONSULTANT               
         BANKING                  
         CONSTRUCTION             
         PUBLIC-RELATIONS         
         HR                       
         DESIGNER                 
         ARTS                     
         TEACHER                  
         APPAREL                  
         DIGITAL-MEDIA            
         AGRICULTURE              
         AUTOMOBILE               
         BPO
       

  note:

    Please be aware that the classification models integrated into our API are designed to classify resumes written exclusively in English.
        Resumes in other languages will not be accurately processed or classified by the current models.
  

API

• Layout

  

• EndPoints

  • '/pdf'

      GET
      GET /{filename} - get specific pdf by id
      POST
      DELETE
    

  • '/curriculum'

      GET
      GET /classification - Get the curriculum classification
    

• Flowchart

  • 

Machine Learning Graphics

• Job

  1. Job Classifier classification report
  • 
  2. Job Classifier Confusion Matrix
  • 

• Seniority

  1. Seniority Classifier classification report
  • 
  2. Seniority Classifier Confusion Matrix
  • 

Contribute

• GitHub
• Linkedin

  Contributors

  • Luan Silveira

    • Linkedin
    • GitHub

  • Caio Coutinho

    • Linkedin
    • GitHub

(Back to top)