Skip to content

Latest commit

 

History

History
37 lines (24 loc) · 1.25 KB

README.md

File metadata and controls

37 lines (24 loc) · 1.25 KB

🗺️ SIC Mapper

The high-level goal of the SicMapper is to map a given job advert to Standardised Industrial Classification (SIC) code.

Installation

  1. Create a Python 3.10 conda environment (we strongly recommend conda over e.g. venv due to usage of the faiss package)
  2. Install the requirements file conda install --yes --file requirements.txt

🔨 Core functionality

To map job adverts to SIC codes, you can use the SicMapper class in sic_mapper.py:

from sic_mapper.sic_mapper import SicMapper

job_ad = {'id': 1, 'job_text': 'We are looking for a software engineer to join our team. This company sits in the software engineering industry.'}
sm = SicMapper()

sm.load() # load relevant models, tokenizers and datasets

sic_code = sm.get_sic_codes(job_ad) # get SIC codes for job advert

>> {1: {'company_description': 'This company sits in the software engineering industry..',
  'sic_code': '62012',
  'sic_name': 'Business and domestic software development',
  'sic_method': 'closest distance',
  'sic_confidence': 0.62}}

🖊️ Methodology

The SIC Mapper can be described in the following diagram: