|
|
--- |
|
|
language: |
|
|
- en |
|
|
metrics: |
|
|
- accuracy |
|
|
- precision |
|
|
- f1 |
|
|
- recall |
|
|
pipeline_tag: token-classification |
|
|
library_name: spacy |
|
|
tags: |
|
|
- spacy |
|
|
- nlp |
|
|
- python |
|
|
- skill-extraction |
|
|
- ner |
|
|
--- |
|
|
|
|
|
# Skill Extraction Model using spaCy |
|
|
|
|
|
This is a custom **Named Entity Recognition (NER)** model built with **spaCy** to identify and extract skills from resumes and job descriptions. |
|
|
|
|
|
## Why This Model? |
|
|
|
|
|
To improve flexibility and accuracy, we transitioned from a static skill extraction approach to a dynamic one. This new method leverages spaCy to fine-tune a pre-trained Named Entity Recognition (NER) model, enabling the extraction of skills directly from resumes and job descriptions. By removing the dependency on predefined skill lists, the model can recognize context-specific, domain-relevant, and even newly emerging skills. This dynamic strategy offers a more adaptive and scalable solution for real-world skill extraction and talent-matching applications. |
|
|
|
|
|
--- |
|
|
|
|
|
## How to Use |
|
|
|
|
|
### 1. Load the Model from Hugging Face |
|
|
|
|
|
```python |
|
|
from huggingface_hub import snapshot_download |
|
|
import spacy |
|
|
|
|
|
# Download the model from the Hub |
|
|
model_path = snapshot_download("amjad-awad/skill-extractor", repo_type="model") |
|
|
|
|
|
# Load the model with spaCy |
|
|
nlp = spacy.load(model_path) |
|
|
|
|
|
# Example usage |
|
|
text = "Experienced in Python, JavaScript, and cloud services like AWS and Azure." |
|
|
doc = nlp(text) |
|
|
|
|
|
# Extract skill entities |
|
|
skills = [ent.text for ent in doc.ents if "SKILLS" in ent.label_] |
|
|
print(skills) |
|
|
``` |
|
|
```output |
|
|
['Python', 'JavaScript', 'cloud', 'AWS', 'Azure'] |
|
|
``` |
