Articles & Papers

Hybridizing Fuzzy String Matching and Machine Learning for Improved Ontology Alignment

[DeezyMatch: A Flexible Deep Learning Approach to Fuzzy String Matching](https://funky-crawdad-49c.notion.site/DeezyMatch-A-Flexible-Deep-Learning-Approach-to-Fuzzy-String-Matching-1ca03e448153807b8e4bdfb6c6affbe4)

Training:

String Matching V1

String matching V2

Description

This project is designed to compare two JSON objects with different structures and field names, identify corresponding fields, and compute a similarity score. It leverages intelegent text matching and similarity metrics to provide accurate matching between fields.

Input

Two JSON objects with potentially different structures and field names.
Fields may contain text, numbers, or nested structures.

Output

A JSON object containing:
- Overall match score (percentage-based).
- Matched fields with:
  - Corresponding field names from both JSONs.
  - Similarity scores for each field.

Sample

{
  "full_name": "John Doe",
  "dob": "1993-05-15",
  "contact": {
    "email": "[email protected]",
    "phone": "+123456789"
  },
  "location": "New York, USA"
}

{
  "name": "Jonathon Doe",
  "birth_date": "15-05-1993",
  "email_id": "[email protected]",
  "address": "NY, United States",
  "mobile": "123456789"
}

{
  "match_score": 82,
  "matched_fields": {
    "full_name": {
      "matched_with": "name",
      "similarity": 90
    },
    "dob": {
      "matched_with": "birth_date",
      "similarity": 95
    },
    "contact.email": {
      "matched_with": "email_id",
      "similarity": 75
    },
    "contact.phone": {
      "matched_with": "mobile",
      "similarity": 100
    },
    "location": {
      "matched_with": "address",
      "similarity": 80
    }
  }
}