Neural Name Match Business
Introduction
Name Match is a difficult problem because there are a huge number of variations of the same name due to different reasons like OCR or human errors, alternate spellings, different order of business names, skipping words, initials, business structures, short forms of business names etc.
To make this problem simple, this LLM Powered API from signzy not only gives a comprehensive match result and match score but also gives a match reason in response.
This API holds the ability to work with Business (Organization) name match verification.
How to call the API
You will need to login before sending search request. You are required to pass the access token received from the login call, as authorization header in the Name match request.
Sample Curl
curl --location 'https://api-preproduction.signzy.app/api/v3/nameMatchBusiness/humanlike' \
--header 'Content-Type: application/json' \
--header 'Authorization: <Auth_Token>' \
--data '{
"nameBlockv3": {
"name1": "{name1}",
"name2": "{name2}"
}
}
'Response Parameter
PARAMETER NAME | DESCRIPTION |
|---|---|
name1_vs_name2_matchResult |
|
name1_vs_name2_matchScore | Name Match score between 0 to 1 |
name1_vs_name2_matchReason | Message output for explaining the reason for the score as a means of text based feedback on what happens inside the code |
Details on Output parameters
matchResult
Gives a single class denoting how well the names match. These classes are a grouped representation of a range of scores as categorized below.
Table 1: Score Categorization
MATCHRESULT (V1) | MATCHRESULT (V2) | MATCHRESULT (V3) | MATCHSCORE RANGE |
|---|---|---|---|
Direct Match | Direct Match | Direct Match | 1 |
Partial Match | Good Partial Match | High Partial Match | 0.9 |
| Moderate Partial Match | Good Partial Match | 0.8 |
| Poor Partial Match | Moderate Partial Match | 0.6 |
| No Match | Poor Partial Match | 0 - 0.3 |
No Match | No Match | Poor Partial Match | 0 - 0.3 |
name1_vs_name2_matchScore
Gives the raw matching score as a number between 0 to 1. A score of 0 means totally different names and a score of 1 means exactly the same names.
name1_vs_name2_matchReason
Gives a transparent reason why the Name Match model thought the input names are similar or different. This is directly inferred from the modular structure of our algorithm.
Getting help
Please feel free to contact us if you have any questions, require clarification, or have ideas for how to make the documents or any of our services better.
You can reach out to us at [email protected].