Counterparty Model
Input:
Transaction details such as amount, description, and other metadata.
Processing:
Advanced text analysis techniques are used to determine the counterparty of the transaction.
Managed data assets:
The model leverages curated directories of known merchants and institutions to assist with counterparty identification.
Output:
The counterparty of the transaction is identified, providing valuable information for insights and other product uses.
Overview
Cleansing Phase
Performs text manipulation and annotations on the description fields to create a cleaner baseline for the counterparty extraction phase.
Extraction Phase
Basically the counterparty recognition: using advanced logics to extract the counterparty from the clean description.
Matching Phase
Using our managed counterparties assets, considering all the information we’ve got so far on the transaction, which includes the data from the DI part and the counterparty from the extraction part.
Examples
Raw description | Counterparty |
---|---|
NETS PURCHASE AT SHENG SIONG SUPERMARKET | Sheng Siong |
PAYPAL PAYMENT TO SPOTIFY SINGAPORE PTE LTD | Spotify |
COURTS (S) PTE LTD PURCHASE | Courts |
NETS E-PAYMENT AT SINGAPORE POLYTECHNIC | Singapore Polytechnic |
FAST TRANSFER TO DBS BANK LTD | DBS |
Updated about 1 year ago