Finance solutions for your business. Make it simple, make it yours. Hand to hand with digital integrations.
Coinscrap’s categorisation engine is able to detect and analyse:
17 categories and 120 sub-categories. The engine can detect custom categories.
Automated ratios about user’s financial standing and creditworthiness.
Auto-detection of user’s insurance products: Type of insurance, amount, due date, frequency.
Auto-detection of non-financial assets that are relevant for financial planning and financial score.
A research article about the results of the categorization of Coinscrap data was published in the Computer Science Q1 research journal IEEE ACCESS. This journal has an impact factor of 4.098 and an article influence score of 0.835 (per 2018 JCR). Oscar Barba, co-founder and CTO of Coinscrap was one of the co-author of the paper.
The system combines Natural Language Processing techniques with Machine Learning algorithms to classify banking transaction descriptions for personal finance management, financial scoring and lead discovering.
APIs run on cloud therefore the solution is highly scalable.
Parallel analysis and stateless requests allow the engine to respond from any node and increase the scalability through a load balancer.
Non-relational databases allow the engine to speed up response and provide high volume of data processing.
Replica Set and Sharding enables increased availability and provide a fully scalable solution.
The engine works in three languages: Spanish, English & Portuguese.
The system could be trained to almost every language and the training process takes around to 4 weeks.
Scalable architecture is able to be installed within the client’s own infrastructure. Prepared for containers deployment in a orchestrator like kubernetes (Openshift). The engine meets the highest security standards.
The system is ready to be consumed via API on demand or setting a defined batch process through API interface or files interchanging.
Coinscrap’s categorisation engine is able to detect and analyse:
The engine is periodically trained to improve the efficiency.
The accuracy ratio is a comparison between machine and human results.
17 standard categories and more than 120 specific subcategories are provided.
In each stage the system provides more data than the conventional categorization.
GTI-Atlanttic is an investigation group from Vigo University specialized in natural language processing and semantic analysis. They have won several prizes because of their research:
Winners of the XI Edition of Vodafone Connecting for Good innovation in telecommunications.
Winners of the II Edition of Ayudas a Proyectos de Investigación de Tecnologías Accesibles Indra + Universia
