With the rise of KYC regulations (AML, terrorism fundings, …) it is now necessary for financial institutions to screen their clients on the web in order to assess their public probity. The process is for now manual and requires people to go through search engines in order to find the client. Starting from this point, a great added value can be found in automatizing the screening process in order to reallocate times on the analysis and decision making. Thus, Sia Partners has developed a bot whose purpose is to extract news article from the web and detect KYC related event concerning corporate clients.
The bot use web scraping and machine learning in order to automate the fastidious screening time and implement an automatic categorization of the content found on the web. Web scraping (or web crawling) is the process of navigating through the web by the means of programs in order to visit pages and collect data automatically.
The bot is thus able to visit news website and collect articles, that are then stored for analysis and logging. Once the articles are collected, machine learning is used for two processes:
- Named Entity Recognition: identify names' entities in the articles in order to relate it to a client;
- Classification of the news depending on the KYC event they concerned (Internal Fraud, Tax Evasion, Justice Investigation, …).