Data extraction from Unstructured documents

Tasks

• MetaBot sends unstructured report to Zanran server
• Zanran server recognises tables, text and images on each page
• MetaBot retrieves outputs and passes structured data on to your Bot for further processing.

PROBLEM: Unstructured files like PDFs do not allow you to extract text and tables easily. If you have a large volume of reports it's impossible to automate your processing without some smart software.

SOLUTION: Zanran's Scaffolder uses computer-vision techniques to identify tables, text, and graphics on each page. It outputs tables and text in structured form so your Automation Anywhere Bots can do further processing. The Scaffolder saves time compared with manual extraction, and improves accuracy.

INPUTS: Reports that are a mix of text, tables and images - especially financial reports.

ACTIONS: the MetaBot takes PDFs and returns Excel, XML etc for further processing. The MetaBot makes it very easy and fast to test your overall solution.

OUTPUTS: Excel, XML or JSON. The Excel is for tables only; the XML and JSON include text, tables and images

USE CASE 1 - Automating part of the Credit-Checking Process. Zanran's Scaffolder software automatically extracts tables from the financial statements of the bank's corporate clients. Automation Anywhere Bots find the relevant rows in the balance sheet and other financial tables and extract data.

USE CASE 2 - Automating the Performance Analysis of Funds. The Scaffolder automatically extracts tables from the quarterly statements of the fund's assets. Automation Anywhere Bots find the relevant rows in the tables, then extracts data. Data is stored in a structured database, and interrogated for performance analysis etc.

MORE DETAILS:
This MetaBot works with a Zanran server on the cloud. It also works with on-premise servers and servers in private clouds.
Zanran provides an XML viewer for manual tagging, quality assurance, and table comparison.