Case Study: Text Analytics for Information Extraction (Financial Services)

… for a financial services company

Business Challenge

• Reduce manual efforts and turn around time in analyzing:
  • More than 20,000 sources of text information (Regulatory Filings, Company Websites, e-mails, 3rd party sources, etc..)
  • More than 3.5 Million documents/yr to process
  • 70+ events to cover
  • Multilingual content
  • Diversity & noise in data sets

Analytics Approach

• Natural Language Processing (NLP) and Machine Learning algorithms developed to scan text sources to classify events and extract key information

Impact

• Our solution saved ~5000 Person hours of effort per year for the client
  • Event Classification:
    • 94.3% Precision (True positives)
    • 93.5% Recall (ratio of the number of relevant records to the total number of relevant records)
  • Key Information extraction:
    • 90-98% Precision
    • 93.5% Recall

Results of the document-load handled and classified by the text analytics software