Document Exploitation
VALUE PROPOSITION
- Enabling a major Digital Transformation leap, by utilising ELEMENT’s revolutionary Knowledge Graph to enrich a detailed network of entity attributes and relationships
- Improving adaptability to evolving business needs and regulatory changes, by generating dashboards and reports in near real-time
- Lowering Total Cost of Ownership by shifting to OPEX via cloud infrastructure and PaaS, and decreasing labour-intensive tasks by providing a process-centric, AI-embedded application
Enablement of tier 1 international bank to more efficiently extract entity information from internal and external documents, by utilising text analytics
BACKGROUND
- The bank’s systems were unable to automatically extract and disambiguate entities and information about them from internal and external documents, resulting in heavy backlogs of pending documents and a high total cost of ownership
- Current systems and evaluated competitor offerings were found inadequate in accurately extracting information, thereby requiring extensive manual data enhancement, and these systems could not handle project scope and scalability requirements
- The bank was unable to achieve its digital transformation goals within time and budget constraints, due to high development costs and long deployment steps
TARGETS
- Automate the extraction and disambiguation of entities from documents, to improve processing efficiency, prevent heavy backlogs of pending documents, and reduce total cost of ownership
- Create a unified knowledge base, seamlessly correlating information about entities and related documents, and making the data available for multiple bank processes (e.g., debt issuance, compliance, due diligence)
- Implement a solution with low development costs and short deployment time, to achieve digital transformation goals within schedule and budget constraints
ADDRESSING THE NEED
- Assembled a comprehensive Knowledge Graph integrating the bank’s core infrastructure — including document repositories and related systems, to automatically enrich knowledge about entities (e.g., clients, creditors, debtors, suppliers), along with their attributes
- Configured a document parsing engine to identify various structured document types (e.g., salary statements, foreclosure notices, tax statements, vendor contracts, HR documents, margin calls), and to extract all entities involved as well as their attributes, using Natural Language Processing
- Trained adaptive AI algorithms to identify multiple types of unstructured documents and extract the metadata (type, language, date of issue, etc.), while analysing the context to disambiguate the entities involved as well as their attributes
Other Use Cases
