Semantic AI Transforms User Search for Biopharma Organization

To improve the identification and awareness of standards, procedures, and responsibilities associated with a role, the organization uses a hub and spoke architecture where centralized knowledge models manage multiple business functions and extract the critical facts from documents to support R&D quality objectives.

Within the enterprise, the management of systems and business processes are highly regulated and subject to precise and specific guidelines. As new employees are hired and individuals assume new roles, the process of identifying the required knowledge, skills, and training requirements from a corpus of more than 500K documents, is long and complicated.

Requirements, which are managed in a repository, contain long standard operating procedure (SOP) documents that support new hires and new roles required skills/training. The systems contain sensitive information and the process to identify the appropriate training requirements requires a manual stare and compare process of locating each document, opening it, and reading it to determine if it is relevant.

Already deployed within the enterprise, Semaphore was used for central ontology management and the classification of public documents for current awareness. By extending the use of Semaphore and creating a unique and innovative classification strategy, the necessary metadata was created and the user experience transformed.

Using Semaphore’s Knowledge Model Management (KMM) capabilities they extended existing R & D models to include the relevant concepts, topics and relationships associated with role standards, procedures, and responsibilities

The model is published and used by Semaphore Classification and Language Services (CLS) to automatically examine each document, identify and extract the facts necessary to populate key fields such as roles, responsibilities, scope and purpose.

The classification process uses sophisticated table handling techniques that push classification to the next level - to identify important parts of each document, model them, and use rule-based classification to apply precise and consistent metadata. The metadata is used to create filters that employees use to quickly identify the skills and training that are relevant to their role.

Semaphore FACTS modeling and CLS refine extraction rules to run against the full set of documents, identify additional relevant facts, and develop appropriate fact extraction strategies. Semaphore worked closely with the client to transfer knowledge that supports existing and future classification and fact extraction outcomes.