ComputableFacts: a Secure System to Store Documents and Graphs

Back to Schedule

Abstract

This 20 minutes talk describes an automated data processing system, ComputableFacts, whose goal is to recover information from unstructured data in a variety of formats (such as Microsoft Office or Adobe PDF documents, emails, web pages, etc.) and convert it into a more usable form. Its key features are: Security:

• Enforce authorizations across multiple access models to the database: batch, interactive and real-time.

Data Engineering:

• Extract data and metadata from a variety of sources and file formats
• Provides a uniform representation of all data, regardless of its initial structure or format

Knowledge Engineering:

• Build facts databases manually and/or automatically
• Automatically derive new facts using rules
• Execute complex queries

Knowledge Dissemination:

• Allow users to create alerts
• Allow users to share and comment on documents
• Allow users to create and export query-focused datasets
• Allow users to rate documents. Later, recommend them documents of interest

Speakers

Cyrille Savelief

CEO, MNCC

Cyrille SAVELIEF is 34 years old and has worked 5 years as a Data Engineer for FactSet Research Systems Inc where he specialized in assisted and automated facts extractions from analysts research reports and financial statements. In 2011 he co-founded MNCC, a company providing services in Data Engineering & Analytics. During the past 3 years he specialized in applying NLP in the context of exploratory analysis of heterogeneous documents collections.