There are three fundamental shapes of data in the enterprise: structured, unstructured and semantic. Epinomy is designed to leverage all three kinds of data to work together. Most businesses have relational data in structured databases, e-mail and business documents in unstructured content repositories. The idea of semantic relations is quite new in most organizations, but it is a powerful new way to think of linked data.
Structured Data (Tables)
For all the attention it is given, a relatively small amount of enterprise data is structured. This structured data is often the most important data in the company - and contains the operational metrics of the organization.
Structured data is usually stored in relational databases, is organized in rows and columns, and is used for business analytics. Sales, logistics, statistics, inventory, manufacturing data and many other business critical financial and operational data is stored in structured data.
Structured data answers questions like "how many units did we sell to whom for how much and when?", "what was our operating margin for Q3 of last year?" and "what region was responsible for the highest sales over the past 4 years?"
The preferred way to analyze this data is to use online analytical processing (OLAP) and business intelligence (BI) tools to look for trends and store actual metrics of the operating system of the business.
Unstructured Data (Text)
The majority of enterprise data, often 80% or more, is unstructured or semi-structured. Items like word processing documents, narratives, Tweets, blog postings, memos and operating manuals make up the unstructured information in an enterprise.
Unstructured data is usually stored in files in a file system or document management system, in e-mails or in other content repositories. NoSQL databases are the modern up-and-coming way to handle unstructured databases by shoving content into large blobs of text loosely coupled to a metadata structure.
The traditional tools for working with unstructured data include search engines, document management systems, content management systems and online repositories like e-mail.
Unstructured data answers questions like "Where is the marketing proposal for Spacely Sprockets?", "Where are the instructions for installing the Illudium Q36 explosive space modulator?" and "What did our customers say about our product on Twitter"?
Semantic Data (Triples)
Semantic Data, or "links" are networks of facts that describe relationships among objects. They are stored in "triple stores" as short sentences made up of a subject, predicate and object. Triples form an elegant and powerful way to organize information without requiring rigid schemas and complex "join" relationships.
The Semantic Web is one manifestation of triples, but they are finding utility throughout industry. Semantic graphs can represent a surprisingly diverse set of information. They are increasingly common on sites like the Internet Movie Database (IMDb) and LinkedIn, where concepts are related in networks of interconnections.
Semantic data answers questions like "What actors in Star Wars have worked on movies that also featured Kevin Bacon?", "What actors in the Star Trek franchise also own a home in Beverly Hills?" and "What parts from the Illudium Q36 Explosive Space Modulator were manufactured Spacely Sprockets in Taiwan?"
What does Epinomy do?
Epinomy takes advantage of modern NoSQL database technology to be able to leverage semantic concepts to make text and tables easier to find. We realized that semantic features like "taxonomies", "concepts", "terms", "dimensions", "nodes" and "columns" were all fundamentally the same thing. They are metadata that can be used to tie together the unstructured and the structured world through a network of concepts.