Concept Search

Concept Search  


Easily gather intelligence by vetting your unstructured data using our unparalleled Concept Search


What is Concept Search?

What makes SIGINT’s Concept Search so unique is that it is search with the understanding of the concept you’re looking for, not merely search for a keyword. When you search with a Concept Search engine like SIGINT’s, it does not just look for those words that literally match your query, it looks for the whole idea by comprehending the meaning of the words in context.

How does Concept Search Work?

SIGINT’s concept search technology is based on math and statistics and works across any language. Concept search works by learning the meaning of words in the same way that people do –  from the words that accompany and surround each word. For example, if you hear the word “court” in isolation, you may not know what that word means.  But if you hear the word “court” in the same sentence with the word “tennis,” or the word “judge,” now you know what “court” means. 

Keyword search is shown to 95% of information in unstructured data. Our unique concept search enables you to:

  • Identify relevant documents without knowing the exact words to search for
  • Uncover specific jargon or code words
  • Spend less time trying to come up with the right keywords
  • Find documents that are most about the concept first
  • Discover words used in frequent proximity to another

Key benefits of Concept Search include:

  • Works independent of dictionaries or language barriers
  • Increases the likelihood of finding a larger number of relevant documents
  • Finds relevant documents that keyword search cannot
  • Decreases the time spent perusing irrelevant documents
  • Reduces the time and effort when searching for pertinent information

Example of Concept Search

With concept search words that have uninteresting ordinary meanings may be found to be used in peculiar ways as a kind of code. A good example is “Rawhide” in the Enron emails. “Rawhide” could mean a kind of leather or an old TV show, but in the context of the Enron emails, “Rawhide” actually refers to one of its off-books partnerships. Rawhide was found to be associated with words like “Osprey”, “Cayco” and “LJM,” not “leather” or “suede.” These words signify some of the other off-books partnerships that were central to Enron’s fraud scheme.