Semantic Intelligence – NextGen BI
A must read about Semantic Intelligence, the next generation business intelligence. Very similar to the concept of semantic web.
Semantic intelligence provides early identification and analysis of consumer sentiment, purchasing trends, market deals, and competitive information – and uncovers this data not only from within a organization’s network, but also from the most unstructured corners of the Web. You may be thinking that a normal Google search can uncover any Web-based information, but unlike simple keyword search, semantic intelligence uncovers the meaning the words express, in their proper context, no matter the number (singular or plural), gender (masculine or feminine), verb tense (past, present, or future), or mode (indicative or imperative).
For example, say you’re a chef and you’re looking for details on how to make soup with healthier ingredients, so, you keyword search “apple stock.” Try it right now – you’ll get dozens upon dozens of pages about Apple, the company. If you try to narrow the search and type, “apple stock and cook,” you will still get hundreds of erroneous search results about Tim Cook and Apple, the company.
Semantic intelligence incorporates morphological, logical, grammatical, and natural language analysis that translates into higher precision and recall when searching for information. By providing information in the requested context and form, semantic intelligence helps organizations strategize, analyze, and make predictions because you’re getting the correct data – and in these economic times, having the right foresight can save a business.
I agree that there are many opportunities in briging the world of semantic information to the BI world. That’s something we will do at Litebi with our liteConcepts module. I think though that Semantic Intelligence should not be focused on answering search inquiries, Google will do that better than any BI tool. We think that a BI tool that “understands” the nature of your business and is able to find and organize information from the Internet that is relevant to you is something very powerful. By the way, as a follower, congratulations for your blog, we would love to see an article about our BI SaaS tool :) We want to change the way people think about BI.
Best regards,
Javier
Javier
26 Mar 09 at 1:37 am
If you call this stuff “semantic intelligence,” noone is going to want to use it, eh? Really it’s business intelligence, brought back to its roots in unstructured (document) sources, best delivered via suitably extended search and BI interfaces and line-of-business applications.
By the way, Google already has a basic level of “semantic intelligence.” Type “4+5″ or “map philadelphia” or “ORCL” or “weather 20912″ into the Google search box and you’ll see.
Seth
Seth Grimes
26 Mar 09 at 5:27 pm
As usual, the race to build products has technologists falling into a black hole. Semantic Intelligence is a type of mental model human beings use to make sense of the world, improve fitness, and compete for resources in the evolutionary race for survival. Semantic Intelligence is the cognitive facility that allows us to integrate information from six different bio-physical facets, and store these as simulations about how the world works. To bridge the gap from science to technology, including how to implement Semantic Intelligence as a BI platform, you only need to understand the algorithm and data model behind “measures of semantic relatedness”. The Data Warehouse Institute is starting to offer courses in this area and Semantic Intelligence. Once you learn to see patterns hidden in all manner of digital artifacts, the concept of structured and unstructured data disappears. The only tools you need to implement these systems are an RDF triple store, SPARQL, AJAX and jQuery. Over the internet we are seeing transaction rates, single-threaded, of 1M+ per second, on $1000 commodity servers running either Linux or Windows. The back-ends are terabyte file systems with open source indexing schemes. If you need 100 Million transactions per second, simply run it threaded, or using parallel code. Join the revolution.
Art Conroy
29 Aug 10 at 6:13 am