Forget big data, here comes ‘smart data’: Semantic Web Meetup

Paul Brock built a search system for scientists at Janseen Scientific Affairs, LLC, subsidiary of Johnson & Johnson. Despite what you’d think by the famous company name, he did not have deep pockets and a large budget, he said at the first Philadelphia Semantic Web Meetup in two years.

Brock, associate director of search solutions for Janseen Scientific Affairs, was able to get a good system up and running by focusing on the things that would deliver the most impact. For example, he created facets for quick searching (similar to what are on most major shopping websites, Zappos being one). The vast majority of his system users make use of the facets for searching. (A technical detail: the facets are stored as a taxonomy in Protégé.)

During the meetup, attendees also heard from Mike Grove, Chief Software Architect for Washington, D.C.-based semantic web company Clark & Parsia.

He discussed a business-motivating aspect of the RDF database Stardog that he built. Instead of encoding business rules in programming code as is normally required, the rules can be encoded separately from the codebase. Since non-programmers can be involved in encoding the rules, programmers and the IT department are no longer a bottleneck in development of the product. In addition, actual experts in business logic can implement the business rules rather than programmers.

Here are some more highlights from the SmartData: Enterprise Semantics and Search meetup:

What is Smart Data vs Big Data?

There’s been a lot of talk about “Big Data.” So what is “Smart Data”? Adding more data isn’t the solution to all problems. Getting value from data comes down to how easily you can do something with it. Smart data is data with semantics attached. Semantics gives data meaning. More specifically it creates data with computer understandable meaning. This means the computer can help. This makes it easier to use the data. Analysis, business intelligence, decision support, etc are all made easier.

What is Stardog?

Stardog is the leading RDF graph database that has semantic search capabilities built-in. Stardog has come a long way in the past few years, like many fast-moving tech solutions. It’s still easy to set up and get working with it. It’s posted some impressive performance gains and now has built-in reasoning capabilities.

Use Cases for semantic reasoning in the enterprise:

Best Buy: Stardog powers Best Buy’s “Like for Like” service (given a product SKU, the most similar products are returned). Stardog also powers searches of the Best Buy product catalog. It’s also used for seasonal promotions searches, like the 2013 Stocking Stuffer promo.

SAIC: Security policies are represented in XACML. These are automatically converted to OWL. Once converted to OWL, the rules are loaded into Stardog and semantic reasoning is used to determine if any inconsistencies exist in the policies. Do the policies do what you think they do? Do any policies allow actions that other policies deny? Are there any policies that are redundant?

NASA: Via Stardog, semantic reasoning infers new connections in the data. This allows them to search existing data to extract information that had previously not been findable. This new search capability saved NASA $38 million a year. It is an official W3C case study for the use of semantic technologies.

History of this type of information science problems/ solutions:

We also got a history lesson on how old some of these information problems and solutions are. For example facet categorization has been around as a concept since the 1930s. It was not a popular way of categorizing data at the time, but in modern computing of course it is used extensively.

For example, most shopping websites use facets as a way for customers to quickly navigate the inventory. These facets will often be represented by a taxonomy or ontology. Once in an ontology, semantic reasoning can be applied to establish any inconsistencies in the ontology and to derive new information.

Small things you can do to improve search results:

There are many pieces of metadata that might seem unimportant, but if known and used will improve search results. For example, knowing authorship of a document can be used to calculate relevance and importance of a document for ranking results.