Register for Higher On-Demand Daily Limit

If you've used the Extractiv On-Demand service, you know that getting semantic markup for your documents is as easy and restful as forming a URL.  But you may not have known that the free limit for this service is 10x less if you are not using an API token.  

If you submit queries anonymously, the limit is 100 documents a day.  This anonymous modality is mostly intended for users submitting a couple of documents manually through an input form.  Users who are submitting automated queries should register for a free account, through which you'll receive an API token that will allow you to process up to 1,000 documents per day.  See our documentation for more instructions.

Extractiv for DIY Data Analysis

Extractiv has been mentioned in several news articles this week.  Pete Warden at O'Reilly Radar discusses 4 Free Data Tools for Journalists (and Snoops). Extractiv is mentioned in its connection to the democratization of data analysis in terms of revolutionizing journalism.

The technology is also getting easier to use. Companies like Extractiv and Needlebase are creating point-and-click tools for gathering data from almost any site on the web, and every other stage of the analysis process is getting radically simpler too.

Seth Grimes at Information Week discusses 5 Paths To The New Data Integration.

Here, I am using a Web API to annotate a White House blog page on the recent Tucson, Arizona shootings. Click on Gabrielle Giffords in the center pane and scroll until you see the Details area on the right. That area links, via a URI for Rep. Giffords, to a DBpedia page on the congresswoman, whom I chose for this example in order to honor her. This is an example of content enrichment made possible by the New Data Integration.

Both of these articles stress the importance of data, and that while more data is available than ever before, you have to have tools to make sense of it all. It's like the Twilight Zone episode where Burgess Meredith is at a library with all the books and time in the world, but then breaks his glasses. These articles show that the public is becoming aware of this data need, but that they aren't sure yet what to do. It's nice to see Extractiv mentioned as one of the tools that fill this need.

We've Launched!

After several months and a lot of hard work, we've publicly launched Extractiv!

To get started, you can sign up at http://portal.extractiv.com.  When you first sign up, you'll have access to all of the basic features, which have limitations on your usage.  To take advantage of our full feature set, such as unlimited access to Semantic On-Demand, you'll want to upgrade your account, which you can do from the portal.

If you have any questions, we recommend reviewing the documentation or submitting a support ticket.

We look forward to seeing what semantic applications are developed with the help of our service!

New Crawling Options: Domain Packs and More!

For those of you who have run Extractiv Crawling jobs recently, you'll notice some new additions to the create job panel.  The first is domain packs.  These are useful for specialized interests, and give a whole set of types that are topically related.  Currently available domain packs are computer, entertainment, finance, medical, politics, and sports. 

After selecting the entity types you want to use, you're given the option to enable entity coreference.  Then, in addition to selecting specific relation types of interest, you can choose to run generic relation extraction.  With these changes, many of the features previously available only on the On-Demand Platform are now available on the Crawling Platform as well.

Semantic On-Demand Adds Entity Linking with DBpedia Support

Our Entity Linking service is now in beta! Entity linking is a huge milestone for us because it transforms Extractiv's semantic output into linked data by automatically determining the identity of each entity mentioned in your text and associating it with a DBpedia URI whenever one exists.

Entity linking is tricky because names often have multiple meanings. For example, "Python" might refer to a type of snake, a programming language, a comedy troupe, or a number of other things. By intelligently using context, the linking service figures out which meaning makes the most sense and generates a DBpedia link accordingly.

Any type of entity may be linked as long as it has a URI in DBpedia. DBpedia is a companion project to Wikipedia whose goal is to provide a machine-readable version of Wikipedia in RDF format. We plan to eventually add links to other public knowledge bases, but we started with DBpedia because its URIs are already used to denote real-world entities in a wide variety of data sets, and because, with millions of topics, it provides very broad coverage.

All of our output formats (HTML, JSON, and of course RDF/XML) support entity links, but linking is currently only available using the On-Demand platform. Support for linking in the Crawling platform is coming soon.  Check our documentation for more details.