About Zemanta

Zemanta API is state of the art in text analysis.

For any input text you get entities, related images, articles, hyperlinks, and tags to use to your liking.

Start building

It's easy. Just three steps:

  1. Create developer account
  2. Obtain API key
  3. Get sample code

Zemanta RDF response

Structure of Zemanta's RDF/XML response was inspired by Linking Open Data initiative, other APIs offering semantic responses and most importantly ideas campioned by W3C.

Structure

RDF format allows for many ways of expressing the same relationships. We chose relatively flat approach that is very easy to comprehend even for people that are not experts in semantic web. An example response can be found here.

Namespace:

Responses use only three vocabularies and consequentially namespaces: golden standard rdf namespace (http://www.w3.org/1999/02/22-rdf-syntax-ns# as rdf:), owl namespace (http://www.w3.org/2002/07/owl#) and Zemanta's own namespace (http://s.zemanta.com/ns# as z:). We leave it to the Semantic web stack to offer reconciliation of the terms between different vocabularies.

Base URI:

Each request in Zemanta's system gets unique identifier called RID. Using RID as part of the URI an unique URI is assigned to every response. This URIs is not dereferencable (for example http://d.zemanta.com/rid/54695302-1264-4103-9a2a-dfee88fdcab2). All top-level objects are then related to this document via "z:doc" relation. This means that you can load many responses from Zemanta API into a single semantic graph and there will be no clashes.

There are a few types of top-level objects:

Also very important object is Target which determines outside references and their basic properties (such as title, publishing date, etc)

Document:

There's only one Document-typed object in each response, it includes information about the request, input text and title, and rid and HTML signature. Properties are the following:

Related:

Related object represents an relationship to certain document that Zemanta found to be related to the input text. It also offers basic description of that document, such as title, time of publishing, and the question if Zemanta was used when producing the related document.

Image:

Image object represents an relationship to certain image that Zemanta found to be related to the input text. It also offers basic description of that image, such as title, time of publishing.

Category:

Category object represents an relationship to certain category that Zemanta found to be related to the input text. It also offers basic description of that category, such as title.

Keyword:

Keyowrd object represents an relationship to certain keyword that Zemanta found to be related to the input text. It also offers basic description of that keyword, such as name.