Nuxeo present FISE, a RESTful semantic engine
Open source enterprise content management experts Nuxeo have announced that, as part of the IKS European project, they are working with partners in the project to develop an open source semantic engine with a RESTful interface, dubbed fise. Fise, which stands for Furtwangen IKS Semantic Engine, was initially created in March at the IKS Semantic Engine Hackathon and now Nuxeo have made a demonstration system available for users to get a feel for what a semantic engine can achieve.
Semantic engines are applications designed to process and catagorise documents that are presented to them. This would include determining what language the document is in, what type of article is contained within the document, suggesting useful and meaningful tags for the document, extracting known entities such as people or places, and detecting as yet unknown entities. With this information, it's then possible to, for example, find related documents based on their tags and catagorisation. An example of how these capabilities can be used is provided by the BBC who used semantic engines to process World Cup 2010 articles.
Fise allows developers to run their own semantic engine which offers web based access to the engine with a /engines endpoint allowing a user to submit text for analysis and view the results, a /stores endpoint to view stored and processed documents and a /sparql endpoint for making SPARQL queries of the stored documents. Internally it uses OpenNLP to identify entities, Apache Lucene to index and search entities and it calls on DBpedia to associate entities with locations.
Nuxeo says that although there are semantic engines with web-based APIs, there aren't many which can be distributed under an open source licence for offline use or within a private infrastructure and that fise fills this gap. In the future, the developers plan to add multi-lingual support, the ability to extract relationships and to integrate it with Nuxeo's own content and data management offerings.
The BSD licensed code base of Fise is available on Google Code and there is also a snapshot jar (direct download, 67MB), which can be run on a Java 6 virtual machine locally; more information about fise and instructions for running it locally can be found an a Nuxeo blog posting.