We’ve overcome the first two complexity barriers by introducing standard protocols, encoding and formats for representing information.The next step would be establishing an abstraction layer and standards for meaning.Once this is done, web APIs and data sources can exchange information on the meaning level, and do this transparently.Data driven applications are expressed as data flow graphs, and the problem of building these applications is reduced to a path finding problem.
A typical data flow, starts from a set of inputs, goes through a transformation & processing flow, and then output as a machine readable, human readable format, persisted or used as a trigger for external applications.Sensors can be for example the user’s GPS coordinates.
99.9% of the information on the web is free form text
Web data is partial, noisy, inaccurate, not objective
Abstraction stack - towards data aware apps
A typical data flow Input Process Output XML/ JSON Order Search Filter Transform Data aware UI components Interaction Aggregate Format Persist Sensors Generate Compute Augment Stream Trigger
Data aware web - Pragmatic approach Data aware app host Today’s web Semantic Abstraction layer Reasoner & data flow manager Query Enrichment APIs Result refinement Abstractions Library Synthesis Unstructured Web pages Semantic Store Entity & Fact extraction
What we’ve done at SemantiNet High performance graph storefor Wikipedia & Linked-data Extensible & Dynamic query enginewith a simple XPath-like query language Semantic abstraction layer over APIsincludes semantic query refinements A complete stack of NLP librariesinteroperable with other APIs Ontology aware contextual disambiguation Templating engine Web development environment for data aware apps
Thank you More coverage on: semanticweb.com/build-data-aware-apps-without-the-hassle_b17465 blog.headup.com Sagied [at] semantinet [dot ] com