Solarman:

Live Demo of an Event-Based Denotational Semantics for the Semantic Web


Natural Language Interface

A demonstration of our new event-based denotational semantics using a SPARQL endpoint as the triplestore

This is a live demo of the Solarman application for Natural Language queries to event based semantic web data. Previously our demonstration used an in-program triplestore. To demonstrate the real-world applicability of our semantics, we have redesigned our demo to use a SPARQL endpoint instead as its triplestore. This improved version is faster and now operates directly on the semantic web.

NEW: This version of our interface is experimental, running our semantics directly in your web browser using WebAssembly. All processing is performed entirely on your own device except for the actual SPARQL queries made to our triplestore. There are no intermediaries -- your browser is running our Haskell code directly. In effect, our approach bridges your browser directly to our triplestore. The goal of this research is to seamlessly turn your web browser into a first class citizen of the Semantic Web. WebAssembly support required! Our original non-WebAssembly interface is accessible here.

Query interface
Query result

Examples:

which moons orbit jupiter

who discovered a moon in 1877

More Examples

Novel additions

In addition to these changes, we demonstrate our support for chained prepositional phrases in our semantics here. We also demonstrate a novel way to handle the word by, as used in the passive form of a verb (e.g., discovered by), by treating it directly as a preposition. In doing this, we can support queries such as which moon was discovered in 1877 by hall without any added complexity to the semantics.

We also demonstrate chained superlative phrases! Try using the most in a query.

Query interface

In the input field above, you may type and run any valid query using the semantic functions.

Queries are written in lower case English with no punctuation.

Link to our SPARQL endpoint

List of triples in the graph

Link to Source Code on Hackage