The consequence is that numbers and booleans need to be casted for being used.ģ. In addition, each literal is expressed as a string value with a datatype. As a consequence, this information is bounded at a deeper level in the JSON structure than the one the developer expects. Reducing and parsing.The value of a property is always wrapped in an object with at least the attributes type (uri or literal) and value, that contain the information. In practice, the developer ignores completely this part and he checks for the availability of a certain property directly in the JSON tree.Ģ. This is the case of the head part that contains the list of variables that one might find in the results. Skip irrelevant metadata.A typical SPARQL output contains a lot of metadata that are usually not useful for web developers. We describe four different needs that developers face when manipulating SPARQL JSON output:ġ. Current habits, expertise and coding technologies for web applications make the manipulation of nested objects easier for developers than the manipulation of triples.Īs an example, let's retrieve a list of Italian cities, with their name and an illustration from DBpedia 1, limited to the first 100 results. ) that satisfies the query, which is not handy for efficient client processing.
In practice, the output of SPARQL endpoints is a set of all possible bindings (of the form However, the nature of data models between the Web of Data (RDF, graph-oriented) and the Web of Applications (JSON, document-oriented) is very different. The latter one, in particular, has been introduced with the purpose of easing the consumption of the data by web (and not only) applications, through its complete compatibility with JavaScript. The SPARQL W3C recommendation defines not only a query language for data represented in RDF, but also the protocol for retrieving the data, including a standardised format for the answers in XML, CSV/TSV and in JSON .
0 kommentar(er)
