Home

Awesome

<meta charset="UTF-8"/>

What is Sparklis?

Sparklis is a query builder in natural language that allows people to explore and query SPARQL endpoints with all the power of SPARQL and without any knowledge of SPARQL, nor of the endpoint vocabulary.

Sparklis is a Web client running entirely in the browser. It directly connects to SPARQL endpoints to retrieve query results and suggested query elements. It covers a large subset of SPARQL 1.1 SELECT queries: basic graph patterns including cycles, UNION, OPTIONAL, NOT EXISTS, FILTER, BIND, complex expressions, aggregations, GROUP BY, ORDER BY. All those features can be combined in a flexible way, like in SPARQL. Results are presented as tables, and also on maps. A configuration panel offers a few configuration options to adapt to different endpoints (e.g., GET/POST, labelling properties and language tags). Sparklis also includes the YASGUI editor to let advanced users access and modify the SPARQL translation of the query.

Sparklis reconciles expressivity and usability in semantic search by tightly combining a Query Builder, a Natural Language Interface, and a Faceted Search system. As a Query Builder it lets users build complex queries by composing elementary queries in an incremental fashion. An elementary query can be a class (e.g., "a film"), a property (e.g., "that has a director"), a RDF node (e.g., "Tim Burton"), a reference to another node (e.g., "the film"), or an operator (e.g., "not", "or", "highest-to-lowest", "+", "average").

As a Faceted Search system, at every step, the query under construction is well-formed, query results are computed and displayed, and the suggested query elements are derived from actual data - not only from schema - so as to prevent the construction of non-sensical or empty results. The display of results and data-relevant suggestions at every step provides constant and acurate feedback to users during the construction process. This supports exploratory search, serendipity, and confidence about final results.

As a Natural Language Interface, everything presented to users - queries, suggested query elements, and results - are verbalized in natural language, completely hidding SPARQL behind the user interface. Compared to Query Answering (QA) systems, the hard problem of spontaneous NL understanding is avoided by controlling query formulation through guided query construction, and replaced by the simpler problem of NL generation. The user interface lends itself to multilinguality, and is so far available in English, French, Spanish, and Dutch.

When refering to Sparklis in scientific documents, please use the following citation.

Sébastien Ferré: Sparklis: An expressive query builder for SPARQL endpoints with guidance in natural language. Semantic Web 8(3): 405-418 (2017)

Where can I try Sparklis?

Simply follow those steps:

  1. Go to the online application
  2. Select a SPARQL endpoint in the dropdown list at the top (the default one is Core English DBpedia, a core subset of DBpedia)
  3. Build your query incrementally by clicking suggested query elements (in the three lists of suggestions, and through the hamburger menu in the query), and clicking different parts of the query to change focus. The suggestions are relative to the current focus.

We recommend to visit the Examples page where there are 100+ example queries over several datasets. Every example query can be opened in Sparklis in one click, and the query can then be further modified. For a number of them, there is a YouTube screencast to show how they are built step by step.

How can I use Sparklis on my own RDF dataset?

It is enough to have a SPARQL endpoint for your dataset that is visible from your machine. It can be a publicly open endpoint (like for DBpedia or Wikidata), or a localhost endpoint (I personally use Apache Jena Fuseki but other stores should work too). The one important condition is that the endpoint server be CORS-enabled so that HTTP requests can be made to it from your browser, where Sparklis runs.

Here a few recommendations about the contents of your store for best results:

The Configure menu offers a number of options to adapt Sparklis to your endpoint, and control the display. Here is a non-exhaustive list:

Sparklis makes use of standard and non-standard properties to get more control on the building of queries, and on the display of suggestions and results. For ech property, there is generally a configuration option to activate its use.

Once you have found a good configuration of your endpoint, you can generate a permalink with the button at the top, which you can share to the endpoint users. Those permalinks also include the current query and current focus, so you can also share example queries and template queries. You can also save queries by simply adding bookmarks in your browser.

If you find your endpoint of general interest, you are welcome to suggest me to add it to the list of SPARQL endpoints.

How do I reuse Sparklis in my web site?

As Sparklis is only client-side code, it is possible to integrate Sparklis into your website by simply copying the contents of the webapp folder among your website files, and adding links to it from your web pages, with URL arguments containing the endpoint, the configuration, and possibly an initial query. To get those URLs, simply navigate in Sparklis and copy (and adapt as needed) the browser URL.

You can adapt the appearance of the main HTML file (osparklis.html, osparklis.css) as long as you retain the Sparklis name, and the credits in the page footer. You can for instance hide some configuration options and elements, you can change the look-and-feel, and the layout of elements. Be careful not to delete element ids and classes that are used by the JS code of Sparklis.

Let me know of successful integrations, and also of problems you encounter in the process.

Compiling Sparklis from the source code

Sparklis is developed in OCaml, and compiled to Javascript with the js_of_ocaml tool. It is strongly recommended to use the opam tool to manage OCaml dependencies.

The following build steps were found to work on Ubuntu (20.04 LTS) by waldenn:

    bash -c "sh <(curl -fsSL https://raw.githubusercontent.com/ocaml/opam/master/shell/install.sh)"
    opam install csv lwt js_of_ocaml js_of_ocaml-lwt lwt_ppx js_of_ocaml-ppx str unix num xmlm xml-light lablgtk2
    eval $(opam env)
    sudo apt-get install camlp5
    git clone https://github.com/sebferre/sparklis.git
    cd sparklis
    make

The following build steps were found to work on Mac OS X 10.15.7 (using Macports as a package manager) by CaptSolo:

sudo port install opam

# this also sets up environment variables in ~/.bash_profile
opam init

# OCaml version 5 is too recent (some packages missing), installing 4.14.0 instead
opam switch create 4.14.0
eval $(opam env --switch=4.14.0)

# required as a dependency when doing opam install
sudo port install expat

opam install csv lwt js_of_ocaml js_of_ocaml-lwt lwt_ppx js_of_ocaml-ppx num xmlm xml-light camlp5

git clone https://github.com/sebferre/sparklis.git
cd sparklis
make

Run Sparklis as Docker Container

Sparklis is available as Docker image.

Pull the image:

docker pull sferre/sparklis:latest

Run a container with name sparklis:

docker run -d --name sparklis -p 8080:8080 sferre/sparklis:latest

For Apple Silicon (M1/M2):

docker pull  --platform linux/x86_64 sferre/sparklis:latest

and

docker run -d --name sparklis --platform linux/x86_64 -p 8080:8080 sferre/sparklis:latest

The webserver exposes two paths:

Credits

Author: Sébastien Ferré

Affiliation: Univ. Rennes 1, team SemLIS at IRISA

Copyright © 2013 Sébastien Ferré, IRISA, Université de Rennes 1, France

Licence: Apache Licence 2.0

Citation: Ferré, Sébastien. ‘Sparklis: An Expressive Query Builder for SPARQL Endpoints with Guidance in Natural Language’. Semantic Web 8(3) : 405-418. IOS Press, 2017. PDF