Sparklis: An Expressive Query Builder for SPARQL Endpoints with Guidance in Natural Language

Tracking #: 1161-2373

Sébastien Ferré

Responsible editor: 
Eero Hyvonen

Submission type: 
Tool/System Report
Sparklis is a Semantic Web tool that helps users explore and query SPARQL endpoints by guiding them in the interactive building of questions and answers, from simple ones to complex ones. It combines the fine-grained guidance of faceted search, most of the expressivity of SPARQL, and the readability of (controlled) natural languages. No knowledge of the vocabulary and schema are required for users. Many SPARQL features are covered: multidimensional queries, union, negation, optional, filters, aggregations, ordering. Queries are verbalized in either English or French, so that no knowledge of SPARQL is ever necessary. All of this is implemented in a portable Web application, Sparklis, and has been evaluated on many endpoints and questions. No endpoint-specific configuration is necessary as the data schema is discovered on the fly by the tool. Online since April 2014, thousands of queries have been formed by hundreds of users over more than a hundred endpoints.
Full PDF Version: 

Minor Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Eetu Mäkelä submitted on 22/Sep/2015
Minor Revision
Review Comment:

I find all my remarks in my previous review addressed and can heartily recommend the manuscript for publication.

I do have a couple of minor suggestions still, though:

When talking about the new Figure 6, the current explanation is not optimal, as it relates the feature names to SPARKLIS terms, which on the other hand are themselves not that familiar to an outside reader (for example, the term 'any' doesn't actually seem to be explained in the previous exposition). I suggest you see if you can replace the SPARKLIS term explanations with for example equivalent SPARQL terms or fragments.

Another minor detail is that when you mention, with regard to no-step users, that the cross-domain issues were solved in December 2014, it would be interesting to know if this affected the percentage of such users in any way.

Further, I would like your opinion on if (or how many of) the outliers at either end of the scale of use (0 steps and 1400 steps) are actual users, and not for example bots or random tests run amok.

Typo: "The short searches can be interpreted as directed
searches, where the user as a clear information need
in mind." as->has

When you say that you've improved the tool in response to user mishandling analysis, it would be really interesting to know exactly how.

Review #2
By Vanessa Lopez submitted on 31/Oct/2015
Review Comment:

The authors have answered all the questions I posed in my previous review. In particular the paper presents now more extended discussions on both strengths, limitations and open issues of the approach presented, which are very useful to understand the different paradigms to query the Web of Data.

There is a lot of work behind the system presented here, I think it is a great working showcase for guided interfaces using natural language, with convincing evaluations and an interesting analysis of results in terms of reconciling guidance and expressivity, scalability and portability.

The paper is very well written and I recommend its publication.