- From: Alfredo Serafini <seralf@gmail.com>
- Date: Mon, 4 Aug 2014 22:13:02 +0200
- To: Ruben Verborgh <ruben.verborgh@ugent.be>
- Cc: public-linked-data-fragments@w3.org
- Message-ID: <CADawF4N35AM05oyM0NZeYEe+TdpBXXoDZ4rNnhCRfMdw5rCtFQ@mail.gmail.com>
Yes Ruben, I agre the current examples are yet in that direction! The point is just to improve user experience in a more graphical way for common users. Talking with some colleagues recently on a mailing list about Open Data here in Italy it appears that there is a lack of good tools for exploring datasets: someone suggested something built on top SPARQL, and even if this can be useful, I think start introducing the Linked Data Fragment this way could also be a good idea, since one of the things that I'd like to have is the ability to mashup resources which are not actually obtained by a SPARQL query at all :-) and this is probably possible by introducing LDF. Yes I have some concrete idea in mind, basically I was thinking on a faceted exploration and/or on a "cloud" graph, which can be used to explore by serendipity. Nothing too much original maybe, but those patterns are recognizable, and if such a tool exist, I'd be curious about the chances to explore on the boundaries of the so-called "linked data cloud" :-) Good to know that a facet browser is already in production (which is one of the options we have cited on the ML, in some way). An interactive "cloud" diagram could be also good for tutorials etc. In the past I started writing something using directly SPARQL with a minimal backend: the basic idea was to query by type, then uri, and perform a DESCRIBE over the uris. The problems of that approach are in handling resources on the client, maybe something could be done now with a combination of D3/angualr or similar libraries, I didn't have yet time to trying something in this direction. The main problem for this I could see with LDF is in re-aggregating them to recompose the unity of a specific resource, but since this is more or less what we have to do with triples most of the times, I suppose is not very far from what the existing client already do. Probably an intermediate component which acts as a lens is also needed (something similar to the "old" Fresnel lens), to handle some special properties in a peculiar way (for example dc:title, rdfs:label etc) for the visualizations. Hope I express the idea in a clear way: the point is not only on fancyness, but it appears several users are scared by too simple or let's say too textual interfaces, even if one provides them the explanations on how to use them. I think the approach of LDF could greatly simplify access to things which are too much complex for a common user (such as SPARQL endpoints, or even RDF repositories), but in order to test this it could be useful to introduce a more friendly UX, just that. What do you think about? Thanks for the reply, sorry if I'm going off-topic Alfredo 2014-08-04 21:25 GMT+02:00 Ruben Verborgh <ruben.verborgh@ugent.be>: > Hi Alfredo, > > > I suggest that the next step should be constructing a visualizer for > Linked Data Fragments > > How would you envision such a visualizer? > I presume you mean for triple pattern fragments, right? > Would it be a facet browser? > An interactive cloud diagram kind of thing? > > What we have now is not "visualization" as such, > i.e., nothing fancy or very graphical, > but it does make the fragments "visible". > I.e., you can see actual fragments of a dataset > by just following links and forms from > http://data.linkeddatafragments.org/dbpedia. > > One of my colleagues is working on a facet browser > that works with triple pattern fragments. > Would this go in the direction of what you say? > > > tIt could be great if there were examples a user can play with (also > graphically), since they are helpful for understanding what's behind at an > intuitive level. > > Absolutely. Do you have concrete things in mind? > > Best, > > Ruben
Received on Monday, 4 August 2014 20:13:30 UTC