Feedback

[csarven]

Apologies for being lazy and not separating this into specific issues. I hope the feedback is of use to you.

Initial feedback on LD Viewer at DBpedia:

• Looks like a good step up in some aspects from the out of the box Pubby view. I still find it geared towards developer/SW/LD folks - was that the intention?
• UI has too many borders/lines. You can often achieve the same effect (separating objects) simply by colour (i.e., with sufficient contrast) alone. Borders/Lines costs more to process cognitively.
• Not enough contrast between the gray background and blue blinks.
• I like the tooltips quite a bit. I also like the header information for the resource to get a quick feel.
• Share action is not working for me on Firefox/Ubuntu. I don't fancy this.
• I don't think the alternative views (LodLive, RelFinder) should be here. While I like what they offer, I think this clutters up the interface.
• Consider using only the Wikipedia link and remove Share and Alternative views. So, taking that further, I would frame this a bit differently: "Source/External". For most of the resources, it would be Wikipedia, but for the others, you can use their links and/or icons (e.g., for sameAs/exactMatch/seeAlso/homepage values)
• Consider showing the images or their thumbnails instead of their local paths/filenames. This is a data issue, I know, but property:img "CH Bern Kramgasse.jpg @en" is a bit weird and completely useless. If you are focused on making the interface better, consider how you can improve that e.g., if there is a common path to where all those images can be dereferenced, use it regardless of what the data (i.e., triples) provides. Perhaps you need to hardcode that base URL into LD Viewer until the data is fixed. (Aside: In your JS, check if the the values of certain pre-determined properties do not begin with http, then prefix the most applicable base URL for them)

Other feedback:

I took a quick peek into LD Viewer code because I wanted to see whether it can be re-used on other endpoints easily. I installed it and tested on the endpoints under http://270a.info/ . My issues:

• Invalid SPARQL queries for Jena e.g., should use "(?p AS ?x)". I don't know if there are others due to queries being only good for Virtuoso or whatever.
• The labeling "system" is hardcoded and limited to rdfs:label. Instead of sending a query out that's only specific to rdfs:label, get the subject resource in question and inspect to see which "labels" you can use. For example, cycle through some common checks rdfs:label, skos:prefLabel, dcterms:title etc., and use whatever is available.
• DRY
• The config (cfg.js) is not really a "configuration" file. That looks as if it is part of the rest of the code. A configuration would contain a minimal set of things that would be useful for the administrator/developer to set up the system. Minimal!
• LD Viewer acts like a stand-alone JS application. IMHO, that's not very Web friendly. Consider the response for:

curl -LH "Accept: text/html" "http://ldv.dbpedia.org/#/page/Switzerland"

.. and how would standard HTML crawlers would see this resource?