On 11/10/11 12:29 AM, Robert Siemer wrote:
On 10/11/11 10:34, Kingsley Idehen wrote:On 11/9/11 8:50 PM, Robert Siemer wrote:What I actually want to query: -Android devices with a display of at least 800x480Out of obvious reasons I reduced that to: -list all android devices with display information included After a couple of days I came up with this query: SELECT DISTINCT ?subject, ?display { { ?subject<http://purl.org/dc/terms/subject> <http://dbpedia.org/resource/Category:Android_devices> . } UNION { ?subject a<http://dbpedia.org/class/yago/AndroidDevices> . } OPTIONAL { ?subject<http://dbpedia.org/property/display> ?display } } My problem: Where is that “raw” infobox dataset, which promises “complete coverage of all Wikipedia properties” with minimal clean-up? The downloadable infobox_properties file and the http://dbpedia.org/snorql/ sparql endpoint return only crap for the display property like: “4” or empty values! (Try the query yourself!) The live.dbpedia.org/sparql endpoint returns more, but still useless. I’m aware of the missing ontology mappings for the mobile phone infoboxes (http://en.wikipedia.org/wiki/Template:Infobox_mobile_phone), but: Should dbpedia live not import the raw values when there are no mappings? The wikipedia template uses micro-templates like {{convert|2.1|in|mm|abbr=on}}, how does dbpedia handle that? How does IntermediateNodeMapping separate the property string?? By spaces alone? Then how to handle this? | display = [[TFT LCD]], {{convert|3.2|in|mm|abbr=on}} diagonal<br /> 320×480 px HVGA<br /> 1.5:1 aspect-ratio wide-screen<br /> 256K colors As far as I understand, CustomMappings are not implemented via media-wiki, is that right? – Would be nice to have some kind of RegexMapping, with: 1) a regular expression retrieves one or more values (named groups) 2) multiple regular expressions can be given 3) values retrieved can be subject to some mathematical/conditional cleanup (e.g. if first_var< second_var then “short_side” = first_var; “orientation” = portrait) 3b) and some more examples: if xyGA = HVGA then “short_side” = 320 3c) and maybe some extra calulations: “dpi” = sqrt(...+...)/... So, how do I get that display info out of dbpedia at all? And how to improve the situation for easy retrieval of both display dimensions? Thanks, Robert ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussionRobert, Orthogonal to your quest, since this is via the LOD cloud cache which has DBpedia + other datasets: http://lod.openlinksw.com/fct/facet.vsp?qxml=%3C%3Fxml%20version%3D%221.0%22%20encoding%3D%22UTF-8%22%20%3F%3E%3Cquery%20inference%3D%22%22%20same-as%3D%22%22%20view3%3D%22%22%20s-term%3D%22%22%20c-term%3D%22%22%3E%3Ctext%3EAndroid%3C%2Ftext%3E%3Cclass%20iri%3D%22http%3A%2F%2Fdbpedia.org%2Fclass%2Fyago%2FAndroidDevices%22%20%2F%3E%3Cproperty%20iri%3D%22http%3A%2F%2Fdbpedia.org%2Fproperty%2Fdisplay%22%3E%3Cview%20type%3D%22list%22%20limit%3D%2220%22%20offset%3D%220%22%20%2F%3E%3C%2Fproperty%3E%3C%2Fquery%3E On DBpedia itself you have: http://dbpedia.org/fct/facet.vsp?cmd=load&fsq_id=4What is that good for? My problem is that the property values in dbpedia don’t reflect the infobox of Wikipedia. Example: property:display of Galaxy_Nexus should be: HD Super AMOLED PenTile[6] 4.65 in (118 mm) diagonal 720×1280 px at 316 ppi 16:9 aspect-ratio 16M colors 100,000:1 contrast ratio (typical) 10 μs response time But I get a couple of properties, with integer values of 10, 16, 720 and 100000. Same problem with HTC_Flyer. Instead of 7-inch 1024×600 px I get only “7” out of dbpedia. I see myself where the “7” comes form, but I nevertheless need the raw value from dbpedia to be useful to me... Robert ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Robert,Its good for understanding the data across many dimensions. Quality is such a dimension. You can triangulate back to the mappers and even further back to the source data.
There are two problems: 1. Mappers 2. Resulting data quality 1&2 ultimately effect the ability to query the data in sensible ways. As I said, my response was orthogonal :-) -- Regards, Kingsley Idehen President& CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca handle: @kidehen Google+ Profile: https://plus.google.com/112399767740508618350/about LinkedIn Profile: http://www.linkedin.com/in/kidehen
smime.p7s
Description: S/MIME Cryptographic Signature
------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1
_______________________________________________ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion