Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
Hi Roberto, You can define constant mappings in the mappings wiki [1]. For example in the actor mapping you can define {{ConstantMapping | ontologyProperty = occupation | value = Actor }} and everyone will get an additional occupation Actor, We have a deduplication step so don't worry if it gets extracted twice ;) Cheers, Dimitris [1] http://mappings.dbpedia.org/index.php/How_to_edit_DBpedia_Mappings#Constant_Mappings On Thu, Apr 10, 2014 at 10:59 PM, Roberto Alsina roberto.als...@canonical.com wrote: Thanks everyone for all the awesome answers. You surely have given me a lot of links to follow and a lot of things I need to learn about! I'll take a few days to digest all the information and finish some pending tasks, and then I'll get back to this. One thing I did in our copy of the data was deduce some extra properties from existing data. For example, if there are 3 or more starring pointing at the same person, I added a occupation::actor to him. Maybe there could be some way to automate that process (although this does mark the Dalai Lama as an actor ;-) On Thu, Apr 10, 2014 at 9:56 AM, Marco Fossati hell.j@gmail.comwrote: Hi Roberto, Do you need multilingual support for your app? If so, mapping infobox properties in different languages would be the way to go. Otherwise, raw infobox properties may be enough. You can find them under the under the http://dbpedia.org/property namespace. See my replies below for your examples. On 4/9/14, 9:09 PM, Roberto Alsina wrote: For example: most actors don't have occupation::Actor. http://dbpedia.org/property/occupation Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. http://dbpedia.org/property/tradedAs For the latter, adding mappings in http://mappings.dbpedia.org/index.php/Main_Page should be enough, right? Yep, if you want more homogeneous data in general and support for multiple languages. Hope this helps! -- Marco Fossati http://about.me/marco.fossati Twitter: @hjfocs Skype: hell_j -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Kontokostas Dimitris -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
On 4/11/14 6:50 PM, Paul Houle wrote: One thing to watch out for is that many people have types that are correct but strange. Looking at this report http://basekb.com/subjectiveEye/typeReport/linkBased/Athlete.html Is there a particular reason why you don't expose DBpedia URIs in that page? It would make a world of difference :-) -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen smime.p7s Description: S/MIME Cryptographic Signature -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
One thing to watch out for is that many people have types that are correct but strange. Looking at this report http://basekb.com/subjectiveEye/typeReport/linkBased/Athlete.html we see Bob Hope is the #2 Athlete of all time. Well, perhaps Bob Hope is in a class by himself, but you don't expect him to be a member of that class -- except he is, because he was a boxer early in his career. Other types have this problem (particularly cities) but because people often do different things, and because types like actor and musician are duck types (i.e. you are an actor because you acted in something) you get unexpected results. For instance, I think Bodybuilder, Actor, Politician are all acceptable one-word descriptions for Arnold Schwarzenegger, but you wouldn't expect Author (even though he co-wrote some influential books) or Musician (because his voice is on some fitness recordings) Anyway, I went and did my thing with the Amazon cloud and produced the type assignment file that I promised. If you go to s3://basekb-misc/0.9/freebaseTypesForDbpedia you will find that directory has 12 files in it, and those files together have Freebase types. You can do queries like prefix : http://rdf.basekb.com/ns/ select COUNT(?s) { ?s a :people.person } and get much better accuracy than DBpedia ontology types. That bucket is public access and you should be able to access it with s3cmd, S3 Browser or any other S3 client. On Thu, Apr 10, 2014 at 9:59 AM, Roberto Alsina roberto.als...@canonical.com wrote: Thanks everyone for all the awesome answers. You surely have given me a lot of links to follow and a lot of things I need to learn about! I'll take a few days to digest all the information and finish some pending tasks, and then I'll get back to this. One thing I did in our copy of the data was deduce some extra properties from existing data. For example, if there are 3 or more starring pointing at the same person, I added a occupation::actor to him. Maybe there could be some way to automate that process (although this does mark the Dalai Lama as an actor ;-) On Thu, Apr 10, 2014 at 9:56 AM, Marco Fossati hell.j@gmail.com wrote: Hi Roberto, Do you need multilingual support for your app? If so, mapping infobox properties in different languages would be the way to go. Otherwise, raw infobox properties may be enough. You can find them under the under the http://dbpedia.org/property namespace. See my replies below for your examples. On 4/9/14, 9:09 PM, Roberto Alsina wrote: For example: most actors don't have occupation::Actor. http://dbpedia.org/property/occupation Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. http://dbpedia.org/property/tradedAs For the latter, adding mappings in http://mappings.dbpedia.org/index.php/Main_Page should be enough, right? Yep, if you want more homogeneous data in general and support for multiple languages. Hope this helps! -- Marco Fossati http://about.me/marco.fossati Twitter: @hjfocs Skype: hell_j -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Paul Houle Expert on Freebase, DBpedia, Hadoop and RDF (607) 539 6254paul.houle on Skype ontolo...@gmail.com -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
On 4/9/14 6:53 PM, Pablo N. Mendes wrote: I like very much the contributing back aspect of this. Thanks for offering! One problem is that some pages have no template, making it impossible to use the template-type mappings defined on the mappings wiki. Other people have implemented type inferencing from categories, lists and even from the text. Others, by cross-referencing with Freebase, Cyc, etc. I am wondering if the type statements obtained through all these approaches should not be imported back to DBpedia through some semiautomatic curation method (read mappings wiki beyond templates). I guess we could also use the wiki, and allow people to also add mappings for Lists, Categories, Tables, and other features generated by these approaches? Cheers Pablo Or folks could just publish their mapping documents from a Web accessible URL :-) -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen smime.p7s Description: S/MIME Cryptographic Signature -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
Hi Roberto, Do you need multilingual support for your app? If so, mapping infobox properties in different languages would be the way to go. Otherwise, raw infobox properties may be enough. You can find them under the under the http://dbpedia.org/property namespace. See my replies below for your examples. On 4/9/14, 9:09 PM, Roberto Alsina wrote: For example: most actors don't have occupation::Actor. http://dbpedia.org/property/occupation Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. http://dbpedia.org/property/tradedAs For the latter, adding mappings in http://mappings.dbpedia.org/index.php/Main_Page should be enough, right? Yep, if you want more homogeneous data in general and support for multiple languages. Hope this helps! -- Marco Fossati http://about.me/marco.fossati Twitter: @hjfocs Skype: hell_j -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
Thanks everyone for all the awesome answers. You surely have given me a lot of links to follow and a lot of things I need to learn about! I'll take a few days to digest all the information and finish some pending tasks, and then I'll get back to this. One thing I did in our copy of the data was deduce some extra properties from existing data. For example, if there are 3 or more starring pointing at the same person, I added a occupation::actor to him. Maybe there could be some way to automate that process (although this does mark the Dalai Lama as an actor ;-) On Thu, Apr 10, 2014 at 9:56 AM, Marco Fossati hell.j@gmail.com wrote: Hi Roberto, Do you need multilingual support for your app? If so, mapping infobox properties in different languages would be the way to go. Otherwise, raw infobox properties may be enough. You can find them under the under the http://dbpedia.org/property namespace. See my replies below for your examples. On 4/9/14, 9:09 PM, Roberto Alsina wrote: For example: most actors don't have occupation::Actor. http://dbpedia.org/property/occupation Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. http://dbpedia.org/property/tradedAs For the latter, adding mappings in http://mappings.dbpedia.org/index.php/Main_Page should be enough, right? Yep, if you want more homogeneous data in general and support for multiple languages. Hope this helps! -- Marco Fossati http://about.me/marco.fossati Twitter: @hjfocs Skype: hell_j -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
It's more accurate to say that the Infovore software is a bridge between Freebase, DBpedia, and other RDF data sets. It merges data sets in batch job and creates data sets that are normalized. I think also it :BaseKB is a family of products produced from Freebase and Dbpedia data using the Infovore software. The main product, :BaseKB Now, is a cleaned up version of the Freebase RDF dump that is much easier to work with than the raw dump. :BaseKB is similar to DBpedia and could be used as a substitute in many applications, but Dbpedia has some unique and valuable information that is not in Freebase. As for vocabulary conversion I get asked about that a lot. One reason I haven't done it is that every transformation you do to data risks messing things up and the data quality issues are up in the air enough that a half-baked effort at conversion will cause more problems than it solves. If you keep the vocabulary separate, you can query Freebase's opinion and query Dbpedia's opinion and know things haven't been worse. The mapping process would be done one predicate at a time and would probably be guided by how prevalent the predicates are. Some of the predicates are going to be easy to process (just rewrite them) but other ones might need more work if compound value types are involved or if the types are literal types that have a system and domain dependent meaning that needs to be preserved (is it feet or meters?) It might also be useful to map to some third vocabulary. I know people would like to see DBpedia and Freebase through schema.org eyes and I think that would be a good idea. Common types and properties will get handled quickly but if somebody is interested in some vertical, say boats (20,000 known in Freebase) they probably personally will need to do the work to figure things out. For instance, Freebase is missing a lot of facts about boats that are in DBpedia. A union database will benefit from that, and there ought to be some community process where those fixes can be expressed as rules and added to the system. On Wed, Apr 9, 2014 at 5:25 PM, Kingsley Idehen kide...@openlinksw.com wrote: On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. :baseKB could (and maybe should) pitched as a human-and-machine curated bridge between Freebase, DBpedia, and Wikidata (I think). Have you considered mapping the classes and properties across DBpedia, Freebase, and Wikidata? -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
On 4/10/14 2:37 PM, Paul Houle wrote: It's more accurate to say that the Infovore software is a bridge between Freebase, DBpedia, and other RDF data sets. It merges data sets in batch job and creates data sets that are normalized. Yep! I think also it :BaseKB is a family of products produced from Freebase and Dbpedia data using the Infovore software. The main product, :BaseKB Now, is a cleaned up version of the Freebase RDF dump that is much easier to work with than the raw dump. :BaseKB is similar to DBpedia and could be used as a substitute in many applications, but Dbpedia has some unique and valuable information that is not in Freebase. No need for the replacement pitch. You have a curation-branded dataset culled from DBpedia, Freebase etc... Once loaded, this dataset can serve many useful purposes in conjunction with existing DBpedia and Freebase data. As for vocabulary conversion I get asked about that a lot. One reason I haven't done it is that every transformation you do to data risks messing things up and the data quality issues are up in the air enough that a half-baked effort at conversion will cause more problems than it solves. Mapping at the definitions level (data dictionary, schema, vocabulary, ontology) has more power and longevity than at the instance data level. TBox (entity types definitions) RBox (entity relation type definitions) driven tours are eternally superior to ABox driven tours, across the Linked Open Data cloud :-) If you keep the vocabulary separate, you can query Freebase's opinion and query Dbpedia's opinion and know things haven't been worse. The TBox, RBox, and ABox relations should always be loosely coupled. Conflation is our worst enemy in the Data Economy. The mapping process would be done one predicate at a time and would probably be guided by how prevalent the predicates are. Some of the predicates are going to be easy to process (just rewrite them) but other ones might need more work if compound value types are involved or if the types are literal types that have a system and domain dependent meaning that needs to be preserved (is it feet or meters?) It might also be useful to map to some third vocabulary. I know people would like to see DBpedia and Freebase through schema.org eyes and I think that would be a good idea. Yes, there should be many of these, all loosely coupled. Common types and properties will get handled quickly but if somebody is interested in some vertical, say boats (20,000 known in Freebase) they probably personally will need to do the work to figure things out. For instance, Freebase is missing a lot of facts about boats that are in DBpedia. A union database will benefit from that, and there ought to be some community process where those fixes can be expressed as rules and added to the system. Yes, and there is value here for those who want functional business models in the Data Economy. Kingsley On Wed, Apr 9, 2014 at 5:25 PM, Kingsley Idehen kide...@openlinksw.com wrote: On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
On 9 April 2014 20:09, Roberto Alsina roberto.als...@canonical.com wrote: For example: most actors don't have occupation::Actor. Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. [Resending to list; apologies to Roberto] I've done some work to get things like people's occupations/ reason for notability, and gender, added to their infobox on the English Wikipedia, but have met a lot of resistance (details on request), so it's slow going. An example of success is the role alpine skier and gender symbol in the infobox on on: https://en.wikipedia.org/wiki/Tina_Maze I've also done a considerable amount of work to deploy sub-templates in infoboxes, to improve machine-readability and data granularity, for things like dates, and multiple values, about which I've posted here from time to time. Again, there is resistance in some quarters, but we've had more successes there. I'm always interested to hear about how these are or are not useful and what else Wikipedians could do to improve the reusability of our content. -- Andy Mabbett @pigsonthewing http://pigsonthewing.org.uk -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. On Wed, Apr 9, 2014 at 3:09 PM, Roberto Alsina roberto.als...@canonical.com wrote: Hi! First, a brief introduction. My name is Roberto Alsina, and my team at Canonical is using DBPedia in the upcoming ubuntu touch phone operating system to improve a suggestions engine. What we are doing is, when a user searches for something, we look it up in wikipedia, and then use the entity name in dbpedia to get properties, which we then associate with different results. For example: User types Metallica = Wikipedia matches = DBPedia says Type::band = We suggest searching in grooveshark and youtube All in all, the approach works remarkably well, but we are finding some missing mappings, and we'd like to help improve DBPedia :-) For example: most actors don't have occupation::Actor. Or, publicly traded companies (example: Microsoft) have a Traded as field in their infoboxes but no matching data in DBPedia. For the latter, adding mappings in http://mappings.dbpedia.org/index.php/Main_Page should be enough, right? Looking forward to working on this :-) -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Paul Houle Expert on Freebase, DBpedia, Hadoop and RDF (607) 539 6254paul.houle on Skype ontolo...@gmail.com -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. :baseKB could (and maybe should) pitched as a human-and-machine curated bridge between Freebase, DBpedia, and Wikidata (I think). Have you considered mapping the classes and properties across DBpedia, Freebase, and Wikidata? -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen smime.p7s Description: S/MIME Cryptographic Signature -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
I like very much the contributing back aspect of this. Thanks for offering! One problem is that some pages have no template, making it impossible to use the template-type mappings defined on the mappings wiki. Other people have implemented type inferencing from categories, lists and even from the text. Others, by cross-referencing with Freebase, Cyc, etc. I am wondering if the type statements obtained through all these approaches should not be imported back to DBpedia through some semiautomatic curation method (read mappings wiki beyond templates). I guess we could also use the wiki, and allow people to also add mappings for Lists, Categories, Tables, and other features generated by these approaches? Cheers Pablo On Apr 9, 2014 2:29 PM, Kingsley Idehen kide...@openlinksw.com wrote: On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. :baseKB could (and maybe should) pitched as a human-and-machine curated bridge between Freebase, DBpedia, and Wikidata (I think). Have you considered mapping the classes and properties across DBpedia, Freebase, and Wikidata? -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees___ Dbpedia-discussion mailing list Dbpedia-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
Hi Pablo, We're working on the topic you are discussing. We're semi-automatically committing things back to to Mappings Wiki with bots [1]. Currently we are limiting ourselves to label translations, but will move on to the inferred mappings generated by Airpedia. There is also an Google Summer of Code proposal that includes checking DBpedia info against other language chapters and external data sources such as freebase, wikidata, geonames, musicbrainz etc, and creating a feedback loop to Wikipedia [3] (you can read the full proposal in Google melange if you have access, the allocation of a slot for this proposal is still being debated though). Roberto: If you want to help improve DBpedia, please register in the mappings wiki and request an editor status, we will gladly give you access. DBpedia is a community effort and we don't have the financial backing of Google or the donations Wikipedia gets. In conclusion, any small edits you can make to improve the mappings are greatly appreciated. Cheers, Alexandru [1] https://github.com/ag-csw/missingBot [2] http://www.airpedia.org/ [3] http://wiki.dbpedia.org/gsoc2014/ideas#h359-20 On Thu, Apr 10, 2014 at 12:53 AM, Pablo N. Mendes pablomen...@gmail.comwrote: I like very much the contributing back aspect of this. Thanks for offering! One problem is that some pages have no template, making it impossible to use the template-type mappings defined on the mappings wiki. Other people have implemented type inferencing from categories, lists and even from the text. Others, by cross-referencing with Freebase, Cyc, etc. I am wondering if the type statements obtained through all these approaches should not be imported back to DBpedia through some semiautomatic curation method (read mappings wiki beyond templates). I guess we could also use the wiki, and allow people to also add mappings for Lists, Categories, Tables, and other features generated by these approaches? Cheers Pablo On Apr 9, 2014 2:29 PM, Kingsley Idehen kide...@openlinksw.com wrote: On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. :baseKB could (and maybe should) pitched as a human-and-machine curated bridge between Freebase, DBpedia, and Wikidata (I think). Have you considered mapping the classes and properties across DBpedia, Freebase, and Wikidata? -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about LinkedIn Profile: http://www.linkedin.com/in/kidehen -- Put Bad Developers to Shame Dominate Development with Jenkins Continuous Integration Continuously Automate Build, Test Deployment Start a new project now. Try Jenkins in the cloud. http://p.sf.net/sfu/13600_Cloudbees ___ Dbpedia-discussion mailing
Re: [Dbpedia-discussion] Fwd: Hello DBPedia!
I'm replying to my own email: Hi Pablo: I'm sorry, it's late and I noticed I didn't read your mail correctly. I see you are talking about extending the mappings wiki for tables, lists and the main article text. This approach is also under discussion in GSoC , under the idea 4.18 [1] . [1] http://wiki.dbpedia.org/gsoc2014/ideas#h359-23 On Thu, Apr 10, 2014 at 1:14 AM, Alexandru Todor to...@inf.fu-berlin.dewrote: Hi Pablo, We're working on the topic you are discussing. We're semi-automatically committing things back to to Mappings Wiki with bots [1]. Currently we are limiting ourselves to label translations, but will move on to the inferred mappings generated by Airpedia. There is also an Google Summer of Code proposal that includes checking DBpedia info against other language chapters and external data sources such as freebase, wikidata, geonames, musicbrainz etc, and creating a feedback loop to Wikipedia [3] (you can read the full proposal in Google melange if you have access, the allocation of a slot for this proposal is still being debated though). Roberto: If you want to help improve DBpedia, please register in the mappings wiki and request an editor status, we will gladly give you access. DBpedia is a community effort and we don't have the financial backing of Google or the donations Wikipedia gets. In conclusion, any small edits you can make to improve the mappings are greatly appreciated. Cheers, Alexandru [1] https://github.com/ag-csw/missingBot [2] http://www.airpedia.org/ [3] http://wiki.dbpedia.org/gsoc2014/ideas#h359-20 On Thu, Apr 10, 2014 at 12:53 AM, Pablo N. Mendes pablomen...@gmail.comwrote: I like very much the contributing back aspect of this. Thanks for offering! One problem is that some pages have no template, making it impossible to use the template-type mappings defined on the mappings wiki. Other people have implemented type inferencing from categories, lists and even from the text. Others, by cross-referencing with Freebase, Cyc, etc. I am wondering if the type statements obtained through all these approaches should not be imported back to DBpedia through some semiautomatic curation method (read mappings wiki beyond templates). I guess we could also use the wiki, and allow people to also add mappings for Lists, Categories, Tables, and other features generated by these approaches? Cheers Pablo On Apr 9, 2014 2:29 PM, Kingsley Idehen kide...@openlinksw.com wrote: On 4/9/14 4:53 PM, Paul Houle wrote: The type assignments in DBpedia are very precise (few false statements) but not accurate in the sense that recall is poor; many things fall through the cracks. The real problem is that the the mappings are the map, not the territory. Wikipedia is an encyclopedia for humans, not for machines, so DBpedia has to parse whatever unsane markup they give us. Systems like Wikidata and Freebase can be edited by machines and human ontologists and get better recall for types. http://basekb.com/ is a conversion from Freebase to industry standard RDF. You could use :BaseKB as a substitute for DBpedia, but DBpedia has advantages too because in addition to the 4 million things important enough to be in DBpedia, there is another 37 million unimportant things in :BaseKB that matter only to librarians, video store clerks and professional discographers. These unimportant things will drive you crazy unless you master them, and the easiest way to turn down the noise is to restrict search to the 4 million things. I could make you an RDF file that has statements such as ?dbpediaTopic a ?freebaseType . you could load that together with the rest of DBpedia. That would get you a long way towards good lists. The trouble at this point is that you don't have the freebase types connected to the DBpedia types so you can't join them against the schema to find properties and such. Mapping the types to the DBpedia types would not be that hard either, since the two systems are well aligned. Then you get something that looks like DBpedia but has more accurate types. Freebase has more accurate and better populated data for things like ticker symbols, geo-coordinates, genders, birth dates and the like. It would not be hard to rewrite Freebase statements to ?dbpediaTopic ?freebasePredicate ?anotherDbpediaTopic . and that would produce something that would be remarkably user friendly. :baseKB could (and maybe should) pitched as a human-and-machine curated bridge between Freebase, DBpedia, and Wikidata (I think). Have you considered mapping the classes and properties across DBpedia, Freebase, and Wikidata? -- Regards, Kingsley Idehen Founder CEO OpenLink Software Company Web: http://www.openlinksw.com Personal Weblog: http://www.openlinksw.com/blog/~kidehen Twitter Profile: https://twitter.com/kidehen Google+ Profile: https://plus.google.com/+KingsleyIdehen/about