Hello everyone,
I'm Ismael Rodríguez, a Software Engineering Master student at UPC,
Barcelona, Spain.
I am very interested in the "DBpedia Mappings Front-End Administration"
<http://wiki.dbpedia.org/ideas/idea/326/dbpedia-mappings-front-end-administration/>
idea,
as I see it fits my knowledge and my topic of interest. If I understood
right, although the Extraction Framework was adapted to support RML
mappings by a project of GSoC 2016, the user interface to create mappings
is still done by the MediaWiki, not supporting RML mappings and needing
expertise on Semantic Web.
However, I have some doubts about how the new Frontend would replace the
MediaWiki. You describe it as a project to "*facilitate lay users
interaction with DBpedia Mapping rules in RML, data transformations, and
overall DBpedia dataset generation*". So, in addition to creation and
interface for the edition of DBpedia Mapping rules:
- Should it also provide discussion support for users being able to
discuss and propose changes to existing rules?
- Should it also serve to DBPedia administrations to schedule and/or
execute data transformation jobs?
As the main goal is to facilitate lay users to contribute to the mapping of
infoboxes to the DBPedia ontology, I think that the mapping editor should
be the priority. I though about some functionalities that it should have.
1. To provide help with RML syntax tips, structures and code completion.
2. To give suggestions based on already existing data on dbPedia.
I was thinking that the editor should provide suggestions with two goals:
to make the mapping process easier for not technical users, and for
achieving greater consistency and less errors in the mappings. Some ideas
are:
- Based on the property labels of the infobox the user is mapping,
properties of the DBPedia ontology could be suggested. For example, if a
user wants to map a Soccer Player's infobox, the system could suggest
DBPedia property URIs for each of the infobox labels, such as
http://dbpedia.org/ontology/team in the case of "football_team" property.
- The suggestions could be based both by lexical similarity and/or the
properties that are mostly used by instances of a similar type. For
instance, if the user wants to create a mapping for the infobox of a
football player, the system could suggest those DBPedia properties that are
most used on instances of Football Players.
As the suggestions would be based on actual data already in DBPedia, it is
more likely that, in the case that a infobox property could be mapped to
several DBPedia properties (footballTeam, team, playsIn), the user would
choose the property that is more common in the context of the type of
infobox he is mapping. Therefore, the quality of the mappings and the
generated data would be better.
The topic of the suggestions came to my mind because I was working for a
whole year in a system that helped users to simply create RDF data and
Wikipedia Infoboxes, by making suggestions based on DBPedia RDF dataset.
(URL of prototype: http://sid01.cps.unizar.es).
Let me know what you think about these ideas. Also, please add me on the
slack channel.
Thank you for your time.
Regards,
Ismael Rodríguez.
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Dbpedia-gsoc mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbpedia-gsoc
