OnAGUI is a software to help align vocabularies, thesauri, ontologies. "Aligning" means creating a mapping between the items of the 2 vocabularies.

OnAGUI can :

• read both vocabularies and display them in trees (one on the left, one the right)
• run automated algorithms to match vocabulary items based on their label proximity
• display the proposed alignment in a table in the center, so that we can review, adjust, delete or create more correspondances

To install OnAGUI :

1. you first need to install Java
2. download the latest OnAGUI release
3. unzip it
4. run either "run.bat" on Windows or "run.sh" on Linux

To get started :

1. You need a vocabulary in SKOS. Many tools can generate this format; if you start from plain CSV or Excel files, you can use the SKOS-Play Excel converter to generate SKOS files;
2. You need another vocabulary in SKOS, obviously the one with which you need to get aligned; you could start by looking at the UNESCO Thesaurus or the Eurovoc thesaurus, both available in SKOS, if you want to play with something;
3. Download, install and run OnAGUI;
4. Load your vocabulary from the menu "File > Load vocabulary 1 > Load in SKOS"; you can browse it in the left tree;
5. Load the vocabulary you need to align to with the menu "File > Load vocabulary 2 > Load in SKOS"; you can browse it in the right tree;
6. Now run a linguistic alignment from the menu "Alignement > I-Sub distance", and here select :
   • the part of the vocabularies you want to align (whole vocabulary or only a sub-branch);
   • the language you want to use for automatic alignment;
   • a threshold score for the alignement; 0.9 is proposed by default, you can lower this a bit to 0.8; the lower the value, the more mappings you will get, but also the more noise !
7. Mappings are shown in the table in the center of the screen; you can edit the table by :
   • editing the types of mappings (exactMatch are produced by default);
   • deleting a mapping line in the table;
   • adding a new mapping line in the table (select concept on both trees, select mapping type, and click on "Add" button)
8. Export your mapping from the menu "File > Export Alignement > Export in SKOS" to generate SKOS mapping properties, or "Export in RDF" to generate an RDF EDOAL file, more expressive;

• I-Sub distance works best. The algorithm is in this java class, from this paper
• Levensthein distance is a classic
• Exact match on the labels

The algorithms do not take into account upper/lower case or accented characters ("Sémantique" will be matched exactly to "SEMANTIQUE").

OnAGUI is available in French and English. You can contribute new translations by translating these message files.

• SKOS thesauri/vocabularies
• OWL ontologies
• arbitrary RDF with common labelling properties; the supported label properties are :
  • rdfs:label
  • skos:prefLabel
  • dcterms:title
  • dc:title
  • schema:name
  • foaf:name
  • and as alternative label properties :
    • schema:alternateName
    • skos:altLabel
    • foaf:nick
    • dcterms:alternative
  • for more details see the class RDFModelContainer

• EDOAL is the most comprehensive model
• SKOS alignment
• CSV

Laurent Mazuel and Jean Charlet at INSERM, in France.

File an issue here !