Resource population using Wikidata and Wiktionary
There are an increasing number of lexical resources available online which are machine-friendly and can be accessed by linked data techniques.
In this repository, we provide a program which collects linguistic information for a given word, in raw text or triples, and converts the collected data to Lemon-OntoLex ontology. The current resources being used are Wikidata and Wiktionary. However, any resource which can be accessed via a SPARQL endpoint can be used in this program.
The following image shows information for the Italian word “acroterio” in Wikidata.
How does it work?
Given a list of words in text or in RDF, this program collects all the relevant information in Wikidata and Wiktionary.
Change categories
In order to collect data in a specific domain, you can add your desired category to the subjects. Note that this may affect the performance of the queries as it will take longer to check all the domains.
subjects = {"architecture":"Q12271", "archeology": "Q10855079", "law" : "Q7748", "legal science" : "Q382995", "social issue" : "Q1920219", "jurisprudence" : "Q4932206", "rule" : "Q1151067", "Economy" : "Q159810", "Economics" : "Q8134", "labour law" : "Q628967", "human action" : "Q451967", "legal concept" : "Q2135465"}
Include external ontologies
If you would like to link your Lemon-Ontolex data to an external ontology, you can use the ontolex:reference property. One of the example given datasets provide the URIs in a tab-separated format.
Conversion to Lemon-OntoLex
The conversion is done using predefined templates for each part of each part of the data, such as, entry description, vatrans, linguistic information (pos, gender etc.). The following figure illustrated the Lemon-OntoLex core model.
Limitations
Please note that the current version of the program only takes into account very specific information and is not creating a fine-grained Lemon-Ontolex representation. For instance, only “OntoLex:LexicalEntry” for representing entries are used (no components).
Our codes are available at https://github.com/sinaahmadi/sparql4respop
Last updated on 21 May 2019.