claimskg_generator

The data lifting module for ClaimsKG that creates the RDF/LOD dataset instantiation from the ClaimsKG model (illustrated hereafter).

Installation & Requirements

This program requires Python 3.x to run.

To install the dependencies you may use: pip3 install -r requirements.txt

Command-line usage

• For usage information you may use

    python3 export.py -h

• The options are the following:

  • --input [file] Indicated the location of the zip file generated by the fake new extractor (mandatory)
  • --output [file] Specifies the output file for the model (default: out.ttl)
  • --format [format] Specifies the format of the output serialization. You may use any of the supported formats in the rdflib package (xml', 'n3', 'turtle', 'nt', 'pretty-xml', 'trix', 'trig' and 'nquads'; default: turtle)
  • --model-uri The base URI of the model (by default http://data.gesis.org/claimskg/public/)
  • --resolve Specifies whether to resolve the Wikipedia page identifiers for TagMe annotations to DBPedia URIs. If this option is activated, the resolution is performed through SPARQL queries to the official DBPedia endpoint, which requires you to have an active Internet connection. Additionally, you will need a running instance of redis-server as the results of the queries are cached to prevent unnecessary queries from being performed. If resolve is not supplied, all entities will have URIs of the form tagme://wikpediaPageID.
  • --threshold [float_value] If --resolve is present, specifies the cutoff confidence threshold to include a TagMe annotations as a mention. The TagMe documentation recommends a value between 0.1 and 0.3 (default 0.3)
  • --include-body If --include-body is supplied, the body of the claim review is included in the schema:ClaimReview instances through the schema:reviewBody property.

  Claim Matching Evaluation

  In the comtext of the claim matching approach, we have produced a annotated dataset for evaluation purposes. The dataset contains 318 matching claims categorized in several types of matches:

  • E: Exact match
  • E*: Same claim but different claimee
  • ST: Same topic, meaning that two claims are about the same occurence or event

  The first columnt contains the type of match, the second column, the URI of the first claim, the third column, the URI of the second claim.

  The gold file can be downloaded here: https://drive.google.com/open?id=1evf67t_p0LqF5ZvNiL-geDCrEBlppZVF

  The URIs correspond to that of the following dataset: http://andon.tchechmedjiev.eu/files/claimskg_20_12_2018.ttl.gz

  Latest releases of core data from en.wikipedia.org (ttl file): https://databus.dbpedia.org/dbpedia/collections/latest-core https://downloads.dbpedia.org/repo/dbpedia/generic/categories/2021.03.01/categories_lang=en_skos.ttl.bz2