VADIS project

VAriable Detection, Interlinking and Summarization

VADIS Knowledge Graph (VADISKG)

The resulting data corpus containing links between scholarly articles, survey variables, and research datasets has been published as Knowledge Graph.

Data model - Dataset - License - Contact

Data model

The following figure illustrates a simplified version of the VADISKG data model.

VADISKG

Namespaces and prefixes

The data model of the VADISKG reuses classes and properties of the following vocabularies while introducing few own classes and properties for which no appropriate equivalent was existing.

Scientific resources in the KG

Resource name Class
Publication schema:ScholarlyArticle
Variable disco:Variable
Dataset schema:Dataset

Additional entities in the KG

Entity name Class
Person schema:Person
Keyword schema:DefinedTerm

Entities generated by VADIS

Entity name Description Class Subclass of
Extractive summary most comprehensive sentence of abstract sentences vadiskg:ExtractiveSummary vadiskg:Summary
Abstractive summary tldr extreme summary generated out of the abstract vadiskg:AbstractiveSummary vadiskg:Summary
Abstract sentence sentence of the abstract vadiskg:AbstractSentence vadiskg:Sentence
Variable sentence sentence containing a variable mention vadiskg:VariableSentence vadiskg:Sentence
Variable reference Detected variable in a variable sentence gesiskg:VariableReference gesiskg:Reference
Metadata name Description Property
Sentence confidence score Computed confidence score for the detected sentence containing variable mentions vadiskg:score
Method type Method which has been used to detect the variable sentences vadiskg:methodType
Common words Common words used in the variable sentence vadiskg:commonWords
Link reason Part of text which is the basis for the link detection gesiskg:linkReason
Variable similarity score Computed similarity score for the detected variable in a variable sentence gesiskg:linkScore
Link type Specifying that the link was automatically generated gesiskg:linkType

URI paths

Dataset

The VADISKG can be accessed and queried via its SPARQL endpoint. Additionally, the KG and the underlying ontology will be available for download soon.

SPARQL endpoint

The data within the VADIS Knowledge Graph can be explored using SPARQL queries at the following SPARQL endpoint: https://data.gesis.org/vadiskg/sparql

You can find some example SPARQL queries here:

Example query #1: List all resources from a particular type

The following query lists all publications which are included in the VADISKG. Result

SELECT ?id ?title
WHERE {?id ?p <https://schema.org/ScholarlyArticle>.
       ?id <https://schema.org/name> ?title.
}

To retrieve resources from a different type, change https://schema.org/ScholarlyArticle accordingly to, e.g., https://schema.org/Dataset.

Example query #2: List the tldr summaries for all publications

The following query retrieves all publications with their ID and title together with all generated tdlr summaries. Result

SELECT ?pub_id ?pub_title ?tldr
WHERE {?pub_id <https://schema.org/name> ?pub_title .
       ?pub_id <https://data.gesis.org/vadiskg/schema/abstractiveSummary> ?summary .
       ?summary <https://schema.org/text> ?tldr .
}
Example query #3: List all publications with their linked variables

The following query retrieves all publications (ID and title) with detected variables (ID and question texts). Result

SELECT ?pub_id ?pub_title ?var_id ?var_text
WHERE {?pub_id <https://schema.org/name> ?pub_title .
       ?pub_id <https://data.gesis.org/gesiskg/schema/variableReference> ?var_ref .
       ?var_ref <https://data.gesis.org/vadiskg/schema/detectedVariable> ?var_id .
       ?var_id <http://rdf-vocabulary.ddialliance.org/discovery#questionText> ?var_text
}

The following query adds all linked datasets to the previous query #3. Result

SELECT ?pub_id ?pub_title ?var_id ?var_text ?dataset_id
WHERE {?pub_id <https://schema.org/name> ?pub_title .
       ?pub_id <https://data.gesis.org/gesiskg/schema/variableReference> ?var_ref .
       ?var_ref <https://data.gesis.org/vadiskg/schema/detectedVariable> ?var_id .
       ?var_id <http://rdf-vocabulary.ddialliance.org/discovery#questionText> ?var_text .
       ?var_id <https://data.gesis.org/gesiskg/schema/dataset> ?dataset_id .
}

Download

The VADIS Knowledge Graph will be available for download as a full RDF dump as well as its underlying ontology from the following links:

License

The VADIS Knowledge Graph is available for access, download, and reuse under a Creative Commons Attribution 4.0 license since the license of some input sources is CC-BY as well.

Contact

Back to the VADIS homepage