Here, is a short overview of the contributions of the WSE group to evaluating Knowledge Graph Question Answering (KGQA) systems. All are located in their own GitHub repositories.
-
The following question answering datasets were created to establish a common benchmark for KGQA systems:
-
QALD-9-plus is the dataset for Knowledge Graph Question Answering based on well-known QALD-9. It enables to train and test KGQA systems over the DBpedia and Wikidata knowledge graphs using high-quality questions in 10 different languages:
-
English, German, Russian, French, Spanish, Armenian, Belarusian, Lithuanian, Bashkir, and Ukrainian.
-
-
QALD-10 is the dataset for the 10th Question Answering over Linked Data Challenge. It contains questions in 4 different language over the Wikidata knowledge graph:
-
English, Chinese, Russian, and German.
-
-
-
Our QADO Question Answering Dataset RDFizer is dedicated to the common challenge of working with different datasets. It converts the datasets into the RDF Turtle format and provides a unified interface to access the data.
-
The Question Answering Dataset Ontology (QADO) is used to describe the datasets in RDF.
-
The following datasets are supported:
-
QALD - Question Answering over Linked Data: supported versions: 5, 6, 8, 9, 9-plus, and 10
-
LC-QuAD - Largescale Complex Question Answering Dataset: supported versions: 1.0 and 2.0
-
RuBQ - A Russian Knowledge Base Question Answering and Machine Reading Comprehension Data Set: supported versions: 1.0 and 2.0
-
Mintaka - A complex, natural, and multilingual dataset for end-to-end question answering
-
ComplexWebQuestions - A dataset for answering complex questions that require reasoning over multiple web snippets
-
-
-
The Knowledge Graph Question Answering systems and their evaluation are collected at the KGQA leaderboard. It is dedicated to transparency and reproducibility of research results. The collection consists of more than 100 KGQA systems and their evaluation results using several knowledge graphs and their corresponding datasets:
-
DBpedia knowledge graph: LC-QuAD 1.0, LC-QuAD 2.0, rewordQALD-9, QALD-9-plus, QALD-9, QALD-8, QALD-7, QALD-6, QALD-5, QALD-4, QALD-3, QALD-2, QALD-1, SimpleDBpediaQA, MLPQ, Monument
-
Wikidata knowledge graph: KQA Pro, RuBQ 1.0, RuBQ 2.0, Compositional Wikidata Questions, TimeQuestions, CronQuestions, CLC-QuAD, SimpleQuestionsWikidata, MKQA, Mintaka
-
Freebase knowledge graph: Free917, WebQuestions, ComplexQuestions, GraphQuestions, WebQuestionSP, ComplexWebQuestions, 30M Factoid QA, PathQuestion, Compositional Freebase Questions, GrailQA, TempQuestions, freebaseQA
-
Other knowledge graphs: MetaQA, EventKQ/EventQA, WC2014QA
-