Multiple Knowledge GraphDB (MKGDB)

Farid Yusifli(1), Stefano Faralli(2), Paola Velardi(1) (1) University of Rome Sapienza (2) University of Rome Unitelma Sapienza

MKGDB is a large-scale graph database created as a combination of multiple taxonomy backbones extracted from 5 existing knowledge graphs, namely: ConceptNet, DBpedia, WebIsAGraph, WordNet and the Wikipedia category hierarchy. MKGDB, thanks the versatility of the Neo4j graph database manager technology, is intended to favour and help the development of open-domain natural language processing applications relying on knowledge bases, such as information extraction, hypernymy discovery, topic clustering, and others. Our resource consists of a large hypernymy graph which counts more than 37 millions nodes and more than 81 millions hypernymy relations.

Installing the Multiple Knowledge Graph Database

MKGDB is avaliable in two format:

• The compressed zip copy of the graph database is avaliable at: https://drive.google.com/open?id=1aREGClCuh8HRlaYya_-6Yuu2Itap4YbS) file or download;
• A dump of the graph databse is avaliable at: https://drive.google.com/open?id=1Ytvzb0S1SXy3HxlitdXAnuafHqkrbJXs.

If you chose to download the copy of the database instance, you will need to modify the neo4j.conf file to target the download graph database and restart the neo4j server. Instead, if you want to use to load the dump version, you can follow the instructions provided at: https://neo4j.com/docs/operations-manual/current/tools/dump-load/.

Related Resources

In this section we provide the link to download the dataset we used to generate the MKGDB:

• [Wikipedia category](http://downloads.dbpedia.org/3.9/en/skos_categories_en.nt.bz2);
• [ConceptNet](https://s3.amazonaws.com/conceptnet/downloads/2019/edges/conceptnet-assertions-5.7.0.csv.gz);
• [DBpedia Instances](http://downloads.dbpedia.org/3.9/en/instance_types_en.nt.bz2);
• [WebIsAGraph](https://drive.google.com/open?id=1iNe8BcUu5Ineu3IpmjQMn2e_f3MImOLI);
• [Wordnet](https://wordnet.princeton.edu/download/current-version);
• [DBpedia Ontology](https://drive.google.com/open?id=1XwVkT40DvutyvXgyhOmUUMaW1rUJVFVc).

Source code

We release also the java project we developed to the generation of the graph database starting from the above mentioned dataset. To correctly run the generation of the resource from the above dataset it is required to edit the config.properties file.

Contacts

Farid Yusifli, Stefano Faralli, Paola Velardi