PATTY Data Set Release 2012-05-21 contact: nnakasho@mpi-inf.mpg.de This data accompanies the publication below. Please cite this publication if you use the data. Ndapandula Nakashole, Gerhard Weikum, Fabian Suchanek PATTY: A Taxonomy of Relational Patterns with Semantic Types. In Proceedings of the International Conference on Empirical Methods in Natural Language Processing (EMNLP 2012). This data is downloadable at: http://www.mpi-inf.mpg.de/yago-naga/patty/ An online demo is at: https://d5gate.ag5.mpi-sb.mpg.de/pattyweb/ 1) File Format The schema of each file is the first line in the file, columns are tab-separated-values (TSV) . 2) Pattern Synsets with Type-Signatures - file: wikipedia-patterns.txt Contains all pattern synsets derived from Wikipedia (June 2011, Version) The Wikipedia version contains ~ 350, 000 pattern sysnets - file: nyt-patterns.txt Contains all pattern sunsets derived from the New York Times archive (1987-2007) The New York Times version contains ~ 80, 000 pattern sysnets 3) Pattern Subsumptions - file: wikipedia-subsumptions.txt Contains subsumptions derived from Wikipedia - file: nyt-subsumptions.txt Contains subsumptions derived from the New York Times archive 4) Pattern Instances - file: wikipedia-instances.txt Contains entity-pairs and the patterns they occur with in Wikipedia - file: nyt-instances.txt Contains entity-pairs and the patterns they occur with in the New York Times archive 5) Relation Paraphrases -file: dbpedia-relation-paraphrases.txt Contains paraphrases of DBpedia relations using PATTY patterns -file: yago-relation-paraphrases.txt Contains paraphrases of YAGO relations using PATTY patterns 6) Evaluation -file: patty-evaluation-coverage-music.txt Contains coverage evaluation of music relations for P: Patty, Y: Yago , Db: DBpedia, Fb: Freebase, N: NELL -file: wikipedia_top100.txt: Contains evaluations of top 100 pattern sysnets derived from Wikipedia -file: wikipedia_random100.txt: Contains evaluations of 100 pattern sysnets randomly selected among Wikipedia-derived patterns. -file: nyt_top100.txt: Contains evaluations of top 100 pattern sysnets derived from the New York Times archive. -file: wikipedia_random100.txt: Contains evaluations of 100 pattern sysnets randomly selected among New York Times archive-derived patterns.