Elamur Research⌕
☰
home / papers / 2012 / PATTY: A Taxonomy of Relational Patterns with Semantic Types

PATTY: A Taxonomy of Relational Patterns with Semantic Types

Summary

Large knowledge bases store relations as abstract labels such as bornIn, but they do not record the many natural-language phrasings that express those relations. Existing pattern collections such as ReVerb also leave their phrases unorganized: they are not grouped into synonyms and not arranged hierarchically. PATTY addresses this by building a WordNet-like resource for binary relational phrases.

The pipeline works as follows:

  • Wikipedia sentences are parsed with the Stanford Parser.
  • Entity mentions are linked to YAGO2.
  • For each entity pair, the shortest dependency path between them becomes a candidate pattern.
  • Each pattern is then abstracted into a SOL pattern, which combines syntactic placeholders, ontological type signatures for the arguments and lexical words.
  • Frequent itemset mining groups patterns into synsets and infers subsumptions from overlapping entity-pair support.

The resulting taxonomy has 350,569 pattern synsets with about 84.7% sampled accuracy and 8,162 subsumption links with about 75% precision. This matters because typed, organized paraphrases support information extraction, knowledge-base population and question answering.

Key points

  • Pattern extraction: Stanford Parser dependency paths are used, keeping only shortest paths that start with subject-like dependencies (nsubj, rcmod, partmod). These paths are expanded with adverbial and adjectival modifiers.

  • Entity linking: Entities are recognized and typed with the YAGO2 knowledge base and disambiguated with a simple context-similarity prior. The reported linking accuracy is well above 80%.

  • Pattern-language comparison (top-100 precision, Table 5):

    VariantPatternsPrecision
    ReVerb-style5,9960.96
    PATTY without types184,6290.74
    Full PATTY350,5690.95

    Full PATTY also has 8,162 subsumptions at 0.83 precision and 3.89M facts at 0.88 precision.

  • Relation paraphrasing: PATTY covered 225 DBpedia relations with 127,811 patterns and 25 YAGO2 relations with 43,124 patterns. The average paraphrase precision was 0.76.

  • Availability: The resource can be browsed and downloaded at www.mpi-inf.mpg.de/yago-naga/patty/.

Interesting ideas

  • I1. Typing pattern arguments with a knowledge base. Replacing entities with semantic classes such as ⟨singer⟩ or ⟨song⟩ separates ambiguous phrases by context, and dropping types caused a large precision loss. The same trick could help any relation-mining effort that has access to a typed entity dictionary.
  • I2. Two complementary ways to generalize. Syntactic generalization swaps words for POS placeholders, for example collapsing "her"/"his" into [prp]. Semantic generalization moves argument types up the class hierarchy. Treating both together yields compact, reusable patterns from noisy text.
  • I3. Inferring subsumption from instance support. One pattern is taken to imply another when its entity-pair set is largely contained in the other's. This cheap, data-driven route to a hierarchy could transfer to clustering paraphrases, verbs or query templates.
  • I4. Framing mining as frequent itemset mining. Treating pattern generation this way keeps the exponential space of candidate patterns tractable at Web scale. This is a useful engineering choice for any combinatorial pattern-discovery task.
  • I5. Extrinsic evaluation via paraphrasing. Measuring how well mined phrases express existing DBpedia and YAGO2 relations gives a practical usefulness check beyond intrinsic sampling. It also exposes failure cases, such as overly general phrases offered for specific relations.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Natural language processing

Cited by

84 papers of our library cite it (of the 140884 whose bibliographies we read); 5 of them have a page here, 79 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{nakashole2012patty,
  title = {{PATTY: A Taxonomy of Relational Patterns with Semantic Types}},
  author = {Nakashole, Ndapandula and Weikum, Gerhard and Suchanek, Fabian},
  booktitle = {Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning},
  year = {2012},
  url = {https://aclanthology.org/D12-1104/}
}

References

40 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Ion Androutsopoulos, Prodromos Malakasiotis: A Survey of Paraphrasing and Textual Entailment Methods. Journal of Artificial Intelligence Research 38: 135– 187, 2010 find our card
  2. Rakesh Agrawal, Tomasz Imielinski, Arun N. Swami: Mining Association Rules between Sets of Items in Large Databases. SIGMOD Conference 1993 find
  3. Enrique Alfonseca, Marius Pasca, Enrique Robledo-Arnuncio: Acquisition of instance attributes via labeled and related instances. SIGIR 2010 find
  4. Sören Auer, Christian Bizer, Georgi Kobilarov, Jens Lehmann, Richard Cyganiak, Zachary G. Ives: DBpedia: A Nucleus for a Web of Open Data. ISWC 2007, data at http://dbpedia.org ↗
  5. Michele Banko, Michael J. Cafarella, Stephen Soderland, Matthew Broadhead, Oren Etzioni: Open Information Extraction from the Web. IJCAI 2007 find our card
  6. Jonathan Berant, Ido Dagan, Jacob Goldberger: Global Learning of Typed Entailment Rules. ACL 2011 find
  7. Kurt D. Bollacker, Colin Evans, Praveen Paritosh, Tim Sturge, Jamie Taylor: Freebase: a collaboratively created graph database for structuring human knowledge. SIGMOD Conference 2008, data at http://freebase.com ↗
  8. Lawrence D. Brown, T.Tony Cai, Anirban Dasgupta: Interval Estimation for a Binomial Proportion. Statistical Science 16: 101–133, 2001 find
  9. Andrew Carlson, Justin Betteridge, Bryan Kisiel, Burr Settles, Estevam R. Hruschka Jr., Tom M. Mitchell. Toward an Architecture for Never-Ending Language Learning. AAAI 2010, data at http://rtw.ml.cmu.edu/rtw/ Timothy Chklovski, Patrick Pantel: VerbOcean: Mining the Web for Fine-Grained Semantic Ve… ↗
  10. EMNLP 2004; data available at http://demo.patrickpantel.com/demos/verbocean/ Anthony Fader, Stephen Soderland, Oren Etzioni: Identifying Relations for Open Information Extraction. ↗
  11. EMNLP 2011, data at http://reverb.cs.washington.edu Christiane Fellbaum (Editor): WordNet: An Electronic Lexical Database. ↗
  12. Jiawei Han, Jian Pei , Yiwen Yin : Mining frequent patterns without candidate generation. SIGMOD 2000. find
  13. Chikara Hashimoto, Kentaro Torisawa, Kow Kuroda, Stijn De Saeger, Masaki Murata, Jun’ichi Kazama: Large-Scale Verb Entailment Acquisition from the Web. EMNLP 2009 find
  14. Catherine Havasi, Robert Speer, and Jason Alonso. ConceptNet 3: a Flexible, Multilingual Semantic Network for Common Sense Knowledge, RANLP 2007; data available at http://conceptnet5.media.mit.edu/ Iris Hendrickx, Su Nam Kim, Zornitsa Kozareva, Preslav Nakov, Diarmuid O Seaghdha, Sebastian Pado, Ma… ↗ our card
  15. Johannes Hoffart, Fabian Suchanek, Klaus Berberich, Edwin Lewis-Kelham, Gerard de Melo, Gerhard Weikum: YAGO2: Exploring and Querying World Knowledge in Time, Space, Context, and Many Languages. WWW 2011, data at http://yago-knowledge.org ↗
  16. Raphael Hoffmann, Congle Zhang, Daniel S. Weld: Learning 5000 Relational Extractors. ACL 2010 find
  17. Vigo Kann: On the approximability of NP-complete optimization problems. PhD thesis, Department of Numerical Analysis and Computing Science, Royal Institute of Technology, Stockholm. 1992. find
  18. Karin Kipper, Anna Korhonen, Neville Ryant, Martha Palmer, A Large-scale Classification of English Verbs, Language Resources and Evaluation Journal, 42(1): 21-40, 2008, data available at http://verbs.colorado.edu/˜mpalmer/projects/verbnet/downloads.html ↗
  19. Zornitsa Kozareva, Eduard H. Hovy: Learning Arguments and Supertypes of Semantic Relations Using Recursive Patterns. ACL 2010 find
  20. Girija Limaye, Sunita Sarawagi, Soumen Chakrabarti: Annotating and Searching Web Tables Using Entities, Types and Relationships. PVLDB 3(1): 1338-1347 (2010) find
  21. Dekang Lin, Patrick Pantel: DIRT: discovery of inference rules from text. KDD 2001 find
  22. Marie-Catherine de Marneffe, Bill MacCartney and Christopher D. Manning. Generating Typed Dependency Parses from Phrase Structure Parses. LREC 2006. find our card
  23. Thahir Mohamed, Estevam R. Hruschka Jr., Tom M. Mitchell: Discovering Relations between Noun Categories. EMNLP 2011 find
  24. Ndapandula Nakashole, Martin Theobald, Gerhard Weikum: Scalable Knowledge Harvesting with High Precision and High Recall. WSDM 2011 find
  25. Preslav Nakov, Marti A. Hearst: Solving Relational Similarity Problems Using the Web as a Corpus. ACL 2008 find
  26. Vivi Nastase, Michael Strube, Benjamin Boerschinger, Cäcilia Zirn, Anas Elghafari: WikiNet: A Very Large Scale Multi-Lingual Concept Network. LREC 2010, data at http://www.h-its.org/english/research/nlp/download/wikinet.php ↗
  27. Roberto Navigli, Simone Paolo Ponzetto: BabelNet: Building a Very Large Multilingual Semantic Network. ACL 2010 data at http://lcl.uniroma1.it/babelnet/ Marius Pasca, Benjamin Van Durme: What You Seek Is What You Get: Extraction of Class Attributes from Query Logs. ↗
  28. Marius Pasca, Benjamin Van Durme: Weakly-Supervised Acquisition of Open-Domain Classes and Class Attributes from Web Documents and Query Logs. ACL 2008 find
  29. Andrew Philpot, Eduard Hovy, Patrick Pantel: The Omega Ontology, in: Ontology and the Lexicon, Cambridge University Press, 2008, data at http://omega.isi.edu/ Simone Paolo Ponzetto, Michael Strube: Deriving a Large-Scale Taxonomy from Wikipedia. ↗
  30. AAAI 2007, data at http://www.h-its.org/english/research/nlp/download/wikitaxonomy. php Joseph Reisinger, Marius Pasca: Latent Variable Models of Concept-Attribute Attachment. ACL/AFNLP 2009 ↗
  31. Ramakrishnan Srikant, Rakesh Agrawal: Mining Sequential Patterns: Generalizations and Performance Improvements. EDBT 1996 find
  32. Fabian M. Suchanek, Gjergji Kasneci, Gerhard Weikum: YAGO: a Core of Semantic Knowledge. WWW 2007 find
  33. Fabian M. Suchanek, Mauro Sozio, Gerhard Weikum: SOFIE: a self-organizing framework for information extraction. WWW 2009 find
  34. Lin Sun, Anna Korhonen: Hierarchical Verb Clustering Using Graph Factorization. EMNLP 2011 find
  35. Petros Venetis, Alon Y. Halevy, Jayant Madhavan, Marius Pasca, Warren Shen, Fei Wu, Gengxin Miao, Chung Wu: Recovering Semantics of Tables on the Web. PVLDB 4(9): 528-538, 2011 find
  36. Tom White: Hadoop: The Definitive Guide, 2nd Edition. O’Reilly, 2010. find
  37. Fei Wu, Daniel S. Weld: Automatically refining the wikipedia infobox ontology. WWW 2008 find
  38. Wentao Wu, Hongsong Li, Haixun Wang, Kenny Q. Zhu: Towards a Probabilistic Taxonomy of Many Concepts. Technical Report MSR-TR-2011-25, Microsoft Research, 2011 find
  39. Limin Yao, Aria Haghighi, Sebastian Riedel, Andrew McCallum: Structured Relation Discovery using Generative Models. EMNLP 2011 find
  40. Jun Zhu, Zaiqing Nie, Xiaojiang Liu, Bo Zhang, Ji-Rong Wen: StatSnowball: a statistical approach to extracting entity relationships. WWW 2009 find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/nakashole-2012-patty-taxonomy-relational-patterns-semantic/