Elamur Research⌕
☰
home / papers / 2016 / Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags

Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags

Paper · 2016 · Niket Tandon, Charles Hariman, Jacopo Urbani et al.

Summary

Machines lack everyday knowledge about how things are composed: which objects have which parts, who belongs to which group, and what things are made of. Earlier resources such as WordNet and ConceptNet are small and hand-built. Automatic collections are noisy, lump all part-whole links into one relation, use ambiguous words instead of senses, and say nothing about whether a part can be seen or how many there are.

The authors built PWKB, a knowledge base learned automatically from the Wikipedia full text and Google n-grams. Bootstrapped pattern extraction uses a new scoring scheme with weighted seeds that rewards patterns which separate the relation types. Arguments, including multi-word phrases, are mapped to WordNet senses. Logical rules then remove inconsistent candidates and derive new ones. Flickr image tags indicate whether a part is visible, and multilingual Google Books n-grams supply part counts.

The result has about 6.75 million assertions, with 89% accuracy in sampled manual checks. That is far larger than prior resources at comparable or better quality. It is useful for search, question answering and vision.

Key points

  • PWKB separates three WordNet-based relations, each with domain and range type restrictions: physicalPartOf, memberOf and substanceOf.
  • Pattern extraction starts from 1,200 WordNet seed pairs. Disambiguation extends the IMS tool with OpenNLP noun-phrase chunking, so that phrase heads get resolved to synsets.
  • Visibility comes from tags of 100 million Flickr images (YFCC100M). Cardinality comes from Google Books n-grams, with German and Italian grammar used for cases that are hard to resolve in English.
  • The final KB holds about 6.75M assertions at 89% sampled accuracy. ConceptNet, by comparison, has only 1,086 part-whole instances.
  • In zero-shot image classification over 200 unseen categories, using 811 part classifiers, PWKB raised top-5 accuracy from 23.8% to 25.5%.

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

 Institutions

🎓University of California, BerkeleyBerkeley · United States243 works · 251,774 citations🔬Max Planck Institute for InformaticsSaarbrücken · Germany22 works · 4,687 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksArts and Humanities · 17,959 worksLanguage and Linguistics · 17,789 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 1,948 works

Cite

Cite as Niket Tandon, Charles Hariman, Jacopo Urbani, Anna Rohrbach, Marcus Rohrbach and Gerhard Weikum. 2016. Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags. AAAI-16.

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

@inproceedings{tandon2016commonsense,
  title = {{Commonsense in Parts: Mining Part-Whole Relations from the Web and Image Tags}},
  author = {Niket Tandon, Charles Hariman, Jacopo Urbani, Anna Rohrbach, Marcus Rohrbach and Gerhard Weikum},
  booktitle = {AAAI-16},
  year = {2016},
  doi = {10.1609/aaai.v30i1.9992},
  url = {https://doi.org/10.1609/aaai.v30i1.9992}
}

References

32 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Berland, M., and Charniak, E. 1999. Finding parts in very large corpora. In ACL.
  2. Bouma, G. 2009. Normalized (pointwise) mutual information in collocation extraction. Proceedings of GSCL 31–40. find
  3. Brants, T., and Franz, A. 2006. tWeb 1T 5-gram Version 1u. find
  4. Brown, L. D.; Cai, T. T.; and DasGupta, A. 2001. Interval estimation for a binomial proportion. find
  5. Statistical Science. Carlson, A.; Betteridge, J.; Kisiel, B.; Settles, B.; Jr., E. H.; and Mitchell, T. 2010. Toward an architecture for never-ending language learning. In AAAI, 1306–1313. AAAI Press. find
  6. Chen, X.; Shrivastava, A.; and Gupta, A. 2013. Neil: Extracting visual knowledge from web data. In CVPR. find
  7. Chen, X.; Shrivastava, A.; and Gupta, A. 2014. Enriching visual knowledge bases via object discovery and segmentation. In CVPR. find
  8. Fellbaum, C., ed. 1998. WordNet: An Electronic Lexical Database. The MIT Press. find
  9. Girju, R.; Badulescu, A.; and Moldovan, D. 2003. Learning semantic constraints for the automatic discovery of part-whole relations. In NAACL.
  10. Girju, R.; Badulescu, A.; and Moldovan, D. 2006. Automatic discovery of part-whole relations. Computational Linguistics. find
  11. Havasi, C.; Speer, R.; and Alonso, J. 2007. ConceptNet 3: a flexible, multilingual semantic network for common sense knowledge. In RANLP. find
  12. Hoffman, J.; Guadarrama, S.; Tzeng, E.; Donahue, J.; Girshick, R.; Darrell, T.; and Saenko, K. 2014. LSDA: Large scale detection through adaptation. In NIPS. find
  13. Ittoo, A., and Bouma, G. 2010. On learning subtypes of the part-whole relation: do not mix your seeds. In ACL. find
  14. Ittoo, A., and Bouma, G. 2013. Minimally-supervised extraction of domain-specific part–whole relations using wikipedia as knowledge-base. Data & Knowledge Engineering 85:57–79. find
  15. Keet, C. M., and Artale, A. 2008. Representing and Reasoning over a Taxonomy of Part-Whole Relations. Applied Ontology. find
  16. Lenat, D. B. 1995. CYC: A large-scale investment in knowledge infrastructure. Commun. ACM. find
  17. Ling, X.; Clark, P.; and Weld, D. S. 2013. Extracting meronyms for a biology knowledge base using distant supervision. In Proceedings of the 2013 workshop on Automated knowledge base construction, 7–12. find
  18. ACM. Matuszek, C.; Witbrock, M.; Kahlert, R.; Cabral, J.; Schneider, D.; Shah, P.; and Lenat, D. 2005. Searching for common sense: Populating Cyc from the Web. In Proc. find
  19. McCarthy, D. 2001. Lexical acquisition at the syntax-semantics interface: diathesis alternations, subcategorization frames and selectional preferences. PhD. thesis. find
  20. Mitchell, T.; Cohen, W.; Hruschka, E.; Talukdar, P.; Betteridge, J.; Carlson, A.; Dalvi, B.; Gardner, M.; Kisiel, B.; Krishnamurthy, J.; Lao, N.; Mazaitis, K.; Mohamed, T.; Nakashole, N.; Platanios, E.; Ritter, A.; Samadi, M.; Settles, B.; Wang, R.; Wijaya, D.; Gupta, A.; Chen, X.; Saparov, A.; Grea… find
  21. Pantel, P., and Pennacchiotti, M. 2006. Espresso: Leveraging generic patterns for automatically harvesting semantic relations. In ACL.
  22. Ravichandran, D., and Hovy, E. 2002. Learning surface text patterns for a question answering system. 41–47. ACL.
  23. Rohrbach, M.; Stark, M.; and Schiele, B. 2011. Evaluating Knowledge Transfer and Zero-Shot Learning in a Large-Scale Setting. In CVPR. find
  24. Ruiz-Casado, M.; Alfonseca, E.; and Castells, P. 2007. Au-tomatising the learning of lexical patterns: An application to the enrichment of wordnet by extracting semantic relationships from wikipedia. Data & Knowledge Engineering 61(3):484–499. find
  25. Shamma, D. 2014. One hundred million Creative Commons Flickr images for research. http://labs.yahoo.com/news/yfcc100m/. ↗
  26. Shaoul, C. 2010. The westbury lab wikipedia corpus. Edmonton, AB: University of Alberta. Smith, B.; Ceusters, W.; Klagges, B.; Köhler, J.; Kumar, A.; Lomax, J.; Mungall, C.; Neuhaus, F.; Rector, A. L.; and Rosse, C. 2005. Relations in biomedical ontologies. Genome Biology 6(R46). find
  27. Speer, R., and Havasi, C. 2012. Representing general relational knowledge in conceptnet 5. In LREC. find
  28. Tandon, N.; de Melo, G.; Suchanek, F. M.; and Weikum, G. 2014. Webchild: harvesting and organizing commonsense knowledge from the web. In WSDM. find
  29. Tandon, N.; de Melo, G.; and Weikum, G. 2011. Deriving a web-scale common sense fact database. In AAAI.
  30. Varzi, A. C. 2014. Mereology. In Zalta, E. N., ed., Stanford Encyclopedia of Philosophy. find
  31. Winston, M. E.; Chaffin, R.; and Herrmann, D. 1987. A Taxonomy of Part-Whole Relations. Cognitive Science. find
  32. Zhong, Z., and Ng, H. T. 2010. It makes sense: A wide-coverage word sense disambiguation system for free text. In ACL. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/tandon-2016-commonsense-parts-mining-part-whole/