Rel-122 Relatedness Norms

The Rel-122 relatedness norms are a set of mean human-assigned relatedness scores for 122 pairs of English-language nouns. This dataset was created as part of my work toward my dissertation at the University of Central Florida. Details of our experimental procedure for eliciting these data, as well as a discussion of the need for and utility of this dataset, can be found in our 2013 ACL short paper. An expanded discussion can be found in Chapter 6 of my dissertation. Questions and general discussion are warmly welcome and may be directed to me (Sean Szumlanski) via e-mail.

rel122-norms.pdf A table of the norms, excerpted from my dissertation.
rel122-norms.txt The same norms, but in plaintext (and with slightly higher precision for the means, but no standard deviations).

[1] S. Szumlanski, F. Gomez, and V. Sims. 2013. A New Set of Norms for Semantic Relatedness Measures. ACL '13. pp. 890—895.  [Download: PDF]

© 2010-2013 Sean Szumlanski. The artwork on this page is derived from works by
Courtney Martin, which are modified and used here with permission.