Summary of the paper
| Title | Comparing Computational Models of Selectional Preferences - Second-order Co-Occurrence vs. Latent Semantic Clusters |
|---|---|
| Authors | Sabine Schulte im Walde |
| Abstract | This paper presents a comparison of three computational approaches toselectional preferences: (i) an intuitive distributional approach thatuses second-order co-occurrence of predicates and complementproperties; (ii) an EM-based clustering approach that models thestrengths of predicate--noun relationships by latent semantic clusters(Rooth et al., 1999); and (iii) an extension of the latent semanticclusters by incorporating the MDL principle into the EM training, thusexplicitly modelling the predicate--noun selectional preferences byWordNet classes (Schulte im Walde et al., 2008). Concerning thedistributional approach, we were interested not only in how well themodel describes selectional preferences, but moreover whichsecond-order properties are most salient. For example, a typicaldirect object of the verb 'drink' is usually fluid, might be hot orcold, can be bought, might be bottled, etc. The general question weask is: what characterises the predicate's restrictions to thesemantic realisation of its complements? Our second interest lies inthe actual comparison of the models: How does a very simpledistributional model compare to much more complex approaches, andwhich representation of selectional preferences is more appropriate,using (i) second-order properties, (ii) an implicit generalisation ofnouns (by clusters), or (iii) an explicit generalisation of nouns byWordNet classes within clusters? We describe various experiments onGerman data and two evaluations, and demonstrate that the simpledistributional model outperforms the more complex cluster-based modelsin most cases, but does itself not always beat the powerful frequencybaseline. |
| Language | Statistical and machine learning methods |
| Topics | Semantics, Lexicon, lexical database, Statistical and machine learning methods |
Full paper  |
Comparing Computational Models of Selectional Preferences - Second-order Co-Occurrence vs. Latent Semantic Clusters |
| Bibtex | @InProceedings{SCHULTEIMWALDE10.632, author = {Sabine Schulte im Walde}, title = {Comparing Computational Models of Selectional Preferences - Second-order Co-Occurrence vs. Latent Semantic Clusters}, booktitle = {Proceedings of the Seventh conference on International Language Resources and Evaluation (LREC'10)}, year = {2010}, month = {may}, date = {19-21}, address = {Valletta, Malta}, editor = {Nicoletta Calzolari (Conference Chair), Khalid Choukri, Bente Maegaard, Joseph Mariani, Jan Odjik, Stelios Piperidis, Mike Rosner, Daniel Tapias}, publisher = {European Language Resources Association (ELRA)}, isbn = {2-9517408-6-7}, language = {english} } |