Analogical Prediction of English Verb Past Tense (York)

Application domain Natural Language Processing
Source: Dataset of English verbs
Dataset size: $\sim$500 entries
Data format: Prolog, pairs (present,past)
Systems used: Progol4.4 - Analogical Prediction
References: (Muggletona and Bain 1999)
Pointers: ftp://ftp.cs.utexas.edu/pub/mooney/nl-ilp-data/alphabetic-past-data
ftp://ftp.cs.york.ac.uk/pub/ML_GROUP/progol4.4/

The best prediction of English Past Tense so far has been obtained with Progol4.4 using Analogical Prediction (AP) (Muggleton and Bain 1999). Analogical Prediction (AP) takes Background Knowledge B and training examples E and then for each test example <x,y> forms an hypothesis Hx from B,E,x. Evaluation of AP is based on estimating the probability that Hx(x) = y for a randomly chosen <x,y> AP has been implemented within Progol4.4. Experiments show that on English past tense data AP has significantly higher predictive accuracy on this data than both previously reported results and Progol in inductive mode. AP has advantages for domains in which a large proportion of the examples must be treated as exceptions with respect to the hypothesis vocabulary. The relationship of AP to analogy and instance-based learning is discussed.

Bibliography

  1. S. Muggleton and M. Bain. Analogical prediction. In Proc. of the 9th International Workshop on Inductive Logic Programming (ILP-99), Berlin, 1999. Springer-Verlag.


back to index