A Very Very Large Corpus Doesn’t Always Yield Reliable Estimates
41264Knowledge Triples
| Source Entity | Relationship | Target Entity |
|---|---|---|
|
One Billion Word Corpus
Material |
addresses |
Sparseness Problems
Task |
|
One Billion Word Corpus
Material |
estimates |
Unigram Probabilities
Metric |
|
Sparseness Problems
Task |
requires |
Statistical Modelling
Method |
|
Training Corpora
Material |
improves |
Unigram Probabilities
Metric |