Next: Linear interpolation Up: Language model smoothing: modelling Previous: Problem formulation

## Linear discounting and backing-off

We start with linear discounting in the so-called backing-off model [Katz (1987), Jelinek (1991)]. This model is exceptional in that it allows virtually closed-form solutions in conjunction with leaving-one-out . Thus we can discuss all the important formulae of the approach without recourse to numeric calculations. The model is defined by the equation:

It is easy to verify that the probabilities p(w|h) sum up to unity. In particular, is the total probability mass to be distributed over the unseen words for a given history h:

Here we have two types of parameter sets to be estimated:

• the discounting parameters for each history h;
• the backing-off distribution for a generalised history . Note that for each history h the generalised history must be well defined in order to have a backing-off distribution .

The unknown parameters in the discounting models will be determined with the so-called leaving-one-out method.   leaving-one-out This method can be obtained as an extension of the cross-validation  or held-out method [Duda & Hart (1973), Efron & Tibshirani (1993)], where the training text  is split into two parts: a ``retained" part for estimating the relative frequencies and a ``held-out" part for estimating the optimal interpolation  parameters. Here, the training text is divided into (N-1) samples as the ``retained" part and only 1 sample as the ``held-out" part; this process is repeated N times so that all N partitions with 1 ``held-out" sample are considered. The basic advantage of this approach is that all samples are used both in the ``retained" part and in the ``held-out" part and thus very efficient exploitation of the training text  is achieved. In particular, singleton events are reduced from 1 to 0 so that the effect of not seeing events is simulated.

The application of leaving-one-out  to the problem under consideration is as follows. By doing some elementary manipulations as shown in the appendix, we can decompose the log-likelihood function into two parts, one of which depends only on and the other depends only on :

The dependent part is:

Taking the partial derivatives with respect to and equating them to zero, we obtain the closed-form solution:

The same value is obtained when we compute the probability mass of unseen words in the training data  for a given history h:

The same value is obtained when using the Turing-Good formula [Good (1953), Nadas (1985), Katz (1987), Ney & Essen (1993)]. This formula is extremely useful when we want to estimate the total probability for events that were not seen in the training data. This formula tells us simply to take the number of singleton events, i.e. , divide it by the total number of events, i.e. , and use this fraction as estimate for the probability of new events. This rule can be applied at various levels such as single words (unigrams ), word bigrams  and word trigrams . For example, to check the coverage    of the vocabulary,  we can use this formula to estimate the probability with which we will encounter words that are not part of the vocabulary, i.e. that we have not seen in the training data .

For the backing-off distribution, the standard method is to use the relative frequencies as estimates:

However, as shown in [Generet et al. (1995)], the leaving-one-out  method can be applied to this task, too. In the appendix, we show how an approximate solution is obtained for this case. The resulting estimation formula is:

where is defined as:

For convenience, we have chosen the normalisation . This type of backing-off distribution will be referred to as singleton backing-off distribution or singleton distribution for short. Considering bigram  modelling as a specific example, we see that the singleton backing-off distribution takes into account such word bigrams as ``has been'', ``United States'', ``bona fide'' etc., where the second word is strongly correlated with the immediate predecessor word. In these cases, the conventional unigram  distribution tends to overestimate the corresponding probability in the backing-off distribution.

Next: Linear interpolation Up: Language model smoothing: modelling Previous: Problem formulation

EAGLES SWLG SoftEdition, May 1997. Get the book...