[SRILM User List] Data preparation for building language model using ngram-count

[Andreas Stolcke]

On 1/15/2010 8:30 PM, Abbas Malik wrote:
> Dear All,
>
> Do we really need to add <s> at the start of each sentence and </s> at 
> the end of each sentence for the preparation of a language model using 
> ngram-count.
>
> my data looks like:
>
> =============
> <s> sentnce1 </s>
> <s> sentence2 </s>
> so on...
> =============
>
> De we really need <s> and </s> tags?
No.  It is done automatically by ngram-count .

Andreas