<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html; charset=GB2312" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
<br>
KN smoothing modifies the lower-order counts before estimating
probabilities.<br>
Therefore, the unigram counts in the two cases you ran are not the same.<br>
(You could use the -write option to dump out the modified counts.)<br>
For details, see the Chen & Goodman paper.<br>
<br>
BTW, please join the srilm-user group if you plan to use SRILM in a
serious way, and address questions like these to the group.<br>
<br>
Andreas <br>
<br>
On 3/8/2010 11:25 PM, ÂÞ»ªÇå wrote:
<blockquote cite="mid:SNT125-DS169A88B5E0F69A9CECC9E6D4340@phx.gbl"
type="cite">
<meta content="text/html; charset=GB2312" http-equiv="Content-Type">
<meta name="GENERATOR" content="MSHTML 8.00.6001.18876">
<div><font face="Calibri">Dear sir:</font></div>
<div><font face="Calibri">I am new to SRILM. When I tried to build a
language model from the count file "nettalk.count", I got the following
error:</font></div>
<div><img
style="margin: 10px 0px 0px; position: static; display: inline-block; float: none;"
src="cid:part1.01010102.00060001@speech.sri.com" border="0" width="639"
height="182"></div>
<div> </div>
<div><font face="Calibri">but when input other command, the error was
gone:</font></div>
<font face="Calibri">
<div><img
style="margin: 10px 0px 0px; position: static; display: inline-block; float: none;"
src="cid:part2.06040604.01010301@speech.sri.com" border="0" width="640"
height="219"></div>
<div>Why did I get different values of n(1, 2, 3, 4) when both
commands above were using modkn smoothing for 1-grams?</div>
<div>Thanks.</div>
<div> </div>
<div> &nbs
p; Huaqing
Luo</div>
<div> &nbs
p; Department
of EE, TsingHua University, China. </div>
</font></blockquote>
<br>
</body>
</html>