... Yes. But it has to be defined at the factory level. We really need a tutorial on this. If you have a filtered tokenizer FilterTokenizer with the obvious ...
Hello, Currently I have a scenario where a list of curse words is put to a trie, and then words entered by users (not text, but individual words) are run...
I'm afraid edit distance won't be sufficient due to the example you mention. This isn't an implementation issue, though it will slow down as you get more and...
Little confused, I understand when using approximate matcher, as the size of the dictionary grows the speed goes down. Question: approximate matcher tries to...
may I add the language is turkish. I have looked at both soundex and metaphone, but i think these are language dependent. I could have a list of clean words,...
... The results won't be equivalent. The string vs. string edit distance matches the whole strings against each other, whereas the approximate chunker looks...
Dear Bob, Thank you for your explanatory post. What I have done was to only check words with more then 6+ inputs, with a max distance of 2. It works fairly...
Thanks to an anonymous bug reporter for submitting these issues with $LINGPIPE/demos/tutorial/posTags First the typo: The read-me.html gives the wrong...
Has anyone successfully compiled LingPipe 3.5.1 using GNU's JDK? We received an error report for some classes that work just fine in the Sun JDKs. Here are...
We just rolled out LingPipe 3.6.0. As usual, find it on our homepage: http://alias-i.com/lingpipe Here are the details: Intermediate Release The latest...
Dear All, my name is Marco, and I'm new in this group and as lingpipe user, so do not hate me for some silly questions :-) My research project requires to...
... We don't distribute any data, but our named entity tutorial points to some sources of data. ELRA and LDC also distribute data, but it's expensive. Most...
Dear Bob, my problem is multilanguage in the sense that I handle documents that can be written in German, French, Spanish and so on, each document is written...
... Nothing free and fast, I'm afraid. We don't have corpora in French or German. Spanish is easy -- you can get it from the CoNLL data. You can get Spanish...
Hello- I'm having a blank. In the context of spell checking, is the edit distance used between the user-entered term and the suggested term, or the reverse? I...
... I should make this clearer in the doc for each operation. ... Right. It's the noisy-channel setup, so it's edits going from the suggested term to the...
As of 3.0, the chunking interface completely changed so it's no longer backward compatible with 2.x code. The last version of LingPipe to support the...
I recently downloaded LingPipe 3.6.0 and tried the ChineseToken tutorial sample. I always got NullPointerException. My environment is: - Window XP - Java...
... It sure does. Thanks for the detailed bug report. The culprit is the following file: $LINGPIPE/src/com/aliasi/spell/CompiledSpellChecker.java The method...
Hi Bob, I have a question on Model Quality. I used the ChineseToken sample to generated a words-zh-as.CompiledSpellChecker model, which has size 78,303KB. I...
... The other way to control model size is take longer n-grams and prune out low-count sequences. If you follow the tutorial, you'll see where we run standard...
Hi Bob, Thanks for replying. Does longer n-grams model mean more accuracy? How do I prune out low-count sequences from model using LingPipe? I have some...
... Usually longer n-grams means more accuracy up to a point at which accuracy plateaus. Longer n-grams can overfit in some situations compared to shorter...
Thanks, Bob. The goal of making English Chinese word alignment is to create some TMX files for "translation memory" tools used by translators. We have some MT...
LingPipe 3.7.0 is now available from: http://alias-i.com/lingpipe The only significant change is an update to the MEDLINE DTDs used by the MEDLINE parser....
LingPipe 3.7.0 will generate a warning when compiling all jars because of an issue with non-ASCII chars in one of our java files. It works with Windows...
Hello, I was looking at NGramBoundaryLM and wondering why exactly those begin and end characters are needed and when you'd use NGramBoundaryLM instead of...
... The spelling corrector and HMMs wrap this all up the right way by default. Classifiers require you to specify which to use, and the decision is based on...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
