I have a string 01a_b-_-c-d which is tokenized as
01a_b
c
d

and the string a_b-_-c_d which is tokenized as
a
b
c
d

why is there a difference when there is a digit at the beginning? I am
using standard unstemmed tokenizer.

Hi Nilesh,

Which version of Lucene are you using?  StandardTokenizer behavior changed in v3.1.

Steve

-----Original Message-----
From: Nilesh Vijaywargiay [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, March 27, 2012 2:04 PM
To: [EMAIL PROTECTED]
Subject: Lucene tokenization

I have a string 01a_b-_-c-d which is tokenized as 01a_b c d

and the string a_b-_-c_d which is tokenized as a b c d

why is there a difference when there is a digit at the beginning? I am using standard unstemmed tokenizer.

---------------------------------------------------------------------

Nilesh,

the StandardAnalyzer is full of generally useful special cases, including emails and numbers detection.
I am supposing you met one such special case which has a justification of some sort.
I can't tell you why but I can tell it's really hard to change because others rely on this somehow (I think).

paul
Le 27 mars 2012 à 20:03, Nilesh Vijaywargiay a écrit :

> I have a string 01a_b-_-c-d which is tokenized as
> 01a_b
> c
> d
>
> and the string a_b-_-c_d which is tokenized as
> a
> b
> c
> d
>
> why is there a difference when there is a digit at the beginning? I am
> using standard unstemmed tokenizer.
---------------------------------------------------------------------