On 02/25/2011 03:02 PM, Koji Sekiguchi wrote:
> (11/02/25 18:30), Tarjei Huse wrote:
>> Hi,
>> On 02/25/2011 02:06 AM, Koji Sekiguchi wrote:
>>> (11/02/24 20:18), Tarjei Huse wrote:
>>>> Hi,
>>>>
>>>> I got an index where I have two fields, body and caseInsensitiveBody.
>>>> Body is indexed and stored while caseInsensitiveBody is just indexed.
>>>>
>>>> The idea is that by not storing the caseInsensitiveBody I save some
>>>> space and gain some performance. So I query against the
>>>> caseInsensitiveBody and generate highlighting from the case sensitive
>>>> one.
>>>>
>>>> The problem is that as a result, I am missing highlighting terms. For
>>>> example, when I search for solr and get a match in caseInsensitiveBody
>>>> for solr but that it is Solr in the original document, no highlighting
>>>> is done.
>>>>
>>>> Is there a way around this? Currently I am using the following
>>>> highlighting params:
>>>> 'hl' => 'on',
>>>> 'hl.fl' => 'header,body',
>>>> 'hl.usePhraseHighlighter' => 'true',
>>>> 'hl.highlightMultiTerm' => 'true',
>>>> 'hl.fragsize' => 200,
>>>> 'hl.regex.pattern' => '[-\w ,/\n\"\']{20,200}',
>>>
>>> Tarjei,
>>>
>>> Maybe silly question, but why no you make body field case insensitive
>>> and eliminate caseInsensitiveBody field, and then query and
>>> highlight on
>>> just body field?
>> Not silly. I need to support usage scenarios where case matters as well
>> as scenarios where case doesn't matter.
>>
>> The best part would be if I could use one field for this, store it and
>> handle case sensitivity in the query phase, but as I understand it, that
>> is not possible.
>
> Hi Tarjei,
>
> If I understand it correctly, you want to highlight case insensitive way.
> If so, it is easy. You have:
>
> body: indexed but not stored
> caseInsensitiveBody: indexed and stored
>
> and request hl.fl=caseInsensitiveBody ?
But I also want to be able to do it the other way around - i.e. I need
to keep both options open so that I at duntime can select if I want to
do a query that is or is not case insensitive. That is why I'm storing
the non lowercased version of the field - with that I do not loose
information.
Regards,
Tarjei
>
> Koji
--
Regards / Med vennlig hilsen
Tarjei Huse
Mobil: 920 63 413
