Just investigating the same problem myself so I thought I would add some 
notes in case anyone else is doing the same...

This problem only occurs when full text indexing is enabled. 
SolrServiceImpl.writeDocument instructs Solr to index the full text 
document in a field called 'fulltext'. Solr is configured to use Apache 
Tika to index the document which automatically inserts some metadata at the 
beginning of the field, it is this metadata that is being matched by the 
search causing a chunk of the fulltext field to appear in the highlights. 
So what can we do about this? I'm not sure why we have Solr configured to 
use Apache Tika since the text has already been extracted, but that is more 
than I want to investigate right now, so I am stuck with it. There doesn't 
appear to be an easy way of telling Solr/Tika to not bother inserting the 
metadata, so I am stuck with that too. I could suppress the display of that 
highlight, but that is not really what I want to do, I really want it not 
to match against the inserted metadata information. Suppressing the display 
would just result in the situation where Solr fnds a match, we display it, 
but without the highlighted field explaining where the match was found. 
Looks like I need to get stuck into that investigation I was trying to 
avoid, why are we using Apache Tika to index something that is already a 
flat text file?

For info - https://wiki.apache.org/solr/ExtractingRequestHandler

Cheers, Robin.





On Wednesday, August 26, 2015 at 7:40:18 PM UTC+1, George Kozak wrote:
>
> Hi…
>
>  
>
> I sent this to the list back when Open Repositories was goingon, and I 
> suspect that it was lost in the shuffle, because I haven’t heard anything.  
> So, I am resending.
>
>  
>
> I am not sure if this has already been reported, but I found something 
> strange with my DSpace 5.2 install which I have in development and am 
> planning to move to production.
>
> I am using the XMLUI with Mirage2 theme.
>
>  
>
> I found that in the search results display, if there is no abstract, the 
> system displays random information.
>
>  
>
> For instance:
>
>  
>
> Pellegrini Vineyards 
>
> Hearn, Russell (2008)
>
> stream_source_info Hearn_PellegriniVineyards_WIWP_2008.pdf.txt 
> stream_content_type text/plain stream_size 2036 Content-Encoding UTF-8 
> stream_name Hearn_PellegriniVineyards_WIWP_2008.pdf.txt Content-Type 
> text/plain; charset=UTF-8...
>
> Pellegrini Vineyards 4/30/08 Pellegrini Vineyards • FIRST VINTAGE: 1991 • 
> OPERATED BY: Bob and Joyce Pellegrini • WINEMAKER: Russell Hearn • WINES: • 
> East End Select- Chardonnay, Rosé, Merlot • Pellegrini Vineyards- 
> Chardonnay, Merlot...
>
>  
>
> If I add an abstract to the item, then the search displays changes to:
>
>  
>
> Pellegrini Vineyards 
>
> Hearn, Russell (2008)
>
> Added abstract here…
>
>  
>
> Has anyone else seen this?  Does anyone know what is going on or where I 
> can go to fix this?
>
>  
>
> George Kozak
>
> Digital Library Specialist
>
> Cornell University Library Information Technologies (CUL-IT)
>
> 218 Olin Library
>
> Cornell University
>
> Ithaca, NY 14853
>
> 607-255-8924
>
>  
>
>  
>
>  
>

-- 
You received this message because you are subscribed to the Google Groups 
"DSpace Technical Support" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/dspace-tech.
For more options, visit https://groups.google.com/d/optout.

Reply via email to