Hi,
I successfully combined Solr and UIMA with the help of
https://wiki.apache.org/solr/SolrUIMA and other pages (and am happy to
provide some help about how to reach this step).
Right now I can run an analysis engine and get some "primitive"
feature/fields which I specify in the schema.xml automatically recognized
by Solr. But if the features itself are objects, I do not know how to
capture them in Solr.
I provided the relevant solrconfig.xml in [1], and the schema.xml addition
in [2] for the following small example, they are using the AE directly
provided by the UIMA example.
With the input "This is a sentence with an email at u...@host.com", Solr
correctly adds the field:
"UIMAname": [
"36"
]
since this is the index where the email token starts. I could also
successfully capture the feature
<str name="feature">end</str> to indicate where the found email token ends.
However, example.EmailAddress has the features: "begin, end, sofa". sofa is
not a primitive feature, but an "object" which itself has features
"sofaNum, sofaID, sofaString, ..."
How can I access fields in Solr from an annotation like
example.EmailAddress that are not simple strings but itself objects?
I made an image of the CAS Visual Debugger with this AE and the sentence to
show which fields I mean, I hope this makes it more clear:
http://tinypic.com/view.php?pic=34rud1s&s=8#.VN5bF7s2cWN
Does anyone know how to access such fields with Solr and UIMA?
Thanks a lot for any help,
Tom
[1]
<updateRequestProcessorChain name="uima" default="true">
<processor
class="org.apache.solr.uima.processor.UIMAUpdateRequestProcessorFactory">
<lst name="uimaConfig">
<lst name="runtimeParameters">
</lst>
<str
name="analysisEngine">/home/toliwa/javalibs/uimaj-2.6.0-bin/apache-uima/examples/descriptors/analysis_engine/UIMA_Analysis_Example.xml</str>
<!-- Set to true if you want to continue indexing even if text
processing fails.
Default is false. That is, Solr throws RuntimeException and
never indexed documents entirely in your session. -->
<bool name="ignoreErrors">false</bool>
<!-- This is optional. It is used for logging when text processing
fails.
If logField is not specified, uniqueKey will be used as
logField.
<str name="logField">id</str>
-->
<str name="logField">id</str>
<lst name="analyzeFields">
<bool name="merge">false</bool>
<arr name="fields">
<str>text</str>
</arr>
</lst>
<lst name="fieldMappings">
<lst name="type">
<str name="name">example.EmailAddress</str>
<lst name="mapping">
<str name="feature">begin</str>
<str name="field">UIMAname</str>
</lst>
</lst>
</lst>
</lst>
</processor>
<processor class="solr.LogUpdateProcessorFactory" />
<processor class="solr.RunUpdateProcessorFactory" />
</updateRequestProcessorChain>
[2]
<field name="UIMAname" type="string" indexed="true" stored="true"
multiValued="true" required="false"/>
