search

brmson / yodaqa

A Question Answering system built on top of the Apache UIMA framework.
http://ailao.eu/yodaqa
Other
619 stars 205 forks source link

Waiting until AnalysisEngine is finished before continuing #68

Closed ercanse closed 7 years ago

ercanse commented 7 years ago

I've been working for some time on creating a Dutch version of YodaQA. Within my particular setup, I'd like to be able to do some collective processing on all results returned from the Solr searches, before continuing with processing them further in the pipeline. Concretely, I want to prevent PassageExtractorAE from kicking into action before the SolrFullPrimarySearch multiplier has finished outputting all its CASes.

Is this possible? Would it be a matter of configuring SolrFullPrimarySearch differently, or would I need to adapt the flow control logic somewhere in cz.brmlab.yodaqa.flow or cz.brmlab.yodaqa.flow.asb?

Any hints would be much appreciated.

ercanse commented 7 years ago

Figured it out: a CAS multiplier can be put in between that just 'collects' the first n-1 times it is called, does its collective processing the nth time, and then starts outputting its CASes one by one.