I'm trying to use the General Tag Filter to get rid of person terms. However, no matter whether "Filter matching" is enabled or not, the result always returns only the terms with the tag that I want to filter out. I guess it's a bug. Has anyone faced a similar issue?
Thanks in advance,
Hi, It seems to work for me.
Are you sure you used the OpenNLP Tagger node first and selected Person in the node. Then use the BoW Creator node, and then use the General Tag Filter node and highlight the Person box. For me, only the Person tags are retained, and if I tick the Filter matching box, then everything is retained except for the Person tags.
I am on KNIME 2.9.2
Thanks for your prompt reply. After further testing, I noticed that the problem occurs only if more than one tag is applied to the term. For example, if a BoW has both NE and POS tags, the filter does not work properly. This is still a problem as I don't want to filter out the NE terms before adding the POS tags, because the POS analysis might fail because of the missing terms. Any suggestion on that? Btw, I'm using Knime 2.9.2 too.
this is an issue in the General Tag Filter. Thank you for pointing this out. Alternatively you can use the Modifiable Term Filter node. This node filters (out or in) all terms that have been set unmodifiable before (which is done be default by NE taggers). You can keep only those terms or filter them out. Attached you find an example workflow.