Configuration to enable/disable Stanford NER

[srijiths]

It would be great if it has a configiration to enable / disable Stanford NER. In my case i wanted to use the Geo tagging part , but not Stanford NER

[rahulbot]

Hi, the NER is used to identify candidate strings for geolocation, so we can't disable that part of the pipeline. We get the orgs and people for free because of he way the NER works. We could add an option to not process those, but they are computationally insignificant compared to the geographic disambiguation, so it wouldn't serve much of a purpose.

————————————

Rahul Bhargava

http://datatherapy.org

http://rahul.connectionlab.org

@rahulbot

On Thu, Jun 11, 2015 at 8:31 AM, Sreejith.S notifications@github.com
wrote:

It would be great if it has a configiration to enable / disable Stanford NER. In my case i wanted to use the Geo tagging part , but not Stanford NER

Reply to this email directly or view it on GitHub:
#37

[charlieg]

@rahulbot is correct, you can't resolve/disambiguate the place names if you don't have Stanford NER (or a similar tool) to extract the place names themselves from the raw input text. It would kinda defeat the whole purpose to disable Stanford NER.

Recommend closing this issue as invalid.

[srijiths]

Ok. Let me clarify my point one more time. I wanted to use CLIFF to identify the place names and disambiguiate. But i dont want any thing other than Places in the output json. Because I am using another NER engine already in my pipeline. Then i thought why add another additional Stanford NER output in CLIFF.

If this is still invalid please close the issue.

[charlieg]

But i dont want any thing other than Places in the output json.

Can't you just ignore that part of the output? Seems much easier to do that than make changes upstream.

Because I am using another NER engine already in my pipeline. Then i thought why add another additional Stanford NER output in CLIFF.

If you don't like the way CLIFF does it, you might want to explore CLAVIN (which is used by CLIFF). With CLAVIN, you'll have more options to replace Stanford NER with the entity extraction of your choice and just use the core geospatial entity resolution components.

[srijiths]

Yes , the first option is what i am thinking of.

Thank You @charlieg @rahulbot . This is really a cool project