Opener Webservices Quick Start with CURL

You can try the webservices using the popular command line tool CURL. For the best result, put your text in a textfile, for the example called some_file.txt and then pass it through the webservices like this:

cat some_file.txt | \
curl -F 'input=<-' "http://opener.olery.com/language-identifier" | \
curl -F 'input=<-' "http://opener.olery.com/tokenizer" | \
curl -F 'input=<-' "http://opener.olery.com/pos-tagger"

The trick here is to use curls -F option with input=<- as argument. The <- in this case means it reads from STDIN and posts it to the webservices.

An alternative way is to use -F 'input=<some_file.txt'. Both options read the textfile and put them through the webservices.

Chaining Webservices

You can chain multiple webservices together. The more callbacks you add the longer your pipeline will be.

So the structure of the curl request looks like this:

cat some_file.txt | \
curl -F 'input=<-' "http://opener.olery.com/language-identifier" | \
curl -F 'input=<-' "http://opener.olery.com/tokenizer" | \
curl -F 'input=<-' "http://opener.olery.com/extra_callback1" | \
curl -F 'input=<-' "http://opener.olery.com/extra_callback2"

Then you will get the result on your terminal.

Returning JSON

If you prefer to receive JSON instead of KAF at the end of the pipline. Append “.json” to the outlet URL. For example:

If you do the following request:

cat some_file.txt | \
curl -F 'input=<-' "http://opener.olery.com/language-identifier" | \
curl -F 'input=<-' "http://opener.olery.com/tokenizer" | \
curl -F 'input=<-' "http://opener.olery.com/pos-tagger" | \
curl -F 'input=<-' "http://opener.olery.com/polarity-tagger" | \
curl -F 'input=<-' "http://opener.olery.com/opinion-detector-basic" | \
curl -F 'input=<-' "http://opener.olery.com/kaf2json"

You will get a JSON representation of the KAF file.

Examples

A Full Pipeline

cat some_file.txt | \
curl -F 'input=<-' "http://opener.olery.com/language-identifier" | \
curl -F 'input=<-' "http://opener.olery.com/tokenizer" | \
curl -F 'input=<-' "http://opener.olery.com/pos-tagger" | \
curl -F 'input=<-' "http://opener.olery.com/constituent-parser" | \
curl -F 'input=<-' "http://opener.olery.com/ner" | \
curl -F 'input=<-' "http://opener.olery.com/ned" | \
curl -F 'input=<-' "http://opener.olery.com/property-tagger" | \
curl -F 'input=<-' "http://opener.olery.com/polarity-tagger" | \
curl -F 'input=<-' "http://opener.olery.com/opinion-detector"

Rule based opinion detection

The opinion-detector endpoint is based on machine learning. The opinion-detector-basic however is rule base. Feel free to swap the one for the other.

to do so use:

curl -F 'input=<-' "http://opener.olery.com/opinion-detector-basic"

instead of

curl -F 'input=<-' "http://opener.olery.com/opinion-detector"

Use the tree tagger instead of the pos-tagger

curl -F 'input=<-' "http://opener.olery.com/tree-tagger"

Beware: the tree tagger is considerably slower and less stable than the standard pos-tagger. If after ± 10 seconds your document is not ready yet, it probably failed.

List of services with their endpoints