QueryClassifier#

class QueryClassifier(config)#

Bases: BatchedStep

Step that uses ONNX model to infer the type of the query.

Input - the input field needs to be of type str.

Output - the output field is filled with data of type str.

Parameters

step (str, "app") – app
type (str, "query_processing") – query_processing
name (str, "query_classifier") – query_classifier
model (str, "svm-query-classifier") – Model name
input_field (str, "user_terms_str") – Field on which the query classifier gets applied
path (str, ".") – path
force_question_characters (str, "?¿") – if query ends with one of those characters, enforce question_or_statement prediction
force_question_length (int, 4) – amount of user query terms that enforce question_or_statement prediction

Methods Summary

`pre_session_hook`()	Download and extract the model if it's not existing.
`process_doc`(doc)	Process a document
`shortcut_question_or_statement`(doc)

Methods Documentation

pre_session_hook()#

Download and extract the model if it’s not existing.

process_doc(doc)#

Process a document