FAQ

This recognizer uses a frozen Universal Sentence Encoder TensorFlow model to encode, using sentence embedding vectors, a list of Frequently Asked Questions and tag sentences that match a question/answer pair given a specified threshold of accuracy with the question/answer from the FAQ.

The recognizer also gives the option to use a python model instead of TensorFlow. You'll need the Python Bridge running for this.

Configuration

Configuration using TensorFlow:

• Match threshold ( type=double | default=0.8 | required ) - The minimum similarity match between the vectors of a sentence and a question/answer to consider the question to tag the input sentence.
• Eval Answer ( type=boolean | default=false | required ) - Whether or not to include the answer vectors when evaluating an input sentence.
• Add Vectors ( type=boolean | default=false | required ) - Whether or not to include the vector generated in the metadata.

Configuration using Python:

• Host Name ( type=string | default=localhost | required ) - Python Bridge host name
• Port ( type=integer | default=5000 | required ) - Python Bridge host port
• Select Model ( type=string | default=bert-base-uncased | required ) - Python model to be used by the recognizer
• Match threshold ( type=double | default=0.8 | required ) - The minimum similarity match between the vectors of a sentence and a question/answer to consider the question to tag the input sentence.
• Select Version ( type=integer | default=1 | required ) - Version of the python model to use

Adding a Question/Answer pair

Click on the  button which will popup the "Add new FAQ" dialog

• Write question here ( type=string | required ) - Question to add to the FAQ
• Write answer here ( type=string | required ) - Answer to the question
• Url ( type=string | optional ) - If provided, the source of the question/answer. (Usually a link to the HTML FAQ page).
• Confidence Adjustment ( type=double | default=1 | required ) - Confidence Adjustment for this entry
  
  • 0.0 to < 1.0  decreases confidence value
  • 1.0 confidence value remains the same
  • > 1.0 to  2.0 increases confidence value

Load FAQ for evaluation

Clicking on the button will encode all question/answer pairs for evaluations.

When the icon shows in purple, it means that the vectors have been loaded and its ready to use for matches.

Click onicon to copy to the clipboard the sentence embedding vector of the question (top) or answer (bottom) for debugging purposes.

General Settings

The general settings can be accessed by clicking on

More settings could be displayed in the same dialog, it varies per recognizer.

• Enable - Enable the processor to be use in pipelines.
• Base Pipeline - Indicates the last stage, from a pipeline, needed by the recognizer.
  
• Skip Flags ( optional ) - Lexical items flags to be ignored by this processor.
• Boundary Flags  ( optional ) - List of vertex flags that indicate the beginning and end of a text block.
  
• Required Flags ( optional ) - Lexical items flags required by every token to be processed.
• At Least One Flag ( optional ) - Lexical items flags needed by every token to be processed.
• Don't Process Flags ( optional ) - List of lexical items flags that are not processed. The difference with "Skip Flags" is that this will drop the path in the Saga graph, skip just skips the token and continues in the same path.
• Confidence Adjustment - Adjustment factor to apply to the confidence value of 0.0 to 2.0 from (Applies for every match).
  • 0.0 to < 1.0  decreases confidence value
  • 1.0 confidence value remains the same
  • > 1.0 to  2.0 increases confidence value
• Debug - Enable debug logging.