The FAQ stage does a semantic comparison of a sentence against questions and its respective answer (using TensorFlow), if the confidence value is in the threshold, it will create a tag holding the question and answer.

This recognizer uses a frozen Universal Sentence Encoder TensorFlow model to encode (using sentence embedding vectors) a list of Frequently Asked Questions. Then tags sentences that match a question/answer pair given a specified threshold of accuracy with the question/answer from the FAQ. Most recently the FAQ also can use Saga Python Bridge to encode the sentence with different algorithms.

Operates On:  Lexical Items with TOKEN and possibly other flags as specified below.

Library: saga-fag-stage

Stage is a Recognizer for Saga Solution, and can also be used as part of a manual pipeline or a base pipeline


Generic Configuration Parameters

  • boundaryFlags ( type=string array | optional ) - List of vertex flags that indicate the beginning and end of a text block.
    Tokens to process must be inside two vertices marked with this flag (e.g ["TEXT_BLOCK_SPLIT"])
  • skipFlags ( type=string array | optional ) - Flags to be skipped by this stage.
    Tokens marked with this flag will be ignored by this stage, and no processing will be performed.
  • requiredFlags ( type=string array | optional ) - Lex items flags required by every token to be processed.
    Tokens need to have all of the specified flags in order to be processed.
  • atLeastOneFlag ( type=string array | optional ) - Lex items flags needed by every token to be processed.
    Tokens will need at least one of the flags specified in this array.
  • confidenceAdjustment ( type=double | default=1 | required ) - Adjustment factor to apply to the confidence value of 0.0 to 2.0 from (Applies for every pattern 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 ( type=boolean | default=false | optional ) - Enable all debug log functionality for the stage, if any.
  • enable ( type=boolean | default=true | optional ) - Indicates if the current stage should be consider for the Pipeline Manager
    • Only applies for automatic pipeline building

Configuration Parameters

  • questions ( type=string | required ) - Name of the resource containing the questions of the FAQ
  • tagWith ( type=string | default=questionTag | optional ) - Indicates the tag which is going to be use for tagging
  • modelPath ( type=string | required ) - Path to the TensorFlow model in case the TensorFlow model is going to be used
  • threshold ( type=double | default=0.6 | optional ) - Threshold indicating when a predictions is acceptable
  • useTensorFlow ( type=boolean | default=true | optional ) - Indicates if the FAQ is going to use TensorFlow or Saga Bridge
  • evalAnswers ( type=boolean | optional ) - Indicates if the sentence must be compared against the answer as well
  • modelName ( type=string | default=bert-base-uncased | required ) - Model name registered in the python bridge
  • version ( type=string | default=1 | optional ) - Model version registered in the python wrapper to query
  • hostname ( type=string | default=localhost | optional ) - Python server communication hostname
  • port ( type=integer | default=5000 | optional ) - Python server communication port


"questions": "saga_provider:saga_faq",
"tagWith": "question",
"modelPath": "./tf-model",
"threshold": 0.6,
"useTensorFlow": true,
"evalAnswers": true,
"modelName": "bert-base-uncased",
"version": "1",
"hostname": "localhost",
"port": 5000,


Example Output

V---------------------[Any analytics available for Internal Sites? Why are dogs always hungry?]---------------------V 
^----------[Any analytics available for Internal Sites? ]-----------V---------[Why are dogs always hungry?]---------^ 
^-[Any]-V-[analytics]-V-[available]-V-[for]-V-[Internal]-V-[Sites?]-^-[Why]-V-[are]-V-[dogs]-V-[always]-V-[hungry?]-^ 
^----------------------------[{question}]---------------------------^


Output Flags

Lex-Item Flags:

  • SEMANTIC_TAG - Identifies all lexical items which are semantic tags.
  • ANSWER- Identifies a lexical item holding an answer from the FAQ.

Vertex Flags:

No vertices are created in this stage


Resource Data

Description of resource.

Resource Format

"_id" : "KGAAJGsBemSwA0nZTLXA",
"tag": "animals",
"question" : "This is a cat?",
"answer" : "That's a tiger",
"fields" : {
    "questionVector" : [ 0.058885570615530014, -0.002637901809066534, ... ],
    "answerVector" : [ 0.34535465645635530014, 0.002342342343453453, ... ],
    "url" : "https://isthisacat.com"
    . . . additional fields as needed go here . . . 
}
"confAdjust": 0.95
  • Multiple entries can have the same pattern. If the pattern is matched, then it will be tagged with multiple (ambiguous) entry IDs.
  • Additional fielded data can be added to the record; as needed by downstream processes.

Fields

  • question ( type=string | required ) - question you want to compare against the sentence
  • answer ( type=string | required ) - answer of the question, that can be also be compare against the sentence
  • fields ( type=json | required ) - metadata of the FAQ entry
    • questionVector ( type=double array | required ) - context vector of the question, size can change depending of the algorithm used to build it
    • answerVector ( type=double array | required ) - context vector of the answer, size can change depending of the algorithm used to build it
    • url ( type=string | optional ) - for use in external applications
  • tag ( type=string | required ) - Tag which will identify any match in the graph, as an interpretation

    • These will all be added to the interpretation graph with the SEMANTIC_TAG flag.

      Tags are hierarchical representations of the same intent. For example, {city} → {administrative-area} → {geographical-area}

  • _id ( type=string | required ) - Identifies the entity by unique ID. This identifier must be unique across all entries (across all dictionaries).

  • confAdjust ( type=boolean | required ) - Adjustment factor to apply to the confidence value of 0.0 to 2.0

    • This is the confidence of the entry, in comparison to all of the other entries. (Essentially, the likelihood that this entity will be randomly encountered.)
    • 0.0 to < 1.0  decreases confidence value
    • 1.0 confidence value remains the same
    • > 1.0 to  2.0 increases confidence value
  • updatedAt ( type=date epoch | required ) - Date in milliseconds of the last time the entry was updated
  • createdAt ( type=date epoch | required ) - Date in milliseconds of the creation time of the entry