Fine tune Knowledge Graph to display maximum number of otpions to user

Hi Team,

We have 1000+ questions which are grouped under different nodes automatically generated from auto-ontology generator.

Even though user typed intent is matching with many questions, bot usually returns only 1-2 questions since KG engine looks for number of terms in the utterance and qualify the path of the nodes based on threshold set in “Path Coverage”

Example:

How to deal with programming errors in NodeJS?
How to deal with programming errors in python?
How to deal with programming errors in Java?
How to deal with programming errors in Javascript?

User utterance : programming errors

Expected result : Should return all the four questions in this case
Actual result : Each of these questions are grouped under different node names, its returning only 1-2 questions.

But I want to retrieve as many as possible.

Straight forward solution is to group nodes under correct nodes names. But since node names are automatically generated, I can’t make changes to it. So I am looking for alternative way to accomplish this? Please suggest

Can the “minimum and maximum threshold” be redefined/adjusted to retrieve as many as possible match questions?
Can we add common tags to all these questions so that KG will pick up them up regardless of the node path?

@vinothini.rangaraj

Did you try using the KG suggestion count? It can display up to 5. Does it work for you
image

More than a search engine, the KG is for addressing conversation time questions based on well-defined ontology. If you can you please share a screenshot of your KG ontology with paths clearly visible, I can try to suggest a solution.

yes KG suggestion count is already set to 5.

User utterance : apply binding rules
Questions containing same terms falls under different node path as below,
with this pattern
How to apply binding rules to ***

MicrosoftTeams-image