PEGA NLP topic detection based on Phrase or sentence
We are using keyword based model to detect topic. In should words we have given multiword phrase like - "New Account".
Our expectation is if an email has "New Account" (continuous words ) then the topic needs to be identified. But in our case, if email has New in one sentence and account is in another sentence, still it is detecting the topic.
eg.
eg. 1. I want top open New Account - topic should be detected based on "New Account" Keyword Phrase
eg. 2. I'm new to bank. I want to check account type - Topic should not be detected based on "New Account" since "New Account" is not continuous words in eg 2
In both the examples, New Account topic is getting detected. How to limit individual keyword based search when we have to determine topic based on continuous words
Accepted Solution
Updated: 22 Feb 2023 2:19 EST
Pegasystems Inc.
IN
@poorna18 if you want continuous text to be detected, you should enclose all words in double quotes when defining keywords for that topic in taxonomy CSV.