{"id":18149,"date":"2019-03-04T07:00:45","date_gmt":"2019-03-04T07:00:45","guid":{"rendered":"https:\/\/developer.kore.ai\/v10-1\/?post_type=docs&p=16989"},"modified":"2023-06-26T13:05:24","modified_gmt":"2023-06-26T12:05:24","slug":"nlp-detection","status":"publish","type":"post","link":"https:\/\/developer.kore.ai\/v10-1\/docs\/bots\/nlp\/nlp-detection\/","title":{"rendered":"Ranking and Resolver"},"content":{"rendered":"
\r\n\t
\r\n\t\t
\r\n\t\t\t

The Kore.ai NLP engine uses Machine Learning, Fundamental Meaning, and Knowledge Graph (if any) models to match intents. All the three Kore.ai engines finally deliver their findings to the Kore.ai Ranking and Resolver component as either exact matches or probable matches. <\/span>Ranking and Resolver<\/strong> determines the final winner of the entire NLP computation.<\/span><\/p>\n

The Ranking & Resolver engine receives the outputs from the above engines and further processes them.<\/span><\/p>\n

<\/div><\/div>\n\n\n\n
\n

Notes on Ranking and Resolver Version 2<\/span><\/span><\/h1>\n

With the introduction of Few-shot models in ML and KG engines, rescoring by Ranking & Resolver is no longer required for intent identification. Therefore, we have introduced a new version of Ranking & Resolver (Version 2) for Few-shot models that only ranks intents based on scores from ML and KG engines. The version significantly improves the accuracy of intent identification.You can <\/span>choose one of the following versions of the Ranker and Resolver to define how to rank matched intents from NLP engines and determine the winning intent:<\/span><\/p>\n