Intent detection (NLP)
Use NLP intent detection and predefined phrases to trigger scenarios. Below you can find information about supported models and their attributes.
Intent detection models in Ebbot
Intent Manager - OVH & v2
NeuralSpace - OVH
Dialogflow - v2
Intent Manager
Intent Manager is Ebbot's own intent detection model that will replace NeuralSpace for all OVH customers by 31/1-2025. Dialogflow will still be supported on v2.
For most bots, changing from NeuralSpace or Dialogflow to Intent Manager can be completed without additional steps. However, there are two things that are good to know about Intent Manager:
Intent Manager performs optimally with a confidence level set around 0.80.
If your bot's confidence level is very high (e.g., 0.99) or low (e.g., 0.60), we recommend adjusting it to a range between 0.70 and 0.80. After making this adjustment, test the bot with example phrases to ensure the correct scenarios are triggered and that unintended scenarios are not triggered. Based on your test feel free to change the level with 0.05 in either direction until you get your desired result.
Intent Manager merges all NLP-phrases into one intent and cares less about outlying phrases or keywords.
Adding a phrase that is significantly different from the other phrases is more likely to be ignored, as Intent Manager aims to identify a single intent that aligns with all the provided phrases.
NeuralSpace
Will no longer be supported after 31/1-2025
Dialogflow
Dialogflow will continue to be supported, but Intent Manager will become the default intent detection model on v2.
Last updated
Was this helpful?