If you haven’t already read my other posts on creating my #chatbot, then check them out:
- Create a Bot using an Azure App Service Bot Service
- Most common chatbot questions and how to answer them
- Publishing a chatbot using Bot Services and LUIS
- How I tested / debugged my chatbot that I created using the Bot Services on Azure
- How my chatbot remained statefull using Azure Bot Services
- C# Bot Builder Samples on GitHub
- Top 10 must have Phrase List Features for your chatbot or any bot LUIS
- 1000 must have utterances for your chatbot using LUIS
- Adding LUIS Entities to my chatbot
- How I improved my chatbot
Anyway, in a post I am currently working on (How I improved my chatbot) I cover how I decided to create my LUIS entities. Although the entities were good and they helped my chatbot #chatbot answer or respond to questions better, there are still some words missing. So I looked for synonyms for my entities, shown in Figure 1 and added the synonymous.
Figure 1, LUIS entities for a #chatbot
Here is a table that contains the Phrase List for each of the Entities.
|Bad||amiss,atrocious,coarse,contaminated,contemptible,corrupt,crappy,cruddy,crummy,defective,deficient,deplorable, depraved,disagreeable,dismal,dissatisfactory,evil,execrable,faulty,foul,ghastly,grungy,harmful,heinous,icky, improper,infamous,inferior,injurious,junky,lousy,nasty,nefarious,off,poor,quality,putrid,rotten,scandalous,sinful,sinister,snide,spoiled,substandard,tainted,thepits,uncouth,wicked|
Then when I entered an utterance with a word that is in the LUIS Phrase List, it gets recognized in the window automatically, as shown in Figure 2.
Figure 2, LUIS chatbot utterance with Phrase List content automatically identified
If you look at the JSON output of the same question you see, that shown below.
"query": "You have accomplished a lot",
I can use both the Intent=None and Intent=State as the probabilities are high, the Entity Type: Good and the Entity entity: accomplished to provide a valid and quality response.