How to get one word answers into entities

[tanmaysinghal98 (Tanmay Singhal)]

I am using 0.13.1 and i want to implement this:-

user: create a new folder

bot: what should be the name of the folder?

user: let the name be Applications

bot: done

and

user: create a new folder

bot: what should be the name of the folder?

user: Applications

bot: done

and

user: create a new folder name Applications

bot: done

I have made a formAction for this. But the problem is, for single word answers it is guessing wrong intent like greet or goodbye. How to force nlu to look for specific intent if the entity is not captured in first intent. Even after forcing, how to train nlu for capturing entities from those single word answers.

[tanmaysinghal98 (Tanmay Singhal)]

intent:mkdir

• create folder
• new folder
• new Directory
• create a new directory
• make folder
• make a new folder
• make Directory
• create a new folder
• create a folder named Downloads
• new folder Tanmay
• new folder sfesds
• new folder Tasdascanmay
• new folder Tasfscsdfvdddzscnmay
• new folder Taenmdfgdtgdrsay
• make directory with name Applications
• make directory with name qdhebdefwer
• make directory with name gftgbdgdrf
• make directory with name edawdawde
• make directory with name rdtgtg
• please make a new folder and keep the name of folder Goa
• please make a new folder and keep the name of folder rwrerfs
• please make a new folder and keep the name of folder Grw4erw3oa
• please make a new folder and keep the name of folder Goaxvxfvdvs
• please make a new folder and keep the name of folder Godawdeqwewea
• create a new directory with name India
• create a new directory with name aefgajds
• create a new directory with name Indefsrfsdawia
• create a new directory with name Inqwsfdswqswdsfedia
• create a new directory with name Indwewefdia
• create a new folder tanmay
• create a folder named laudjba
• create a new folder rushabh
• create a new folder sejfh
• create a new folder woisdfsut
• create a new folder rushsrggrafabh
• create a new folder ruswewfwehabrgh
• create a folder named djhf
• create a folder named laudrglaa
• create a folder named ljrgefjqerauda
• create a folder named oweuro

intent:new_folder_mkdir

• name of new directory is Pakistan
• name of new directory is sefss
• name of new directory is Pakisfsfaadstan
• name of new directory is awsefsPaawaawakistan
• name of new directory is Pafesfsejsajkistan
• Pictures
• Music
• Downloads
• Movies
• madarjknnkjnchod
• efsefs
• madarsvsdcschod
• madadsxasdawsrchod
• awdsefsf
• dcddcdcsd
• sdsdfsfse
• keep US as the name
• keep shgdfnsdf as the name
• keep ssfsefse as the name
• keep sdfsdczsdsd as the name
• keep jyuijyyujt as the name
• keep UdfgdthS as the name
• name should be Australia
• name should be jfgjse dse
• name should be Aussgdhgachmxaedaetralia
• name should be Austdgdfvsdvsefralia
• name should be efawaefefse
• name should be efsrgrtr
• keep it Kenya
• keep it adsghfksjef
• keep it sefjjeopoedpawjdka
• keep it zdnasbda
• keep it sfjksfefse
• keep it as sfjksfefse
• keep it as dfbsmesr
• keep it as sfjkswadasxsfefse
• keep it as sfjdgthgyjukuiksfefse
• keep it as sfuikyujtyjksfefse
• it should be named Priyanshi
• it should be named fgjdgfdd
• it should be named sdvxdvf
• it should be named Priydgbfdanshi
• it should be named Priyfvsdsdanshi

regex:new_folder

• [a-zA-Z]+

[Juste]

Hey @tanmaysinghal98. The best approach for one word entity extraction is to use Lookup tables. Did you have a chance to look into them yet?

[lgrinberg (Leo@Vocinity)]

If your entities will come from a fixed set of values, then use lookup tables, as Juste has recommended. If your entities can have totally random values and they will overlap with other entities, then use the approach described here.

Providing conversation context to the NLU using microservices

Many thanks @lgrinberg for sharing your concept. Your example is related to word with different meaning (e. g. last name vs. city). This is a good example where context matters, but there are so much more situation where context is key. Also words, which are not ambiguous, can have a different meaning depending on the context. Let’s say we have a chatbot that talks about renting or buying real estate: Context: None / User message: “apartment” Since we have no context, we basically do not kn…