Rasa: Most common entity extraction issues

[akelad]

We're trying to collect a list of the most common issues people have with entity extraction. What we have so far:

Please vote for the ones you also have issues with, or comment with any other common issues you have 🙂

[kumarv1988]

Names matching rwgex

[akelad]

@kumarv1988 could you elaborate on what "names matching regex" are?

[AbdurRub]

Looking forward to phrase_matcher merger!

[amn41]

@akelad what about entity roles? e.g. origin_location, destination_location ?

[venky14]

@AbdurRub what does phrase_matcher do?

[ravirkd]

regex for person names? e.g. First_name, last_name?

[AbdurRub]

https://github.com/RasaHQ/rasa_nlu/pull/822 @venky14

[geojolly]

The entity recognition is not good for custom locations : for instance when i model it for hospitality industry, each hotel may have many restauraants within, other places for us all of them are treated as a place within hotel : restaurants, spas, gyms, reception, hotel shop etc. the entity recognition is not accurate always in this case.
Also, often the model takes the stop words as entities : eg: user utterance: "Where is the sea food restaurant ? " the model takes "the seafood restaurant". at present I am cleaning up it with python .

(these are few that I noted, but I could list many ;)

[NikhilBansal21]

@akelad not able to extract a custom regex based alphanumeric entity with regex pattern like
(?i)(SR)[0-9]* in rasa nlu

[91ns]

Are there any news on whether Entity roles will be supported or not?

[Manvendra01]

Entity recognition is not working well with entities not included in training model.
entity extraction is mostly having problem in entity with spaces.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale[bot]]

This issue has been automatically closed due to inactivity. Please create a new issue if you need more help.