Hi guys. So I am working on a project which has several entities. My question is that if I have an entity named “Location” and I provide it with some examples with my nlu examples such as city names such has “Lahore” “New York” and so on can I use lookup tables to provide my bot with a list of possible values that entity “Location” can hold? Like I have given examples of 5-10 different cities. But I have 100 possible city names. And if so how can I do that?
I have read the documents regarding the training data format but Im still confused. Your help would be appreciated!
Hi @zaidalvi, lookup tables provide an extra features to the entity extractor (depending on the rasa version you are using, this would be DIETClassifier or CRFEntityExtractor) to help recognize entities with a limited number of possible values. So, in your example, providing a lookup table of city names means that if one of those city names is found in a user message, the feature for “location lookup found” (not literally, just a way to talk about it) is set. This will increase the chances of the location entity being extracted, but there still needs to be sufficient representation of location entities in your examples. What kind of results are you seeing with the data you have?