Hello! I am trying to use lookup tables to save username.Problem in that if I enter name which listed in username intent rasa understands it as name but if I enter name which listed in names.txt and not in username intent it does not recognize it as name,but I have provided path to names.txt
@MMustafa As mentioned in the lookup table documentation, you need to make sure that you are using some of the names in your
names.txt file in the NLU training data. Please make sure to follow this pattern.
Just to be sure the entity
names you have in your training data is not the same as the
@Tanja lets say we do not use some some of the entity values from the lookup table, instead we fill the inform intent(an intent to handle entity values in the form action) with all the values in the lookup table with annotation. So will our extractor be able to extract the entities used other than the inform intent? or will our lookup be working in this case as well? Thanks
If you have a lookup table in place and defined a
RegexFeaturizer in your pipeline, the entity extractor does not know how to handle the features created by the
RegexFeaturizer if you don’t have some entities with values from the lookup table in your training data.
So, in case you want to make sure your entity extractor extracts entities present in the lookup tables more likely, you need to have some examples in the training data. It does not matter how you handle the form intent in that case. The only thing that matters is that you have some example of entities with values of the lookup table in your training data. You can also put those under the
inform intent, if you like.
Does that answer your question?