I have a dictionary of 10000+ entities. So I want to use database table as the lookup table. Is it possible?
@abhishekborse No I dont think so. I came across the same problem. For me entity(e.g. company names) list was not fixed and company name could be chinese or anything, which makes very hard to train a model on. One option is you write every entity_value in the lookup table. Or if the entity_type is quite random, then you could write a small API to perform text search, to search for Database for Named Entity Recognition and call it in the Custom Action or on the Interface level. You can use simple matching techniques using ngrams and Apache Lucene.
I think, it is not a good practice to maintain numbers of data files in the application.