Named Entity Recognition Tool - Basic Question

[hellyars]

I have a general question regarding the Text Mining NER tool.

I want to use the train with new entities option.

1.   How many custom name entities should I use -- to extract names from a paragraph (test case) or from 25,000 pages of text?

2.   Does every potential value need to be in the custom list OR (given enough examples) will the model learn to recognize and entity as fitting the pattern of the custom list even if that entity is not itself in the custom list?

[Manoj_k]

hi @hellyars for

1. you can use a 20% of random data from the 25000 pages.
2.The model can learn to identify entities based on the patterns and context present in the training data, even if those specific entities are not included in the custom list.