NER_Multiple_entities

I am working on a problem of entity extraction which requires me to extract variables of interest from a text document. My challenge is that the text contains multiple entities of a variable, for ex. For a variable "name", there are multiple names but I only have to extract a particular name of interest. Likewise there are other variable with multiple entities. I have 2 questions regarding this.

1. How should I prepare my data for training?. Should I annotate all the entities for a variable or just the entity of interest. For ex: Out of names like "james", "dave" and "harry", if name of interest is "james", then should I annotate only "james" or all the names for training.

2. Is there any particular method to approach such problem?