Which type of entities are considered pre-trained entities based on rules?

Pre-trained entities based on rules refer to those that have been constructed using predefined standard or custom-defined rules without relying on machine learning techniques or requiring large datasets for training. These entities typically utilize logical conditions, patterns, or regular expressions to recognize and categorize data inputs effectively.

In contrast, entities derived from machine learning alone rely on the training data and might require extensive samples to learn and identify patterns, which does not align with the concept of being rule-based. Entities trained from scratch would necessitate building the model from the ground up, which also does not represent pre-trained entities. Additionally, entities that require manual intervention indicate a process that involves human oversight or adjustments, which strays from the automatic and predefined nature of rule-based entities. Thus, the choice identifying entities based on standard or custom-defined rules accurately captures the essence of pre-trained entities within this context.