this piece has been extracted regularly before, and its function is relatively perfect, but sometimes it needs to be fine-tuned, so if you want to implement an AI, you can automatically extract and learn.

for example, the following data

12345678

123123123

mainly want to extract the name of a certain company or person, a bank and other numbers, but due to the existence of interfering words, regular matching can only be processed manually, so is there any algorithm that can be learned automatically? what is the general direction of the references? I have searched some information now, the main search direction is "how to extract keywords", and I am looking at Word2Vec.

I hope God will give me a reference or search direction. Thank you