Help: Which ML/DL model approach is most suitable for my problem?

[amarpredicts]

Hello dear community,

I am a student and a total beginner in the topics ML and neural networks.
For my first real Uni project I need to build my own ML/DL model from scratch.

It should have a context sensitive algorithm to distinguish between person names and objects. For example

Input:

"Alfred Winter (a person name) wants to visit the USA in the Winter (a Season)".

At the end, I should only replace the identified person names with asterisk characters and thus anonymize them like this output:

"***** ***** wants to visit the USA in the Winter"

My problem is that I can't use a predefined model like Spacy because it was already used earlier. So I have to build a model from scratch.
Some of my colleagues have recommended me to use RNN (Recurrent Neural Networks) and LSTM (Long short-term memory). Others say I should concentrate on NER (Named Entity Recognition).
But I am not very sure since I have little to no experience, so this all is quite confusing for a beginner and i dont know where to start. I would also like to know, which one of the above just mentioned approaches is the simplest direction before deep diving into the topic.

Does anyone have good tips on how I can best approach my project?
Thank you in advance!

[mrdbourke]

Hey there!

NER is where I would head first.

You'd want to use a NER model to detect different person's names and then once they're found, replace the strings (the characters which make up their names) with asterisks.

[amarpredicts]

Thank you very much!