Predict if a word is an ending of sentence

Hi everyone,

I want to predict if a word is an ending of a sentence or not in a paragraph; I mean if a word is an ending it should be predicted as 1 else 0. It’s a binary classification problem.

Is there any model for that?


Why do you need a model for that in the first place? You can just use a parser that does sentence segmentation (e.g. spacy or stanza) and then select the last word from it.

Hi Bram,

Thanks for your reply. That’s not exactly what I am trying to solve, sorry for not posting the original problem. Actually, I am trying to do Scene segmentation for Movies, in which shots’ representations are given, I have to predict if a shot is the ending of a scene; where the scene boundary detection could be regarded as a binary classification problem on shot boundaries.

Please check this link:

I want to use a RNN or a BERT model for which I want to input the representations of the shots i.e. like embeddings for each word. I want to make use of the output of each RNN hidden unit (with each shot as input).

In that case you use any XXXForTokenClassification model with two possible labels: end of scene or not. This example is a good start.

Thanks, Bram. I will try that.