Text Similarity with tensorflow.hub

Hello, I’m workink on the text similarity between two variables which have str and number as a type(police number and text about explaination).
So, I found a tutorial on TensorFlow Blog on semantic with tf_hub and I want to know if is good for my project.

HI Armel,

if you the problem you are trying to solve is: given sentence A and Sentence B, are they a paraphrase of each other?
is this it?

if so, you can use this: Solve GLUE tasks using BERT on TPU  |  Text  |  TensorFlow

more specifically the QQP benchmark.

1 Like

Hello Gus,

Thank you for your help. I checked your link and I have a question: Paraphrase means similarity? If yes, it work.

So, here is my work in screenshot before getting your link. It’s about give a similarity score to 2 sentences depending police number (Police number is just columns name from my dataset).

Yes, paraphrase means express the meaning using different words so I guess it might help you.
If you look into the dataset for some of the training data, you will get the idea.

(you didn’t attached the image)

It’s ok . So, I can’t upload image, I get this message error “Sorry, you can’t embed media items in a post.” sorry

I guess it’s because it depends on more activity on the forum.

I think so. Thank you for yor help.

Yes, check Understanding Discourse Trust Levels

1 Like