Repository files navigation

This repo contains various ways to calculate the similarity between source and target sentences. You can choosethe pre-trained models you want to use such asELMo,BERT andUniversal Sentence Encoder (USE).

And you can also choosethe method to be used to get the similarity:

1. Cosine similarity2. Manhattan distance3. Euclidean distance4. Angular distance5. Inner product6. TS-SS score7. Pairwise-cosine similarity8. Pairwise-cosine similarity + IDF

You can experiment with (The number of models) x (The number of methods) combinations!

• This project is developed underconda enviroment
• After cloning this repository, you can simply install all the dependent libraries described inrequirements.txt withbash install.sh

conda create -n sensim python=3.7conda activate sensimgit clone https://github.com/Huffon/sentence-similarity.gitcd sentence-similaritybash install.sh

• Totest your own sentences, you should fill outcorpus.txt with sentences as below:

I ate an apple.I went to the Apple.I ate an orange....

• Then,choose themodel andmethod to be used to calculate the similarity between source and target sentences

python sensim.py    --model    MODEL_NAME  [use, bert, elmo]    --method   METHOD_NAME [cosine, manhattan, euclidean, inner,                            ts-ss, angular, pairwise, pairwise-idf]    --verbose  LOG_OPTION (bool)

• In this section, you can see the example result ofsentence-similarity
• As you know, there is a nosilver-bullet which can calculateperfect similarity between sentences
• You should conduct various experiments with your dataset
  • Caution:TS-SS score might not fit withsentence similarity task, since this method originally devised to calculate the similarity between long documents
• Result:

• Universal Sentence Encoder
• Deep contextualized word representations
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
• Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
• BERTScore: Evaluating Text Generation with BERT
• A Hybrid Geometric Approach for Measuring Similarity Level Among Documents and Document Clustering

• TF-hub's Universal Sentence Encoder
• Allen NLP's ELMo
• Sentence Transformers
• BERTScore
• Vector Similarity

• An Overview of Sentence Embedding Methods
• Comparing Sentence Similarity Methods
• The Current Best of Universal Word Embeddings and Sentence Embeddings