Sentence to Emoji Translation (or 📄🗜➡️😆)

A novel chunking approach to summarizing a sentence to emoji translation using sent2vec and some of emoji2vec's dataset. This is a final senior research project for my undergraduate degree at Clarion University. The paper and explanation of the algorithm can be seen on my website.

Folder Structure

• EmojiTranslation ➡️ Emoji Translation module
• JupyterNotebooks ➡️ Literate programming-esque emoji translation algorithm
• plots ➡️ Plots for the paper

Getting off the Ground

• This project requires Python 3
• Install dependencies with pip install -r requirements.txt
• Install sent2vec according to the github repo
• Download a sent2vec model to the models/ directory
• Download the emoji2vec dataset to the data/ directory

Jupyter Notebook

The JupyterNotebook directory contains all of the Jupyter Notebooks that were used to develop this algorithm. Most of them are accurately named and a lot of them have a ton of comments and markdown to explain everything. the command cd JupyterNotebooks && jupyter notebook starts the jupyter notebook server and allows you to explore the algorithm

Module

The module contains the two separate algorithms in two separate classes. To use them simply import the Translators submodule from the EmojiTranslation module. In that submodule there are two classes of note ExhaustiveChunkingTranslation and PartOfSpeechEmojiTranslator. They both are instantiated with three paramters. An example of use is shown below

# Import the translation submodule 
from EmojiTranslation import Translators

# Paramererize the constructor arguments
emoji_file = "./data/emoji_joined.txt"         # Emoji keyword file from emoji2vec
sent2vec_model = "../models/wiki_unigrams.bin" # sent2vec model
nothing_lemma_func = lambda x: x               # Lemmatization function that does nothing

# Instantiate the exhaustive and part of speech translation algorithms
exh = Translators.ExhaustiveChunkingTranslation(emoji_file, sent2vec_model, nothing_lemma_func)
pos = Translators.PartOfSpeechEmojiTranslator(emoji_file, sent2vec_model, nothing_lemma_func)

# Translate a sentence
sent = "the dog ran fast"
print(exh.summarize(sent))
print(pos.summarize(sent))

Citation

If you want to cite this work please use the following bibtex:

@inproceedings{day2020confet,
  title           = {CoNFET: An English Sentence to Emojis Translation
                  Algorithm},
  author          = {Alex Day and Chris Mankos and Soo Kim and Jody Strausser},
  booktitle       = {Proceedings of the 35th Annual Spring Conference of the
                  Pennsylvania Computer and Information Science Educators},
  organization    = {Pennsylvania Computer and Information Science Educators},
  year            = 2020,
  month           = {April},
}