The DeepMoji project by @bfelbo et. al. uses a neural network trained on billions of tweets to select emoji that most accurately represent the emoji of input text. Their code is written in Python 2.7 using Tensorflow, and is available on Github
This repository is a docker container for a web service to call the neural network as an API. Forked.
After deploying the container to 127.0.0.1, you can send a POST request with sentences to find the emoji for.
curl -X POST \
http://127.0.0.1 \
-H 'Cache-Control: no-cache' \
-H 'Content-Type: application/json' \
-H 'Postman-Token: 748a5055-d93a-4c2a-91ef-1b3bd1135845' \
-d '{
"sentences": ["I love this band!", "I hate this album."]
}'
The respose is, for each sentence, the a set of tuples containing the emoji, and the probability that emoji is related to the sentence:
{
"emoji": [
[
{
"emoji": "😅",
"prob": 0.00611169869080185890
},
{
"emoji": "😒",
"prob": 0.00045027132728137076
},
{
"emoji": "😫",
"prob": 0.00971820019185543060
},
{
"emoji": "😭",
"prob": 0.01060504186898469925
}, ...
Content type
Image
Digest
Size
1.6 GB
Last updated
over 7 years ago
docker pull thandaanda/deepmoji