Package for extracting emotions from social media text. Tailored for financial data.

Overview

EmTract: Extracting Emotions from Social Media Text Tailored for Financial Contexts

EmTract is a tool that extracts emotions from social media text. It incorporates key aspects of social media data (e.g., non-standard phrases, emojis and emoticons), and uses cutting edge natural language processing (NLP) techniques to learn latent representations, such as word order, word usage, and local context, to predict the emotions.

Details on the model and text processing are in the appendix of EmTract: Investor Emotions and Market Behavior.

User Guide

Installation

Before being able to use the package python3 must be installed. We also recommend using a virtual environment so that the tool runs with the same dependencies with which it was developed. Instruction on how to set up a virtual environment can be found here.

Once basic requirements are setup, follow these instructions:

  1. Clone the repository: git clone https://github.com/dvamossy/EmTract.git
  2. Navigate into repository: cd EmTract
  3. (Optional) Create and activate virtual environment:
    python3 -m venv venv
    source venv/bin/activate
    
  4. Run ./install.sh. This will install python requirements and also download our model files

Usage

Our package should be run with the following command:

python3 -m emtract.inference [args]

Where args are the following:

  • --model_type: can be twitter or stocktwits. Default is stocktwits
  • --interactive: Run in interactive mode
  • --input_file/-i: input to use for predictions (only for non interactive mode)
  • --output_file/-o: output location for predictions(only for non interactive mode)

Output

For each input (i.e., text), EmTract outputs probabilities (they sum to 1!) corresponding to seven emotional states: neutral, happy, sad, anger, disgust, surprise, fear. It also labels the text by computing the argmax of the probabilities.

Modes

Our tool can be run in 2 execution modes.

Interactive mode allows the user to input a tweet and evaluate it in real time. This is great for exploratory analysis.

python3 -m emtract.inference --interactive

The other mode is intended for automating predictions. Here an input file must be specified that will be used as the prediction input. This file must be a csv or text file with 1 column. This column should have the messages/text to predict with.

python3 -m emtract.inference -i tweets_example.csv -o predictions.csv

Model Types

Our models leverage GloVe Embeddings with Bidirectional GRU architecture.

We trained our emotion models with 2 different data sources. One from Twitter, and another from StockTwits. The Twitter training data comes from here; it is available at data/twitter_emotion.csv. The StockTwits training data is explained in the paper.

One of the key concerns using emotion packages is that it is unknown how well they transfer to financial text data. We alleviate this concern by hand-tagging 10,000 StockTwits messages. These are available at data/hand_tagged_sample.parquet.snappy; they were not included during training any of our models. We use this for testing model performance, and alternative emotion packages (notebooks/Alternative Packages.ipynb).

We found our StockTwits model to perform best on the hand-tagged sample, and therefore it is used as the default for predictions.

Alternative Models

We also have an implementation of DistilBERT in notebooks/Alternative Models.ipynb on the Twitter data; which can be easily extended to any other state-of-the-art models. We find marginal performance gains on the hand-tagged sample, which comes at the cost of far slower inference.

Citation

If you use EmTract in your research, please cite us as follows:

Domonkos Vamossy and Rolf Skog. EmTract: Investor Emotions and Market Behavior https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3975884, 2021.

Contributing and Feedback

This project welcomes contributions and suggestions.

Our goal is to provide a unified framework for extracting emotions from financial social media text. Particularly useful for research on emotions in financial contexts would be labeling financial social media text. We plan to upload sample text upon request.

Code for "Learning From Multiple Experts: Self-paced Knowledge Distillation for Long-tailed Classification", ECCV 2020 Spotlight

Learning From Multiple Experts: Self-paced Knowledge Distillation for Long-tailed Classification Implementation of "Learning From Multiple Experts: Se

27 Nov 05, 2022
ARAE-Tensorflow for Discrete Sequences (Adversarially Regularized Autoencoder)

ARAE Tensorflow Code Code for the paper Adversarially Regularized Autoencoders for Generating Discrete Structures by Zhao, Kim, Zhang, Rush and LeCun

19 Nov 12, 2021
Explaining Hyperparameter Optimization via PDPs

Explaining Hyperparameter Optimization via PDPs This repository gives access to an implementation of the methods presented in the paper submission “Ex

2 Nov 16, 2022
HyperDict - Self linked dictionary in Python

Hyper Dictionary Advanced python dictionary(hash-table), which can link it-self

8 Feb 06, 2022
Official implementation of Rethinking Graph Neural Architecture Search from Message-passing (CVPR2021)

Rethinking Graph Neural Architecture Search from Message-passing Intro The GNAS can automatically learn better architecture with the optimal depth of

Shaofei Cai 48 Sep 30, 2022
OpenFace – a state-of-the art tool intended for facial landmark detection, head pose estimation, facial action unit recognition, and eye-gaze estimation.

OpenFace 2.2.0: a facial behavior analysis toolkit Over the past few years, there has been an increased interest in automatic facial behavior analysis

Tadas Baltrusaitis 5.8k Dec 31, 2022
Python Multi-Agent Reinforcement Learning framework

- Please pay attention to the version of SC2 you are using for your experiments. - Performance is *not* always comparable between versions. - The re

whirl 1.3k Jan 05, 2023
Object tracking implemented with YOLOv4, DeepSort, and TensorFlow.

Object tracking implemented with YOLOv4, DeepSort, and TensorFlow. YOLOv4 is a state of the art algorithm that uses deep convolutional neural networks to perform object detections. We can take the ou

The AI Guy 1.1k Dec 29, 2022
Calibrate your listeners! Robust communication-based training for pragmatic speakers. Findings of EMNLP 2021.

Calibrate your listeners! Robust communication-based training for pragmatic speakers Rose E. Wang, Julia White, Jesse Mu, Noah D. Goodman Findings of

Rose E. Wang 3 Apr 02, 2022
Punctuation Restoration using Transformer Models for High-and Low-Resource Languages

Punctuation Restoration using Transformer Models This repository contins official implementation of the paper Punctuation Restoration using Transforme

Tanvirul Alam 142 Jan 01, 2023
Audio Domain Adaptation for Acoustic Scene Classification using Disentanglement Learning

Audio Domain Adaptation for Acoustic Scene Classification using Disentanglement Learning Reference Abeßer, J. & Müller, M. Towards Audio Domain Adapt

Jakob Abeßer 2 Jul 06, 2022
Google Recaptcha solver.

byerecaptcha - Google Recaptcha solver. Model and some codes takes from embium's repository -Installation- pip install byerecaptcha -How to use- from

Vladislav Zenkevich 21 Dec 19, 2022
Vector Quantized Diffusion Model for Text-to-Image Synthesis

Vector Quantized Diffusion Model for Text-to-Image Synthesis Due to company policy, I have to set microsoft/VQ-Diffusion to private for now, so I prov

Shuyang Gu 294 Jan 05, 2023
Implementation of SE3-Transformers for Equivariant Self-Attention, in Pytorch.

SE3 Transformer - Pytorch Implementation of SE3-Transformers for Equivariant Self-Attention, in Pytorch. May be needed for replicating Alphafold2 resu

Phil Wang 207 Dec 23, 2022
Pytorch Implementation of the paper "Cross-domain Correspondence Learning for Exemplar-based Image Translation"

CoCosNet Pytorch Implementation of the paper "Cross-domain Correspondence Learning for Exemplar-based Image Translation" (CVPR 2020 oral). Update: 202

Lingbo Yang 38 Sep 22, 2021
Select, weight and analyze complex sample data

Sample Analytics In large-scale surveys, often complex random mechanisms are used to select samples. Estimates derived from such samples must reflect

samplics 37 Dec 15, 2022
A cross-document event and entity coreference resolution system, trained and evaluated on the ECB+ corpus.

A Comprehensive Comparison of Word Embeddings in Event & Entity Coreference Resolution. Introduction This repo contains experimental code derived from

2 May 09, 2022
Sibur challange 2021 competition - 6 place

sibur challange 2021 Решение на 6 место: https://sibur.ai-community.com/competitions/5/tasks/13 Скор 1.4066/1.4159 public/private. Архитектура - однос

Ivan 5 Jan 11, 2022
Understanding Convolutional Neural Networks from Theoretical Perspective via Volterra Convolution

nnvolterra Run Code Compile first: make compile Run all codes: make all Test xconv: make npxconv_test MNIST dataset needs to be downloaded, converted

1 May 24, 2022
Graph WaveNet apdapted for brain connectivity analysis.

Graph WaveNet for brain network analysis This is the implementation of the Graph WaveNet model used in our manuscript: S. Wein , A. Schüller, A. M. To

4 Dec 17, 2022