Getting Started

(Recommended operating systems: macOS or Linux)

Follow the steps provided below in order to get everything set up:

1. Install Python

$ python3 --version
Python 3.6.7rc1

2. Install Git and download this repository

$ sudo apt install git
$ git clone https://github.com/jumayel06/Tension-Analysis.git

3. Download the following files from this link: https://goo.gl/PUDAKv

• model.h5
• variables.p
• stanford-corenlp-full-2018-02-27.zip

4. Copy the first two files (model.h5 and variables.p) in a new folder called models.
5. Copy the last file (stanford-corenlp-full-2018-02-27.zip) in the resources folder and unzip it.
6. Install the following packages from terminal:

$ sudo apt install python3-pip
$ sudo apt install default-jre
$ pip3 install bs4
$ pip3 install mammoth

7. Install nltk and other dependencies

$ pip3 install nltk
$ python3
> import nltk
> nltk.download('punkt')
> nltk.download('wordnet')
> exit()

8. Install other packages

$ pip3 install psutil
$ pip3 install vaderSentiment
$ pip3 install numpy
$ pip3 install keras
$ pip3 install tensorflow
$ pip3 install emoji
$ pip3 install sklearn

For Users

In order to use the tension analysis tool, run the following command on the terminal:

$ python3 main.py models/model.h5 models/variables.p datasets/interview_transcripts/BertheKayitesi.docx output.csv

NOTE: The above command requires five arguments:

1. main.py (this file can be found in Tension Analysis folder)
2. models/model.h5 (relative path for the trained model for emotion recognition)
3. models/variables.p (relative path for stored variables)
4. datasets/interview_transcripts/BertheKayitesi.docx (path to the interview file)
5. output.csv (Name of the output file)

The above command will analyze the provided interview transcript and generate a csv file similar like this:

Important: Provided interview transcript should be in the right format for this tool to work correctly.

Tutorial: https://goo.gl/cegbjB