ScribWord

6.835: Intelligent Multimodal User Interfaces - Final Project

File structure

├── public                      # Folder with publicly available resources
│   ├── css                     # Folder containing .css files
│   |    ├── style.css              # Stylizes all .html files in the application
│   ├── img                     # Folder containing .gif files
│   |    ├── mic-animate.gif        # .gif for when mic is recording
│   |    ├── mic-slash.gif          # .gif for when mic is not available
│   |    ├── mic.gif                # .gif for when mic is available
│   ├── js                      # Folder containing .js files
│   |    ├── history.js             # .js file for history.html
│   |    ├── index.js               # .js file for index.html
│   |    ├── setupSpeech.js         # .js file for continuous speech recognition used in history.html
├── topic_modelling             # Folder containing topic modeling files
│   |    ├── venv               # Virtual environment for Python3
│   |    ├── compute_input.py   # Interfaces between topic modeling backend and app.js for frontend
│   |    ├── guided_lda.py      # One of the approaches to topic modeling. --Unused
│   |    ├── guidedlda_model.pkl  # Pickled model of guided lda topic model --Unused
│   |    ├── lda.py             # One of the approaches to topic modeling. --Used
│   |    ├── requirements.txt   # Requirements for Python3
│   |    ├── test_we.py         # Tests word embedding implementation
│   |    ├── topic_modelling.py # Uses pretrained models to categorize input text with topics
│   |    ├── vectorize.py       # Converts words to vectors based on training corpus
│   |    ├── vectorize_on_brown.pkl # Pickled model of vectorizer trained on brown corpus
│   |    ├── vectorizer.pkl     # Pickled model of vectorizer
│   |    ├── word_embedding.py  # Attempt at using neural network for topic modeling --Unused
├── .gitignore                  # Lists files or types of files to ignore when committing via git
├── README.md             
├── app.js                      # Holds node.js server
├── data.json                   # .json file to store conversations in memory
├── history.html                # Holds html for Past Conversations tab
├── index.html                  # Holds html for Create Conversations tab
├── package-lock.json           # Automatically generated when npm modifies either the node_modules or package.json
├── package.json                # File containing all dependencies for node.js app
└── vectorize_on_brown.pkl      # Pickled model of vectorization model trained on brown corpus

Instructions to run code

1. Clone the repository.
2. Install node.js version recommended for most users
3. Run npm install to install dependencies. This will create a node_modules file at the root.
4. Install Python 3.6
5. Run pip3 install -r requirements.txt in /topic_modelling/ to install required dependencies for topic modeling.
6. Run npm start to start Express server.

Other notes

No need to train lda.py or guided_lda.py. Simply run either of them using python3. Test lda.py by changing text on line 42 to a string passage of your choice. Test guided_lda.py by changing text on line 66 to a string passage of your choice.