Missing Position Prediction + Story Completion Demo System

This is a prototype demonstration system for Human Story Writing Assistance. Based on the "Missing Position Prediction (MPP)" approach we have proposed [1][2].

You can run this demo system on your own computing environment or on Colaboratory.

Run on your own Computing Environment

Requirements

You can install the required packages as follows.

pip install -r requirements.txt

After preparing an environment, please see the trained_model directory and download the trained model for running the demo system.

Run

streamlit run mppsc-streamlit.py

Run on Colaboratory

You can quickly run the system on Colaboratory (Google Colab) by just getting ngrok auth token from https://dashboard.ngrok.com/auth.
Please open mppsc_demo_streamlit.ipynb in Colab, then input your auth token appropriately. Run all cells in the file, and now you can access the demo system.

References

1. Yusuke Mori, Hiroaki Yamane, Yusuke Mukuta, Tatsuya Harada, “Finding and Generating a Missing Part for Story Completion,” 4th Joint SIGHUM Workshop on Computational Linguistics for Cultural Heritage, Social Sciences, Humanities and Literature (LaTeCH-CLfL) (COLING 2020, Workshop), 2020. [PDF] [Code]
2. Yusuke Mori, Hiroaki Yamane, Yusuke Mukuta, Tatsuya Harada, “The Nectar of Missing Position Prediction for Story Completion,” Text2Story 2021 (ECIR 2021, Workshop), 2021. [Video] [(PDF is to appear)]