TwIPS

Project Overview

TwIPS (Texting with Interpret, Preview, and Suggest) is a research prototype texting application that uses large language models (LLMs) to help autistic users interpret tone and intent in online messaging.

---

Key Features

Interpret

Explains the overall tone and meaning of an incoming message and highlights potentially ambiguous elements (e.g., sarcasm, metaphors, emojis). The UI can mark ambiguous fragments and provide short explanations.

Preview

Lets users preview how their message might be perceived by the recipient (e.g., the recipient's emotional reaction). Preview can flag messages that may come across as blunt.

Suggest

When a draft may come across as blunt, Suggest generates an alternative phrasing that preserves intent while softening the tone.

---

Model & Prompting

TwIPS uses GPT-4 via Azure OpenAI.

---

Evaluation

TwIPS was evaluated in an in-lab study with autistic participants using a two-phase design:

• A scripted conversation to evaluate Preview and Suggest.
• An AI-based simulation to evaluate Interpret.

---

Tech Stack

• React/JS (frontend)
• Node/Express (backend)
• Azure OpenAI API

---

How to Run (Local)

This project has two parts:

• Frontend in public/ (contains its own package.json and src/)
• Backend in server/ (contains its own package.json and index.js)

Prerequisites

• Node.js 16+ recommended
• npm (or yarn)

Install dependencies

Frontend:

cd public
npm install
cd ..

Backend:

cd server
npm install
cd ..

Configure environment variables in LLMInterpretation.jsx

Run (two terminals)

Terminal 1 — backend:

cd server
npm start

Terminal 2 — frontend:

cd public
npm start

Open the UI (commonly http://localhost:3000)

Paper

TwIPS: A Large Language Model Powered Texting Application to Simplify Conversational Nuances for Autistic Users.
In Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2024).
PDF: https://rukhshan23.github.io/twips.pdf

Citation

@inproceedings{10.1145/3663548.3675633,
        author = {Haroon, Rukhshan and Dogar, Fahad},
        title = {TwIPS: A Large Language Model Powered Texting Application to Simplify Conversational Nuances for Autistic Users},
        year = {2024},
        isbn = {9798400706776},
        publisher = {Association for Computing Machinery},
        address = {New York, NY, USA},
        url = {https://doi.org/10.1145/3663548.3675633},
        doi = {10.1145/3663548.3675633},
        booktitle = {Proceedings of the 26th International ACM SIGACCESS Conference on Computers and Accessibility},
        articleno = {24},
        numpages = {18},
        location = {St. John's, NL, Canada},
        series = {ASSETS '24}
        }