LAUNCH Lab
LAUNCH = LAnguage Understanding and generatioN researCH
The LAUNCH lab aims to build efficient and robust Natural Language Processing (NLP) models to extract knowledge from large-scale texts as well as teach machines to talk and write like humans do. Our goal is to make machines more intelligent through creating natural language understanding and generation tools.
Our lab has active projects in summarizing documents of various domains and genres, generating text with complex narratives, characterizing human arguments and reasoning process, extracting fine-grained opinions from user-generated content, and solving interdisciplinary challenges in computational social science and AI for education.
We can be found on Twitter as @launchnlp.
📋 If you're interested in joining the LAUNCH lab for research projects, please check out this page.
Recent News
[November 2023] Received grants from LG AI and Cisco to work on "Effective and Fine-grained Feedback for Enhanced Language Model Reasoning and Alignment" and "Multi-Document Reasoning with Large Language Models".
[July 2023] Received a grant from NSF to work on "Argument Graph Supported Multi-Level Approach for Argumentative Writing Assistance".
[June 2023] Check out our new preprint on efficient long document summarization and reasoning with large language models.
[June 2023] Check out our recent work on fast inference-time controlled text generation and word category arcs in narratives.
[June 2023] Shuyang Cao won the Bloomberg Data Science Ph.D. Fellowship for 2023-2024. CSE news
[April 2023] ReadingQuizMaker, our collaborative work with Xu Wang's group, wins a Best Paper Honorable Mention Award at CHI 2023.
[Feb 2023] Check out our new preprint on multi-hop question answering using large language models (LLMs).
[Nov 2022] Check out our recent work on entity-to-entity stance detection, multimodal ideological content analysis, media bias study, time-aware prompting for generation, and generative models for aspect-based sentiment analysis.
[Sep 2022] Check out our ideology prediction and stance detection tool, which is made available online: https://politics.eecs.umich.edu/ (Paper, Codebase, Huggingface).
[Aug 2022] Received a grant from LG AI Research to work on "Knowledge-grounded Scientific Reasoning".
[May 2022] Received a grant from AFOSR to work on "Computational Modeling of Narrative Representation, Shaping, and Influence". CSE news
[Apr 2022] Check out our recent work on ideology prediction and stance detection and question generation system design for education.
[Mar 2022] Check out our recent work on long document summarization with hierarchical question-summary pairs and efficient structure extraction methods for argument mining.
[Feb 2022] Welcome Winston Wu to join the lab as Postdoc researcher. He will work on cross culture narrative understanding (together with Rada Mihalcea) and media content analysis.
[Jan 2022] Shuyang Cao received an award of $50k Oracle for Research Cloud Credits for improving factuality of summarization systems.