Summary of Combining Hierachical Vaes with Llms For Clinically Meaningful Timeline Summarisation in Social Media, by Jiayu Song et al.
Combining Hierachical VAEs with LLMs for clinically meaningful timeline summarisation in social media
by Jiayu Song, Jenny Chim, Adam Tsakalidis, Julia Ive, Dana Atzil-Slonim, Maria Liakata
First submitted to arxiv on: 29 Jan 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper introduces a novel hybrid approach for generating clinically meaningful summaries from social media user timelines, suitable for mental health monitoring. The proposed method combines hierarchical Variational Autoencoder (VAE) with Large Language Models (LLMs), specifically LlaMA-2. The generated summaries are composed of two narrative points of view: clinical insights in third person, useful for clinicians, and a temporally sensitive abstractive summary of the user’s timeline in first person. The proposed method is assessed through automatic evaluation against expert summaries and human evaluation with clinical experts, showing that the hierarchical VAE (TH-VAE) results in more factual and logically coherent summaries rich in clinical utility, outperforming LLM-only approaches in capturing changes over time. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper develops a new way to summarize what people write on social media about their mental health. The goal is to help doctors track patients’ mental well-being by generating accurate and helpful summaries. Two types of summaries are created: one that provides clinical insights for doctors, and another that tells the story of the user’s timeline in their own words. The new approach combines two techniques: a type of machine learning called hierarchical VAE and large language models like LlaMA-2. Researchers tested this method by comparing it to expert-written summaries and asking doctors to evaluate its usefulness, showing that it produces more accurate and logical summaries that can help doctors understand patients’ mental health. |
Keywords
» Artificial intelligence » Llama » Machine learning » Variational autoencoder
