Summary of Comparing Pre-trained Human Language Models: Is It Better with Human Context As Groups, Individual Traits, or Both?, by Nikita Soni et al.
Comparing Pre-trained Human Language Models: Is it Better with Human Context as Groups, Individual Traits, or Both?
by Nikita Soni, Niranjan Balasubramanian, H. Andrew Schwartz, Dirk Hovy
First submitted to arxiv on: 23 Jan 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
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 In this paper, researchers investigate the limitations of pre-trained language models in understanding human context. Current models consider neighboring words and documents but neglect the author’s personal attributes, social situations, and environmental factors. To bridge this gap, two approaches are explored: group-wise attributes (e.g., over-45-year-olds) and individual traits. While group attributes provide a simple yet coarse representation, individual traits require more complex modeling and data. The study compares these approaches on five user- and document-level tasks, revealing that no single approach outperforms the others. Instead, human-centered language modeling offers avenues for different methods. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how well computers can understand what people are writing about. Right now, computer models think about the words around them but don’t know anything about who wrote it or why. To make things better, researchers try two ways to add human information: group attributes (like being over 45) and individual traits (like someone’s personality). The study tests these methods on different tasks and finds that there isn’t one “best” way. Instead, making language models more human-centered can lead to new ideas. |
