GISTBench: Evaluating LLM User Understanding via Evidence-Based Interest Verification
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914433367801856 |
|---|---|
| author | Fostiropoulos, Iordanis Azhar, Muhammad Rafay Sawwan, Abdalaziz Fang, Boyu Liu, Yuchen Liu, Jiayi Yu, Hanchao Guo, Qi Wang, Jianyu Liu, Fei Fan, Xiangjun |
| author_facet | Fostiropoulos, Iordanis Azhar, Muhammad Rafay Sawwan, Abdalaziz Fang, Boyu Liu, Yuchen Liu, Jiayi Yu, Hanchao Guo, Qi Wang, Jianyu Liu, Fei Fan, Xiangjun |
| contents | We introduce GISTBench, a benchmark for evaluating Large Language Models' (LLMs) ability to understand users from their interaction histories in recommendation systems. Unlike traditional RecSys benchmarks that focus on item prediction accuracy, our benchmark evaluates how well LLMs can extract and verify user interests from engagement data. We propose two novel metric families: Interest Groundedness (IG), decomposed into precision and recall components to separately penalize hallucinated interest categories and reward coverage, and Interest Specificity (IS), which assesses the distinctiveness of verified LLM-predicted user profiles. We release a synthetic dataset constructed on real user interactions on a global short-form video platform. Our dataset contains both implicit and explicit engagement signals and rich textual descriptions. We validate our dataset fidelity against user surveys, and evaluate eight open-weight LLMs spanning 7B to 120B parameters. Our findings reveal performance bottlenecks in current LLMs, particularly their limited ability to accurately count and attribute engagement signals across heterogeneous interaction types. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_29112 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GISTBench: Evaluating LLM User Understanding via Evidence-Based Interest Verification Fostiropoulos, Iordanis Azhar, Muhammad Rafay Sawwan, Abdalaziz Fang, Boyu Liu, Yuchen Liu, Jiayi Yu, Hanchao Guo, Qi Wang, Jianyu Liu, Fei Fan, Xiangjun Artificial Intelligence Computation and Language H.3.3; I.2.7 We introduce GISTBench, a benchmark for evaluating Large Language Models' (LLMs) ability to understand users from their interaction histories in recommendation systems. Unlike traditional RecSys benchmarks that focus on item prediction accuracy, our benchmark evaluates how well LLMs can extract and verify user interests from engagement data. We propose two novel metric families: Interest Groundedness (IG), decomposed into precision and recall components to separately penalize hallucinated interest categories and reward coverage, and Interest Specificity (IS), which assesses the distinctiveness of verified LLM-predicted user profiles. We release a synthetic dataset constructed on real user interactions on a global short-form video platform. Our dataset contains both implicit and explicit engagement signals and rich textual descriptions. We validate our dataset fidelity against user surveys, and evaluate eight open-weight LLMs spanning 7B to 120B parameters. Our findings reveal performance bottlenecks in current LLMs, particularly their limited ability to accurately count and attribute engagement signals across heterogeneous interaction types. |
| title | GISTBench: Evaluating LLM User Understanding via Evidence-Based Interest Verification |
| topic | Artificial Intelligence Computation and Language H.3.3; I.2.7 |
| url | https://arxiv.org/abs/2603.29112 |