According to Me: Long-Term Personalized Referential Memory QA

TL;DR

This paper introduces ATM-Bench, a comprehensive benchmark for evaluating multimodal, multi-source personalized memory question answering, highlighting the challenges and potential improvements in long-term personalized AI assistant memory systems.

Contribution

The paper presents ATM-Bench, the first benchmark for multimodal, multi-source personalized memory QA, and proposes Schema-Guided Memory to better represent diverse memory sources.

Findings

Current models perform poorly on complex personalized memory tasks.

Schema-Guided Memory improves over traditional descriptive memory.

State-of-the-art systems achieve under 20% accuracy on challenging tasks.

Abstract

Personalized AI assistants must recall and reason over long-term user memory, which naturally spans multiple modalities and sources such as images, videos, and emails. However, existing Long-term Memory benchmarks focus primarily on dialogue history, failing to capture realistic personalized references grounded in lived experience. We introduce ATM-Bench, the first benchmark for multimodal, multi-source personalized referential Memory QA. ATM-Bench contains approximately four years of privacy-preserving personal memory data and human-annotated question-answer pairs with ground-truth memory evidence, including queries that require resolving personal references, multi-evidence reasoning from multi-source and handling conflicting evidence. We propose Schema-Guided Memory (SGM) to structurally represent memory items originated from different sources. In experiments, we implement 5 state-of-the-art memory systems along with a standard RAG baseline and evaluate variants with different memory ingestion, retrieval, and answer generation techniques. We find poor performance (under 20\% accuracy) on the ATM-Bench-Hard set, and that SGM improves performance over Descriptive Memory commonly adopted in prior works. Code available at: https://github.com/JingbiaoMei/ATM-Bench