Saved in:
Bibliographic Details
Main Authors: Fisher, Jillian, Neville, Jennifer, Park, Chan Young
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.14473
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911596774686720
author Fisher, Jillian
Neville, Jennifer
Park, Chan Young
author_facet Fisher, Jillian
Neville, Jennifer
Park, Chan Young
contents A common approach to personalization in large language models (LLMs) is to incorporate a subset of the user memory into the prompt at inference time to guide the model's generation. Existing methods select these subsets primarily using similarity between user memory items and input queries, ignoring how features actually affect the model's response distribution. We propose Response-Utility optimization for Memory Selection (RUMS), a novel method that selects user memory items by measuring the mutual information between a subset of memory and the model's outputs, identifying items that reduce response uncertainty and sharpen predictions beyond semantic similarity. We demonstrate that this information-theoretic foundation enables more principled user memory selection that aligns more closely with human selection compared to state-of-the-art methods, and models $400\times$ larger. Additionally, we show that memory items selected using RUMS result in better response quality compared to existing approaches, while having up to $95\%$ reduction in computational cost.
format Preprint
id arxiv_https___arxiv_org_abs_2604_14473
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Response-Aware User Memory Selection for LLM Personalization
Fisher, Jillian
Neville, Jennifer
Park, Chan Young
Artificial Intelligence
A common approach to personalization in large language models (LLMs) is to incorporate a subset of the user memory into the prompt at inference time to guide the model's generation. Existing methods select these subsets primarily using similarity between user memory items and input queries, ignoring how features actually affect the model's response distribution. We propose Response-Utility optimization for Memory Selection (RUMS), a novel method that selects user memory items by measuring the mutual information between a subset of memory and the model's outputs, identifying items that reduce response uncertainty and sharpen predictions beyond semantic similarity. We demonstrate that this information-theoretic foundation enables more principled user memory selection that aligns more closely with human selection compared to state-of-the-art methods, and models $400\times$ larger. Additionally, we show that memory items selected using RUMS result in better response quality compared to existing approaches, while having up to $95\%$ reduction in computational cost.
title Response-Aware User Memory Selection for LLM Personalization
topic Artificial Intelligence
url https://arxiv.org/abs/2604.14473