ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries

Junhyuk Kwon, Seungjoon Lee, Hyejin Park, Kyle Min, Jungseul Ok · 2026

Read ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries by Junhyuk Kwon, Seungjoon Lee, Hyejin Park, Kyle Min, Jungseul Ok (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-06223.txt.

Description

Junhyuk Kwon and four co-authors build a navigation agent that asks smart questions when a user's request is ambiguous. A request may match several similar objects, and demanding a full description up front burdens the user. Worse, agents often lock onto a distractor that fits the gathered details and decide too early, wasting the whole dialogue.

ProCompNav first builds a candidate pool, then narrows it through recursive comparative judgment. Each round picks an attribute that splits the pool, asks the user a binary question, and drops candidates that clash with the answer. No attribute needs to be unique to the target for the process to work. On two benchmarks the method led on success rate while sharply shortening user responses.

Catalog record

CREATORJunhyuk Kwon, Seungjoon Lee, Hyejin Park, Kyle Min, Jungseul Ok
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.06223

Browse the full library · About the Tower of Babel