{"id":"244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba","title":"Junhyuk Kwon and coauthors propose ProCompNav for ambiguous robot navigation","argument":"The paper’s authors propose ProCompNav, a way for agentic robots to identify which of several similar objects a user means without relying on long, open-ended answers. In interactive instance navigation, the robot first builds a candidate pool, then uses Recursive Comparative Judgment (RCJ): it asks binary questions about attribute-value pairs that divide the candidates and removes options inconsistent with the user’s answers. The method does not require an attribute unique to the intended target. The authors report that ProCompNav substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and has the highest Success Rate among evaluated methods on a non-interactive benchmark. In two human studies, participants preferred its binary questions over open-ended ones, and batched delivery over interleaved interaction. The argument is that robot design should consider task performance and the burden and experience of interaction together. The abstract reports benchmark and study outcomes, but does not give their numerical results or details of the evaluation settings.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper's abstract, which does not provide numerical benchmark results or details of the evaluation settings. The abstract reports findings from two human studies but does not describe their participants or procedures.","speaker":{"name":"Kwon, Junhyuk, Lee, Seungjoon, Park, Hyejin, Min, Kyle, Ok, Jungseul","roleAtSource":"Article author"},"source":{"title":"ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2605.06223","publishedAt":"2026-05-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose a two-stage method that builds a candidate pool and identifies the intended target through Recursive Comparative Judgment.","excerpt":"we propose Proactive Instance Navigation with Comparative Judgment (ProCompNav), a two-stage framework that first constructs a candidate pool and then identifies the target through Recursive Comparative Judgment (RCJ).","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"RCJ narrows the candidate pool by asking binary questions about attribute-value pairs that divide the candidates, then removes candidates inconsistent with the user's answer; it does not require an attribute unique to the target.","excerpt":"selecting an attribute-value pair that divides the candidates, asking the user a binary question, and removing inconsistent candidates, without requiring an attribute unique to the target.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"The authors report improved Success Rate and reduced Response Length on an interactive navigation benchmark, and the highest Success Rate among evaluated methods on a non-interactive benchmark.","excerpt":"substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and achieves the highest Success Rate among evaluated methods on a non-interactive benchmark.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and batched delivery to interleaved interaction.","excerpt":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and its batched delivery to interleaved interaction.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"The authors argue that robot design should consider task performance alongside user-friendly interaction strategies.","excerpt":"Our results highlight the need to jointly consider task performance and user-friendly interaction strategies when designing agentic robots.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.06223","checkedAt":"2026-11-08T19:10:10.171Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What benchmark settings and measurements underlie the reported improvements in Success Rate and Response Length?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Junhyuk Kwon and coauthors propose ProCompNav for ambiguous robot navigation","summary":"The paper’s authors propose ProCompNav, a way for agentic robots to identify which of several similar objects a user means without relying on long, open-ended answers. In interactive instance navigation, the robot first builds a candidate pool, then uses Recursive Comparative Judgment (RCJ): it asks binary questions about attribute-value pairs that divide the candidates and removes options inconsistent with the user’s answers. The method does not require an attribute unique to the intended target. The authors report that ProCompNav substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and has the highest Success Rate among evaluated methods on a non-interactive benchmark. In two human studies, participants preferred its binary questions over open-ended ones, and batched delivery over interleaved interaction. The argument is that robot design should consider task performance and the burden and experience of interaction together. The abstract reports benchmark and study outcomes, but does not give their numerical results or details of the evaluation settings.","question":"What benchmark settings and measurements underlie the reported improvements in Success Rate and Response Length?","claims":[{"text":"The authors propose a two-stage method that builds a candidate pool and identifies the intended target through Recursive Comparative Judgment.","excerpt":"we propose Proactive Instance Navigation with Comparative Judgment (ProCompNav), a two-stage framework that first constructs a candidate pool and then identifies the target through Recursive Comparative Judgment (RCJ).","support":"source-stated"},{"text":"RCJ narrows the candidate pool by asking binary questions about attribute-value pairs that divide the candidates, then removes candidates inconsistent with the user's answer; it does not require an attribute unique to the target.","excerpt":"selecting an attribute-value pair that divides the candidates, asking the user a binary question, and removing inconsistent candidates, without requiring an attribute unique to the target.","support":"source-stated"},{"text":"The authors report improved Success Rate and reduced Response Length on an interactive navigation benchmark, and the highest Success Rate among evaluated methods on a non-interactive benchmark.","excerpt":"substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and achieves the highest Success Rate among evaluated methods on a non-interactive benchmark.","support":"source-stated"},{"text":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and batched delivery to interleaved interaction.","excerpt":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and its batched delivery to interleaved interaction.","support":"source-stated"},{"text":"The authors argue that robot design should consider task performance alongside user-friendly interaction strategies.","excerpt":"Our results highlight the need to jointly consider task performance and user-friendly interaction strategies when designing agentic robots.","support":"source-stated"}],"limitations":["The supplied evidence is the paper's abstract, which does not provide numerical benchmark results or details of the evaluation settings.","The abstract reports findings from two human studies but does not describe their participants or procedures."],"mediaId":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","correctionReason":null,"id":"244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba","publicationId":"5196fdc1-2997-471d-817c-5033dc382464","version":1,"sourceUrl":"https://arxiv.org/abs/2605.06223","sourceTitle":"ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-05-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-06T05:12:37.574Z","createdAt":"2026-11-08T18:38:54.673Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","kind":"article","url":"https://arxiv.org/abs/2605.06223","sourceUrl":"https://arxiv.org/abs/2605.06223","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-08T18:37:57.780Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.06223","checkedAt":"2026-11-08T19:10:10.171Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","confidence":0.8,"provider":"typesafe","model":"jev-1.13.0","inputHash":"be9e1006b2d7d468e53c281d9c3a72664e073317283ceec90fc6b960c7a36df0","decidedAt":"2026-11-08T18:37:59.758Z"},"provenance":{"textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observationId":"e9e57aeefded69285e1b6876a88e2095ce8a63a41277272df530fee625cd3fa3","retrievedAt":"2026-11-08T18:37:57.313Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"41f0eafb-43c2-4b4f-801b-303b23c87d55"},"reviewer":{"model":"gpt-6-luna","runId":"bcd7d93d-e19b-4b0f-82c2-92be926a1185"}},"audit":{"status":"passed","reviewedAt":"2026-11-08T18:38:54.673Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Kwon, Junhyuk","url":null,"kind":"unknown"},{"name":"Lee, Seungjoon","url":null,"kind":"unknown"},{"name":"Park, Hyejin","url":null,"kind":"unknown"},{"name":"Min, Kyle","url":null,"kind":"unknown"},{"name":"Ok, Jungseul","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Kwon, Junhyuk","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kwon, Junhyuk","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Lee, Seungjoon","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lee, Seungjoon","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Park, Hyejin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Park, Hyejin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Min, Kyle","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Min, Kyle","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Ok, Jungseul","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ok, Jungseul","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2605.06223","snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba/source"}},"canonicalUrl":"https://escapemeridian.com/readings/junhyuk-kwon-and-coauthors-propose-procompnav-for-ambiguous-244e269f","formats":{"html":"/readings/junhyuk-kwon-and-coauthors-propose-procompnav-for-ambiguous-244e269f","markdown":"/readings/junhyuk-kwon-and-coauthors-propose-procompnav-for-ambiguous-244e269f.md","json":"/readings/junhyuk-kwon-and-coauthors-propose-procompnav-for-ambiguous-244e269f.json"},"attribution":"Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.","maintainedGuideLinks":[],"readerContext":{"version":"reader-context-v1","attribution":"Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.","terms":[{"id":"term-agentic","name":"Agentic AI","explanation":"AI systems that plan and carry out a series of steps toward a goal, often by using other software, rather than giving one answer.","referenceUrl":"https://developers.google.com/machine-learning/glossary#agentic"},{"id":"term-benchmark","name":"Benchmark","explanation":"A defined set of tests used to compare systems. Results depend on the tasks and measures selected.","referenceUrl":"https://arxiv.org/abs/2211.09110"}],"entities":[{"id":"arxiv","name":"arXiv","kind":"organization","context":"arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.","referenceUrls":["https://arxiv.org/","https://info.arxiv.org/help/api/index.html"],"path":"/entities/arxiv","json":"/entities/arxiv.json","markdown":"/entities/arxiv.md"}]},"researchQuestions":[],"withdrawn":false,"references":{"schemaVersion":1,"target":{"type":"reading","id":"244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba","revision":"5196fdc1-2997-471d-817c-5033dc382464","revisionKind":"publication-id","title":"Junhyuk Kwon and coauthors propose ProCompNav for ambiguous robot navigation","url":"https://escapemeridian.com/readings/junhyuk-kwon-and-coauthors-propose-procompnav-for-ambiguous-244e269f"},"source":{"type":"source","id":"https://arxiv.org/abs/2605.06223","revision":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","revisionKind":"sha256-source-text","title":"ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries","url":"https://arxiv.org/abs/2605.06223"},"status":"available","items":[],"next":null,"scope":"Explicit citations in public readings and public file attachments to this reading. Shared topics are not citations. Other artifact types need an explicit reference adapter."},"revisions":[{"id":"244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba","title":"Junhyuk Kwon and coauthors propose ProCompNav for ambiguous robot navigation","argument":"The paper’s authors propose ProCompNav, a way for agentic robots to identify which of several similar objects a user means without relying on long, open-ended answers. In interactive instance navigation, the robot first builds a candidate pool, then uses Recursive Comparative Judgment (RCJ): it asks binary questions about attribute-value pairs that divide the candidates and removes options inconsistent with the user’s answers. The method does not require an attribute unique to the intended target. The authors report that ProCompNav substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and has the highest Success Rate among evaluated methods on a non-interactive benchmark. In two human studies, participants preferred its binary questions over open-ended ones, and batched delivery over interleaved interaction. The argument is that robot design should consider task performance and the burden and experience of interaction together. The abstract reports benchmark and study outcomes, but does not give their numerical results or details of the evaluation settings.","topicIds":["intelligence-and-discovery"],"domain":"science","claimType":"attributed reading","limits":"The supplied evidence is the paper's abstract, which does not provide numerical benchmark results or details of the evaluation settings. The abstract reports findings from two human studies but does not describe their participants or procedures.","speaker":{"name":"Kwon, Junhyuk, Lee, Seungjoon, Park, Hyejin, Min, Kyle, Ok, Jungseul","roleAtSource":"Article author"},"source":{"title":"ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries","publisher":"arXiv cs.AI","format":"Article","url":"https://arxiv.org/abs/2605.06223","publishedAt":"2026-05-07","dateBasis":"Date basis not provided."},"evidence":[{"text":"The authors propose a two-stage method that builds a candidate pool and identifies the intended target through Recursive Comparative Judgment.","excerpt":"we propose Proactive Instance Navigation with Comparative Judgment (ProCompNav), a two-stage framework that first constructs a candidate pool and then identifies the target through Recursive Comparative Judgment (RCJ).","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"RCJ narrows the candidate pool by asking binary questions about attribute-value pairs that divide the candidates, then removes candidates inconsistent with the user's answer; it does not require an attribute unique to the target.","excerpt":"selecting an attribute-value pair that divides the candidates, asking the user a binary question, and removing inconsistent candidates, without requiring an attribute unique to the target.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"The authors report improved Success Rate and reduced Response Length on an interactive navigation benchmark, and the highest Success Rate among evaluated methods on a non-interactive benchmark.","excerpt":"substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and achieves the highest Success Rate among evaluated methods on a non-interactive benchmark.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and batched delivery to interleaved interaction.","excerpt":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and its batched delivery to interleaved interaction.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"},{"text":"The authors argue that robot design should consider task performance alongside user-friendly interaction strategies.","excerpt":"Our results highlight the need to jointly consider task performance and user-friendly interaction strategies when designing agentic robots.","url":"https://arxiv.org/abs/2605.06223","relationship":"Source-stated"}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.06223","checkedAt":"2026-11-08T19:10:10.171Z","items":[],"note":"No suitable media was found by the supported page extraction."},"assumptions":[],"openQuestion":{"attribution":"Open question","text":"What benchmark settings and measurements underlie the reported improvements in Success Rate and Response Length?"},"whatWouldChangeView":{"attribution":"Change condition not recorded","text":""},"editorialComparisons":[],"live":{"title":"Junhyuk Kwon and coauthors propose ProCompNav for ambiguous robot navigation","summary":"The paper’s authors propose ProCompNav, a way for agentic robots to identify which of several similar objects a user means without relying on long, open-ended answers. In interactive instance navigation, the robot first builds a candidate pool, then uses Recursive Comparative Judgment (RCJ): it asks binary questions about attribute-value pairs that divide the candidates and removes options inconsistent with the user’s answers. The method does not require an attribute unique to the intended target. The authors report that ProCompNav substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and has the highest Success Rate among evaluated methods on a non-interactive benchmark. In two human studies, participants preferred its binary questions over open-ended ones, and batched delivery over interleaved interaction. The argument is that robot design should consider task performance and the burden and experience of interaction together. The abstract reports benchmark and study outcomes, but does not give their numerical results or details of the evaluation settings.","question":"What benchmark settings and measurements underlie the reported improvements in Success Rate and Response Length?","claims":[{"text":"The authors propose a two-stage method that builds a candidate pool and identifies the intended target through Recursive Comparative Judgment.","excerpt":"we propose Proactive Instance Navigation with Comparative Judgment (ProCompNav), a two-stage framework that first constructs a candidate pool and then identifies the target through Recursive Comparative Judgment (RCJ).","support":"source-stated"},{"text":"RCJ narrows the candidate pool by asking binary questions about attribute-value pairs that divide the candidates, then removes candidates inconsistent with the user's answer; it does not require an attribute unique to the target.","excerpt":"selecting an attribute-value pair that divides the candidates, asking the user a binary question, and removing inconsistent candidates, without requiring an attribute unique to the target.","support":"source-stated"},{"text":"The authors report improved Success Rate and reduced Response Length on an interactive navigation benchmark, and the highest Success Rate among evaluated methods on a non-interactive benchmark.","excerpt":"substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark, and achieves the highest Success Rate among evaluated methods on a non-interactive benchmark.","support":"source-stated"},{"text":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and batched delivery to interleaved interaction.","excerpt":"In two human studies, participants preferred ProCompNav's binary questions to open-ended questions and its batched delivery to interleaved interaction.","support":"source-stated"},{"text":"The authors argue that robot design should consider task performance alongside user-friendly interaction strategies.","excerpt":"Our results highlight the need to jointly consider task performance and user-friendly interaction strategies when designing agentic robots.","support":"source-stated"}],"limitations":["The supplied evidence is the paper's abstract, which does not provide numerical benchmark results or details of the evaluation settings.","The abstract reports findings from two human studies but does not describe their participants or procedures."],"mediaId":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","correctionReason":null,"id":"244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba","publicationId":"5196fdc1-2997-471d-817c-5033dc382464","version":1,"sourceUrl":"https://arxiv.org/abs/2605.06223","sourceTitle":"ProCompNav: Proactive Instance Navigation with Comparative Judgment for Ambiguous User Queries","publisher":"arXiv cs.AI","publishedAt":null,"originalPublished":{"value":"2026-05-07","precision":"day","basis":"publisher_citation_date"},"firstObservedAt":"2026-11-06T05:12:37.574Z","createdAt":"2026-11-08T18:38:54.673Z","historical":true,"entityIds":["arxiv"],"topicIds":["intelligence-and-discovery"],"media":[{"id":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","kind":"article","url":"https://arxiv.org/abs/2605.06223","sourceUrl":"https://arxiv.org/abs/2605.06223","label":"Read original","provenance":"source_link","verification":{"status":"verified","checkedAt":"2026-11-08T18:37:57.780Z","method":"acquired_source","contentType":"text/html"},"recommendationReason":"Recommended by Jev for understanding this source; checked by Luna."}],"sourceMedia":{"status":"not_found","sourceUrl":"https://arxiv.org/abs/2605.06223","checkedAt":"2026-11-08T19:10:10.171Z","items":[],"note":"No suitable media was found by the supported page extraction."},"editorialDecision":{"outcome":"accept","reasonCodes":["material_development"],"rubricVersion":"stream-admission-v5","mediaId":"9723d4cad0451f06e2a222791287673ee1c9184097a669511eb450df1871c2cd","confidence":0.8,"provider":"typesafe","model":"jev-1.13.0","inputHash":"be9e1006b2d7d468e53c281d9c3a72664e073317283ceec90fc6b960c7a36df0","decidedAt":"2026-11-08T18:37:59.758Z"},"provenance":{"textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observationId":"e9e57aeefded69285e1b6876a88e2095ce8a63a41277272df530fee625cd3fa3","retrievedAt":"2026-11-08T18:37:57.313Z","kind":"publisher-page","preparer":{"model":"gpt-6-luna","runId":"41f0eafb-43c2-4b4f-801b-303b23c87d55"},"reviewer":{"model":"gpt-6-luna","runId":"bcd7d93d-e19b-4b0f-82c2-92be926a1185"}},"audit":{"status":"passed","reviewedAt":"2026-11-08T18:38:54.673Z","model":"anthropic/claude-opus-5.5"},"sourceAttribution":{"authors":[{"name":"Kwon, Junhyuk","url":null,"kind":"unknown"},{"name":"Lee, Seungjoon","url":null,"kind":"unknown"},{"name":"Park, Hyejin","url":null,"kind":"unknown"},{"name":"Min, Kyle","url":null,"kind":"unknown"},{"name":"Ok, Jungseul","url":null,"kind":"unknown"}],"authorRole":"article-author","hosts":[],"status":"stated","authorRoster":{"truncated":false,"omitted":0,"assertionGroupsOmitted":0},"participation":[{"actor":{"name":"Kwon, Junhyuk","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Kwon, Junhyuk","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:0","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Lee, Seungjoon","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Lee, Seungjoon","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:1","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Park, Hyejin","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Park, Hyejin","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:2","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Min, Kyle","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Min, Kyle","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:3","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}},{"actor":{"name":"Ok, Jungseul","url":null,"kind":"unknown","actorId":null},"role":"author","scope":{"kind":"work-credit","locator":null},"literalCredit":"Ok, Jungseul","status":"stated","provenance":{"basis":"head-meta","sourceUrl":"https://arxiv.org/abs/2605.06223","locator":"head/meta[name=author|citation_author]"},"assertionId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337:participation:4","identity":{"version":"entity-identity-v1","status":"unresolved","actorId":null,"canonicalActorId":null,"candidateActorIds":[],"assertions":[],"sourceObservedAt":"2026-11-08T19:10:10.171Z","knowledgeAsOf":null},"sourceVersion":{"snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","sourceUrl":"https://arxiv.org/abs/2605.06223","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z"}}],"participationCoverage":{"total":5,"returned":5,"complete":true,"fullAssertions":"source-metadata-endpoint"},"distribution":null,"publisher":null,"sourceUrl":"https://arxiv.org/abs/2605.06223","snapshotId":"0b9b7e71ce405b22d8197b4e329dbbc2801f6631c2bceb59529973b2ecc57337","contentSha256":"481a94691255b791b3e6de131c8881fc74e57c4b6338e446be0c996a9bba7c56","textHash":"460afe35c2ccb139e6795af98e5886bdc1b1e45984d60bdfce14da4463e9d9da","observedAt":"2026-11-08T19:10:10.171Z","basis":"publisher-page-metadata","sourceLinks":49,"linksTruncated":false,"metadataUrl":"/api/pipeline/stream/244e269f0d056470b5b04acff0451f62d0008753b9145fc3836a965e881424ba/source"}}}],"audits":[{"publicationId":"5196fdc1-2997-471d-817c-5033dc382464","verdict":"passed","findings":[{"dimension":"attribution","status":"pass","explanation":"The title credits Junhyuk Kwon and coauthors, matching the listed authors. The summary attributes the method, results and argument to the paper's authors, and frames outcomes as what the authors report rather than established fact.","excerpt":"Authors:Junhyuk Kwon, Seungjoon Lee, Hyejin Park, Kyle Min, Jungseul Ok"},{"dimension":"summary","status":"pass","explanation":"The reading leads with the actor and the problem: a robot must work out which of several similar objects a user means without long answers. It explains the two-stage method and RCJ in plain language, reports results with their scope, and preserves the authors' closing argument about performance and interaction burden. 'Batched delivery' versus 'interleaved interaction' is left unexplained, but it is secondary and the abstract does not define it. The question addresses a genuine gap. The reading also omits the authors' critique of prior methods' premature decisions; this is minor.","excerpt":"an agent must identify and reach the user-intended target among similar objects"},{"dimension":"date","status":"pass","explanation":"originalPublished is 2026-05-07, which matches the v1 submission date. It is not the v5 revision date of 5 Nov 2026 or the feed date.","excerpt":"Submitted on 7 May 2026 (v1), last revised 5 Nov 2026 (this version, v5)"},{"dimension":"support","status":"pass","explanation":"Every claim is directly backed by the abstract. The benchmark gains are kept as author-reported results limited to the evaluated benchmarks. The human-study preferences are not inflated. The limitations correctly note that the abstract gives no numerical results, evaluation details or study participant details. The reading makes no unsupported claims about benefits or real-world reliability.","excerpt":"ProCompNav substantially improves Success Rate and reduces Response Length on an interactive navigation benchmark"},{"dimension":"media","status":"pass","explanation":"The selected media is the verified direct arXiv abstract page for this paper, which is the original source. No other media candidates were found.","excerpt":null}],"createdAt":"2026-11-08T18:38:54.673Z","model":"anthropic/claude-opus-5.5"}],"correctionUrl":"https://escapemeridian.com/corrections/"}