{"schema_version":"papers/paper-detail-v1","title":"ProRL: Effective Reinforcement Learning for Proactive Recommendation via Rectified Policy Gradient Estimation","surface":"papers","opportunity_kernel":{"paper_id":"dd3c0e9d-c2f2-42b4-a0f6-f7a3a375f975","title":"ProRL: Effective Reinforcement Learning for Proactive Recommendation via Rectified Policy Gradient Estimation","authors":["Hongru Hou","Tiehua Mei","Denghui Geng","Jinhui Huang","Ao Xu","Hengrui Chen","Jiaqing Liang","Deqing Yang"],"arxiv_id":"2605.28293v1","doi":null,"published_at":"2026-05-27T10:43:37.000Z","score_object":{"overall":{"value":6,"scale":"0-10","confidence":0.85,"confidence_reason":"Aligned to persisted papers.viability_score for public contract consistency.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:19:42.104Z","fresh_until":"2026-06-27T16:19:42.104Z","is_stale":false,"source_count":1,"missingness":[]},"technical":{"value":1.4,"scale":"0-10","confidence":0.55,"confidence_reason":"Backfilled from paper_extraction_scorecards.reconstruction_score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:20:52.442Z","fresh_until":"2026-06-11T16:20:52.442Z","is_stale":false,"source_count":1,"missingness":["reproducibility_results.reproducibility_score","deployability_scores.score"]},"commercial":{"value":5,"scale":"0-10","confidence":0.75,"confidence_reason":"Backfilled from persisted commercial_flags and repo availability.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:19:42.104Z","fresh_until":"2026-06-27T16:19:42.104Z","is_stale":false,"source_count":2,"missingness":[]},"market":{"value":5.5,"scale":"0-10","confidence":0.8,"confidence_reason":"Backfilled from distribution_readiness_scores.score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:53:04.832Z","fresh_until":"2026-06-11T16:53:04.832Z","is_stale":false,"source_count":1,"missingness":[]},"team":{"value":4.7,"scale":"0-10","confidence":0.28,"confidence_reason":"Heuristic fallback from paper author count.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:19:42.104Z","fresh_until":"2026-06-11T16:19:42.104Z","is_stale":false,"source_count":1,"missingness":["paper_author_affiliations"]},"methodology":{"value":4.7,"scale":"0-10","confidence":0.82,"confidence_reason":"Backfilled from paper_extraction_scorecards.total_score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-28T16:20:52.442Z","fresh_until":"2026-06-27T16:20:52.442Z","is_stale":false,"source_count":2,"missingness":[]}},"evidence_receipt":{"freshness":"fresh","proof_status":"unverified","repo_status":"active","references_count":0,"source_count":4,"coverage":0.6667,"missingness":["references","proof_status"],"unresolved_unknowns":["proof verification has not been recorded yet"],"last_verification_at":"2026-05-28T16:53:04.832Z"},"lineage_hash":"8268f4068f7ee5769ce4736050b48afa23f6fb502a852205f496b89ebf3789e8"},"distribution":null,"replication_evidence":[],"author_dna":[]}