{"closest_prior_art":[{"name":"ReleGuard automated search-quality monitoring","overlap":"Uses scheduled read-only tests for critical queries, expected-result and rank conditions, historical results, gradual-regression detection, alerts, and team accountability before users report failures. This substantially overlaps the register's core query-panel, cadence, trend, threshold, and escalation mechanism.","remaining_difference":"It is a generic commercial-search product and does not document curator-adjudicated library records, name/subject/date/identifier paths, facet and link checks, catalog-to-discovery comparison, two-round persistence bands, or routing to library-specific metadata, authority, crosswalk, and reindexing owners.","source_ids":["SRC2"]},{"name":"Brown University Library search relevancy tests","overlap":"Predetermined phrases test expected Solr/Blacklight results, systematically protect against regressions after indexing-schema or term-weight changes, and have identified and resolved title-weighting and identifier-search defects.","remaining_difference":"The page does not describe a fixed cadence, longitudinal condition bands, persistence or acceleration thresholds, multiple access-path signals, cross-interface disagreement, owned remediation tickets, or post-repair condition recalibration.","source_ids":["SRC1"]},{"name":"University of Houston Metadata Clean Sweep","overlap":"An ongoing digital-collection audit sampled each collection, established a baseline, checked controlled-name formatting and broken links, recorded collection and field-level patterns, maintained correction lists outside the live system, required staff review, performed repairs, and continued periodic targeted batch reviews.","remaining_difference":"It primarily inspects metadata records against guidelines rather than repeatedly executing adjudicated searches and measuring retrieval presence, rank bands, facets, identifier resolution, or disagreement between discovery interfaces.","source_ids":["SRC4"]},{"name":"Known-item zero-hit evaluation in EconBiz","overlap":"A judged query set measured whether held items appeared within the top 30 results and diagnosed failures including incomplete or erroneous metadata, demonstrating a direct query-to-record evaluation method.","remaining_difference":"It was a research analysis of transaction-log queries, not a privacy-preserving stable panel with recurring trend memory, warning thresholds, named repair authority, and post-repair reruns.","source_ids":["SRC3"]}],"contrastive_claim_falsifier":"The narrowed claim is falsified if blinded review finds that persistent or accelerating multi-signal warnings do not predict reproducible, correctable losses of intended access paths earlier or more precisely than ordinary relevancy regression tests and periodic metadata audits, or if scoped repairs do not restore the flagged paths on rerun.","contrastive_claim_remaining":"For digitized local-history collections, a curator-adjudicated, privacy-preserving panel that jointly tracks query retrieval, rank bands, facets, identifiers, links, and catalog-to-discovery disagreement—and escalates only persistent or accelerating deterioration through named library authorities—may identify correctable access-path decay earlier and with more actionable localization than generic search-regression monitoring or recurring metadata audits alone.","experiment_id":"eoa_inverse_innovation_exp13_second_slot_policy60_20260806","gates":{"adequate_source_search":{"rationale":"The bounded search covered the exact proposal title, retrievability and known-item terminology, library relevancy testing, metadata audits and dashboards, commercial monitoring products, and relevant preservation/discovery standards. Four opened sources from four publishers were retained; an exact-title miss was not treated as novelty.","source_ids":["SRC1","SRC2","SRC3","SRC4"],"status":"PASS"},"bounded_next_test":{"rationale":"The proposed eight-week, two-collection, 40-query shadow pilot has fixed duration, sample, signals, reviewers, classifications, continuation criterion, and no production repair. Prior implementations show that predetermined query tests, scheduled read-only monitoring, sampled audits, and staff-reviewed offline correction lists are operationally feasible.","source_ids":["SRC1","SRC2","SRC4"],"status":"PASS"},"distinct_testable_claim":{"rationale":"Although the broad monitoring architecture substantially collides with prior art, a narrower claim remains testable: whether library-specific multi-path signals plus persistence bands and explicit authority routing improve early, correctable detection over generic regression tests and metadata audits.","source_ids":["SRC1","SRC2","SRC4"],"status":"PASS"},"no_obvious_safety_or_authority_stop":{"rationale":"No obvious stop applies to the authorized first step because it is read-only, excludes identifiable patron histories and production changes, preserves existing access controls, and leaves descriptive and indexing decisions with existing owners. Comparable sources describe read-only testing and keeping proposed changes outside the live system for staff approval. Restricted-record exposure and workload remain pilot halt conditions rather than unavoidable blockers.","source_ids":["SRC2","SRC4"],"status":"PASS"},"supported_problem":{"rationale":"The problem is visible: Brown reports regression risk from indexing-schema and term-weight changes and resolved identifier/title defects; the EconBiz study found held items missed because of incomplete or erroneous metadata; the Houston case found metadata inconsistencies and broken links affecting discovery and requiring ongoing maintenance. These sources support access failures and silent regression risk, though not the full claimed gradual trajectory for local-history collections.","source_ids":["SRC1","SRC3","SRC4"],"status":"PASS"}},"prior_art_disposition":"SUBSTANTIAL_COLLISION","problem_evidence":{"finding":"PARTLY SUPPORTED: library discovery failures attributable to metadata defects, indexing changes, identifier handling, and broken links are directly documented, as are recurring audits and silent search regressions. The retained evidence does not independently demonstrate that authority-file, crosswalk, and reindexing changes usually produce measurable directional deterioration across repeated rounds in digitized local-history collections.","source_ids":["SRC1","SRC2","SRC3","SRC4"],"status":"PARTLY_SUPPORTED"},"research_id":"eoa_inverse_innovation_exp13_light_screen_20260806","schema_version":1,"screen_id":"E13P029","screen_survival":false,"search_lanes":{"component_combination":{"no_result_note":null,"queries":["search relevance regression testing judged query set continuous monitoring","library discovery relevance regression test query set Solr","digital library metadata quality monitoring dashboard recurring assessment"],"source_ids":["SRC1","SRC2","SRC4"]},"direct_problem_and_intervention":{"no_result_note":"The exact-title query produced no exact match; that phrase miss was not treated as evidence of novelty.","queries":["\"Retrievability Condition Register\" digitized collections","\"retrievability monitoring\" library collections","digital collections retrievability audit known item search recurring monitoring","library digital collections discovery access paths metadata quality audit"],"source_ids":["SRC1","SRC4"]},"products_practices_and_standards":{"no_result_note":null,"queries":["site:quepid.com regression testing search relevance judgments queries alerts","digital library metadata quality monitoring dashboard recurring assessment","site:niso.org library discovery service search relevance standard metadata quality","ISO 16363 digital repository access discovery metadata audit requirement","NDSA Levels digital preservation metadata discovery access monitoring"],"source_ids":["SRC2","SRC4"]},"synonyms_and_historical_terms":{"no_result_note":null,"queries":["library discovery system regression testing known-item searches metadata index","\"search relevancy tests\" library known-item regressions","Known-item searches resulting in zero hits discovery systems","metadata drift monitoring digital library authority file reindexing discovery"],"source_ids":["SRC1","SRC3","SRC4"]}},"sources":[{"claims_supported":["Predetermined library search phrases can verify expected results systematically.","Such tests protect against regressions when Solr indexing schema or term weighting changes.","The tests identified and helped resolve title-weighting and identifier-search deficiencies."],"publisher":"Brown University Library","source_id":"SRC1","source_type":"FIRST_PARTY_PRODUCT","title":"Search relevancy tests","url":"https://library.brown.edu/create/digitaltechnologies/search-relevancy-tests/"},{"claims_supported":["Scheduled read-only search tests can enforce result-presence and rank expectations.","Historical results can expose gradual search-quality degradation.","Failures can generate immediate alerts and support team accountability."],"publisher":"ReleGuard","source_id":"SRC2","source_type":"FIRST_PARTY_PRODUCT","title":"Search Quality Monitoring & Regression Testing","url":"https://www.releguard.com/"},{"claims_supported":["A primary study evaluated 708 known-item queries against judged correct results.","314 queries failed to return the correct item, including 62 cases where a held item had incomplete or erroneous metadata.","The method distinguished no-result failures from failures to place the correct item within the top 30."],"publisher":"The Journal of Academic Librarianship","source_id":"SRC3","source_type":"PRIMARY_RESEARCH","title":"Known-item searches resulting in zero hits: Considerations for discovery systems","url":"https://searchstudies.org/wp-content/uploads/2021/06/Pre-print_Behnert_Lewandowski_2017_Known-items-69091.pdf"},{"claims_supported":["A digital-library metadata audit used collection sampling, a baseline, controlled-vocabulary checks, link checks, trend summaries, and correction lists.","The project continued through periodic batch reviews targeted by collection and field.","Suggested changes were kept outside the live system until staff review, and the audit documented broken links and metadata defects affecting discovery."],"publisher":"D-Lib Magazine","source_id":"SRC4","source_type":"PRIMARY_RESEARCH","title":"Metadata Clean Sweep: A Digital Library Audit Project","url":"https://www.dlib.org/dlib/may12/westbrook/05westbrook.html"}],"world_novelty_boundary":"This coarse public-web screen found substantial overlap in both the library domain and generic search-monitoring products. It does not establish that every element has previously been combined in one implementation, nor can it establish world novelty, patentability, freedom to operate, market size, expert acceptance, or realized value. The remaining boundary is the specific combination of library-adjudicated multi-path condition signals, persistence or acceleration bands, and named remediation authority, which requires targeted literature, product, implementation, and patent searching plus the proposed comparative pilot."}