{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp04_retrieval_first_paired20_20260802","cell_id":"invariant_mode_decomposition_design__library_information_science","round_index":0,"assessments":[{"hypothesis_id":"H1","search_queries":["metadata crosswalk validation SVD reconstruction residual rare records library","metadata crosswalk quality evaluation semantic loss record classes","site:dublincore.org metadata crosswalk mapping validation","site:loc.gov metadata crosswalk validation MARC MODS"],"sources":[{"source_id":"H1-S1","title":"MARC to Dublin Core Crosswalk","publisher":"Library of Congress","url":"https://www.loc.gov/marc/marc2dc.html","source_class":"OFFICIAL_GUIDANCE","claims_supported":["The official crosswalk maps multiple MARC fields into simpler Dublin Core elements.","The Library of Congress explicitly warns that accurate round-trip mapping is not expected because information is lost.","The crosswalk omits some MARC fields and permits application-specific choices."]},{"source_id":"H1-S2","title":"Conversions: Metadata Object Description Schema: MODS","publisher":"Library of Congress","url":"https://www.loc.gov/standards/mods/mods-conversions.html","source_class":"OFFICIAL_GUIDANCE","claims_supported":["The MODS Editorial Committee supplies conversion stylesheets for MARC, Dublin Core, MODS, and BIBFRAME.","The stylesheets are not exhaustive and may omit uncommon data elements.","Organizations are advised to customize mappings for local requirements."]},{"source_id":"H1-S3","title":"Metadata Interoperability and Standardization—A Study of Methodology Part I: Achieving Interoperability at the Schema Level","publisher":"D-Lib Magazine","url":"https://www.dlib.org/dlib/june06/chan/06chan.html","source_class":"PRIMARY_RESEARCH","claims_supported":["Crosswalks ordinarily operate as element-level semantic mappings.","Differences in repeatability, hierarchy, value constraints, and conditional obligations can create conversion-quality problems beyond simple field correspondence."]},{"source_id":"H1-S4","title":"Converting VRA Core Records to MARC Records: A Study in Crosswalking","publisher":"San José State University ScholarWorks","url":"https://scholarworks.sjsu.edu/libphilprac/1076/","source_class":"PRIMARY_RESEARCH","claims_supported":["A library-domain case study examined record conversion through an established metadata crosswalk.","The study reports practical challenges in converting heterogeneous visual-resource records to MARC."]}],"closest_analogue":"Library of Congress crosswalk/conversion guidance combined with item-level crosswalk case studies and multidimensional metadata-quality evaluation.","overlap":"Existing work recognizes lossy many-to-one mappings, omitted uncommon elements, application-specific conversion rules, record-level challenges, and semantic quality dimensions beyond aggregate field completeness.","remaining_difference":"The screened sources do not describe representing the crosswalk as a rectangular operator, selecting retained singular modes through held-out reconstruction error, and enforcing separate residual tolerances for governed rare record classes. That difference is testable by comparing class-specific semantic-defect and review-cost outcomes against conventional rule and field validation.","classification":"POSSIBLE_DISTINCTION","disposition":"ADVANCE","rationale":"Loss and heterogeneity are established, but the proposed class-stratified modal stopping rule is not present in the closest sources. It is a concrete methodological distinction rather than a claim inferred from search failure."},{"hypothesis_id":"H2","search_queries":["library discovery click feedback ranking exposure popularity bias dynamical","recommender systems feedback loop exposure concentration dynamic ranking primary research","ranking feedback loop eigenvalue stability exposure mode","site:acm.org recommender feedback loops popularity bias exposure"],"sources":[{"source_id":"H2-S1","title":"Only the first results count: user-feedback-modified relevance ranking in E-LIB Bremen","publisher":"Insights / UKSG","url":"https://insights.uksg.org/articles/10.1629/uksg.235","source_class":"PRIMARY_RESEARCH","claims_supported":["A library discovery implementation incorporated click statistics and popularity factors into relevance ranking.","The work directly situates user-feedback ranking within a library catalogue or discovery context."]},{"source_id":"H2-S2","title":"Feedback Loop and Bias Amplification in Recommender Systems","publisher":"ACM CIKM / arXiv","url":"https://arxiv.org/abs/2007.13019","source_class":"PRIMARY_RESEARCH","claims_supported":["Offline repeated-update simulation showed that recommendation feedback amplifies popularity bias over time.","The study measured declining catalogue coverage, homogenization, and disproportionate effects on a minority group.","It explicitly proposes evaluating popularity-control algorithms across repeated feedback iterations."]},{"source_id":"H2-S3","title":"Fairness of Exposure in Dynamic Recommendation","publisher":"arXiv","url":"https://arxiv.org/abs/2309.02322","source_class":"PRIMARY_RESEARCH","claims_supported":["The study models exposure bias under repeated recommendation and feedback updates.","It adapts a mitigation method for long-run exposure fairness and evaluates it on real data.","The reported objective combines reduced long-term exposure disparity with maintained recommendation accuracy."]},{"source_id":"H2-S4","title":"The Dynamics of Recommender Systems","publisher":"University of California, Berkeley","url":"https://www2.eecs.berkeley.edu/Pubs/TechRpts/2022/EECS-2022-178.pdf","source_class":"PRIMARY_RESEARCH","claims_supported":["The work treats recommendation and collected feedback as a coupled process evolving over time.","It analyzes interventions for feedback loops through simulation and causal methods.","It identifies recommendation-induced bias, homogenization, and rich-get-richer effects."]}],"closest_analogue":"Fairness of Exposure in Dynamic Recommendation, supplemented by Feedback Loop and Bias Amplification in Recommender Systems and the E-LIB Bremen click-ranking implementation.","overlap":"The analogue already contains repeated ranking-feedback updates, popularity or exposure concentration, offline longitudinal evaluation, exposure mitigation, protection of recommendation accuracy, and concern for underexposed or minority interests. A library discovery system using click-derived popularity ranking has also been implemented.","remaining_difference":"The remaining difference is the internal control parameterization: linearize a query-cluster-by-item-group update, identify unstable eigenmodes, and damp only those modes. The screened outcome claim is otherwise substantially covered by dynamic exposure-fairness work.","classification":"OBVIOUS_COLLISION","disposition":"REJECT","rationale":"The proposed problem, evaluation loop, mitigation goal, and relevance-preservation constraint closely collide with established dynamic recommender-feedback research; replacing existing dynamic or causal mitigation with eigenmode-specific damping is too narrow for this elimination screen."},{"hypothesis_id":"H3","search_queries":["digital preservation audit sampling risk assessment format stratified integrity checks","digital preservation risk assessment correlated factors PCA latent modes audit","site:ndsa.org Levels of Digital Preservation fixity sampling","site:loc.gov digital preservation risk assessment integrity audit sampling"],"sources":[{"source_id":"H3-S1","title":"Levels of Digital Preservation","publisher":"National Digital Stewardship Alliance","url":"https://www.ndsa.org/publications/levels-of-digital-preservation/","source_class":"OFFICIAL_GUIDANCE","claims_supported":["NDSA provides a structured framework and assessment tool for evaluating preservation programs.","Its guidance covers decision points involving collections, infrastructure, integrity, metadata, formats, and preservation strategy."]},{"source_id":"H3-S2","title":"DiAGRAM—The Digital Archiving Graphical Risk Assessment Model","publisher":"The National Archives (UK)","url":"https://diagram.nationalarchives.gov.uk/","source_class":"GOVERNMENT_OR_REGULATOR","claims_supported":["DiAGRAM explicitly models linked digital-preservation risk events.","It uses statistical methods to estimate preservation probability and supports counterfactual scenario testing."]},{"source_id":"H3-S3","title":"Digital Preservation Framework for Risk Assessment and Preservation Planning","publisher":"U.S. National Archives and Records Administration","url":"https://www.archives.gov/preservation/digital-preservation/risk","source_class":"GOVERNMENT_OR_REGULATOR","claims_supported":["NARA prioritizes preservation action using file-format risk analysis.","The framework organizes plans by record type and identifies significant properties to retain."]},{"source_id":"H3-S4","title":"Identifying Threats to Successful Digital Preservation: the SPOT Model for Risk Assessment","publisher":"D-Lib Magazine","url":"https://www.dlib.org/dlib/september12/vermaaten/09vermaaten.html","source_class":"PRIMARY_RESEARCH","claims_supported":["SPOT defines outcome-oriented preservation properties and multiple threats to them.","Related preservation models represent relationships among risks, objects, and environments.","Repository-specific risk registers and ongoing risk analysis are established practices."]}],"closest_analogue":"DiAGRAM's linked probabilistic preservation-risk model, alongside SPOT/DRAMBORA-style repository risk registers and NARA record-type risk prioritization.","overlap":"Existing approaches combine multiple preservation risks, represent dependencies among risks and environments, estimate preservation outcomes, test scenarios, and prioritize by format or record type.","remaining_difference":"The unresolved difference is object-level audit allocation to poorly reconstructed or outcome-sensitive latent regions. Bounded research question: does any repository retain a joinable object-level dataset containing longitudinal storage, software, provenance, handling, and format features plus independently verified integrity failures at sufficient scale for an equal-budget retrospective sampling comparison?","classification":"INDETERMINATE_RESEARCH_NEEDED","disposition":"REJECT","rationale":"The intervention is distinguishable from program-level risk modeling, but its evaluability depends on a specific dataset condition not established by this shallow search. Search failure is not treated as novelty; an independent critic can resolve the bounded data-availability question."},{"hypothesis_id":"H4","search_queries":["interlibrary loan network eigenvector centrality routing lender load","interlibrary loan network analysis centrality resource sharing consortium","OCLC interlibrary loan network analysis centrality","library resource sharing network resilience routing centrality"],"sources":[{"source_id":"H4-S1","title":"Application of social network analysis in interlibrary loan services","publisher":"Webology","url":"https://www.webology.org/2013/v10n1/a108.html","source_class":"PRIMARY_RESEARCH","claims_supported":["Researchers analyzed ten years of national ILL logs covering 240 institutions.","They calculated weighted degree, closeness, and betweenness centralities to reveal institutional roles and information flow.","The analysis was presented as decision support for resource-sharing cooperation and collection planning."]},{"source_id":"H4-S2","title":"A library network model","publisher":"Journal of the American Society for Information Science","url":"https://asistdl.onlinelibrary.wiley.com/doi/10.1002/asi.4630270204","source_class":"PRIMARY_RESEARCH","claims_supported":["A distributed queueing-network model evaluates ILL request satisfaction, delay, cost, and processing load at each member library.","The model compares centralized and distributed network structures."]},{"source_id":"H4-S3","title":"Limits of Concentration and Dispersion of Lending Service Based on a Model of Interlibrary Loan Network","publisher":"Annals of Japan Society of Library Science / J-STAGE","url":"https://www.jstage.jst.go.jp/article/ajsls/41/3-4/41_130/_article/-char/en","source_class":"PRIMARY_RESEARCH","claims_supported":["The study models assignment of ILL demand to eligible lenders.","It analyzes concentration and dispersion under lender-capacity constraints.","Maximum lender capacity and the distribution of service capacity bound achievable concentration and dispersion."]},{"source_id":"H4-S4","title":"WorldShare Interlibrary Loan","publisher":"OCLC","url":"https://www.oclc.org/en/worldshare-ill.html","source_class":"COMMERCIAL_FIRST_PARTY","claims_supported":["WorldShare automates borrowing and lending across a network of more than 10,000 libraries.","Operational reporting includes lending volume, fill rates, and turnaround time.","The platform supplies a plausible source of routing and fulfillment logs for counterfactual analysis."]}],"closest_analogue":"Application of social network analysis in interlibrary loan services, combined with established ILL queueing and lender-concentration models.","overlap":"ILL transaction networks have already been analyzed with centrality measures; institutional bridge roles, lender load, routing structure, delays, capacity constraints, and concentration-versus-dispersion are established concerns.","remaining_difference":"The screened sources do not combine dominant-eigenvector dependency scores with simulated node-removal sensitivity and a live or replayed routing rule that deliberately diversifies requests while constraining fulfillment time. This can be tested against degree-, volume-, and capacity-based routing on historical logs.","classification":"POSSIBLE_DISTINCTION","disposition":"ADVANCE","rationale":"Network analysis and load concentration are prior art, but the proposed removal-sensitive spectral routing policy remains a concrete, falsifiable distinction. Its advantage must be demonstrated against the already strong degree, betweenness, capacity, and queueing baselines."},{"hypothesis_id":"H5","search_queries":["authority file vocabulary evolution spectral graph eigenvector drift","controlled vocabulary change detection graph spectral semantic drift library authority","Library of Congress authority records change frequency review corrections","thesaurus evolution network analysis semantic drift eigenvector"],"sources":[{"source_id":"H5-S1","title":"MARC 21 Format for Authority Data","publisher":"Library of Congress","url":"https://www.loc.gov/marc/authority/","source_class":"STANDARD","claims_supported":["MARC authority records formally encode headings, references, linking entries, notes, provenance, and update history.","The authority format continues to receive documented revisions, providing structured graph and temporal inputs."]},{"source_id":"H5-S2","title":"Library of Congress Policy Regarding Requests for the Creation and Modification of Catalog Records","publisher":"Library of Congress","url":"https://www.loc.gov/catdir/cpso/modification-catalog-records.html","source_class":"OFFICIAL_GUIDANCE","claims_supported":["Authority-record corrections and duplicate deletions are governed human-review activities.","Library of Congress staff limitations make comprehensive enhancement impractical.","Correction requests require supporting evidence, creating a potential labeled outcome stream."]},{"source_id":"H5-S3","title":"Orbit signatures for semantic drift detection in NLP networks","publisher":"Journal of Computational Social Science / ScienceDirect","url":"https://www.sciencedirect.com/science/article/pii/S2590005626003644","source_class":"PRIMARY_RESEARCH","claims_supported":["The paper proposes structural diagnostics for dynamic vocabulary networks.","Its indicators include a spectral drift index and graph-structure measures.","The authors state that correlation with real semantic-change benchmarks remains empirically unresolved."]},{"source_id":"H5-S4","title":"Detection of Temporal Shifts in Semantics Using Local Graph Clustering","publisher":"NSF Public Access Repository","url":"https://par.nsf.gov/biblio/10417882","source_class":"PRIMARY_RESEARCH","claims_supported":["The study represents terms through time-evolving contextual-word graphs.","It detects semantic shifts from structural changes using dynamic graph methods.","The demonstrated target is corpus-language change rather than maintenance events in a governed authority file."]}],"closest_analogue":"Orbit signatures for semantic drift detection in NLP networks, with dynamic local-graph semantic-shift detection as a second close analogue.","overlap":"Dynamic vocabulary graphs, structural drift measurement, spectral indicators, and detection of changes that surface counts may miss are already described. Authority files also provide explicit terms, references, revisions, and governed correction events.","remaining_difference":"The proposed test uses authority-reference and bibliographic co-assignment graphs, tracks dominant-subspace rotation or spectral-gap collapse, and asks whether alerts predict later governed correction clusters better than edit-volume and term-frequency baselines. The closest semantic-drift methods have not demonstrated that authority-maintenance prediction task.","classification":"POSSIBLE_DISTINCTION","disposition":"ADVANCE","rationale":"Spectral vocabulary-drift detection is a close methodological neighbour, so broad novelty is not supported. The narrower authority-governance outcome—capacity-matched prediction of subsequent correction clusters—remains distinct and directly testable."}],"nominated_ids":["H1","H4","H5"],"replenishment_recommended":false}