{"schema_version":1,"research_id":"eoa_inverse_innovation_exp04_external_evaluation_20260802","source_assessment_id":"invariant_mode_decomposition_design__library_information_science:SENTINEL:v0","cell_id":"invariant_mode_decomposition_design__library_information_science","search_queries":["site:help.oclc.org WorldShare ILL smart lender strings queue depth turnaround time","site:nlm.nih.gov docline routing work random lenders","site:openlibraryfoundation.atlassian.net ReShare Resource Sharing Settings ratio load balancing","interlibrary loan concentration lenders workload consortium study","site:projectreshare.org consortium governance steering committee resource sharing libraries PALCI","site:palci.org resource sharing load balancing consortium strategic plan interlibrary loan","ISO 18626 interlibrary loan transactions standard official data elements","interlibrary loan transaction logs network analysis centrality libraries primary research","\"Interlibrary loan in primary access libraries\" closures 1996 Colorado results","\"Application of social network analysis in interlibrary loan services\" Jalalimanesh Yaghoubi pdf","\"Distributed Location of the Critical Nodes\" spectral analysis node removal","site:crl.acrl.org interlibrary loan lending workload concentration consortium"],"sources":[{"source_id":"S1","title":"An Analysis of Member Participation in a Document Delivery Service Using Transaction Data","publisher":"Korea Citation Index / Korean Society for Information Management","url":"https://www.kci.go.kr/kciportal/landing/article.kci?arti_id=ART001912205","source_class":"PRIMARY_RESEARCH","publication_date":"2014","accessed_at":"2026-08-02","claims_supported":["Nine years of KERIS document-delivery transactions were analyzed as a library request-and-supply network.","More than 80% of member libraries requested more than they supplied.","The service depended heavily on a relatively small set of top suppliers, and supply-side concentration exceeded request-side concentration.","Subject networks exhibited single-center, multi-center, and distributed structures."]},{"source_id":"S2","title":"An Analysis of the Current State and Changes in the Interlibrary Loan Service Focused on KERIS Data From 2004 to 2014","publisher":"Korean Society for Information Management","url":"https://journal.kci.go.kr/kosim/archive/articlePdf?artiId=ART002035198","source_class":"PRIMARY_RESEARCH","publication_date":"2015-09","accessed_at":"2026-08-02","claims_supported":["The study analyzed 115,522 requests and 97,790 fulfilled transactions from an eleven-year KERIS dataset.","Transaction records included requesting and supplying institutions, processing status, stage dates, failure reasons, and bibliographic fields, demonstrating that institution-level flow and outcome analysis is technically possible.","The paper reports workload as a barrier to participation and records interview evidence that shrinking acquisition budgets increased the need for external resource sharing."]},{"source_id":"S3","title":"Smart lender strings through the Automated Request Manager","publisher":"OCLC Support","url":"https://help.oclc.org/Resource_Sharing/WorldShare_Interlibrary_Loan/Smart_fulfillment/040Smart_lender_strings_through_the_Automated_Request_Manager","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"undated; current documentation accessed 2026-08-02","accessed_at":"2026-08-02","claims_supported":["WorldShare constructs lender strings after excluding ineligible or deflecting libraries.","Eligible lenders are ordered using turnaround time and hourly queue depth specifically to avoid overwhelming individual lenders.","The system already supports automated, dynamic lender ordering and operational turnaround monitoring."]},{"source_id":"S4","title":"Load balancing algorithm","publisher":"Open Library Foundation ReShare","url":"https://openlibraryfoundation.atlassian.net/wiki/spaces/PR/pages/2551775240/Load%2Bbalancing%2Balgorithm","source_class":"OFFICIAL_PRODUCT_DOCUMENTATION","publication_date":"2021-12-20","accessed_at":"2026-08-02","claims_supported":["ReShare dynamically orders eligible suppliers in a request rota.","Its deployed approach compares each library's current lending with a library-defined target loan-to-borrow ratio.","The routing workflow provides an implementation point for testing a topology-aware ordering policy but is already a strong load-balancing comparator."]},{"source_id":"S5","title":"How does DOCLINE routing work?","publisher":"U.S. National Library of Medicine","url":"https://www.nlm.nih.gov/docline/context_help/context_help_routing_choices.html","source_class":"OFFICIAL_GUIDANCE","publication_date":"2018-05-08","accessed_at":"2026-08-02","claims_supported":["DOCLINE checks holdings, participation, delivery, service, payment, exclusion, format, and embargo constraints before routing.","Within eligible routing-table cells and broader groups, requests are randomly routed.","Randomized eligible-lender diversification is established practice and must be included as a comparator."]},{"source_id":"S6","title":"ReShare Steering Committee","publisher":"Project ReShare","url":"https://projectreshare.org/about/steering-committee/","source_class":"OFFICIAL_ORGANIZATION_DATA","publication_date":"undated; current roster accessed 2026-08-02","accessed_at":"2026-08-02","claims_supported":["The ReShare Steering Committee has authority over the strategic roadmap, resource allocation, risk, and community funding.","Its membership includes representatives of library consortia and operating institutions, identifying a credible platform-level authorizer or research partner.","The source does not show commitment to this spectral proposal or authority over an individual tenant consortium's transaction data."]},{"source_id":"S7","title":"ISO 18626:2021 Information and documentation — Interlibrary loan transactions","publisher":"International Organization for Standardization","url":"https://www.iso.org/standard/79013.html","source_class":"STANDARD","publication_date":"2021-12","accessed_at":"2026-08-02","claims_supported":["ISO 18626 specifies transactions and messages between libraries and other agencies for handling item requests.","A current formal interoperability standard exists for ILL transaction exchange.","The standard supports workflow interoperability but does not establish that historical eligible-candidate sets or counterfactual outcomes are retained."]},{"source_id":"S8","title":"Distributed Location of the Critical Nodes to Network Robustness based on Spectral Analysis","publisher":"IEEE / National Laboratory for Scientific Computing","url":"https://www.ic.unicamp.br/~edmundo/proceedings/html/fullPapers/88298_1.pdf","source_class":"PRIMARY_RESEARCH","publication_date":"2011","accessed_at":"2026-08-02","claims_supported":["The paper uses localized spectral analysis to rank nodes by network robustness importance.","It validates rankings through node-removal experiments and compares effects with highest-degree removal.","Its experiments establish adjacent spectral critical-node and removal-testing prior art, but use generic undirected connectivity rather than ILL fulfillment outcomes."]}],"problem_evidence":{"support":"MODERATE","rationale":"The problem's concentration and workload premises are visible: KERIS transaction research found disproportionate dependence on a small supplier set, while OCLC and ReShare explicitly use queue-aware or ratio-based ordering to prevent overload. Generic network research also shows why degree alone may not identify removal-critical nodes. No opened source directly demonstrates, in an ILL consortium, that dominant-eigenvector scores identify outage-sensitive lenders missed by volume, degree, queue depth, or turnaround; that central empirical premise remains unverified.","source_ids":["S1","S2","S3","S4","S8"]},"stakeholder_evidence":{"support":"MODERATE","rationale":"ReShare has an identifiable Steering Committee empowered to set roadmap, resources, risk, and funding, and its product already implements dynamic supplier load balancing. OCLC and NLM likewise operate routing systems addressing overload, efficiency, and eligible-lender selection. This establishes credible authorizers and expressed operational need for balancing, but no consortium or platform has expressed demand for the proposed spectral method or committed data, staff, or pilot authority.","source_ids":["S3","S4","S5","S6"]},"prior_art":{"proximity":"SUBSTANTIAL_COLLISION","closest_analogues":[{"name":"OCLC WorldShare smart lender strings","similarity":"A deployed ILL workflow filters eligible lenders and dynamically orders them using turnaround and current queue depth to improve fulfillment without overwhelming lenders.","remaining_difference":"The documentation does not disclose rolling request-flow spectral scores, fulfillment-specific node-loss replay, or a sensitivity threshold controlling diversification.","source_ids":["S3"]},{"name":"ReShare ratio-based supplier load balancing","similarity":"A deployed resource-sharing platform dynamically scores and orders eligible suppliers to balance lending load.","remaining_difference":"The documented score is based on target loan-to-borrow ratios rather than network topology or simulated loss disruption.","source_ids":["S4"]},{"name":"DOCLINE randomized eligible-lender routing","similarity":"An official ILL router diversifies assignments among libraries satisfying holdings, policy, service, payment, and exclusion rules.","remaining_difference":"Its documented randomization does not use dominant modes, centrality, or node-loss sensitivity.","source_ids":["S5"]},{"name":"Spectral critical-node localization and removal testing","similarity":"Spectral structure is used to find robustness-critical nodes and removal experiments test consequences against degree-based selection.","remaining_difference":"The work concerns generic undirected network connectivity, not directed weighted ILL transactions, eligible-supplier routing, fulfillment, load share, or turnaround noninferiority.","source_ids":["S8"]},{"name":"KERIS transaction-network concentration analysis","similarity":"ILL transactions are represented as a network, supplier concentration is measured, and structurally different centralized and distributed patterns are identified.","remaining_difference":"The study is descriptive and does not implement eigenvector-triggered routing, held-out removal replay, or constrained diversification.","source_ids":["S1","S2"]}],"distinctive_claim_remaining":"For a consortium whose historical eligibility and outcome records permit faithful replay, a policy that combines rolling request-flow dominant-mode dependency, fulfillment-specific lender-removal sensitivity, threshold-triggered diversification only among already eligible lenders, and a median-fulfillment-time noninferiority constraint will reduce both maximum lender-load share and held-out lender-loss disruption more than the existing router and volume-, degree-, random-, queue-aware-, and ratio-based comparators.","confidence":"MODERATE"},"implementation_evidence":{"support":"MODERATE","rationale":"Technical feasibility is credible: published ILL datasets contain institution identifiers, process states, dates, outcomes, and failure reasons; standards define transaction exchange; and operational systems already compute dynamic supplier orderings. Eigenvector estimation and node-removal simulation are routine at much larger network scales. Workflow feasibility is strongest for an offline shadow analysis inserted after existing eligibility logic. Key unresolved data issues are whether eligible-but-unselected alternatives, reroutes, declines, expirations, and comparable lender outcomes are retained well enough for counterfactual replay. Legal and authority feasibility require a tenant consortium's written data-use approval; institution pseudonyms and exclusion of patron identity and request content reduce risk but do not substitute for local confidentiality, retention, and contractual review. Safety risks include stigmatizing high-score libraries and shifting load to smaller lenders; the offline-only first step, score interpretation contract, subgroup and tail checks, gap/stability fallbacks, and prohibition on eligibility changes bound those risks.","source_ids":["S2","S3","S4","S5","S6","S7","S8"]},"scores":{"meaningful_impact":{"score":3,"rationale":"Supplier concentration, workload, fulfillment, and turnaround matter operationally, but the magnitude of avoidable outage harm in a target consortium is unknown.","source_ids":["S1","S2","S3"]},"stakeholder_pull":{"score":3,"rationale":"Named platform and consortium representatives govern resource-sharing risk and already support load balancing, but none has requested or committed to the spectral intervention.","source_ids":["S4","S6"]},"incremental_advantage":{"score":2,"rationale":"The proposal faces strong queue-aware, ratio-based, and randomized baselines; no evidence yet shows that spectral dependency adds predictive or routing value.","source_ids":["S3","S4","S5"]},"distinctiveness_plausibility":{"score":3,"rationale":"The complete ILL-specific combination was not found, although nearly every component has close prior art and proprietary routing logic remains unexamined.","source_ids":["S3","S4","S5","S8"]},"technical_implementability":{"score":4,"rationale":"Institution-level transaction networks, dynamic ordering, interoperability standards, spectral scoring, and removal simulations are individually established and computationally tractable.","source_ids":["S2","S3","S4","S7","S8"]},"adoption_authority_feasibility":{"score":3,"rationale":"A credible governance body exists and controls roadmap, risk, resources, and funding, but tenant data authority and consortium consent have not been obtained.","source_ids":["S6"]},"evidence_readiness":{"score":2,"rationale":"Historical logs likely contain many necessary fields, but faithful replay may fail without eligible-alternative histories and counterfactual lender outcomes.","source_ids":["S2","S3","S5"]},"safety_net_benefit":{"score":3,"rationale":"If valid, diversification could reduce dependence on pivotal suppliers; benefits remain simulated and load-transfer or labeling harms require explicit controls.","source_ids":["S1","S3","S4"]},"scalability":{"score":3,"rationale":"The computation and standardized transaction workflow can scale, but portability depends on heterogeneous logging, eligibility reconstruction, policy, and vendor integration.","source_ids":["S2","S4","S7","S8"]}},"score_confidence":"MODERATE","costs":{"first_evidence":{"band_2026_usd":"50K_TO_250K","scope":"One consenting consortium; data-use review and extraction; eligibility-history audit; pseudonymized data mart; preregistered operator and thresholds; offline fit/held-out replay; comparator analysis; uncertainty, tail-delay, drift, and load-transfer reporting.","confidence":"MODERATE","assumptions":["No live routing change.","Six to nine person-months across a data engineer, quantitative analyst, ILL domain lead, privacy/security reviewer, and project manager.","Existing logs are exportable and require no vendor-built historical reconstruction.","The study stops early if eligible alternatives or outcomes are not reconstructable."],"source_ids":["S2","S3","S4","S5","S7"]},"initial_deployment_startup":{"band_2026_usd":"250K_TO_1M","scope":"Production-grade ingestion, spectral and replay services, routing-adapter integration, access controls, monitoring, audit trail, fallback router, security testing, staff interfaces, documentation, and governance approval for one platform or consortium.","confidence":"LOW","assumptions":["A vendor or open-source maintainer must integrate with an existing routing platform.","Approximately two to five technical FTE-years plus product, security, and ILL subject-matter effort.","Existing eligibility logic is reused rather than rebuilt.","This estimate excludes consortium-wide replacement of the underlying ILL platform."],"source_ids":["S3","S4","S6","S7"]},"operational_launch":{"band_2026_usd":"50K_TO_250K","scope":"Separately authorized limited live pilot, staff training, shadow period, staged activation, incident response, independent evaluation, and rollback support for a subset of requests or libraries.","confidence":"LOW","assumptions":["Eight-to-twelve-week controlled launch after offline success.","Existing platform infrastructure and startup integration are complete.","No changes to eligibility, licensing, payment, or collection policies.","Dedicated operational monitoring and immediate baseline-router rollback are available."],"source_ids":["S3","S4","S5","S6"]},"annual_recurring":{"band_2026_usd":"50K_TO_250K","scope":"Data quality operations, periodic re-estimation, spectral-gap and drift monitoring, model and policy review, security maintenance, incident handling, reporting, and approximately 0.5–1.5 FTE of technical and governance support.","confidence":"LOW","assumptions":["One consortium-scale deployment on existing infrastructure.","No major vendor license increase or full platform redevelopment.","Quarterly governance review and continuous automated health monitoring.","Material topology or policy changes trigger revalidation funded within the band."],"source_ids":["S3","S4","S6","S7"]}},"verified_pipeline_gates":{"externally_supported_problem":{"status":"YES","reason":"Direct transaction research documents supplier-side concentration and operational systems explicitly address lender overload and balancing, although the spectral-hidden component is not yet demonstrated in ILL.","source_ids":["S1","S2","S3","S4"]},"externally_credible_adopter_or_authorizer":{"status":"YES","reason":"The ReShare Steering Committee is an identifiable resource-sharing authority responsible for roadmap, resources, risk, and funding, with consortium representatives; actual tenant participation remains to be secured.","source_ids":["S6"]},"distinct_testable_incremental_claim":{"status":"YES","reason":"The remaining claim specifies a spectral-plus-removal trigger, named routing comparators, two required resilience outcomes, and a fulfillment-time noninferiority constraint.","source_ids":["S3","S4","S5","S8"]},"bounded_next_evidence_step":{"status":"YES","reason":"A single-consortium, no-more-than-12-month retrospective audit and held-out replay can terminate with a go/no-go report and no live routing action.","source_ids":["S2","S3","S4","S5"]},"no_unresolved_safety_or_authority_stop":{"status":"YES","reason":"The proposed first step is offline, institution-pseudonymized, excludes patron-level evaluation and live changes, and can halt before analysis if consent, confidentiality review, or replay validity fails. No opened source reveals a categorical prohibition.","source_ids":["S5","S6","S7"]},"credible_cost_scope_and_range":{"status":"YES","reason":"All four resource-equivalent ranges have bounded scopes and staffing or integration assumptions. Confidence is only moderate for the retrospective study and low for deployment because no consortium data audit or vendor quotation exists.","source_ids":["S2","S3","S4","S6","S7"]}},"next_evidence_step":"Recruit one ReShare-affiliated or comparable consortium and obtain written tenant-level data-use authorization. For a fixed window of at most 12 months, first audit whether at least 90% of routed requests have reconstructable eligibility sets, routing events, and terminal outcomes; stop if this threshold is missed or missingness is materially lender-correlated. Preregister the directed weighted operator, fit window, spectral-gap and score-stability requirements, sensitivity trigger, maximum rerouting share, and a consortium-set median-time noninferiority margin. Fit on the earlier segment and replay on a later held-out segment: (1) the recorded router, (2) request-volume ordering, (3) degree ordering, (4) random diversification, (5) documented queue-aware or ratio-based routing where reconstructable, and (6) spectral-threshold diversification. Report maximum lender-load share, unfilled share, median and 90th-percentile fulfillment time, per-lender removal disruption, cross-load transferred to other lenders, spectral gap, score stability across windows, and confidence intervals. Falsify the intervention if it fails to reduce both maximum load share and removal disruption versus the existing router, provides no reproducible advantage over any simpler comparator, breaches the predeclared time margin, concentrates load on smaller lenders, or loses its spectral gap or ranking stability. The output is an offline go/no-go report only.","blocking_evidence":["No target consortium has committed proprietary transaction, eligibility, and outcome logs or staff time.","The prevalence and magnitude of outage-sensitive lender dependence beyond volume, degree, queue, and turnaround signals are unknown.","Eligible-but-unselected lender outcomes may be unobservable, making faithful counterfactual replay impossible without stronger modeling assumptions or prospective shadow data.","No evidence yet shows that the dominant mode is separated and stable across operational windows.","No empirical estimate exists for improvement over queue-aware, ratio-based, randomized, volume, and degree comparators.","Tenant-specific confidentiality, retention, contractual, and data-sharing authority have not been reviewed.","Deployment and recurring cost bands lack vendor quotations and a platform-specific integration assessment.","Proprietary routing logic, patents, local consortium implementations, and terminology-mismatched studies remain outside the bounded prior-art screen."],"research_disposition":"PARTNERED_RESEARCH_PROGRAM","world_novelty_boundary":"This evaluation is a bounded open-web evidence and prior-art screen using eight opened sources. It does not measure world novelty, patentability, freedom to operate, market size, or realized impact. The reviewed record establishes ILL concentration, transaction-network analysis, dynamic load-aware routing, randomized diversification, and generic spectral node-removal analysis as prior art, but did not identify the complete ILL-specific combination. Proprietary product logic, patents, non-indexed local consortium practices, and research using different terminology could eliminate the remaining distinction.","arm":"SENTINEL","candidate_version":0,"controller_recommendation":{"action":"STOP_EMPIRICAL_RESEARCH_NEEDED","repairable":false,"material_progress_observed":true,"progress_targets":["Secure written participation and data-use authorization from one consortium-level routing authority.","Complete a data audit showing at least 90% reconstructable eligibility sets, routing events, and terminal outcomes without materially lender-correlated missingness.","Demonstrate a stable, predeclared spectral gap and reproducible lender ranking across rolling fit windows; otherwise abandon the single-mode policy.","On held-out replay, reduce both maximum lender-load share and lender-removal fulfillment disruption relative to the recorded router and all simpler comparators while satisfying the fulfillment-time noninferiority bound.","Show that gains persist after tail-delay, unfilled-share, scarce-holdings, lender-size, and cross-load-transfer checks.","Obtain tenant-specific privacy, confidentiality, contractual, and governance approval before any prospective shadow or live test.","Replace low-confidence deployment and recurring estimates with a platform integration assessment and two independent staffing or vendor quotations.","Conduct a separate bounded technical-documentation and patent differentiation review before making any novelty or freedom-to-operate representation."],"reason":"Web evidence establishes a meaningful concentration problem, credible authorizers, feasible technical components, and substantial prior-art collision, but it cannot determine whether the spectral signal adds value in a real consortium. The decisive evidence requires proprietary eligibility and outcome logs, consortium authorization, counterfactual replay, and potentially prospective shadow testing. Those are empirical and authority-dependent rather than repairable by further ordinary web research."}}