{ "schema": 2, "generated_at": "2026-11-02T21:26:32Z", "script": "scripts/build-research-graph.mjs", "source_db": "/Users/vidigalp/dev/github.com/vidigalp/literature/data/literature.db", "source_db_fingerprint": "92f36c5d4fa83aef:1048577", "counts": { "papers_total": 2669, "papers_verified": 2659, "edges_curated": 2391, "nodes_drawn": 746, "edges_drawn": 1299, "semantic_edges": 638, "cites_edges": 661, "dangling_dropped": 81, "authored_total": 23, "authored_drawn": 10, "authored_without_edges": 13, "retracted_shown": 1, "relation_counts": { "CITES": 661, "EXTENDS": 210, "COMPLEMENTS": 262, "USES_METHOD": 36, "OPERATIONALIZES": 61, "CONTRASTS_WITH": 64, "SHARES_BENCHMARK": 2, "REFUTES": 3 } }, "by_relation_drawn": { "CITES": 661, "EXTENDS": 210, "COMPLEMENTS": 262, "USES_METHOD": 36, "OPERATIONALIZES": 61, "CONTRASTS_WITH": 64, "SHARES_BENCHMARK": 2, "REFUTES": 3 }, "keep_rules": [ "endpoint of a curated semantic (non-CITES) edge with >=2 curated connections", "authored record with >=1 curated edge, plus its full 1-hop neighborhood", "all 16 evidence-chain spine anchors plus their semantic 1-hop neighborhood", "top-8 most central per display area (directed PageRank over curated relations) plus strongest neighbor", "post-filter prune: any node with 0 induced edges in keepEdges is removed", "excluded alias and companion keys: 219" ], "foundation_split_rules": [ { "id": "foundations-viz", "rule": "/visuali[sz]|visual display|visual data|data communication|chart|graphical perception|infographic|dashboard|\\bvega\\b|scatter|infovis|\\bdotplots?\\b|\\bplots?\\b|y-axis|colou?r|rainbow|visual encod|graphical encod|perceptual|tufte|cartograph|glyph|data-ink|small multiples|screen reader/i", "records": 90 }, { "id": "foundations-stats", "rule": "/statistic|regression|survival|\\bcox\\b|accelerated failure time|inferen|probability|calibrat|bootstrap|hypothes|experiment|sampl|bayes|estimat|reliab|life.table|turing|capture.recapture|missing data|maximum likelihood|power analysis|meta.analy|causal|confound|uplift|\\bab test|a\\/b|nonparametric|heavy.traffic|parallel trends|censored|empirical distribution|synthetic control|stepped wedge|cluster randomized|epidemiol|false-positive psychology|monetary management|goodhart|outbreak|syndromic|surveillance|changepoint|cusum|ewma|poisson|inspection schemes|precision-recall|\\broc\\b|inter-coder|agreement|measurement error|comparing partitions|structural equation|bifactor|item response|specification curve|multiverse|many analysts|over-dispersion|judgment under uncertainty|heuristics and biases|research findings are false|selection bias|composite indicators/i", "records": 199 }, { "id": "foundations-telemetry", "rule": "/\\blog\\b|\\blogs\\b|sherlog|loggrep|console|anomal|monitor|alert|incident|fault|debug|troubleshoot|root[- ]cause|telemetr|profil|trace|\\bbugs?\\b|bug report|ticket|call cent|helpdesk|help desk|outage|crash|gray failure|service level|confvalley|bursty|opprentice|aberration|complex systems fail|queue?ing/i", "records": 162 }, { "id": "foundations-systems", "rule": "/network|database|storage|filesystem|disk|cache|memory|cluster|distributed|replica|consensus|protocol|\\btcp\\b|latency|cloud|virtualiz|container|\\bgpu\\b|hardware|energy|grid|power|server|operating system|scheduling|kernel|index|b-?tree|\\blsm\\b|string search|algorithm|vldb|sigmod|asplos|sosp|osdi|nsdi|eurosys|\\bfast\\b|cidr|icde|dataflow|millwheel|datafusion|simdjson|mison|hyperscan|columnar|parquet|arrow|duckdb|redshift|flink|spark|kafka|raft|paxos|quorum|\\bwal\\b|fsync|rocksdb|monarch|query|workload|benchmark|hash coding|\\bbloom\\b|structured streaming|parallel operations|mimalloc|free list|schema evolution|data provenance|liveness|microservice|service availability/i", "records": 161 }, { "id": "foundations-se", "rule": "/software|program|requirement|testing|\\bcode\\b|maintain|engineering|agile|story point|refactor|version control|compiler|decomposing systems into modules|\\bmodular|reproducib|replication|documentation|code review|good practice|silver bullet|organizational|accessib|wcag|usability|human.(computer|factors|interface)|customer service|routing and staffing|satisfaction|SERVQUAL|SERVPERF|productivity|copilot|targets and gaming|psychological safety|clinical diagnosis/i", "records": 142 }, { "id": "foundations-other", "rule": "remainder", "records": 297 } ], "violations": { "chain_anchors_missing": [], "authored_aliases_excluded": [ "aghili2025protecting", "ahmed2023rootcausellm", "ahocorasick1975", "aiops_llm_survey_2025", "alexander-2025-arxiv", "alvaro2015ldfi", "andy-2024-arxiv", "angelopoulos-ltt-2022", "angelopoulos-ppi-2023", "angelopoulos2024conformal", "angelopoulos2025ltt", "aoyang-2026-arxiv", "bairavasundaram-2007", "barnett-2005", "bellare-1996", "benjamini1995controlling", "benzhao2025loglite", "bevan-2006", "beyondcal2026", "borusyak2024", "borusyak2024revisiting", "brooks1987nosilver", "bufalino-2025", "c:22", "callaway-2021", "callaway2021", "cameron2008", "cameron2008bootstrap", "cameron2008wildcluster", "campbell-1979", "carbone2017flinkstate", "carbone2017state", "chen-2019", "chen-2020", "chen-2024", "chen-rcacopilot-2024", "chen2007hashjoin", "chen2023frugalgpt", "chernozhukov2018", "chicco-2020", "companion_gorust2026", "companion_paper05_pv2", "companion_paper08_savi", "companion_paper09_p04", "companion_paper10_p05", "companion_wtlp1_2026", "corbett-2012", "corbett2013spanner", "cui2024logeval", "daly2020edm", "davidovic2018", "dawid1979em", "deeks-2004", "defraeye-2016", "deng2013", "divlog2024", "dixon2010effort", "efron1975data", "elkan2001cost", "field-2007", "fleetcanary605", "fleming2023hunter", "frenay-2014", "geifman2019bias", "ghosh2022fightincidents", "gilardi-2023", "glerum-2009", "gregg-2012", "gunawi-2014", "gunawi-2016", "gunawi-2018", "gwet-2008", "heckman-1979", "herzig-2013", "howard2021", "huang-2017", "huang-2020", "ibrahim-2016", "jeong-2009", "jev2026axioms", "jevagent2026", "jha-itbench-2025", "jiaqi2025llmsrclog", "kadavath2022language", "kaldor-2017", "keene-1983", "kerr-1975", "khan-2022", "kikuchi2025chaoseater", "kulkarni2014hlc", "kuschewski-2023", "leskovec-2007", "li2026beyondcalibration", "li2026calibratecurate", "li2026jevasajudge", "lianmin-2023-arxiv", "librelog2024", "lin-2016", "llm_srclog2024", "logeval2024", "loghub2_2024", "lou2020omegagen", "lu2026indepth", "luan-2024-arxiv", "luo2020lsm", "ma2024librelog", "ma2026llm4log", "mackinnon-cr3j-2023", "mackinnon2023cr3j", "malkov2020efficient", "mantel1959", "messoudi2026calibration", "montgomery2020escalationfe", "muennighoff2022mteb", "murphy-1973", "naeini2015bbq", "nosek2018prereg", "nosek2018preregistration", "ong2024routellm", "ongaro2014raft", "overconfidencejudge2025", "panickssery2024llm", "peeters2025llmem", "pelkonen-2015", "polyzotis2018datalifecycle", "polyzotis2018lifecycle", "preprocessing_2024", "putze2007bloom", "putze2009cache", "qiaolin2024preprocessing", "qin2024preprocessing", "rabanser-2019", "raglog2024", "ramdas2023", "roberts-2017", "robertson2009bm25", "romera-paredes-2023", "rousseeuw1993qn", "saerens-2002", "schroeder-2026", "shi2025judging", "shreya-2024-arxiv", "siraaj-2025-arxiv", "siyu2026delog", "steiner2025finetuning", "sun2021", "sun2022sieve", "sun2025llmsrclog", "taeyoon-2026-arxiv", "taft-2020", "tang2015", "tang2015configerator", "tian2023just", "tibshirani-2019", "tidb2026hardware", "tu-2019", "verbitski2017aurora", "vidigal-2026", "vidigal-2026-textblob", "vidigal2026", "vidigal2026deonticmem", "vidigal2026mcpegressdlp", "vidigal2026memory", "vidigal2026p01", "vidigal2026p02", "vidigal2026precog", "vidigal2026rcacalib787", "vidigal2026sentiment", "vidigal2026textblob", "vidigal2026upgraderisk", "vidigal2026wtljoss", "vidigal2026zenododlp", "vidigal2027canary", "vidigal2027cranfield", "vidigal2027fleet", "vidigal2027mcpfirewall", "vidigal2027memorygatekeeper", "vidigal2027paper07a", "vidigal2027paper08", "vidigal2027paper09", "vidigal2027rca", "vidigal2027rcacalib", "vidigal2027upgraderisk", "viktor-2025-arxiv", "vovk2021", "wang-ramdas-2022", "wang2022", "wang2024faireval", "wang2024openhands", "waudby2024betting", "waudbysmith2024", "wei-2023", "wilson1927", "wu2025incremental", "wu2025incsum", "xia2025agentless", "xiao-2026-arxiv", "xiong2024can", "xu2009console", "xu2009detecting", "xu2016early", "yang-2026", "yang2018hownot", "yinfang-2025-arxiv", "yu2026delog", "yuan2012logging", "zeng-2023", "zenodo-agentsearch-2026", "zenodo-deonticmem-2026", "zenodo-fleetcanary-2026", "zenodo-rcacalib-2026", "zenodo_fleetcanary605", "zhang2025logdb", "zhihan-2023-arxiv", "zhou-2024", "zhu-2019", "zhu2026jevagent", "zhu2026jevharn", "zimmermann2010bugreport" ], "anonymous_companions_not_authored": [ "companion_gorust2026", "companion_paper05_pv2", "companion_paper08_savi", "companion_paper09_p04", "companion_paper10_p05", "companion_wtlp1_2026" ], "authored_without_edges": [ "vidigal2026agentsearch56k", "vidigal2026collector", "vidigal2026cranfieldrag", "vidigal2026fleetcanary605", "vidigal2026logitfusion", "vidigal2026pv1", "vidigal2026rcacalib", "vidigal2026yugabytedbatscale", "vidigal2027csat", "vidigal2027memory", "vidigal2027paper05", "vidigal2027paper10", "vidigal2027revenuephysics" ], "peer_reviewed_without_identifier": [ "Ando2005", "Hinton06", "Samuel59", "abadie2010", "abdallah2025rerankers", "abdelnabi2025firewalls", "abitbol2026revolut", "advani-2026", "aetherlog2024", "agentarch2025", "aggarwal2024automix", "akhtar2025llm_event_log_survey", "andrew2007scalable", "angelopoulos2021ltt", "angelopoulos2023conformal", "angelopoulos2024conformal", "angelopoulos2024crc", "angelopoulos2024ppipp", "anomaly_datasets", "ansari2024chronos", "arkhangelsky2021", "asai2024selfrag", "askell2021alignment", "assuncao2025modernization", "aws-dsql-docs-2026", "aws-limitless-docs-2026", "bai2022training", "bala2010upgrades", "bansal2026uber", "barbieri-etal-2020-tweeteval", "barker2025faster", "bartlett2026compliance", "bavaresco2025beyond", "beck2025log_parsing_llm_review", "bell2002bias", "benzhao2025loglite", "besker2018ictd", "besker2019td", "besker2020morale", "beyondcal2026", "bhagwan2014", "biasedllm2025", "boehm1981software", "bousetouane2026acc", "brecko2023versioning", "brodersen2015", "brooker-2026", "brundage2026frontier", "brutlag2000aberrant", "brynjolfsson-2025", "c:83", "c:84", "calibration2025", "caminal2025filtered", "candes2023survival", "cedlog2024", "chen-rcacopilot-2024", "chen2024agentpoison", "chen2024frugalgpt", "chen2026raise", "cheng2025brt", "chhikara2025mem0", "chillarege1992odc", "chirkova2025provence", "choudhary2007saas", "chuang2025confident", "cisco2025codes", "citus-repo-2026", "coalm2025", "condorcet2024ensemble", "confidencebench2026", "coqueret2026randomness", "corbett-2012", "corbett2012spanner", "cormack2009rrf", "crdb-docs-2026", "crosslingualpii2026", "crossmodel2026disagreement", "crossvendor2025classification", "crossvendor2025writing", "daly2020changepoint", "das2024larimar", "dash2026mpbench", "demszky-etal-2020-goemotions", "deusser2026jev", "dietz2025principles", "ding2024hybridllm", "dixon2010ces", "dong2021rocksdb", "dong2025minja", "du2024improving", "dubey2024llama3", "dutta2024accuracy", "edge2024graphrag", "efron2026ppi", "egami2022", "elkan2001cost", "elkan2001foundations", "emnlp2025smalllarge", "enrichlog2024", "fagerland2013mcnemar", "falkner2018bohb", "farzi2025criteria", "feder2022", "feng2023kuzu", "ffrenchconstant2026confidencebench", "fleetcanary605", "flores2025clinical", "frantar2023gptq", "fu2024autoraghp", "garciahernandez2026loginsight", "garg-memfail-2026", "geifman2017selective", "gibbs2024online", "gim2024promptcache", "gloaguen-2026", "goswami2024moment", "governatori2018deontic", "gretton2012", "gretton2012mmd", "groot2024overconfidence", "grunwald2024", "guha2023legalbench", "guha2024smoothie", "guo2025lightrag", "guo2025retrieval", "gupta-etal-2014-sentiment", "gutierrez2024hipporag", "gutierrez2025hipporag2", "hakim2026preserving", "han2024lasrca", "han2024wildguard", "han2025metadata", "han2025tale", "hasan2025mcp", "hcrt:83", "hendrycks2021cuad", "holm1979", "howard2020", "howard2021", "hu2026cbstail", "hu2026jevadv", "hu2026jevadvbench", "huang2024crashtranslator", "huang2024relation", "huang2026mcp", "huang2026pmpa", "huang2026runoff", "hybrid_lstm_transformer_2025", "ieee2024helpdesk", "ignatiadis2024", "inan2023llamaguard", "incremental2025summarize", "iscan2026rscbmc", "jain2024livecodebench", "jamieson2016nonstochastic", "jeong2024adaptiverag", "jev2026axioms", "jev2026evaluating", "jeva2026", "jevadv2026", "jevagent2026", "jin2023sensitivity", "jin2024timellm", "jin2025searchr1", "johari2022", "jonkers2024wcps", "judgebench2025", "jung2025trustescalate", "k4_2024", "kadavath2022know", "kadavath2022language", "kamath2020selective", "kang2023libro", "kang2024diverse", "keith2020", "kelvich-2022", "keough2026granularity", "kerr1998", "killick2012optimal", "kim2023propile", "kim2024autorag", "kim2025correlated", "kingman1961single", "klein2017fast", "kohli2026ninejudges", "kopiczko2025bitune", "krichevsky1981", "kroll2026unsafe", "kumar2026verdicts", "kurtic2024bf16", "kwon2023vllm", "landauer2022aitldsv2", "landauer2023deep_learning_log_survey", "langley00", "laurer2024deberta", "le2022narya", "leakagebench2026", "lee2023eadro", "lee2024readagent", "lee2026ppi", "lehuy2026cacd", "lei2021counterfactual", "levy2020narya", "li2018hyperband", "li2021mfeshb", "li2021tracerca", "li2026axioms", "li2026beyond", "li2026beyondcalibration", "li2026jeva", "li2026jevasajudge", "li2026jevjudge", "li2026scoringbias", "librelog2024", "lin2018ming", "lin2022microcbr", "lin2024awq", "lindon2024", "lindon2026", "lipcon-2015", "liu2023thinkinmemory", "liu2024itransformer", "liu2025guardreasoner", "liu2026homogenization", "llm_srclog2024", "llmpersonality2025", "log2graphs2024", "logdb2025", "logelectra2024", "logeval2024", "logformer2024", "loghub2_2024", "logllama2025", "logllm2024", "logparser_llm2024", "logrca2024", "logsage2024", "logsd2024", "longcontext2025", "loria2018textblob", "lu2026doublycal", "luk2024", "luo2020lsm", "luo2026nudgethejudge", "ma2026krone", "mackinnon1985", "mackinnon2023", "magnusson2025signal", "maharana2024locomo", "majorityrules2025", "mann1945nonparametric", "matthews1975comparison", "mckeeman1998differential", "mcpitp2026", "mehra2012upgrades", "mehta2020rex", "messoudi2026calibration", "messoudi2026ltt", "mo2025kggen", "mt2st2025", "mtprompt2025", "nagaraj2012distalyzer", "nannini2026aiagents", "neon-docs-2026", "nie2023patchtst", "nissenbaum2004privacy", "nixon2019measuring", "norman2026reliability", "ocean2025", "odonoghue2005stable", "ong2025routellm", "ongaro-2014", "ontologx2025", "openrca2-2026", "oraby-etal-2017-sentiment", "orbach2025analysis", "ouyang2022instructgpt", "overconfidencejudge2025", "overthinking2025", "packer2023memgpt", "pan2024llmlingua2", "panickssery2024llm", "panickssery2024selfpref", "paparrizos2022vus", "park2023generative", "peeters2025entity", "peeters2025llmem", "peft_log_2025", "personalityrisk2025", "pournaderi2026sketch", "pradeep2023rankzephyr", "pradeep2025trecrag", "preprocessing_2024", "prism1_2026", "qin2024prp", "r:80x", "rabanser-2026", "rabanser2019failing", "raglog2024", "rahmani2025judging", "ramdas2023gametheoretic", "rasmussen2025zep", "rasooli-tetrault-2015", "rath2026quantization", "rau2024bergen", "reflectevo2025", "robbins1956empirical", "romano2019cqr", "runoff2026", "safetytax2025", "sahoo2026sycophancy", "santoni2026cmv", "schuirmann1987tost", "sculley2015hiddendebt", "selfexplanation2025", "selfserveer2026", "shafer2021", "sharifymoghaddam2025rankllm", "shi2025judging", "shin2025survival", "shinn2023reflexion", "silentchurn2025", "sinq2025", "sitaram-etal-2019-survey", "siyu2026delog", "slmensemble2026", "snr2026", "spanner-docs-2026", "springer2024metatokenization", "srivastava2025memorygraft", "staab2024beyond", "stanislawek2021kleister", "stekhoven2012missforest", "sumita2024cogbiassurvey", "sun2026bm25", "sun2026scoregranularity", "sun2026typesafe", "systemone2026benchmark", "tacl2025liu", "tan2025judgebench", "tang2026tdmchecklist", "tao2024medicine", "tarvo2015", "taylor2018forecasting", "temporal_logical_attention_2024", "teng2026prism1", "thinkingstyle2025", "tian2025overconfidence", "tian2026gala", "tibshirani-2019", "tibshirani2019conformal", "tickit2024bytedance", "tidb-docs-2026", "tplogad2024", "traub2024overcoming", "truong2020selective", "tu2018careful", "tunstall2022setfit", "uddin2026conformalrouting", "upadhyay2024largescale", "upadhyay2024umbrela", "upadhyay2025trecrag", "veitch2020", "veitch2020text", "verga2024poll", "vidigal-2026-q1", "vidigal-2026-q2", "vidigal2026p01", "vidigal2026p02", "vidigal2027canary", "vidigal2027cranfield", "vidigal2027csat", "vidigal2027fleet", "vidigal2027mcpfirewall", "vidigal2027memory", "vidigal2027memorygatekeeper", "vidigal2027paper05", "vidigal2027pv2", "vidigal2027rca", "vidigal2027rcacalib", "vidigal2027revenuephysics", "vidigal2027upgraderisk", "ville1939", "vonwright1951deontic", "vovk2024symmetry", "wang2004peerpressure", "wang2023selfconsistency", "wang2023voyager", "wang2024docllm", "wang2024memoryllm", "wang2024scm", "wang2025argus", "wang2025astuterag", "wang2025comem", "wang2025mcptox", "wang2025mindguard", "wang2025openhands", "wang2025rankedvoting", "wang2026cloudopsbench", "wang2026contamination", "wang2026jevlitemas", "wang2026lean4agent", "wataoka2025selfpreference", "watermark2026forensic", "wee2025alignment", "wei2022chain", "wen2016locus", "wen2019robuststl", "woo2024moirai", "wu-clasheval-2024", "wu2023timesnet", "wu2025incremental", "wu2025incsum", "wu2025longmemeval", "wu2025realtimeprp", "wu2026quantization", "xiong2024uncertainty", "xminstitute2025", "xu2016pcheck", "xu2025amem", "xu2025openrca", "yan2024corrective", "yan2024crag", "yang2024doubly", "yang2024sweagent", "yang2025gateddeltanet", "yang2025qwen3", "yang2025swesmith", "yao2023react", "ye2025justice", "yoran2024making", "yu2023nezha", "zecchin2025wcrc", "zeller1999", "zeng2024shieldgemma", "zeng2025shieldgemma2", "zeng2026autoragtuner", "zhang2021cloudrca", "zhang2023diagfusion", "zhang2023repocoder", "zhang2024anymatch", "zhang2024autocoderover", "zhang2024edc", "zhang2024incontext", "zhang2024lmpace", "zhang2024monitorassistant", "zhang2025tamo", "zhang2026piitrace", "zhao2023cloudigger", "zhao2025languagemodelcouncil", "zheng2023judging", "zheng2024sglang", "zheng2025degrag", "zhou2012buglocator", "zhou2026micp", "zhu2026fcps", "zhu2026jevagent", "zhu2026jevagentharness", "zhu2026jevharn", "zou2025seara" ], "future_dated_records": [ "companion_paper05_pv2", "companion_paper08_savi", "companion_paper09_p04", "companion_paper10_p05", "vidigal2027canary", "vidigal2027cranfield", "vidigal2027csat", "vidigal2027fleet", "vidigal2027mcpfirewall", "vidigal2027memory", "vidigal2027memorygatekeeper", "vidigal2027paper05", "vidigal2027paper07a", "vidigal2027paper08", "vidigal2027paper09", "vidigal2027paper10", "vidigal2027pv2", "vidigal2027rca", "vidigal2027rcacalib", "vidigal2027revenuephysics", "vidigal2027upgraderisk" ], "year_reconciled_from_key": [ "athey2019surrogate", "brodie2005symptom", "chen1998metamorphic", "christmansson1996errorset", "kitchenham2004ebse", "mcc2025imbalance", "montgomery2020escalationfe", "shneiderman1996eyes", "slagell2005sharing", "ustun-2019-riskslim", "xu2002prefix", "yu2024denum", "zhu2025judgelm" ], "temporal_direction_normalized": 23, "dangling_edges_dropped": 81 }, "limits": { "cites_rule": "The drawn dataset also holds 661 mechanically extracted citation edges (from 1413 publisher reference-list links in the vault); they are hidden by default and can be switched on in paper views.", "authored_note": "13 of 23 canonical authored records have no curated relation yet; the authored index lists all of them, drawn or not.", "edge_note": "Relation notes are agent-drafted (drafting model on every edge) and reviewed in the study pipelines. Nothing on this page is labelled human-curated. Peer-reviewed status is only shown where a resolvable external identifier exists." } }