43:["$","title",null,{"children":"Linear A: which signs appear together? | Smoked Haddock | Observable"}] 4b:T4dc,

Colocation analysis of the Linear A corpus.

Method. Each inscription is treated as a basket containing the distinct Linear A glyphs in its text. The Apriori algorithm finds pairs of signs that occur together in at least 1% of inscriptions. For each pair, lift compares how often the signs co-occur with how often they would if they were independent: 1 means independent, higher means they appear together more than chance. Co-occurrence is per inscription, not adjacency, and pair counts are small (roughly 18–35 inscriptions), so treat the results as exploratory. Sign values (such as "ka") are conventional readings from the dataset, not settled decipherments.

Display note. Linear A glyphs need a font that covers the Unicode block (for example Noto Sans Linear A). Without one they may appear as empty boxes; each label also gives the sign's phonetic value or ID.

Data: Navarre-AI/linear-a (JSON under linear_a/data/), compiled from GORILA, RILA Supplement 1, SigLA, Younger and lineara.xyz. SigLA-derived records are CC BY-NC-SA 4.0; the rest is CC BY 4.0. See the repo's README, CREDITS and LICENSE.

4c:[] 47:{"id":"c6163b23c4fd3943","slug":"linear-a-which-signs-appear-together","title":"Linear A: which signs appear together?","thumbnail":"88247c961a28a1c4c81653eb30b70fecaeb08b2ec40d04229d23581c3005c8b8","access_level":"public","create_time":"$D2026-10-29T10:16:04.829Z","update_time":"$D2026-10-30T10:02:49.628Z","view_time":null,"first_public_time":"$D2026-10-29T11:05:51.087Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"e7154098d804ccd3","login":"mycroft","name":"Smoked Haddock","avatar_url":"https://avatars-next.observableusercontent.com/0de24321ffaf4363ff2bdd86f04762092b5e049101dfafce274db05f65c8d923","tier":"pro_2026","bio":"Pattern radar for symbol streams","followed_by_you":false,"follows_you":false},"version":82,"publish_version":82,"role":null,"authored":false,"authors":[{"type":"individual","id":"37a2ff115a9312f2","login":"mycroft","name":"Smoked Haddock","avatar_url":"https://avatars-next.observableusercontent.com/0de24321ffaf4363ff2bdd86f04762092b5e049101dfafce274db05f65c8d923","tier":null,"bio":"Pattern radar for symbol streams","followed_by_you":false,"follows_you":false}],"members":[{"type":"team","id":"e7154098d804ccd3","login":"mycroft","name":"Smoked Haddock","avatar_url":"https://avatars-next.observableusercontent.com/0de24321ffaf4363ff2bdd86f04762092b5e049101dfafce274db05f65c8d923","tier":"pro_2026","role":"editor"}],"comments":[{"id":"7c2576403dba7397","reply_to_id":null,"content":"NLP for the dispossessed.","create_time":"$D2026-10-29T10:51:15.486Z","update_time":null,"resolved":false,"user":{"type":"individual","id":"37a2ff115a9312f2","login":"mycroft","name":"Smoked Haddock","avatar_url":"https://avatars-next.observableusercontent.com/0de24321ffaf4363ff2bdd86f04762092b5e049101dfafce274db05f65c8d923","tier":null}}],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Linear A: which signs appear together?","cells":[{"id":1,"value":"# Linear A: which signs appear together?","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":2,"value":"$4b","pinned":false,"hidden":false,"mode":"html","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"// Fetch any JSON file from the Navarre-AI/linear-a repo (via jsDelivr, which serves CORS-friendly raw GitHub content).\nconst repoBase = \"https://cdn.jsdelivr.net/gh/Navarre-AI/linear-a/\";\nasync function loadRepoJSON(path) {\n const response = await fetch(repoBase + path);\n if (!response.ok) throw new Error(`Failed to load ${path}: ${response.status}`);\n return response.json();\n}","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":8,"value":"const corpusRaw = await loadRepoJSON(\"linear_a/data/corpus.json\");\nconst signsRaw = await loadRepoJSON(\"linear_a/data/signs.json\");","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":17,"value":"

The Tablets

","pinned":false,"hidden":true,"mode":"html","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":9,"value":"const corpus = Object.values(corpusRaw);\nconst signs = Object.entries(signsRaw).filter(([k]) => k !== \"provenance\").map(([, v]) => v);\n// One row per sign occurrence (SigLA-derived), with document context\nconst occurrences = corpus.flatMap((doc) =>\n (doc.signs ?? []).map((s) => ({doc: doc.id, site: doc.site, type_doc: doc.type, period: doc.period, ...s})));\n// Word tokens per document\nconst wordTokens = corpus.flatMap((doc) =>\n (doc.words ?? []).map((word, i) => ({doc: doc.id, site: doc.site, period: doc.period, position: i, word})));\ndisplay(Inputs.table(corpus, {columns: [\"id\", \"site\", \"type\", \"period\", \"sign_count\", \"words\", \"unicode_text\"]}));\ndisplay({documents: corpus.length, signs: signs.length, occurrences: occurrences.length, wordTokens: wordTokens.length});","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":10,"value":"import * as aprioriModule from \"npm:apriori\";","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":11,"value":"const Apriori = aprioriModule.default;\nconst isLinearA = (ch) => { const c = ch.codePointAt(0); return c >= 0x10600 && c <= 0x1077f; };\n// Transactions: one per document, the distinct Linear A glyphs in unicode_text\nconst transactions = corpus\n .filter((d) => d.unicode_text)\n .map((d) => ({id: d.id, items: [...new Set([...d.unicode_text].filter(isLinearA))]}))\n .filter((d) => d.items.length > 0);\ndisplay({transactions: transactions.length});","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":12,"value":"const minSupport = 0.01;\nconst minConfidence = 0.3;\nconst result = new Apriori.Algorithm(minSupport, minConfidence, false).analyze(transactions.map((d) => d.items));\nconst supportOf = new Map(Object.values(result.frequentItemSets).flat().map((s) => [s.itemSet.slice().sort().join(\"\"), s.support]));","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":13,"value":"const glyphInfo = new Map(signs.filter((s) => s.unicode).map((s) => [s.unicode, s]));\n// Readable label: phonetic value if known, else sign id, else the bare glyph\nconst glyphLabel = (g) => {\n const s = glyphInfo.get(g);\n return s ? `${g} ${s.phonetic ?? s.id}` : `${g} (not in signs.json)`;\n};","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":14,"value":"const pairs = result.frequentItemSets[2].map(({itemSet: [a, b], support}) => {\n const sa = supportOf.get(a), sb = supportOf.get(b);\n return {\n a, b,\n pair: `${glyphLabel(a)} + ${glyphLabel(b)}`,\n documents: Math.round(support * transactions.length),\n support,\n lift: support / (sa * sb),\n jaccard: support / (sa + sb - support)\n };\n});\ndisplay(Inputs.table(pairs.slice().sort((x, y) => y.lift - x.lift), {\n columns: [\"pair\", \"documents\", \"support\", \"lift\", \"jaccard\"],\n format: {support: d3.format(\".3f\"), lift: d3.format(\".2f\"), jaccard: d3.format(\".3f\")}\n}));","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":16,"value":"

Is there anything to notice?

","pinned":false,"hidden":false,"mode":"html","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":15,"value":"const topPairs = pairs.slice().sort((x, y) => y.lift - x.lift).slice(0, 25);\nconst plot = display(Plot.plot({\n title: \"Linear A signs that turn up together far more often than chance\",\n subtitle: `Top 25 sign pairs by lift, among pairs co-occurring in ≥${Math.round(minSupport * 100)}% of ${transactions.length.toLocaleString()} documents. Lift 1 = independent.`,\n marginLeft: 150,\n height: 560,\n x: {label: \"Lift\", grid: true},\n y: {label: null, domain: topPairs.map((d) => d.pair)},\n color: {legend: true, label: \"Documents\", scheme: \"blues\"},\n marks: [\n Plot.barX(topPairs, {y: \"pair\", x: \"lift\", fill: \"documents\", tip: true}),\n Plot.ruleX([1])\n ]\n}));","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":18,"value":"

The Tablets

Is there anything to notice?

Stay tooned or pitch in with your suggestion!