45:["$","title",null,{"children":"Thematic cross referencing in Wikipedia | Smoked Haddock | Observable"}] 4d:T50f,The file `atlas2.json` holds a single key, `hymap`, whose value is a list of 110 Wikipedia page entries. Seven of these are sub-sections of pages that are already included, with titles carrying URL fragments such as `#cite note-…`. I removed them, which leaves **103 distinct pages**. Each page has a `title`, the article text split into `sentences` (about 39,800 in total), and a `network` of keyword nodes and weighted edges. The nodes are salient terms with occurrence counts, such as *Parts*, *Properties* and *Meaning* in the "Holism" article. The edges link terms that co-occur, and each carries a `count` and per-occurrence `confidence` scores. A single page has up to 66 nodes and 755 edges. Across all pages there are 3,168 nodes and 12,041 edges, covering 1,913 distinct terms. The `itemsets` and `rules` fields are empty on every page. The pages range widely: philosophy and logic (Holism, Gödel's incompleteness theorems, Derrida, Plato), AI and language models (Geoffrey Hinton, OpenAI, Gemini, Claude), and a scattering of unrelated biographical and historical entries (Fiona Bruce, George III, the Suez Canal obstruction). Terms that recur across many networks, such as *Language*, *Theory*, *Models* and *Systems*, are natural anchors for cross-referencing the pages by theme.4e:T530,const selectedPages = pages.filter((p) => [selection].flat().includes(p.title)); if (!selectedPages.length) throw new Error("Select at least one page"); const selectedTitles = new Set(selectedPages.map((p) => p.title)); // Combine the selected pages into one page-shaped object: node counts and edge counts are summed const selectedPage = (() => { if (selectedPages.length === 1) return selectedPages[0]; const nodes = new Map(), edges = new Map(); for (const p of selectedPages) { for (const n of p.network.nodes) { const m = nodes.get(n.data.id); if (m) m.data.count += n.data.count; else nodes.set(n.data.id, {...n, data: {...n.data}}); } for (const e of p.network.edges) { const k = edgeKey(e.data.source, e.data.target); const m = edges.get(k); if (m) m.data.count += e.data.count; else edges.set(k, {...e, data: {...e.data}}); } } return { title: selectedPages.map((p) => p.title).join(" + "), sentences: selectedPages.flatMap((p) => [`== ${p.title} ==`, ...p.sentences]), network: {nodes: [...nodes.values()], edges: [...edges.values()]} }; })(); display({title: selectedPage.title, pages: selectedPages.length, sentences: selectedPage.sentences.length, nodes: selectedPage.network.nodes.length, edges: selectedPage.network.edges.length});4f:T1912,const graph = view((() => { const width = 800, height = 560; const nodes = selectedPage.network.nodes.map((d) => ({ id: d.data.id, count: d.data.count, x: width / 2 + (Math.random() - 0.5) * 100, y: height / 2 + (Math.random() - 0.5) * 100 })); const ids = new Set(nodes.map((d) => d.id)); // `shared` = number of OTHER pages whose network contains the same term pair const links = selectedPage.network.edges .map((d) => ({ source: d.data.source, target: d.data.target, count: d.data.count, shared: Math.max(0, (edgeIndex.get(edgeKey(d.data.source, d.data.target))?.size ?? 1) - 1) })) .filter((d) => ids.has(d.source) && ids.has(d.target)); const nShared = links.filter((d) => d.shared > 0).length; const sharedColor = "#1c7ed6", plainColor = "#aaa"; const baseStroke = (d) => (d.shared > 0 ? sharedColor : plainColor); const baseOpacity = (d) => (d.shared > 0 ? 0.75 : 0.45); const r = d3.scaleSqrt(d3.extent(nodes, (d) => d.count), [5, 22]).clamp(true); const w = d3.scaleLinear(d3.extent(links, (d) => d.count), [0.8, 5]); const color = d3.scaleSequentialLog(d3.extent(nodes, (d) => Math.max(1, d.count)), d3.interpolateBlues); const pad = 6, top = 50, labelW = 40, labelH = 14; const clampX = (d) => Math.max(r(d.count) + pad + labelW / 2, Math.min(width - r(d.count) - pad - labelW / 2, d.x)); const clampY = (d) => Math.max(top + r(d.count) + labelH, Math.min(height - r(d.count) - pad, d.y)); const svg = d3.create("svg").attr("viewBox", [0, 0, width, height]).attr("width", width) .style("max-width", "100%").style("height", "auto").style("border", "1px solid #ddd"); svg.append("text").attr("x", 10).attr("y", 18).attr("font-size", 12).attr("fill", "#666") .text(`${selectedPage.title}: ${nodes.length} terms, ${links.length} links. Click a term or a link to filter sentences; drag to move.`); // legend const lg = svg.append("g").attr("transform", "translate(10,36)").attr("font-size", 11).attr("fill", "#444"); lg.append("line").attr("x1", 0).attr("x2", 28).attr("y1", 0).attr("y2", 0).attr("stroke", sharedColor).attr("stroke-width", 3).attr("stroke-opacity", 0.75); lg.append("text").attr("x", 34).attr("y", 4).text(`solid blue: also linked in other pages (${nShared})`); lg.append("line").attr("x1", 290).attr("x2", 318).attr("y1", 0).attr("y2", 0).attr("stroke", plainColor).attr("stroke-width", 2).attr("stroke-dasharray", "4,3"); lg.append("text").attr("x", 324).attr("y", 4).text(`dashed grey: only on this page (${links.length - nShared})`); const linkG = svg.append("g"); const link = linkG.selectAll("line.vis").data(links).join("line").attr("class", "vis") .attr("stroke", baseStroke).attr("stroke-opacity", baseOpacity) .attr("stroke-dasharray", (d) => (d.shared > 0 ? null : "4,3")) .attr("stroke-width", (d) => w(d.count)); const hit = linkG.selectAll("line.hit").data(links).join("line").attr("class", "hit") .attr("stroke", "transparent").attr("stroke-width", 10).style("cursor", "pointer"); hit.append("title").text((d) => `${d.source.id ?? d.source} – ${d.target.id ?? d.target}: ${d.count} co-occurrences; ${d.shared ? `also in ${d.shared} other page${d.shared === 1 ? "" : "s"}` : "only on this page"}`); const node = svg.append("g").selectAll("g").data(nodes).join("g").style("cursor", "pointer"); node.append("circle").attr("r", (d) => r(d.count)).attr("fill", (d) => color(Math.max(1, d.count))) .attr("stroke", "#333").attr("stroke-width", 1); node.append("text").text((d) => d.id).attr("font-size", 11).attr("text-anchor", "middle") .attr("dy", (d) => -r(d.count) - 3).attr("paint-order", "stroke").attr("stroke", "white").attr("stroke-width", 3); node.append("title").text((d) => `${d.id}: ${d.count} occurrences`); const sim = d3.forceSimulation(nodes) .force("link", d3.forceLink(links).id((d) => d.id).distance(60)) .force("charge", d3.forceManyBody().strength(-120)) .force("collide", d3.forceCollide((d) => r(d.count) + 3)) .force("x", d3.forceX(width / 2).strength(0.1)) .force("y", d3.forceY((height + top) / 2).strength(0.1)) .on("tick", () => { nodes.forEach((d) => { d.x = clampX(d); d.y = clampY(d); }); for (const sel of [link, hit]) { sel.attr("x1", (d) => d.source.x).attr("y1", (d) => d.source.y).attr("x2", (d) => d.target.x).attr("y2", (d) => d.target.y); } node.attr("transform", (d) => `translate(${d.x},${d.y})`); }); node.call(d3.drag() .on("start", (e) => { if (!e.active) sim.alphaTarget(0.3).restart(); e.subject.fx = e.subject.x; e.subject.fy = e.subject.y; }) .on("drag", (e) => { e.subject.fx = clampX({x: e.x, count: e.subject.count}); e.subject.fy = clampY({y: e.y, count: e.subject.count}); }) .on("end", (e) => { if (!e.active) sim.alphaTarget(0); e.subject.fx = null; e.subject.fy = null; })); const el = svg.node(); el.value = null; const key = (s) => (s ? s.terms.join("|") : ""); const isSel = (l, terms) => terms.length === 2 && ((l.source.id === terms[0] && l.target.id === terms[1]) || (l.source.id === terms[1] && l.target.id === terms[0])); function select(sel) { el.value = sel; const terms = sel ? sel.terms : []; const keep = new Set(terms); if (sel && sel.type === "node") links.forEach((l) => { if (l.source.id === terms[0]) keep.add(l.target.id); if (l.target.id === terms[0]) keep.add(l.source.id); }); node.attr("opacity", (d) => (!sel || keep.has(d.id) ? 1 : 0.15)); node.select("circle").attr("stroke", (d) => (terms.includes(d.id) ? "crimson" : "#333")).attr("stroke-width", (d) => (terms.includes(d.id) ? 3 : 1)); const on = (l) => sel && (sel.type === "edge" ? isSel(l, terms) : l.source.id === terms[0] || l.target.id === terms[0]); link.attr("stroke", (l) => (on(l) ? "crimson" : baseStroke(l))) .attr("stroke-opacity", (l) => (!sel ? baseOpacity(l) : on(l) ? 0.9 : 0.05)); el.dispatchEvent(new Event("input", {bubbles: true})); } node.on("click", (e, d) => { e.stopPropagation(); const s = {type: "node", terms: [d.id]}; select(key(el.value) === key(s) ? null : s); }); hit.on("click", (e, d) => { e.stopPropagation(); const s = {type: "edge", terms: [d.source.id, d.target.id]}; select(key(el.value) === key(s) ? null : s); }); svg.on("click", () => select(null)); invalidation.then(() => sim.stop()); return el; })());50:T129f,const sentencesView = (() => { const terms = graph ? graph.terms : []; const esc = (s) => s.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"); const colors = ["#ffe066", "#8ce99a"]; // word-start match, so "Claim" also matches "claims" const matchers = terms.map((t) => new RegExp(`\\b${esc(t)}`, "i")); const splitRe = terms.length ? new RegExp(`(\\b(?:${terms.map(esc).join("|")}))`, "gi") : null; const highlight = (s) => s.split(splitRe).map((part, i) => { if (i % 2 === 0) return part; const k = terms.findIndex((t) => part.toLowerCase() === t.toLowerCase()); return html`${part}`; }); const pill = (t, i) => html`${t}`; // Walk the sentences, tracking the current section heading let section = ""; const rows = []; for (const raw of selectedPage.sentences) { const s = raw.trim(); if (!s) continue; const h = s.match(/^(={2,6})\s*(.*?)\s*=+$/); if (h) { section = h[2]; rows.push({heading: h[2], level: h[1].length, section}); } else rows.push({text: s, section}); } let content = [], summary; if (!terms.length) { summary = `${selectedPage.sentences.length} sentences. Click a term or link in the graph to filter.`; content = rows.map((d) => { if (d.heading) { const el = document.createElement("h" + Math.min(6, d.level + 1)); el.style.margin = "1em 0 0.3em"; el.textContent = d.heading; return el; } return html`

${d.text}.

`; }); } else { const hits = rows.filter((d) => d.text && matchers.every((m) => m.test(d.text))); summary = html`${hits.length} sentence${hits.length === 1 ? "" : "s"} containing ${terms.length === 2 ? "both " : ""}${terms.map((t, i) => html`${i ? " and " : ""}${pill(t, i)}`)}`; // For a clicked edge, FIRST list every page that uses the link; click one to add it to (or remove it from) the menu selection if (graph.type === "edge") { const users = [...(edgeIndex.get(edgeKey(terms[0], terms[1])) ?? [])] .sort((a, b) => b[1] - a[1] || d3.ascending(a[0], b[0])); const nNew = users.filter(([t]) => !selectedTitles.has(t)).length; content.push(html`

${users.length} page${users.length === 1 ? "" : "s"} use this link${nNew ? ` (click to add ${nNew === 1 ? "the other" : "the others"} to the selection)` : " (all already selected)"}

`); for (const [title, count] of users) { const on = selectedTitles.has(title); const toggle = () => setPage(title, !on); content.push(html`

{ if (e.key === "Enter" || e.key === " ") { e.preventDefault(); toggle(); } }}> ${on ? "✓ " : "+ "}${title} ${pill(terms[0], 0)} – ${pill(terms[1], 1)} ${count} co-occurrence${count === 1 ? "" : "s"}

`); } content.push(html`

Matching sentences in the selected page${selectedTitles.size === 1 ? "" : "s"}

`); } content.push(...(hits.length ? hits.map((d) => html`

${d.section ? html`

${d.section}

` : ""}

${highlight(d.text)}.

`) : [html`

No sentence on this page contains ${terms.length === 2 ? "both terms" : "this term"}.

`])); } return html`

${selectedPage.title}: ${summary}

${content}

`; })(); display(sentencesView);51:Taa0,const clustering = (() => { // Only the pages currently selected in the menu take part const P = selectedPages; // Feature sets: edges (undirected term pairs) with enough co-occurrences on a page const feats = P.map((p) => { const s = new Set(); for (const e of p.network.edges) if (e.data.count >= minCount) s.add(edgeKey(e.data.source, e.data.target)); return s; }); const df = new Map(); feats.forEach((s) => s.forEach((e) => df.set(e, (df.get(e) ?? 0) + 1))); const ok = (e) => df.get(e) >= 2 && df.get(e) <= maxDf; const sets = feats.map((s) => new Set([...s].filter(ok))); const n = P.length; const sim = Array.from({length: n}, () => new Float64Array(n)); for (let i = 0; i < n; i++) for (let j = i + 1; j < n; j++) { let inter = 0; for (const e of sets[i]) if (sets[j].has(e)) inter++; const u = sets[i].size + sets[j].size - inter; sim[i][j] = sim[j][i] = u ? inter / u : 0; } // Only pages with at least one neighbour sharing an edge take part; the rest are "unclustered" const active = d3.range(n).filter((i) => d3.max(sim[i]) > 0); const kk = Math.min(k, active.length); let cl = active.map((i) => [i]); const avg = (a, b) => d3.sum(a, (i) => d3.sum(b, (j) => sim[i][j])) / (a.length * b.length); while (cl.length > kk) { let best = -1, bi = 0, bj = 1; for (let i = 0; i < cl.length; i++) for (let j = i + 1; j < cl.length; j++) { const s = avg(cl[i], cl[j]); if (s > best) { best = s; bi = i; bj = j; } } cl[bi] = cl[bi].concat(cl[bj]); cl.splice(bj, 1); } cl.sort((a, b) => b.length - a.length); const cluster = new Array(n).fill(-1); cl.forEach((m, c) => m.forEach((i) => (cluster[i] = c))); const summary = cl.map((m, c) => { const cnt = new Map(); m.forEach((i) => sets[i].forEach((e) => cnt.set(e, (cnt.get(e) ?? 0) + 1))); const top = [...cnt].filter(([, v]) => v >= 2).sort((a, b) => b[1] - a[1]).slice(0, 5) .map(([e, v]) => `${e.replace("|", " – ")} (${v})`); return {cluster: c + 1, size: m.length, sharedEdges: top.join("; "), pages: m.map((i) => P[i].title).sort().join(", ")}; }); const unclustered = d3.range(n).filter((i) => cluster[i] < 0).map((i) => P[i].title).sort(); if (unclustered.length) summary.push({cluster: "none", size: unclustered.length, sharedEdges: "(no edge shared with another selected page under these settings)", pages: unclustered.join(", ")}); return {sets, sim, cluster, summary, usedEdges: [...df.keys()].filter(ok).length, nClusters: cl.length}; })(); display({pagesClustered: selectedPages.length, usedEdges: clustering.usedEdges, clusters: clustering.nClusters, unclustered: clustering.cluster.filter((c) => c < 0).length});52:Tbf3,const clusterGraph = (() => { const width = 900, height = 600; const {sim, cluster, sets} = clustering; const nodes = selectedPages.map((p, i) => ({i, title: p.title, c: cluster[i], deg: sets[i].size})); const links = []; for (let i = 0; i < nodes.length; i++) for (let j = i + 1; j < nodes.length; j++) if (sim[i][j] > 0 && sim[i][j] >= minSim) links.push({source: i, target: j, s: sim[i][j]}); const pal = d3.scaleOrdinal(d3.schemeTableau10.concat(d3.schemeSet3)).domain(d3.range(30)); const color = (c) => (c < 0 ? "#e5e5e5" : pal(c)); const svg = d3.create("svg").attr("viewBox", [0, 0, width, height]).style("max-width", "100%").style("border", "1px solid #ddd"); svg.append("text").attr("x", 10).attr("y", 18).attr("font-size", 12).attr("fill", "#666") .text(`${nodes.length} selected page${nodes.length === 1 ? "" : "s"}, ${clustering.nClusters} cluster${clustering.nClusters === 1 ? "" : "s"}; links = Jaccard similarity of shared edges ≥ ${minSim}; grey = no shared edges. Change the page selection to alter the pages clustered.`); const link = svg.append("g").attr("stroke", "#999").selectAll("line").data(links).join("line") .attr("stroke-opacity", 0.5).attr("stroke-width", (d) => 0.5 + d.s * 6); const node = svg.append("g").selectAll("circle").data(nodes).join("circle") .attr("r", (d) => 4 + Math.sqrt(d.deg) * 1.5).attr("fill", (d) => color(d.c)).attr("stroke", "#333") .on("mouseenter", function () { d3.select(this).attr("stroke", "crimson").attr("stroke-width", 3); }) .on("mouseleave", function () { d3.select(this).attr("stroke", "#333").attr("stroke-width", 1); }); node.append("title").text((d) => `${d.title} (${d.c < 0 ? "unclustered" : "cluster " + (d.c + 1)}, ${d.deg} shared edges)`); const label = svg.append("g").attr("font-size", 9).attr("pointer-events", "none").selectAll("text").data(nodes).join("text") .text((d) => d.title).attr("paint-order", "stroke").attr("stroke", "white").attr("stroke-width", 2); const sim2 = d3.forceSimulation(nodes) .force("link", d3.forceLink(links).distance(40).strength((d) => d.s)) .force("charge", d3.forceManyBody().strength(-40)) .force("collide", d3.forceCollide(10)) .force("x", d3.forceX((d) => d.c < 0 ? width - 60 : width / 2 + ((d.c % 5) - 2) * 150).strength(0.08)) .force("y", d3.forceY((d) => d.c < 0 ? height / 2 : height / 2 + (Math.floor(d.c / 5) % 3 - 1) * 170).strength(0.08)) .on("tick", () => { nodes.forEach((d) => { d.x = Math.max(10, Math.min(width - 10, d.x)); d.y = Math.max(30, Math.min(height - 10, d.y)); }); link.attr("x1", (d) => d.source.x).attr("y1", (d) => d.source.y).attr("x2", (d) => d.target.x).attr("y2", (d) => d.target.y); node.attr("cx", (d) => d.x).attr("cy", (d) => d.y); label.attr("x", (d) => d.x + 7).attr("y", (d) => d.y + 3); }); invalidation.then(() => sim2.stop()); return svg.node(); })(); display(clusterGraph); display(Inputs.table(clustering.summary, {rows: 10, width: {cluster: 50, size: 40, pages: 500, sharedEdges: 300}}));53:[] 49:{"id":"7161183199e8cdfe","slug":"thematic-cross-referencing","title":"Thematic cross referencing in Wikipedia","thumbnail":"e96eee277b7f255510eea1ca054f0022e5015ad6d77f4edd363a9b0a556cbf48","access_level":"public","create_time":"$D2026-11-02T11:27:05.158Z","update_time":"$D2026-11-02T18:48:04.563Z","view_time":null,"first_public_time":"$D2026-11-02T12:19:59.506Z","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/e728d900ac38fc0183e4c4156893ff8c845be004d0502664062eb5a51a0604b8","tier":"pro_2026","bio":"Pattern radar for symbol streams","followed_by_you":false,"follows_you":false},"version":66,"publish_version":66,"role":null,"authored":false,"authors":[{"type":"individual","id":"37a2ff115a9312f2","login":"mycroft","name":"Smoked Haddock","avatar_url":"https://avatars-next.observableusercontent.com/e728d900ac38fc0183e4c4156893ff8c845be004d0502664062eb5a51a0604b8","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/e728d900ac38fc0183e4c4156893ff8c845be004d0502664062eb5a51a0604b8","tier":"pro_2026","role":"editor"}],"comments":[],"custom_thumbnail":true,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Thematic cross referencing in Wikipedia","cells":[{"id":1,"value":"# Thematic cross referencing in Wikipedia","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":3,"value":"const atlas = await d3.json(\"https://mycroft-pull.b-cdn.net/assets/json/atlas2.json\", {proxy: true});\ndisplay(Array.isArray(atlas) ? `array of ${atlas.length}` : Object.keys(atlas).slice(0, 30));","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":4,"value":"const allPages = atlas.hymap.map(d => typeof d.summary === \"string\" ? JSON.parse(d.summary) : d.summary);\n// Drop sub-section entries (titles with a URL fragment such as \"#cite note-…\")\nconst pages = allPages.filter(p => !p.title.includes(\"#\"));\nconst info = {\n loaded: allPages.length,\n kept: pages.length,\n removed: allPages.filter(p => p.title.includes(\"#\")).map(p => p.title.split(\"#\")[0]),\n summaryKeys: Object.keys(pages[0])\n};\ndisplay(info);","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":5,"value":"const stats = {\n pages: pages.length,\n totalSentences: d3.sum(pages, p => p.sentences.length),\n totalNodes: d3.sum(pages, p => p.network.nodes.length),\n totalEdges: d3.sum(pages, p => p.network.edges.length),\n nodesRange: d3.extent(pages, p => p.network.nodes.length),\n edgesRange: d3.extent(pages, p => p.network.edges.length),\n uniqueTerms: new Set(pages.flatMap(p => p.network.nodes.map(n => n.data.id))).size,\n nonEmptyItemsetsOrRules: pages.filter(p => p.itemsets.length || p.rules.length).length,\n repeatedTitles: d3.rollups(pages, v => v.length, p => p.title).filter(d => d[1] > 1)\n};\ndisplay(stats);","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":6,"value":"$4d","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"const titles = pages.map((p) => p.title).sort(d3.ascending);\n// keep a reference to the element so other cells can switch page programmatically\nconst menu = Inputs.select(titles, {label: \"Pages\", value: [\"Holism\"], multiple: true, size: 8});\nconst selection = view(menu);\n// Replace the selection with a single page, as if the user had picked it\nconst goTo = (title) => {\n const sel = menu.querySelector(\"select\");\n for (const o of sel.options) o.selected = o.textContent === title;\n sel.dispatchEvent(new Event(\"input\", {bubbles: true}));\n};\n// Add (on = true) or remove (on = false) one page, keeping the rest of the selection\nconst setPage = (title, on) => {\n const sel = menu.querySelector(\"select\");\n const chosen = [...sel.options].filter((o) => o.selected);\n if (!on && chosen.length <= 1) return; // always keep at least one page\n for (const o of sel.options) if (o.textContent === title) o.selected = on;\n sel.dispatchEvent(new Event(\"input\", {bubbles: true}));\n};","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":8,"value":"$4e","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":9,"value":"$4f","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":11,"value":"// Index of every edge across all pages: undirected term pair -> Map(page title -> summed co-occurrence count)\nconst edgeKey = (a, b) => [a, b].map((s) => s.toLowerCase()).sort().join(\"|\");\nconst edgeIndex = new Map();\nfor (const p of pages) {\n for (const e of p.network.edges) {\n const k = edgeKey(e.data.source, e.data.target);\n if (!edgeIndex.has(k)) edgeIndex.set(k, new Map());\n const m = edgeIndex.get(k);\n m.set(p.title, (m.get(p.title) ?? 0) + e.data.count);\n }\n}\ndisplay({edges: edgeIndex.size, sharedAcrossPages: [...edgeIndex.values()].filter((m) => m.size > 1).length});","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":10,"value":"$50","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":13,"value":"const minCount = view(Inputs.range([1, 10], {label: \"Min co-occurrences for an edge to count on a page\", step: 1, value: 1}));\nconst maxDf = view(Inputs.range([2, 60], {label: \"Ignore edges found in more than N pages\", step: 1, value: 25}));\nconst k = view(Inputs.range([2, 30], {label: \"Number of clusters\", step: 1, value: 8}));\nconst minSim = view(Inputs.range([0, 0.5], {label: \"Min similarity to draw a link\", step: 0.01, value: 0.05}));","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":14,"value":"$51","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":15,"value":"$52","pinned":false,"hidden":false,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W53","stdlib":"2"}} 44:null