1#!/usr/bin/env python32"""WP44 — generate LaTeX tables (booktabs) from result files into manuscript/tables/*.tex. Every table is regenerated from3CSV/JSON artefacts; the inventory manuscript/tables/TABLES.json records source files and captions for Appendix G."""4import json, os, sys, csv, glob5import pandas as pd6R = sys.argv[1] if len(sys.argv) > 1 else os.path.dirname(os.path.dirname(os.path.abspath(__file__)))7OUT = os.path.join(R, "manuscript/tables"); os.makedirs(OUT, exist_ok=True); INV = []8910def tex_escape(s):11 s = str(s).replace("&", "\\&").replace("%", "\\%").replace("_", "\\_").replace("#", "\\#")12 for a, b in (("↔", "$\\leftrightarrow$"), ("→", "$\\rightarrow$"), ("…", "\\ldots{}"), ("∨", "$\\vee$"), ("∧", "$\\wedge$"), ("≥", "$\\geq$"), ("≤", "$\\leq$"), ("–", "--"), ("—", "---"), ("×", "$\\times$")):13 s = s.replace(a, b)14 return s151617def write(name, df, caption, label, sources, fmt=None, col_names=None, longtable=False, note=None):18 df = df.copy()19 if col_names: df.columns = col_names20 for c in df.columns:21 if pd.api.types.is_float_dtype(df[c]):22 df[c] = df[c].map(lambda v: "" if pd.isna(v) else (fmt or {}).get(c, "{:,.3f}").format(v).replace(",", "{,}"))23 elif pd.api.types.is_integer_dtype(df[c]):24 df[c] = df[c].map(lambda v: f"{v:,}".replace(",", "{,}"))25 else:26 df[c] = df[c].map(lambda v: tex_escape("" if pd.isna(v) else v))27 cols = "l" + "r" * (len(df.columns) - 1)28 head = " & ".join(f"\\textbf{{{tex_escape(c)}}}" for c in df.columns) + " \\\\"29 body = "\n".join(" & ".join(map(str, r)) + " \\\\" for r in df.itertuples(index=False))30 env = "longtable" if longtable else "tabular"31 s = f"% generated by tables/make_tables.py from {', '.join(sources)}\n"32 if longtable:33 s += f"\\begin{{longtable}}{{{cols}}}\n\\caption{{{caption}}}\\label{{{label}}}\\\\\n\\toprule\n{head}\n\\midrule\n\\endfirsthead\n\\toprule\n{head}\n\\midrule\n\\endhead\n{body}\n\\bottomrule\n"34 if note: s += f"\\multicolumn{{{len(df.columns)}}}{{p{{0.95\\linewidth}}}}{{\\footnotesize {note}}}\\\\\n"35 s += "\\end{longtable}\n"36 else:37 s += f"\\begin{{table}}[htbp]\\centering\\caption{{{caption}}}\\label{{{label}}}\n\\begin{{threeparttable}}\\small\n\\begin{{tabular}}{{{cols}}}\n\\toprule\n{head}\n\\midrule\n{body}\n\\bottomrule\n\\end{{tabular}}\n"38 if note: s += f"\\begin{{tablenotes}}\\footnotesize\\item {note}\\end{{tablenotes}}\n"39 s += "\\end{threeparttable}\\end{table}\n"40 open(os.path.join(OUT, name + ".tex"), "w").write(s); INV.append({"file": name + ".tex", "caption": caption, "label": label, "sources": sources, "rows": int(len(df))})414243def main():44 p = f"{R}/data/metadata/prisma_counts.json"45 if os.path.exists(p):46 P = json.load(open(p)); rows = [("Identified: OpenAlex works with a CS topic, 2018–2026 (snapshot 2026-10-14, both passes)", P["identified"]["openalex_total_records_loaded"])]47 labels = {"e1_outside_window": "E1 outside the window", "e2_type": "E2 non-research document type", "e3_xpac": "E3 XPAC expansion record (not DBLP-linked)", "e4_no_title": "E4 no title", "e5_duplicate": "E5 preprint merged into its published version", "e7_generic_deposit": "E7 generic deposit platform without DBLP/arXiv support"}48 rows += [("Excluded " + labels.get(k, k), v) for k, v in P["excluded_sequential"].items()]49 rows += [("After exclusions", P["after_exclusions"]), ("CS core (mapping corpus)", P["cs_core_eligible"]), ("CS penumbra (sensitivity set)", P["cs_penumbra_eligible_otherwise"]), ("Strict core: DBLP-indexed and primary field CS", P["strict_core_dblp_and_cs_primary"])]50 write("tab_prisma", pd.DataFrame(rows, columns=["Stage", "Records"]), "Record flow of the mapping corpus (PRISMA-style, sequential exclusions).", "tab:prisma", ["data/metadata/prisma_counts.json"])51 write("tab_eligible_by_year", pd.DataFrame(sorted(P["eligible_by_year"].items()), columns=["Year", "Eligible works"]), "Eligible works by publication year (2026 incomplete).", "tab:eligible_year", ["data/metadata/prisma_counts.json"])52 write("tab_eligible_by_type", pd.DataFrame(P["eligible_by_type"].items(), columns=["OpenAlex type", "Eligible works"]), "Eligible works by document type.", "tab:eligible_type", ["data/metadata/prisma_counts.json"])53 p = f"{R}/data/metadata/linkage_report.json"54 if os.path.exists(p):55 L = json.load(open(p)); rows = [("DBLP records 2018–2026", L["dblp_2018_2026"]), ("… with DOI", L["dblp_with_doi"]), ("DBLP ↔ OpenAlex links by DOI", L["link_dblp_oa"].get("doi", 0)), ("… by normalised title + year (+ first author)", L["link_dblp_oa"].get("title_year", 0)), ("DBLP records linked (distinct)", L["dblp_linked_distinct"]), ("DBLP records without OpenAlex record", L["dblp_unlinked"]),56 ("arXiv cs records created 2018–2026", L["arxiv_created_2018_2026"]), ("arXiv ↔ OpenAlex by arXiv id", L["link_arxiv_oa"].get("arxiv_id", 0)), ("… by DOI", L["link_arxiv_oa"].get("doi", 0)), ("… by title + year", L["link_arxiv_oa"].get("title_year", 0)), ("ACL entries 2018–2026", L["acl_2018_2026"]), ("ACL ↔ OpenAlex linked (2018–2026)", L["acl_linked_distinct_2018_2026"])]57 write("tab_linkage", pd.DataFrame(rows, columns=["Quantity", "Count"]), "Record linkage between sources.", "tab:linkage", ["data/metadata/linkage_report.json"])58 for f, name, cap, cols in [("volume_yearly", "tab_volume_yearly", "Yearly volume and composition of the eligible corpus.", ["year", "n", "n_dblp", "n_cs_primary", "n_arxiv", "n_abstract", "n_oa", "n_preprint", "n_conf", "n_article", "mean_authors", "mean_refs"]),59 ("citation_windows_yearly", "tab_citation_windows", "Fixed-window citation statistics by publication year (c12: publication year + 1; c36: + 2 calendar years).", ["year", "n", "mean_c12", "mean_c36", "med_c36", "share_uncited_36", "p99_c36", "gini_c36", "gini_c12", "mean_fwci"]),60 ("team_size_yearly", "tab_team_size", "Team size and collaboration by year.", ["year", "n", "mean_authors", "med_authors", "single", "five_plus", "ten_plus", "intl", "multi_inst"]),61 ("industry_share_yearly", "tab_industry", "Works with a company affiliation by year.", ["year", "n_with_inst", "n_company", "share_company"]),62 ("venue_hhi_yearly", "tab_venue_hhi", "Venue concentration by year.", ["year", "n", "n_sources", "hhi", "top1_share"]),63 ("diversity_yearly", "tab_diversity", "Rao–Stirling and Leinster–Cobbold diversity of the category mix by year (bootstrap 95\\% CI).", ["year", "n", "rao_stirling", "rs_ci_low", "rs_ci_high", "leinster_cobbold_q2"]),64 ("cagr_by_category", "tab_cagr", "Compound annual growth by category, baseline (2018–2022) and post (2023–2025).", ["category", "n_2018", "n_2022", "n_2025", "cagr_baseline", "cagr_post", "share_2018", "share_2025"])]:65 p = f"{R}/results/bibliometrics/{f}.csv"66 if os.path.exists(p):67 d = pd.read_csv(p)68 if "period" in d.columns: d = d[d.period.isin(["baseline", "post"])]69 d = d[[c for c in cols if c in d.columns]]70 if "team" in f or "volume" in f: d = d[d.iloc[:, 0] <= 2026]71 write(name, d, cap, f"tab:{name[4:]}", [f"results/bibliometrics/{f}.csv"], longtable=len(d) > 30)72 p = f"{R}/results/bibliometrics/countries_yearly.csv"73 if os.path.exists(p):74 d = pd.read_csv(p); top = d.groupby("country").n_frac.sum().nlargest(15).index; piv = d[d.country.isin(top)].pivot_table(index="country", columns="year", values="n_frac").loc[top]75 piv = piv.round(0).astype(int).reset_index(); write("tab_countries", piv, "Fractional output of the fifteen largest countries by year.", "tab:countries", [p])76 p = f"{R}/results/econometrics/breaks.csv"77 if os.path.exists(p):78 d = pd.read_csv(p); cols = ["series", "chatgpt_level", "chatgpt_level_p", "chatgpt_slope", "chatgpt_slope_p", "placebo_level_p", "placebo_slope_p", "bp_breaks_bic", "pelt"]79 write("tab_breaks", d[[c for c in cols if c in d.columns]], "Structural-change estimates by series: ITS level and slope changes at 2022-12 (HAC p-values), placebo p-values, Bai–Perron (BIC) and PELT break dates.", "tab:breaks", [p], longtable=len(d) > 25)80 p = f"{R}/taxonomy/validation/metrics.json"81 if os.path.exists(p):82 M = json.load(open(p)); rows = []83 for k, v in M.items():84 if isinstance(v, dict) and "per_category" in v:85 for c, m in v["per_category"].items():86 rows.append({"comparison": k, "category": c, "precision": m["precision"], "recall": m["recall"], "f1": m["f1"], "tp": m["tp"], "fp": m["fp"], "fn": m["fn"]})87 if rows: write("tab_validation", pd.DataFrame(rows), "Taxonomy validation: precision, recall and F1 of the rule-based assignment against each rater and the rater consensus, by category.", "tab:validation", [p], longtable=True)88 for f in glob.glob(f"{R}/results/emerging/metrics_h*.json"):89 M = json.load(open(f)); rows = [{"model": m, "AUC": M[m]["auc"], "P@20": M[m]["precision_at_20"], "P@50": M[m]["precision_at_50"]} for m in ("zscore_baseline", "logistic", "gbm") if m in M]90 write("tab_emerging_h%d" % M["horizon"], pd.DataFrame(rows), f"Out-of-time performance of the emerging-topic detectors at horizon {M['horizon']} months (test cutoffs {M['test_cutoffs'][0]}–{M['test_cutoffs'][-1]}).", f"tab:emerging_h{M['horizon']}", [f])91 p = f"{R}/results/emerging/known_set_lead_times.csv"92 if os.path.exists(p):93 d = pd.read_csv(p)[["concept", "total_works", "detection_month", "peak_growth_month", "lead_months"]]; write("tab_known_set", d, "Preregistered known-emergence set: detection month, peak-growth month and lead time.", "tab:known_set", [p])94 tables_part08() # Part VIII tables (appended 2026-11-03)95 json.dump(INV, open(os.path.join(OUT, "TABLES.json"), "w"), indent=1); print(len(INV), "tables")969798# ---------------- Part VIII (Scientific Knowledge Networks): tables appended by Agent L/I, 2026-11-03 ----------------99def _ascii(s, n=None):100 """Titles and venue names contain characters (U+2010, curly quotes, ...) that inputenc does not define; normalise to ASCII."""101 import unicodedata102 t = str(s).replace("‐", "-").replace("‑", "-").replace("–", "-").replace("—", "-").replace("’", "'").replace("‘", "'").replace("“", '"').replace("”", '"').replace("×", "x")103 t = unicodedata.normalize("NFKD", t).encode("ascii", "ignore").decode("ascii").replace("$", "\\$").replace("^", "").replace("~", "-").replace("{", "").replace("}", "")104 return (t[: n - 3].rstrip() + "\\dots") if (n and len(t) > n) else t105106107def tables_part08():108 B = f"{R}/results/bibliometrics"; G = f"{R}/results/graph"109 p = f"{B}/team_size_yearly.csv"; q = f"{B}/international_collab_yearly.csv"110 if os.path.exists(p):111 d = pd.read_csv(p); d = d[d.year <= 2026]112 d["sum_authors"] = d.mean_authors * d.n113 g = d.groupby("year").agg(n=("n", "sum"), sum_authors=("sum_authors", "sum"), single=("single", "sum"), five_plus=("five_plus", "sum"), ten_plus=("ten_plus", "sum"), multi_inst=("multi_inst", "sum"), no_country=("no_country", "sum")).reset_index()114 t = pd.DataFrame({"Year": g.year, "Works": g.n, "Mean authors": g.sum_authors / g.n, "Single (%)": 100 * g.single / g.n, "5+ (%)": 100 * g.five_plus / g.n, "10+ (%)": 100 * g.ten_plus / g.n, "Multi-inst. (%)": 100 * g.multi_inst / g.n, "No country (%)": 100 * g.no_country / g.n})115 if os.path.exists(q):116 i = pd.read_csv(q); t = t.merge(pd.DataFrame({"Year": i.year, "Intl. (%)": 100 * i.share_intl}), on="Year", how="left")117 t["Year"] = t.Year.astype(str)118 write("tab_collaboration", t, "Team size and collaboration indicators by publication year (all eligible works; periods of the source file summed).", "tab:collaboration", [p, q], fmt={"Mean authors": "{:.3f}", "Single (%)": "{:.1f}", "5+ (%)": "{:.1f}", "10+ (%)": "{:.1f}", "Multi-inst. (%)": "{:.1f}", "No country (%)": "{:.1f}", "Intl. (%)": "{:.1f}"},119 note="Columns: works with one author, five or more, ten or more authors, and more than one institution. Shares are over all eligible works of the year, including works without affiliation data (column `No country'); the international share is over works with at least one resolved country (ICS, Part IV). 2026 is incomplete.")120 p = f"{B}/institution_types_yearly.csv"121 if os.path.exists(p):122 d = pd.read_csv(p); d = d[d.year <= 2026]; tot = d.groupby("year").n.transform("sum"); d["share"] = 100 * d.n / tot123 piv = d.pivot_table(index="itype", columns="year", values="share").fillna(0); piv = piv.loc[piv.sum(axis=1).sort_values(ascending=False).index].reset_index(); piv.columns = ["Institution type"] + [str(c) for c in piv.columns[1:]]124 write("tab_institution_types", piv, "Share of typed affiliations by OpenAlex institution type and year (per cent of affiliation-work pairs).", "tab:institution_types", [p], fmt={c: "{:.2f}" for c in piv.columns[1:]}, note="A work with affiliations of several types counts once under each type; shares therefore describe the mix of affiliation types, not of works. Types follow the OpenAlex institution ontology (education, facility, company, government, healthcare, nonprofit, funder, archive, other).")125 p = f"{B}/countries_yearly.csv"126 if os.path.exists(p):127 d = pd.read_csv(p); d = d[d.year <= 2026]; tot = d.groupby("year")[["n_full", "n_frac"]].sum()128 top = d.groupby("country").n_frac.sum().nlargest(15).index129 sh = d[d.country.isin(top)].copy(); sh["share"] = 100 * sh.n_frac / sh.year.map(tot.n_frac); piv = sh.pivot_table(index="country", columns="year", values="share").loc[top].reset_index(); piv.columns = ["Country"] + [str(c) for c in piv.columns[1:]]130 write("tab_country_shares", piv, "Share of fractional output of the fifteen largest countries by year (per cent).", "tab:country_shares", [p], fmt={c: "{:.2f}" for c in piv.columns[1:]}, note="Fractional counting: each work is divided equally among its distinct author countries, so the shares of all countries sum to 100 within a year. ISO 3166-1 alpha-2 codes. 2026 is incomplete.")131 rows = []132 for y, g in d.groupby("year"):133 pf = g.n_frac / g.n_frac.sum(); cn = g[g.country == "CN"]; us = g[g.country == "US"]134 rows.append({"Year": str(y), "Works": int(round(g.n_frac.sum())), "Countries": int(len(g)), "HHI": float((pf ** 2).sum()), "Top-5 (%)": 100 * pf.nlargest(5).sum(), "Top-10 (%)": 100 * pf.nlargest(10).sum(), "CN works (%)": 100 * float(cn.n_full.sum()) / g.n_frac.sum(), "US works (%)": 100 * float(us.n_full.sum()) / g.n_frac.sum(), "CN full/frac": float(cn.n_full.sum()) / float(cn.n_frac.sum()), "US full/frac": float(us.n_full.sum()) / float(us.n_frac.sum())})135 write("tab_country_concentration", pd.DataFrame(rows), "Concentration of output across countries by year, and the two largest producers under full counting.", "tab:country_concentration", [p], fmt={"HHI": "{:.4f}", "Top-5 (%)": "{:.1f}", "Top-10 (%)": "{:.1f}", "CN works (%)": "{:.2f}", "US works (%)": "{:.2f}", "CN full/frac": "{:.3f}", "US full/frac": "{:.3f}"},136 note="Works: fractional total of works with at least one country; Countries: number of countries with any output; HHI: Herfindahl--Hirschman index of fractional country shares; Top-5/Top-10: fractional share of the five and ten largest countries. CN/US works: share of works with at least one author in the country among works with a country (full counting); CN/US full/frac: ratio of the country's full count to its fractional count, which exceeds one by the extent to which its works are co-authored with other countries.")137 p = f"{B}/country_pairs_by_period.csv"138 if os.path.exists(p):139 d = pd.read_csv(p); d = d[d.period.isin(["baseline", "post"])]140 rows = []141 for per in ["baseline", "post"]:142 g = d[d.period == per]; w = g.n.sum(); cu = g[g.c1.isin(["CN", "US"]) & g.c2.isin(["CN", "US"])].n.sum(); st = pd.concat([g[["c1", "n"]].rename(columns={"c1": "c"}), g[["c2", "n"]].rename(columns={"c2": "c"})]).groupby("c").n.sum().sort_values(ascending=False)143 rows.append({"Period": per, "Edges": int(len(g)), "Countries": int(len(set(g.c1) | set(g.c2))), "Weight": int(w), "CN--US": int(cu), "CN--US (%)": 100 * cu / w, "Top strength": st.index[0] + ", " + st.index[1] + ", " + st.index[2]})144 write("tab_country_network", pd.DataFrame(rows), "The country co-authorship network by period: edges with at least fifty joint works.", "tab:country_network", [p], fmt={"CN--US (%)": "{:.2f}"}, note="Edges: country pairs with at least fifty joint works; Weight: total joint works over these edges; CN--US: weight of the China--United States edge and its share of the total; Top strength: the three countries with the largest strength (sum of edge weights). Baseline: 2018-01-01 to 2022-11-29; post: 2022-11-30 to 2026-11-02 (incomplete).")145 b = d[d.period == "baseline"].sort_values("n", ascending=False).head(20).reset_index(drop=True); q = d[d.period == "post"].sort_values("n", ascending=False).head(20).reset_index(drop=True)146 t = pd.DataFrame({"Rank": [str(i + 1) for i in range(20)], "Pair (baseline)": b.c1 + "--" + b.c2, "Joint works (baseline)": b.n.astype(int), "Pair (post)": q.c1 + "--" + q.c2, "Joint works (post)": q.n.astype(int)})147 write("tab_country_pairs", t, "The twenty strongest country pairs by period (number of joint works).", "tab:country_pairs", [p], note="Pairs are unordered; HK and MO are OpenAlex country codes for Hong Kong and Macau and are distinct from CN in the affiliation data. The post period is incomplete.")148 p = f"{B}/top_institutions_by_period.csv"149 if os.path.exists(p):150 d = pd.read_csv(p); b = d[d.period == "baseline"].set_index("inst_id"); q = d[d.period == "post"].set_index("inst_id"); j = b.join(q, lsuffix="_b", rsuffix="_p", how="inner")151 rows = []152 for per, g in [("baseline", b), ("post", q)]:153 pf = g.n_frac / g.n_frac.sum()154 rows.append({"Period": per, "Institutions": int(len(g)), "Fractional works": int(round(g.n_frac.sum())), "Full-count works": int(g.n_full.sum()), "Top-10 (%)": 100 * pf.nlargest(10).sum(), "Top-50 (%)": 100 * pf.nlargest(50).sum(), "Top-100 (%)": 100 * pf.nlargest(100).sum()})155 rows.append({"Period": "both periods", "Institutions": int(len(j)), "Fractional works": int(round(j.n_frac_b.sum())), "Full-count works": int(j.n_full_b.sum()), "Top-10 (%)": float("nan"), "Top-50 (%)": float("nan"), "Top-100 (%)": float("nan")})156 t = pd.DataFrame(rows); write("tab_institution_concentration", t, "Institutions with at least two hundred works per period: coverage and concentration (fractional counting).", "tab:institution_concentration", [p], fmt={"Top-10 (%)": "{:.2f}", "Top-50 (%)": "{:.2f}", "Top-100 (%)": "{:.2f}"},157 note=f"Institutions: number with at least two hundred works in the period; Top-10/50/100: fractional share of the largest listed institutions relative to the listed institutions, not to all works. The row `both periods' counts the institutions present in both lists (its work columns are their baseline totals); the Spearman rank correlation of their fractional counts across periods is {j.n_frac_b.rank().corr(j.n_frac_p.rank(), method='spearman'):.3f}.")158 top = j.sort_values("n_frac_p", ascending=False).head(25)159 t = pd.DataFrame({"OpenAlex id": ["I" + str(i) for i in top.index], "Full, base": top.n_full_b.astype(int).values, "Frac., base": top.n_frac_b.values, "Full, post": top.n_full_p.astype(int).values, "Frac., post": top.n_frac_p.values, "Ratio": (top.n_frac_p / top.n_frac_b).values})160 write("tab_top_institutions", t, "The twenty-five institutions with the largest fractional output in the post period, with their baseline output.", "tab:top_institutions", [p], fmt={"Frac., base": "{:,.0f}", "Frac., post": "{:,.0f}", "Ratio": "{:.2f}"}, longtable=False,161 note="Full and fractional counts in the baseline and post periods; Ratio is the post-to-baseline ratio of fractional counts. Institution names are not stored in the result file; the identifiers resolve at https://openalex.org/I<id>. The post period is about ten months shorter than the baseline, so a ratio below the corpus-wide ratio does not indicate decline.")162 p = f"{B}/top_sources_by_period.csv"163 if os.path.exists(p):164 d = pd.read_csv(p); d = d[d.period.isin(["baseline", "post"])]165 rows = []166 for per in ["baseline", "post"]:167 g = d[d.period == per]; bt = g.groupby("source_type").n.sum(); w = bt.sum()168 r = {"Period": per, "Sources": int(len(g)), "Works": int(w)}169 for ty in ["journal", "conference", "repository", "book series", "ebook platform"]:170 r[ty.replace("ebook platform", "ebook").replace("book series", "series") + " (%)"] = 100 * float(bt.get(ty, 0)) / w171 rows.append(r)172 write("tab_source_types", pd.DataFrame(rows), "Distribution of works over source types among the listed sources, by period (per cent).", "tab:source_types", [p], fmt={c: "{:.1f}" for c in ["journal (%)", "conference (%)", "repository (%)", "series (%)", "ebook (%)"]}, note="Source types are OpenAlex's (series = book series; ebook = ebook platform). The listed sources are those retained by the bibliometric script for each period; works whose primary source is not listed are excluded from the denominators.")173 b = d[d.period == "baseline"].sort_values("n", ascending=False).head(20).reset_index(drop=True); q = d[d.period == "post"].sort_values("n", ascending=False).head(20).reset_index(drop=True)174 t = pd.DataFrame({"Rank": [str(i + 1) for i in range(20)], "Source (baseline)": [_ascii(x, 30) for x in b.source_name], "Type": [_ascii(x).replace("ebook platform", "ebook").replace("book series", "series").replace("repository", "repos.").replace("conference", "conf.") for x in b.source_type], "Works": b.n.astype(int), "Source (post)": [_ascii(x, 30) for x in q.source_name], "Type ": [_ascii(x).replace("ebook platform", "ebook").replace("book series", "series").replace("repository", "repos.").replace("conference", "conf.") for x in q.source_type], "Works ": q.n.astype(int)})175 write("tab_top_sources", t, "The twenty largest sources (venues) by period, with OpenAlex source type.", "tab:top_sources", [p], longtable=True, note="Source names are OpenAlex display names, truncated; types abbreviated (repos. = repository, conf. = conference, series = book series, ebook = ebook platform).")176 p = f"{G}/cited_age_yearly.csv"177 if os.path.exists(p):178 d = pd.read_csv(p); rows = []179 for y, g in d.groupby("citing_year"):180 if y < 2018 or y > 2026: continue181 n = g.n.sum(); neg = g[g.age < 0].n.sum(); gg = g[g.age >= 0].sort_values("age"); w = gg.n.values; ag = gg.age.values182 import numpy as np183 cum = np.cumsum(w) / w.sum(); med = int(ag[np.searchsorted(cum, 0.5)])184 rows.append({"Citing year": str(y), "References": int(n), "Negative (%)": 100 * neg / n, "Mean age": float((w * ag).sum() / w.sum()), "Median": med, "0--2 y (%)": 100 * gg[gg.age <= 2].n.sum() / w.sum(), "0--5 y (%)": 100 * gg[gg.age <= 5].n.sum() / w.sum(), "over 10 y (%)": 100 * gg[gg.age > 10].n.sum() / w.sum()})185 write("tab_cited_age", pd.DataFrame(rows), "Age of cited works by citing year (resolved references of eligible works; corpus version of the result file at generation time).", "tab:cited_age", [p], fmt={"Negative (%)": "{:.2f}", "Mean age": "{:.2f}", "0--2 y (%)": "{:.1f}", "0--5 y (%)": "{:.1f}", "over 10 y (%)": "{:.1f}"},186 note="References: resolved references of eligible works of the citing year. Age is the citing year minus the cited work's publication year. Negative ages (cited work dated after the citing work: preprints cited before their published version, or year-only dates) are excluded from the age statistics and reported as a share. The 2026 row is incomplete.")187 p = f"{B}/top_cited_by_year.csv"188 if os.path.exists(p):189 d = pd.read_csv(p); d = d[(d.year >= 2018) & (d.year <= 2025) & (d.rk <= 10)].sort_values(["year", "rk"])190 t = pd.DataFrame({"Year": d.year.astype(str).values, "Rk": d.rk.astype(str).values, "Title (truncated)": [_ascii(x, 44) for x in d.title], "c36": d.c36.astype(int).values, "Total": d.cited_by_count.astype(int).values})191 write("tab_top_cited", t, "The ten most cited works of each publication year by 36-month window citations (c36), with total citations to date.", "tab:top_cited", [p], longtable=True, note="As recorded in OpenAlex (snapshot 2026-10-14 plus increment). Several entries are duplicate or misdated records of canonical works, or records whose citation counts are inconsistent with their total (c36 larger than the total, or a total of zero); they are kept because the table documents what the database returns, and are discussed in the text.")192 p = f"{B}/dblp_only_yearly.csv"193 if os.path.exists(p):194 d = pd.read_csv(p); piv = d.pivot_table(index="year", columns="rec_type", values="n", aggfunc="sum").fillna(0).astype(int); piv["total"] = piv.sum(axis=1); piv["with DOI (%)"] = 100 * d.groupby("year").with_doi.sum() / d.groupby("year").n.sum()195 piv = piv.reset_index(); piv["year"] = piv.year.astype(str); piv.columns = ["Year"] + [str(c) for c in piv.columns[1:]]196 write("tab_dblp_only", piv, "DBLP records of 2018--2026 without an OpenAlex counterpart, by DBLP record type and year.", "tab:dblp_only", [p], fmt={"with DOI (%)": "{:.1f}"}, note="These records are in DBLP but were not linked to any OpenAlex work by DOI, arXiv identifier or normalised title (Part IV). They are not in the eligible corpus; the table bounds what DBLP-based studies would see that this corpus does not.")197 p = f"{B}/indexing_lag.csv"198 if os.path.exists(p):199 d = pd.read_csv(p); d["year"] = d.pub_month.str[:4].astype(int); rows = []200 for y, g in d.groupby("year"):201 jan = g[g.pub_month.str.endswith("-01")]; jun = g[g.pub_month.str.endswith("-06")]202 rows.append({"Year": str(y), "Works": int(g.n.sum()), "Lag, Jan. (d)": float(jan.lag_days_p50.iloc[0]) if len(jan) else float("nan"), "Lag, June (d)": float(jun.lag_days_p50.iloc[0]) if len(jun) else float("nan"), "Within 90 d (%)": 100 * g.n_indexed_within_90d.sum() / g.n.sum(), "After 180 d (%)": 100 * g.n_indexed_after_180d.sum() / g.n.sum()})203 write("tab_indexing_lag", pd.DataFrame(rows), "The OpenAlex \\texttt{created\\_date} relative to the publication date, by publication year.", "tab:indexing_lag", [p], fmt={"Lag, Jan. (d)": "{:.0f}", "Lag, June (d)": "{:.0f}", "Within 90 d (%)": "{:.1f}", "After 180 d (%)": "{:.1f}"},204 note="Lag is created\\_date minus publication date; the two lag columns give the median lag in days for works published in January and in June of the year; the last two columns give the shares of the year's works whose record was created within 90 days and after 180 days of publication. The near-constant decline of the median by one month per publication month in 2018--2024 identifies a single creation date for most records (a re-ingestion), not an indexing delay; see Part III.")205 # ---- corpus v3 graph tables (appended by Agent L/M, 2026-11-03): reference coverage and field-level inflows ----206 p = f"{G}/reference_coverage_yearly.csv"207 if os.path.exists(p):208 d = pd.read_csv(p); d = d[(d.year >= 2018) & (d.year <= 2026)].copy(); res = d.n_internal + d.n_external209 t = pd.DataFrame({"Year": d.year.astype(str), "Works": d.n_works.astype(int), "Declared refs.": d.n_refs_declared.astype(int), "Refs./work": d.n_refs_declared / d.n_works, "Resolved (%)": 100 * res / d.n_refs_declared, "Internal": d.n_internal.astype(int), "External": d.n_external.astype(int), "Internal (%)": 100 * d.n_internal / res})210 write("tab_reference_coverage", t, "Reference coverage by citing year: declared, resolved, internal and external references of eligible works.", "tab:reference_coverage", [p], fmt={"Refs./work": "{:.2f}", "Resolved (%)": "{:.2f}", "Internal (%)": "{:.2f}"},211 note="Works: eligible works of the year (including works with no reference list); Declared: entries of their OpenAlex \\texttt{referenced\\_works} lists; Resolved: declared references that resolve to an OpenAlex record (internal = cited work in the eligible corpus; external = cited work outside it), as a share of the declared count; Internal (\\%): internal share of resolved references. The resolved share of 2025--2026 is higher than in earlier years because the recent cohorts carry fewer parsed references per work (reference extraction lags publication) and the parsed ones resolve more often; it describes the database, not citing behaviour. The 2026 row is incomplete.")212 p = f"{G}/field_flows_yearly.csv"213 if os.path.exists(p):214 d = pd.read_csv(p); d = d[(d.citing_year >= 2018) & (d.citing_year <= 2026)]; tot = d.groupby("citing_year").n.sum()215 names = {17: "CS", 22: "Eng.", 33: "Soc.", 27: "Med.", 26: "Math.", 18: "Dec.", 32: "Psy.", 31: "Phys."}216 ext = d[d.kind == "external"]; internal = d[d.kind == "internal"].groupby("citing_year").n.sum().reindex(tot.index).fillna(0)217 byf = ext.pivot_table(index="citing_year", columns="cited_field_id", values="n", aggfunc="sum").reindex(tot.index).fillna(0)218 cs = byf[17.0] if 17.0 in byf.columns else 0 * tot219 t = pd.DataFrame({"Year": [str(y) for y in tot.index], "Inside": 100 * internal.values / tot.values, "CS out": 100 * cs.values / tot.values, "Other": 100 * (tot.values - internal.values - cs.values) / tot.values})220 for fid in [22, 33, 27, 26, 18, 32, 31]:221 t[names[fid]] = 100 * (byf[float(fid)].values if float(fid) in byf.columns else 0) / tot.values222 write("tab_field_flows", t, "Destination of the resolved references of eligible works by citing year: inside the corpus, computer science outside it, and the seven largest other OpenAlex fields (per cent of resolved references).", "tab:field_flows", [p], fmt={c: "{:.2f}" for c in t.columns[1:]},223 note="Shares of all resolved references of the citing year. Inside: cited work is an eligible work (computer science, 2018--2026); CS out: cited work outside the corpus whose primary OpenAlex field is Computer Science (field 17; mostly pre-2018 work); Other: all remaining external references, including those without a field. The seven field columns are the largest components of `Other': Eng. = Engineering (OpenAlex field 22), Soc. = Social Sciences (33), Med. = Medicine (27), Math. = Mathematics (26), Dec. = Decision Sciences (18), Psy. = Psychology (32), Phys. = Physics and Astronomy (31). The rise of the inside-corpus share is mechanical: the corpus starts in 2018, so later citing years have more corpus to cite. The 2026 row is incomplete.")224225226if __name__ == "__main__":227 main()228