关于「套话密度与超额词汇:泊松尾概率与群体下界」,《机器文气与句长节律》给出的定量结论是什么?

结论:套话的问题在密度,不在某个词。p41 给出每千字频次的算法,p42 说明短稿子的密度即使是基线三倍,也常常达不到统计显著,判红门槛必须随长度变化。p43 重演了科巴克等人的超额词汇思路:比值适合稀有词,差值适合常用词,差值给出的是群体下界,拆不回任何一篇稿子。

数据口径:书中示例均为作者自写的合成文字或公有领域文本,参数为示意,全部计算由书中程序生成。 第一至六章依次讲句长指纹与对数正态、突发度与顺序量、检测器的贝叶斯账与偏差、套话密度的泊松检验与群体超额词汇、三元连发与压缩冗余、只动句读的动态规划与真实细节账本:合成草稿只改 5 处句读,变异系数就从 0.123 升到 0.546,而字词一个不动。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 43

textunitshitsper 1,000
zh steady157957.3
zh breath8100.0
en steady439209.3
en breath4600.0
查看完整验算程序
# Chapter 4. p41: stock-transition density per 1,000 units.
# The two phrase lists are ILLUSTRATIVE, written by the author for this
# book; they are not a validated lexicon and do not prove machine
# authorship. Texts: the four SYNTHETIC passages of p11 (shortened
# copies). Density = hits per 1,000 hanzi (zh) or per 1,000 words (en).
import re

ZH_LIST = ["首先", "其次", "此外", "值得注意的是", "总之", "综上所述",
           "与此同时", "因此", "不可或缺", "在当今"]
EN_LIST = ["firstly", "moreover", "additionally", "furthermore",
           "it is worth noting", "in conclusion", "therefore", "crucial",
           "delve", "in today's"]
ZH = {
    "steady": "在当今快速发展的时代,写作能力已经成为每个人不可或缺的重要"
              "素养。首先,清晰的结构能够帮助读者快速理解文章的核心观点。"
              "其次,准确的用词可以有效提升表达的专业性。此外,合理的段落"
              "安排有助于增强文章的逻辑性。值得注意的是,不同类型的文本有"
              "不同要求。因此,写作者需要灵活调整表达方式。与此同时,持续"
              "阅读也是重要途径。总之,只有不断练习,才能写出优秀的文章。",
    "breath": "货到了。三只纸箱,两只完好,第三只的一角压出一道斜折痕。先"
              "不拆。拍照,记下运单尾号,再拆。里面的书脊没有伤,只有塑封"
              "起了皱。可客人问的是那道折痕。所以回信只写看见的东西。她回"
              "了两个字。收到。",
}
EN = {
    "steady": "In today's fast-paced world, clear writing is essential. "
              "Firstly, structure helps readers. Moreover, precise word "
              "choice enhances clarity. Additionally, paragraphs improve "
              "logic. It is worth noting that genres differ. Therefore, "
              "writers must adapt. Furthermore, reading is crucial. In "
              "conclusion, practice makes excellent texts.",
    "breath": "The parcel came at noon. Three boxes, one creased at a "
              "corner. We did not open it at once. Photo first, then the "
              "waybill number, then the knife. The spines were fine. But "
              "the customer had asked about the crease. She answered in "
              "one word. Thanks.",
}


def zh_density(text):
    units = len(re.findall(r"[一-鿿]", text))
    hits = 0
    for p in ZH_LIST:
        hits += text.count(p)
    return units, hits, 1000 * hits / units


def en_density(text):
    low = text.lower()
    units = len(re.findall(r"[a-z][a-z'\-]*", low))
    hits = 0
    for p in EN_LIST:
        hits += len(re.findall(r"\b" + re.escape(p) + r"\b", low))
    return units, hits, 1000 * hits / units


print("text        units  hits  per 1,000")
for name in ("steady", "breath"):
    u, h, d = zh_density(ZH[name])
    print("zh {:<7} {:>6} {:>5} {:>10.1f}".format(name, u, h, d))
for name in ("steady", "breath"):
    u, h, d = en_density(EN[name])
    print("en {:<7} {:>6} {:>5} {:>10.1f}".format(name, u, h, d))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《机器生成文本统计指纹剥离、句长突发律动重构与宗师级文气呼吸感母典》第 4 章「套话密度与超额词汇:泊松尾概率与群体下界」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-394 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026394,
  author    = {Fan, Yafu},
  title     = {Stripping Statistical Fingerprints of Machine Prose, Sentence-Length Burstiness Reconstruction and Master-Level Stylistic Cadence},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论