关于「三元排比与冗余:正则计数、连发长度与压缩比」,《机器文气与句长节律》给出的定量结论是什么?
结论:三元排比不是机器的发明,也不是毛病本身。p51 把它拆成两个量:每千字频次与最长连发;短文本的频次很容易被一处列举拉高,连发长度更接近读者对「机械」的感受。p52 把视野放宽到句框层面的重复:句子字面全不相同,词三元组重复率仍可以很高;压缩比与重复率都是冗余指标,不是作者指标,而且只对足够长的稿子有意义。
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。
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
| text | units triads per 1,000 max run | |||
| zh steady | 106 | 3 | 28.3 | 3 |
| zh breath | 72 | 1 | 13.9 | 1 |
| en steady | 38 | 3 | 78.9 | 3 |
| en breath | 41 | 1 | 24.4 | 1 |
| en listy | 24 | 3 | 125.0 | 3 |
查看完整验算程序
# Chapter 5. p51: three-part lists per 1,000 units, and their runs.
# Patterns (author's heuristics, not a validated parser):
# zh "A, B he/yu/ji C" with A, B, C of 1 to 8 hanzi, and three short
# comma clauses whose lengths differ by at most 2 hanzi;
# en "X, Y, and Z" or "X, Y and Z" (also "or"), items of 1 to 3 words.
# A "run" is a stretch of consecutive sentences that each hold a triad.
# Texts: the SYNTHETIC passages of p11, plus a synthetic list-heavy
# English paragraph written by the author to show a run.
import re
ZH = {
"steady": "首先,清晰的结构能够帮助读者理解文章的核心观点与主要内容。"
"其次,准确的用词可以提升表达的专业性、可信度和整体感染力。"
"此外,合理的段落安排有助于增强文章的逻辑性、连贯性与层次感。"
"总之,只有不断练习、认真反思和虚心学习,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。所以回信只写看见的东西:"
"哪一只箱子,压在哪个角,书有没有受损。她回了两个字。收到。",
}
EN = {
"steady": "Moreover, precise word choice can enhance the clarity, "
"credibility, and impact of writing. Additionally, good "
"paragraphs improve the logic, coherence, and flow of the "
"argument. In conclusion, only practice, reflection, and "
"learning make excellent texts. Writers must adapt.",
"breath": "The parcel came at noon. Three boxes: two intact, the third "
"creased across one corner. We did not open it at once. So "
"the reply said which box, which corner and which "
"attachment showed it. She answered in one word. Thanks.",
"listy": "Our service is fast, reliable, and affordable. We value "
"trust, quality, and care. Clients enjoy speed, savings, "
"and peace of mind. Call us today.",
}
H = r"[一-鿿]{1,8}"
ZH_LIST = re.compile(H + r"、" + H + r"[和与及]" + H)
ZH_COMMA = re.compile(r"([一-鿿]{2,9}),([一-鿿]{2,9})"
r",([一-鿿]{2,9})(?=[,。])")
ITEM = r"[A-Za-z']+(?: [A-Za-z']+){0,2}"
EN_LIST = re.compile(ITEM + r", " + ITEM + r",? (?:and|or) " + ITEM)
def zh_sents(t):
return [s for s in re.split(r"(?<=[。!?])", t) if s]
def en_sents(t):
return [s for s in re.split(r"(?<=[.!?])\s+", t.strip()) if s]
def zh_hits(s):
n = len(ZH_LIST.findall(s))
for a, b, c in ZH_COMMA.findall(s):
if max(len(a), len(b), len(c)) - min(len(a), len(b), len(c)) <= 2:
n += 1
return n
def report(lang, name, sents, hit, units):
flags = [hit(s) > 0 for s in sents]
total = 0
for s in sents:
total += hit(s)
run = best = 0
for f in flags:
run = run + 1 if f else 0
best = max(best, run)
print("{} {:<7} {:>5} {:>6} {:>9.1f} {:>8}".format(
lang, name, units, total, 1000 * total / units, best))
print("text units triads per 1,000 max run")
for name, t in ZH.items():
report("zh", name, zh_sents(t), zh_hits,
len(re.findall(r"[一-鿿]", t)))
for name, t in EN.items():
report("en", name, en_sents(t), lambda s: len(EN_LIST.findall(s)),
len(re.findall(r"[A-Za-z][A-Za-z'\-]*", t)))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 5 章「三元排比与冗余:正则计数、连发长度与压缩比」(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}
}