The Geometry of Citation Weight in Generative Search: Retrieval and Reranking Features and the Paragraph Topology of Highly Cited Bilingual Answer Pages: chapter answers
A common belief holds that repeating keywords will make answer engines cite a page more often. This book tests that belief with the mathematics of retrieval and ranking. After checking the GEO paper by Aggarwal and colleagues, published at KDD in 2024, it corrects its founding blueprint: on the paper's benchmark keyword stuffing changed position-adjusted word count by -0.0829 relative to the baseline, quotation addition raised it by 0.4093, but a fluency rewrite that adds no information gained almost as much; the later C-SEO Bench found most such rewrites largely ineffective on new tasks, wit…
DOI pending · monograph page goes live after publication · 5DL-B-260 · 中文版
- Chapter 1 · Measuring Visibility: Why Keyword Stuffing Fails and Why Sources, Figures and Quotations Help
Chapter 1 rebuilds the paper's table of effects and uses a proportional-share model to show that a rewrite adopted alone buys only a 0.3027 gain in share and nothing once all adopt. - Chapter 2 · The Geometry of Hybrid Scoring: BM25 Saturation, Cosine Direction and Rank Fusion
Chapter 2 covers BM25 saturation, cosine direction and rank fusion: one new true matching term earns nearly three times what one repeated term does. - Chapter 3 · Paragraph Length and the Point of the Cut: Chunking, Truncation and an Interior Optimum
Chapter 3 covers chunking and truncation: at a cut of 128 words a conclusion placed first is read 4.5000 times as often as one placed last, and paragraph length has an interior optimum of about 60 to 130 words. - Chapter 4 · The Discount by Citation Position: Cumulated Gain, Position-Adjusted Word Count and the Sag in the Middle
With $\tau=1$ the five prior shares are 0.6364, 0.2341, 0.0861, 0.0317 and 0.0117. A rewrite of the first source gains only +0.1181, because it already holds the bulk; a rewrite of the fifth gains +0.4026, close to the… - Chapter 5 · Verifiability: Why Sources Raise the Chance of Citation, and the Trap of Inferring Features from Outside
Chapter 5 treats verifiability and shows by simulation how fragile it is to infer feature weights from outside: leaving out length overstates the weight on statistics by 0.3252. - Chapter 6 · The Paragraph Topology of a Bilingual Direct-Answer Page: Self-Contained Paragraphs, One Set of Numbers on Four Surfaces, and Compliant Structured Data
The narrative page has five paragraphs; only 0.2000 of them stand alone, none names a source, and a single lifted paragraph has an expected value of 0.4400. The inverted-pyramid page has four paragraphs; half stand alon… - Chapter 7 · Classical Sources East and West: the Three Gauges of the Mozi and Garfield's Citation Index
Chapter 7 examines the three gauges of the Mozi and Garfield's citation index of 1955: cumulative advantage lowers the rank correlation between quality and citations from 0.8688 to 0.6773. - Chapter 8 · Closing the Loop: an A/B Test of Answer-Paragraph Templates, the Citability Scorer and a Pre-Registered Test
Chapter 8 gives an A/B test of answer-paragraph templates, the Answer-Paragraph Citability Scorer and a pre-registered protocol (524 checks per group, simulated power 0.8085). Marked boxes separate published evidence, r…
Illustrative model calculation, not a live quote