Feature Fusion for Computational Intertextuality Detection in Tesserae V6

Computational detection of intertextuality has so far used individual features such as word n-grams, edit distance, or word embeddings to identify text reuse or allusion, with only limited efforts to combine two such features for improved detection. This article presents Tesserae V6, a freely available platform that searches, weights, and combines into a single fused score ten distinct signals: shared headwords; a single shared headword; identical surface forms; character-trigram sound similarity; near-identical strings by edit distance; shared meaning by sentence embedding; curated cross-headword synonym pairs; matching dependency-tree patterns at shared headwords; identical dependency frames without any shared word; and shared rare words. The fused score also incorporates word-rarity weighting from corpus-wide frequencies. On the founding Lucan–Vergil benchmark, the previous V3 lexical baseline recovered about 27% of known parallels. Against five commentator-attested benchmarks comprising 862 parallel passages from Latin epic, the new fused V6 system recovered 91.5% in its full ranked output. Of the top distinct results for the Lucan–Vergil comparison, 79% (34 of 43) are meaningful instances of intertextuality. The platform was built by a classicist directing a sequence of AI coding assistants in natural language, with no traditional programming, and with each stage of development checked against the same benchmarks. The platform is available at https://tesserae.caset.buffalo.edu. This preprint has not been peer reviewed.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24206123
Primary Topic
Digital Humanities and Scholarship
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Feature Fusion for Computational Intertextuality Detection in Tesserae V6

Neil Coffee, Walter J. Scheirer
Zenodo (CERN European Organization for Nuclear Research)
Digital Humanities and Scholarship
preprint

Feature Fusion for Computational Intertextuality Detection in Tesserae V6

Neil Coffee, Walter J. Scheirer
preprint en

Abstract

Computational detection of intertextuality has so far used individual features such as word n-grams, edit distance, or word embeddings to identify text reuse or allusion, with only limited efforts to combine two such features for improved detection. This article presents Tesserae V6, a freely available platform that searches, weights, and combines into a single fused score ten distinct signals: shared headwords; a single shared headword; identical surface forms; character-trigram sound similarity; near-identical strings by edit distance; shared meaning by sentence embedding; curated cross-headword synonym pairs; matching dependency-tree patterns at shared headwords; identical dependency frames without any shared word; and shared rare words. The fused score also incorporates word-rarity weighting from corpus-wide frequencies. On the founding Lucan–Vergil benchmark, the previous V3 lexical baseline recovered about 27% of known parallels. Against five commentator-attested benchmarks comprising 862 parallel passages from Latin epic, the new fused V6 system recovered 91.5% in its full ranked output. Of the top distinct results for the Lucan–Vergil comparison, 79% (34 of 43) are meaningful instances of intertextuality. The platform was built by a classicist directing a sequence of AI coding assistants in natural language, with no traditional programming, and with each stage of development checked against the same benchmarks. The platform is available at https://tesserae.caset.buffalo.edu. This preprint has not been peer reviewed.

Zenodo (CERN European Organization for Nuclear Research)
University of Notre Dame (US), University at Buffalo, State University of New York (US)
Digital Humanities and Scholarship
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon