https://grafemas.net/tags/computational/ Recent content in computational on Γραφεμας Hugo -- gohugo.io en Fri, 03 Jul 2026 00:00:00 +0000 https://grafemas.net/blog/race/embedding-explorer/ Fri, 03 Jul 2026 00:00:00 +0000 https://grafemas.net/blog/race/embedding-explorer/ The analyses published in this series have relied on word2vec models trained on a lemmatized ancient Greek corpus, sliced by century from the 8th to the 1st BCE. The tool embedded below makes those models directly explorable. The Neighbourhood Browser returns the nearest neighbours of any lemma across one or more century models simultaneously, so you can see how a term’s distributional company changes over time. The Similarity Probe takes a focal term and a user-defined concept list and returns a colour-coded cosine-similarity table across selected models — this is the view underlying most of the quantitative claims in the series. https://grafemas.net/blog/race/diachronic-embeddings/ Sun, 28 Jun 2026 00:00:00 +0000 https://grafemas.net/blog/race/diachronic-embeddings/ The four previous notes in this series used per-author fastText models to compare how βάρβαρος is distributed across the vocabularies of Herodotus, Euripides, Plato, and Xenophon. The CADE (Compass-Aligned Distributional Embeddings) framework extends that analysis longitudinally: a compass model trained on the full ancient Greek corpus serves as an alignment anchor, and separate models are trained on century-sliced sub-corpora — 6BC through 1BC in this dataset. The century slices allow a question the per-author models cannot answer: whether the differences between Herodotus and Plato are specific to those authors or representative of their periods. https://grafemas.net/blog/race/euripides-embeddings/ Sun, 28 Jun 2026 00:00:00 +0000 https://grafemas.net/blog/race/euripides-embeddings/ A previous note looked at what a fastText model trained on Herodotus reveals about the discourse of group difference in that corpus. The same framework applied to Euripides produces a useful comparison — not because the models are large enough to settle interpretive questions, but because systematic differences between two authors at the corpus level can generate hypotheses that close reading can then test. The Euripides model has 13,189 tokens against Herodotus’s 9,278, which reflects the larger size of the extant tragic corpus. https://grafemas.net/blog/race/herodotus-embeddings/ Sun, 28 Jun 2026 00:00:00 +0000 https://grafemas.net/blog/race/herodotus-embeddings/ I have a fastText model trained on Herodotus (using CADE, a diachronic embedding framework) and wanted to see whether it captures anything meaningful about how group difference works in the text — and whether it could be useful for a broader project on racializing discourse in ancient Greek sources. The first thing the model shows is where βάρβαρος actually lives in Herodotean discourse. Its nearest neighbors are Σαλαμίς, στρατόπεδον, διώκω, ναῦς, ναυμαχία, φυγή, πολέμιος, Ἀρτεμίσιον. https://grafemas.net/blog/race/plato-embeddings/ Sun, 28 Jun 2026 00:00:00 +0000 https://grafemas.net/blog/race/plato-embeddings/ Two previous notes applied fastText models to Herodotus and Euripides, tracing how the semantic neighborhood of βάρβαρος shifts between a military cluster (Herodotus) and a political-social cluster (Euripides). The Plato model extends the comparison to a third register — philosophical prose — and produces results that complicate any simple story of progressive naturalization. The first thing the Plato model shows is that βάρβαρος and Ἕλλην have become each other’s nearest neighbors, at a similarity of 0. https://grafemas.net/blog/race/xenophon-embeddings/ Sun, 28 Jun 2026 00:00:00 +0000 https://grafemas.net/blog/race/xenophon-embeddings/ Three previous notes in this series have traced the semantic neighborhood of βάρβαρος across Herodotus, Euripides, and Plato, watching the term shift from a military label (Herodotus) to a political-social category (Euripides) to a consolidated dyad with Ἕλλην (Plato). The Xenophon model, trained on the same surface-form fastText framework, completes the four-author comparison and produces the most distinctive profile of the series: βάρβαρος in Xenophon is not simply a regression to Herodotus’s military cluster.