v1.1.0

FLAME v1.1.0 is a Python-based text reuse analysis tool featuring both CLI and GUI interfaces, specifically designed to detect and map formulaic language in historical corpora like medieval charters. At its core, the software implements a flexible Leave-N-Out (LNO) n-gram approach paired with subword tokenization, allowing it to look past scribal variations, regional dialects, and spelling shifts that typically break traditional n-gram models. The framework operates autonomously through self-supervised engines, leveraging an automated Auto-Tune system to discover ideal window properties, Otsu's method to dynamically set similarity cutoffs, and vectorized NumPy lookups for adaptive character normalization. Results are delivered as memory-efficient sparse matrices (dist_mat.npz), linguistic variation spreadsheets, and highly interactive side-by-side HTML reports featuring live fuzzy-matching sliders for effortless philological exploration.

v1.0.0.

FLAME is a Python-based tool with both Command-Line (CLI) and Graphical (GUI) interfaces, designed for identifying and analyzing formulaic language and text reuse, particularly in historical corpora like medieval charters. It uses a Leave-N-Out (LNO) n-gram approach, which is highly effective for detecting variant forms of expressions that differ due to scribal variations, regional dialects, or other textual modifications. It automatically learn normalization rules from the corpus itself (handling medieval ligatures and special characters), uses subword tokenization to handle rare words and morphological variants. Automatically suggest an optimal vocabulary size for the tokenizer based on the corpus's statistical properties. It perform both intra-corpus and inter-corpus comparisons, and automatically determine an optimal similarity cutoff score using Otsu's method.