UST-alignment

To align Hebrew source text with an English UST translation, you provide the Hebrew USFM and English UST draft. The system generates index-based mappings from Hebrew to English phrases and outputs a machine-readable JSON mapping.

Meaning-based alignment handles implied information and radical restructuring by mapping phrases based on semantic equivalence rather than strict word-for-word matching. It constructs index-based mappings to accurately represent split alignments and superscriptions in the output.

Yes, you can convert alignment mappings into USFM format. The workflow uses a two-step process where the AI generates index-based JSON mappings, and a subsequent script converts those mappings into properly formatted aligned USFM.

To generate a JSON mapping for Hebrew-to-UST alignment, you need Hebrew source text in USFM format and an English UST draft. Optional hints can also be provided to help produce deterministic, validated alignments.

Yes, the alignment workflow supports JSON-based mapping for downstream publication tools. It outputs a JSON structure specifically designed to be compatible with downstream USFM tools for publication or quality assurance tasks.

Aligning superscriptions and split alignments requires careful handling of radical restructuring. The system addresses these limitations by using meaning-based index mappings to ensure the JSON output accurately represents complex semantic relationships.