under construction

LLM-based Text Analysis for
Translation of Literary Texts

Master's ThesisUniversity of InnsbruckPhilipp Parzer

This paper introduces a novel enhancement of human-in-the-loop literary translation processes by including large language models (LLMs). LLMs annotate the source text hinting at possible pitfalls, hard to translate sections, and other peculiarities. This increases human translators' efficiency, as translators can focus on translation problems and spend less time post-editing standard language. The system outlined uses OpenAI's function-calling API.

playground preview

Playground is
currrently in closed preview

request access

or

login

Examples

The texts provided below have been used to evaluate the efficacy of this tool in the paper. It's important to note that the paper's focus on Russian and German literary texts presents an insightful yet limited perspective that highlights the need for expanded research to encompass a broader range of language pairs, genres, etc.

Russian, Viktor Pelevin, Spi, p. 1
Russian, Katya Kachur, Gen Rafaila, pp. 5-6
German, Janne Teller, Nichts. Was im Leben wichtig ist, pp. 8-10
German, Caroline Wahl, 22 Bahnen, pp. 9-10

more examples soon

Prompts

section coming soon

The Paper

section coming soon

Further Research

section coming soon

Public API

section coming soon

Architecture

section coming soon

Local Setup

section coming soon

open-source