Translation-Service

A novel retrieval-augmented NMT framework that uses contrastive retrieval and hierarchical encoding to leverage diverse, non-redundant translation memories for improved translation quality.

Analysis of degeneracies in inverse propensity scoring for offline machine translation learning from deterministic logs, with proposed solutions and formal insights.

An analysis comparing printed/online dictionaries and term bases, focusing on their evolution, reliability, and future in translation technology.

Analysis of the first application of Neural Machine Translation to Arabic, comparing performance against phrase-based systems and evaluating preprocessing impact.

Introducing WOKIE, an open-source pipeline for automated translation of SKOS thesauri using external services and LLM refinement to enhance accessibility and cross-lingual interoperability in Digital Humanities.

A research paper exploring how Translation Memories (TMs) can be used as prompts to significantly enhance the translation performance of Large Language Models (LLMs) through in-context learning.

An analysis of Machine Translation systems developed for Indian languages, covering approaches like Direct, Rule-Based, and Corpus-Based methods, key systems, and future research directions.

Analysis of using LLMs for few-shot translation of English semantic parsing datasets to train multilingual parsers, outperforming translate-train baselines across 50 languages.

An in-depth exploration of Neural Machine Translation, covering its history, core neural network concepts, language modeling, encoder-decoder architectures, refinements, and future challenges.

Analysis of a novel post-training domain adaptation method for NMT, exploring incremental specialization, experimental results, and future applications.

An analysis of challenges and opportunities in legal translator training, exploring updated models, technology integration, and the evolving role of translators as intercultural mediators.

Introduces Structure-Invariant Testing (SIT), a metamorphic testing method for validating machine translation software by analyzing structural consistency in translated outputs.

An analysis of how professional translation ethics codes apply to non-professional contexts like crowdsourcing, fan translation, and FOSS localization, highlighting differences and innovations.

An analysis of fuzzy match algorithms for Translation Memory systems, evaluating their correlation with human judgments and proposing a new weighted n-gram precision method.

Analysis of TM-augmented NMT from a probabilistic retrieval view and variance-bias decomposition, proposing a method to address contradictory performance in high/low-resource scenarios.

Analysis of modern QA tools for translation, their integration with CAT tools, industry standards, and practical evaluation of standalone QA software outputs.

Analysis of how globalization and technology redefine translation demands, positioning translators as cultural mediators and strategic business assets.

Analysis of a variational encoder-decoder model for NMT that introduces continuous latent variables to explicitly model underlying semantics, improving translation quality.