Conference Program 2025

memoQ co-CEO, Peter Reynolds, will give a progress report on memoQ’s activities over the last year.

At Alpha we have built a custom AI to act as a reviewer. It is trained to comply with ISO specifications for LQA (ISO 5060) and provide feedback on segments that are deemed less than perfect. It is integrated in the server based workflow, so when a translator delivers a file, it is sent to the AI for review, and then automatically reimported with the comments, and segments that are commented have changed status to Edited, so they are easy to filter. The translator can read the comments in the preview/comments pane and either disagree and reconfirm the segment as it is, or implement the suggested changes or make other changes. It is not designed to replace a human reviewer, but to provide feedback in workflows where there isn’t one, and can also be deployed to assess MT.

Machine translation engines and LLMs have reached a level where serious translation mistakes are rarely due to flaws in the technology itself but rather to how they are applied. In fact, many of the most critical errors today arise from a lack of contextual awareness, where translations fail to maintain consistency, coherence, and accuracy across segments.

Yet not a single MT plugin currently available in memoQ, including memoQ’s own AGT, enables context-aware translation—all still translate one segment at a time in complete isolation—an astonishing limitation that leads to inconsistencies, mistranslations, and avoidable errors.

To address this, we enhanced our customizable GPT-powered solution, PromptLoc, first introduced at last year’s memoQfest, adding functionality specifically designed to overcome segment-based constraints. The improved system allows multiple segments to be translated simultaneously while retaining segmentation, and, in addition to processing batches of segments at a time, it also allows users to customize memory length (i.e. how many previous source segments and their translations are retained), ensuring that context is maintained not only across segments but also across batches.

It also integrates seamlessly with memoQ’s Resources API, automatically enriching each translation batch with relevant TM and TB matches. Additionally, users retain complete flexibility to customize prompts to suit diverse project requirements.

We’ve successfully adopted this approach internally at Edimart for MTPE and raw MT tasks, greatly improving our localization workflow and translation quality. In this talk, I will share practical insights from our experience, explain the technical aspects behind this solution, and demonstrate how this new approach can benefit other memoQ users facing similar challenges.

The topics of master classes will be published shortly.

The memoQ QA check contains a lot of useful checks, but there might be more things you want to do.
Using auto-translation rules and activating them for the QA check can already catch some of them, like making sure that there is a non-breaking space between a number and a measurement. But even that might not be enough, because there might not be any element in the source to compare the target to.
Here the QA check can look at the target language only, for example catching any half-width katakana characters in a translation into Japanese or making sure the currency symbol is in the right place or the date formats are consistent…

The topics of master classes will be published shortly.

The master classes on TMS/server optimization at the memoQfest 2023 & 2024 provided a good opportunity for advanced server users to meet and discuss various performance issues with memoQ’s developers.

The goal of this session would be to follow up on the issues discussed last year and provide a forum for server managers to share insights and discuss persistent challenges with the developers.

In the era of neural machine translation, maintaining consistent terminology remains a crucial challenge. AI-TermInator demonstrates how artificial intelligence can enhance, rather than replace, human terminology work. This presentation showcases a novel term extraction tool that combines AI’s capabilities with human expertise through domain-specific extraction, smart filtering, and context-aware processing. Unlike traditional statistical extractors, AI-TermInator understands context and validates terms through multi-step verification, while respecting human-curated exclusion lists and existing terminology. We’ll demonstrate live examples of how translation teams can reduce terminology extraction time while maintaining high quality through this human-AI collaboration approach.

In translation projects, handling URL slugs presents unique challenges due to specific formatting rules that differ from standard text translation. In this case study, we explore how we were able to exploit various memoQ functions and AI-driven automation to optimise our workflow for a website localization project involving six languages. After a first trial with traditional pre-translation and regex-based processing that proved inefficient, we decided to leverage ChatGPT to automate slug conversion. This approach drastically reduced manual effort while ensuring linguistic accuracy. The final review by human linguists confirmed the effectiveness of AI-driven processing. This presentation showcases how AI can enhance translation efficiency and reduce translation costs while maintaining human expertise at the core of quality control.

I don’t remember the last session that covered what a Project Manager working in a LSP means, or should mean/do… (admittedly, I didn’t participate in many sessions along the years, skipped many… shhh!).

So, my proposal will try and paint a picture of MY idea of what a Project Manager in a translation company should look like (and if it’s really feasible).

• Will try to match each part (or most of them) of the body to a feature, functionality, something from memoQ.
• Will add some AI into the mix since it’s still the go to acronym…
• By the end of the session, I will have created a monstrosity of skills, mashed up into one entity.

Let the production team guide you through the most recent developments, and get a glimpse of what’s cooking at memoQ right now.