Masterarbeit
Context-Aware Real-Time Webpage Text Translation Using Large Language Models
Research Area
Intelligent Information Management
Advisers
Bai Li M.Sc.
Description
The increasing volume of multilingual web content has spurred a demand for more accurate access to multilingual information, particularly in online systems related to transnational higher education. While existing browser translation tools offer basic support, they typically translate text sentence by sentence, ignoring the structure and context of web pages. This can lead to inaccurate, inconsistent, or unusable translations. Recent advancements in Large Language Models (LLMs) have opened up new possibilities for more context-aware translation. As part of the central topic "EduLoom: A Multimodal AI Framework for Integrating Educational Web Resources“, this thesis aims to design and implement an LLM-based real-time web page translation system. This system analyzes not only the rendered page but also structural information such as HTML or DOM data to better understand the page's content and context. It should extract relevant text, generate context-aware translations, and replace or expand the original content while maintaining layout and usability. Since this project is applied to a university environment, domain-specific requirements need to be considered, such as simultaneously displaying translated course names and original titles and identifiers, as well as accessibility issues such as screen reader compatibility and providing simplified language versions when necessary. The prototype should be evaluated in terms of translation quality, latency, cost, and usability, and compared with existing browser-based translation tools. The objective of this thesis is the creation of a solution or the combination of existing techniques to solve the problem of web translation through language model as described above. This comprises the analysis of the state of the art of AI-based translation as well as the demonstration of the solution by implementation.
