Our research competes at the top level in high impact scientific areas.

Research Tasks

Human Language Technology is a multi-disciplinary research area that addresses the automatic processing of human language for a range of tasks. In particular, our research unit focuses on:

• Automatic speech recognition, that is the conversion of a speech signal into a readable text
• Machine Translation, that is the translation of speech or text from one language into another
• Content Processing, that includes the extraction, integration and retrieval of information from texts

HLT draws on several disciplines, such as Computational Linguistics or Natural Language Processing, Speech Processing, Information Retrieval, Machine Learning, Pattern Recognition, ....

Approach

Research in HLT is mainly experimental and is currently dominated by the statistical approach, also referred to as data-driven or machine learning approach. In particular, statistical models are developed to capture properties of language relevant to the task, as well as a recognition or classification algorithm. Model parameters are typically learned from large data sets that represent a specific application domain. Briefly, experimental research in HLT is conducted according to the research cycle shown in the nearby figure.

Organization

Statistical methods and large-scale experimental evaluations are nowadays at the forefront of HLT. This has increased over time the importance of setting up an infrastructure for HLT, to collect and organize language resources and to manage the necessary computing resources.