Scientist in the field of GeoAI-with a focus on multilingual geo-parsing - Anwendungsentwicklung, Datenbankentwicklung/BI, Ingenieur

The Institute of Data Science in Jena supports all of the DLR’s areas of application (aviation, space, energy, transport, security) with data science concepts, methods and tools. To this end, methods with a focus on applications are researched and developed through interdisciplinary collaboration. The results are put into practice in cooperation with other DLR institutes and external partners. The research and development work carried out by the Data Acquisition and Mobilization department (DMO) is aimed at providing high-quality data for industry and academia. The key challenges here are usability, availability and access. These challenges form the basis for the department’s key priorities, in particular the development of effective methods to ensure data quality during data collection, the exploitation of previously untapped data sources, and the creation of effective and target-group-oriented data access. The research objective of the Multimodal and Spatial Information Retrieval (DMO-MIR) group is to make unstructured, multimodal and heterogeneous data usable. The group conducts research into practical methods and develops modular software for the collection, processing, filtering, geolocation and content evaluation of unstructured data. The aim is to make data that was previously not machine-readable usable as a source of information for various DLR applications, as well as for decision support. ## Your tasks - Contribution to the DFG project GeoMMCorpus+Eval (https://gepris.dfg.de/project/575831912)

• Familiarisation with the current state of research in the field of GeoAI, e.g. on multilingual geo-parsing, the creation and annotation of geographical text corpora, and LLM-supported annotation and quality assurance procedures
• Designing scientific methods and workflows for the extraction, processing, selection and annotation of large, heterogeneous and multilingual text datasets
• Development and application of data science methods for the extraction, cleaning, filtering and structured processing of large multilingual text corpora from various text genres, such as websites, news articles, social media, academic publications and historical documents
• Planning experimental studies on LLM-assisted annotation and iterative quality assurance using human-in-the-loop methods

Design and application of suitable quality metrics for evaluating the results ## Qualifications - A completed academic degree (Master’s / University Diploma) in Computer Science, Data Science, Geoinformatics, Computational Linguistics or another relevant field of study

• In-depth knowledge and practical experience in the areas of natural language processing, information extraction and machine learning
• Experience in working with large language models (LLMs), particularly in their application and customisation, e.g. through prompting and fine-tuning
• Programming experience in Python, as well as experience with relevant software development and version control tools, e.g. Git
• Knowledge and experience in the processing and analysis of large and heterogeneous text datasets
• Ability to work independently and autonomously on scientific research questions, as well as to systematically plan, carry out and evaluate scientific experiments
• Very good written and spoken English