Master thesis »Extraction of structured materials data through large language models«

At the Fraunhofer IWM, 330 employees conduct research on materials and components with the aim of better understanding, developing, processing and using them. In our projects, we bridge the gap between the properties of materials and the durability, safety and function of technical systems.

We show ways and solutions for more energy efficiency and sustainability. After all, materials are crucial for climate neutrality, for the careful use of our planet's limited resources and for the sustainable transformation of our economy.

 

What you will do

 

Within our group »Meso- and Micromechanics«, we amongst others work on the development of data-driven methods for the accelerated characterization and development of materials.

 

A wide variety of tasks await you:

The aim of this master's thesis is to extract materials information from a literature corpus and from image data, in a targeted and automated manner and to convert it into a structured data format. For this purpose, language models or language model systems that can handle multimodal data are to be conceptualized, applied, and refined. Combinations of different language models with so-called “tools”, which can be called up autonomously by the language model, can also be used to solve the task. These tools can represent database queries, API calls or the inference of models/functions. The information should be extracted in such a way that conformity with a material science application ontology is ensured. To this end, schema languages such as JSON schema are to be used to provide the language models with information on the permitted vocabulary and its definitions. The aim is to build a bridge between unstructured data sources and existing, elaborately generated knowledge graphs.

 

The exemplary use case within the thesis is materials fatigue, as there is a particularly pronounced scarcity of data in this subdomain and preliminary work such as prepared data sets and data schemas already exist.

 

What you bring to the table

 

Your professional background:

You are studying computer science, computer engineering, computational methods in engineering or a related subject. Familiarity with programming in Python and the functional principle of language models is beneficial. Initial experience in the use of API endpoints for the inference of language models and preliminary experiences in fine-tuning language models are advantageous.

 

Qualifications that complement your profile:

You enjoy working in interdisciplinary teams and are characterized by a systematic way of working and an analytical way of thinking. Proficiency in English language helps you to communicate in multicultural teams in your day-to-day work. You have an inherent interest in applying new computer science methods to scientific problems.

 

What you can expect

 

Always something new:

Working at the Fraunhofer IWM means researching innovative topics in an established research institute. Thanks to our flat hierarchies and open working culture, you will always have the right support at your disposal.

 

What we offer you:

Through flexible working hours and mobile working, we are committed to ensuring that you can balance your free time and work. In family emergencies, we offer emergency childcare and allow you to bring your children to our parent-child office if necessary, in order to react flexibly to unforeseen situations or childcare bottlenecks. You can also benefit from a wide range of employee discounts on events, furniture, clothing and much more.

 

We value and promote the diversity of our employees' skills and therefore welcome all applications - regardless of age, gender, nationality, ethnic and social origin, religion, ideology, disability, sexual orientation and identity. Severely disabled persons are given preference in the event of equal suitability. 

 

With its focus on developing key technologies that are vital for the future and enabling the commercial utilization of this work by business and industry, Fraunhofer plays a central role in the innovation process. As a pioneer and catalyst for groundbreaking developments and scientific excellence, Fraunhofer helps shape society now and in the future. 

 

 

Apply online and shape the future of Fraunhofer IWM together with us!

 

 

 

Our recruiter Lea Hauserstein will be happy to answer all your questions about the position and our application process. 
You can reach us by e-mail at recruiting@iwm.fraunhofer.de or by phone at +49(0) 761 5142-234.

 

All topic-specific questions will be answered by Dr. Ali Riza Durmaz under +49(0) 761 5142-195 or ali.riza.durmaz@iwm.fraunhofer.de.
 

 

Nothing suitable for you?

We always have new projects coming up and a wide variety of vacancies. Perhaps you will hit the mark with an unsolicited application to write your thesis in one of our scientific business areas.

Apply online and we will add you to our application pool.

 

Fraunhofer Institute for Mechanics of Materials IWM 

www.iwm.fraunhofer.de 

 

Requisition Number: 76042