Student work and projects
The work carried out at our chair is concerned with software and systems engineering in the field of digital farming. For this purpose, we consider systems along the entire value chain. Our location in Rhineland-Palatinate offers an excellent basis due to the versatile coverage of different areas, such as wine, vegetable and grain cultivation, but also the immediate proximity to research and industry.
Interested students looking for theses and projects are always welcome to contact us to discuss interests and possible topics.
Our research focus:
- Software and systems engineering in the field of digital farming
- Requirements engineering of the different actors in the agricultural ecosystem
- Improvement of interoperability and networking between actors and systems
- Improvement of the acceptance of digital farming solutions (e.g. FMIS, decision support systems, agricultural machinery)
Geometric Representation of AGROVOC for Improved Semantic Linking
OBJECTIVES OF THE MASTER THESIS
The aim of the master's thesis is to investigate whether geometric representations of AGROVOC can improve semantic linking in the context of the PrOWhEAT project. Euclidean, hyperbolic and asymmetric/Finsler approaches will be studied. As the main research contribution, the thesis will develop and investigate a relation-aware asymmetric representation that models relation type and direction. The best structural representation will be integrated into PrOWhEAT alongside text similarity.
MOTIVATION
In PrOWhEAT we extracts agricultural information from PDFs, Excel and CSV files and links terms to AGROVOC. Current linking mainly uses text similarity and local ontology information. Yet AGROVOC is a directed, typed graph: relation type and direction carry meaning that normal vector spaces may not capture well. A geometry that preserves this structure could improve concept selection while keeping AGROVOC in RDF/SKOS.
KEYWORDS
AGROVOC, semantic linking, knowledge graphs, Euclidean/Hyperbolic/Finsler geometry, relation aware embeddings, PrOWhEAT
MILESTONES OF THE WORK
Search for the state of the art in ontology/knowledge-graph embeddings and geometry-based semantic linking
Prepare the AGROVOC RDF/SKOS graph and establish the current PrOWhEAT linking approach as a baseline
Implement Euclidean and hyperbolic representations of AGROVOC
Develop and investigate an asymmetric/Finsler representation for directional relations
- Extend it to a relation-aware representation for different AGROVOC relation types
Evaluate hidden relation recovery, relation direction, and alternative-label-to-concept retrieval
Perform an external semantic-relatedness check (e.g. ConceptNet) and integrate the best representation into PrOWhEAT
Compare with existing text-based linking, analyse the results, and document the thesis.
WE OFFER
Collaboration in a pioneering research project.
Opportunity to investigate a novel research idea with potential for a scientific contribution.
WHAT YOU BRING WITH YOU
- Background in Computer science, data science or related studies
- Good Python programming skills
- Interest in machine learning, knowledge graphs/ontologies and mathematical representations
- Interest in agricultural applications
- RDF/SKOS/SPARQL experience is helpful but not required
Contact:
M.Sc. Anish Bipin Jagadale (anish.jagadale(at)rptu.de)
Design and Implementation of a RAG‑Based Chatbot for Knowledge Extraction from the Agricultural Knowledge Sources
In cooperation with the 'Friends of Digital Farming'
OBJECTIVES OF THE MASTER THESIS
The objective of this thesis is to design and implement a fully local RAG chatbot capable of delivering precise, context‑aware answers from the Agricultural knowledge sources like papers, books, etc. This involves developing a robust PDF ingestion pipeline to extract and pre-process handbook content, constructing a vector‑based retrieval system using Chroma DB, and integrating a local LLM to generate high‑quality responses. The work further aims to build a production‑ready FastAPI backend that exposes the chatbot as an accessible interface for agricultural practitioners, while evaluating the system’s accuracy, latency, and usability in real‑world digital farming scenarios. Ultimately, the thesis should serve as a proof of concept and seeks to demonstrate how retrieval‑augmented generation can transform large agricultural documents into interactive, user‑friendly knowledge systems.
MOTIVATION
The motivation for this thesis is to make the agricultural knowledge sources like papers, books, etc. easier to access by transforming its extensive content into an interactive, question‑answering system. Farmers and students often struggle to locate relevant information quickly, and a local RAG chatbot enables fast, precise retrieval without relying on cloud services. By combining semantic search with a locally running LLM, the project supports digital adoption in agriculture while ensuring data sovereignty and practical usability.
KEY POINTS
- PDF ingestion pipeline
- Vector embeddings
- RAG workflow
- Evaluation Framework
- User-centric design
MILESTONES OF THE WORK
- Requirement analysis for user needs, the legal requirements (copyrights, etc.) and handbook structure.
- PDF extraction and pre-processing pipeline, setting up Vector database.
- RAG pipeline implementation combining retrieval and LLM reasoning.
- FastAPI backend development for serving queries.
- User interface prototype for interacting with the chatbot.
- Evaluation and benchmarking of accuracy, latency, and user satisfaction.
- Thesis documentation and final write-up including architecture, experiments, and conclusions.
WHAT YOU BRING WITH YOU
- Studied computer science, bioinformatics and business informatics.
- Interest in scientific work and literature research methods.
- Interest in Machine Learning Application.
- Enthusiasm for agricultural issues.
CONTACT
M.Sc. Vishal Sharbidar Mukunda (mukunda(at)rptu.de)
Optimierung der Interoperabilität von IT-Systemen in der Wertschöpfungskette im Obst- und Gemüsebau
Research Objectives:
- Different IT systems are used along the value chain in fruit and vegetable production. These IT systems are partially interconnected.
- However, interoperability between the systems is often insufficient. This leads to the necessity of manually re-entering data that is already available in digital form into other IT systems.
- This master's thesis aims to develop a concept for optimizing interoperability using quality assurance and complaint management systems in fruit and vegetable production as an example.
- The project is conducted in cooperation with Pfalzmarkt für Obst und Gemüse eG.
Expected Contributions:
- Familiarization with the state of the art in the field of interoperability.
- Familiarization with the working processes at Pfalzmarkt für Obst und Gemüse eG.
- Analysis of quality assurance and complaint processes at Pfalzmarkt für Obst und Gemüse eG.
- Modeling the data models of the different IT systems.
- Semantic description of the data to be exchanged.
- Design of a / alternative concepts for improved interoperability.
- Possible prototype development.
- Possible evaluation of the prototype with end users from Pfalzmarkt für Obst und Gemüse eG.
Important note: Proficiency in German is mandatory.
Contact: