heiBOOKS

Abstract: Chatbots and virtual assistants are becoming increasingly popular for user questions and support. With FAIR GPT, the Mannheim University Library released a virtual assistant for research data management (RDM) in 2024, designed to help researchers and institutions in making their data FAIR (Findable, Accessible, Interoperable, Reusable). FAIR GPT provides various RDM services, e.g. metadata enhancement, repository selection and FAIR assessment. However, FAIR GPT has numerous disadvantages: As a ‘Custom GPT’ of OpenAI, it is proprietary software that only outputs sources for the generated answers if it uses its internal web search tool (which cannot be controlled by the user) and therefore lacks transparency. Reliance on external cloud-based services leads to privacy 
concerns when dealing with sensitive (meta)data and the chatbot is still prone to hallucinations, thus reducing its trustworthiness. These issues led us to explore alternative open-source solutions. We searched for opensource 
alternatives to Perplexity.ai, a system known for its ability to provide citations for the information it retrieves through web searches. We identified three candidates available on GitHub: Perplexica, sensei, and farfalle. These tools use local instances of the metasearch engine SearXNG to perform internet searches, using the results as input 
for Large Language Models (LLMs). We modified these tools to focus specifically on RDM tasks, releasing the new versions on GitHub openly under the names FAIRplexica, FAIR-sensei and FAIR-farfalle. 
Keywords: Research Data Management, LLMs, Chatbot, RDM Assistants, FAIR data