AI services at the University Library
AI tools for literature search
The library's training team keeps track of current AI tools for literature search.
Search tools: These tools allow you to search using search terms.
Semantic Scholar (login with UniAccount)
Semantic Scholar offers access to freely available, licensed and partly also unlicensed scientific full-text articles. Searches are carried out using keywords; it is not possible to search by research question. Citations, references and thematically similar articles are also displayed. When using the library, it is indicated whether a paper cites other titles in the library or is itself cited by them.
Note: The sources are automatically indexed using crawlers. This means that articles from so-called predatory journals may also be included. A check using other verification tools (e.g. DOAJ blacklist) is therefore recommended.
For some Open Access publications, for example from arXiv, the Semantic Reader is available, which enables computer-aided text analysis.
ORKG Ask
After entering a research question on the CORE database (80 million open access publications), ORKG Ask generates a tabular overview of the 5 most important publications. The existing columns (Answers, Insights, TL;DR, Conclusions, Results) can be expanded to include additional columns. The research question cannot be refined. The tool is not suitable for metadata queries and author searches. The user query is followed by a semantic search and information extraction (LLM) as well as the output of filtered results. Bibliographies can be created via a login (free account). The tool is multilingual. The focus of the sources is on the STEM subjects.
Perplexity
Perplexity answers a research question through a web search and returns web sources or hits in Semantic Scholar.
In incognito mode, threads are not saved and are automatically deleted after 24 hours.
Research tools: These tools can be used to find thematically related publications and display them in knowledge maps.
Inciteful
Inciteful creates thematic networks of publications (selected or uploaded). Based on this, the tool shows other similar titles on the topic or scientists researching on this topic. In addition, further metrics are displayed, which can be used to find further literature on the selected topic. The tool has its own Zotero plug-in, which can be used to search directly from Zotero.
Open Knowledge Maps
After entering keywords, Open Knowledge Maps generates so-called knowledge maps, which are visually reminiscent of intersections in set theory and in which 100 of the most relevant results are sorted by subject area. The aim is to provide an overview of the most important sub-areas of a topic, whereby initial documents are also displayed. Open Access titles are clearly displayed, and in some cases PDFs are also attached. If this is not the case, a DOI refers to the original source. If the tool refers to a fee-based platform, only one article can be referenced. Citation aids for the most common English-language citation styles are available. The tool's result lists can be filtered by publication type (OA or not) or sorted by relevance, year of publication, author or title. There is also a search option within the results lists.
When using AI tools, data protection and copyright laws must be observed!
AI projects in libraries
Libraries are also exploring the potential of using AI in library background work beyond literature research. The following projects illustrate the many ways in which AI is already being tested and used in libraries.
BibKI - a chatbot for libraries
The AI-supported chatbot BibKI was developed by the KIT library. It helps users with questions about services, opening hours, etc. The virtual assistant relieves the information desk, makes information about the library more quickly accessible and thus creates low-threshold access to frequently required content. BibKI will also soon be available on the UB/LMB Kassel website.
ORKG Ask
In 2024, TIB Hannover launched ORKG Ask, a service that uses language models, semantic searches and the Open Research Knowledge Graph to provide answers to research questions. This gives researchers quick access to findings from over 80 million scientific publications and enables them to explore complex topics more efficiently.
Automatic content indexing with AI (German National Library)
The DNB tests and operates machine learning methods for the automatic assignment of keywords and notations, including with the open source software Annif. The aim is to significantly increase the quality and efficiency of content indexing and to transfer this into library practice. This makes sources (books, journal articles, websites and others) easier to find.
B!SON - Open Access Journal Recommendation (TIB Hannover & SLUB Dresden)
B!SON is a library recommendation service that uses semantic and bibliometric methods to suggest suitable Open Access journals for the title/abstract/references of a manuscript. The freely usable service is operated by TIB and SLUB. It helps authors who would like to publish open access to find alternatives to closed access journals.
FAQ
The use depends on your subject and the course. Consult your teacher and, if necessary, check the guidelines and examination regulations.
First of all, you should prepare your literature search well. Generative language models from the Academic Cloud of the University of Kassel can help you to develop suitable search terms and search strings. However, specialist databases and the KARLA library catalog remain the most important tools for the actual academic literature search. There you will find the original sources that you can cite and get as complete an overview as possible of the available literature. An AI search, on the other hand, often only provides a random selection of individual sources.
Yes, but they do not replace your intellectual work. You can use the Academic Cloud LLMs of the University of Kassel, for example. Critically review the text summaries and continue working extensively with the literature. AI can support your literature work, but it cannot learn for you.
AI-generated references must ALWAYS be checked, as AI systems can generate hallucinations or provide incorrect or distorted references. Research the original source in specialist databases or in the KARLA library catalog and check whether it actually contains the statements attributed by the AI. You can find out how to check and evaluate sources in the University Library's training courses.
The AI Assistant in Citavi (version 7.0) supports you with literature research and summarizing texts.
Zotero offers the integration of AI into the reference management program with selected plug-ins.
The following also applies here: literature sources and text summaries must ALWAYS be checked.
If you use texts from articles or books in AI tools (e.g. upload or copy them in), check beforehand whether this is permitted under copyright law. Short text excerpts are usually less problematic than uploading complete PDFs or entire chapters. If possible, use the chat AI LLMs of the University of Kassel that are not marked "External" - your entries will not be forwarded to external LLM services. If in doubt, the Citavi and Zotero team at the University Library can help you.
Yes, the University Library offers regular training courses on literature search with AI tools and AI assistants. It gives you an overview of generative language models and supports you in effective prompting. Dates can be found in the course program of the University Library.
You can try out various generative language models via the university-wide Chat AI service. You should make sure that you only use models for content with confidential, sensitive or personal data where the data entered is not processed externally or used for training purposes. Models that are marked "External" in the interface use external services to process the requests. No confidential, legally protected or personal data may be entered into these models (see Chat AI service documentation).
First points of contact are your Faculty and your teaching staff. For literature research with artificial intelligence, please contact the service desk at the University Library or book an individual consultation appointment via our "Meet an Expert" service.
Do not upload any personal or copyrighted texts to AI services without reflection and observe the terms of use and license conditions.