Primo Research Assistant
When using Primo Research Assistant, you should write your search as a question in natural language. It's just like interacting with a chatbot, such as Microsoft Copilot or ChatGPT. This is a significant difference from traditional Primo search, where you normally use keywords, topics, titles, authors and the like.
So Primo Research Assistant and Primo Search are two different tools, each with its own functionality. But just like Primo Search, the answers you get from Primo Research Assistant are based on the materials and sources that the library provides and makes available. Therefore, if you want to use generative AI in your literature search, Primo Research Assistant is a more qualified AI service than, for example, Copilot and ChatGPT.
How does the Primo Research Assistant work?
Primo Research Assistant uses generative AI. Specifically, it transforms your question into a search string that Primo can understand using a large language model (LLM).
Based on the search in Primo, the top 30 results are selected. The LLM then selects the five sources that, based on their abstracts, appear to be the most relevant to answer your question.
The LLM will now use the selected sources to answer your question and will make a reference to these sources.
The assistant presents the answer, along with the five sources selected, with links back to Primo so you can explore them further. It also gives you a link to a custom search in Primo that you can use to find additional material related to your question.
For each source, you will be able to read the corresponding abstract, which forms the basis of the answer, and find links to access the material through the library's resources. If we don't have access to the material, you can of course request it through interlibrary loan.
Key facts about Primo Research Assistant
- You must be logged in to Primo to use the Primo Research Assistant
- To answer your question, the LLM does not use its own training data. Instead, the LLM uses the five selected sources from Primo, which reduces the risk of false answers (hallucinations)
- Primo Research Assistant access source data from most of the library's collections, with the exception of a few local library resources such as the Danish Article Index, AAU Student Projects, VBN or printed books
- While the risk of false answers is minimised, you should still be aware of any limitations, inaccuracies or biases in the answer when using a tool based on generative AI
- The Primo Research Assistant answer itself is not subject to copyright and you are free to use it in academic, professional and research contexts as long as you cite the selected sources. However, always remember to attribute the use of generative AI!
- Unlike most other chatbots, your questions are not part of a conversation. Therefore, each question is independent and you can't refer to previous answers, for example
- You can ask questions in languages other than English. For example, if you ask a question in German, the LLM and Primo will try to include German sources in the answer, in addition to the English ones. You can also ask a question in English and ask for an answer in Danish
- Some publishers have declined to provide source data including APA, Elsevier and JSTOR
- Your questions will not be used as additional training data for the language model. Data will not be shared with third parties and all processing will take place within EU borders
- By providing feedback with đź‘Ť or đź‘Ž you contribute to the improvement of the tool. However, please note that your feedback and questions may be read and reviewed by our supplier