The research assistant is a retrieval augmented generation (RAG) service combining language capabilities of large language models (LLM) with high quality library resources from Library Search.

Ask it a research question and it will:

• Turn your question into a search query with different variations
• Retrieve results of the most relevant sources
• Use the top five results to generate an overview from the abstracts and link to the sources
• Generate related questions to dive deeper into the topic
• Link through to Library Search for more results

Use this tool for:

1. An overview of a topic that is new to you
2. If English is not your first language, use this tool in other languages to help you understand a complex topic
3. A source of background information with references and links to the content
4. Translating your question into a complex Boolean search to find more resources

General advice:

• Check with your educator and assessment requirements about the use of generative artificial intelligence tools
• Verify the accuracy of the information and the citation details of the sources

Get started

More details

How does it work?

RAG tools like the research assistant use LLMs to query an index of resources, in this case, the Library Search index. An overview of the process is:

1. Your question is sent to the LLM and turned into a Boolean search query with different variations
2. The query is then sent to Library Search to get results
3. Up to 30 top results are re-ranked using embeddings to find the best five sources
4. The LLM uses the abstracts of the five best sources to create an overview with references

What sources are not included?

Not everything in Library Search is included in the research assistant:

• The library's physical resources, like books as well as ebooks
• News content such as newspaper articles, newsletters and text resources
• Articles from ScienceDirect, JSTOR and APA (PsycArticles and PsycInfo)
• Sources that don't have abstracts or have brief records in Library Search

You can use the 'View more results' option displayed at the end of the sources to search all Library Search resources. You won't get an AI-generated summary, but you will find resources not included in the research assistant.

What are the limitations?

• Always check the accuracy of the results and summaries. RAG tools like the research assistant can sometimes misinterpret sources. For example, a source might be referenced in the summary, but the interpretation may be incorrect.
• You can't ask for specific source types, like peer-reviewed articles, or ask for results for a specific time period in your initial question. You can instead use the filter options to search by resource types and dates.
• It's not possible to ask follow-up questions, instead, you have to include all relevant information in each question.