Finding Additional Sources

Finding Additional Sources#

In addition to keyword searches, there are a number of techniques you can use to find additional literature that may help you answer your research question. To get a more complete picture, we recommend using these techniques only in addition to keyword-searching.

Standard Practice

Citation-Based Searching
Look at the sources of a relevant article to find additional sources

AI Assist 1

Literature Mapping
Use specific tools to map citations and uncover new literature

AI Assist 2

Semantic Searching
Use natural language in academic databases to uncover articles potentially not found by a keyword search

AI Assist 3

Agent-Based Searching
Use deep search techniques in academic databases to uncover more literature

Recap Standard Practice: Citation-Based Searching#

A common technique to find additional papers is looking at the citations of relevant article to find related information (“snowballing”). Depending on who cites an article, or who the article cites, there may be relevant information present. Learn more about basic snowballing techniques in this video:


"Alternative Ways of Searching" by TU Delft Library Education Support is licensed CC-BY

AI Assist 1: Literature Mapping#

AI tools can aid finding additional literature in multiple ways. There are AI-tools that can support this process by helping you explore the literature around a key article. To use this technique effectively, it helps to start with identifying a reliable and relevant key article for your specific research question. A literature mapping tool subsequently finds additional articles based on the references of the initial source you provided.

For example, with Litmaps you can get an overview of who has cited and who is cited by the article. Via ConnectedPapers you can get a map of prior works and derivative works, and it also creates maps based on semantic similarity. ResearchRabbit can map and visualise different citations and authors.

Guided Activity: Literature Mapping

Tool Suggestions: Research Rabbit, Connected Papers, Litmaps, Inciteful

Step 1: Upload a Key Article

  • Start with a key article in your field—one you already know is useful.

  • Upload the article to the Literature Mapping tool

  • Explore the network of citing and cited papers the tool suggests.

Step 2: Reflect on the Results Questions to guide your exploration:

  • How do the results compare to the results of your keyword search strategy?

  • What additional articles seem promising?

  • Are any of them newer or from different disciplines?

AI Assist 2: Semantic Searching#

In addition to citation searching, there are new AI-assisted search techniques emerging alongside traditional keyword-searching that can help you find additional information. Academic tools like Elicit and SciSpace use semantic searching to find papers. The tools search semantic scholar and retrieve articles where the language used best matches the language of a query written in natural language.

This technology can be used to find papers in addition to those found through a keyword search. You can use these tools by simply putting in your research question and ask it to find relevant peer-reviewed articles.

A lot of these tools also offer possibilities to build further on the information you found, by connecting it to citations or similar results.

Guided Activity: Semantic Searching

Tool suggestion: Elicit

Step 1: Enter Your Research Question

  • Enter your research question (not just keywords!) in the semantic search bar.

  • Review the top results returned.

Step 2: Explore the results

  • Select 1-2 promising articles from Elicit

  • Use Elicit’s follow-up questions or search features to: explore related literature (“search citation trails”) or map the paper’s structure (add columns for methods, outcomes, claims).

Step 3: Reflect on the Outcomes: How do the results compare to the results of your keyword search strategy? Are the additional articles helpful? What other things do you notice about your results?

AI Assist 3: Agent-Based Searching#

Instead of just looking for keyword matches, agent-based searching techniques use a combination of semantic and keyword searching and then ask an LLM to rate the relevance of the results. It imitates iterative searching and as a result finds more hidden papers. It can take a long time (about 3 minutes) to run a search. This approach can be an interesting addition for researchers looking for highly specific papers. Example tools are Ai2 Paper Finder, Future House Platform and Undermind.ai

Guided Activity: Agent-Based Searching

Tool suggestion: Undermind.ai

Step 1: Enter Your Research Question

  • Enter your research question into the chat-based interface.

  • Chat with Undermind to refine your query before searching for literature

Step 2: Reflect on the Results

  • How do the results compare to the results of your keyword search strategy?

  • Are the articles high-quality and/or peer-reviewed?

  • Do the summaries seem reliable or overly generic?

Summary and Prompts#

Task

Standard Practice

AI Assist

Tools

Example Prompts

Find additional literature

Use citations from key articles to find more references

Use literature mapping tools to explore key references and citations further

Academic tools, e.g. Research Rabbit, Connected Papers

Upload or find a relevant paper and explore the citation trail

Use semantic searching to find additional papers

Academic tools, e.g. Elicit, Semantic Scholar

Enter your research question and explore the literature found (potentially doing a citation search)

Use agent-based searching to find additional papers in an extensive and detailed search

Academic tools, e.g. Undermind.ai

Enter your research question and explore the literature found

References#

Contents