Tools for Literature Mapping

In the ever-expanding web of academic literature, exploring the research landscape and searching for publications can often feel complicated, overwhelming, and repetitive. A common approach is to start with a Google Scholar keyword search or a subject-specific database, using filters to narrow down results. However, even with these refinements, it frequently requires trial and error to identify the most relevant papers.

Over the past few years, tools have emerged that aim to streamline this process. By leveraging existing scholarly metadata, these tools generate literature maps (or network graphs) that visually connect related publications. This provides a snapshot of a particular topic, theme, or field, typically stemming from one or more ‘seed papers’. While these tools do not replace ‘traditional’ literature searches, they can complement them by using seed papers as starting points to locate other relevant works.

In this post, I test out three such tools—Connected Papers, Litmaps, and Research Rabbit to get a feel for their potential in mapping academic literature. This won’t be an in-depth guide to using each tool, instead I’m interested in learning more about how these can aid researchers in navigating and organising their reading more efficiently.

Connected Papers

In a nutshell, Connected Papers creates a graph-based network that clusters similar works together, even if they don’t directly cite or reference each other. Instead of showing a citation tree, it uses data from Semantic Scholar to visualise academic relationships based on bibliographic coupling and co-citation analysis. This means you can quickly see groups of publications that share a strong thematic or methodological connection.

To begin, you enter a keyword or a work’s title, DOI, URL etc. This generates a list of potential publications from which an ‘origin paper’ (i.e., seed paper) is selected. Only one origin paper can be used. 

Connected Papers then generates a network graph, where:

• Node size represents the number of citations.
• Node colour indicates publication year (lighter = older, darker = more recent).
• Strong connecting lines show close relationships between publications.

Selecting a node reveals more details about a work—its abstract, citation count etc., and also provides links to Semantic Scholar and Google Scholar. The graph can also be expanded by adding new origin papers, helping to refine exploration. I found that adding three or four papers improved the relevance of results—it’s a shame that selecting multiple origin papers from the get-go isn’t an option as this would save some time. It does offer flexibility after initial generation, allowing unlimited addition or removal of origin papers. 

A network graph in Connected Papers. The node with a purple border is the origin paper.

The filtering options allow users to narrow results based on keywords, publication dates, or accessibility features such as open access status or PDF availability. This can be useful for visualising the number of works published each year, for example, or narrowing down the results if the number of nodes feels excessive or unmanageable.

There is also an option to view the data in a list format. By default this is organised by ‘Rating’ (i.e., the publication’s similarity to the origin paper). However, it can be organised by year, number of citations, author, and number of references. The list can be downloaded as a BibTex file for easy import into your preferred reference manager. 

Two noteworthy features are ‘Prior works’ and ‘Derivative works’. Prior works identifies publications most commonly cited by papers in your graph, often surfacing seminal works in the field. When one of these works is selected, any publication in your graph that references it highlights in blue, creating a visual map of influence. Derivative works functions similarly but in reverse, showing publications that have frequently cited papers in your graph. This can be used to identify emerging research directions, potential offshoots, and connected themes.

A network graph with prior works highlighted in blue.

Connected Papers is fast, has a simple user interface that is easy to grasp, and is great for getting a quick snapshot of a research area. While testing it, I gained a clearer understanding of related topics and spotted a few connections that hadn’t been obvious to me through traditional keyword searches.

However, the free version is limited to five graphs per month (although all features are included), which was sufficient for my initial exploration but would require upgrading to the premium account (6 USD monthly at the time of writing) for more extensive research. 

Litmaps

Litmaps takes a different approach by focusing specifically on citation networks. Like Connected Papers, you begin by searching for publications by title, keyword, author, or DOI. However, Litmaps immediately differentiates itself by allowing users to select multiple works as origin points, offering greater flexibility in how you construct your initial visualisation.

The platform also provides alternative starting methods, including the option to upload BibTex, RIS, or PubMed files. Premium users (12.50 USD monthly at the time of writing) can link their Zotero library directly to Litmaps, streamlining the process of incorporating existing reference collections.

Litmaps generates a citation-based graph, where papers are positioned along an X/Y axis. By default:

• The X-axis represents publication year (older papers on the left, newer on the right).
• The Y-axis represents citation count (higher up means more citations).
• Node size reflects the number of citations.
• Hollow nodes are suggested literature, which have been referenced by the origin paper(s).

One of Litmaps’ strengths is its customizability. Users can modify the node size and X/Y axes to display different metrics, including connectivity, reference count, and momentum. 

A network graph in Litmaps. The green nodes are seed papers, the hollow nodes are suggested papers.

As with Connected Papers, selecting a node displays information about the publication, including its abstract, citation count, number of references, and DOI. You can expand the graph by using the ‘Add to Litmap’ option, which then generates additional nodes and connections based on your selection. Litmaps promotes iterative searching, so as you add more works to the graph, you can run the explore again to find new connections.

The tagging functionality allows users to categorise publications by subtopics or custom metrics, adding an organisational layer that can also help identify trends or themes within the graph. 

Beyond visualisation, Litmaps offers a list view that displays all the publications in your graph, which can be downloaded as a BibTex file. A convenient feature is its public sharing option, which generates a unique link to your graph—ideal for collaborative research or presentations. Alternatively, you can download your visualisation as a static figure.

Some of Litmaps features are restricted to premium accounts, including advanced filtering options for publication date, keywords, and journals. The free account limits users to two network graphs at any given time and monthly email notifications about relevant publications, while the premium subscription offers unlimited graphs and weekly notifications. Suggested papers are also capped at 20 at a time with a free account (you can run the explore again and new suggestions will be shown), but I found this to be plenty. 

Despite some limitations, Litmaps is a powerful platform that allows users to easily create and customise citation networks. Personally, I found Litmaps the most helpful for locating works that were directly linked to the topic I’m researching – perhaps this is due to the fact that it draws on data providers beyond Semantic Scholar, claiming to have the largest database size among leading research platforms.

Research Rabbit

If Connected Papers is about snapshots and Litmaps is about tracking citations, Research Rabbit is about endless exploration. It stands out for its comprehensive approach to publication discovery and its completely free access model. At first glance, Research Rabbit’s interface may appear more complex and chaotic than the other tools, with multiple cascading pop-ups offering various options. Don’t let this put you off!

The platform organises research into collections (themes) and categories (folders), allowing for hierarchical organisation. Publications within a collection can be added by searching for a keyword, title, DOI etc., through uploading a BibTex or RIS file, or connecting to your Zotero library—any updates in Zotero then sync automatically with Research Rabbit and vice versa. 

When you select a work from your collection, a pop-up provides an abstract, DOI, link to PDF etc. A second pop-up displays engagement options, including exploring similar works, viewing more by the same author(s), and searching related content (in my experience these are mostly Wikipedia articles). Each of these options opens another pop-up, so it can feel as though one is descending down a rabbit hole. Fortunately, you can navigate back and forth through the pop-ups via a scrollbar to retrace your steps.

A selection of pop-up menus in Research Rabbit.

Choosing one of the ‘Explore Paper’ or ‘Explore People’ options generates an interactive and expandable graph:

• Node size reflects the number of citations.
• Node colour indicates if the work is already in your collection (green) or a suggested work (blue). Suggested works can then be added to your graph and collection.
• Nodes can be moved around (drag and drop), which might be useful for visualising or re-structuring data.

A network graph in Research Rabbit.

The works in the graph are also available in a list view, sorted by relevance. Users can reorder them by date, citation count, or alphabetical order and filter them by keyword(s). The list can be exported in multiple formats, including BibTeX, RIS, and CSV. A ‘Shared with me’ section allows for collaborative work, and like Litmaps, there is an option to receive email updates on collections to stay informed about new publications.

One of Research Rabbit’s standout features is its timeline view, which shows connections between papers chronologically. This is handy for visualising how research has evolved over time. Like Connected Papers, it also allows you to explore earlier and later works related to your selected publications, creating an increasingly detailed graph of connections. This makes it particularly valuable for exhaustive literature searches where breadth is important. 

The timeline feature in Research Rabbit.

The detailed level of exploration this enables surpasses other tools, but it comes with a risk—you could easily lose days to playing with all its features and lose track of what you’re doing and why. I spent an enjoyable afternoon browsing various connections, only to realise I had drifted into a different, albeit loosely-related, topic. Although this yielded some interesting observations, it didn’t align with my original aim or support my work. For this reason, I feel Research Rabbit is most effective when one has a firm focus or plan of action in terms of what they’re searching for. 

The interface can feel overwhelming, with a steeper learning curve than Connected Papers or Litmaps—I recommend this guide to get you started. However, for researchers seeking comprehensive coverage, Research Rabbit’s power and flexibility make it worth the initial learning investment. As a Zotero user, I find the integration a brilliant add-on that streamlines my workflow. The fact that all of this is free—and will remain so—is impressive.

Conclusion

Each tool lends itself to different aims of a literature review process and/or different needs. Perhaps rather than viewing them as competitors, we might benefit from using them in complementary ways, with the choice of tool depending on preferred visualisations, integration with existing workflows, and research focus. 

I think it’s important to note here that all three rely on Semantic Scholar’s metadata, which has its limitations. Coverage is significantly stronger in the sciences than in the humanities, so publications are missing. Additionally, they focus more on journal articles, leaving other sources such as books and blogs underrepresented. Metadata handling can also be inconsistent, especially for specialised journals or publications in non-English languages. For example, for some publications in Études mongoles et sibériennes, centrasiatiques et tibétaines, the journal name is listed in place of article titles. While author information and publication details are still provided, allowing users to eventually track down the correct information, this creates unnecessary work—especially if multiple entries have this issue.

Although not perfect, literature mapping tools can make the process of identifying, organising, and visualising literature more efficient, and certainly helped me piece together a more comprehensive understanding of existing work on my topic.