Connected Papers: Literature Maps, Free Access & Limits

What Connected Papers does

Connected Papers helps researchers explore papers through a graph of similarity. Its documented method uses co-citation and bibliographic coupling, drawing on Semantic Scholar’s corpus. This profile covers literature discovery: a nearby paper is a candidate to read, not an automatically verified source for your argument.

Where it fits, and where it falls short

A suggested Connected Papers workflow

1. 1

   Define the research question and choose a paper whose population, topic or method genuinely fits it.

2. 2

   Build a similarity graph and keep the origin’s identifier with your search notes.

3. 3

   Inspect candidate papers and prior or derivative works; record why each belongs or does not belong.

4. 4

   Read the relevant passages and carry verified references and limitations into your own evidence record.

Read a similarity edge as a lead to investigate

The About page explicitly distinguishes Connected Papers from a citation tree. Co-citation concerns papers cited together by other work; bibliographic coupling concerns shared references. These relationships can connect papers even when neither directly cites the other. Proximity therefore answers a different question from “did this paper cite that paper?”

When a connection matters to an essay, open the relevant publications. Inspect whether they investigate comparable populations, methods and outcomes, and locate the passage behind any claim. A closely related topic can still contain a disagreement, an inappropriate comparison or a result that does not address your question. Treat the graph as a reading route rather than replace that inspection with its layout.

Supporting documentation: [1]

Use prior and derivative works to choose a reading direction

The provider describes Prior Works as common ancestral papers of the graph’s papers, and Derivative Works as common descendants. Its homepage presents these as routes toward background work and later literature, including reviews. They are directions from a selected neighborhood, not a chronological list of everything published on a topic.

Use the direction to decide what to inspect next. A prior work might explain an instrument or concept that several candidates use. A later review might supply terminology and further references. Neither role establishes that the paper is methodologically strong or supports your planned conclusion. Keep the reason you followed it, then evaluate what it contributes after reading.

Supporting documentation: [2] [4]

Refine the neighborhood with another origin

The team’s multi-origin explanation describes adding another paper as an origin to construct a graph selected by similarity to the origins. The current pricing page still lists multi-origin graphs among the included features. This is a way to change the discovery question, not simply combine two complete bibliographies.

Choose the second origin for an explicit reason: a different method, a relevant population or an alternative account of the same issue. Save both identifiers and the reason for changing the set. If interesting candidates disappear or new ones appear, inspect their fit rather than interpret the changed layout as evidence that a study became better or worse. More origins do not, by themselves, establish more complete coverage.

Supporting documentation: [3] [5]

A suggested check: apply the same inclusion rule after following a link

Start with a question such as how short-answer practice relates to delayed recall in university courses. Define the population, task and assessment timing you need before choosing an origin. This is a proposed research exercise, not a Connected Papers search we performed or an accuracy benchmark.

Consider two entirely fictional candidates for practising the decision. Candidate A examines university students and a delayed assessment; Candidate B examines school pupils and an immediate assessment. No papers, graph positions or findings are supplied as real results. Even if two such papers were nearby in a graph, that would not make their populations and outcomes interchangeable.

For an actual evaluation, open each candidate’s methods and record the relevant population, practice task, comparison and assessment timing. Keep a precise source locator and an inclusion reason. A paper outside the immediate scope may still be useful background, but label that role instead of quietly adding it to the evidence for your narrower claim.

Then compare the graph-derived reading list with the terms and databases appropriate to the assignment. Preserve what you searched, which origins you chose and what you excluded. A useful map can improve your next reading decision without establishing that the literature search is exhaustive. Your conclusion should reflect the sources actually inspected.

Keep a recoverable research record outside the graph

The current feature list includes saved papers and graph history. Keep your own identifiers, relevance notes and passage locations as well. The terms say users are responsible for their own backup copies; a saved list in a service should not be your only record of a dissertation’s source decisions.

The privacy page describes collecting information entered into text boxes, including search queries, as well as saved-paper lists and third-party usage analytics. It also describes provider processing for service functions. Use permitted research information and inspect the current sharing and account controls before putting sensitive project details into a query. This guide does not establish private-link permissions, deletion timing or an institutional confidentiality approval.

For a handoff, first identify the actual artifact you need: a recoverable reference, a reading note or a record of why a paper was included. Verify any file you obtain against that need. We do not claim to have tested an export or established every supported export format. A standard reference manager and an evidence matrix can hold the checked record after discovery.

Supporting documentation: [3] [6] [7]

Choose a graph, citation context or structured extraction

Connected Papers fits a question about related literature around an origin. ResearchRabbit and Litmaps are other discovery workflows to compare; inspect their current collection, monitoring and reference-manager handoffs when those are the bottleneck. Scite addresses citation context around a reference, while Elicit can help organize structured extraction. These are task distinctions, not results of a head-to-head test.

Before subscribing, use the free graph allowance to investigate a defined reading task. Academic and Business plans cover different use contexts, and the terms describe annual or quarterly commitments with automatic renewal. Cancelling renewal and obtaining a refund are separate questions to check in the current terms. Choose the relevant plan and billing period for your actual use rather than infer an unrestricted monthly subscription from a displayed monthly equivalent.

Supporting documentation: [3] [7]

Pricing, free access and plan checks

The public Free plan includes five graphs per month and lists all features. Academic and Business Premium plans list unlimited graphs, with annual or quarterly billing. Confirm the total for the selected period and current renewal terms before paying.

Confirm billing period, renewal terms, limits and regional availability on the provider’s current plan page before paying.

Sources and research notes

This guide synthesizes official product information with our editorial assessment of academic use. Suggested exercises are examples you can repeat; they are not results of tests we performed. We do not infer accuracy from a provider’s marketing claims.

1. Connected Papers: how similarity and the underlying corpus work
2. Connected Papers: discovery uses and prior/derivative works
3. Connected Papers: current plans and included features
4. Connected Papers team: original graph and prior/derivative explanation
5. Connected Papers team: multi-origin graphs announcement
6. Connected Papers: privacy policy
7. Connected Papers: terms, account records and subscriptions

Found a changed feature or a factual error? Send the exact page and an official source to info@ai-essay.com. Read our editorial policy for how we handle corrections.