Trace where a paper came from and where it led
Start from one paper, branch back to the work it built on and forward to the work that cited it, and read the whole lineage on one map.
Explore the reference map
A citation map built around Kahneman and Tversky's prospect theory, zoomed out to show the whole lineage. Zoom in on any paper to read what it argued.
How it works
Enter a seed paper and ask the model for its citation context. Each direction you follow gets its own branch, and every question under a branch keeps that line of influence in context.
-
Start from a seed paper. Name a paper, a finding or a theory. Ask which works it built on, which papers came out alongside it, and which major works cite it.
-
Branch back to its sources. Under the backward branch, ask about the assumptions, methods and debates the paper drew on. Follow any source further back with another follow-up.
-
Branch forward to its influence. Under the forward branch, ask how later work extended, applied or challenged the idea. Split it by decade if the field changed over time.
-
Follow the idea into other fields. Ask where it was picked up outside its home discipline, such as a psychology result used in economics. Give each crossing its own branch.
The model answers from what it learned in training, or from the web when it searches. Check every reference it names against Google Scholar or your library database before you cite it.
Why a map instead of a reference list
A reference list or a spreadsheet gives you an index of papers. A map shows how they group: which lines of work are still active and which went quiet, and where two lines met.
Zoom out and each node shortens to a phrase, then a keyword, so decades of work fit on one screen. Zoom back in on one paper for the detail.
Example
Say you are tracing where transformer models in NLP came from. Starting from “Attention Is All You Need” (2017), a backward branch leads to Bahdanau attention (2014) and sequence-to-sequence models (2014). Forward branches split into BERT-style masked models and GPT-style autoregressive models. A branch on scaling laws shows that line of work arrived in 2020, and you ask under it which earlier research it drew on.