.png%3Ftable%3Dblock%26id%3D5084f370-3565-4b5b-9037-eae55ff6e1f1%26cache%3Dv2&optimizer=image&quality=80&width=280)
Do not index
Do not index
Conducting a review of pre-existing literature relies on finding seminal works, i.e. foundational research that is heavily cited across a subfield. When producing new research within that field, it’s important to know which seminal works should be cited to provide sufficient background context.
Finding these works can be painstaking and time consuming. To make matters worse, doing an incomplete job can result in a rejection when looking to publish a new paper or file new intellectual property.
At Epsilon, we’re looking to make that process easier by launching Key Texts. Here’s how it works.
Key Texts
From Epsilon’s main page, you can click on “Find Papers” in the left side bar and type in any search term. From here, you can search for any topic and Epsilon will return search results like this:
Epsilon groups papers into different categories. At the top is “Latest Research” which highlights some of the latest works related to the search topic. Beneath that is “Key Texts” which highlights the seminal works. Here I can see the first result is the paper “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks”, which was the first paper in the AI field to introduce the concept of semantic search using LLMs.
To identify key texts, Epsilon relies on multiple pieces of data including overall citation count, influential citation count, and other citation graph based signals. We’re also continually adding new signals to make the algorithms smarter and more helpful.
Citation Exploration
From the search results above, we can dive deeper into the citation graph by clicking the dropdown button on the left of the paper. For example, I can expand the paper mentioned above to see all the references and citations. References refers to other works that this paper has cited whereas citations are papers that have cited this work in their research.
By navigating this tree, I can start to built a mental model of the citation tree that was used to build this subfield of research. This is great when starting a new literature review, conducting prior art research, or catching up on a new field. I can then save any of these papers to one of my libraries using the ‘+’ dropdown on the right side of the paper to run search queries and other analysis on that document. (note: this is only available for papers that have a free, public PDF).
Feedback
We’re continually improving this functionality, so reach out to us at hello@epsilon-ai.com to let us know what you think!
