Laniakea

With the overwhelming amount of digital information available to us, most people rely on a search function to find what they need in large document collections. But searches only reveal results directly tied to the terms you’ve entered, missing the surrounding context of related ideas that deepen your understanding of a topic.

With Laniakea, we’ve focused on the experience of exploring a massive amount of information. Through a combination of algorithmic mapping of content, topic modeling, and refined interaction design, Laniakea allows users to navigate an archive in an intelligent and intentional way. The first layer of Laniakea maps out each document or article based on its content, creating a cartographic view of the information. The second layer uses topic modeling and color to reveal additional relationships between areas of the collection.

Instead of relying on algorithms and computation to define and surface everything, we’ve worked on creating a fluid interface to empower users to do what humans do best &endash; draw inferences and explore connections. No area is “titled,” but instead a list of key terms suggests what topics might form that region of the archive. Hovering over a key term reveals where that term is found throughout the entire landscape, allowing users to preview where they are headed before diving deeper.

While direct lines of connection are not drawn, users can easily find their own relationships. For example, in looking at Wikipedia, an interesting relationship between politics, crime, and business can be found – is this a result of how these topics are all written about? Or does this reveal something about how our society functions?

Laniakea is available online for three datasets – Wikipedia’s top articles, PubMed, and all the Wikipedia articles containing the word ‘ghost’ (a Halloween special).

If you or your organization is interested in a version of Laniakea to run on your own document sets, please contact us at inquire@fathom.info.