miska knapek
people map
The People Map is a Aalto University project to build an online visual information tool to primarily help the university’s researchers, find other university staff of relevant knowledge. En route to making such a tool, this project addresses the increasingly important topic of being able to overview, navigate and find digital entities in a world of increasingly populated by entities described between genres. With more online, the need for multidimensional overview, navigation and search is only more needed. Otherwise, peoples’ ability to find and see things far from main established genres, that we currently navigate by, will continue to be rather limited.
why is holistic/hybrid search important?
In a very specific sense, the project addresses the need for researchers at the Aalto University to find relevant collaboration partners, when the traditional way of categorising people ( in single categories such as departments ) makes it very difficult for people of similar interests, but in different general categories, to find one another. This project allows people to overview and find people of specific hybrid interests. ( And it also solves similar issues in regards to being able to overview research interests, generally, at the Aalto University, as might be relevant for various policy makers ).

The general project idea – of finding people of relevant hybrid competencies – is relevant to all sorts of organisations where people might have hybrid interests and it being useful to find and use people of hybrid interests – such that people get to have multiple roles in an organisation, rather than merely a single one.

In a general sense, the project addresses some rather important overviewing, searching and finding questions, that are growing increasingly important. The traditional ways we have of navigating lots of information – typically one dimensional search engine listings – can’t handle and do injustice to increasing amount of information on the net.

While this might sound a little abstract, one can say that most the interesting things on the net are described across categories, hybridly. As a case in point, one can look at cultural production, such as songs, writing, performance art of various kinds, etc… Back in the days when there were few alternatives to chose from, it was easy to overview them using traditional methods. Now when the net allows many more a platform for the cultural doings, and with the net increasingly being filled with many new cultural artefacts, there’s a significant risk many practitioners’ works will be sidelined thanks to the traditional ways of navigating these things only has the ability to overview very few things. (I’d jokingly call it a tyranny of one dimensional lists… ).

Using something like the self organising map interface developed in this project, allows easier access to hybridly described artefacts, such as cultural artefacts, or just many of the things increasingly populating the net. Putting machine learning based interfaces ‘into the wild’ is something yet to happen properly.
interface details
term map
The term map is an interface for people map searches, on the right. The terms are the aggregated research terms extracted from researchers' profiles. The people map arranges and develops an overview of the people/researchers engaged in these terms, according to the importance given the terms in the terms map. The weight can be changed by pressing on desired terms.

If one wants to find researchers interested in information visualization and health, for instance, one would increase the importance of these terms, and the people map would organize to show an overview of the people interested in these terms, and the different/similar ways they're interested in the same research terms.

Technically speaking, the term map is a (self organized map) automatically generated by datamining people.aalto.fi researcher profiles. Wikipedia linguistic data is used to generate the layout of the terms, as a (self organised map) a semantically related layout of terms, to help users overview and find relevant terms.
people map
The people map shows the which researchers are similar to one another, given the importance given different terms, on the term map to the right. To help viewers see what researchers are interested in, in different regions of the people map, a grid-like term cloud view of the most active terms in particular regions, can be overlaid the people layout.

Researchers' positions are based on a self organizing map algorithm, bringing similar items together, thus naturally forming continuums of similarity. By arranging similarly interested researchers together, the user can see the different categories of ways researchers can investigate particular topics in, and better get an overview of research interests and relevant researchers.
future developments
The sensemaking, overview and matchmaking of researcher interests that the People map concept allows, can also be a timely help for the otherwise large amounts of unordered and unrelated material increasingly populating the internet.

For example, if someone gives you a collection of two thousand, or just twenty unknown musicians/poems/articles/or other unfamiliar digital cultural matter, how does one go through it, other than at arbitrary length? Combining the People map with existing working music/text/linguistic similarity analysis algorithms one can get an ordered overview of what's all where. From having a random collection of matter, one gets a map of the territory and an informed route through the territory.( Excitingly, as the self organizing map algorithm groups similar items together, its analysis also produces genres automatically, without people needing to specify what a relevant genres might be. Thus one can discover genres no-one knew existed.)

As people increasingly put more and more of their thoughts, doings and work, the net facilitates an increasingly democratic access to cultural publishing and the enjoyment of all the world's corners and cultures. However, this democratic access is mostly stopped in its tracks by today's search and overviewing facilities. It's simply not easy to find anything but the big names and successes. Today's search engines, reducing all the world's matter and complexity into one-dimensional lists of more-to-less relevant, reduce the world's richness and variety, removing variety, overview and access to all. Machine learning based data analysis tools, like self organizing maps, and new multidimensional search interfaces, return the world's richness, make it accessible, and better (re)present the vast culture producing world that we have.

team & collaborators

Miska (Michael) Knapek and Timo Honkela

Supporting institutions:

Aalto University / Information and Computer Science
Aalto University / Media Factory

And not to forget... Kitchen + Co, Klaus K hotel/Livingroom, Cafe Carousel, Bulevardin Kahvilasalonki, Korjaamo Bar + Cafe, Cafe Strindberg, Johtu Cafe, Kaivapuisto