Groop.us is a neat web service that is to visualize your social context and tags that you use. It will then try to group areas of interest and people that relate to you. The interface used is very interesting and has a great way of splitting the groups up into circles with tag coulds inside the circles. They do provice registration if you have received an activation code before. If not, you can view the demos on their site. The interface confused me a little at first, but lets take a look.
When in the demo (http://groop.us/users/kaiman, for example) you will first see circles on the left and center column. The left column is your personal tag cluster. The center shows related clusters of similar people matching your clusters of tags. You will now was to find one of your tag clusers that you are interested in and click on it. You will then see the middle column shift and select the most related cluster from other people. Also notice that you now will have a right column that lists links that you have added in your personal cluster that you have selected. Quick overview of your activity and who relates, but thats not all you can do.
If you focus back to the middle column, you will see next to all the bubbles is a link, “Get Recommendations.” The interface will then change around again and change the left column to blocks of tags containing your foci and personal clusters. This is pretty much just a summary of your tag activity in relation with others again. You can select a box from in the column to show different results on this page. In the middle column, you will see the selected tag cloud for recommendations on the top and a groups area below this. What the groups area does is group people that are highly active but also similar to the selected tag cloud. Now, for the groups, you are supposed to be able to select a name, but only if the person allows their data to be publicized. So, you may only get an alert box saying that the person didn’t publicize.
The column that I find really useful is the “My Recommendations” column. You will see a list of links with a relevance rank of the left and the person from the group in the middle column, or the recommender if you will. Now, if you select a link, it will bring another area up with the title, the link, and some basic information like the date and tags used for that link. I really like the recommendations column because you only see links from the people that relate to you the most. This allows for a more fresh quality selection of links then just jumping in a pool of del.icio.us links.
The framework that they used in getting the recommendations uses four major steps that you can find in their master thesis. The first step is “Structuring of Individual Resources” which is grouping up your data so you can find matching clusters from other groups of people. The second step is”Mapping Individual’s Interests to Foci.” What this will do is take your data and match it with others to create separate foci groups of matching people with their tags and resources. Third is the “Building a Social Network.” The thesis states this step as, “groups combine people who are most similar (in terms of their tags), interact most frequently (in a positive manner) and share one or more foci. The latter – also called multiplexity – indicates a shared context based on the assumption that people who share several interests are likely to have a common understanding in one area of interest.” Basically, it is the grouping of groups to get the most relevant recommendations for the next step, “Recommendations and Feedback.” The recommendations are focused on being automatic for the user while the feedback is manual. The recommendations are passed to each person from his closest neighbors in the focus to bring you the most related resources. I have not however seen the Feedback functionality, and I am assuming you will see this when you can get logged into Groop.us. From what it sounds, the feedback step will allow you to rate and send feedback about a recommended resource or article. This is to help produce more accurate recommendations for you.
This is by far the most complex visualization I have seen yet. It is very interesting to see how your tags are group with other people and then grouped with other groups of people to create the most relevant recommendations possible. It does seem to work rather well and I am very excited to see its final state. Great work Groop.us team, this looks amazing.