Slashdot Mirror
Visualizing Stories On Current Events With Newsmap
hrbrmstr writes "Marcos Weskamp and Dan Albritton have created Newsmap, an extremely cool way of visualizing news stories. The site takes the aggregated content from Google News (globally) and maps it out into a visual space. That way, you get an immediate feel for news patterns (what the media in any particular region is gravitating to) - there's quite a bit of potential here."
While this is really cool, it is ofcourse a snapshot of current state of affairs: how many times is a certain news item highlighted.
The very small items could however be interesting too:
Take for example a small accident that gets catched on by more and more news companies as time goes on, simply because it is found out that an important person was involved.
Thus, 'small' news items that have a 'high rate of increase' across various sites should be voted more important than static ones.
For simplicity sake, perhaps this could be done visually (simply animate the news from a certain point in time forward to the now, and you see developments more clearly).
This thing is certainly an eye-opener however, applauds to the designer.
Slashdot: stuff for news, nerds that matter, matter for news, stuff that nerd
I worked on something similar once. Used a fairly obscure mathematical technique known as formal concept analysis to group objects (which were documents in this system) according to their relationships to each other. Each document would be defined in terms of keywords (which were generated using an automatic text summarisation tool) and the system would output a graphical representation of the relationships between them, with the most general documents at one end, and the most common topics at the other. If you want to find out more, look up formal concept analysis and it should be fairly apparent how it would've worked.
I extended the basic technique, which usually produces "concept lattices", to have different strengths of links, and made it all work in an oh-so-1998 3D environment. There was a load of potential in it, but it was just something I knocked up in a summer placement and I don't think anything major was done with it.
It's not new : see swissmap.ch
The idea is excellent and the implementation works well (for a beta). But I can't see why the programmer used flash when it seems like a tool that could be done so easily (and in a bandwidth-friendly fashion) using colored HTML tables.
---- scrm
Another area that could benefit from it is Google Zeitgeist
I can't see anyone having mentioned this, and I don't know if it has featured before on /. but:
0 20 01.jpg
...
http://www.cs.umd.edu/hcil/treemap-history/all1
show a 'treemap' of usenet. it's kind of inevitable that 'sex' and 'erotica' should be so large
Hmmm... Interesting. France and Canada have a similar ratio of national and international stories to the US!
The German Google news has a whopping huge ratio of entertainment to news!
However, India and Australia are WAY low on National news! (Even the UK despite the highest proportion of international news has more national news)
That's all folks, I'm sick of waiting 5 hours for each page to load up (even if it is subst minutes hours)...
-- *~()____) This message will self-destruct in 5 seconds...
This is a great technology for those studying media and culture. It reminds me a bit of the HP technology that tracks the spread of stories on web logs. What would be interesting is a combination of the following:
- the information and utility of Newsmap.
- the tracking of the HP blog project.
- the ability to track the author, source, and parent company of each article.
It is interesting to see how much press a given story is picking up, but it is even more interesting to track what media giant is publishing that story in as many of its subsidiaries as it can. This would allow people to see just how much control each conglomerate has over what news the public is allowed to consume. By the same token, what stories are seeing the least coverage? What potentially important news is being "obscured by shit"? Who publishes the news first? What companies merely follow stories that others have already broken?------- "One of the joys of travel is visiting new towns and meeting new people." -- G. KHAN
It is a cool idea, but an identical service, called I believe "Newsmaps" was operating in 1999-2000, so its not such a new concept. The old Newsmaps service presented the data as a hilly island, with bigger stories generating tall peaks. There were separate islands for different types of news - breaking stories, business news, entertainment drivel, etc. I am not sure why it failed, but perhaps not having Google around to do all the hard work of data collection had something to do with it.
Interesting problem he has with choosing to emphasize the font size vs. the size of the story's block. Two stories with equal-sized blocks have different fonts based on the size of their headline. Being used to newspapers, I tend to think the headline with larger font is more important story. I think he is right to go with block-size as the indicator of a story's prominence in the media. Think of the opposite approach: A story with a one-word headline, but a huge font ("War") would have the same-sized block as a story with a multiple-word headline that was less important. I think that would result in a more confusing visual metaphor.
So, I think the programmer had a difficult design choice, but made the right decision. In order to use this effectively, I have to retrain my eye to judge importance according to the amount of real-estate being taken up, not by the size of the font.
I'm bad at filing stuff in any sort of systematic
Me, too.
I've tried to clean up my top level home directory so that there's only a screenfull of concise subdirectories listed, then everything goes into those.
Problem is, some of those subdirectories become chock full at the next level. I have a directory called "tmp/src" that includes about every imagined release of some interesting application tarball ever made.
Then, documents can hide way down in some particular project directory.
Instead of a static view of my files and work, I'd like VFolders that could be generated a lot like Google Searches, including criterion such as file type, time last accessed, keywords in the document.
I remember reading once of some crazy guy that used CVS for his home directory, but I think CVS is too clunky. But he had gem of an idea: time travel - "I want to see my desktop from 8 months ago".
And, yes, while a graphical tree is really nice, I'd like to be able to navigate through any tree using pure text-based tools, terminals if I desired.
Maybe then I could make my own "/usr/bin" sane instead of what my sysadmin thinks is a good idea.
"Provided by the management for your protection."
Newsmap is based on Treemaps, which is both a conceptual GUI idiom as well as a commercial product. This is the work that Ben Schneiderman is most well-known for, and he's been working on different forms of interactive information visualization for decades.
The parent was asking about projects like KDE and Gnome picking up cool concepts like this. The HCI world is full of 'hey neat' ideas that on the surface really seem like they should be brought into the fold, but aren't for a variety of reasons. One company in particular that I worked for (and won't name) has a really cool project that I feel could become a standard UI idiom like radio buttons and scrollpanes, but the product is doomed to failure because the company is horribly mismanaged and (having been the sole coder--as an intern, even) I also know the code to be completely inflexibly designed. Furthermore, they want to make all sorts of money on the thing, which means they're charging customers an arm and a leg to use it.
The Linux desktop environment projects have issues equally as inibitive as the one described above, but rather than being financially oriented, their problems are more about ego and (with the exception of some of the KDE guys) a complete misunderstanding of what HCI is all about. I really wish KDE/Gnome would use these experimental UI metaphors, but alas, I think their structures prohibit this sort of thing.