Hanan Samet discusses the NewsStand map-query interface in this video accompaniment to "Reading News with Maps by Exploiting Spatial Synonyms," a Contributed Article in the October 2014 Communications of the ACM.
00:00-00:18 VOICEOVER: In the beginning was the news. And it was good. And the news multiplied... and multiplied... and multiplied. Drowning itself out in riotous cacophony.
00:18-00:43 So researchers at the University of Maryland developed NewsStand to help readers cut through the noise and find news that matters to them. It collects at least 50,000 news items from over 10,000 RSS feeds every day; automatically tags them by location and other attributes; and presents them on a map-based interface.
00:43-00:55 Join us as we talk with Hanan Samet, the 2011 ACM Paris Kanellakis Theory and Practice Award winner, as he tells us about Reading News with Maps.
00:55-01:03 [Intro graphics/music]
01:03-01:26 While the Internet devastated the print newspaper, it gave us something in its place: The newsfeed. That's led to mashups, such as this colorful one at MSNBC. But map-based browsing has always required a bit more work.
01:26-01:46 DR. SAMET: In other words, if I'm looking for Los Angeles, in some application it could be a point, another one could be a boundary or a big area. Here, I just put Los Angeles. So it's aspects of inheritence or polymorphic types. The down side is that there's ambiguity. If I give you Dublin, is it Dublin Ohio, Dublin Ireland?
01:46-01:57 VOICEOVER: NewsStand uses established toponym recognition techniques to match names to locations. Then it lets visitors choose which "Dublin" they mean.
01:57-02:24 DR. SAMET: So when you're at a location, it shows you the sort of representative headline, what it thinks the most important in that location, and below that it has a little mini map. So the mini map is actually summarizing with the orange balls all the other locations that have some information about this topic. And the blue balls give you all the other locations with that name.
02:24-02:48 VOICEOVER: But location mapping is only part of how NewsStand interprets its sources. Visitors can filter by such features as language and country, while layers show where the media are talking about specific people, diseases, and brands. It also tries to determine what's most important to you, based on the news source, article topic, and your own location.
02:48-02:57 DR. SAMET: As you zoom out, you can't put in a million markers because you can't tell anything. So it has to make a decision what is important.
02:57-03:02 VOICEOVER: As its history grows, new uses of the data may emerge.
03:02-03:18 DR. SAMET: And one of the things that we're currently working on in introducing the temporal dimension. So you could look at news articles over a ten-year period, and see the spread of SARS as it's going from place to place by the dates in which it was mentioned.
03:18-03:22 VOICEOVER: Sometimes, those decisions may lead to unexpected insights.
03:22-03:35 DR. SAMET: It finds things what you're not looking for. It enables you, it's sort of serendipitous in a sense. So it's quite an interesting way to spend an afternoon.
03:35-03:40 VOICEOVER: But above it all, NewsStand is about location.
03:40-03:52 DR. HAMET: For many people location, in this business, location trumps content in a sense that they want to know if something's happening in a particular place because they have oil shipping from there or something.
03:52-04:00 VOICEOVER: Find out more in this month's Communications of the ACM, in the contributed article, "Reading News with Maps".
04:00-04:14 [Outro and credits]