We have created a live taxonomy of three metadata sets: Google’s top 10 searched keywords, the word content of corresponding blogs, and the top Flickr image whose tag matches the Google keyword. The information is arranged into a comparative taxonomy. For each topic, we used text mining to analyze the word content of the related blog article. The top 5 most common words in the corresponding articles are displayed structurally in order of appearance next to each keyword. Each Google keyword and a snippet of the article is juxtaposed with the Flickr image that has the same tag word (when corresponding images exist). Every item is linked to its original content: news articles, blogs, and images. Because it is a constantly updating process, the results will change each time sNews loads.

sNews displays a live meta-cultural archive of the most up to date human interests. It compares the standard organizational concept of taxonomy with the modified organizational concept of folksonomy. The taxonomical hierarchy in sNews is not created by any standard classification system. Instead, it is a hierarchy of terms created by the popularity of internet search phrases. The internet users are actually the ones classifying the keywords based on their collective popular interests at that moment. Both the keywords from Google and the tag words from Flickr integrate the concept of folksonomy into the project because they are terms mutually agreed upon by the internet users. The other element of sNews displays the "structure" of the content of the related news article. The text of each article is broken down and the five words that appear the most are shown in order of use. We omitted most of the unnecessary words hoping to reveal the relationship between the keyword and the article text, similar to the ideas of content mining or data mining. For example, the keyword "Obama Speech" resulted in "young » years » who » white » where" as the most frequently used words. Most of the technology used for content mining is extremely advanced. Our programming does not analyze the content, but we present the information in such a way that a meaning can be extracted from all of the elements of the project and their juxtaposition.

All of the programming for sNews was done in PHP which is a server-side scripting language that allows us to write HTML dynamically before it is sent to the client's browser. This allows us to grab data from other sources (Google Trends as well as the links contained there) and manipulate the data before displaying it to the user. Our process looks roughly like this:

1. We first get the latest Trends from Google (http://google.com/trends/hottrends?sa=X) and parse those into an array for reference later.
2. We then get the snippets of information as well as links to the containing sites from each Trend's Google page (http://google.com/trends/hottrends?q=trend+name+goes+here) and again store the snippets and links as arrays for later reference.
3. Next, we pass off the Trends values to Flickr through a PHP modification of the Flickr API.

4. Finally, we get the HTML from the link Google pointed us to and perform a word count on the HTML after stripping all of the HTML and PHP tags from it.

The "scraping" of these websites is accomplished through PHP's "file_get_contents" command which allows you to get the entire contents of a website (or file on a server) as a string. The ability to grab entire web pages as strings proved extremely useful when trying to count / manipulate the contents of that string.

If you have any questions, comments, leers, or jeers please feel free to email me.