Silobreaker Experiment

By kfreelskfreels (1261532767|%a, %b %e at %I:%M%p)

In this blog post, I'll be examining how Silobreaker, a news analysis tool, works. Since I finally picked my term project topic (fast food industry - hooray!), I'll be delving into all kinds of fast food industry news.

silobreaker.png

Silobreaker Overview

Silobreaker is a site that helps the user find news articles and analyze them. On the home page, headline articles are listed along with the date they were first posted, the last time they were updated, how many documents are available, its category, and links to related entities. Silobreaker also has some category links at the top of the home page. These include global issues, technology, science, business, energy, and the world. So it might not be the best site to find news articles about John & Kate + 8 developments…

When I click on one of the headlining stories, I'm taken to a Silobreaker page dedicated to that story that provides a summary of the story, what others have reported on it, quotes regarding it, related stories, and many more tools that I will get into when I do my search experiment. The bottom line is: Silobreaker is not just a website that coughs up links to articles when you do a news search, it uses textual analysis to categorize stories, summarize the internet buzz about those stories, and figure out how various stories are interconnected.

Now, onto the experiment

mcdonalds.png

I began with a generic search for [mcdonalds]. Silobreaker automatically asked me if I meant [McDonald's Corporation (Company)], and thank goodness it did, because the results for my [mcdonalds] search were less than impressive. Once I clicked on their suggested search, my results improved. At the top of results, Silobreaker had a section with the McDonald's logo and the type, name, nationality, and fact sheet about the organization (see right). This would be helpful when researching a company you don't know a lot about. Instead of scanning through various search results, you could get the current, need-to-know info right up front. Now I will dive a little deeper into my analysis, and tell you about the unique features you can find on Silobreaker

Unique Features of Silobreaker & How they Helped my News Search!

Mouse-Over Information

In Silobreaker, you don't even have to click on anything to get in-depth information! When I mouse-over the title of any search result, a box pops up that shows a longer summary of the article, in-text content, and more documents about that story. This is a feature not available in other search engines, and it's useful because the user can quickly find out more about the search result right on that page. Below you will see what popped up when I moused over the article called "Why the Fat Police of South LA are Failing". The most important thing that the mouse-over showed me was the main point of the article in the "In Context" box. Now, I don't even need to read the full article - I know exactly what it was talking about because of the mouse-over function.

mouse.png
Content Volume

Many of the search tools we've learned so far this semester do offer a tool that shows the popularity of a specific topic over a period of time. Silobreaker "Content Volume" feature not only shows how much stuff was posted about the topic over a period of time, it also breaks it down into whether or not that content was news articles, blogs, or audio/video. The one thing that did disappoint me about this feature was that it isn't interactive at all. For instance, I can't change the period of time it covers, or see any information about why the content peaked on a particular day. I can't even enlarge the graph! So all it really told me was that for some reason, between September 6th and 13th, there was a peak of content volume about McDonald's.

volume.png
Network

The Network tool is probably the coolest tool that Silobreaker offers. It shows the relationships that McDonald's has with other entities in the news. In the image below, you can see that the network connects McDonald's with companies (i.e. Burger King), organizations (i.e. NYSE), locations (i.e. Kuala Lumpur), and keyphrases (i.e. Food Chain). You can also see that I have the ability to select the period of time the network covers, and I can adjust the inputs to the network. So if I only wanted to see the companies that McDonald's shares a network with, I could do so.

Picture1.png

The best thing is, I can mouse-over any name in the network to get snapshot information about it. Here's what comes up when I mouse-over Burger King Holdings, Inc:

bk.png

Additionally, I can hover my mouse over any connection in the network to see the article that corresponds to that link. For example, when I mouse-over the midpoint between Taxes and McDonald's, a box pops up that shows documents and stories indicating a relationship between McDonald's Corporation and Taxes, and also allows me to search further is this particular connection is of interest to me. I believe that this is the most useful tool that Silobreaker offers because it allows the user to see a big picture explanation of the most current McDonald's news stories and how they are linked to other entities. Since my goal for this semester is to research the Fast Food industry, I will keep checking up with Silobreaker's Network tool to see how it evolves.

Trends

Another very neat tool that Silobreaker provides is "Trends". While a few of the blog search tools I analyzed in my last blog post had cool trending features, I would have to argue that Silobreaker's "Trends" feature is THE coolest (and I should mention, about news, obviously). The image below shows the relative share % of McDonald's news compared to other entities. The grey bar graph represents the total news article volume for all the entities combined. The very cool thing is that I can choose whatever, and however many entities I want to compare McDonald's to. Silobreaker even gives me suggestions from various categories like company, keyphrase, and city. Additionally, I can select different lengths of time for the graph to depict (4 weeks, 3 months..etc.), AND can also manually insert a range of time I want to see.

Since the Content Volume feature told me that there was a peak in articles about McDonald's between September 6th and 13th, I decided to see how Mickey-D's stacked up against some other companies during that time period. Sad face - it didn't work. All I got was a grey bar graph and no indication of what % share McDonald's had. So my conclusion is that the Trends feature is COOL, but you need to give it a long enough range so that it has something to work with.

trend.png
Blogs

Silobreaker really does a great job of analyzing news, not just spitting out lists of news. It tries to collect and decipher as much information about a particular news topic from all types of media. For instance, on the right hand side of any search results page, Silobreaker has a section called Blogs that lists blog posts about your search topic. The mouse-over feature works on these too, so you can quickly get the gist of the post without having to click on it. In addition, the Blogs section can be sorted by relevance or date, much like the blog search tools that I've previously analyzed. This was helpful during my search about McDonald's because not only was I getting news about the company, but I could see how people were reacting to that news.

blog.png
Conclusion

Silobreaker is a great news search tool. Its unique tools actually help the user analyze news, as opposed to just find it. Additionally, it's focused on legitimate, newsworthy entities which makes me trust the results more than I would any regular full news site. The most useful features are Network, Mouse-Over and Trends. Silobreaker could improve its Content Volume tool by making it more interactive. I did like that they included the Blogs tool with my search results but if I was really looking for relevant blogs, I would probably use a dedicated blog search site. Ultimately, Silobreaker is a site I will use again to learn more about the Fast Food industry!

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License