There’s a great article in the Globe today about Radian6: Starting at the top. Radian6 is a Fredricton NB startup that provides advanced blog monitoring services. Exvisu uses other data sources to do our primary analysis work, but Radian6 is a very nice compliment to what we do, and we’re happy to be able to use such a robust tool to help us provide ongoing services to our clients.

There are quite a few good resources to help you follow the online face of the current Canadian Federal election campaign, but probably the most comprehensive set of tools has been built by our friends* at the Infoscape Research Lab in Toronto, who have built a really interesting suite in partnership with the CBC’s Ormiston Online. My personal favourite is the Blog MediaBuzz widget.

* Zach Devereaux, one of Infoscape’s researchers, is a long-term friend of Exvisu’s. Zach is a top-notch researcher with whom we’ve worked on some really great projects.

With the US Political Convention season in full swing, we thought we’d present a short demonstration of how Exvisu’s blogosphere analysis can shed light on the discourse related to political campaigns, candidates, and issues. In mid-July, we performed an analysis of the English blogosphere related to the upcoming US Presidential election. We focused specifically on Free Trade and NAFTA, which became a key issue during Primary season.

The following map was generated using a dataset including all blog posts (there were 946 total posts after filtering for spam) for a structured series of queries between July 14 and July 18. We then performed a lexical analysis that identified and mapped the relationship between the top 150 words in the dataset.

Obama - McCain - NAFTA

Observations

• McCain and Obama are both located at the center of the map - which is to be expected, since these were used as keywords to establish the dataset. All of the words on the map are arranged by their relationship to their nearest 5 other nodes (words), and the relative size of each node is directly proportional to its resonance in the dataset.
• When performing such an analysis, words that appear on the map tend to self-organize into clusters that define concepts that are important. In this case, there were 6 significant conceptual clusters related to important events of that week: Gov. Mark Sanford’s gaffe in an interview on CNN; Michigan speeches by both Obama and McCain; McCain’s speech at the La Raza Convention; issues related to free trade and trade agreements; the “flip-flop” question, and general political and social issues that are at play during the election season.
• In terms of positives for McCain during that week, the strong cluster related to his speech at La Raza is particularly interesting. In this speech, McCain was quoted as saying that he’s an “unapologetic supporter” of free trade. “Unapologetic” and “supporter” are the words which connected this cluster with the general free trade cluster to its left (and this connection is largely responsible for their proximity on the map). That is not to say that this declaration is the only thing in play for McCain on this issue - as we read the blog posts that make up the free trade cluster, we noticed that the controversy surrounding McCain’s trip to Ottawa in June was still very evident in the text.
• The other negative cluster related to McCain was related to Sanford’s inability to distinguish specific ways that McCain’s economic policy proposals differed from those of Bush in an interview with Wolf Blitzer on CNN. One of the important innovations of this kind of lexical analysis is that it allows us to identify which stories resonate the most in a given period of time, and the fact that a single story related to a single on-air interview by a McCain surrogate even appears on this graph demonstrates the influence that even a small problem can have once it’s picked up in the blogs.
• It’s also interesting to note the other individuals who appear among the top 150 words on the map. Other than Obama and McCain, we also see the following people: Hillary Clinton, Bill Clinton, Phil Gramm, Austan Goolsbee, and Sanford.
  • Gramm, the former McCain Campaign co-chair, was in the news due to his assertion that America had “become a nation of whiners” following which he resigned his position in the campaign. His name is directly connected to the words “nation,” “political,” “economic,” and “advisor.”
  • Austan Goolsbee is Obama’s key economic advisor and appears because he is widely cited by bloggers that take a close look at Obama’s economic and trade policies.
  • The fact that both Clintons appear among the top 150 words in the map is an indication of the importance they still wield related to this year’s election cycle.
• The last cluster to note is related to the “flip-flop” question. This is a relatively small cluster but contains words directly related to Obama including “fisa” and “public finance.” At the time of this analysis, this question was still very much alive.

As the election draws closer we’ll perform at least one additional analysis of this kind that will allow us to see how these issues have developed since mid-July. Feel free to ask questions or note any additional trends in the comments.

Last month I was working with Zach Devereaux (you can read some of his academic work with the team at Ryerson’s Infoscape Media Lab on the blogosphere surounding the liberal leadership race), who pointed out an excellent blog post from the Oxford Internet Institute that characterises some of the problems we here at Exvisu have come across regularly with Google Blog search.

A more insidious bug occurs with the language preferences. Google Blog Search will return fewer results when the Google interface language that you have specified in your Google preferences doesn’t match the language you want to search in.

Yesterday, Michael and I came across a brand new type of bug at Google Blog search: searches starting on Jan 1 2008 until the current date were only returning results dated after July 15. Of course now that I am writing this post, this problem seems to have been rectified. Strange indeed. Did anything special happen on July 15?

Feel free to try to reproduce all of the other bugs described in the OII post, those at least are longstanding and repeatable.

I don’t know about you but these problems make me want to cheer for all of those underdog search engines that are trying to break into market these days.

It has been nearly a month and a half since the latest installment of Webcom in Montreal; however, the study we did and passed around at K3 Media’s conference booth was just too interesting to let lie…

Working with Aleece Germano, we produced a report and found many insights into how people use and perceive BareMinerals makeup, a top-selling foundation in the US from BareEscentuals.

Aleece is an internet consultant with 8 years of experience in the cosmetics industry. She’s also a musician, an event organizer, and luckily for us, our neighbour at Station C. Yet more proof that Station C is more than just a beautiful co-working space - it is a spawning ground for new ideas and a crossroads of top notch Web 2.0 artisans.

The results of our research were very interesting, to say the least. A hint: just because something is organic doesn’t mean you should put it on your face.

Download our BareMinerals blogosphere graph and lexical analysis fact sheet here.

One of the key questions we get when we present our services to potential clients is, “how many people really blog, though?” and, “Do people - ‘real’ people - really pay attention to blogs?”

At Exvisu we don’t make quantitative claims about the extent to which the blogosphere is representative - in our work, this varies on a case-by-case basis, and a big part of our job is to ensure that the analysis we do relies on a substantial dataset that we construct.

The other argument on this point relates to the sheer size of the blogosphere. For years, people have tried to assert that there are X number of blogs. Another approach - one that we prefer - is to talk about what proportion of people blog themselves and read blogs. The numbers are pretty impressive.

In Quebec, the most reliable information comes from NETendance CEFRIO and Léger Marketing, whose survey data suggests that 8.3% of Quebeckers publish a blog (March 2007) and that over 25% of Quebeckers read blogs (July 2007).

Earlier today Sebastien Provencher from Praized Media published a post summarizing new results from e-Marketer.com about the US blogosphere on The Praized Blog. In terms of bloggers, the numbers tell an even more impressive tale. In 2008 they estimate that 13% of Americans are bloggers and over half the population (54%) reads blogs. By any standard, figures like these are impressive.