Uncategorized

Linkfluence is known for its web cartography expertise, but that’s not all we do. Maps are tools, they define a territory, a “playing field”, which we can then mine and analyze. One way of analyzing the context and content of discussions is by creating word clouds: word clouds provide a convenient albeit imperfect overview of the gist of a large volume of content. Unfortunately, word clouds are typically used as ‘snapshots’ of said content at a given time. This is fine when looking at homogeneous content, but becomes counterproductive when trying to analyze trends, rise or decline of certain words, brands or concepts over time, particularly when considering the ebb and flow of online conversations.

We have been working on this issue for some time, with the objective of combining the convenience of word clouds with a chronological dimension to follow semantic trends. And here it is: “timecloud” allows you to manipulate the time line and update the associated word clouds accordingly.

In the following example, which represents a “timecloud” for the brand Apple within the Mac-fans community, one can easily visualize which words, brands, adjectives come and go over the past 90 days. Unsurprisingly, “iphone” and “app store” remain very prominent throughout the entire period, while product launches like “snow leopard” or “tablet” pop up at the time of the launch then recede in the background.

Hence, this “timecloud” not only displays the most frequently-used words over the selected time period, it also shows the evolution of the conversation through the appearance and disappearance of those key words and concepts. It is also possible to toggle the “list mode” (icon in the upper right hand corner, next to the magnifying glass) to display all the words with the same size, including “unused” words (in light grey) for the selected period. When selecting a word on the tag cloud, a graph will appear at the bottom, showing relative volume of mentions over the entire timeline. Using the ctrl key, one can select up to four words and follow their evolution over time.

Plurk This Post

It’s been a busy week in the politicosphere, with heavy coverage and discussions about three prominent news topics: the war in Afghanistan and the need for more troops, the U.N. General Assembly and Security Council resolution, and the perennial health care reform debate.

News Trends - Content published across a sample of 13 000 sites

What’s newsworthy –or blogworthy- however largely depends on which part of the web you’re looking at, or which community one hails from. By and large, health care reform remains the dominant topic of discussion within the conservative and progressive communities.

Out of a sample of 1202 conservative blogs and 1190 progressive blogs, health care reform accounted for 9.7% and 12.3% of new content respectively during the week of Sept. 21 to Sept. 27. By comparison, discussions about the need for more troops in Afghanistan “only” accounted for 4.8% of content published on the conservative side vs. 3.5 on the liberal side. The same topic however fired up the Defense (23.1%) and International Affairs (15.7%) communities which naturally dedicated more content to this topic than any other this week.

The International Affairs community got even busier by the second half of the week with a series of news coming out of the United Nations General Assembly, environment summit and security council resolution. The environmental community got stirred up as well by the U.N. Climate Summit in the run-up to Copenhagen.

Let’s see next week if the G-20 summit manages to garner as much attention.

Plurk This Post

(cross-posted on politicosphere.net) A couple of weeks ago, the New York Times dedicated a very clear and concise article to the ambitious research spearheaded by Jon Kleinberg of Cornell University on “Meme-tracking and the Dynamics of the News Cycle”. The research team “studied the news cycle by looking for repeated phrases and tracking their appearances on 1.6 million mainstream media sites and blogs. Some 90 million articles and blog posts, which appeared from August through October, were scrutinized with their phrase-finding software”.

The research concluded that blogs typically trail behind “MSM”, except in rare instances (3.5% of story lines to be precise) where a handful of blogs picked up on the news first. Unsurprisingly, these are not just any blogs but professional and quasi-professional political blogs with the connections, competence and credibility that in effect puts them on par with traditional news organizations.

We recently had the opportunity to look up close at one such instance where an influential political blog “broke the news”, initiating a new news cycle as national media followed suit. The following video is a visualization of the propagation of the story as it moves across the online political landscape.

Blogs breaking news – firedoglake case study from linkfluence on Vimeo.

note: this visualization was created afterwards, based on a sample of articles covering the story, to dynamically represent the speed at which this particular piece of information spread from a well-identified source to many other blogs and media sites. It is not “real-time” propagation and does not represent every article and every site referencing the story. For methodology questions regarding the map itself, please refer to the “Keys” section on this page .

Plurk This Post

Leave it to your users to speak about your products better than you could.

Peter Corbett with iStrategyLabs posted this great video-demo of our linkscape dashboard (tagline: “social graph insight”).

Mapping Linkfluence – Financial & Swine Flu Example from Peter Corbett on Vimeo.

If you’re looking for “catch-all” social media monitoring, linkscape is probably not the right tool for you. But if you believe like us that “less is more”, that social media analysis at the community level can provide a fuller picture than individual posts, and if you’re desperate to measure trends and share of voice over time without falling victim to the “Google Dance”, you ought to give linkscape a try.

Plurk This Post