This guide introduces the major features of the Carbon Capture Report.
Several of the Carbon Capture Report displays integrate information on the tone of the underlying content. The following indicators are available:
- Tone/Positivity/Negativity. Specially-tuned linguistic algorithms examine the text of each document and determine the overall "tone" of the wording it uses. This does not measure how positive or negative the underlying events being reported on are, but rather how positive or negative the writer is portraying them as. For example, an announcement of a new wind turbine plant might be viewed as an extremely positive event for wind energy proponents, while local townspeople might view it as an extremely negative development to be fought. Both sides will likely generate news coverage casting the turbine plant in either a positive or negative light, respectively. This measure therefore offers insight into local and global reaction to new developments. Positivity and Negativity measures indicate the raw intensities of positive and negative content, while Tone is a composite of the two, with a positive score indicating the text is more positive, and a negative score indicating a more negative text. Tone can range from -100 to +100, but in practice usually falls into the range -10 to +10.
- Polarity. Polarity measures the overall "emotional charge" of a text, independent of its tone. An article with equal levels of positive and negative content will have a Tone score of neutral, but if it contained significant emotional wording on both sides, it will still have a high polarity score. Polarity therefore measures how "emotional" an article is, ranging from 0 to 100, with lower scores indicating more "clinical" texts that simply recount facts, and higher scores indicating more impassioned pleas and discussions.
- Activity. Activity measures the intensity of "active language" in a text, indicating whether a text is more passive and scientific in nature, recounting a series of factual statements, or more active, encouraging the reader to take action and using intense language to discuss information. This indicator ranges from 0 to 100, with lower scores suggesting factual recollections and higher scores suggesting emotional calls to action.
- Personalization. Personalization measures the degree to which the writer attempts to bring the reader into the fold as "part of the story" versus maintaining clinical detachment in describing events. For example, a blog post could simply criticize a new power plant and suggest its construction should be protested, or it could invoke the reader as a fellow activist, encouraging him or her to "join in" on the fight against the plant. Personalization ranges from 0 to 100, with higher personalization levels usually indicating calls to action and other more subtle mechanisms of encouraging the reader to be "one of the group."
- Questions/Exclamations. Tweets have two further tone indicators: questions and exclamations. Given their 140-character size limit, Tweets make heavier use of punctuation as tone carriers, and these metrics reflect tone carried through non-word mechanisms.
The Activity Timeline allows you to view coverage characteristics over time, seeing how negativity interacts with personalization, for example, or how changes in news coverage have correlated with blog volume. These displays are rendered as interactive Flash applications and so may not work properly on devices that do not support Flash.
As you move your mouse over the display, you should see a set of dots, one per data line, follow your mouse, and the date of the datapoint under your mouse is displayed in the upper-left. The actual numeric values of the currently displayed data lines for the date under your mouse should be shown along the top of the display, color-coded to match each data line.
In the bottom of the display is a log-scaled version of the timeline. You can click and drag the right or left edges of this portion of the display towards each other to reduce the size of the time window currently displayed. Alternatively you can click on the preset time windows in the top-left of the display, such as "5d" (5 days), "3m" (3 months), or "Max" to restore the display to the entire time window. When working with a smaller window of time, the scrollbar at the bottom of the display allows you to move the currently-displayed time window through the available time, or you can also click and drag directly on the timeline display itself.
Using the checkboxes below the display, you can turn individual data series on and off to explore relationships between different variables. Note that variables have different scales associated with them (some may range from 0 to 100, while others might reach maximums of 10,000 or more) and so some combinations may result in one or more data lines displaying as flat lines due to incompatible scales.
The default timeline display shown on the summary page of a Daily Report or Biographical Profile shows you the major temporal trends in coverage, but for further analysis you will likely want to be able to see the actual coverage of each day overlaid onto the timeline. Clicking on the By News Articles, By Blog Posts, or By Tweets links on the lower-right under a timeline display opens the timeline display in full-screen mode, along with additional data series. The most unique aspect of full-screen mode, however, is the list of the top news article, blog post, or tweet of each day. Clicking on a data point will jump to that article's entry in the list at the right side, with a link to the article. Alternatively, you can keyword-search titles using the search box at top-right to limit to just those days with top articles having titles mentioning a specific keyword. Using the annotated timeline display makes it possible to instantly see the most popular stories corresponding to surges in coverage volume, drops in tone or polarity, or characteristics of coverage.
DASHBOARD / VISUAL SUMMARY
The Visual Summary display offers a "dashboard" visualization of the key indicators of a given day or profile entry's coverage. The four dials in the upper-left display the overall tone across all news, blog, and Twitter coverage and the individual tone averages for each, to help with media breakdown. To the right of the tone gauges are three more gauges, showing the average Polarization, Activity, and Personalization across all content.
At the bottom-left of the dashboard is a pie chart showing the relative breakdown by volume of the coverage into news, blog, and Twitter sources. For Biographical Profiles, this display makes it immediately clear where the majority of coverage of that entity comes from, especially whether it its primary coverage base is the traditional mainstream or the emerging social media.
Daily Reports display the Overall Tone dial again in the lower-right at a larger size to make it easy to quickly see whether tone that day is high or low. Biography Database profiles instead display a dial showing what percentage of the days monitored by this topic have included at least one reference to that entry. This is useful for spotting the "usual suspects" versus brief or sporadic mentions. For example, a high-profile event might lead to the organizer's name being mentioned heavily for several days, leading to a high overall volume ranking, but if the person is never mentioned again in the news, this indicator would help segment that person from those that have lower overall volume, but are regularly mentioned.
Many topics have a built-in list of categories that help further refine coverage by content focus. These topics will include bar charts that show the overall percentage of stories falling into each category. Articles are classified into as many categories as apply, so percentages may sum to greater than 100%. For Daily Reports, this can be used to quickly estimate the major focus of the day day's coverage, while for Biographical Profiles, it allows individual entities to be organized by their primary role and associations in the news.
Categories are also available in the Activity Timeline, allowing you to see ebbs and flows in each topic area.
The Carbon Capture Report examines the full text of each news article, blog posting, and tweet, and automatically identifies and disambiguates all globally geographic references down to the centroid city level. This information is then used to create both thematic and precise maps of the geographic focus of coverage. Thematic Maps aggregate references to the country level, allowing quick views of the major geographic corridors. Google Maps and Google Earth KML files display coverage as precise locations on earth (to the city level), making it possible to precisely trace coverage at the subnational level.
Google Earth files contain additional date information, tying each geographic reference to the date of the article mentioning it. When viewing the file in Google Earth, a "time slider" will be displayed in the upper-left of the program, allowing you to set a specific time window and move the slider to see how coverage has changed over time at the city level. For example, major events will result in strong clustering of coverage around small geographic points, while over a period of weeks, one will often see strong ebbs and flows in coverage across a country.
Using DNS registrations, geographic context, and other information, the Carbon Capture Report also attempts to pinpoint the geographic location of each outlet's controlling organization, allowing coverage to be stratified by geographic source. This makes it possible to display a day's coverage by major originating locations and focus locations.
BIOGRAPHICAL PROFILE ENTRIES
The Carbon Capture Report analyzes the full contents of every news article, blog post, and tweet it monitors, and extracts a list of all individuals, organizations, and locations mentioned in that content, as well as recording a list of the outlets that have published on the topic. This allows you to interact with coverage not as a massive pile of documents, but as higher-order knowledge. Of particular interest is the ability to track emerging players, such as new companies just entering the market. As the saying goes, "you don't know what you don't know," but our Biographical Profile entries help reduce this, giving you daily insights into the "new names in the news".