If there’s one thing most bloggers almost universally love, it is tools that let us compare things and identify trends. Compete charts. Google Trends charts and Insights for Search maps. Technorati graphs. We eat those types of sites up because they’re supremely helpful when doing research for posts. They’re also very useful for academic researchers, web site marketers, and advertising execs.
One of those useful tools is Lexicon from Facebook, which graphs mentions of words and phrases on profile Walls over time. We recently called Lexicon one of our top ten way to track buzz.
Today, Inside Facebook pointed to a preview version of a coming update to Lexicon that adds a number of new metrics. The preview version is limited to searches on 17 pre-selected terms, but offers a glimpse at what is coming.
The new Lexicon will provide access to exponentially more data than the current version. Below is a list of the various pages in the preview version:
- Dashboard – Essentially the current Lexicon on steroids. It graphs terms based on the number of posts, posters, and as a percentage of all posts. Also includes gender and age information on posters for each term. The time frame controls are still really bad, though.
- Demographics – Demographic trends for age, gender, and country of origin for terms and phrases expressed as a total number of posters or percentage of all posters. In the preview, country stats only work for the US, Canada, and the UK, hopefully more country options will be added when the new Lexicon actually launches.
- Associations – Plots additional words and phrases mentioned on profiles that also mention the searched for word or phrase. The results are plotted with an interesting visualization that not only shows off how often associated words are mentioned, but also the average age and gender of those making the association.
- Sentiment – Shows the percentage of posts that are positive vs. negative for a phrase or word. Demo version allows comparison of up to three terms. We assume they use some sort of pre-defined ontology for determining if phrases are negative or positive. It’s easy to imagine that sentiment scores can be tripped up — for example, “I hate all the negative press about Walmart,” would likely appear to a computer to be a negative sentence about Walmart, when in actuality the opposite is true. With that in mind, it might be wise to take sentiment scores with a grain of salt, at least until we know more about Facebook’s method for determining negative vs. positive.
- Pulse – Pulse shows the raw counts for related terms that appear in the profiles of users who also mention the searched for term or phrase. This could be extremely useful for determining interests related to your core users. Lists can be broken down into terms that appear on the Interests, Music, Movies, TV Shows, and Books sections of user profiles.
- Maps – Creates heat maps based on the buzz surrounding certain products. Right now it does Canada, the UK, and the United States.