Blog > February 2012

We’ve just posted a shiny new dataviz on our website, which pits the different host cities of Social Media Week against each other in a royal rumble of skyscraper proportions.

Each city is represented by the tallest building located in that city, from London’s amazing new Shard building to the Washington monument in DC. We’ve created queries using the Brandwatch tool to track all mentions (across Twitter and Facebook) of each city with regard to Social Media Week.

When composing the queries we stumbled across a few hurdles, so we thought this would be a good opportunity to enlighten you all about how users can approach the query-creating process when using Brandwatch.

The widget was originally designed to pull in data from a solitary query in Brandwatch, which we had set up with the following string, using the languages of Portugese, Brazilian Portugese, French, English, German and Spanish:

((“social media week” OR socialmediaweek OR smw*) AND site:(twitter.com OR facebook.com))

This will pick up all mentions of any of the three terms shown in bold, as long as it is from either Twitter or Facebook. It will also pick up all Twitter mentions and replies to the official Twitter accounts, as they all feature ‘smw’ in their handles.

To add to the query, we included some Twitter handles that have been set up for use with each city’s SMW events, so that all of their tweets would be picked up too, even if they didn’t include any of the above terms.

OR

(url: twitter.com/smwhamburg/ OR url: twitter.com/smwhk/ OR url: twitter.com/smwldn/ OR url: twitter.com/smwmiami/ OR url: twitter.com/smwnyc/ OR url: twitter.com/smwparis/ OR url: twitter.com/smwsf/ OR url: twitter.com/smwsp/ OR url: twitter.com/smw_sg/ OR url: twitter.com/smwtok/ OR url: twitter.com/smwto/ OR url: twitter.com/smwwdc/)

This served as our master query, which pulled in all mentions of Social Media Week in general into one dataset, totalling around 40,000 mentions at the beginning of this week.

Next we broke down the data into categories. We did this by performing the following searches for each city.

url:twitter.com/smwldn/ OR london OR ldn OR smwldn

The search would then return all mentions of terms like LDN within the dataset we created, as well as all tweets from the official SMWLDN Twitter profile, finding around 8,000 mentions out of the 40,000 to be related to London in some way.

After repeating this for each city, we found that some cities were lacking. It’s impossible to perform a concurrent search between alphabets, so adding Japanese and Chinese to the query was problematic. The results showed only mentions of SMW Tokyo and Hong Kong that were composed in the Latin alphabet.

We also had to consider that while our coverage of Twitter is absolute, Asian markets aren’t as dominant in their use of the Twitter platform.

In China, for example, Weibo is the prevailing social network. It’s a site that we are attempting to improve upon with regard to our coverage, but permission to access their API isn’t so easy to gain.

To remedy this as best as possible, we modified the structure of the dataviz, so that it instead pulled in twelve separate queries: one query for each city, rather than one category each within a global query.

(((“social media week” OR socialmediaweek OR smw*) AND site:(twitter.com OR facebook.com))

OR

(((“social media week” OR socialmediaweek OR smw*) AND site:(twitter.com OR facebook.com))

OR

(url: twitter.com/smwhamburg/ OR url: twitter.com/smwhk/ OR url: twitter.com/smwldn/ OR url: twitter.com/smwmiami/ OR url: twitter.com/smwnyc/ OR url: twitter.com/smwparis/ OR url: twitter.com/smwsf/ OR url: twitter.com/smwsp/ OR url: twitter.com/smw_sg/ OR url: twitter.com/smwtok/ OR url: twitter.com/smwto/ OR url: twitter.com/smwwdc/)) AND (url:twitter.com/smwparis/ OR paris OR smwparis OR raw:smw_paris OR raw:SMW_paris OR raw:SMW_PARIS)

This query track everything to do with Social Media Week (the Twiitter profiles, the key terms and hashtags), but then insists it has to also be related to Paris, the Paris SMW Twitter profile or a string of other hashtags and terms.

We also set up specific additional queries in Japanese and Chinese (as well as English) to contribute to the Tokyo and Hong Kong totals, boosting the Roman-only language queries.

That is how we set up the queries that power the new Social Media Week widget that can be found on our homepage, here. If you have any questions or comments about the dataviz, or about how queries work in Brandwatch, please don’t hesitate to contact us.