Measuring Buzz

Tylenol & Advil - brand perception and usage

Previously I touched on how using technology to measure buzz and online perceptions around products and services on the web while possible, still required a human element to interpret the resultant findings.
The reason, human traits such as sarcasm and sentiment could skew machine based analytics. An example being the word 'bad' could be interpreted by a human as actually being good, alternatively the context of an entire sentence could be reversed with the word not! at the end
marketing technology
An interesting development in this area from Crimson Hexagon could mean working out who is saying what about your brands could be a far simpler process than using technology to count words and frequency, with humans providing time and cost intensive interpretation.
At the heart of the technology is an algorithm which uses statistical analysis to gleen meaning and understanding from the blogs, websites and other social sites it trawls to produce these unbiased ( through optimization ) insights.
In order to represent this data statistically they use a number of techniques, simplifying the data to create a ' bag of words ', weighting ( the second instance of the word love in a sentence does not provide as much information as the first ) and categorising this ' bag ' of information allowing it to be subsequently reported against, without human intervention.

No comments: