For the most part, online advertising, be that image or link-based, is triggered either by context (content-type or keyword), search term or the known (or assumed) characteristics of the individual. The fuzzy logic supporting many of these targeting systems usually skim an appropriate target audience, but are better known for a minority of howlers, like “Swept off rocks while fishing”, accompanied by a no-expense spared tower ad spruiking bait and tackle!
Now a group of researchers at China’s Zhejiang University are proposing that the typical contextually-based trigger of a topic word(s) be supplemented by a further trigger, namely a sentiment classification developed via a process of ‘opinion mining’. The proposal carries the acronym DASA, or Dissatisfaction-orientated Advertising based on Sentiment Analysis.
Sentiment-Assisted Ad Targeting
Once the topic words are extracted, the consumer’s sentiment on those extracted words is determined. The words used in a negative context are then chosen for advertising keyword selection. In other words, an ad is considered to be relevant to a Web page only if it is associated with the topic words towards which consumers have negative attitudes. The point of this exercise is to effectively ambush a brand which is experiencing negative sentiment with a competitor’s offer or message.
The research concluded that ad selection based on a DASA system delivered a more appropriate ad (i.e. more relevant offer or creative execution) almost twice as often as a system relying totally on keyword selection.
Leave a Reply