Identification of targeting-relevant Information

Based on the existing classification scheme from The AdInsider, a number of classes are selected for which a more fine-grained contextual analysis (“situation”) should be performed. This allows for a more precise setup during the targeting process, e.g. to select adverts for a concert based on the discussion of a CD within a forum page. The currently available classification schemes are not sufficient for such purposes. In order to make the classification applicable in commercial terms, a methodology for the automated training and retraining of classes is required.

Depending on the textual context, related words may have different meanings. Such variants are the possible interpretations of words. An initial prototype can already recognise the different interpretations, and thus allows for a fine detection of positive and negative contexts (e.g. a blog post, dealing critically with a specific product, vendor or theme (e.g. travel destination).