3 ms·
It's also prob not the best source of features (in a ML sense) for choosing ads. When you read an email, your intent is either to understand what the sender is
by ecesena 9y ago
It's also prob not the best source of features (in a ML sense) for choosing ads.
When you read an email, your intent is either to understand what the sender is asking you (new email), or get the info you asked for (response to your email).
In the majority of cases a contextual ad is 1. pretty weird, 2. pretty against your primary intent.
Might as well showing you ads related to your recent browsing history, imo.
- linkregister 9y agoI see it as an opportunity to use ML to improve targeting. Ad clicks could be used as the training data since users are less likely to click on non-relevant or inappropriate ads. This would help with avoiding inappropriate keywords, e.g. death in the family.