Hunch.com is a Web site that provides personalized recommendations for its users. More conventional recommendation engines used by Amazon.com, Netflix, Youtube, and others rely on knowledge about your preferences (products you’ve bought, or movies you’ve watched) and correlates that to information it has about other users. This works fine reasonably often, but also may lead to irrelevant or undesirable suggestions.
Hunch builds a “taste profile” by asking a user questions, and generates recommendations from there. It’s not perfect either, but consider the limited amount of questions it asks, it can be amazing how well it fits.
For this lab exercise, explore Hunch by answering the initial set of questions (you don’t need to create an account), and by trying to determine how the Web site came up with the recommendations it presented to you. Try to identify three to five methods Hunch may use (we don’t know for sure - they’re a bit guarded about this).
Submission
I’ll try to put together a Web or Blackboard questionnaire for this by Thursday. For now, you can use the table below to take some notes.
Student Name:
| Methods | Hunch’s use of the method | Evidence |
| 1 | | |
| 2 | | |
| 3 | | |
| 4 | | |
| 5 | | |
| Summary | | |