Met Recommendation

Gabriel Gianordoli and John Choi

MET Media Lab

Spring, 2015

To-dos

DATA

• Get from scrapi:

  • Check whether or not object has image
  • Check whether or not object is on display
  • Get the gallery number (from the website)
  • Get the department
  • Get the image url

• Change the data format to:

{
	item_id: 16584,
	item_title: 'George Washington',
	gallery_number: 140/null, // null if object is not on display
	department: 'egyptian art'/null,
	img_url_web: 'http://...', // null if it doesn't have image?
	img_url_big: 'http://...', // null if it doesn't have image?
	similar_items: [
			{ item_id: 004327, similarity: 0.01 },
			{ item_id: 052345, similarity: 0.005 },
			...
	]
}

• 10 items based on user's collection
• 10 related to the last object the user picked
• 5 random

SERVER-SIDE

• Program the interface with the data (json file)

• Program the interface with the user

  • Kart
  • Combine recommendations for each new item (add latest)

CLIENT-SIDE

• Visual design

• Program the data display

• Program the interaction