Read this word problem: 'Maya, who has red sneakers and loves art, buys 3 packs of stickers. Each pack has 12 stickers. It is a sunny Tuesday. How many stickers does she have?' To answer, you need only 3 packs and 12 stickers in each: 3×12=36. The sneakers, the art, and the weather do not change the answer, so you set them aside. That is abstraction: keep what matters for the question and ignore the rest.
Computer scientists do this all the time. A map app treats roads as lines with lengths and speed limits and ignores trees and house colours. A video game stores a ball as just its position, size, and speed, not every stitch on it. A button labelled 'Print' hides all the steps that send a page to the printer, so you only need to know what the button does.
What counts as extra depends on the question. If the problem asked 'What colour are Maya's shoes?', red sneakers would be the one detail to keep. And abstraction is not the same as being vague. 'She bought some stickers' throws away the numbers you need; that is not a simpler model, just a wrong one. A good abstraction is exact about what it keeps.
Think of it like this. Abstraction is like packing your school bag for one day. You check the timetable and pack only what today's lessons need, leaving the rest at home. Pack too little and your maths book is missing; pack everything and the bag is too heavy to carry. A different day, like a different question, means packing differently.
Core idea. Good abstraction reveals what matters and hides what doesn't.