Money packages
Ask a computer what you can get by putting in a money amount in and it shall tell you a package of goods and services that when combined offer something useful to you in return for that money.. so if you put enough cash for a home renovation it shall show you ideas for a house renovation and be fully costed. If you have certain electronics it shall recommend what electronics to buy based on what you have such as accessories.
Imagine a database of products and services. Then you match packages of products and services of a particular theme together and sell someone a package of them together.
So if you also had a database of what you have you can predict what someone might want based on what they have. And what they don't have you can recommend something to buy.
Would this be like Google maps, where you could get a list of all locations that you can travel to with a given amount of fuel? (i.e., you enter the amount of fuel, and it shows all the roads that could possibly be taken from a location X) We could do that kind of map with money too: filter of product combinations (possible baskets), but combinatorially that would be explosive -- it is already explosive even with simple maps and fuel, because every forking on the road leads to new possible travel combination. You have to constrain your query to, say, -- what's the furthest you could go with given amount of fuel / money.
Mindey I want that for walking and public transport usage.
For example, I want to see how far I can get within 10-20 minutes. There may be shops near bus stations or train stations I don't know of.
But yes the number of items you can purchase for a set amount of money is interesting computationally. I think it's a variation of the calculating coins to use to give change or to bin pack items in a crate or shipping package.
Like an extrapolating recommender based on your current trend and the money threshold?
Yes recommendation systems could be used to calculate what other people have and could use reviews for feedback.
Amazon has a section on a product page that says "people that bought this also bought this." And then it says buy these for £a+b+c. I'm not sure if it lets you buy it for cheaper but I've never bought one of them. I saw it for my book I bought recently..there was another book that people bought with my book.
The reason I want this is that I want to do this for salaries.
Imagine registering for a company that does economy of scale to provide services to you. So you allocate 100% of your salary to them and they allocate you food, transport and shelter.
// "people that bought this also bought this."
Amazon recommends generally based on collaborative filtering.
// buy these for £a+b+c
Perhaps that's just k-means-like clustering.
There are many more complex methods, but those do not do true intent inference: those are associations based on trends, not understanding true intent of user.