4 Replies - 699 Views - Last Post: 08 October 2013 - 06:21 PM Rate Topic: -----

#1 praveen2710  Icon User is offline

  • New D.I.C Head

Reputation: 0
  • View blog
  • Posts: 5
  • Joined: 17-April 13

Recommendation System?

Posted 08 October 2013 - 11:28 AM

Hi,
I would like to learn more about recommendation system and also is there any way to build a small system for college project.I would like to know about software to use for building such a system if possible
Is This A Good Question/Topic? 0
  • +

Replies To: Recommendation System?

#2 modi123_1  Icon User is offline

  • Suitor #2
  • member icon



Reputation: 9073
  • View blog
  • Posts: 34,115
  • Joined: 12-June 08

Re: Recommendation System?

Posted 08 October 2013 - 11:30 AM

What do you mean by "recommendation system"?
Was This Post Helpful? 0
  • +
  • -

#3 praveen2710  Icon User is offline

  • New D.I.C Head

Reputation: 0
  • View blog
  • Posts: 5
  • Joined: 17-April 13

Re: Recommendation System?

Posted 08 October 2013 - 02:36 PM

When we watch a movie on netflix how does it give 10 more suggestion which are actually valid and have some strong reason to be suggested to you.It is not a static set of movies it changes for every indiviual on movies he sees.That is done by a recommended system.
Was This Post Helpful? 0
  • +
  • -

#4 modi123_1  Icon User is offline

  • Suitor #2
  • member icon



Reputation: 9073
  • View blog
  • Posts: 34,115
  • Joined: 12-June 08

Re: Recommendation System?

Posted 08 October 2013 - 05:18 PM

I would assume the back end is a system of taggings to movies.. an aggregate of the tags of movies you have watch is compared to other movies with similar tags.. displayed as it goes.
Was This Post Helpful? 0
  • +
  • -

#5 Atli  Icon User is offline

  • D.I.C Lover
  • member icon

Reputation: 3717
  • View blog
  • Posts: 5,979
  • Joined: 08-June 10

Re: Recommendation System?

Posted 08 October 2013 - 06:21 PM

The basic concept of systems like these are, like modi123_1 says, to just add meta-data to items; tag them with a list of categories or something, and then cross match the tags from the user's history with the current item. That would result in a pretty good list of recommendations for each users.

However when you're dealing with databases on a massive scale, like those for Netflix or Amazon, there are other techniques they use as well. It's often referred to as Big Data; it's essentially data-mining on a huge scale, where the system analyzes the data and looks for patterns that would be hard to spot or even predict using "normal" methods. This can lead to some interesting results.

For instance, you will sometimes find random things in the suggestions on Amazon that make little sense in the context you're in. You may be looking into books on database design, but all of a sudden find yourself getting suggestions relating to hiking or fishing. It makes no sense on the surface, but behind the scenes the system has found a pattern indicating that people with histories similar to yours, looking up database design books, end up buying fishing and hiking gear. Thus, it suggests that to you as well.

That might be a bit beyond what you are looking for, but it's an interesting topic of you want to expand beyond the basics.
Was This Post Helpful? 0
  • +
  • -

Page 1 of 1