Datamapper - Get most associated N records for association ( ex: 5 most liked posts )

Question

I am working on my blog, a Sinatra application that uses Datamapper as its ORM. I just added a feature for hash-tags. A hash-tag is as such, its a many-to-many relationship with a 'Story'. My need is to get the most popular/used 5 hash-tags to display on the sidebar.

Here is an abstract of my modals that are of interest.

The hash-tag model

class Hashtag

  include DataMapper::Resource

  has n, :stories, through: Resource

  def self.default_repository_name
    :default
  end

  property        :id,                Serial
  property        :created_at,        DateTime
  property        :updated_at,        DateTime

  property        :hashtag,           String
end

The story model:

# File: models/story.rb


class Story

  include DataMapper::Resource

  def self.default_repository_name
    :default
  end

  belongs_to :person
  belongs_to :category

  has 1, :story_content

  has n, :comments
  has n, :hashtags, through: Resource


  property        :id,                Serial
  property        :created_at,        DateTime
  property        :updated_at,        DateTime

  property        :published,         Boolean,    default: false
  property        :privacy_level,     Enum[ :private, :friends, :public ] # SNS access control for Twitter/FB only

end

Need: The most efficient way to get the most used n hash-tags.

• Tag 1 ~> 1 story
• Tag 2 ~> 2 stories
• Tag 3 ~> 3 stories

For n = 2, I need ~> [Tag 2, Tag 3]

I am thinking of maintaining a counter and did it for some thing like 'Likes'. Wonder if there is any other way than cache the count of Stories per HashTag.

Answer 1

Since the top tags are not very likely to change very frequently, you could alternatively compute and cache them for a longer period of time, for example a day. Then, you can recompute the top five from scratch in a nightly job.

Pros:

• No extra stories_count column
• No maintenance of the stories_count (not that easy to handle!)
• Fast queryability

Cons:

• Top 5 may be out of date (not an option if you expect frequent changes)
• Need to build and configure a nightly job