CakePHP - trying to order posts by relevance to tags shared with main post and date created - mysql

Question

I'm trying to display eight of the most relevant posts, in order of relevance, based on tags shared with the current post being viewed and the date created...
Models:
Tag habtm Post
Post habtm Tag
DB:
posts(id, slug, ...)
tags(id, tag, ...)
posts_tags(post_id, tag_id)
Within controller action:
$post = $this->Post->find('first', array('conditions' => array('slug' => $slug)));
$this->set('post', $post);
$tags = $post['Tag'];
$relOrd = '';
foreach($tags as $tag){
$tagId = $tag['id'];
$relOrd .= " + (CASE WHEN PostsTag.tag_id = ".$tagId." THEN 1 ELSE 0 END)";
}
$relOrd = '(' . substr($relOrd, 3) . ') AS Relevance';
$morePosts = $this->Post->find('all', array(
'joins' => array(
array(
'table' => 'posts_tags',
'alias' => 'PostsTag',
'type' => 'LEFT',
'conditions' => array(
'PostsTag.post_id = Post.id',
)
)
),
'group' => 'Post.id',
'fields' => array($relOrd, 'Post.*'),
'order' => array('Relevance' => 'DESC', 'Post.created' => 'DESC'),
'limit' => 8,
));
$this->log($morePosts);
$this->set('morePosts', $morePosts);
It's almost working, although the relevance value is being treated as if each post has only one tag (being only 0 or 1). So it seems that the relevance value for each post is taking either 0 or 1 depending on the posts LAST tag rather than being accumulative based on ALL tags.

Answer 1

First of all, I'd take all of the logic out of the controller. Consider this:
$post = $this->Post->find('first', array('conditions' => array('slug' => $slug)));
$this->set('post', $post);
$this->set('morePosts', $this->Post->findRelevant($post));
Now your controller is easy to read and does it's job. You essentially start by describing what data you want by naming an imaginary model function, and then you write the model code to fulfill that request.
So here's a stab at the model code:
var $actsAs = array('Containable');
function findRelevant($post, $limit = 8) {
// create an array of ids of the tags from this post
$tags = array();
foreach($post['Tag'] as $num => $tag) {
$tags[$tag['id']] = $tag['id'];
}
// find other posts that have any of those tags
$relevant = $this->find('all', array(
'conditions' => array('Post.id <>' => $post['Post']['id']),
'order' => 'Post.created desc',
'contain' => array('Tag' => array('conditions' => array(
'Tag.id' => $tags
))),
));
// count the number of tags of each post and call it relevance
// (this number is essentially the number of tags in common
// with the original post because we used contain to get only
// the tags from the original post)
foreach($relevant as &$p) {
$p['Post']['relevance'] = count($p['Tag']);
}
// sort by relevance
$relevant = Set::sort($relevant, '{n}.Post.relevance', 'desc');
// limit the number of posts returned (defaults to 8)
return array_splice($relevant, 0, $limit);
}
Obviously it would be great to use the database logic to fetch the records (as you are attempting to do) so that it's as fast as possible and so that you minimize the amount of data that you retrieve, but I can't see how to do that for what you're trying to achieve.
This method should work fine and is not database specific. :)