measures a cosine similarity between two one-hot encoding vectors
Demo code
$dot = Similarity::dot(["a", "b", "c", "d"]);
$score = Similarity::cosine(["a", "b"], ["b", "c"], $dot);Example articles:
Array
(
[0] => Array
(
[article] => Data Mining: Finding Similar Items
[tags] => Array
(
[0] => Algorithms
[1] => Programming
[2] => Mining
[3] => Python
[4] => Ruby
)
)
[1] => Array
(
[article] => Blogging Platform for Hackers
[tags] => Array
(
[0] => Publishing
[1] => Server
[2] => Cloud
[3] => Heroku
[4] => Jekyll
[5] => GAE
)
)
[2] => Array
(
[article] => UX Tip: Don't Hurt Me On Sign-Up
[tags] => Array
(
[0] => Web
[1] => Design
[2] => UX
)
)
[3] => Array
(
[article] => Crawling the Android Marketplace
[tags] => Array
(
[0] => Python
[1] => Android
[2] => Mining
[3] => Web
[4] => API
)
)
)
Target article:
Array
(
[0] => Publishing
[1] => Web
[2] => API
)
Sorted result similarity:
Array
(
[Data Mining: Finding Similar Items] => 0
[Blogging Platform for Hackers] => 0.23570226039552
[UX Tip: Don't Hurt Me On Sign-Up] => 0.33333333333333
[Crawling the Android Marketplace] => 0.51639777949432
)