Hi, very long time ago i made post small search engine.
And that post i use fts vector for search in your postgress database.
But in time my project needed similar items and i thought it is necessary to take some complex algorithm and implement it, but i can make small and easy for use algorithm search a similar items from my database.
I start search Google and ... oh my god, very much interest algorithms, but i dont see for NodeJS. Sorry, may be i not good use Google.
Hmmm Stop! I have fts vector and we have fast algorithm Levenstein. Yeah.
FTS vector - this is small matrix with very important word/words in your text. I have this structure in my database:
item
id | title | price | description | fts
fts vector generic sum of title and description.
We can compared this vectors in algorithm Levenstein or another. That is easy.
Okay, how start this ?
First i writed himself algorithm Levenstein, but he is sooo slooooowwwwllllyyyyy... and i found npm-package -> fast-levenstein
Okay, go write code.
npm install fast-levenstein
Next, i created this file.
var levenshtein = require('fast-levenshtein');
var getSimilarItems = function(id,func) {
//....
//code of get items, sort this items
//....
let levenshtein_number = levenshtein.get(data[0].fts, category_items[i].fts)
//compare numbers and return necessary
}
module.exports.getSimilarItems = getSimilarItems
And, thats all. What do you think about this and what i do better ?
Top comments (0)