gistfile1.js

function cron_analyze_users(){
    // get a user that needs to be analyzed
    var subject = storage.users.sort(lastAnalyzed).first()
    
    // make a hash of their interests and the difference between the volume of 
    // the term for the user and the volume of the term on the whole site. 
    var interests = []
    storage.interests.filter({user:subject,relevant:true}).forEach(function(i){
        interests[i.term.string] = i.difference
    })

    // this is a long way of sorting users by the last time they were compared to the subject user
    var analyzed = []
    storage.relations.filter({subject:subject}).forEach(function(r){
        analyzed[r.object.id] = r.analyzed
    })
    storage.users.sort(function(a,b){
        if(!analyzed[a.id] && analyzed[b.id]) return 1
        else if(analyzed[a.id] && !analyzed[b.id]) return -1
        else if(analyzed[a.id] && analyzed[b.id]) return analyzed[a.id] - analyzed[b.id]
        else return a.created - b.created
    }).limit(20).forEach(function(object){
        if(object.email != subject.email){ // don't compare a user to themselves
            relation = storage.relations.filter({subject:subject, object:object}).first()
            if(!relation) relation = storage.relations.add({subject:subject, object:object})
            
            relation.euclidean = euclidean(subject, object, interests)
            relation.pearson = pearson(subject, object, interests)
            relation.tanimoto = tanimoto(subject, object, interests)
            relation.analyzed = new Date()

            // copy all that to the flip side of this relationship, screw normal-form!
            flip = storage.relations.filter({subject:object, object:subject}).first()
            if(!flip) flip = storage.relations.add({subject:object, object:subject})
            flip.euclidean = relation.euclidean
            flip.pearson = relation.pearson          
            flip.tanimoto = relation.tanimoto 
            flip.analyzed = new Date()
        }
    })
    subject.analyzed = new Date()
}
function euclidean(subject, object, interests){
    sum = 0
    storage.interests.filter({user:object,relevant:true}).forEach(function(i){
        if(interests[i.term.string]){
            sum += Math.pow(interests[i.term.string] - i.difference, 2)
        }
    })
    if(sum) return 1/(1+Math.sqrt(sum)) // convert to number between 1-0
    else return false
}
function pearson(subject, object, interests){
    var shared = 0
    var subjectSum = 0
    var objectSum = 0
    var subjectSqSum = 0
    var objectSqSum = 0
    var productSum = 0
    storage.interests.filter({user:object,relevant:true}).forEach(function(i){
        if(interests[i.term.string]){
            shared++
            subjectSum += interests[i.term.string]
            objectSum += i.difference
            subjectSqSum += Math.pow(interests[i.term.string],2)
            objectSqSum += Math.pow(i.difference,2)
            productSum += interests[i.term.string] * i.difference
        }
    })
    if(shared) {
        num = productSum - (subjectSum * objectSum / shared)
        den = Math.sqrt(
            (subjectSqSum-Math.pow(subjectSum,2) / shared) * 
            (objectSqSum -Math.pow(objectSum ,2) / shared)
        )
        if(den) return Math.round((num / den) * 100)/100
    }
    return false
}
function tanimoto(subject, object, interests){
    var shared = 0
    var subjectSum = 0
    var objectSum = 0
    var termSum = 0
    storage.interests.filter({user:object,relevant:true}).forEach(function(i){
        if(interests[i.term.string]){
            shared++
            subjectSum += interests[i.term.string]
            objectSum += i.difference
            termSum += i.term.difference
        }
    })
    if(shared) {
        num = subjectSum * objectSum / shared
        den = (Math.pow(termSum,2) / shared) * 2 - (subjectSum * objectSum) / shared
        if(den) return num/den
    }
    return false
}