i am using neo4j to retrieve news feed using this query.
MATCH (u:Users {user_id:140}),(p:Posts)-[:CREATED_BY]->(pu:Users)
WHERE (p)-[:CREATED_BY]->(u) OR (p:PUBLIC AND (u)-[:FOLLOW]->(pu)) OR
(p:PRIVATE AND (p)-[:SHARED_WITH]->(u))
OPTIONAL MATCH (p)-[:POST_MEDIA]->(f)
OPTIONAL MATCH (p)-[:COMMENT]->(c)<-[:COMMENT]-(u3) RETURN
(p.meta_score+0.2*p.likes+0.1*p.dislikes + 10/(((".time()."-
p.created_time)/3600)+0.1)) as score,
{user_id:pu.user_id,firstname:pu.firstname,lastname:pu.lastname,
profile_photo:pu.profile_photo,username:pu.username} as pu, p,
collect({user_id:u3.user_id,profile_photo:u3.profile_photo,text:c.text}) as comment,
collect(f) as file ORDER BY score DESC,
p.post_id DESC LIMIT 25
In this equation for getting score right now i am using mainly this equation p.meta_score+0.1*p.likes-0.05*p.dislikes + 10/(((current_time-
p.created_time)/3600)+0.1)) as score here i hace added 0.1 to prevent infinity error as current_time may be nearly equal to post created_time( as p refer post class)
Here its nice for single day but after a day the time part doesn't contribute well total score as the way i am calculating time decay factor is not consistent i need a equation which plays its role consistently (I means decrease score at lesser rate) for first seven days and than start decreasing its contribution towards score at an higher rate. one way was using trigonometry's tan or cot functions but the problem is that after some intervals they changes there signs.I shall be thankfull to everybody gives me further suggestions.
At a basic level, it is common to use an exponential time decay function here. Something like:
score = score / elapsedTime^2
As elapsed time since the post increases, the value of the score decreases exponentially. Sites like Reddit and Hacker News use much more complicated algorithms, but that is the basic idea.
Related
I run a music website for amateur musicians where we have a rating system based on a score out of 10, which is then calculated into an overall score out of 100. We have a "credibility" points system for users which directly influences the average score at the point of rating, but the next step is to implement a chart system which uses this data effectively.
I'll try and explain exactly how it all works so you can see which data I have at my disposal.
A site member rates a track between 1 and 10.
That site member has a "credibility" score, which is just a total of points accumulated for various activities around the site. A user gains, for example, 100 points for giving a rating so the more ratings they give, the higher their "credibility" score. Only the total credibility score is saved in the database, updated each time a user performs an activity with a points reward attached. These individual activities are not stored.
Based on the credibility of this user compared to other users who have rated the track, a weighted average is calculated for the track, which is then stored as a number between 1 and 100 in the tracks table.
In the tracks table, the number of times a track is listened to (i.e. number of plays) is also stored as a total.
So the data I have to work with is:
Overall rating for the track (number between 1 and 100)
Number of ratings for the track
Number of plays for the track
In the chart system I want to create a ranking that uses the above 3 sets of data to create a fair balance between quality (overall rating, normalized with number of ratings) and popularity (number of plays). BUT the system should factor quality more heavily than popularity, so for example the quality aspect makes up 75% of the normalized ranking and popularity 25%.
After a search on this site I found the IMDB Bayesian-style system which is helpful for working out the quality aspect, but how do I add in the popularity (number of plays) and have it balanced in the way I want?
The site is written in PHP and MySQL if that helps.
EDIT: the title says "number of clicks" but this is basically the direct equivalent of "number of plays".
You may want to try the following. The IMDB equation you mentioned uses weighing to lean toward either the average rating of the movie or the average rating of all movies:
WR = (v/(v+m)) × R + (m/(v+m)) × C
So
v << m => v/(v+m) -> 0; m/(v+m) -> 1 => WR -> C
and
v >> m => v/(v+m) -> 1; m/(v+m) -> 0 => WR -> R
This should generally be fair. Calculating a popularity score between 0 and 100 based on the number of plays is pretty tricky unless you really know your data. As a first try calculate the average number of plays avg(p) and the variance var(p) you can then use these to scale the number of plays using a technique call whitening:
WHITE(P) = (p - avg(p))/var(p)
This will give you a score between -1 and 1 by assuming your data looks like a bell curve. You can then scale this to be in the range 0 - 100 by scaling again:
POP = 50 * (1 + WHITE(P))
To combine the score based on some weighting factor w (e.g. 0.75) you'd simply do:
RATING = w x WR + (1 - w) x POP
Play with these and let me know how you get on.
NOTE: this does not account for the fact that a use can "game" the popularity buy playing a track many times. You could get around this by penalising multiple plays of a single song:
deltaP = (1 - (Puser - 1)/TPuser)
Where:
deltaP = Change in # plays
Puser = number of time this user has played this track
TPuser = total number of tracks (not unique) played by the user
So the more times a user plays just the one track the less it counts toward the total number of plays for that track. If the users listening habits are diverse then TPuser will be large and so deltaP will tend back to 1. This still can be gamed but is a good start.
I'm trying to convert this adaptive bayesian rating formula into PHP code: see here.
Here are the details of the various parts of the formula..
nvotes : total number of votes so far
nlinks : total number of links
nvotes(k) : number of votes cast to rth link.
deltarank(k, m) : rank increment caused by kth vote that is casted to mth link.
nsaves(i) : number of users that save ith link to their linkibol.
a : save exponent (an ad-hoc value close to 1)
age(i) : the difference (in days) between date link added and current date.
b : decay exponent (an ad-hoc value close to 0)
(full details of the formula can be found at http://blog.linkibol.com/2010/05/07/how-to-build-a-popularity-algorithm-you-can-be-proud-of/ - scroll down to the "How Do We Implement Popularity in linkibol?" section)
I can convert most of this function into PHP code easily, but the bit I'm not understanding is the sigma and deltarank bit. I'm not sure what that bit is supposed to do or what values to pass to k and m.
If anyone has any tips or could break the complex bit of the formula down that'd be great, then I can look at what would be the best way to implement it in PHP - there might be functions I could make use of etc..
They define the delta rank as the change in rank when the kth vote is cast on the mth link... it seems like that's arbitrary, since their rank change is based on the karma of the users casting the vote.
As for the sigma, it's just the sum of the contents from (k=1) to (k=whatever), so you'll implement that with a loop.
The sigma part is summation. Use the values provided as the loop counter. (so k=1 to 10, use those values in the functions that take k).
Delta is simply a difference, but that particular function surely has a more precise definition.
There are a few hundred of book records in the database and each record has a publish time. In the homepage of the website, I am required to write some codes to randomly pick 10 books and put them there. The requirement is that newer books need to have higher chances of getting displayed.
Since the time is an integer, I am thinking like this to calculate the probability for each book:
Probability of a book to be drawn = (current time - publish time of the book) / ((current time - publish time of the book1) + (current time - publish time of the book1) + ... (current time - publish time of the bookn))
After a book is drawn, the next round of the loop will minus the (current time - publish time of the book) from the denominator and recalculate the probability for each of the remaining books, the loop continues until 10 books have been drawn.
Is this algorithm a correct one?
By the way, the website is written in PHP.
Feel free to suggest some PHP codes if you have a better algorithm in your mind.
Many thanks to you all.
Here's a very similar question that may help: Random weighted choice The solution is in C# but the code is very readable and close to PHP syntax so it should be easy to adapt.
For example, here's how one could do this in MySQL:
First calculate the total age of all books and store it in a MySQL user variable:
SELECT SUM(TO_DAYS(CURDATE())-TO_DAYS(publish_date)) FROM books INTO #total;
Then choose books randomly, weighted by their age:
SELECT book_id FROM (
SELECT book_id, TO_DAYS(CURDATE())-TO_DAYS(publish_date) AS age FROM books
) b
WHERE book_id NOT IN (...list of book_ids chosen so far...)
AND RAND()*#total < b.age AND (#total:=#total-b.age)
ORDER BY b.publish_date DESC
LIMIT 10;
Note that the #total decreases only if a book has passed the random-selection test, because of short-circuiting of AND expressions.
This is not guaranteed to choose 10 books in one pass -- it's not even guaranteed to choose any books on a given pass. So you have to re-run the second step until you've found 10 books. The #total variable retains its decreased value so you don't have to recalculate it.
First off I think your formula will guarantee that earlier books get picked. Try to set your initial probabilities based on:
Age - days since publication
Max(Age) - oldest book in the sample
Book Age(i) - age of book i
... Prob (i) = [Max (age) + e - Book Age (i)] / sum over all i [ Max (age) + e - Book age(i) ]
The value e ensures that the oldest book has some probability of being selected. Now that that is done, you can always recalc the prob of any sample.
Now you have to find an UNBIASED way of picking books. Probably the best way would be to calculate the cumulative distribution using the above then pick a uniform (0,1) r.v. Find where that r.v. is in the cumulative distribution and pick the book nearest to it.
Can't help you on the coding. Make sense?
I'm making a digg-like website that is going to have a homepage with different categories. I want to display the most popular submissions.
Our rating system is simply "likes", like "I like this" and whatnot. We basically want to display the submissions with the highest number of "likes" per time. We want to have three categories: all-time popularity, last week, and last day.
Does anybody know of a way to help? I have no idea how to go about doing this and making it efficient. I thought that we could use some sort of cron-job to run every 10 minutes and pull in the number of likes per the last 10 minutes...but I've been told that's pretty inefficient?
Help?
Thanks!
Typically Digg and Reddit-like sites go by the date of the submission and not the times of the votes. This way all it takes is a simple SQL query to find the top submissions for X time period. Here's a pseudo-query to find the 10 most popular links from the past 24 hours using this method:
select * from submissions
where (current_time - post_time) < 86400
order by score desc limit 10
Basically, this query says to find all the submissions where the number of seconds between now and the time it was posted is less than 86400, which is 24 hours in UNIX time.
If you really want to measure popularity within X time interval, you'll need to store the post and time for every vote in another table:
create table votes (
post foreign key references submissions(id),
time datetime,
vote integer); -- +1 for upvote, -1 for downvote
Then you can generate a list of the most popular posts between X and Y times like so:
select sum(vote), post from votes
where X < time and time < Y
group by post
order by sum(vote) desc limit 10;
From here you're just a hop, skip, and inner join away from getting the post data tied to the returned ids.
Do you have a decent DB setup? Can we please hear about your CREATE TABLE details and indices? Assuming a sane setup, the DB should be able to pull the counts you require fast enough to suit your needs! For example (net of indices and keys, that somewhat depend on what DB engine you're using), given two tables:
CREATE TABLE submissions (subid INT, when DATETIME, etc etc)
CREATE TABLE likes (subid INT, when DATETIME, etc etc)
you can get the top 33 all-time popular submissions as
SELECT *, COUNT(likes.subid) AS score
FROM submissions
JOIN likes USING(subid)
GROUP BY submissions.subid
ORDER BY COUNT(likes.subid) DESC
LIMIT 33
and those voted for within a certain time range as
SELECT *, COUNT(likes.subid) AS score
FROM submissions
JOIN likes USING(subid)
WHERE likes.when BETWEEN initial_time AND final_time
GROUP BY submissions.subid
ORDER BY COUNT(likes.subid) DESC
LIMIT 33
If you were storing "votes" (positive or negative) in likes, instead of just counting each entry there as +1, you could simply use SUM(likes.vote) instead of the COUNTs.
For stable list like alltime, lastweek, because they are not supposed to change really fast so that I think you should save the list in your cache with expiration time is around 1 days or longer.
If you concern about correct count in real time, you can check at every page view by comparing the page with lowest page in the cache.
All you need to do is to care for synchronizing between the cache and actual database.
thethanghn
Queries where the order is some function of the current time can become real performance problems. Things get much simpler if you can bucket by calendar time and update scores for each bucket as people vote.
To complete nobody_'s answer I would suggest you read up on the documentation (if you are using MySQL of course).
I'd like to populate the homepage of my user-submitted-illustrations site with the "hottest" illustrations uploaded.
Here are the measures I have available:
How many people have favourited that illustration
votes table includes date voted
When the illustration was uploaded
illustration table has date created
Number of comments (not so good as max comments total about 10 at the moment)
comments table has comment date
I have searched around, but don't want user authority to play a part, but most algorithms include that.
I also need to find out if it's better to do the calculation in the MySQL that fetches the data or if there should be a PHP/cron method every hour or so.
I only need 20 illustrations to populate the home page. I don't need any sort of paging for this data.
How do I weight age against votes? Surely a site with less submission needs less weight on date added?
Many sites that use some type of popularity ranking do so by using a standard algorithm to determine a score and then decaying eternally over time. What I've found works better for sites with less traffic is a multiplier that gives a bonus to new content/activity - it's essentially the same, but the score stops changing after a period of time of your choosing.
For instance, here's a pseudo-example of something you might want to try. Of course, you'll want to adjust how much weight you're attributing to each category based on your own experience with your site. Comments are rare, but take more effort from the user than a favorite/vote, so they probably should receive more weight.
score = (votes / 10) + comments
age = UNIX_TIMESTAMP() - UNIX_TIMESTAMP(date_created)
if(age < 86400) score = score * 1.5
This type of approach would give a bonus to new content uploaded in the past day. If you wanted to approach this in a similar way only for content that had been favorited or commented on recently, you could just add some WHERE constraints on your query that grabs the score out from the DB.
There are actually two big reasons NOT to calculate this ranking on the fly.
Requiring your DB to fetch all of that data and do a calculation on every page load just to reorder items results in an expensive query.
Probably a smaller gotcha, but if you have a relatively small amount of activity on the site, small changes in the ranking can cause content to move pretty drastically.
That leaves you with either caching the results periodically or setting up a cron job to update a new database column holding this score you're ranking by.
Obviously there is some subjectivity in this - there's no one "correct" algorithm for determining the proper balance - but I'd start out with something like votes per unit age. MySQL can do basic math so you can ask it to sort by the quotient of votes over time; however, for performance reasons, it might be a good idea to cache the result of the query. Maybe something like
SELECT images.url FROM images ORDER BY (NOW() - images.date) / COUNT((SELECT COUNT(*) FROM votes WHERE votes.image_id = images.id)) DESC LIMIT 20
but my SQL is rusty ;-)
Taking a simple average will, of course, bias in favor of new images showing up on the front page. If you want to remove that bias, you could, say, count only those votes that occurred within a certain time limit after the image being posted. For images that are more recent than that time limit, you'd have to normalize by multiplying the number of votes by the time limit then dividing by the age of the image. Or alternatively, you could give the votes a continuously varying weight, something like exp(-time(vote) + time(image)). And so on and so on... depending on how particular you are about what this algorithm will do, it could take some experimentation to figure out what formula gives the best results.
I've no useful ideas as far as the actual agorithm is concerned, but in terms of implementation, I'd suggest caching the result somewhere, with a periodic update - if the resulting computation results in an expensive query, you probably don't want to slow your response times.
Something like:
(count favorited + k) * / time since last activity
The higher k is the less weight has the number of people having it favorited.
You could also change the time to something like the time it first appeared + the time of the last activity, this would ensure that older illustrations would vanish with time.