+1 vote
ago by
We have a game, its rating is 66.0% and a second game, its rating is 99.9%.

That's just the rating is calculated likes - dislikes.

First game, 10,000,000 likes, and 3,400,000 dislikes.
The second game, 1,000 likes and 1 disliked.

Sorting by rating becomes useless.

What are the alternatives? Ranking? Weight of usefulness, significance? How to count?

The question is not about how to produce relevant results, but how to assess whether the content is useful to the community and how much.
That is clearly the first game is 10,000+ times more useful to the community.
(dislikes are not counted 10,000,000 / 1,000)

If this is a game, the parameters of usefulness would be approximately
- How many people have installed
- And how many bought
- How much time is spent in the game
- How many people have left a positive review of the game
- How many now play, for example, an old game, but it is played, it means it is interesting, content, updates

In the case of video
- how many views
- likes (no dislikes)
- whether the user has watched all the videos, or how much time
- to see if the user returned to the video
- whether the user shared the link to the video via social networks

Books, manga, comics
- views
- How much each user has read shows the degree of interest

- How much listens
- How many were listening
- how many users have repeated, that is, when a user listens again, or plays a game, or watches a video, it means the content is interesting

What other parameters can you think of?
How would you rate the usefulness of the content?

What's Googling

content that adds valuevaluable contentcalcualte content valueestimate contents value

and other similar things.

UPD interesting algorithm reddit a little more about personalization

( Total Reviews = Positive Reviews + Negative Reviews )( Review Score = frac{Positive Reviews}{Total Reviews} )( Rating = Review Score - (Review Score - 0.5)*2^{-log_{10}(Total Reviews + 1)} )function GetRating( $positiveVotes, $negativeVotes ) { $totalVotes = $positiveVotes + $negativeVotes; $average = $positiveVotes / $totalVotes; $score = $average - ( $average - 0.5 ) * 2 ** -log10( $totalVotes + 1 ); return $score * 100;}
ago by
And how did you come to the conclusion that the first game was more useful than the second? Put your thoughts in the form of an algorithm, and there will be no such questions.
ago by
BonBon Slick Now you've added to your question and got an answer for yourself.)
Now it remains to understand why we need to know the usefulness of the content for the entire community, not just its consumers and poatform
ago by
Vasily Bannikov , little-used content, is less useful to the community as a whole. For those few hundred, yes, but not for the most part. Sorting by rating is useless. So I sorted the games in my library by rating, and what do I see? A game from 1996 with a rating of 98%, Witcher 3 has a rating of 96%. Witcher 3 has 10,000,000 reviews, and that niche game, 1,000 likes, the sorting is broken.
The game was good for the 90s, now it's trash, according to my observations, myself the same, the old games get a nickel out of nostalgia. Playing games that are the same as 20-25 years ago is nonsense. Do not believe me? Let a schoolboy play TES Morrowind or Deus X 1 the same Halva Life 1 in the original, cheesy, and then ask his opinion.
What they sell Halva is all remastered.
This is all my thinking how to calculate the usefulness of the content, since the user can post it and would like to remove useless.
ago by
BonBon Slick Well, then count only the number of grades, and give the ratio purely as a reference.
Really, I don't know who would benefit from such a rating.
Imho, personalization is always more valuable.
And if, for example, I really like the second, less popular variant, it will be more useful to me than the first, popular
ago by
Why useless?
The first can be characterized as "steamed turd.
And the second apparently some very niche and very good for everyone who played it.
ago by
Dmitry Belyaev , widened the question
ago by
BonBon Slick Now for the big question.
Who needs such a service, which stores only pops content?
It would be like yt would only have a "trends" section, and there would be little point in subscribing to small or downgraded channels, because they would be deleted anyway.
Or if all the old and indie games disappeared from steam.

Each user falls into a particular niche audience, and if suddenly the service began to remove the content that he likes - the user will simply go to a competitor who does not live on disk space.
ago by
Vasily Bannikov personalization is the importance of content for each user. It is counted by other formulas. It's about the content that brings the most traffic, users and has the most impact, brings the most value to the resource and the community.
Everyone is unrealistic to please and store dozens of petabytes just to give every ten years, a crumbling article, a game, because 1 user it suits his personal preferences, utopia.
ago by
Vasily Bannikov , what to find out? Metrics, analytics. Little popular content or unused content, go away. Or let it pay for storage, but it will not appear in search.
Get rid of unnecessary garbage, irrelevant, outdated content.
Look at the first youtube video from 2005, it's unviewable.
What is the purpose of keeping such a thing? In my case, I would like to delete, there is little point in storing such garbage, if the client does not pay for it, will only litter the server, even if removed from the search. But here's how to determine whether it is worth it to remove or not, what is its value and relevance to society.
The review left 20 years ago is clearly outdated today. The game was great, 20 years ago, it's in the past. The same goes for the likes and dislikes. If no one downloads or plays it, what's the point of keeping this garbage? Is it a dump or what? Not enough trash?

2 Answers

0 votes
ago by
Best answer
You forget that "communities" are different. Each game has its own target audience. And these audiences may not overlap at all. So a great benefit to the first audience will mean nothing to the second audience.

If, for example, likes are possible only after you buy/download the game, the percentage of likes reflects not the usefulness, but justified expectations. That is, if you write in the description that the game sucks, although the game is not so bad, it will download 1001 people, of which 1000 people will be pleasantly surprised, and will put a "like" (for simplicity we will assume that all assess, no one ignores the "like" button). And if your description says "super game", it will be downloaded by 13,400,000 people, a decent proportion of whom will be disappointed in the description and screenshots of the game, and as a result will set the disliked.

You can present your game as a special niche product at a high price. And fans of this particular rare genre will buy. Let them be few, but there will be no alternatives to your game in this niche, which is very profitable. If the game will fit the description and promises + small bonus, of which the description is silent, the percentage of likes will be high. And those who are not interested in the genre, just pass by, because the description immediately understand that they will not get the game, and the dislikes from them you do not get.

ExampleOne of my acquaintances did a video on hydrogen liquefaction in the game ONI. People came in and wrote angry comments that they were looking for information on physics, and here was some gamer with his silly game that had nothing to do with reality. An acquaintance added the name of the game itself to the title of the video, and the conversions from people uninterested in this content stopped. The percentage of likes started to grow. Although the content did not change at all. Strange, isn't it? The content is the same, but the "usefulness" suddenly went up because of the title. That's because it wasn't usefulness that went up.
+2 votes
ago by
All that is listed is the wild ravings of a schoolboy. Determine what the business requirements for this functionality and what it is more profitable to give the user in the first positions. From this already choose metrics and sort by them. even use parrots and put them by hand I do not know