The system picks front-page recommendations by randomly selecting a few games with the highest ratings, excluding games that you've told us you've already played. @@ -36,35 +30,6 @@ confidence in the game, by adding five "fake" ratings to the average (one 1-star, one 2-star, one 3-star, one 4-star, and one 5-star rating) and subtracting the standard deviation from the result. - - -
The system picks front-page recommendations by looking for other -members with similar patterns of likes and dislikes to your own, as -expressed in the ratings they gave to the games they reviewed. IFDB -tries to match you up with a few other members, then recommends games -that those other members rated highly. - -
In principle, the more ratings you and other members provide, -the more accurate the matching will become. So the recommendations -should get better and better as you rate more games. - -
This approach is sometimes called "collaborative filtering." Some -people think it's great, others are skeptical. The obvious objection -is that it doesn't capture the reasons that you like the games -you like, so it might match you up with someone who happens to like -some of the same games, but for completely different reasons. There's -obviously no guarantee that the approach will actually produce good -advice, but we hope that it gives you at least a few leads on games -that you might otherwise have overlooked. - -
If the algorithmic recommendations on the home page don't work for -you, remember that IFDB still offers several ways to get -personal recommendations from other users, such as member -reviews and Recommended lists. -