Recommendation Diversity Calculator

Work out recommendation diversity instantly with clear inputs, formula shown and shareable results.

About Recommendation Diversity Calculator

Intra-list diversity is one minus the average pairwise similarity across the n(n-1)/2 pairs in a recommendation list. A list of ten near-identical products scores near zero and feels repetitive even when every item is individually well-ranked. Category spread is a complementary, coarser measure that is easier to explain to merchandising teams and easier to constrain in a ranker.

Formula

Intra-list diversity

pairs = n(n-1)/2; diversity = 1 - average pairwise similarity; category ratio = distinct categories / list size

Worked example

With the default values below, Recommendation Diversity Calculator returns Intra-list diversity (1 - similarity) of 0.5800. Change any input above and every figure here is recalculated in your browser.

Inputs used

Items in the recommendation listaccepts 2–100: 10
Average pairwise similarityaccepts 0–1: 0.42
Distinct categories in the listminimum 1: 4

Results

Intra-list diversity (1 - similarity): 0.5800
Item pairs in the list: 45
Distinct categories per slot: 0.400
Average redundancy between items: 42.00%

Step by step

1Values used
2Intra-list diversity
3Intra-list diversity (1 - similarity)= 0.5800
4Item pairs in the list= 45
5Distinct categories per slot= 0.400
6Average redundancy between items= 42.00

Assumptions and limits

The result is derived only from the 3 inputs listed above. Anything not among them is not part of the calculation.
In particular, this calculator does not take inflation, tax, fees or charges or compounding frequency into account.
Figures are rounded for display to 4 decimal places, so totals built from several rounded parts can differ slightly from the unrounded arithmetic.

How these calculators are built and tested is described on the Methodology page. If a figure looks wrong, please report it.

Frequently Asked Questions

Does more diversity reduce accuracy?

There is a trade-off, but a small diversity constraint usually costs little precision and measurably improves engagement and satisfaction. Maximal marginal relevance is the standard way to tune it.

Which similarity should be used?

Whatever matches the user's notion of sameness — content embeddings, category taxonomy or co-purchase similarity. The metric is only as meaningful as that choice.