您可以为此使用定制收集器：

Multimap<String, Foo> map = list.stream().collect(
    ImmutableMultimap::builder,
    (builder, value) -> value.getTags().forEach(tag -> builder.put(tag, value)),
    (builder1, builder2) -> builder1.putAll(builder2.build())
).build();

这不会引起额外的副作用（请参见此处），并且是并发的并且更加惯用。

您还可以将这些临时lambda提取到成熟的收集器中，如下所示：

public static <T, K> Collector<T, ?, Multimap<K, T>> toMultimapByKey(Function<? super T, ? extends Iterable<? extends K>> keysMapper) {
    return new MultimapCollector<>(keysMapper);
}

private static class MultimapCollector<T, K> implements Collector<T, ImmutableMultimap.Builder<K, T>, Multimap<K, T>> {
    private final Function<? super T, ? extends Iterable<? extends K>> keysMapper;

    private MultimapCollector(Function<? super T, ? extends Iterable<? extends K>> keysMapper) {
        this.keysMapper = keysMapper;
    }

    @Override
    public Supplier<ImmutableMultimap.Builder<K, T>> supplier() {
        return ImmutableMultimap::builder;
    }

    @Override
    public BiConsumer<ImmutableMultimap.Builder<K, T>, T> accumulator() {
        return (builder, value) -> keysMapper.apply(value).forEach(k -> builder.put(k, value));
    }

    @Override
    public BinaryOperator<ImmutableMultimap.Builder<K, T>> combiner() {
        return (b1, b2) -> b1.putAll(b2.build());
    }

    @Override
    public Function<ImmutableMultimap.Builder<K, T>, Multimap<K, T>> finisher() {
        return ImmutableMultimap.Builder<K, T>::build;
    }

    @Override
    public Set<Characteristics> characteristics() {
        return Collections.emptySet();
    }
}

然后该集合将如下所示：

Multimap<String, Foo> map = list.stream().collect(toMultimapByKey(Foo::getTags));

如果顺序对您不重要，则也可以EnumSet.of(Characteristics.UNORDERED)从characteristics()方法返回。这可以使内部收集机制更有效地运行，尤其是在并行缩减的情况下。