int ir_eval_tests()
{
system("rm -rf ceeaus-inv");
create_config("file");
auto idx = index::make_index<index::inverted_index, caching::splay_cache>(
"test-config.toml", uint32_t{10000});
index::okapi_bm25 ranker;
index::ir_eval eval{"test-config.toml"};
int num_failed = 0;
num_failed += testing::run_test("ir-eval", [&]()
{
for (size_t i = 0; i < 5; ++i)
{
corpus::document query{idx->doc_path(doc_id{i}), doc_id{0}};
auto ranking = ranker.score(*idx, query);
double f1 = eval.f1(ranking, query_id{i});
double p = eval.precision(ranking, query_id{i});
double r = eval.recall(ranking, query_id{i});
ASSERT(f1 >= 0 && f1 <= 1);
ASSERT(p >= 0 && p <= 1);
ASSERT(r >= 0 && r <= 1);
}
});
system("rm -rf ceeaus-inv test-config.toml");
return num_failed;
}
int ir_eval_bounds()
{
return testing::run_test(
"ir-eval-bounds", [&]()
{
system("rm -rf ceeaus-inv");
create_config("file");
auto idx = index::make_index<index::inverted_index,
caching::splay_cache>(
"test-config.toml", uint32_t{10000});
index::okapi_bm25 ranker;
index::ir_eval eval{"test-config.toml"};
// sanity test bounds
for (size_t i = 0; i < 5; ++i)
{
corpus::document query{idx->doc_path(doc_id{i}), doc_id{0}};
auto ranking = ranker.score(*idx, query);
auto f1 = eval.f1(ranking, query_id{i});
auto p = eval.precision(ranking, query_id{i});
auto r = eval.recall(ranking, query_id{i});
auto avg_p = eval.avg_p(ranking, query_id{i});
auto ndcg = eval.ndcg(ranking, query_id{i});
ASSERT(f1 >= 0 && f1 <= 1);
ASSERT(p >= 0 && p <= 1);
ASSERT(r >= 0 && r <= 1);
ASSERT(avg_p >= 0 && avg_p <= 1);
ASSERT(ndcg >= 0 && ndcg <= 1);
}
auto map = eval.map();
auto gmap = eval.gmap();
ASSERT(map >= 0 && map <= 1);
ASSERT(gmap >= 0 && gmap <= 1);
system("rm -rf ceeaus-inv test-config.toml");
});
}
std::string disk_index::doc_name(doc_id d_id) const
{
auto path = doc_path(d_id);
return path.substr(path.find_last_of("/") + 1);
}