void euclid_lsh::neighbor_row_from_hash(
const bit_vector& bv,
float norm,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
// This function is not thread safe.
// Take lock out of this function.
jubatus::util::lang::shared_ptr<const column_table> table =
get_const_table();
const_bit_vector_column& bv_col = lsh_column();
const_float_column& norm_col = norm_column();
const float denom = bv.bit_num();
heap_t heap(ret_num);
jubatus::util::lang::function<heap_t(size_t, size_t)> f =
jubatus::util::lang::bind(
&ranking_hamming_bit_vectors_worker, &bv, &bv_col, &norm_col,
denom, norm, ret_num,
jubatus::util::lang::_1, jubatus::util::lang::_2);
ranking_hamming_bit_vectors_internal(
f, table->size_nolock(), threads_, heap);
vector<pair<float, size_t> > sorted;
heap.get_sorted(sorted);
ids.clear();
for (size_t i = 0; i < sorted.size(); ++i) {
ids.push_back(make_pair(
table->get_key_nolock(sorted[i].second), sorted[i].first));
}
}
void euclid_lsh::neighbor_row_from_hash(
const bit_vector& bv,
float norm,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
jubatus::util::lang::shared_ptr<const column_table> table = get_const_table();
jubatus::core::storage::fixed_size_heap<pair<float, size_t> > heap(ret_num);
{
const_bit_vector_column& bv_col = lsh_column();
const_float_column& norm_col = norm_column();
const float denom = bv.bit_num();
for (size_t i = 0; i < table->size(); ++i) {
const size_t hamm_dist = bv.calc_hamming_distance(bv_col[i]);
const float theta = hamm_dist * M_PI / denom;
const float score = norm_col[i] * (norm_col[i] - 2 * norm * cos(theta));
heap.push(make_pair(score, i));
}
}
vector<pair<float, size_t> > sorted;
heap.get_sorted(sorted);
ids.clear();
const float squared_norm = norm * norm;
for (size_t i = 0; i < sorted.size(); ++i) {
ids.push_back(make_pair(table->get_key(sorted[i].second),
sqrt(squared_norm + sorted[i].first)));
}
}
void nearest_neighbor_base::get_all_row_ids(vector<string>& ids) const {
vector<string> ret;
shared_ptr<const table::column_table> table = get_const_table();
ret.reserve(table->size());
for (size_t i = 0; i < table->size(); ++i) {
ret.push_back(table->get_key(i));
}
ret.swap(ids);
}
void euclid_lsh::neighbor_row(
const std::string& query_id,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
util::concurrent::scoped_rlock lk(get_const_table()->get_mutex());
/* table lock acquired; all subsequent table operations must be nolock */
const pair<bool, uint64_t> maybe_index =
get_const_table()->exact_match_nolock(query_id);
if (!maybe_index.first) {
ids.clear();
return;
}
const bit_vector bv = lsh_column()[maybe_index.second];
const float norm = norm_column()[maybe_index.second];
neighbor_row_from_hash(bv, norm, ids, ret_num);
}
void bit_vector_nearest_neighbor_base::neighbor_row_from_hash(
const bit_vector& query,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
vector<pair<uint64_t, float> > scores;
ranking_hamming_bit_vectors(query, bit_vector_column(), scores, ret_num);
jubatus::util::lang::shared_ptr<const column_table> table = get_const_table();
ids.clear();
for (size_t i = 0; i < scores.size(); ++i) {
ids.push_back(make_pair(table->get_key(scores[i].first), scores[i].second));
}
}
void bit_vector_nearest_neighbor_base::neighbor_row(
const string& query_id,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
const table::column_table& table = *get_const_table();
const pair<bool, uint64_t> maybe_index = table.exact_match(query_id);
if (!maybe_index.first) {
ids.clear();
return;
}
const_bit_vector_column& col = bit_vector_column();
neighbor_row_from_hash(col[maybe_index.second], ids, ret_num);
}
void euclid_lsh::neighbor_row(
const common::sfv_t& query,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
util::concurrent::scoped_rlock lk(get_const_table()->get_mutex());
/* table lock acquired; all subsequent table operations must be nolock */
neighbor_row_from_hash(
cosine_lsh(query, hash_num_, threads_, cache_),
l2norm(query),
ids,
ret_num);
}
void euclid_lsh::neighbor_row(
const std::string& query_id,
vector<pair<string, float> >& ids,
uint64_t ret_num) const {
const pair<bool, uint64_t> maybe_index =
get_const_table()->exact_match(query_id);
if (!maybe_index.first) {
ids.clear();
return;
}
const bit_vector bv = lsh_column()[maybe_index.second];
const float norm = norm_column()[maybe_index.second];
neighbor_row_from_hash(bv, norm, ids, ret_num);
}
void nearest_neighbor_base::pack(framework::packer& packer) const {
get_const_table()->pack(packer);
}
const_bit_vector_column& bit_vector_nearest_neighbor_base::bit_vector_column()
const {
return get_const_table()->get_bit_vector_column(bit_vector_column_id_);
}
const_float_column& euclid_lsh::norm_column() const {
return get_const_table()->get_float_column(first_column_id_ + 1);
}
const_bit_vector_column& euclid_lsh::lsh_column() const {
return get_const_table()->get_bit_vector_column(first_column_id_);
}