Refactored namespaces
This commit is contained in:
emeric
@@ -28,178 +28,174 @@
|
||||
|
||||
#include "som/DataNormalizer.hpp"
|
||||
#include "som/Network.hpp"
|
||||
#include "utils/Utils.hpp"
|
||||
#include "core/Utils.hpp"
|
||||
#include "IEngine.hpp"
|
||||
#include "FeaturesEngineCache.hpp"
|
||||
#include "FeaturesDefs.hpp"
|
||||
|
||||
namespace Database
|
||||
namespace lms::db
|
||||
{
|
||||
class Session;
|
||||
class Session;
|
||||
}
|
||||
|
||||
namespace Recommendation {
|
||||
|
||||
using FeatureWeight = double;
|
||||
|
||||
class FeaturesEngine : public IEngine
|
||||
namespace lms::recommendation
|
||||
{
|
||||
public:
|
||||
FeaturesEngine(Database::Db& db) : _db {db} {}
|
||||
using FeatureWeight = double;
|
||||
|
||||
FeaturesEngine(const FeaturesEngine&) = delete;
|
||||
FeaturesEngine(FeaturesEngine&&) = delete;
|
||||
FeaturesEngine& operator=(const FeaturesEngine&) = delete;
|
||||
FeaturesEngine& operator=(FeaturesEngine&&) = delete;
|
||||
class FeaturesEngine : public IEngine
|
||||
{
|
||||
public:
|
||||
FeaturesEngine(db::Db& db) : _db{ db } {}
|
||||
|
||||
static const FeatureSettingsMap& getDefaultTrainFeatureSettings();
|
||||
FeaturesEngine(const FeaturesEngine&) = delete;
|
||||
FeaturesEngine(FeaturesEngine&&) = delete;
|
||||
FeaturesEngine& operator=(const FeaturesEngine&) = delete;
|
||||
FeaturesEngine& operator=(FeaturesEngine&&) = delete;
|
||||
|
||||
private:
|
||||
void load(bool forceReload, const ProgressCallback& progressCallback) override;
|
||||
void requestCancelLoad() override;
|
||||
static const FeatureSettingsMap& getDefaultTrainFeatureSettings();
|
||||
|
||||
TrackContainer findSimilarTracksFromTrackList(Database::TrackListId tracklistId, std::size_t maxCount) const override;
|
||||
TrackContainer findSimilarTracks(const std::vector<Database::TrackId>& tracksId, std::size_t maxCount) const override;
|
||||
ReleaseContainer getSimilarReleases(Database::ReleaseId releaseId, std::size_t maxCount) const override;
|
||||
ArtistContainer getSimilarArtists(Database::ArtistId artistId, EnumSet<Database::TrackArtistLinkType> linkTypes, std::size_t maxCount) const override;
|
||||
private:
|
||||
void load(bool forceReload, const ProgressCallback& progressCallback) override;
|
||||
void requestCancelLoad() override;
|
||||
|
||||
void loadFromCache(FeaturesEngineCache&& cache);
|
||||
TrackContainer findSimilarTracksFromTrackList(db::TrackListId tracklistId, std::size_t maxCount) const override;
|
||||
TrackContainer findSimilarTracks(const std::vector<db::TrackId>& tracksId, std::size_t maxCount) const override;
|
||||
ReleaseContainer getSimilarReleases(db::ReleaseId releaseId, std::size_t maxCount) const override;
|
||||
ArtistContainer getSimilarArtists(db::ArtistId artistId, core::EnumSet<db::TrackArtistLinkType> linkTypes, std::size_t maxCount) const override;
|
||||
|
||||
// Use training (may be very slow)
|
||||
struct TrainSettings
|
||||
{
|
||||
std::size_t iterationCount {10};
|
||||
float sampleCountPerNeuron {4};
|
||||
FeatureSettingsMap featureSettingsMap;
|
||||
};
|
||||
void loadFromTraining(const TrainSettings& trainSettings, const ProgressCallback& progressCallback);
|
||||
void loadFromCache(FeaturesEngineCache&& cache);
|
||||
|
||||
template <typename IdType>
|
||||
using ObjectPositions = std::unordered_map<IdType, std::vector<SOM::Position>>;
|
||||
// Use training (may be very slow)
|
||||
struct TrainSettings
|
||||
{
|
||||
std::size_t iterationCount{ 10 };
|
||||
float sampleCountPerNeuron{ 4 };
|
||||
FeatureSettingsMap featureSettingsMap;
|
||||
};
|
||||
void loadFromTraining(const TrainSettings& trainSettings, const ProgressCallback& progressCallback);
|
||||
|
||||
using ArtistPositions = ObjectPositions<Database::ArtistId>;
|
||||
using ReleasePositions = ObjectPositions<Database::ReleaseId>;
|
||||
using TrackPositions = ObjectPositions<Database::TrackId>;
|
||||
template <typename IdType>
|
||||
using ObjectPositions = std::unordered_map<IdType, std::vector<som::Position>>;
|
||||
|
||||
template <typename IdType>
|
||||
using ObjectMatrix = SOM::Matrix<std::vector<IdType>>;
|
||||
using ArtistMatrix = ObjectMatrix<Database::ArtistId>;
|
||||
using ReleaseMatrix = ObjectMatrix<Database::ReleaseId>;
|
||||
using TrackMatrix = ObjectMatrix<Database::TrackId>;
|
||||
using ArtistPositions = ObjectPositions<db::ArtistId>;
|
||||
using ReleasePositions = ObjectPositions<db::ReleaseId>;
|
||||
using TrackPositions = ObjectPositions<db::TrackId>;
|
||||
|
||||
void load(const SOM::Network& network, const TrackPositions& tracksPosition);
|
||||
template <typename IdType>
|
||||
using ObjectMatrix = som::Matrix<std::vector<IdType>>;
|
||||
using ArtistMatrix = ObjectMatrix<db::ArtistId>;
|
||||
using ReleaseMatrix = ObjectMatrix<db::ReleaseId>;
|
||||
using TrackMatrix = ObjectMatrix<db::TrackId>;
|
||||
|
||||
FeaturesEngineCache toCache() const;
|
||||
void load(const som::Network& network, const TrackPositions& tracksPosition);
|
||||
|
||||
template <typename IdType>
|
||||
static std::vector<SOM::Position> getMatchingRefVectorsPosition(const std::vector<IdType>& ids, const ObjectPositions<IdType>& objectPositions);
|
||||
FeaturesEngineCache toCache() const;
|
||||
|
||||
template <typename IdType>
|
||||
static std::vector<IdType> getObjectsIds(const std::vector<SOM::Position>& positions, const ObjectMatrix<IdType>& objectsMatrix);
|
||||
template <typename IdType>
|
||||
static std::vector<som::Position> getMatchingRefVectorsPosition(const std::vector<IdType>& ids, const ObjectPositions<IdType>& objectPositions);
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<IdType> getSimilarObjects(const std::vector<IdType>& ids,
|
||||
const ObjectMatrix<IdType>& objectMatrix,
|
||||
const ObjectPositions<IdType>& objectPositions,
|
||||
std::size_t maxCount) const;
|
||||
template <typename IdType>
|
||||
static std::vector<IdType> getObjectsIds(const std::vector<som::Position>& positions, const ObjectMatrix<IdType>& objectsMatrix);
|
||||
|
||||
Database::Db& _db;
|
||||
bool _loadCancelled {};
|
||||
std::unique_ptr<SOM::Network> _network;
|
||||
double _networkRefVectorsDistanceMedian {};
|
||||
template <typename IdType>
|
||||
std::vector<IdType> getSimilarObjects(const std::vector<IdType>& ids,
|
||||
const ObjectMatrix<IdType>& objectMatrix,
|
||||
const ObjectPositions<IdType>& objectPositions,
|
||||
std::size_t maxCount) const;
|
||||
|
||||
ArtistPositions _artistPositions;
|
||||
std::unordered_map<Database::TrackArtistLinkType, ArtistMatrix> _artistMatrix;
|
||||
db::Db& _db;
|
||||
bool _loadCancelled{};
|
||||
std::unique_ptr<som::Network> _network;
|
||||
double _networkRefVectorsDistanceMedian{};
|
||||
|
||||
ReleasePositions _releasePositions;
|
||||
ReleaseMatrix _releaseMatrix;
|
||||
ArtistPositions _artistPositions;
|
||||
std::unordered_map<db::TrackArtistLinkType, ArtistMatrix> _artistMatrix;
|
||||
|
||||
TrackPositions _trackPositions;
|
||||
TrackMatrix _trackMatrix;
|
||||
};
|
||||
ReleasePositions _releasePositions;
|
||||
ReleaseMatrix _releaseMatrix;
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<SOM::Position>
|
||||
FeaturesEngine::getMatchingRefVectorsPosition(const std::vector<IdType>& ids, const ObjectPositions<IdType>& objectPositions)
|
||||
{
|
||||
std::vector<SOM::Position> res;
|
||||
TrackPositions _trackPositions;
|
||||
TrackMatrix _trackMatrix;
|
||||
};
|
||||
|
||||
if (ids.empty())
|
||||
return res;
|
||||
template <typename IdType>
|
||||
std::vector<som::Position> FeaturesEngine::getMatchingRefVectorsPosition(const std::vector<IdType>& ids, const ObjectPositions<IdType>& objectPositions)
|
||||
{
|
||||
std::vector<som::Position> res;
|
||||
|
||||
for (const IdType id : ids)
|
||||
{
|
||||
auto it = objectPositions.find(id);
|
||||
if (it == objectPositions.end())
|
||||
continue;
|
||||
if (ids.empty())
|
||||
return res;
|
||||
|
||||
for (const SOM::Position& position : it->second)
|
||||
Utils::push_back_if_not_present(res, position);
|
||||
}
|
||||
for (const IdType id : ids)
|
||||
{
|
||||
auto it = objectPositions.find(id);
|
||||
if (it == objectPositions.end())
|
||||
continue;
|
||||
|
||||
return res;
|
||||
for (const som::Position& position : it->second)
|
||||
core::utils::push_back_if_not_present(res, position);
|
||||
}
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<IdType> FeaturesEngine::getObjectsIds(const std::vector<som::Position>& positions, const ObjectMatrix<IdType>& objectMatrix)
|
||||
{
|
||||
std::vector<IdType> res;
|
||||
|
||||
for (const som::Position& position : positions)
|
||||
{
|
||||
for (const IdType id : objectMatrix.get(position))
|
||||
core::utils::push_back_if_not_present(res, id);
|
||||
}
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<IdType> FeaturesEngine::getSimilarObjects(const std::vector<IdType>& ids,
|
||||
const ObjectMatrix<IdType>& objectMatrix,
|
||||
const ObjectPositions<IdType>& objectPositions,
|
||||
std::size_t maxCount) const
|
||||
{
|
||||
std::vector<IdType> res;
|
||||
|
||||
std::vector<som::Position> searchedRefVectorsPosition{ getMatchingRefVectorsPosition(ids, objectPositions) };
|
||||
if (searchedRefVectorsPosition.empty())
|
||||
return res;
|
||||
|
||||
while (1)
|
||||
{
|
||||
std::vector<IdType> closestObjectIds{ getObjectsIds(searchedRefVectorsPosition, objectMatrix) };
|
||||
|
||||
// Remove objects that are already in input or already reported
|
||||
closestObjectIds.erase(std::remove_if(std::begin(closestObjectIds), std::end(closestObjectIds),
|
||||
[&](IdType id)
|
||||
{
|
||||
return std::find(std::cbegin(ids), std::cend(ids), id) != std::cend(ids);
|
||||
})
|
||||
, std::end(closestObjectIds));
|
||||
|
||||
for (IdType id : closestObjectIds)
|
||||
{
|
||||
if (res.size() == maxCount)
|
||||
break;
|
||||
|
||||
core::utils::push_back_if_not_present(res, id);
|
||||
}
|
||||
|
||||
if (res.size() == maxCount)
|
||||
break;
|
||||
|
||||
// If there is not enough objects, try again with closest neighbour until there is too much distance
|
||||
const std::optional<som::Position> closestRefVectorPosition{ _network->getClosestRefVectorPosition(searchedRefVectorsPosition, _networkRefVectorsDistanceMedian * 0.75) };
|
||||
if (!closestRefVectorPosition)
|
||||
break;
|
||||
|
||||
core::utils::push_back_if_not_present(searchedRefVectorsPosition, closestRefVectorPosition.value());
|
||||
}
|
||||
|
||||
return res;
|
||||
}
|
||||
}
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<IdType>
|
||||
FeaturesEngine::getObjectsIds(const std::vector<SOM::Position>& positions, const ObjectMatrix<IdType>& objectMatrix)
|
||||
{
|
||||
std::vector<IdType> res;
|
||||
|
||||
for (const SOM::Position& position : positions)
|
||||
{
|
||||
for (const IdType id : objectMatrix.get(position))
|
||||
Utils::push_back_if_not_present(res, id);
|
||||
}
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
template <typename IdType>
|
||||
std::vector<IdType>
|
||||
FeaturesEngine::getSimilarObjects(const std::vector<IdType>& ids,
|
||||
const ObjectMatrix<IdType>& objectMatrix,
|
||||
const ObjectPositions<IdType>& objectPositions,
|
||||
std::size_t maxCount) const
|
||||
{
|
||||
std::vector<IdType> res;
|
||||
|
||||
std::vector<SOM::Position> searchedRefVectorsPosition {getMatchingRefVectorsPosition(ids, objectPositions)};
|
||||
if (searchedRefVectorsPosition.empty())
|
||||
return res;
|
||||
|
||||
while (1)
|
||||
{
|
||||
std::vector<IdType> closestObjectIds {getObjectsIds(searchedRefVectorsPosition, objectMatrix)};
|
||||
|
||||
// Remove objects that are already in input or already reported
|
||||
closestObjectIds.erase(std::remove_if(std::begin(closestObjectIds), std::end(closestObjectIds),
|
||||
[&](IdType id)
|
||||
{
|
||||
return std::find(std::cbegin(ids), std::cend(ids), id) != std::cend(ids);
|
||||
})
|
||||
, std::end(closestObjectIds));
|
||||
|
||||
for (IdType id : closestObjectIds)
|
||||
{
|
||||
if (res.size() == maxCount)
|
||||
break;
|
||||
|
||||
Utils::push_back_if_not_present(res, id);
|
||||
}
|
||||
|
||||
if (res.size() == maxCount)
|
||||
break;
|
||||
|
||||
// If there is not enough objects, try again with closest neighbour until there is too much distance
|
||||
const std::optional<SOM::Position> closestRefVectorPosition {_network->getClosestRefVectorPosition(searchedRefVectorsPosition, _networkRefVectorsDistanceMedian * 0.75)};
|
||||
if (!closestRefVectorPosition)
|
||||
break;
|
||||
|
||||
Utils::push_back_if_not_present(searchedRefVectorsPosition, closestRefVectorPosition.value());
|
||||
}
|
||||
|
||||
return res;
|
||||
}
|
||||
|
||||
} // ns Recommendation
|
||||
|
||||
Reference in New Issue
Block a user