private ProjectedDatabase ConstructProjectedDatabase(Sequence prefix, ProjectedDatabase sourceProjectedDatabase)
{
var prefixLastItem = prefix.LastItem;
var projectedDatabase = new ProjectedDatabase(prefix);
// The count of sequences in a database which dont have to contains prefix
// and projected database can still contains some sequential pattern.
int minSequencesInProjectedDb = sourceProjectedDatabase.Size - (int)_minSupport;
for (int sid = 0, lastSid = sourceProjectedDatabase.Size; sid < lastSid; sid++)
{
// If sequence contains the last item of a prefix, project sequence and add it to db.
int itemsetIndex;
if ((itemsetIndex = sourceProjectedDatabase[sid].Contains(prefixLastItem)) >= 0)
{
var pseudoSequence = PseudoSequence.MakeProjection(sourceProjectedDatabase[sid], prefix, itemsetIndex);
if (pseudoSequence != null && pseudoSequence.IsNotEmpty)
{
projectedDatabase.Add(pseudoSequence);
continue;
}
}
// If projected database can not contains new sequential pattern stop projection sooner.
if (--minSequencesInProjectedDb < 0)
{
return(null);
}
}
return(projectedDatabase);
}
/// <summary>
/// Sets the first instance of prefix in the pseudo sequence.
/// </summary>
/// <param name="sequence"></param>
private void SetFirsInstanceOfPrefix(PseudoSequence sequence)
{
_firstInstances = new List <int>(sequence._firstInstances);
if (_prefix.IsLastItemIExtension)
{
_firstInstances[_firstInstances.Count - 1] = _isLastItemIExtension ? _offset : _offset - 1;
}
else
{
_firstInstances.Add(_isLastItemIExtension ? _offset : _offset - 1);
}
}
/// <summary>
/// Initializes a new instance of the PseudoSequence class.
/// </summary>
/// <param name="isLastItemIExtension">If the last item of the prefix was append by I-Step. </param>
public PseudoSequence(Sequence prefix, PseudoSequence sequence, int offset, bool isLastItemIExtension)
{
_offset = offset;
_prefix = prefix;
_items = sequence._items;
_isLastItemIExtension = isLastItemIExtension;
if (offset < _items.Length)
{
_firstItemset = _items[offset];
}
SetFirsInstanceOfPrefix(sequence);
}
/// <summary>
/// Initializes a new instance of the PseudoSequence class.
/// </summary>
/// <param name="isLastItemIExtension">If the last item of the prefix was append by I-Step. </param>
public PseudoSequence(Sequence prefix, PseudoSequence sequence, int offset, bool isLastItemIExtension)
{
_offset = offset;
_prefix = prefix;
_items = sequence._items;
_isLastItemIExtension = isLastItemIExtension;
if (offset < _items.Length)
{
_firstItemset = _items[offset];
}
SetFirsInstanceOfPrefix(sequence);
}
/// <summary>
/// Initializes a new instance of the PseudoSequence class.
/// </summary>
/// <param name="isLastItemIExtension">If the last item of the prefix was append by I-Step. </param>
public PseudoSequence(Sequence prefix, PseudoSequence sequence, int offset, uint[] firstItemset, bool isLastItemIExtension)
: this(prefix, sequence, offset, isLastItemIExtension)
{
_firstItemset = firstItemset;
}
/// <summary>
/// Sets the first instance of prefix in the pseudo sequence.
/// </summary>
/// <param name="sequence"></param>
private void SetFirsInstanceOfPrefix(PseudoSequence sequence)
{
_firstInstances = new List<int>(sequence._firstInstances);
if (_prefix.IsLastItemIExtension)
{
_firstInstances[_firstInstances.Count - 1] = _isLastItemIExtension ? _offset : _offset - 1;
}
else
{
_firstInstances.Add(_isLastItemIExtension ? _offset : _offset - 1);
}
}
/// <summary>
/// Creates pseudo-projected sequence from a given sequence,
/// the projection will be start at the first occurence of the last prefix item.
/// </summary>
/// <param name="firstItemsetIndex">An index of the first itemset, where should be started finding occurence of the last prefix item.</param>
public static PseudoSequence MakeProjection(PseudoSequence sequence, Sequence prefix, int firstItemsetIndex = 0, bool removeEmpty = true)
{
// Two types of prefix last item: y and (_y).
// Rules for projection:
// 1) y: xyz... => z...
// 2) y: (xyz)... => (_z)...
// 3) (_y): (vyz)... => (_z)...
// 4) (_y): (_xyz)... => (_z)...
// Where x,z can be zero, one or more items and v one or more items,
// and v union y contains all items from the last itemset of prefix.
var prefixLastItem = prefix.LastItem;
var lastItemsetIndex = sequence.Size - 1;
var prefixLastItemset = prefix[prefix.Size - 1];
var isPrefixLastItemsetIExtensions = prefix.IsLastItemIExtension;
// Finds last item of prefix and creates projection.
for (int tid = firstItemsetIndex, lastTid = sequence.Size; tid < lastTid; tid++)
{
var isActualItemsetPostfix = (tid == 0 && sequence.IsLastItemIExtension);
// Projection is not allow in situation: (x): (_x).... => {}
if (!isPrefixLastItemsetIExtensions && isActualItemsetPostfix) continue;
var actualItemset = sequence[tid];
var lastItemIndexInActualItemset = actualItemset.Length - 1;
// Every item in an actual itemset compare with prefix last item.
for (var itemId = 0; itemId < actualItemset.Length; itemId++)
{
if (actualItemset[itemId] != prefixLastItem) continue;
// If is not the end of a sequence makes projection, otherwise stop.
if (removeEmpty && tid == lastItemsetIndex && itemId == lastItemIndexInActualItemset) return null;
// 1) y: xyz... => z...,
// 2) y: (xyz)... => (_z)...
if (!isPrefixLastItemsetIExtensions ||
// OR 4) (_y): (_xyz)... => (_z)...
isActualItemsetPostfix ||
// OR 3) (_y): (vyz)... => (_z)...
prefixLastItemset.IsSubsetOf(actualItemset, itemId))
{
return itemId == lastItemIndexInActualItemset
?
new PseudoSequence(prefix, sequence, sequence._offset + tid + 1, false)
:
new PseudoSequence(prefix, sequence, sequence._offset + tid, sequence[tid].Slice(itemId + 1), true);
}
}
}
return null;
}
/// <summary>
/// Initializes a new instance of the PseudoSequence class.
/// </summary>
/// <param name="isLastItemIExtension">If the last item of the prefix was append by I-Step. </param>
public PseudoSequence(Sequence prefix, PseudoSequence sequence, int offset, uint[] firstItemset, bool isLastItemIExtension)
: this(prefix, sequence, offset, isLastItemIExtension)
{
_firstItemset = firstItemset;
}
/// <summary>
/// Creates pseudo-projected sequence from a given sequence,
/// the projection will be start at the first occurence of the last prefix item.
/// </summary>
/// <param name="firstItemsetIndex">An index of the first itemset, where should be started finding occurence of the last prefix item.</param>
public static PseudoSequence MakeProjection(PseudoSequence sequence, Sequence prefix, int firstItemsetIndex = 0, bool removeEmpty = true)
{
// Two types of prefix last item: y and (_y).
// Rules for projection:
// 1) y: xyz... => z...
// 2) y: (xyz)... => (_z)...
// 3) (_y): (vyz)... => (_z)...
// 4) (_y): (_xyz)... => (_z)...
// Where x,z can be zero, one or more items and v one or more items,
// and v union y contains all items from the last itemset of prefix.
var prefixLastItem = prefix.LastItem;
var lastItemsetIndex = sequence.Size - 1;
var prefixLastItemset = prefix[prefix.Size - 1];
var isPrefixLastItemsetIExtensions = prefix.IsLastItemIExtension;
// Finds last item of prefix and creates projection.
for (int tid = firstItemsetIndex, lastTid = sequence.Size; tid < lastTid; tid++)
{
var isActualItemsetPostfix = (tid == 0 && sequence.IsLastItemIExtension);
// Projection is not allow in situation: (x): (_x).... => {}
if (!isPrefixLastItemsetIExtensions && isActualItemsetPostfix)
{
continue;
}
var actualItemset = sequence[tid];
var lastItemIndexInActualItemset = actualItemset.Length - 1;
// Every item in an actual itemset compare with prefix last item.
for (var itemId = 0; itemId < actualItemset.Length; itemId++)
{
if (actualItemset[itemId] != prefixLastItem)
{
continue;
}
// If is not the end of a sequence makes projection, otherwise stop.
if (removeEmpty && tid == lastItemsetIndex && itemId == lastItemIndexInActualItemset)
{
return(null);
}
// 1) y: xyz... => z...,
// 2) y: (xyz)... => (_z)...
if (!isPrefixLastItemsetIExtensions ||
// OR 4) (_y): (_xyz)... => (_z)...
isActualItemsetPostfix ||
// OR 3) (_y): (vyz)... => (_z)...
prefixLastItemset.IsSubsetOf(actualItemset, itemId))
{
return(itemId == lastItemIndexInActualItemset
?
new PseudoSequence(prefix, sequence, sequence._offset + tid + 1, false)
:
new PseudoSequence(prefix, sequence, sequence._offset + tid, sequence[tid].Slice(itemId + 1), true));
}
}
}
return(null);
}
