C# Class StopGuessing.DataStructures.Sketch

A sketch is a probabilistic data structure that provides an approximate count of the number of times an item has been observed or, more generally, the sum of a set of numbers associated with each item (where the common case is that each observation of an item is the number 1). One can add observations of items (or numbers associated with items) and get an estimate of that number. A sketch always returns a number that is _at least_ as large as the number observed (up to the maximum number that the sketch can store). In other words, it is a lower bound on the values observed. It may falsely return a number too large, but will never return a number too small (unless, again, the value has exceeded the maximum integer the sketch is designe do store.)

ファイルを表示 Open project: Microsoft/StopGuessing

Public Methods

Method	Description
Add ( string s, ulong amountToAdd = 1 ) : ResultOfUpdate	An Add function that is equivalent to calling `increment` multiple (amountToAdd) times. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.
ConservativeAdd ( string s, ulong amountToAdd = 1 ) : ResultOfUpdate	An Add function that is equivalent to calling `conservativeIncrement` multiple (amountToAdd) times. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.
ConservativeIncrement ( string s ) : ResultOfUpdate	Count the observation of the string s by adding one to those values indexed by calling `getIndexesForString(s) for which the value is currently equal to getMin(s). This algorithm is more conservative than increment in that it is less likely to result in overcounting. If the count is already MaxValue, the operation will have no effect.`
Get ( long elementIndexes ) : ResultOfGet	Get the current state of the sketch for a value's row indexes into each column.
Get ( string s ) : ResultOfGet	Get the current state of the sketch for a value's row indexes into each column.
GetColumnTotal ( int column ) : ulong	Get the total value stored in a columnn
GetMin ( string s ) : ulong	Get the minimum of all the values at the indexes identified for the string s with the sketch. This is a lower bound on the number of times the string s has been witnessed by the sketch via a call to increment(s) and/or conservativeIncrement(s). However, counting stops when the maximum value storable in the sketch is reached.
Increment ( string s ) : ResultOfUpdate	Count the observation of the string s by adding one to the value in each table indexed by calling `getIndexesForString(s)` If the count is already MaxValue, the operation will have no effect.
IsNonZero ( string s ) : bool	Test whether a string s has never be witnessed before by seeing if `getMin(s) > 0`. This is guaranteed to return true if increment(s) or conservativeIncrement(s) has been called on the sketch. It is probabilistically likely, but not guaranteed, to return false neither increment(s) or conservativeIncrement(s) have been called.
Sketch ( long numberOfColumns, long numberOfRows, int bitsPerElement ) : System	A sketch can be viewed as either having k tables (one for each hash) of n elements or as a two-dimensional array of k columns * n rows. This constructor creates a Sketch of size specified by k, n, and the number of bits per element.
this ( long elementIndexes ) : ulong	A shortcut to set all the elements of the sketch identified by an array of indexes, or to Get the minimum of all the values at the indexes identified by the array of indexes.
this ( string s ) : ulong	A shortcut to set all the elements of the sketch identified by indexes for the string s or to Get the minimum of all the values at the indexes identified for the string s (equivalent to calling `getMin(s)`.

Protected Methods

Method	Description
Add ( long elementIndexForEachColumn, ulong amountToAdd = 1 ) : ResultOfUpdate	Add to the the value at each [column,index] pair, where the index into elementIndexForEachColumnis the column number and the value is the row (element) index. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.
ConservativeAdd ( long elementIndexForEachColumn, ulong amountToAdd = 1 ) : ResultOfUpdate	Use the conservative algorithm to Add to the data structure such that for each [column][row], the value stored is no less than `getMin(elementIndexForEachColumn) + amountToAdd` (but never exceeding MaxValue).
GetIndexesForString ( string s ) : long[]	Given a string s, provides an index into each column of the sketch. Each index identifies a row (element) to which one can write (if adding members/counts) or read (if testing membership or min counts). SECURITY NOTE: This indexes can be predicted by an adversary. If using the sketch in a scenario in which an adversary may cause harm by targeting certain indexes, include as a prefix of the parameter s a key that is not publicly-known (and hopefully can be kept secret from any adversaries). In other words, ulong[] indexesIntoSketch = getIndexesForString( key + s ).
Read ( long column, long row ) : ulong	Read from the underlying sketch data structure (two-dimensional array) at column (Table) column and row (element) row.
Write ( long column, long row, ulong value ) : void	WriteAccountToStableStoreAsync value to the underlying sketch data structure (two-dimensional array) at column (Table) column and row (element) row.

Method Details

Add() protected method

Add to the the value at each [column,index] pair, where the index into elementIndexForEachColumnis the column number and the value is the row (element) index. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.

protected Add ( long elementIndexForEachColumn, ulong amountToAdd = 1 ) : ResultOfUpdate
elementIndexForEachColumn	long	An array for which the index is a columnn number and the value /// is the row (element) index.
amountToAdd	ulong	The amount to Add to each of the indexed elements.
return	ResultOfUpdate

Add() public method

An Add function that is equivalent to calling increment multiple (amountToAdd) times. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.

public Add ( string s, ulong amountToAdd = 1 ) : ResultOfUpdate
s	string	The string to witness.
amountToAdd	ulong	The number of times to witness it. /// that would result before the Add.
return	ResultOfUpdate

ConservativeAdd() protected method

Use the conservative algorithm to Add to the data structure such that for each [column][row], the value stored is no less than getMin(elementIndexForEachColumn) + amountToAdd (but never exceeding MaxValue).

protected ConservativeAdd ( long elementIndexForEachColumn, ulong amountToAdd = 1 ) : ResultOfUpdate
elementIndexForEachColumn	long	An array for which the index is a columnn number and the value /// is the row (element) index.
amountToAdd	ulong	The desired increase such that `getMin()` /// after the operation returns is + `getMin()` /// (unless doing so would cause the result to exceed MaxValue).
return	ResultOfUpdate

ConservativeAdd() public method

An Add function that is equivalent to calling conservativeIncrement multiple (amountToAdd) times. If adding would cause the value to exceed MaxValue, MaxValue will be stored instead.

public ConservativeAdd ( string s, ulong amountToAdd = 1 ) : ResultOfUpdate
s	string	The string to witness.
amountToAdd	ulong	The number of times to witness it. /// that would result before the Add.
return	ResultOfUpdate

ConservativeIncrement() public method

Count the observation of the string s by adding one to those values indexed by calling

getIndexesForString(s) for which the value is currently equal to getMin(s). This algorithm is more conservative than increment in that it is  less likely to result in overcounting. If the count is already MaxValue, the operation will have no effect.

public ConservativeIncrement ( string s ) : ResultOfUpdate
s	string	The string to witness.
return	ResultOfUpdate

Get() public method

Get the current state of the sketch for a value's row indexes into each column.

public Get ( long elementIndexes ) : ResultOfGet
elementIndexes	long	The row indexes into each column of the sketch
return	ResultOfGet

Get() public method

Get the current state of the sketch for a value's row indexes into each column.

public Get ( string s ) : ResultOfGet
s	string	The string to query the sketch for occurrence/frequency information.
return	ResultOfGet

GetColumnTotal() public method

Get the total value stored in a columnn

public GetColumnTotal ( int column ) : ulong
column	int	The column
return	ulong

GetIndexesForString() protected method

Given a string s, provides an index into each column of the sketch. Each index identifies a row (element) to which one can write (if adding members/counts) or read (if testing membership or min counts). SECURITY NOTE: This indexes can be predicted by an adversary. If using the sketch in a scenario in which an adversary may cause harm by targeting certain indexes, include as a prefix of the parameter s a key that is not publicly-known (and hopefully can be kept secret from any adversaries). In other words, ulong[] indexesIntoSketch = getIndexesForString( key + s ).

protected GetIndexesForString ( string s ) : long[]
s	string	A string to map to indexes pointing into each column (table) of the sketch. /// If used in adversarial contexts, create a key string and include /// prefix all calls to this method with that key. ///
return	long[]

GetMin() public method

Get the minimum of all the values at the indexes identified for the string s with the sketch. This is a lower bound on the number of times the string s has been witnessed by the sketch via a call to increment(s) and/or conservativeIncrement(s). However, counting stops when the maximum value storable in the sketch is reached.

public GetMin ( string s ) : ulong
s	string	The string to query for the minimum occurrence count of.
return	ulong

Increment() public method

Count the observation of the string s by adding one to the value in each table indexed by calling getIndexesForString(s) If the count is already MaxValue, the operation will have no effect.

public Increment ( string s ) : ResultOfUpdate
s	string	The string to witness.
return	ResultOfUpdate

IsNonZero() public method

Test whether a string s has never be witnessed before by seeing if getMin(s) > 0. This is guaranteed to return true if increment(s) or conservativeIncrement(s) has been called on the sketch. It is probabilistically likely, but not guaranteed, to return false neither increment(s) or conservativeIncrement(s) have been called.

public IsNonZero ( string s ) : bool
s	string	The string to test.
return	bool

Read() protected method

Read from the underlying sketch data structure (two-dimensional array) at column (Table) column and row (element) row.

protected Read ( long column, long row ) : ulong
column	long	The column (table) to read from.
row	long	The row (element) within the table (column).
return	ulong

Sketch() public method

A sketch can be viewed as either having k tables (one for each hash) of n elements or as a two-dimensional array of k columns * n rows. This constructor creates a Sketch of size specified by k, n, and the number of bits per element.

public Sketch ( long numberOfColumns, long numberOfRows, int bitsPerElement ) : System
numberOfColumns	long	The number of columns in the sketch, which is equivalent to the number of tables (one table per hash index).
numberOfRows	long	The number of rows, which is equivalent to the number of elements per table.
bitsPerElement	int	The size of each element, in bits, such that the maximum value that can be stored (MaxValue) /// in the sketch elements is 2^(n)-1.
return	System

Write() protected method

WriteAccountToStableStoreAsync value to the underlying sketch data structure (two-dimensional array) at column (Table) column and row (element) row.

protected Write ( long column, long row, ulong value ) : void
column	long	The column (table) to write to.
row	long	The row (element) within the column (table).
value	ulong	The value to write.
return	void

this() public method

A shortcut to set all the elements of the sketch identified by an array of indexes, or to Get the minimum of all the values at the indexes identified by the array of indexes.

public this ( long elementIndexes ) : ulong
elementIndexes	long	An array of indexes into the sketch, one for each column (table).
return	ulong

this() public method

A shortcut to set all the elements of the sketch identified by indexes for the string s or to Get the minimum of all the values at the indexes identified for the string s (equivalent to calling getMin(s).

public this ( string s ) : ulong
s	string	A string to map to indexes pointing into each column (table) of the sketch.
return	ulong