Approximate Frequency Counts over Data Streams Gurmeet Singh Manku, Rajeev Motwani Standford University VLDB2002
Introduction Data come as a continuous “ stream ” Differs from traditional stored DB The sheer volume of a stream over its lifetime is huge Queries require timely answer
Frequent itemset mining on offline databases vs data streams Often, level-wise algorithms are used to mine offline databases At least 2 database scans are needed Ex: Apriori algorithm Level-wise algorithms cannot be applied to mine data streams Cannot go through the data stream multiple times
Challenges of streaming Single pass Limited Memory Enumeration of itemsets
Purpose Present algorithms computing frequency exceeding threshold Simple Low memory footprint Output approximate, guaranteed not exceed a user specified error parameter. Deployed for singleton items, handle variable sized sets of items. Main contributions of the paper: Proposed 2 algorithms to find frequent items appear in a data stream of items Extended the algorithms to find frequent itemset
Notations Some notations: Let N denote the current length of the stream Let s (0,1) denote the support threshold Let (0,1) denote the error tolerance << s
Approximation guarantees All itemsets whose true frequency exceeds sN are reported No itemset whose true frequency is less than ( s- ) N is output Estimated frequencies are less than the true frequencies by at most N
Example s = 0.1% ε should be one-tenth or one-twentieth of s. ε = 0.01% Property 1, elements frequency exceeding 0.1% output. Property 2, NO element frequency below 0.09% output Elements between 0.09% ~ 0.1% may or may not be output. Property 3, frequencies are less than their true frequencies at most 0.01%
Problem definition An algorithm maintains an ε- deficient synopsis if its output satisifies the aforementioned properties Devise algorithms support ε- deficient synopsis using little main memory as possible
The Algorithms for frequent Items Each transaction contains only 1 item Two algorithms proposed: Sticky Sampling Algorithm Lossy Counting Algorithm Features : Sampling used Frequency found approximate, error guaranteed not exceed user-specified tolerance level For Lossy Counting, all frequent items are reported
Sticky Sampling Algorithm Create counters by sampling Stream
Sticky Sampling Algorithm User input : Support threshold s Error tolerance Probability of failure Counts kept in data structure S Each entry in S is in the form ( e, f ), where: e : item f : frequency of e since the entry inserted in S Output entries in S where f (s - )N
Sticky Sampling Algorithm r : sampling rate Sampling an element with rate = r means select the element with probablity = 1/r
Sticky Sampling Algorithm Initially – S is empty, r = 1. For each incoming element e if (e exists in S) increment corresponding f else { sample element with rate r if (sampled) add entry (e,1) to S else ignore }
Sampling rate Let t = 1/ ε log(s -1 -1 ) ( = probability of failure) First 2t elements sampled at rate=1 The next 2t at rate=2 The next 4t at rate=4 and so on …
Sticky Sampling Algorithm Whenever the sampling rate r changes: for each entry (e,f) in S repeat { toss an unbiased coin if (toss is not successful) diminsh f by one if (f == 0) { delete entry from S break } } until toss is successful
Lossy Counting Data stream conceptually divided into buckets = 1/ transactions Buckets labeled with bucket ids, starting from 1 Current bucket id is b current,value is N/ f e :true frequency of an element e in stream seen so far Each entry in data structure D is form ( e, f, ) e : item f : frequency of e : the maximum possible error in f
Lossy Counting is the maximum # of times e occurred in the first b current – 1 buckets ( this value is exactly b current – 1) Once a value is inserted into D its value is unchanged
Lossy Counting Initially D is empty Receive element e if (e exists in D) increment its frequency (f) by 1 else create a new entry (e, 1, b current – 1) If bucket boundary prune D by the following the rule: (e,f,) is deleted if f + ≤ b current When the user requests a list of items with threshold s, output those entries in D where f ≥ (s – ε)N
Lossy Counting 1. function prune(D, b) 2. for each entry (e,f,) in D do 3. if f + b do 4. remove the entry from D 5. endif
Lossy Counting At window boundary, remove entries that for them f+∆ ≤ b current D is Empty
Lossy Counting At window boundary, remove entries that for them f+∆≤ b current Next Window +
Lossy Counting Lossy Counting guarantees that: When deletion occurs, b current N Entry ( e, f, ) is deleted, If f e b current f e : actual frequency count of e Hence, if entry ( e, f, ) is deleted, f e N Finally, f f e f + N
Sticky Sampling vs Lossy Counting Sticky Sampling is non- deterministic, while Lossy Counting is deterministic Experimental result shows that Lossy Counting requires fewer entries than Sticky Sampling
Sticky Sampling vs Lossy Counting Lossy counting is superior by a large factor Sticky sampling performs worse because of its tendency to remember every unique element that gets sampled Lossy counting is good at pruning low frequency elements quickly
The more complex case: finding frequent itemsets The Lossy Counting algorithm is extended to find frequent itemsets Transactions in the data stream contains a set of items
Finding frequent itemsets Stream
Finding frequent itemsets Input: stream of transactions, each transaction is a set of items from I N: length of the stream User specifies two parameters: support s, error Challenge: - handling variable sized transactions - avoiding explicit enumeration of all subsets of any transaction
Finding frequent itemsets Data structure D – set of entries of the form (set, f, ) set : subset of items Transactions are divided into buckets = 1/ transactions : # of transactions in each bucket b current : current bucket id
Finding frequent itemsets Transactions not processed one by one. Main memory filled as many transactions as possible. Processing is done on a batch of transactions. β : # of buckets in main memory in the current batch being processed.
Finding frequent itemsets D ’ s operations : UPDATE_SET updates and deletes in D Entry (set, f, ) count occurrence of set in the batch and update the entry If updated entry satisfies f + bcurrent, removed it from D NEW_SET inserts new entries into D If set set has frequency f in batch and set doesn ’ t occur in D, create a new entry (set, f, bcurrent-)
Finding frequent itemsets If f set ≥ N it has an entry in D If (set,f,) E D then the true frequency of f set satisfies the inequality f≤ f set ≤ f+ When user requests list of items with threshold s, output in D where f ≥ (s-)N β needs to be a large number. Any subset of I that occurs β +1 times or more contributes to D.
Buffer: repeatedly reads in a batch of buckets of transactions into available main memory Trie: maintains the data structure D SetGen: generates subsets of item-id ’ s along with their frequency counts in the current batch Not all possible subsets need to be generated If a subset S is not inserted into D after application of both UPDATE_SET and NEW_SET, then no supersets of S should be considered
Three modules BUFFER TRIE SUBSET-GEN maintains the data structure D operates on the current batch of transactions repeatedly reads in a batch of transactions into available main memory implement UPDATE_SET, NEW_SET
Module 1 - Buffer Read a batch of transactions Transactions are laid out one after the other in a big array A bitmap is used to remember transaction boundaries After reading in a batch, BUFFER sorts each transaction by its item-id ’ s Window 1 Window 2 Window 3 Window 4 Window 5 Window 6 In Main Memory
Module 2 - TRIE Sets with frequency counts
Module 2 – TRIE cont… Nodes are labeled {item-id, f, , level} Children of any node are ordered by their item- id ’ s Root nodes are also ordered by their item-id ’ s A node represents an itemset consisting of item- id ’ s in that node and all its ancestors TRIE is maintained as an array of entries of the form {item-id, f, , level} (pre-order of the trees). Equivalent to a lexicographic ordering of subsets it encodes. No pointers, level ’ s compactly encode the underlying tree structure.
Module 3 - SetGen BUFFER Frequency counts of subsets in lexicographic order SetGen uses the following pruning rule: if a subset S does not make its way into TRIE after application of both UPDATE_SET and NEW_SET, then no supersets of S should be considered
Overall Algorithm BUFFER SUBSET-GEN TRIEnew TRIE
Conclusion Sticky Sampling and Lossy Counting are 2 approximate algorithms that can find frequent items Both algorithms produces frequency counts within a user-specified error tolerance level, though Sticky Sampling is non-deterministic Lossy Counting can be extended to find frequent itemsets
Thank you very much…