A Tree-Based Approach for Frequent Pattern Mining from Uncertain Data

Abstract. Mining frequent itemsets from transactional datasets is a well known problem with good algorithmic solutions. In the case of uncertain data, however, several new techniques have been proposed. Unfortunately, these proposals often suffer when a lot of items occur with many different probabilities. Here we propose an approach based on sampling by instantiating "possible worlds" of the uncertain data, on which we subsequently run optimized frequent itemset mining algorithms. As such we gain efficiency at a surprisingly low loss in accuracy. These is confirmed by a statistical and an empirical evaluation on real and synthetic data.

show abstract

“…UF-growth [7] extends the FP-Growth algorithm [6]. It is based on a UF-tree data structure (similar to FP-tree).…”

Section: Related Workmentioning

confidence: 99%

Efficient Pattern Mining of Uncertain Data with Sampling

Calders

Garboni

Goethals

2010

Advances in Knowledge Discovery and Data Mining

View full text Add to dashboard Cite

show abstract

“…Most of uncertain data models use possible world semantics to represent the uncertain relations among item [12]. Since the uncertain transactions are probabilistic in nature, it is impossible to count the frequency of itemsets deterministically.…”

Section: Preliminariesmentioning

confidence: 99%

“…There are two possible worlds for an item x and a transaction t i : 1) W 1 where x t i and 2) W 2 where xt i . The probability for W 1 to be the true world is thus P(x, t), and for W 2 is 1-P(x,t) , where P(x, t) represents the existential probability of x in t [12].…”

Section: Preliminariesmentioning

confidence: 99%

“…As a result, it is not practical to compress databases in active routine using the conventional data compression techniques. Research in uncertain and probabilistic database management is abundant (see [12], [13] for recent surveys). Recently, also the data mining community has started dealing with uncertain data, tackling the problems of clustering, classification and frequent pattern mining, but finding small and informative pattern-based descriptions of uncertain databases has not been addressed before.…”

Section: Literature Surveymentioning

confidence: 99%

See 1 more Smart Citation

An Efficient Compression Algorithm for Uncertain Databases Aimed at Mining Problems

Gabr¹,

Darwish²,

Mohsin³

2015

LNSE

View full text Add to dashboard Cite

Abstract-Many studies on association rule mining have focused on item sets from precise data in which the presence and absence of items in transactions was certainly known. In some applications, the presence and absence of items in transactions are uncertain and the knowledge discovered from this type of data will extracted with approximation manner. Data compression offers a good solution to reduce data size that can save the time of discovering useful knowledge. In this paper we suggest a new algorithm to compress transactions from uncertain database based on modified version of M 2 TQT (Mining Merged Transactions with the Quantification Table) approach and fuzzy logic concept. The algorithm bands the uncertain data to set of clusters using K-Mean algorithm and exploits fuzzy membership function to classify the transaction items as one of those clusters. Finally, the modified version of M 2 TQT has been employed to compress the classified transactions. The key idea of our algorithm is that since uncertain data is probabilistic in nature and frequent item set is counted as expected values so, compressed transactions will give us approximate values for the item set's support. Experimental results show that the proposed algorithm is better than U-Apriori algorithm in case of large uncertain database.Index Terms-Rule mining, database compression, Uncertain database, fuzzy logic.

show abstract

“…Frequent itemset mining [1,10,12,13,14] aims to search for implicit, previously unknown and potentially useful information-in the form of frequently occurring patterns-that might be embedded in data. It is in demand in many real-life applications, and it also plays an essential role in the mining of many patterns other than frequent itemsets (e.g., maximal itemsets, closed itemsets, association rules, correlations, sequences, episodes).…”

Section: Introductionmentioning

confidence: 99%