R. Agrawal, T. Imielinski and A. Swami, Mining associations between sets of items in massive database, Proc. ACM SIGMOD Int. Conf. Management Data, Washington D. C., 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....occur in very few baskets. Basket data is a collection of records (or baskets) each record typically consisting of a transaction date and a collection of items (thought of as the items bought in this 12 transaction) The problem of mining association rules over basket data was introduced in [3]. Formally we consider a domain set I = fi 1 ; i m g of elements called items and we are given a set D of transactions where each transaction T is a subset of I. We say that a transaction T contains a set X of items if X T . An association rule is an implication rule X ) Y where X I ....

....constructing candidate pairs of items only if both items in the pair are frequent. Thus, to nd frequent itemsets, they proceed levelwise, nding rst the frequent items (sets of size 1) then the frequent pairs, the frequent triples, and so on. Mining for frequent itemsets An a priori algorithm [3, 5] which nds rst all frequent items needs to store them in main memory (it assumes that there is enough space) and then create candidate sets of all pairs of frequent items and nds in a second pass all frequent pairs. After the second pass it can create candidate sets of triples such that any ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In SIGMOD, pages 207-216, 1993.

....H.2.8 [Database Management] Database Applications data mining ; I.2.6 [Arti cial Intelligence] Learning; H.3.3 [Information Storage and Retrieval] Information Search and Retrieval General Terms Impact Rule, Association Rule, Numeric Data, Search 1. INTRODUCTION Association rules [1] have demonstrated the ability to detect interesting associations between elds in a database. However, they utilize frequency statistics and hence have limited utility for quantitative analyses. In particular, they cannot directly segment data to optimize a numeric target. Aumann and Lindell [3] ....

....of selecting the records that satisfy the antecedent. Such a rule might be valuable for identifying classes of customers from a mailing list that might be targeted most pro tably in a mailing campaign. Aumann and Lindell [3] propose that such rules be found by identifying frequent itemsets [1] and then, treating each as an antecedent, calculating the appropriate statistics for the target. An itemset is a set of conditions. A frequent itemset is an itemset that covers at least a prede ned minimum number of training set records. The primary diculty with this approach is that frequent ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In SIGMOD-93, pages 207-216, 1993.

....between heterogeneous datasets based on similarity. To the best of our knowledge, our work is the first study of such kind. 3 Association Mining and Itemset Lattice 3. 1 Association Mining The problem of finding all frequent associations among attributes in categorical ( basket ) databases [3], called association mining, is one of the most fundamental and most popular problems in data mining.We present basic concepts on association mining that are relevant to the similarity measure. The presentation here follows that of Agrawal et al. 4] Let D be a database of transactions over the ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993.

....techniques have also been applied to many areas, for example, market basket data, web data, DNA data, text data, and spatial data. Association rule mining is one of the important advances in the area of data mining. The initial application of association rule mining was on market basket data [AIS93]. An association rule is a relationship of the form X= Y, where X and Y are sets of items. X is called the antecedent and Y the consequence. An example of the rule can be customers who buy diaper and milk tend to buy beer. There are two primary quality measures for each rule, support and ....

....measures for each rule, support and confidence. Support indicates the frequency of the occurring pattern while confidence indicates the strength of the implication. The goal of association rule mining is to find all the rules with support and confidence exceeding some user specified thresholds [AIS93]. The problem is typically divided into two steps. The fist first step is to find itemsets with at least minimum support. These itemsets are called frequent itemsets or large itemsets. Once the frequent itemsets are generated, the second step is to derive rules with at least minimum confidence. ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and A. Swami, "Mining Associations Between Sets of Items in Massive Databases," Proceedings of the ACM SIGMOD, Washington, DC, May 1993, pp. 207-216.

....in their shopping baskets. The information discovered by association rules may help to develop marketing strategies, e.g. layout optimization in supermarkets (placing milk and bread within close proximity may further encourage the sale of these items together within single visits to the store) In [1] concrete examples for extracted associations between items are given. The examples are based on supermarket products that are included in a set of transactions collected from customers purchases. One classical anectode is that diapers are purchased together with beer . For the objective of ....

Agrawal, R. and Imielinski, T. and Swami, A.: Mining Associations between Sets of Items in Massive Databases, In Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data, Washington, D.C., May 26-28, 1993.

....algorithm. For example, the edit distance, ed, between the two lexical entries TopHotel and Top Hotel equals 1, ed( TopHotel , Top Hotel ) 1, because one insertion operation changes the string TopHotel into TopIotel . are compared, translating a numeric difference to a similarity value [0, 1] can be difficult. For example comparing the attribute population of a country a difference of 4 should yield a similarity value very close to 1, but comparing the attribute average number of children per woman the same numeric difference value should result in a similarity value close to 0. To ....

....numeric difference value should result in a similarity value close to 0. To take this into account, we first find the maximum difference between values of this attribute and then calculate the the similarity as 1 (Difference max Difference) Definition 15 (Literal similarity) slsira( A, A) [0, 1] mlsim : max sisira(A1, A2) A1 .A A A2 .A lsira(Ai, Aj, A) slsim(Ai, Aj) mlsim(A) And last but not least, unlike for relations the minimal similarity when comparing attributes is always zero. Definition 16 (Similarity for one attribute) if As(A , 11) 0 V As(A, 12) 0 (a As( ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and A. Swami. Mining Associations between Sets of Items in Massive Databases. In Proceedings of the 1993 ACM SIGMOD International Conference on Management of Data, Washington, D.C., May 26-28, 1993, pages 688-692. ACM Press, 1993.

....to performing the type of contrast analysis for which contrast sets were designed. Magnum Opus is a general purpose rule discovery system. It implements the OPUS AR rule discovery algorithm [14] It provides association rule like functionality, but does not use the frequent itemset strategy [1] and hence does not require the specification of a minimum support constraint. C4.5rules derives classification rules by first learning a decision tree and then transforming that tree into a rule format. It and the other two systems are described in more detail in Section 2. To evaluate the ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the 1993.

....mining association rules is to generate all association rules that have support s at least as great as some user specified minimum support Stain and confidence c at least as great as some user specified minimum confidence Cmin . Several algorithms have been presented in the literature [5] 16][17] for finding all such association rules. Many of them are variations of the Apriori algorithm. Apriori algorithm has two phases: 1) it finds all itemsets that have support above the minimum support; and (2) it uses these itemsets to generate all rules whose confidence are above the minimum ....

....We support two criteria: 1) A new session starts when the duration of the whole group of traversals exceeds a time threshold, similarly to [19] 2) The elapsed time between two consecutive Figure 3. A website taxonomy 4.3. Mining Algorithm To support taxonomies, one can use algorithms apriori [17] for mining standard association rules by considering extended transaction that contains not only the items in transactions but also their ancestors. To make this process efficient, certain optimizations are done to restrict the number of itemsets that need to be counted at various stages in the ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and A. Swami, "Mining Associations between Sets of Items in Massive Databases", in Proceedings of the ACM-SIGMOD 1993.

....sampling the training database. However, Bagging and Boosting make predictions hard to understand by users. In this paper, we also consider multiple rule sets, but we disturb a database systematically and use the union of all rule sets. The proposal for association rule mining first appeared in [1]. Most research work has been on how to generate frequent itemsets e#ciently since this may be the bottleneck of association rule mining. Apriori [2] is a widely accepted algorithm. The traditional goal of association rule mining is to solve market basket problems. However, it can also be used to ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proc. of the ACM SIGMOD Int'l Conference on Management of Data, pages 207--216, 1993.

....commercial DBMSs use equi depth his tograms [21, 23] which are in fact quantiles, during query optimization in order to estimate the size of intermedlate results and pick competitive query execution plans. Quantiles can also be used for determining association rules for data mining applications [1, 3, 2]. Quantile distribution helps design well suited user interraces to visualize query result sizes. Also, quantties provide a quick similarity check in coarsely comparing relations, which is useful in data cleaning [16] Finally, they are used as splitters in parallel database systems that employ ....

R. Agrawal, T. Imielinski and A. Swami. Min- ing Associations between Sets of Items in Massive Databases. In Proc. of ACM SIGMOD, pages 207-216, Washington D.C, May 1993.

....an integer threshold t, a V is called t frequent if it is contained in at least t hyperedges of and is called t infrequent otherwise. The generation of maximal frequent and minimal infrequent sets for are important tasks in knowledge discovery and data mining applications (see, for instance, [1, 2, 18]) Since the function f(X) X is polymatroid of range Theorems 3 and 2 imply respectively that the number of maximal frequent sets can be bounded by a quasi polynomial in the number of minimal infrequent sets and the sizes of V, and that the minimal infrequent sets can be generated ....

R. Agrawal, T. Imielinski and A. Swami, Mining associations between sets of items in massive databases, Proc. 1993 ACM-SIGMOD Int. Conf. on Management of Data, pp. 207-216. 12

....by t the family of all minimal t infrequent sets (i.e. those which are infrequent but all proper subsets of them are t frequent. The generation of frequent sets of a given binary matrix A is an important task of knowledge discovery and data mining, e.g. it is used for mining association rules [1, 2, 16, 21, 22], correlations [8] sequential patterns [3] episodes [23] emerging patterns [10] and appears in many other applications. Most practical procedures to generate t are based on the anti monotone Apriori heuristic (see [2] and build frequent sets in a bottom up way, running in time proportional ....

R. Agrawal, T. Imielinski and A. Swami. Mining associations between sets of items in massive databases. In: Proceedings of the 1993 ACM-SIGMOD International Conference on Management of Data, pp. 207-216.

....by t the family of all minimal t infrequent sets (i.e. those which are infrequent but all proper subsets of them are t frequent. The generation of frequent sets of a given binary matrix A is an important task of knowledge discovery and data mining, e.g. it is used for mining association rules [1, 2, 16, 22, 23], correlations [8] sequential patterns [3] episodes [24] emerging patterns [10] and appears in many other applications. Most practical procedures to generate t are based on the antimonotone Apriori heuristic (see [2] and build frequent sets in a bottom up way, running in time proportional ....

R. Agrawal, T. Imielinski and A. Swami. Mining associations between sets of items in massive databases. In: Proceedings of the 1993.

....enough to be of potential interest for a detailed data analysis. The precise interpretation of this term depends on the data model, dataset, and application. Perhaps the best studied framework for data mining uses association rules to describe interesting relationships between sets of data items [AIS93] In this framework, which is typically applied to market basket data (from checkout registers, indicating items purchased together) the critical operation is determining frequent itemsets, which are defined as sets of items that are purchased together often enough to pass a given threshold ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. SIGMOD Record, 22(2):207--216, June 1993.

....in data mining and information retrieval will be applicable to solve the knowledge acquisition problem Data Mining Data mining investigates knowledge discovery techniques that obtain predefined structured knowledge solely from huge databases. Some data mining problems (like e.g. clustering [4] [5] had already been known for a long time, and had been studied by many researchers in other fields of science (like statistics, machine learning, and visualization) Promising, possibly optimal, results were obtained (like K L transform for dimensionality reducing) the size of databases, ....

....in the size of the database, initiated further research. There are several data mining approaches and application areas. Association rules try to md association between set of items in a given database of e.g. sales transactions (or market baskets) where each transaction contains some product [4] [6] Clustering finds classes of closely related (similar) objects within a database of pattern vectors using a distance (similarity) function [5] Sequence matching searches a database of event sequences, and it tries to find the closest to a query sequence [7] In Episode finding a database ....

R. Agrawal, T. Imielinski, A. Swami: "Mining Associations between Sets of Items in Massive Databases," Proc. of the ACM SIGMOD Int'l Conference on Management of Data, Washington D.C., May 1993, pp. 207-216.

.... X , and Y 6= Since all 1 the subsets of a frequent itemset are frequent as well, and the support values of all frequent itemsets are known from the previous step, computing the con dence of the generated rules is straightforward. In this paper we concentrate our attention on the FSC problem [3]. Its search space is exactly P(M ) the power set of M , where M is the set of items contained in the transactions of D. Although P(M) is exponential in m = jM j, e ective pruning techniques exist for reducing it. Unfortunately, for small support thresholds, pruning becomes less e ective, thus ....

....is exponential in m = jM j, e ective pruning techniques exist for reducing it. Unfortunately, for small support thresholds, pruning becomes less e ective, thus making the FSC problem very expensive to solve both in time and space. A lot of proposals regard the ecient solution of the FSC problem [3, 4, 5, 12, 14, 15, 16, 18, 20, 21, 22]. The main goals of these algorithms are to eciently prune or partition P(M ) and to provide e ective strategies for traversing it. The capability of e ectively pruning the search space derives from the intuitive observation that none of the superset of an infrequent itemset can be frequent. The ....

R. Agrawal, T. Imielinski, and Swami A. Mining Associations between Sets of Items in Massive Databases. In Proc. of the ACM-SIGMOD

....architectures. The experimental results con rm that our new algorithm, DCP, sensibly outperforms the others previously proposed. Our test bed was a Pentium based Linux workstation, while the datasets used for tests were synthetically generated. 1 Introduction The Frequent Set Counting (FSC) [1] problem has been extensively studied as a method of unsupervised Data Mining [6, 7, 12] for discovering all the subsets of items (or attributes) that frequently occurs in the transactions of a given database. Knowledge on the frequent sets is generally used to extract Association Rules stating ....

....for c q in the tuples which immediately follow c p in F k 1 , and we stop the search when we nd a c q whose rst k 2 items are not equal to the rst k 2 items of c p . Only if we nd a c q such that c p [i] c q [i] 8i 2 f1; k 2g, then we can create the k itemset c = c p S c q = fc p [1]; c p [k 2] c p [k 1] c q [k 1]g. Note that, due to the same ordering of the tuples in F k 1 , checking whether the subsets of c are included in F k 1 (line 4 in Figure 3) can be done in logarithmic time. Subroutine apriori gen(Fk 1 ) 1: Ck ; 2: for all cp ; cq 2 Fk 1 j cp [i] cq ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and Swami A. Mining Associations between Sets of Items in Massive Databases. In Proc. of the ACM-SIGMOD

....improvements regard the use of an innovative method for storing candidate set of items and counting their support, and the exploitation of e#ective pruning techniques which significantly reduce the size of the dataset as execution progresses. 1 Introduction The Frequent Set Counting (FSC) [1,2,3,4,6,11,12,13,15,17,19] problem has been extensively studied as a method of unsupervised Data Mining [8,9,16] for discovering all the subsets of items (or attributes) that frequently occur in the transactions of a given database. Knowledge of the frequent sets is generally used to extract Association Rules stating how a ....

R. Agrawal, T. Imielinski, and Swami A. Mining Associations between Sets of Items in Massive Databases. In Proc. of the ACM-SIGMOD

....each t 2 T is a subset of I . Our discussions in this paper will be based on cooccurence patterns; however it is also possible to extend these ideas to sequential relations (Feldman, 1997) To determine the interestingness of a pattern, we define a weight function in Appendix A based on confidence (Agrawal et al. 1993) values. However, it is also possible to incorporate other measures such as interest (Silverstein et al. 1997) and improvement (Bayardo et al. 2000) In the subsequent subsections, we give definitions for representing each transaction as a set of these patterns and then we propose a hypergraph ....

....pattern p to exist in a transaction t, it is required not only the existence of the items in p but also the absence of items that will cause any superset of p to exist in t. For the above example, pattern p 1 means that milk and sugar exist, but f lour does not. In association rule mining problem (Agrawal et al. 1993), support of a pattern p (denoted as sup(p) in a dataset D is defined to be the ratio of the number of transactions that contain all the items in p to the total number of transactions in D . To extend this idea to the new semantics of the patterns, we propose the following definition: Definition ....

[Article contains additional citation context not shown here]

Agrawal, R., Imielinski, T., and Swami, A.: `Mining associations between sets of items in massive databases'. In Proc. of the 1993 ACM-SIGMOD Int'l Conf. on Management of Data, pp. 207-216, 1993.

....rules from all of the frequent itemsets, is relatively straightforward, but it can still be very expensive when solving real world problems. We have only briefly described the most basic concepts of association rule discovery. For more detailed information, see related technical publications [1][2] 4] 12] 13] 3. ALGORITHMS In this section we describe the software implementations of the association rule algorithms used in our experiments. The five algorithms evaluated were Apriori, Charm, FP growth, Closet and MagnumOpus. We provide references to articles describing the details of the ....

....us with an updated version of their code to fix bugs and or improve the performance. We reran our experiments with the new versions and noted below when updated versions were received. Apriori: Apriori is Christian Borgelt s implementation of the well known Apriori association rule algorithm [1][2] The source code in C for this implementation is available under the GNU Lesser General Public License from http: fuzzy.cs.unimagdeburg. de borgelt . Apriori takes transactional data in the form of one row for each pair of transaction and item identifiers. It first generates frequent ....

Agrawal, R., Imielinski, T., and Swami, A. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD International Conference on Management of Data, 207216.

.... widespread adoption by computer scientists in a variety of applications (e.g. for model based collaborative filtering, Heckerman et al. 2000) 3 Finding patterns (data mining) rather than global models (statistics) examples of pattern finding algorithms include association rule algorithms [AIS93], sequential association algorithms [MTI95] rule induction algorithms [WB86, SG92, FF99] and contrast set al..gorithms [BP99] These pattern finding algorithms differ from more conventional statistical modeling in that they do not attempt to cover all of the observed data, but instead focus in a ....

Agrawal, R., Imielinski, T., and Swami, A. (1993) Mining associations between sets of items in massive databases, in Proceedings of the 1993 ACM SIGMOD International Conference on the Management of Data, New York, NY: ACM Press, 207--216.

....hidden patterns and that it will generate meaningful intervals. Keywords: multivariate discretization; set mining; data mining 1. Introduction In set mining the goal is find conjunctions (or disjunctions) of terms that meet al..l user specified constraints. For example, in Association Rule Mining (Agrawal et al. 1993) a common first step is to find all itemsets that have support greater than a threshold. Set mining is a fundamental operation of data mining. In addition to association rule mining, many other large classes of algorithms can be formulated as set mining such as classification rules (e.g. Liu et ....

Agrawal R, Imielinski T, & Swami A (1993) Mining associations between sets of items in massive databases. Proceedings of the ACM SIGMOD International Conference on Management of Data, pp. 207--216.

....amount of computation above the data access layer. This computation can be floating point intensive in the case of a neural network application, or may involve a significant number of integer operations for rule based algorithms encountered in tree classification [14] and associations algorithms [1]. High performance data mining applications, therefore, require both fast data access and fast computation. In many ways Java is an ideal language to implement data mining operations. Java supports portable parallel programming on many platforms, which is useful for developing applications that ....

....Each product i has an affinity vector A i , where A ik is the affinity of product i with spending category k. These affinities can be interpreted as the perceived appeal of this product to customers with participation in this spending category. They are determined by pre computing associations [1] at the level of spending categories. The collection of affinity vectors from all products forms the m n affinity matrix A. This matrix is sparse and is organized so that products with the same nonzero patterns in their affinity vectors are adjacent. Therefore, matrix A can be represented by a ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993 International Conference on Management of Data, pages 207--216, Washington, DC, May 1993.

....classification strategy knowing only the information in the Figure 1b will yield an accuracy of 17.7 while random guessing gives 16.67 (assuming equal class priors) 2 1. 1 Relation to Association Rule Mining A closely related area to our work on contrast sets is association rule mining (Agrawal et al. 1993). Association rules are relations between variables of the form X Y . In market basket data X or Y would be items such as bread or milk. In categorical data X and Y are attribute value pairs such as occupation = engineer or income #50K. Finding association rules and mining contrast sets both ....

Agrawal, R., Imielinski, T., & Swami, A. (1993). Mining associations between sets of items in massive databases. Proceedings of the ACM SIGMOD International Conference on Management of Data (pp. 207--216).

....evaluated by some measure that led to identi cation of rules for which the antecedent identi ed subsets of the training set that were dominated by a single value of the class attribute, the intent being to predict the occurrence of that value. The other branch was association rule discovery [1]. Association rule discovery di ers in intent from most other rule discovery paradigms. While the other paradigms have concentrated on nding rules that are predictive of a single, preselected, class variable, association rule discovery has been motivated by nding rules that predict increased ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the 1993 ACM-SIGMOD International Conference on Management of Data, pages 207-216, 1993.

....an association rule with support oe and confidence ffi provided that (1) B=jT j oe, and (2) B=A ffi. Efficiently enumerating association rules has become an important topic in data mining. Typically, a heuristic approach is taken wherein one first enumerates all of the frequent sets of T [4, 7, 40, 5, 17, 37]. A frequent set is any set of attributes S such that the fraction of rows that have all attributes of S set to 1 is at least oe. Efficient algorithms for the conversion problem would be useful as a heuristic for enumerating all maximal frequent sets [17] Our read k restriction translates in this ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proc. of the ACM-SIGMOD 1993 International Conference on Management of Data, Washington D.C., 1993, 207-216.

....to be mined and based on the techniques used are found in the literature. Application of specific data mining techniques greatly depends on the kind of data we have, the questions that we need to answer and also on the computation power at hand. Mining association rules in relational databases [AIS93] is a data mining technique which derives a set of strong association rules, in a given set of transactions, of the form X ) Y , where X and Y are sets of items and each transaction is a set of literals (called items) For example, from a large set of transaction data one may find an association ....

....transaction set D if s of transactions in D contain X [ Y . Confidence is a measure of the strength of implication and support indicates the frequencies of occurring patterns in the rule. Rules with high confidence and strong support are often more interesting and are referred to as strong rules [AIS93] The Apriori algorithm [AS94a] and several modifications of this work have explored mining association rules in greater depth. Another technique that has found several applications is the process of data classification. Classification is the process of finding common properties among a set of ....

R. Agrawal, T. Imielinski, and A. Swami. Mining Associations between Sets of Items in Massive Databases . In Proc. of the ACM SIGMOD International Conference on Management of Data, pages 207--216, May 1993.

....state space complexity of the resulting scheme. Third, the frequency threshold parameter #, specifies the actual number of training set instances that must be supported by each state and not the fraction of training set instances as it is often done in the context of association rule discovery [AIS93] This is done primarily for the following two reasons: i) the trust worthiness of the estimated transition probabilities of a particular state depend on the actual number of training set instances and not on the relative number; ii) the total number of trainingset instances is in general ....

Rakesh Agrawal, Thomas Imielinski, and Arun Swami. Mining associations between sets of items in massive databases. In Proc. of the ACM SIGMOD Int'l Conference on Management of 13 Data, 1993.

....autocorrelation as a function of time. 5.2 Association Rules Discovery Given a set of records, with each record contains some number of items, it is desirable to discover the dependency rules such that the occurrence of an item can be predicted based on occurrences of other items. Agrawal et al. [2] proposed a formal model to define the association rules. Let = I 1 ; I 2 ; I m be a set items and T be a database of transactions. An association rule is an implication of the form X = I j , where X is a set of some items in , and I j is a single item in that is not present in X. ....

R. Agrawal, T. Imielinski, and A. Swami. Mining Associations between Sets of Items in Massive Databases. In Proc. of the ACM-SIGMOD Int'l Conference on Management of Data, pages 207--216, May 1993.

....Each product i has an affinity vector A i , where A ik is the affinity of product i with spending category k. These affinities can be interpreted as the perceived appeal of this product to customers with participation in this spending category. They are determined by pre computing associations [2] at the level of spending categories. The collection of affinity vectors from all products forms the m Theta n affinity matrix A. This matrix is sparse and organized so that products with the same nonzero patterns in their affinity vectors are adjacent. Therefore, matrix A can be represented by a ....

.... for a customer j against all products in a S = 2 6 4 Theta Theta Theta Theta Theta Theta Theta Theta Theta Theta Theta Theta Theta 3 7 5 N = 2 6 4 1 1 1 1 1 1 1 1 1 1 1 1 1 3 7 5 oe 1 = Theta Theta Theta ] J1 = 1 3 5 ] oe 2 = Theta Theta Theta ] J2 = [ 1 2 5 ] oe 3 = Theta Theta Theta Theta ] J3 = 1 2 3 4 ] oe 4 = Theta Theta Theta ] J4 = 2 3 4 ] Figure 2: Structures of the spending matrix S and normalizing matrix N , and representation of matrix S. 3 product group i can be computed in the following way. First, two vectors and j ....

[Article contains additional citation context not shown here]

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993 International Conference on Management of Data, pages 207--216, Washington, DC, May 1993.

....In the mediator environment the dominant cost is that of retrieving data from the remote Web sources which is what we attempt to minimize. Finally, the problem of extracting patterns from queries is somewhat similar to the problem of data mining, particularly that of mining association rules [3]. The problem of mining association rules is that of extracting implications of the form X ) Y from a database where X ae I and Y ae I and X Y = OE, where I = fi 1 ; i 2 ; i mg is a set of data items. The patterns that we extract from queries can also be looked upon as implications. For ....

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM SIGMOD International Conference on Management of Data, Washington D.C., 1993.

No context found.

R. Agrawal, T. Imielinski and A. Swami, Mining associations between sets of items in massive database, Proc. ACM SIGMOD Int. Conf. Management Data, Washington D. C., 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993.

No context found.

Agrawal, R., Imielinski, T., & Swami, A. (1993). Mining associations between sets of items in massive databases. Proc. of ACM SIGMOD.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proc. of ACM SIGMOD, 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM-SIGMOD 1993.

No context found.

Agrawal, R., Imielinski, T. & Swami, A. (1993), Mining Associations between Sets of Items in Massive Databases, In Proc. of the ACM SIGMOD Int'l Conference on Management of Data, Washington D.C., pages 207--216.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining Association between Sets of Items in Massive Databases. In ACM-SIGMOD 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining Association between Sets of Items in Massive Databases. In ACMSIGMOD 1993.

No context found.

R. Agrawal, T. Imielinski and A. Swami. Mining Associations between Sets of Items in Massive Databases. In Proc. of ACM SIGMOD, pages 207--216, Washington D.C, May 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proc. 1993.

No context found.

Agrawal, R., Imielinski, T., and Swami, A. Mining associations between sets of items in massive databases. In Proceedings of the 1993.

No context found.

T. Imielinski R. Agrawal and A. Swami. Mining associations between sets of items in massive databases. In Proceedings of the ACM SIGMOD Int'l Conference on Management of Data, Washington D.C., pages 207--216, May 1993. 34

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining Association between Sets of Items in Massive Databases. In ACM-SIGMOD 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. SIGMOD Record, 22(2):207-216, June 1993.

No context found.

Agrawal, R., Imielinski, T., & Swami, A. (1993). Mining associations between sets of items in massive databases. Proc. of ACM SIGMOD.

No context found.

R. Agrawal, T. Imielinski and A. Swami, Mining associations between sets of items in massive database, Proc. ACM SIGMOD Int. Conf. Management Data, Washington D. C., 1993.

No context found.

R. Agrawal, T. Imielinski, and A. Swami. Mining associations between sets of items in massive databases. In Proc. of the ACM SIGMOD Int'l Conference on Management of Data, 1993.

No context found.

Agrawal, R., Imielinski T., & Swami A. (1993). Mining Associations between Sets of Items in Massive Databases. Proc. of the ACM SIGMOD Int'l Conference on Management of Data, Washington D.C., May 1993. 207-216.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC