• Journal of KIISE:Databases
• /
• v.31 no.2
• /
• pp.91-98
• /
• 2004

DHP mining association rules algorithm maintains previously independent direct hash table to reduce the sire of hash tree containing the frequency number of each candidate large itemset. It performs pruning by using the direct hash table when the hash tree is constructed. The mort large the size of direct hash table increases, the higher the effort of pruning becomes. Especially, the effect of pruning in phase 2 which generate 2-large itemsets is so high that it dominates the overall performance of DHP algorithm. So, following the speedy trends of producing VLM(Very Large Memory) systems, extreme increment of direct hash table size is being tried and one of those trials is perfect hash table in phase 2. In case of using perfect hash table in phase 2, we found that some rearrangement of DHP algorithm got about 20% performance improvement compared to simply ｜H$_2$｜ reconfigured DHP algorithm. In this paper, we examine the feasibility of perfect hash table in phase 2 and propose PHP algorithm, a rearranged DHP algorithm, which uses the characteristics of perfect hash table sufficiently, then make an analysis on the results in experimental environment.

• The KIPS Transactions:PartD
• /
• v.13D no.5 s.108
• /
• pp.651-660
• /
• 2006

Algorithms for mining association rules based on the Apriori algorithm use the hash tree data structure for storing and counting supports of the candidate frequent itemsets and the most part of the execution time is consumed for searching in the hash tree. The DHP(Direct Hashing and Pruning) algorithm makes efforts to reduce the number of the candidate frequent itemsets to save searching time in the hash tree. For this purpose, the DHP algorithm does preparative simple counting supports of the candidate frequent itemsets. At this time, the DHP algorithm uses the direct hash table to reduce the overhead of the preparative counting supports. This paper proposes and evaluates an efficient hashing mechanism for the direct hash table $H_2$ which is for pruning in phase 2 and the hash tree $C_k$, which is for counting supports of the candidate frequent itemsets in all phases. The results showed that the performance improvement due to the proposed hashing mechanism was 82.2% on the maximum and 18.5% on the average compared to the conventional method using a simple mod operation.

• Journal of KIISE:Databases
• /
• v.35 no.3
• /
• pp.199-207
• /
• 2008

The FP-tree(Frequency Pattern Tree) mining association rules algorithm was proposed to improve mining performance by reducing DB scan overhead dramatically, and it is recognized that the performance of it is better than that of any other algorithms based on different approaches. But the FP-tree algorithm needs a few more memory because it has to store all transactions including frequent itemsets of the DB. This paper implements a FP-tree algorithm on a general purpose UNK system and compares it with the DHP(Direct Hashing and Pruning) algorithm which uses hash tree and direct hash table from the point of memory usage and execution time. The results show surprisingly that the FP-tree algorithm is poor than the DHP algorithm in some cases even if the system memory is sufficient for the FP-tree. The characteristics of the test data are as follows. The site of DB is look, the number of total items is $1K{\sim}7K$, avenrage length of transactions is $5{\sim}10$, avergage size of maximal frequent itemsets is $2{\sim}12$(these are typical attributes of data for large-scale convenience stores).

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.7
• /
• pp.2599-2613
• /
• 2015

Multi-index hashing (MIH) is the state-of-the-art method for indexing binary codes, as it di-vides long codes into substrings and builds multiple hash tables. However, MIH is based on the dataset codes uniform distribution assumption, and will lose efficiency in dealing with non-uniformly distributed codes. Besides, there are lots of results sharing the same Hamming distance to a query, which makes the distance measure ambiguous. In this paper, we propose a data-oriented multi-index hashing method (DOMIH). We first compute the covariance ma-trix of bits and learn adaptive projection vector for each binary substring. Instead of using substrings as direct indices into hash tables, we project them with corresponding projection vectors to generate new indices. With adaptive projection, the indices in each hash table are near uniformly distributed. Then with covariance matrix, we propose a ranking method for the binary codes. By assigning different bit-level weights to different bits, the returned bina-ry codes are ranked at a finer-grained binary code level. Experiments conducted on reference large scale datasets show that compared to MIH the time performance of DOMIH can be improved by 36.9%-87.4%, and the search accuracy can be improved by 22.2%. To pinpoint the potential of DOMIH, we further use near-duplicate image retrieval as examples to show the applications and the good performance of our method.

• Journal of KIISE:Databases
• /
• v.31 no.3
• /
• pp.261-273
• /
• 2004

Advances in information and communication technologies have been creating new classes of applications in the area of databases. For example, in moving object databases, which track positions of a lot of objects, or stream databases, which process data streams from a lot of sensors, data Processed in such database systems are usually changed very rapidly and continuously. However, traditional database systems have a problem in processing these rapidly and continuously changing data because they suppose that a data item stored in the database remains constant until It is explicitly modified. The problem becomes more serious in the R-tree, which is a typical index structure for multidimensional data, because modifying data in the R-tree can generate cascading node splits or merges. To process frequent updates more efficiently, we propose a novel update technique for the R-tree, which we call the leaf-update technique. If a new value of a data item lies within the leaf MBR that the data item belongs, the leaf-update technique changes the leaf node only, not whole of the tree. Using this leaf-update manner and the leaf-access hash table for direct access to leaf nodes, the proposed technique can reduce update cost greatly. In addition, the leaf-update technique can be adopted in diverse variants of the R-tree and various applications that use the R-tree since it is based on the R-tree and it guarantees the correctness of the R-tree. In this paper, we prove the effectiveness of the leaf-update techniques theoretically and present experimental results that show that our technique outperforms traditional one.