• The KIPS Transactions:PartD
• /
• v.12D no.3 s.99
• /
• pp.345-354
• /
• 2005

Subsequence matching is one of the most important operations in the field of data mining. The existing subsequence matching algorithms use only one index, and their performance gets worse as the difference between the length of a query sequence and the site of windows, which are subsequences of a same length extracted from data sequences to construct the index, increases. In this paper, we propose a new subsequence matching method based on index interpolation to overcome such a problem. An index interpolation method constructs two or more indexes, and performs search ing by selecting the most appropriate index among them according to the given query sequence length. In this paper, we first examine the performance trend with the difference between the query sequence length and the window size through preliminary experiments, and formulate a search cost model that reflects the distribution of query sequence lengths in the view point of the physical database design. Next, we propose a new subsequence matching method based on the index interpolation to improve search performance. We also present an algorithm based on the search cost formula mentioned above to construct optimal indexes to get better search performance. Finally, we verify the superiority of the proposed method through a series of experiments using real and synthesized data sets.

• Journal of Industrial Technology
• /
• v.21 no.A
• /
• pp.155-166
• /
• 2001

This paper deals with the subsequence searching problem under time-warping. Our work is motivated by the observation that subsequence searches slow down quadratically as the average length of data sequences increases. To resolve this problem, the Segment-Based Approach for Subsequence Searches (SBASS) is proposed. The SBASS divides data and query sequences into a series of segments, and retrieves all data subsequences. Our segmentation scheme allows segments to have different lengths; thus we employ the time warping distance as a similarity measure for each segment pair. For efficient retrieval of similar subsequences, we extract feature vectors from all data segments exploiting their monotonically changing properties, and build a spatial index using feature vectors. The effectiveness of our approach is verified through extensive experiments.

• Journal of KIISE:Databases
• /
• v.34 no.2
• /
• pp.119-132
• /
• 2007

In molecular biology, approximate subsequence search is one of the most important operations. In this paper, we propose an accurate and efficient method for approximate subsequence search in large DNA databases. The proposed method basically adopts a binary trie as its primary structure and stores all the window subsequences extracted from a DNA sequence. For approximate subsequence search, it traverses the binary trie in a breadth-first fashion and retrieves all the matched subsequences from the traversed path within the trie by a dynamic programming technique. However, the proposed method stores only window subsequences of the pre-determined length, and thus suffers from large post-processing time in case of long query sequences. To overcome this problem, we divide a query sequence into shorter pieces, perform searching for those subsequences, and then merge their results. To verify the superiority of the proposed method, we conducted performance evaluation via a series of experiments. The results reveal that the proposed method, which requires smaller storage space, achieves 4 to 17 times improvement in performance over the suffix tree based method. Even when the length of a query sequence is large, our method is more than an order of magnitude faster than the suffix tree based method and the Smith-Waterman algorithm.

• The KIPS Transactions:PartD
• /
• v.13D no.3 s.106
• /
• pp.357-368
• /
• 2006

This paper discusses the way of processing time-series subsequence matching under time warping. Time warping enables finding sequences with similar patterns even when they are of different lengths. The prefix-querying method is the first index-based approach that performs time-series subsequence matching under time warping without false dismissals. This method employs the $L_{\infty}$ as a base distance function for allowing users to issue queries conveniently. In this paper, we extend the prefix-querying method for absorbing $L_1$, which is the most-widely used as a base distance function in time-series subsequence matching under time warping, instead of $L_{\infty}$. We also formally prove that the proposed method does not incur any false dismissals in the subsequence matching. To show the superiority of our method, we conduct performance evaluation via a variety of experiments. The results reveal that our method achieves significant performance improvement in orders of magnitude compared with previous methods.

• The KIPS Transactions:PartD
• /
• v.9D no.4
• /
• pp.555-560
• /
• 2002

Subsequence matching, which consists of index searching and post-processing steps, is an operation that finds those subsequences whose changing patterns are similar to that of a given query sequence from a time-series database. This paper discusses optimization of post-processing for subsequence matching. The common problem occurred in post-processing of previous methods is to compare the candidate subsequence with the query sequence for discarding false alarms whenever each candidate subsequence appears during index searching. This makes a sequence containing candidate subsequences to be accessed multiple times from disk, and also have a candidate subsequence to be compared with the query sequence multiple times. These redundancies cause the performance of subsequence matching to degrade seriously. In this paper, we propose a new optimal method for resolving the problem. The proposed method stores ail the candidate subsequences returned by index searching into a binary search tree, and performs post-processing in a batch fashion after finishing the index searching. By this method, we are able to completely eliminate the redundancies mentioned above. For verifying the performance improvement effect of the proposed method, we perform extensive experiments using a real-life stock data set. The results reveal that the proposed method achieves 55 times to 156 times speedup over the previous methods.

• Journal of the Korea Computer Industry Society
• /
• v.4 no.4
• /
• pp.613-624
• /
• 2003

This paper proposes a design for efficient and accurate retrieval of similar image subsequences using the multi-dimensional time warping distance as similarity evaluation tool in image sequence database after building of two indexing structures implemented with priority queue and suffix tree respectively. Receiving query image sequence, at first step, the proposed method searches the candidate set of similar image subsequences in priory queue index structure. If it can not get satisfied results, it retrieves another candidate set in suffix tree index structure at second step. The using of the low-bound distance function can remove the dissimilar subsequence without false dismissals during similarity evaluating process between query image sequence and stored sequences in two index structures.

• Journal of the Korea Society of Computer and Information
• /
• v.14 no.5
• /
• pp.19-28
• /
• 2009

Prediction problem of the time-series data has been a research issue for a long time among many researchers and a number of methods have been proposed in the literatures. In this paper, a method is proposed that similarities among time-series data are examined by use of Hidden Markov Model and Likelihood and future direction of the data movement is determined. Query sequence is modeled by Hidden Markov Modeling and then the model is examined over the pre-recorded time-series to find the subsequence which has the greatest similarity between the model and the extracted subsequence. The similarity is evaluated by likelihood. When the best subsequence is chosen, the next portion of the subsequence is used to predict the next phase of the data movement. A number of experiments with different parameters have been conducted to confirm the validity of the method. We used KOSPI to verify suggested method.

• Journal of KIISE:Databases
• /
• v.30 no.4
• /
• pp.381-396
• /
• 2003

Subsequence matching is an operation that finds subsequences whose changing patterns are similar to a given query sequence from time-series databases. This paper points out the performance bottleneck in subsequence matching, and then proposes an effective method that improves the performance of entire subsequence matching significantly by resolving the performance bottleneck. First, we analyze the disk access and CPU processing times required during the index searching and post processing steps through preliminary experiments. Based on their results, we show that the post processing step is the main performance bottleneck in subsequence matching, and them claim that its optimization is a crucial issue overlooked in previous approaches. In order to resolve the performance bottleneck, we propose a simple but quite effective method that processes the post processing step in the optimal way. By rearranging the order of candidate subsequences to be compared with a query sequence, our method completely eliminates the redundancy of disk accesses and CPU processing occurred in the post processing step. We formally prove that our method is optimal and also does not incur any false dismissal. We show the effectiveness of our method by extensive experiments. The results show that our method achieves significant speed-up in the post processing step 3.91 to 9.42 times when using a data set of real-world stock sequences and 4.97 to 5.61 times when using data sets of a large volume of synthetic sequences. Also, the results show that our method reduces the weight of the post processing step in entire subsequence matching from about 90% to less than 70%. This implies that our method successfully resolves th performance bottleneck in subsequence matching. As a result, our method provides excellent performance in entire subsequence matching. The experimental results reveal that it is 3.05 to 5.60 times faster when using a data set of real-world stock sequences and 3.68 to 4.21 times faster when using data sets of a large volume of synthetic sequences compared with the previous one.

• The KIPS Transactions:PartD
• /
• v.10D no.4
• /
• pp.577-590
• /
• 2003

The shape-based retrieval is defined as the operation that searches for the (sub) sequences whose shapes are similar to that of a query sequence regardless of their actual element values. In this paper, we propose a similarity model suitable for shape-based retrieval and present an indexing method for supporting the similarity model. The proposed similarity model enables to retrieve similar shapes accurately by providing the combination of various shape-preserving transformations such as normalization, moving average, and time warping. Our indexing method stores every distinct subsequence concisely into the disk-based suffix tree for efficient and adaptive query processing. We allow the user to dynamically choose a similarity model suitable for a given application. More specifically, we allow the user to determine the parameter p of the distance function $L_p$ when submitting a query. The result of extensive experiments revealed that our approach not only successfully finds the subsequences whose shapes are similar to a query shape but also significantly outperforms the sequence search.

• Journal of KIISE
• /
• v.43 no.10
• /
• pp.1173-1178
• /
• 2016

Since the introduction of BitTorrent protocol in 2001, everything can be downloaded through file sharing, including music, movies and software. As a result, the copyright holder suffers from illegal sharing of copyright content. In order to solve this problem, countries have enacted illegal share related law; and internet service providers block pirate sites. However, illegal sites such as pirate bay easily reopen the site by changing the domain name. Thus, we propose a technique to easily detect pirate sites that are reopened. This automated technique collects the domain names using the google search engine, and measures similarity using Longest Common Subsequence (LCS) algorithm by comparing the tag structure of the source web page and reopened web page. For evaluation, we colledted 2,383 domains from google search. Experimental results indicated detection of a total of 44 pirate sites for collected domains when applying LCS algorithm. In addition, this technique detected 23 pirate sites for 805 domains when applied to foreign pirate sites. This experiment facilitated easy detection of the reopened pirate sites using an automated detection system.