研究者詳細 - 石川　佳治

論文 - 石川　佳治

分割表示 320 件中 41 - 60 件目／全件表示 >>

Novelty-based Incremental Document Clustering for On-line Documents 査読有り

Sophoin Khy, Yoshiharu Ishikawa, and Hiroyuki Kitagawa

Proceedings of International Workshop on Challenges in Web Information Retrieval and Integration (WIRI 2006) 頁： 40 2006年4月

記述言語：英語

Document clustering has been used as a core technique in managing vast amount of data and providing needed information. In on-line environments, generally new information gains more interests than old one. Traditional clustering focuses on grouping similar documents into clusters by treating each document with equal weight. We proposed a novelty-based incremental clustering method for on-line documents that has biases on recent documents. In the clustering method, the notion of `novelty' is incorporated into a similarity function and a clustering method, a variant of the K-means method, is proposed. We examine the efficiency and behaviors of the method by experiments.

マルコフ連鎖モデルに基づく動的な移動ヒストグラム構築手法査読有り

町田陽二, 石川佳治, 北川博之

日本データベース学会Letters 5 巻 ( 1 ) 頁： 89-92 2006年6月

記述言語：日本語掲載種別：研究論文（学術雑誌）

リアルタイムに移動する多数のオブジェクトの移動状況の集約・
分析のため，我々は移動データを要約する移動ヒストグラムの構
築手法の開発を進めている．この手法では，マルコフ連鎖モデル
に基づき移動統計量を集約する．移動ヒストグラムの論理モデル
としてはデータキューブを採用し，OLAP 的な分析を支援する．
一方，物理モデルとしては木構造を採用し，ストリーム的に配信さ
れる移動状況データを限られたメモリ内で効率よく要約する．本
稿では提案手法の詳細について述べ，性能を実験により評価する．

Incremental Clustering Based on Novelty of On-line Documents 査読有り

Sophoin Khy, Yoshiharu Ishikawa, and Hiroyuki Kitagawa

DBSJ Letters 5 巻 ( 1 ) 頁： 57-60 2006年6月

記述言語：英語掲載種別：研究論文（学術雑誌）

Clustering has been widely used as a fundamental
method in many areas such as characterization
and classification. Various clustering researches have
been conducted since decades ago. In previous papers,
we presented a novelty-based incremental document
clustering method which considers novelty of
on-line documents in its similarity measure and performs
clustering based on an extended algorithm of
the K-means method. This paper further examines the
performance of the incremental and non-incremental
processing of the clustering method and effect of parameter
values on the method by showing the experimental
results.

A Dynamic Mobility Histogram Construction Method Based on Markov Chains 査読有り

Yoshiharu Ishikawa, Yoji Machida, and Hiroyuki Kitagawa

Proceedings of the 18th International Conference on Scientific and Statistical Database Management (SSDBM 2006) 頁： 359-368 2006年7月

担当区分：筆頭著者記述言語：英語

With the recent progress of spatial information technologies and communication technologies, it has become easier to track positions of a large number of moving objects in real-time. Mobility statistics plays an important role in the interactive analysis of a large collection of moving objects trajectories and its use of movement pattern prediction. The development of an effective mobility statistics measure and its efficient computation method are critical issues. Thus, we propose an approach for constructing a mobility histogram to summarize a number of moving object trajectories. The histogram is based on a mobility statistics model called the Markov chain model. To facilitate an interactive analysis performed by a user, we provide a mobility histogram data cube-like logical representation and support an OLAP-style analysis. Since trajectory data is often received continuously as a trajectory stream, we have to support dynamic histogram construction and maintenance. We introduce a tree structure as the physical representation of a histogram and present histogram construction and maintenance methods that work efficiently within the given upperbound size. We evaluate the performance and the precision of the proposed method by means of experiments.

マルコフ連鎖モデルに基づく移動ヒストグラムの動的構築法査読有り

石川佳治，町田陽二，北川博之

電子情報通信学会論文誌 D J90-D 巻 ( 2 ) 頁： 311-324 2007年2月

担当区分：筆頭著者記述言語：日本語掲載種別：研究論文（学術雑誌）

GPSや通信機器の普及に伴い，移動オブジェクトの追跡が容易となっている．大量の移動オブジェクトの移動状況をリアルタイムにモニタリングし集積することにより，移動状況の把握や予測に利用することが考えられる．そのため，本論文では大量の移動オブジェクトの移動軌跡データをコンパクトに要約するための移動ヒストグラムの構築手法を提案する．ヒストグラムの基礎となる移動統計量のモデルとして，本手法ではマルコフ連鎖モデルを採用し，データキューブに似た移動ヒストグラムの論理表現を与え，OLAP的な分析を可能とする．一方，その物理表現としては木構造を採用し，近似的な表現により記憶容量の削減を図る．また，移動状況データはストリームとして逐次受信されることから，リアルタイム処理に適した効率的なヒストグラムの構築方式を実現する．本論文では，提案手法の効率及び精度について実験に基づく評価についても報告する．

T-Scroll：時間的トピックの推移をとらえる可視化システム査読有り

長谷川幹根，石川佳治

日本データベース学会Letters 6 巻 ( 1 ) 頁： 149-152 2007年6月

記述言語：日本語掲載種別：研究論文（学術雑誌）

インターネット上では，ニュースなどの大量のテキストデータの配信が日々なされている．本論文では，このようなテキストデータにおける，時間的なトピックの推移をとらえるための可視化システムT-Scrollについて述べる．本システムは，下位の時系列的な文書クラスタリングシステムのクラスタリング結果をもとに，クラスタの関連を巻き物（scroll）状に提示する．本論文では，システムのアイデア，機能，実現手法等について述べる

Record Extraction Based on User Feedback and Document Selection 査読有り

Jianwei Zhang, Yoshiharu Ishikawa, Hiroyuki Kitagawa

Proceedings of the Joint Conference of the 9th Asia-Pacific Web Conference and the 8th International Conference on Web-Age Information Management (APWeb/WAIM07) 頁： 574-585 2007年6月

記述言語：英語

In recent years, the research of record extraction from large document data is becoming popular. However there still exist some problems in record extraction. 1) when large document data is used for the target of information extraction, the process usually becomes very expensive. 2) it is also likely that extracted records may not pertain to the user's interest on the aspect of the topic. To address these problems, in this paper we propose a method to efficiently extract those records whose topics agree with the user's interest. To improve the efficiency of the information extraction system, our method identifies documents from which useful records are probably extracted. We make use of user feedback on extraction results to find topic-related documents and records. Our experiments show that our system achieves high extraction accuracy across different extraction targets.

T-Scroll: Visualizing Trend in a Time-series of Documents for Interactive User Exploration 査読有り

Yoshiharu Ishikawa, Mikine Hasegawa

Proceedings of 11th European Conference on Research and Advanced Technology for Digital Libraries (ECDL 2007) 頁： 235-246 2007年9月

担当区分：筆頭著者記述言語：英語

On the Internet, a large number of documents such as news articles and online journals are delivered everyday. We often have to review major topics and topic transitions from a large time-series of documents, but it requires much time and effort to browse and analyze the target documents. We have therefore developed an information visualization system called T-Scroll (Trend/Topic-Scroll) to visualize the transition of topics extracted from those documents. The system takes periodical outputs of the underlying clustering system for a time-series of documents then visualizes the relationships between clusters as a scroll. Using its interaction facility, users can grasp the topic transitions and the details of topics for the target time period. This paper describes the idea, the functions, the implementation, and the evaluation of the T-Scroll system.

トピックを考慮した大規模文書情報源からのレコード抽出査読有り

張建偉，石川佳治，北川博之

情報処理学会論文誌：データベース 48 巻 ( SIG 14(TOD 35) ) 頁： 107-123 2007年9月

記述言語：日本語掲載種別：研究論文（学術雑誌）

近年，大量のテキスト文書からのレコード抽出の研究が行われている．レコード抽出には次の課題が存在する．第一に，大量の文書を情報抽出の対象とした場合に多大な処理コストがかかる．第二に，抽出されたレコードが，必ずしもユーザが興味あるトピックと合致しないことがある．これに対し本論文では，ユーザの意図に合った情報を効率よく抽出するためのレコード抽出手法を提案する．本手法では，効果的な抽出のために，ユーザの意図に適合した情報を含んでいる可能性の高い文書群を特定する．その特定した文書群を優先的に抽出処理に利用することで処理コストの削減を目指す．また，それらの文書群から内容の関連が深いレコードを抽出することで高い抽出精度を達成する．実験結果により，提案手法が抽出精度の低下を防ぎつつ，処理コストの削減を実現できることを示す．

曖昧な位置情報に基づく空間問合せの処理手法査読有り

石川佳治

日本データベース学会Letters 6 巻 ( 2 ) 頁： 49-52 2007年9月

担当区分：筆頭著者記述言語：日本語掲載種別：研究論文（学術雑誌）

センサ環境や移動ロボット応用では，測定の誤差やオブジェクト自身の移動により，オブジェクトの位置自体が曖昧である状況がしばしば生じる．本稿では，問合せオブジェクトの位置が正規分布に基づく確率密度関数で曖昧に指定されている状況における空間データベース問合せについて，その処理手法のアイデアを示す．

T-Scroll：時系列文書のクラスタリングに基づくトレンド可視化システム査読有り

長谷川幹根，石川佳治

情報処理学会論文誌：データベース SIG 20 巻 ( TOD 36 ) 頁： 61-78 2007年12月

記述言語：日本語掲載種別：研究論文（学術雑誌）

インターネット上では，ニュースなどの大量のテキストデータの配信が日々行われている．ここでは，ニュース記事のように，発行時刻などの情報を伴い次々と配信される文書のことを時系列文書と呼ぶ．大量の時系列文書の中から，着目した時期における主要なトピックや，トピックの継続的なつながりにより出現するより大きなトレンドをとらえたいという要求がしばしば発生するが，一般には，文書データの閲覧・分析のために多大な時間と労力が必要となる．そこで本研究では，時系列文書集合中に含まれるトレンドをとらえるための可視化システムであるT-Scroll（Trend/Topic-Scroll）の開発を行った．本システムは，下位の時系列文書を対象とするクラスタリングシステムが定期的に生成するクラスタリング結果をもとに，クラスタ間の関連を巻き物状に可視化して提示する．対話的な機能により，ユーザは文書集合中に含まれるトレンドや個々のトピックの詳細を把握することが可能となる．本論文では，システムのアイデア，機能，実現手法，評価について述べる．

A Novelty-based Clustering Method for On-line Documents 査読有り

Sophoin Khy, Yoshiharu Ishikawa, and Hiroyuki Kitagawa

World Wide Web Journal 11 巻 ( 1 ) 頁： 1-37 2008年3月

記述言語：英語掲載種別：研究論文（学術雑誌）

In this paper, we describe a document clustering method called novelty-based document clustering. This method clusters documents based on similarity and novelty. The method assigns higher weights to recent documents than old ones and generates clusters with the focus on recent topics. The similarity function is derived probabilistically, extending the conventional cosine measure of the vector space model by incorporating a document forgetting model to produce novelty-based clusters. The clustering procedure is a variation of the K-means method. An additional feature of our clustering method is an incremental update facility, which is applied when new documents are incorporated into a document repository. Performance of the clustering method is examined through experiments. Experimental results show the efficiency and effectiveness of our method.

Traceable P2P Record Exchange Based on Database Technologies 査読有り

Fengrong Li and Yoshiharu Ishikawa

Proceedings of the 10th Asia Pacific Web Conference (APWeb 2008) 頁： 475-486 2008年4月

記述言語：英語

Information exchanges in P2P networks have become very popular in recent years. However, tracing how data circulates between peers and how data modifications are performed during the circulation before reaching the destination are not easy because data replications and modifications are performed independently by peers. This creates a lack of reliability among the records exchanged. To provide reliable and flexible information exchange facilities in P2P networks, we propose a framework for a record exchange system based on database technologies. The system consists of three layers: a user layer, a logical layer and a physical layer. Its tracing operations are executed as distributed recursive queries among cooperating peers in a P2P network. This paper describes the concept and overviews the framework.

Monitoring Aggregate k-NN Objects in Road Networks 査読有り

Lu Qin, Jeffrey Xu Yu, Bolin Ding, and Yoshiharu Ishikawa

Proceedings of 20th International Conference on Scientific and Statistical Database Management (SSDBM 2008) 頁： 168-186 2008年7月

記述言語：英語掲載種別：研究論文（学術雑誌）

In recent years, there is an increasing need to monitor k nearest neighbor (k-NN) in a road network. There are existing solutions on either monitoring k-NN objects from a single query point over a road network, or computing the snapshot k-NN objects over a road network to minimize an aggregate distance function with respect to multiple query points. In this paper, we study a new problem that is to monitor k-NN objects over a road network from multiple query points to minimize an aggregate distance function with respect to the multiple query points. We call it a continuous aggregate k-NN (CANN) query. We propose a new approach that can significantly reduce the cost of computing network distances when monitoring aggregate k-NN objects on road networks. We conducted extensive experimental studies and confirmed the fficiency of our algorithms.

Traceable P2P Record Exchange: A Database-Oriented Approach 査読有り

Fengrong Li, Takuya Iida, and Yoshiharu Ishikawa

Frontiers of Computer Science in China 2 巻 ( 3 ) 頁： 257-267 2008年9月

記述言語：英語掲載種別：研究論文（学術雑誌）

In recent years, peer-to-peer (P2P) technologies are used for flexible and scalable information exchange in the Internet, but there exist problems to be solved for reliable information exchange. It is important to trace how data circulates between peers and how data modifications are performed during the circulation before reaching the destination for enhancing the reliability of exchanged information. However, such lineage tracing is not easy in current P2P networks, since data replications and modifications are performed independently by autonomous peers---this creates a lack of reliability among the records exchanged. In this paper, we propose a framework for traceable record exchange in a P2P network. By managing historical information in distributed peers, we make the modification and exchange histories of records traceable. One of the features of our work is that the database technologies are utilized for realizing the framework. Histories are maintained in a relational database in each peer, and tracing queries are written in the datalog query language and executed in a P2P network by cooperating peers. This paper describes the concept of the framework and overviews the approach to query processing.

A Query Language and Its Processing for Time-Series Document Clusters 査読有り

Sophoin Khy, Yoshiharu Ishikawa, and Hiroyuki Kitagawa

Proceedings of the 11th International Conference on Asia-Pacific Digital Libraries (ICADL 2008) 頁： 82-92 2008年12月

記述言語：英語

Document clustering methods for time-series documents produce a sequence of snapshots of clustering over time. Analyzing the contents (topics) and trends in a long sequence of clustering snapshots is hard and requires efforts since there are too many number of clusters; a user may need to access every cluster or read every document contained in each cluster. In this paper, we propose a framework to find clusters of user interest and change patterns called transition patterns involving the clusters. A cluster in a clustering result may persist in another cluster, branch into more than one cluster, merge with other clusters to form one cluster, or disappear in the adjacent clustering result. This research aims at providing users facilities to retrieve specific transition patterns in the clustering results. For this purpose, we propose a query language for time-series document clustering results and an approach to query processing. The first experimental results on TDT2 corpus clustering results are presented.

Range Query Processing for Imprecise Objects with Gaussian Distributions

Yoshiharu Ishikawa

The 4th Korea-Japan Workshop (KJDB 2008) 頁：（招待講演） 2008年9月

Querying Topic Evolution in Time Series Document Clusters 査読有り

Sophoin Khy, Yoshiharu Ishikawa, and Hiroyuki Kitagawa

7 巻 ( 3 ) 頁： 7-12 2008年12月

記述言語：英語掲載種別：研究論文（学術雑誌）

A document clusteringmethod for time series documents produces a sequence of clustering results over time. Analyzing the contents and trends in a long sequence of clustering results is a hard and tedious task since there are too many number of clusters. In this paper, we propose a framework to find clusters of users' topics of interest and evolution patterns called transition patterns involving
the topics. A cluster in a clustering result may continue to appear in or move to another cluster, branch into more than one cluster, merge with other clusters to form one cluster, or disappear in the adjacent clustering result. This research aims at providing users facilities to retrieve specific transition patterns in the clustering results. For this purpose, we propose a query language for time series document clustering results and an approach to query processing. The first experimental results on TDT2 corpus clustering results are presented.

Spatial Range Querying for Gaussian-Based Imprecise Query Objects 査読有り

Yoshiharu Ishikawa, Yuichi Iijima, and Jeffrey Xu Yu

Proceedings of the 25th International Conference on Data Engineering (ICDE 2009) 頁： 676-687 2009年4月

担当区分：筆頭著者記述言語：英語

In sensor environments and moving robot applications, the position of an object is often known imprecisely because of measurement error and/or movement of the object. In this paper, we present query processing methods for spatial databases in which the position of the query object is imprecisely specified by a probability density function based on a Gaussian distribution. We define the notion of a {\em probabilistic range query\/} by extending the traditional notion of a spatial range query, then present three strategies for query processing. Since the qualification probability evaluation of target objects requires numerical integration by a method such as the Monte Carlo method, reduction of the number of candidate objects that should be evaluated has a large impact on query performance. We compare three strategies and their combinations in terms of the experiments and evaluate their effectiveness.

`Pay-as-you-go' Processing for Tracing Queries in a P2P Record Exchange System 査読有り

Fengrong Li, Takuya Iida, and Yoshiharu Ishikawa

Proceedings of the 14th International Conference on Database Systems for Advanced Applications (DASFAA 2009) 頁： 323-327 2009年4月

記述言語：英語

In recent years, data provenance or lineage tracing has become an acute issue in the database research. Our target is the data provenance issue in peer-to-peer (P2P) networks where duplicates and modifications of data occur independently in autonomous peers. To ensure reliability among the exchanged data in P2P networks, we have proposed a reliable record exchange framework with tracing facilities based on database technologies in [5,6]. The framework is based on the "pay-as-you-go" approach in which the system maintains the minimum amount of information for tracing with low maintenance cost and a user pays the cost when he or she issues a tracing query to the system. This paper focuses on its two alternative query processing strategies and compare their characteristics according to the performance.

　前のページ - 次のページ　

<石川 佳治>

論文 - 石川 佳治

<石川　佳治>

論文 - 石川　佳治