DASFAA2014：国际数据库系统应用研讨会论文集

需积分: 10 189 浏览量更新于2024-07-17 收藏 39.13MB PDF 举报

"Database Systems for Advanced Applications - 19th International Conference, DASFAA2014, International Workshops: BDMA, DaMEN, SIM³, UnCrowd" 这篇资源是一本关于高级应用数据库系统的论文集，由一系列在2014年DASFAA（Database Systems for Advanced Applications）国际会议上发表并修订后的精选论文组成。DASFAA是一个重要的学术会议，专注于数据库系统领域的最新研究和发展。会议同时包括了几个专题工作坊：BDMA（Big Data Management and Analytics）、DaMEN（Data Management in Edge Networking）、SIM³（Social Interaction, Information Modeling and Management in 3D）以及UnCrowd（Uncertainty Reasoning for Crowdsourced Data）。这些工作坊探讨了数据库技术在大数据管理、边缘计算、社交网络信息建模与管理以及基于众包数据的不确定性推理等前沿领域。论文集涵盖了数据库系统的多个核心主题，如数据存储、查询处理、索引技术、事务处理、数据安全、数据挖掘、大数据分析、云计算环境下的数据库管理以及新兴的非传统数据库模型（如NoSQL和NewSQL）。通过这些论文，读者可以了解到如何利用先进的数据库技术来解决复杂的应用问题，例如高效的数据检索、大规模数据的处理和分析、数据安全和隐私保护策略，以及适应不断变化的业务需求的弹性数据库设计。此外，这个资源还提到了一系列知名学者的名字，如Wook-Shin Han、Mong Li Lee、Agus Muliantara、Ngurah Agus Sanjaya、Bernhard Thalheim和Shuigeng Zhou，他们都是数据库和计算机科学领域的专家，这暗示了论文集的专业性和权威性。《Lecture Notes in Computer Science》系列是计算机科学领域的知名出版物，自1973年开始出版，收录了许多高质量的学术论文，为研究人员、学生和从业者提供了深入学习和参考的宝贵资料。 "Database Systems for Advanced Applications"是数据库和信息管理专业人士以及对相关领域感兴趣的学生的重要参考资料，它反映了当前数据库系统研究的最新进展和技术趋势。通过阅读这些论文，读者可以提升对高级数据库系统设计、实现和应用的理解，以及如何在实际场景中有效地利用这些技术来解决现实世界的问题。

Solving Multiple Bichromatic Mutual Nearest Neighbor Queries

with the GPU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308

Marta Fort and J. Antoni Sellarès

A Kernel Density Method for Aggregating Boundary Collision Data

into Areal Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317

Ge Cui, Xin Wang, and Dae-Won Kwon

Integrated Indoor Positioning with Mobile Devices for Location-Based

Service Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329

Bei Huang and Yang Gao

A Hybrid Scale-Out Cloud-Based Data Service for Worldwide Sensors . . . . 342

Tania Khalafbeigi, Chih-Yuan Huang, Steve Liang, and Mea Wang

DASFAA Workshop on Uncertain and Crowdsourced Data (UnCrowd)

Uncertainty in Crowd Data Sourcing Under Structural Constraints . . . . . . . . 351

Antoine Amarilli, Yael Amsterdamer, and Tova Milo

Integration of Web Sources Under Uncertainty and Dependencies

Using Probabilistic XML . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360

M. Lamine Ba, Sebastien Montenez, Ruiming Tang, and Talel Abdessalem

Skill Ontology-Based Model for Quality Assurance in Crowdsourcing . . . . . 376

Kinda El Maarry, Wolf-Tilo Balke, Hyunsouk Cho, Seung-won Hwang,

and Yukino Baba

ProbKS: Keyword Search on Probabilistic Spatial Data. . . . . . . . . . . . . . . . 388

Feng Gao, Rohit Jain, Sunil Prabhakar, and Luo Si

Towards Mobile Sensor-Aware Crowdsourcing: Architecture,

Opportunities and Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403

Jiyin He, Kai Kunze, Christoph Lofi, Sanjay K. Madria, and Stephan Sigg

Conditioning Probabilistic Relational Data with Referential Constraints . . . . 413

Ruiming Tang, Dongxu Shao, M. Lamine Ba, and Huayu Wu

Author Index ............................................ 429

Contents XXI

Meme Media and Knowledge Federation

for Exploratory Visual Analytics of Big Data

Yuzuru Tanaka

(&)

Meme Media Laboratory, Hokkaido University, Sapporo, Japan

tanaka@meme.hokudai.ac.jp

Abstract. This paper proposes the use of meme media and knowledge

federation technologies as the basis for a generic framework of exploratory

visual analytics of big data. We ﬁrst propose a ‘‘coordinated multiple views’’

visualization framework, and then extend this to a ‘‘coordinated multiple

analyses’’ visualization framework for the integration of clustering tools, fre-

quent pattern mining tools, and statistical analysis tools into the framework.

The webtop meme media system Webble World works as the enabling tech-

nology to implement these frameworks. Its improvisational knowledge feder-

ation capability allows us to improvisationally federate available external tools

and services to work together in our framework. This provides our exploratory

visual analytics framework with a large library of tools and data sources open

for future extension.

Keywords: Big data analysis

 Exploratory visual analytics  Coordinated

multiple views

Coordinated multiple analyses Meme media Clinical trial 

Social cyber-physical system

1 Introduction

One of the problems we are facing these days in big data R&Ds may be a big gap

between the core technology R&Ds and the application R&Ds. Through the

involvement both in EU FP6 and FP7 projects on the integrated IT support of clinical

trials on cancers and in Japanese government-initiative project on social cyber-

physical systems for optimizing social system services such as the snow plowing and

removing in Sapporo City, which has 1.9 million people and the average annual

snowfall of 6 m, the current author has been also facing the difﬁculties to ﬁll in the gap

between varieties of available data analysis methods and the goals to ﬁnd out new

meaningful personalized medicine or the optimized resource scheduling for the snow

plowing and removing.

Each clinical trial ends up with an accumulation of patient treatment data and

patient diagnosis data including DICOM images and patient genomic data. The

number of patients in each clinical trial may range from hundreds to thousands. While

the data size is not extremely large, their analysis is not so straight forward as applying

any ‘‘planned-for’’ analysis scenarios. The objective there is to ﬁnd out new person-

alized medical treatments, each of which shows the best recovery rate for some

speciﬁc group of patients. We need to ﬁnd out each of these speciﬁc patient groups for

W.-S. Han et al. (Eds.): DASFAA 2014, LNCS 8505, pp. 3–17, 2014.

DOI: 10.1007/978-3-662-43984-5_1,  Springer-Verlag Berlin Heidelberg 2014

which one of the candidate treatment arms after the randomization in the master trial

plan may show the best recovery rate than the others. The analysis process there is

inherently an exploratory repetition of the hypothesis making through the segmen-

tation of some patient group and the hypothesis checking through some analysis and

visualization of the segmented patient group. The personalized medicine has become

more focused since it became obvious that the best treatment for a patient is not

necessarily the best for another patient who shows the same macroscopic properties as

the ﬁrst one. This means that some macroscopic analysis of a patient group that is

segmented with respect to their macroscopic properties does not work well.

The same is true in the analysis of how snowfalls and the snow plowing and

removing may inﬂuence the trafﬁc of each road segment. Using the average speed in

each road segment at every 5 min interval for 24 h, we can characterize each road

segment as a vector of 288 dimensions. The same clustering method classiﬁes the

consecutive road segments along the same major route in the central Sapporo into the

same cluster for the data on the next day of the complete snow plowing and removing,

but in different clusters for the data on a day with a heavy snowfall. This indicates that

the inﬂuence of snow to the trafﬁc is not uniform, and that we cannot apply macro

analysis methods to obtain any meaningful knowledge about winter roads.

These two examples have convinced the current author the necessity of explor-

atory visual analytics.

2 Requirements for Exploratory Visual Analytics

It is well known that exploratory visual analytics [1–4] may use the ‘‘coordinated

multiple views’’ visualization framework [5] as its basis. This framework provides

more than one view for the visualization of the same database D from different aspects

as shown in Fig. 1. Each view V

may be a chart view, a map view, a graph repre-

sentation view, or a calendar view, and shows the evaluation result Q

(D) of some

query Q

associated with this view using its speciﬁc visualization scheme. Each view

allows users to select a set of visualized objects by directly specifying each of them or

enclosing some of them, which deﬁnes a new additional quantiﬁcation condition C to

quantify the objects stored in the underlying database D. This quantiﬁcation deﬁnes a

new database viewD’ deﬁned as follows:

CREATE VIEWD’

SELECT *

FROM D

WHERE C

Each view V

including V

itself then immediately changes its visualization from Q

(D)toQ

(D’), or just highlights the objects in Q

(D’) in the visualization of Q

(D),

depending on the user speciﬁcation of its visualization mode. Multiple visualization

views are mutually coordinated in this sense.

In exploratory visual analytics with a coordinated-multiple-views visualization

system, each user may start with the original databaseD, and repetitively try different

4 Y. Tanaka

selections of visual objects on different views for the exploration of different quan-

tiﬁcations on database objects to ﬁnd out a meaningful group of database objects.

He or she may roll back the preceding visual object selection to try a different

quantiﬁcation through a different visualization view. Figure 2 schematically shows

such a process of exploratory visual analytics.

Each view in coordinated-multiple-views visualization is, however, just a database

visualization view. No analysis is actually applied in such exploratory visual analytics

processes. In order to apply analysis tools to quantiﬁed sets of objects, we need to

integrate these tools together with their analysis result visualizations into coordinated

multiple view visualization. Many researchers emphasized the importance of inte-

grating various analyses and visualizations [6]. To the best of our knowledge, apart

from some statistical chart tools to show histograms, correlations, or heatmaps [7], no

other analysis tools such as clustering and frequent pattern mining tools have ever

been integrated into coordinated multiple views visualization to allow users to directly

select a cluster or some of the mined frequent patterns for further quantifying database

objects and for further analyzing those quantiﬁed database objects. Varieties of

analysis tools are rapidly increasing these days. In order to keep our visual analytics at

the current state of the art, we need to be able to make any new analysis tools available

and operational in our visual analytics.

Exploratory visual analytics requires a coordinated-multiple-views visualization

framework to which we can integrate any analysis tools so that their result visual-

izations may be also coordinated with other visualization views and analysis result

Vis1

Vis2

Visi

selection

Vis1

Vis2

Visn

selection

Visn

Visj

Vis1

Vis2

Visn

map

histogram

24-hour

clock

scatter

Fig. 2. A process of exploratory visual analytics.

Database or Semantic Web

( ) Q

( )

( )Q

( )

map

histogram

24 hour

clock

scatter graph

Fig. 1. A coordinated-multiple-views visualization framework.

Meme Media and Knowledge Federation for Exploratory 5

剩余438页未读，继续阅读

J-10

粉丝: 18
资源: 483

DASFAA2014：国际数据库系统应用研讨会论文集

英文原版高级应用数据库系统

Absolute Database for D7

TimesTen In-Memory Database Installation Guide

Modern Operating Systems,Tanenbaum,3rd

Deploying Advanced Web Crawlers and Building a Cluster: Using Docker to Containerize Crawler ...

[Advanced Chapter] Detailed Explanation of GUI Design and Interactive Applications in MATLAB

Navicat Connecting to MySQL Database: 10 Solutions for Connection Failures

Navicat Connecting to MySQL Database: Unveiling Common Error Codes for Easy Troubleshooting ...

【Advanced】Image Face Detection in MATLAB: Using MTCNN for Image Face Detection

【Database Model Design】: The Golden Rules for Building Efficient Structures with Python and MySQL

最新资源