(sigemb-info 398) [GRACE-staffs] [GRACE-announce] The 21st GRACE Seminar, (May 26th)

["Yumi YAMASAKI" <yumi1007@xxxxxxxxx>]

ご関係者各位

いつもお世話になっております。

本ＭＬをお借りいたしまして
第21回GRACEセミナー（5/26）のご案内を差し上げます。

尚、重複して案内を受け取られた方は
ご了承くださいますようお願い申し上げます。

田口　研治（ＮＩＩ）
Tel：03-4212-2604
e-mail：ktaguchi@xxxxxxxxx

====================================================================

第21回先端ソフトウェア科学・工学に関するGRACEセミナーのご案内
The 21st GRACE Seminar on Advanced Software Science and Engineering

下記の日程で、第21回GRACEセミナーを開催いたします。

  第21回 GRACEセミナー
  日時：2009年5月26日（火） 10:00-12:00
  場所：国立情報学研究所(NII) 12階会議室(1208)

参加費は無料です。
参加をご希望の方は、下記アドレスまで、セミナー前日まで
必要事項を記入し，"21st GRACE Seminar"というタイトルで
電子メールでご登録をお願いします。

 登録アドレス：event-info@xxxxxxxxxxxxxxx
 必要事項：お名前、所属、メールアドレス

I am pleased to announce the 21st GRACE Seminars on Advanced Software
Science and Engineering to be held on May 26th (Tue).

The registration fee is free but it is needed to register your name,
affiliation and e-mail address in advance.
Please send a mail titled "21st Grace Seminar"
including the information to event-info@xxxxxxxxxxxxxxxx

We are looking forward to meeting you at the GRACE seminars.

You can find the detailed information below:

=====================================================================

Details：

First Speaker:
Takuya Maekawa,
NTT Communication Science Laboratories

Title:
Real World Web Data Engineering
: Sensor and Web Data Processing for Real World Applications.

Abstract:
 This talk consists of an outline of studies of NTT CS Labs on ubiquitous
sensor networks and the speaker's studies on web content engineering in
ubiquitous sensor environments.
The speaker mainly introduces his works on ubiquitous sensors:
(1) web based lifelogging and (2) context-aware query-free web search.

(1) The Object-Blog presented here is a lifelog system in which personified
indoor daily objects automatically post weblog entries to a weblog about
sensor data obtained from sensors attached to the objects. Recent reductions
in data storage costs make it possible to store realworld data related to
normal daily living observed with ubiquitous sensors. However, many raw logs
will be write-only logs; they will be written but never accessed.
Raw logs should be processed into meaningful data events and aggregating the
events into a kind of chronicle. To move from raw logs to chronicles in the
ubiquitous-sensor environment, the Object-Blog system presented here
automatically generates web content from realworld phenomena.

(2) The new IR method presented here automatically searches for useful web
pages related to user's current activities of daily living by using sensor
data streams obtained from object usage sensors attached to daily objects.
The retrieved pages are to be shown on Internet-enabled appliances that
exist ubiquitously in our daily lives without rich input interface such as
televisions.
This talk is based on the IEEE Pervasive Computing and SIGIR'09 papers
coauthored with Yasue Kishino (NTT), Yasushi Sakurai (NTT), Koji Kamei
(NTT), Yutaka Yanagisawa (NTT West), and Takeshi Okadome (Kwansei Gakuin
University).

Bio:
 Takuya Maekawa is a researcher at NTT Communication Science Laboratories in
Japan. His research interests include Web content engineering in mobile and
ubiquitous environment. He received his Ph.D.
in information science and technology from Osaka University in 2006.

=======================================

Second Speaker:
Dr. Hideyuki Kawashima,
University of Tsukuba

Title:
Recent Advances in Data Stream Processing

Abstract：
 This talk introduces our recent activities in data stream processing.
Recently there are a number of stream information sources such as routers,
sensor devices or video cameras. On processing data streams, traditional
relational DBMS does not perform effectively.
The reasons are twofold.

 (1) Function aspect: RDBMS provides simple operations such as selection,
projection, join, aggregation, union, etc. The operators are effective for
static, clean, symbol-based data. While on the other hand, sensor data are
inherently dynamic, noisy and signal-based.
Applications of such sensor data usually require new functions such as
probabilistic reasoning or noise filtering.

(2) Performance aspect: RDBMS once stores incoming data items into
persistent devices, and then the data items are read into memories for
processing.
This methodology requires disk accesses, which incurs performance
degradation.
Since sensor data generate highly frequently, new architectural approaches
are required. This talk introduces our recent activities including
provenance management, pattern matching with Kleene plus and probabilistic
reasoning.

Bio:
 Hideyuki Kawashima is currently an assistant professor at Graduate School
of Systems and Information Engineering and at Center for Computational
Sciences, University of Tsukuba. He received the Ph.D. degree in computer
science from

Keio University in 2005. He was a research associate in Keio University from

2005 to 2007. He is a member of ACM, IEEE Computer Society, DBSJ, IEICE, and

IPSJ. His research interest lies in database system kernel, data stream
processing, e-Science, and data mining.

=================== GRACEセンターからのお知らせ ======================

GRACEセンター（先端ソフトウェア工学・国際研究センター）では、
今後GRACEセミナーなどのイベント案内や研究資料など、
GRACEセンターに関する情報をEメールにてお届けしています。

■お申し込みは↓■
http://grace-center.jp/mail_member.html
是非ご購読下さい！購読料は無料です。

[GRACEセンター]
http://www.grace-center.jp/index.html
[メールマガジンについてのお問い合わせ]
owner-grace-bulletin@xxxxxxxxx