YOKAN TABLE
An
Intermediary System between Notables and People
Kazuya Fujisawa, Ryutaro Motora
Graduate School of Media and Governance, Keio University, Minato, Tokyo, Japan
Michiaki Yasumura
Faculty of Environment and Information studies, Keio University, Minato, Tokyo, Japan
Keywords:
Microblog, Notables, Event extraction, Talkshow, Lecture.
Abstract:
There are a lot of people who want to meet active notables by whom he or she may obtain the chance of as a
human being. Currently, the informations to meet notables are scattered in various places on the Web, and we
have to view the Web pages either from news sites, word of mouth sites, blog sites, etc. to look for necessary
information. However it is troublesome to view the web pages carefully everyday, and it is a problem to
overlook the important information. In this study, we aim to make a lot of chances to meet notable people.
In order to collect and analysis the information of notables appearance that scattered on the web are collected
and analysis, we developed a system called YOKAN (presage) table with the function of notification of newly
arriving information on physical table.
1 INTRODUCTION
We are looking for the chance of growthing every day.
Especially, meeting notables is the best opportunity
because it brings not only knowledge and knowhow
but also a lot of inspiration. Currently, there is a lot of
information to meet notables like the notification of
the talk events etc. But then is no much, we have to
view the Web page of the news sites, blog sites. More-
over we also need to look for necessary information
carefully, and it is a problem that we often overlook
the information. In addition, we should always check
the newly arrived information. It is difficult to search-
ing an interesting notable daily, and checking the site
carefully everyday.
Therefore, we’d like to design and implement the
Web service called Yokan (presage) table that collects
notable’s appearance information. In this study, we
aim to make a lot of chance to meet notable people.
We designed and implemented Yokan talble.
2 YOKAN TABLE
The Yokan table is a systems that notifies newly arriv-
ing information on physical table. In this research, we
defined the design policy that people can collect in-
formation everyday without the stress, and can keep
user continuously access system. Hereafter, we ex-
plain the function of the Web part and the table part
of the system.
2.1 Web Part of the System
In this study, we focused on the talk event as a good
opportunity to meet a notable. The talk show and lec-
ture meeting informations that indicated in the future
are collected from web automatically. We defined fu-
ture information is within three months. We utilized
the Twitter that display the information in time line
(Figure 1).
User’s judgment index, includes not only essential
information of (1) the date of talk-events (2) the no-
table name (2) and (3) talk-events name, but also (4)
informationof notable specialized field and (5) degree
of attention are displayed in addition. Information is
updated every 24 hours.
2.2 Physical Table Part of the System
The purpose of the table part is to induce the user
from daily life to the Web page (Figure 2). And it
197
Fujisawa K., Motora R. and Yasumura M. (2011).
YOKAN TABLE - An Intermediary System between Notables and People.
In Proceedings of the 1st International Conference on Pervasive and Embedded Computing and Communication Systems, pages 197-200
DOI: 10.5220/0003366601970200
Copyright
c
SciTePress
Figure 1: Screenshot of Web service.
Figure 2: Yokan Table appearance.
informs the sense that comes close to famous people.
In this study, using metaphor that a surrounding thing
vibrated because of the rumbling of the earth when
the massive creature moved. Desktop coaster part
is prepared, and the glass of water will shake glass.
The size of the shake changes depending on degree
of attention of talk-event, and the color of the coaster
changes depending on notable specialized field.
The users feels the scene that a notable ap-
proached on the YOKAN table in breakfast and din-
ner, and checking detailed information on the Web
service before going to school, commuting, and break
time.
3 IMPLEMENTATION
This chapter describes implementation of the soft-
ware and the hardware of the YOKAN table.
3.1 The Software
To obtain talk-event information of notables, Twitter
of the microblog service was used. It explains that
semantics step of the system as follows (Figure 3).
1. First of all, by using Twitter search API” that
Twitter offers, the comment that is called ”tweet”
where the keyword ”talk-show” or ”lecture” in-
cluded is acquired. These data is brought together
every day, and it processes day by day.
2. Next, to make the quote comment that is called
”retweet” and ”tweet” of the similar content
group, the vector space model is applied to the
similarity calculation of ”in one tweet” and ”an-
other tweet”. To make the feature vector of each
”tweet”, the morphological analysis is done to
each ”tweet”, and elements count of feature vec-
tor is composed of the number of all noun, each
elements weight is composed of the count of the
corresponding noun.
3. For the similarity calculation, using cosine simi-
larity. The cosine similarity of feature vector X
and feature vector Y is calculated in expression
(1). Merge of ”tweet” is done by using the short-
est distance method of a hierarchical clustering,
and merge is repeated until cosine similarity be-
come under the sub threshold. When clustering is
finished, each cluster becomes one event.
sim(X, Y) =
X ·
Y
|X| ·
|Y|
=
i
x
i
y
i
r
i
x
2
i
·
r
i
y
2
i
(1)
x
i
Weight of element
i
of feature vector X
y
i
Weight of element
i
of feature vector Y
4. Finally, name of person, the person’s category and
events name have been extracted from each event
information. The morphological analysis is done
to each event information. When the combination
of last name and first name existed, it extracted
as a name of person. Person’s tag information is
extracted from person search engine ”Fined him
in SPYSEE” (http://spysee.jp/), and the person’s
category is decided from tag information by the
rule that the system decided beforehand. By the
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
198
HTTP access to URL that exists in event informa-
tion, contents of the Title tag in the HTML source
has been extracted as an event name.
3.2 The Hardware
Figure 3 showed the structure of hardware. The
hardware of the YOKAN table used a wooden table
(100mm x 70mm x 70mm) on the market. To keep
space that built in the mechanism internally, the top
board with space was added. And space that holds
the mechanism internally was secured. we used the
DC solenoid (12V) and the 3M Co. pocket projector
(MPro120), for shaking mechanism and for color pro-
jection mechanism respectively. For we used Arduino
to detect the glass on the coaster. And the system also
controls the DC solenoid and projection color.
Figure 3: System configuration diagram.
4 DISCUSSION
We found that more than 90% is next three month
information in the extracted information. Because
Figure 4: Hardware configuration diagram and structure.
this future information often contains URL, an official
Web site of talk-events information can be acquired.
As a result, it has been understood to obtain useful
information by focused on at the date.
Recently, the weight of the feature vector uses
only the number of the nouns, so we have to examine
weight. Moreover, the threshold of the cosine simi-
larity of clustering is set to ”0.5”. But after a detailed
experiment is done, it is necessary to fix this thresh-
old.
From now on, we will explain comparative exper-
iments on the data acquired from micro blog and gen-
eral search engine, and experiment on utility using a
lot of data and examinee.
5 RELATED WORKS
It remarkable that the number of Twitter user is 11
milion in Japan on November, 2010. Because a lot
of services that focused on event informations and re-
searches that using information on the micro blog ex-
ists, and we introduces them in this chapter.
There are some researches that regard each user
who sends information from microblog as one sen-
sor (Fujisawa, 2010). That means regard Micro blog
as network service observes real space, and it tries to
understand various events that occur by a real space
from the sensing from the user. Those analyses are
used the user’s location information and the remark
time as index. This research is same as our study
about focus on event informations. But our study fo-
cuses on not user’s location informations and post-
ing time, but date informations of ”Tweet” to extract
event informations.
Web service called ”KOTOSAGA” that helps to
see event informations (http://www.cotosaga.com/).
It uses the crawler of original development, and it
collects event informations automatically from more
YOKAN TABLE - An Intermediary System between Notables and People
199
than 700 event informations sites. ”KOTOSAGA” is
based on the element of ”date” and ”place”, so we
can also know information around the event hall, for
example, weather, map etc.
6 CONCLUSIONS
In this study, we aim try to give chance for people who
want to meet notable. For this purpose our system
called Yokan Table that collects notable’s event infor-
mation. And the Yokan table that notify newly arrived
information on desktop glass was designed and im-
plemented. Usually, to meet notable, we should view
the Web page from the news site, by word of mouth
site, site of notable himself etc. Moreover we also
need to look for necessary information carefully, and
it is the problem that we often overlook the informa-
tion. Because Yokan table installed the entrance to
web in daily life, users can obtain the chance meeting
notable more comfortably. In the future, we want to
aim at personalizing to offer information matched to
user’s interest.
REFERENCES
T. Fujisaka, R. Lee, K. Sumiya 2010. Estimating Influence
Regions of Social Events by Geo-tagged Micro-Blogs
Analysis, DEIM Forum 2010 D7-4.
KOTOSAGA. http://www.cotosaga.com/
SPYSEE. http://spysee.jp/
PECCS 2011 - International Conference on Pervasive and Embedded Computing and Communication Systems
200