Introduction
Background and Objective
Today, mobile users can access information from various Web sites
anywhere and anytime via browsers on their personal digital
assistants (PDAs). To support such users, the number of Web sites
that provide location-dependent information is continuously
growing. A Web site (denoted by Web Information sources or
WIS from here on) is called location-dependent if it
provides contents which vary according to users' current
locations. Typically, WISs are made of client applications and
backend server. Mobile users can query these WISs' contents
through the client applications (e.g., CGI-forms, java applets, or
ordinary document pages) of WISs, which are downloaded into users'
PDAs and interact with the database backend servers of the WISs
that generate result-pages according to individual input
conditions.
Generally, there are various WISs in the Internet, and some WISs
provide location-dependent contents and some do location-independent
ones. For efficient moving, mobile users must find useful information
from combination of these Web contents. To do so, the users need to
query information from many related WISs. A current solution to avoid
this complexity is integration of such WISs as a new single,
simpler information source. This topic is studied well in the
literature of Web integration [YAT, STRUDEL, WebSurvey,
ARANEUS], but most of the studies assume that such an integration
is made in advance by a system designer. However, in the
above mobile Internet applications, it would be too expensive for a
system designer to prepare all possible combinations of integrations
for all location-dependent/-independent WISs for all users. We think
that which pairs of WISs should be integrated as a new useful WIS is
dependent on individual mobile users; i.e., it is decided by a mobile
user according to his own purpose, his current location, and his other
attributes (e.g., his historical moving path or a set of interesting
WISs he found.) Thus, it will be beneficial that a mobile user can
freely connect contents of one WIS with services of other related WISs
according to his own requirement on his PDA. This paper proposes such
a framework that allows mobile users to define appropriate integration
of WISs on their PDAs whenever they want.
As an example, consider an exhibition of computer companies, and
assume a WIS which announces the list of ten exhibitor booths that
are nearer to current locations of individual users. Then, assume
that a mobile user (i.e., participant) can integrate, on his PDA,
the service of this WIS with the services of other WISs concerned
with the exhibition, such as a time-schedule server of
presentations, or a detailed document server about companies in
each booth. (Here, by the term ``integrate WISs'', we
mean a situation that a mobile user connects functions of several related
WISs together and use them as a single new WIS.) Then the new WIS
resulting from the integration will become a useful portal site
personalized for the user.
A significant point in this example is that individual mobile
users now have the right to decide which pairs of WISs to be
integrated in what ways; a system designer must prepare only an
environment to support such flexibility of integration.
This paper proposes this framework.
For this framework, a style of integrating WISs should be much
simpler for mobile usage. It is also necessary that a mobile user
can specify this integration on his PDA even if the PDA is
disconnected from the network. Here, because of functional
simplicity of PDAs, what to be done by the user on his PDA must be
only to define the integration. (e.g., users only need to write a
kind of database view definition by a query command.)
Materializing such a view as a new WIS must be executed later by
some system-side servers when the PDA is network-connected.
In this paper, we firstly propose a new style of WIS integration for
the above usage. This style is termed ``navigational integration''
[PACRIM99]. Roughly speaking, navigational integration
is to connect functions of given WISs in such a way that each
data-record (about an exhibition, in the aforementioned example) in
one WIS is given a new link which calls services of another WIS
related with that data-record. After this proposal, our system
architecture is described.
Assumed Environment
We model here an assumed environment and identify our problem from a
technical viewpoint.
The above figure describes our assumptions of a system environment.
An area is an autonomous organization of maintaining some
WISs. WISs (such as ordinary Web document servers, CGI or Java
applications with database backend) can provide information about some
areas such as a building, a university or a city. Also, accesses to
some WISs can be restricted to those from certain areas. Further, some
WISs may provide location-dependent information, and some may be
location-independent.
Preparing all useful integration of these WISs would be an expensive
task with much delay for system designers. Rather, it is a better
choice to allow a mobile end-user in such areas to integrate
appropriate WISs which he found during his moving. Under this
assumption, our problem is identified by these two items:
- a style of integration must be simple enough for mobile users
to easily integrate various WISs as a single WIS.
- Our system must support the following activities:
- a mobile user caches minimal metadata from WISs into
his PDA during network connection;
- Thereafter, he defines the navigational integration on his
PDA during network-disconnection. (The definition is given by a
query command on the cached metadata, in the same way as a
database view definition is so.);
- Once the definition is made, he can materialize it as a new
WIS anytime when the PDA is connected to the network. (there
must be system-side wrapper/mediation servers to execute the
materialization.)
As for related works, previous researches [WebOQL, TSIMMIS, YAT,
STRUDEL, ARANEUS] integrate Web contents as new Web structures
through database views. This approach can perfectly manipulate Web
documents but it needs complicated query languages. In contrast,
navigational integration is a way to simply connect functions of WISs
via appropriate data-records. This will be simple enough for a mobile
user to easily integrate a wide class of WISs. In research trends of
Web integration, our study has a direction to export some functions of
Web integration techniques to mobile users and to thereby alleviate
complexity of preparations of system designers. Because new mobile
Internet services (e.g., many location-dependent WISs and smart phones
capable of Web-browsing) are appearing today, our study is worthy to
be examined.
Reference
[WebOQL] G. Arocena and A. Mendelzon. Weboql: Restructuring
documents, databases and webs. In Proc. of 14th. Intl. Conf. on
Data Engineering(ICDE), Orando, Florida, 1998.
[TSIMMIS] S. C. et al. The tsimmis project: Integration of
heterogenous information sources. In Proc. 10th of Transactions
of Information Processing Society of Japan Tech. Rep. DBS, pages
7--18, Tokyo, Japan, 1994.
[YAT] S. C. et al. Your mediator needs data conversion! In
Proc. of the ACM SIGMOD Conf. on Management of Data, pages
177--188, Washington, USA, 1997.
[STRUDEL] M. Fernandez, D. Florescu, A. Levy, and D. Suciu. A
query language for a web site management system. SIGMOD
Record, 3(26):4--11, September 1997.
[WebSurvey] D. Florescu, A. Levy, and A. Mendelzon. Database
techniques for the world-wide web: Survey. SIGMOD Record,
3(27):59--74, September 1998.
[ARANEUS] P. M. P. Atzeni, G. Mecca. To weave the web. In
Proceeding of the 23th VLDB Conference, pages 206--215,
[GARLIC] M. T. Roth and P. Schwarz. Don't scrap it, wrap it! a
wrapper architecture for legacy sources. In Proceeding of the
23th VLDB Conference, pages 266--275, Athens, Greece, 1997.
[PACRIM99] W. Sae-Tung, T. Ohmori, and M. Hoshi. Navigational
integration of autonomous web information sources by mobile users.
Proc. IEEE Pacific Rim Conference on Communications, Computers
and Signal Processing, pages 270--275, August 1999.