zuh/ITDataset · Datasets at Hugging Face

text stringlengths 0 134
contrast, data "recharging" has none of these attributes.

The Data Recharging project is developing a service and corresponding
infrastructure that permits a mobile device of any kind to plug into the Internet at any
location for any amount of time and as a result, end up with more useful data than it
had before [CFZOO]. As with power recharging, the initiation of a data charge simply
requires "plugging in" a device to the network. The longer the device is left plugged
in, the more effective the charge. Although similar to battery recharging, data
recharging is more complex; the type and amount of data delivered during a data
charge must be tailored to the needs of the user, the capabilities of the recharged
device, and the tasks that the recharged data is needed to support.

Different mobile users will have different data needs. A business traveler may
want updates of contact information, restaurant reviews and hotel pricing guides
specific to a travel destination. Students may require access to recent course notes,
required readings for the next lecture and notification about lab space as it becomes
available. Data recharging represents specifications of user needs as profiles.
Profiles can be thought of as long-running queries that continually sift through the
available data to find relevant items and determine their value to the user.

Profiles for data recharging contain three types of information: First, the profile
describes the types of data that are of interest to the user. This description must be
declarative in nature, so that it can encompass newly created data in addition to
existing data. The description must also be flexible enough to express predicates
over different types of data and media. Second, because of limitations on bandwidth,
device-local storage, and recharging time, only a bounded amount of information can
be sent to a device during data recharging. Thus, the profile must also express the
user s preferences in terms of priorities among data items, desired resolutions of
multi-resolution items, consistency requirements, and other properties. Finally, user
context can be dynamically incorporated into the recharging process by
parameterizing the user profile with information obtained from the device-resident
Personal Information Management (PIM) applications such as the calendar, contact
list, and To Do list.

Our previous work on user profiles has focused on 1) efficiently processing
profiles over streams of XML documents (i.e., the XFilter system) [AFOO], 2)
learning and maintaining user profiles based on explicit user feedback [CFGOO], and
3) development of a large-scale, reliable system for mobile device synchronization
[DFOO]. Data recharging can build upon this work, but requires the further
development of a suitable language and processing strategy for highly expressive user
profiles (i.e., that include user preference and contextual information), and the
development of a scalable, wide-area architecture that is capable of providing a data
recharging service on a global basis to potentially millions of users.

Adaptive Dataflow Query Processing

A second important aspect of ubiquitous computing environments is the variable
nature of data availability and the challenges of managing and processing dynamic
data flows. In mobile applications for example, data can move throughout the system
in order to follow the users who need it. Conversely, in mobile applications where
the data is being created at the endpoints (say, for example, a remote sensing
application) data streams into the system in an erratic fashion to be processed, stored,
and possibly acted upon by agents residing in the network. Information flows also
arise in other applications, such as data dissemination systems in whieh streams of
newly created and modified data must be routed to users and shared caches.

Traditional database query processing systems break down in such environments
for a number of reasons: First, they are based on static approaches to query

optimization and planning. Database systems produce query plans using simple cost
models and statistics about the data to estimate the cost of running particular plans.
In a dynamic dataflow environment, this approach simply does not work because
there are typically no reliable statistics about the data and because the arrival rates,
order, and behavior of the data streams are too unpredictable [UFA98].

Second, the exisiting approaches cannot adequately cope with failures that arise
during the processing of a query. In current database systems, if the failure of a data
source goes undetected, the query processor simply blocks, waiting for the data to
arrive. If a failure is detected, then a query is simply aborted and restarted. Neither
of these situations is appropriate in a ubiquitous computing environment in which
sources and streams behave unpredictably, and queries can be extremely long-running
(perhaps even continuous ).

Third, existing approaches are optimized for a batch style of processing in which
the goal is to deliver an entire answer (i.e., they are optimized for the delivery of the
last result). In a ubiquitous computing environment, where users will be interacting
with the system in a fine-grained fashion, such approaches are unacceptable.
Processed data (e.g., query results, event notifications, etc.) must be passed on to the
user as soon as they are available. Furthermore, because the system is interactive, a
user may choose to modify the query on the basis of previously returned information
or other factors. Thus, the system must be able to gracefully adjust to changes in the
needs of the users [HACO-i-99].

The Telegraph project at UC Berkeley [HFCD-l-00] is investigating these issues
through the development of an adaptive dataflow processing engine. Telegraph uses a
novel approach to query execution based on eddies , which are dataflow control
structures that route data to query operators on an item-by-item basis [AHOO].
Telegraph does not rely upon a traditional query plan, but rather, allows the plan to
develop and adapt during the execution. For queries over continuous streams of data.
the system can continually adapt to changes in the data arrival rates, data
characteristics, and the availability of processing, storage, and communication
resources.

In addition to novel control structures. Telegraph also employs non-blocking,
symmetric query processing operators, such as XJoins [UFOO] and Ripple Joins
[HH99], which can cope with changing and unpredictable arrival of their input data.
The challenges being addressed in the Telegraph project include the development of
cluster-based and wide-area implementations of the processing engine, the design of
fault-tolerance mechanisms (particularly for long-running queries), support for
continuous queries over sensor data and for profile-based information dissemination,

contrast, data "recharging" has none of these attributes.

The Data Recharging project is developing a service and corresponding

infrastructure that permits a mobile device of any kind to plug into the Internet at any

location for any amount of time and as a result, end up with more useful data than it

had before [CFZOO]. As with power recharging, the initiation of a data charge simply

requires "plugging in" a device to the network. The longer the device is left plugged

in, the more effective the charge. Although similar to battery recharging, data

recharging is more complex; the type and amount of data delivered during a data

charge must be tailored to the needs of the user, the capabilities of the recharged

device, and the tasks that the recharged data is needed to support.

Different mobile users will have different data needs. A business traveler may

want updates of contact information, restaurant reviews and hotel pricing guides

specific to a travel destination. Students may require access to recent course notes,

required readings for the next lecture and notification about lab space as it becomes

available. Data recharging represents specifications of user needs as profiles.

Profiles can be thought of as long-running queries that continually sift through the

available data to find relevant items and determine their value to the user.

Profiles for data recharging contain three types of information: First, the profile

describes the types of data that are of interest to the user. This description must be

declarative in nature, so that it can encompass newly created data in addition to

existing data. The description must also be flexible enough to express predicates

over different types of data and media. Second, because of limitations on bandwidth,

device-local storage, and recharging time, only a bounded amount of information can

be sent to a device during data recharging. Thus, the profile must also express the

user s preferences in terms of priorities among data items, desired resolutions of

multi-resolution items, consistency requirements, and other properties. Finally, user

context can be dynamically incorporated into the recharging process by

parameterizing the user profile with information obtained from the device-resident

Personal Information Management (PIM) applications such as the calendar, contact

list, and To Do list.

Our previous work on user profiles has focused on 1) efficiently processing

profiles over streams of XML documents (i.e., the XFilter system) [AFOO], 2)

learning and maintaining user profiles based on explicit user feedback [CFGOO], and

3) development of a large-scale, reliable system for mobile device synchronization

[DFOO]. Data recharging can build upon this work, but requires the further

development of a suitable language and processing strategy for highly expressive user

profiles (i.e., that include user preference and contextual information), and the

development of a scalable, wide-area architecture that is capable of providing a data

recharging service on a global basis to potentially millions of users.

Adaptive Dataflow Query Processing

A second important aspect of ubiquitous computing environments is the variable

nature of data availability and the challenges of managing and processing dynamic

data flows. In mobile applications for example, data can move throughout the system

in order to follow the users who need it. Conversely, in mobile applications where

the data is being created at the endpoints (say, for example, a remote sensing

application) data streams into the system in an erratic fashion to be processed, stored,

and possibly acted upon by agents residing in the network. Information flows also

arise in other applications, such as data dissemination systems in whieh streams of

newly created and modified data must be routed to users and shared caches.

Traditional database query processing systems break down in such environments

for a number of reasons: First, they are based on static approaches to query

optimization and planning. Database systems produce query plans using simple cost

models and statistics about the data to estimate the cost of running particular plans.

In a dynamic dataflow environment, this approach simply does not work because

there are typically no reliable statistics about the data and because the arrival rates,

order, and behavior of the data streams are too unpredictable [UFA98].

Second, the exisiting approaches cannot adequately cope with failures that arise

during the processing of a query. In current database systems, if the failure of a data

source goes undetected, the query processor simply blocks, waiting for the data to

arrive. If a failure is detected, then a query is simply aborted and restarted. Neither

of these situations is appropriate in a ubiquitous computing environment in which

sources and streams behave unpredictably, and queries can be extremely long-running

(perhaps even continuous ).

Third, existing approaches are optimized for a batch style of processing in which

the goal is to deliver an entire answer (i.e., they are optimized for the delivery of the

last result). In a ubiquitous computing environment, where users will be interacting

with the system in a fine-grained fashion, such approaches are unacceptable.

Processed data (e.g., query results, event notifications, etc.) must be passed on to the

user as soon as they are available. Furthermore, because the system is interactive, a

user may choose to modify the query on the basis of previously returned information

or other factors. Thus, the system must be able to gracefully adjust to changes in the

needs of the users [HACO-i-99].

The Telegraph project at UC Berkeley [HFCD-l-00] is investigating these issues

through the development of an adaptive dataflow processing engine. Telegraph uses a

novel approach to query execution based on eddies , which are dataflow control

structures that route data to query operators on an item-by-item basis [AHOO].

Telegraph does not rely upon a traditional query plan, but rather, allows the plan to

develop and adapt during the execution. For queries over continuous streams of data.

the system can continually adapt to changes in the data arrival rates, data

characteristics, and the availability of processing, storage, and communication

resources.

In addition to novel control structures. Telegraph also employs non-blocking,

symmetric query processing operators, such as XJoins [UFOO] and Ripple Joins

[HH99], which can cope with changing and unpredictable arrival of their input data.

The challenges being addressed in the Telegraph project include the development of

cluster-based and wide-area implementations of the processing engine, the design of

fault-tolerance mechanisms (particularly for long-running queries), support for

continuous queries over sensor data and for profile-based information dissemination,