Database Reference
In-Depth Information
she sees an opened hatchway on the road. As the user is driving a car, he/she has no
possibility to search for a proper phone, dial it to report a problem. The user just presses
the
“
red
”
button in the corner of the application screen and says
“
I
'
ve seen an opened
hatchway
. Then the last minute of the video is automatically extracted and is uploaded
into the cloud. In this way the user can continue driving his car without the need for
parking to make a call or type text. Besides, the user does not break the law by making
a call while driving. The point of ergonomic usage is very important here. User
”
s voice
annotation and all the metadata like location and timestamp are also attached to the
video and loaded to the cloud. This information is processed by city management
services. After passing some automated
'
ltering, the data becomes available to a
special employee who decides that this information should be passed to the road
service brigade of the current district. It all takes a few minutes before service brigade
moves out to
x the problem. This feature can also be used to report about any city
problems we can think off, for example, rude or dangerous behavior of other drivers,
broken traf
c lights, sewer breakthrough, water spills and much more.
The pseudo code for the
“
driver-initiated alert
”
scenario can be found below:
As it was already mentioned, the main principle of data acquisition is crowd-
sourcing. Crowdsourcing seems to be a strong method of collecting data, but there are
concerns about manual data processing. Though a lot of dispatching work can be
automated, most of it has to be made manually and by professionals.
There are also a number of challenges that would arise in any crowdsourcing
solution:
Redundant data. An open hatchway can be possibly seen and reported by many
drivers. All their reports must be linked to one and only one collective report to
avoid redundancy. This can be done by showing all similar user reports to the
employee who approves the report. The similarity can be found by comparing time,
coordinates and key words that were detected in the users
'
voice annotations.
False data. Someone
s joke can become a waste of time for the service team. One
of the solutions of this problem can be in storing user
'
s credibility ratings in the
database. These ratings can be made by giving marks to loaded videos by personnel
who work with them. Requests from users, who had already produced false or
'
Search WWH ::
Custom Search