Database Reference
In-Depth Information
followers, tweets), with a maximum rate limit of 350 requests per hour. The third
one, the streaming API, is the most suited access for data mining or analytic
research, allowing one to retrieve a 1% filter of all tweets that users are actually
carrying out, possibly using some filtering fields such as keywords, tags, users,
and geographic bounding box. Such a rate limit can be raised by asking Twitter
for a “gardenhose” access, in order to receive a steady stream of tweets, very
roughly 10% of all public statuses. Note that these proportions are subject to
unannounced adjustment as traffic volume varies.
Unlike Twitter, the Foursquare API allows one to view all friends of an
individual but does not allow one, for reasons of privacy, to “stalk” a specified
user. The only way to collect information about users' activity is to select a
set of venues in one or more specific regions, and download all the activities
(check-ins) performed in those locations, with a rate limit of 5,000 requests
per hour. Both Twitter and Foursquare have severe limitations to data retrieval,
enabling one to gather only a very partial subset of users' activities.
Among many other geo-social networks, Flickr poses the fewest limitations.
In such online photo management system, practically all the valuable meta-
data such as tags and geolocation can be accessed by API programs. Anyway,
some experiments carried out by the authors using the same query in different
moments lead to retrieve slightly different results, leaving some uncertainty on
the soundness of results. Applications can produce raw or derived data. Raw
data can be the coordinates (latitude and longitude) of the message generated
by the mobile device, and derived data can be the coordinates' bounding box,
the place type, the place name, or the street name. This information is produced
by the social network application using the coordinates passed by the mobile
device. The information that can be produced and retrieved changes depending
on the system, the device, the privacy settings, and so on.
16.4.4 Geographic Uncertainty of CGI
As we already pointed out, geo-social data can have several sources of uncer-
tainty. Uncertainty aspects should be taken into account when performing sta-
tistical analysis and when developing systems based on these kinds of data. In
this section we discuss some of the most important uncertainty aspects of CGI
data. Other aspects of uncertainty are covered in Chapter 5 .
Uncertainty about Precision
In this category we join issues related to data generation. The first source of
uncertainty is information granularity: each point of the trajectory can have a
different scale, sometimes coordinates of a specific place, sometimes a bounding
box area. The second one is related to devices. A third one is that the precision
can be modified by the social network system: in some applications (such as
Search WWH ::




Custom Search