IEEE Geoscience and Remote Sensing Magazine - December 2022 - 152

same conditions apply to many other social media platforms,
such as Instagram. In contrast, text messages posted
on Twitter are, by default, open, unless users restrict
access; hence, they can be crawled extensively. Twitter also
provides an official application programming interface
(API) that permits massive download of tweets. Another
issue is the geographic location contained in the data as it
is the key to linking the data to geographic applications.
For this reason, this article focuses on geographically harvestable
social media data. We require that photos and
text messages are either geotagged, or it is possible to infer
their geographic locations with reasonable precision (in
a building block level). Table 1 summarizes the available
harvestable social media based on these requirements.
The list is, of course, not exhaustive. It is intended to give
readers the most common sources of social media data
for geoinformation retrieval. Most freely available social
media data are under the creative commons license, often
the CC-BY license. The license of their images vary from
platform to platform. We intend to give readers guidance
on what type of social media sites and what data licenses
are suitable for scientific research. For example, although
CC-BY allows free use of the data, we note that a CC-BY license
does not permit sublicensing. This means that posting
or publishing such data on certain websites or journals
requires special attention such as a sublicense or even the
transfer of ownership, which is sometimes required by the
journal publisher.
Our extensive research shows that among all these options,
Twitter and Flickr are the most generous in terms of
total data volume and data harvestability because of their
license terms and functionality of the APIs. Volunteered
geographic information such as OSM is also accessible in
large volumes. However, they are not strictly social media
data and hence not the focus of this article. We focus on
tweets and Flickr images in the following.
TWITTER
Twitter offers several API packages with different pricing
levels. In this article, we focus solely on the freely available
Twitter API, which allows the user to stream approximately
1% of the daily Twitter stream of an area of interest (AOI).
Further, the API offers several techniques to query tweets.
If a researcher is interested in a special hashtag or keyword
like #COVID19, it is possible to submit such a request to the
API and receive tweets with matching contents as a result.
(Note that hashtags are keywords used to tag a tweet with
a certain topic, event, celebrity, and so on and consist of a
word or word sequence without trailing spaces and a leading
" # " ; for example, #RemoteSensingIsGreat.) An AOI can
be specified via a bounding box of a city, region, or country.
It is also possible to receive the 1% stream without any
keyword or hashtag filtering. A further method used to receive
tweets is to download a user's timeline or " hydrate "
tweet IDs, i.e., retrieve tweet content based on IDs. This is
particularly relevant in research as datasets are usually only
shared in this format due to Twitter's license terms prohibiting
sharing of complete tweet data.
FLICKR
Flickr, a social media image platform, offers a powerful free
API allowing arbitrary spatiotemporal queries. It can thus
be used to create comprehensive worldwide datasets and is
therefore often the first choice for image data. Flickr images
have been used in various studies across all disciplines, e.g., in
[16], [17], [18], and [19]. In contrast, Facebook used to have
an open API in its early days but changed toward a more privacy-preserving
one as it became more popular and allowed
users more fine-grained options on the visibility of posts.
OTHERS
Instagram closed their API in April 2018 and redesigned it to
be used by businesses and external apps to curate the profile
TABLE 1. AVAILABLE SOCIAL MEDIA DATA.
PLATFORM
TYPE
Google image search
Flickr.com
Unsplash.com
Pexels.com
Magdeleine.co
Twitter.com
Instagram.com
OpenStreetMap.org
Mapillary.com
Google Places
Foursquare
Geograph.org.uk
Image
Image
Image
Image
Image
Text and image
Text and image
vector and text
Image
Text and image
Text and image
Text and image
DESCRIPTION
Search engine
LICENSE
Photographer website
Photographer website
Photographer website
Photographer website
Social sharing website
Social sharing website
Online map service
Online map service
POI service
POI service
Online map service
CC: creative commons; DB: database; POI: point of interest.
152
Partly CC or free to use
Partly CC
All CC or free to use
All CC or free to use
All CC0 or CC-BY
Posts public by default
Posts public by default
Open Data Commons
Open Database
CC BY-SA
Proprietary
Proprietary
CC BY-SA
CRAWLABILITY
Third-party crawlers
Official API
Official API
Official API
Official API
Official API 1% of all
Third-party crawlers
Official API and DB dumps
Official API with limited quota
Official API
Official API
Official API and DB dumps
GEOTAG
Partly
Partly
Mostly
Mostly
Few
Partly
Partly
All
All
All
All
All
IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE DECEMBER 2022
http://www.Flickr.com http://www.Unsplash.com http://www.Pexels.com http://www.Magdeleine.co http://www.Twitter.com http://www.Instagram.com http://www.OpenStreetMap.org http://www.Mapillary.com http://Geograph.org.uk
IEEE Geoscience and Remote Sensing Magazine - December 2022

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - December 2022
