| Abstrak/Abstract |
Predicting the location is a critical issue for
conducting any analyses. Location analysis is one of analysis
which processes Geographic Information (GI). Example in
social media, users could include their posts or tweets with
latitude and longitude using the Global Position System (GPS)
in their devices. Unfortunately, not all users or content creators
provide their latitude and longitude. As a result, we investigate
the literature in journals and conferences while being aware of
the best methodology and algorithm employed by current
researchers. Based on the predetermined inclusion and
exclusion criteria, thirty-six location prediction research
articles published between 2013 and 2022 were kept and
selected to be investigated further. A strategy for identifying,
assessing, and interpreting all study materials accessible with
the goal of responding to specific questions is known as a
Systematic Literature Review (SLR). The SLR result show that
the topic of location prediction is still high in publication level.
User text post is the most popular approach used by
researchers, followed by user information and user trajectory.
Almost all reviewed paper use Twitter to get the social media
data. It mostly caused because of the availability throughout
the Application Programming Interface (API), tool library,
and datasets on the internet. Out of the twenty-six ways, four
of the most popular algorithms for predicting the location were
found (RNN, LSTM, CNN and Random Forest). To raise the
performance of machine learning classifiers for position
prediction, some researchers combining any machine learning
algorithms. Therefore, the researcher suggests to using any
combined algorithm, approach, and source of data in social
media to get deeper optimization of location prediction. |
