It is worth noting, that this is a critical step since this task is normally performed
manually. The state of the art of existing databases shows that there is no standard
corpus containing scientific articles classified according to the taxonomy of epidemiological studies. Scientific papers are labeled according to different predefined classes
according to the taxonomy of the epidemiological study. For that reason, in the data
collect process, we were aware that the quality of data plays a vital role in the training
data process and the calculation of the accuracy score of any machine learning classification algorithm. Based on the carefully selected data, the machine learning algorithms can learn the patterns and correlations in the data.
Table 1. Label encoding
Category
code
Category name
Definition
0
Analytic study
Raise etiological hypotheses by comparing the prevalence of
the event in exposed and unexposed subjects
1
Descriptive
study
Describe phenomena and their geographic distribution and
temporal
2
Meta-Analysis
study
Assess the effect of a treatment used on comparable
populations, by combining the results of multiple studies
3
Experimental
study
Intervene on the exhibition status of subjects. It can affect the
factor (s) of exposure, the time of exposure and the people
exposed
Fig. 1. The distribution of articles across the
different values of labels/article.
Fig. 2. Histogram representation of the
articles/label Percentage
352
M. Khadhraoui et al.
manually. The state of the art of existing databases shows that there is no standard
corpus containing scientific articles classified according to the taxonomy of epidemiological studies. Scientific papers are labeled according to different predefined classes
according to the taxonomy of the epidemiological study. For that reason, in the data
collect process, we were aware that the quality of data plays a vital role in the training
data process and the calculation of the accuracy score of any machine learning classification algorithm. Based on the carefully selected data, the machine learning algorithms can learn the patterns and correlations in the data.
Table 1. Label encoding
Category
code
Category name
Definition
0
Analytic study
Raise etiological hypotheses by comparing the prevalence of
the event in exposed and unexposed subjects
1
Descriptive
study
Describe phenomena and their geographic distribution and
temporal
2
Meta-Analysis
study
Assess the effect of a treatment used on comparable
populations, by combining the results of multiple studies
3
Experimental
study
Intervene on the exhibition status of subjects. It can affect the
factor (s) of exposure, the time of exposure and the people
exposed
Fig. 1. The distribution of articles across the
different values of labels/article.
Fig. 2. Histogram representation of the
articles/label Percentage
352
M. Khadhraoui et al.