add an example of duplciate

2_background.tex

@@ -110,8 +110,17 @@ it is impossible for contributors to always be aware of others' activities and g
 Therefore, 
 unintentional relations like dependence, conflict, and duplicate  
 may be found among pull-requests submitted by different contributors.
-% !!要不要加个例子
-\hl{needed a exmple dup?}
+Figure~\ref{fig:dup_example} shows an example of a pair of duplicate pull-requests.
+
+
+\begin{figure*}[h]
+	\centering
+	\includegraphics[width=0.9\textwidth]{resources/dup_example.pdf}
+	\caption{An example of a pair of duplicate pull-requests}
+   \label{fig:dup_example}
+\end{figure*}
+
+
 
 \subsection{Duplicates in community-based collaboration}
 Except from the pull-based development, 

4_rq3.tex

@@ -482,15 +482,13 @@ a pull-request has less likelihood to be preferred}.
 
 
 \subsection{Manual inspection.}
-
-% 周老师：
-% 关于coding（定性研究总结taxonomy那块，以及可能要加的contribution comment总结），这个open coding是有protocal的，需要遵循
-We manually explore the decision elaborations made by reviewers 
+\hl{turn}
+We manually inspect the decision elaborations made by reviewers 
 when they make a choice between duplicates 
 and build a taxonomy of the reasons why reviewers would prefer one pull-request to the other one.
 This work is conducted on 100 pairs  
 of duplicate pull-requests 
-randomly selected from \textit{DupPR}.
+which are randomly selected.
 \hl{This sample represents the overall collected duplicate pull-requests, with a confidence level of **\% and a margin of error of ***\%.}
 % 给出链接
 
@@ -566,20 +564,27 @@ the final taxonomy emerges after the refinement.
 
 \subsubsection{Taxonomy of preference}
 Figure~\ref{fig:taxo} shows the observed taxonomy. 
-Although we have attached the number of observed comments for each preference in the figure, 
-that numbers are just reported as a statistical observation and they cannot reflect the importance of each preference. 
 As said, 
-preferences are not mutually exclusive 
-and reviewers tend to assess crowd contributions based on multiple factors. 
-Therefore, 
-reviewers many only point out their preference for a specific factor   
-when they think a crowd contribution has a problem with that assessment factor. 
-% ！！ 有些可能连看都不看实现细节呢。一看重复了，直接关闭
-For example, 
-the number of observed comments of the preference \textit{Correct implementation} is relatively small, 
-however it is obviously the most fundamental requirement for crowd contributions. 
-
-% 这里还是说一个comment可能要有多个分类吧
+the taxonomy has two levels. 
+In the first level, 
+there are four categories, \ie \textit{technical assessment}, \textit{non-technical consideration}, \textit{no explanation}, and \textit{self-decision}. 
+In the second level, 
+\textit{technical assessment} and \textit{non-technical consideration} 
+includes several sub-categories respectively.
+In the following, 
+we discuss each of the level-1 categories and level-2 sub-categories.
+% Although we have attached the number of observed comments for each preference in the figure, 
+% that numbers are just reported as a statistical observation and they cannot reflect the importance of each preference. 
+% As said, 
+% preferences are not mutually exclusive 
+% and reviewers tend to assess crowd contributions based on multiple factors. 
+% Therefore, 
+% reviewers many only point out their preference for a specific factor   
+% when they think a crowd contribution has a problem with that assessment factor. 
+% % ！！ 有些可能连看都不看实现细节呢。一看重复了，直接关闭
+% For example, 
+% the number of observed comments of the preference \textit{Correct implementation} is relatively small, 
+% however it is obviously the most fundamental requirement for crowd contributions.