summaryrefslogtreecommitdiff
path: root/aied2018/introduction.tex
diff options
context:
space:
mode:
Diffstat (limited to 'aied2018/introduction.tex')
-rw-r--r--aied2018/introduction.tex2
1 files changed, 1 insertions, 1 deletions
diff --git a/aied2018/introduction.tex b/aied2018/introduction.tex
index ab69147..ebe86f5 100644
--- a/aied2018/introduction.tex
+++ b/aied2018/introduction.tex
@@ -16,4 +16,4 @@ Jin et al. use linkage graphs to represent dependencies between individual progr
When comparing programs we must account for superficial differences such as different naming schemes. Rivers et al. canonicalize student programs using equivalency-preserving transformations (renaming variables, reordering binary expressions, inlining functions and so on)~\cite{rivers2015data-driven}. We use their canonicalization as a preprocessing step before extracting patterns.
-The next section describes TREs and AST patterns, and gives a brief quantitative evaluation to show that patterns can be used to discriminate between correct and incorrect programs. Section~\ref{sec:rules} describes our modified version of the CN2 rule-learning algorithm, and analyzes student solutions to two programming exercises in terms of such rules. The last section outlines the directions for our future work.
+The next section describes TREs and AST patterns, and gives a brief evaluation to show that patterns can discriminate between correct and incorrect programs. Section~\ref{sec:rules} describes our modified version of the CN2 rule-learning algorithm, and analyzes student solutions to two programming exercises in terms of such rules. The last section outlines the directions for our future work.