Differences

This shows you the differences between two versions of the page.

--- resarch:nlpa:paper_1 [2014/09/18 18:10]
preethac
+++ resarch:nlpa:paper_1 [2014/10/02 14:43] (current)
preethac
@@ Line 3: / Line 3: @@
 \\ Martin P. Robillard
-[[http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CB4QFjAA&url=http%3A%2F%2Fwww.cs.mcgill.ca%2F~martin%2Fpapers%2Ficse2013.pdf&ei=ZVYbVKb_Gs7qoATytYG4Aw&usg=AFQjCNEM0UobXT16enNDJbSsrcAXjiMuiQ&bvm=bv.75097201,bs.1,d.b2U]]
+[[http://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&ved=0CB4QFjAA&url=http%3A%2F%2Fwww.cs.mcgill.ca%2F~martin%2Fpapers%2Ficse2013.pdf&ei=ZVYbVKb_Gs7qoATytYG4Aw&usg=AFQjCNEM0UobXT16enNDJbSsrcAXjiMuiQ&bvm=bv.75097201,bs.1,d.b2U|Link]]
+**Problem:**
+Determining the code elements salient to an informal developer communications, such as forum posts.
+**Importance/Applications of the technique:**
+\\ Does not require an index of code elements.
+\\ Discovers code elements based on salience.
+**Approach:**
+\\ 1. Developed a tool "Automatic Code Element Extractor (ACE)" which does the following steps:-
+  * Identify code like terms from a document using Island Parser
+  * Creates index of valid code elements
+  * Reparses documents to identify ambiguous terms and resolve them
+  * Outputs the code elements
+. Created a classifier of code element salience
+**Evaluation:**
+P= 0.92
+R= 0.90
+**Critique**
+  Strengths:
+  Discards closed world assumption(where an index is created from before)
+  Discards uniform importance assumption(previously each code element would have been considered equally important)
+  Weaknesses:
+  Unable to identify indirectly referenced code elements(eg: answers to questions in stack overflow is considered not salient, hence miss indirectly referenced code elements which might be salient)