Just pushing what I have

fras2560 · Feb 6, 2018 · 1099408 · 1099408
1 parent 1e94b35
commit 1099408
Show file tree

Hide file tree

Showing 7 changed files with 245 additions and 307 deletions.
diff --git a/src/index/IndexFiles.java b/src/index/IndexFiles.java
@@ -363,10 +363,10 @@ public static void main(String[] args) {
                  + " [-index INDEX_PATH] [-docs DOCS_PATH] [-logfile file] [-update]\n\n"
                  + "This indexes the documents in DOCS_PATH, creating a Lucene index"
                  + "in INDEX_PATH that can be searched with SearchFiles";
-    Path indexPath = Paths.get(System.getProperty("user.dir"), "resources", "index", "temp");
-    Path docsPath = Paths.get("/home", "d6fraser", "Documents", "Research", "Datasets", "NTCIR_TEMP");
-    Path logFile = Paths.get(System.getProperty("user.dir"), "resources", "logs", "NTCIRTemp.log");
-    boolean create = true;
+    Path indexPath = Paths.get(System.getProperty("user.dir"), "resources", "index", "full_arXiv", "firstAttempt");
+    Path docsPath = Paths.get(System.getProperty("user.dir"), "resources", "documents", "TODO");
+    Path logFile = Paths.get(System.getProperty("user.dir"), "resources", "logs", "NTCIRFull.log");
+    boolean create = false;
     for(int i=0;i<args.length;i++) {
       if ("-index".equals(args[i])) {
           indexPath = Paths.get(args[i+1]);

diff --git a/src/index/IndexThreadProducer.java b/src/index/IndexThreadProducer.java
diff --git a/src/programs/FindOptimalQueryWeight.java b/src/programs/FindOptimalQueryWeight.java
@@ -22,7 +22,7 @@ public static void main(String[] args) throws Exception {
             System.out.println(usage);  
             System.exit(0);
         }
-        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "arXiv", "current");
+        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "full_arXiv", "tunaIndex");
         Path queries = Paths.get(System.getProperty("user.dir"), "resources", "query", "NTCIR12-ArXiv.xml");
         Path results = Paths.get(System.getProperty("user.dir"), "resources", "results", "NTCIR12-ArXiv-Math.dat");
         String date = new SimpleDateFormat("dd-MM-yyyy:HH:mm").format(new Date());

diff --git a/src/programs/FindOptimalWeight.java b/src/programs/FindOptimalWeight.java
@@ -23,9 +23,9 @@ public static void main(String[] args) throws Exception {
             System.out.println(usage);  
             System.exit(0);
         }
-        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "ntcir-12-wikipedia", "current");
-        Path queries = Paths.get(System.getProperty("user.dir"), "resources", "query", "NTCIR12-MathWiki-main.xml");
-        Path results = Paths.get(System.getProperty("user.dir"), "resources", "results", "NTCIR12-MathWiki-main.dat");
+        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "full_arXiv", "tunaIndex");
+        Path queries = Paths.get(System.getProperty("user.dir"), "resources", "query", "NTCIR12-ArXiv.xml");
+        Path results = Paths.get(System.getProperty("user.dir"), "resources", "results", "NTCIR12-ArXiv-Math.dat");
         String date = new SimpleDateFormat("dd-MM-yyyy:HH:mm").format(new Date());
         Path output = Paths.get(System.getProperty("user.dir"), "resources", "output", "arXiv", "weights", date + ".csv");
         Path logFile = Paths.get(System.getProperty("user.dir"),

diff --git a/src/programs/NTCIR_TEST.java b/src/programs/NTCIR_TEST.java
@@ -42,24 +42,24 @@ public static void main(String[] args) throws Exception {
             System.out.println(usage);  
             System.exit(0);
         }
-        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "ntcir-12-wikipedia", "current");
-        Path queries = Paths.get(System.getProperty("user.dir"), "resources", "query", "NTCIR12-MathWiki-formula.xml");
-        Path results = Paths.get(System.getProperty("user.dir"), "resources", "results", "NTCIR12-MathWiki-formula.dat");
+        Path index = Paths.get(System.getProperty("user.dir"), "resources", "index", "full_arXiv", "firstAttempt");
+        Path queries = Paths.get(System.getProperty("user.dir"), "resources", "query", "NTCIR12-ArXiv.xml");
+        Path results = Paths.get(System.getProperty("user.dir"), "resources", "results", "NTCIR12-ArXiv-Math.dat");
         String date = new SimpleDateFormat("dd-MM-yyyy:HH:mm").format(new Date());
         Path logFile = Paths.get(System.getProperty("user.dir"),
                                  "resources",
                                  "output",
-                                 "ntcir12-wikipedia",
+                                 "arXiv",
                                  date + ".log");
         Path queryOutput = Paths.get(System.getProperty("user.dir"),
                                      "resources",
                                      "output",
-                                     "ntcir12-wikipedia",
+                                     "arXiv",
                                      date + "-queries.txt");
         Path resultOutput = Paths.get(System.getProperty("user.dir"),
                                       "resources",
                                       "output",
-                                      "ntcir12-wikipedia",
+                                      "arXiv",
                                       date + "-results.txt");
         for(int i = 0;i < args.length;i++) {
           if ("-index".equals(args[i])) {
@@ -106,8 +106,7 @@ public static void main(String[] args) throws Exception {
             config.setQueryType(ConvertConfig.DIFFERENT_WEIGHTED_QUERY);
             // do the actual searching
             Search searcher = new Search(index, config);
-            searcher.setAlpha(0.05f);
-            searcher.setBeta(0.34f);
+            searcher.setAlpha(0.60f);
             searcher.ntcirTest(queries, results, resultsWriter);
             searcher.recordQueries(queries, queryWriter, 100);;
             // close the files

diff --git a/src/tangent/convert.py b/src/tangent/convert.py
@@ -122,7 +122,7 @@ def check_node(node):
         # does it make sense to keep pairs of symbols with no path
         # if one of those symbols is a wildcard
         if len(node) == 3:
-            # if one is a wildcard the dont want to keep it
+            # if one is a wildcard then dont want to keep it
             check = not(check_wildcard(node[0]) or check_wildcard(node[1]))
         else:
             # then both need to be a wildcard