fixed some pattern detection bugs

issuecomp-analysis.R

@@ -9,8 +9,11 @@ date_start <- as.Date("2014-01-01")
 date_end <- as.Date("2014-12-31")
 drange <- as.integer(date_end - date_start)
 drange <- date_start + days(0:drange)
-issues <- data.frame(date = drange)
 
+
+# MATCH TWEETS ------------------------------------------------------------
+
+issues <- data.frame(date = drange)
 issuelist <- xmlToList("issues.xml")
 issueheads <- names(issuelist)
 issues[issueheads] <- 0
@@ -28,44 +31,38 @@ for(d in 1:nrow(issues)) {
     curtext <- as.character(tweets_curday$text[t])
     curtext <- str_replace_all(curtext, "#", "")
     
+    # Now test each single issue (not tag!)
     for(i in 1:length(issuelist)) {
       curtags <- as.character(issuelist[[i]])
       curissue <- names(issuelist)[i]
-      curtags <- str_c("\\W", curtags, "\\W")
-      tags_found <- str_detect(curtext, sprintf("%s", curtags))
-      tags_found <- any(tags_found)
-      
-      ######
-      
-      # Test all tags in ONE issue
+
+      # Now test all tags of a single issue
       for(t in 1:length(curtags)) {
-        curtag <- curtags[t]
-        curchars <- nchar(curtag, type = "chars")
+        curtag <- str_c("\\W", curtags[t], "\\W")
+        curchars <- nchar(curtag, type = "chars") - 4
         
+        # Match current tweet with tag. If >= 5 letters allow 1 changed letter, if >=8 letters allow 2 (Levenshtein distance)
         tags_found <- smartPatternMatch(curtext, curtag, curchars)
-        
         if(tags_found == 1) {
-          cat("Text contains at least the tag:", curtag, "\n")
+          #cat("Matched", curtag, "with", curtext,"\n")
+          issues[d,curissue] <- issues[d,curissue] + 1
           break
         }
-      }
-      
-      
-      ######
-      
-      if(tags_found) {
-        #cat("Positive in", curissue,"from",as.character(drange[d]),"\n")
-        issues[d,curissue] <- issues[d,curissue] + 1
-      }
-      else {
-        #cat("Nothing found\n")
-      }
+        else {
+          #cat("Nothing found\n")
+        }
+      } # /for curtags
       
     } # /for issuelist
   } # /for tweets_curday
 } # /for drange
 
 
+
+# WEEKLY INTERVALS --------------------------------------------------------
+
+
+
 ## Do not use days but week intervals
 
 wrange <- (as.integer(date_end - date_start) / 7)

issuecomp-functions.R

@@ -28,13 +28,13 @@ convertLogical0 <- function(var) {
 
 smartPatternMatch <- function(string, pattern, chars) {
   if(chars < 5) {
-    found <- agrep(pattern, string, max.distance = list(all = 0), ignore.case = TRUE)
+    found <- agrep(pattern, string, max.distance = list(all = 0), ignore.case = TRUE, fixed = FALSE)
   }
-  if(chars > 7) {
-    found <- agrep(pattern, string, max.distance = list(all = 2), ignore.case = TRUE)
+  else if(chars > 7) {
+    found <- agrep(pattern, string, max.distance = list(all = 2), ignore.case = TRUE, fixed = FALSE)
   }
   else {
-    found <- agrep(pattern, string, max.distance = list(all = 1), ignore.case = TRUE)
+    found <- agrep(pattern, string, max.distance = list(all = 1), ignore.case = TRUE, fixed = FALSE)
   }
   found <- convertLogical0(found)
   return(found)

issues.xml

@@ -11,15 +11,80 @@
         <tag>kraftwerk</tag>
         <tag>strom</tag>
     </issue.green>
+    
+    <issue.ukraine>
+        <tag>ukraine</tag>
+        <tag>euromaidan</tag>
+        <tag>krim</tag>
+        <tag>putin</tag>
+        <tag>kiew</tag>
+    </issue.ukraine>
+    
+    <issue.labour>
+        <tag>arbeitsmarkt</tag>
+        <tag>mindestlohn</tag>
+        <tag>arbeitslosigkeit</tag>
+        <tag>hartz4</tag>
+        <tag>arbeitslos</tag>
+    </issue.labour>
+    
+    <issue.nsa>
+        <tag>nsa</tag>
+        <tag>snowden</tag>
+        <tag>bnd</tag>
+        <tag>gchq</tag>
+        <tag>überwachung</tag>
+    </issue.nsa>
+    
+    <issue.wm2014>
+        <tag>wm2014</tag>
+        <tag>weltmeister</tag>
+        <tag>meister</tag>
+        <tag>finale</tag>
+        <tag>halbfinale</tag>
+        <tag>viertelfinale</tag>
+        <tag>achtelfinale</tag>
+        <tag>brager</tag>
+        <tag>gerbra</tag>
+        <tag>argger</tag>
+        <tag>gerarg</tag>
+        <tag>wm</tag>
+        <tag>stadion</tag>
+    </issue.wm2014>
+    
+    <issue.middleeast>
+        <tag>israel</tag>
+        <tag>gaza</tag>
+        <tag>naher osten</tag>
+        <tag>nahen osten</tag>
+        <tag>nahost</tag>
+    </issue.middleeast>
+    
+    <issue.iraq>
+        <tag>irak</tag>
+        <tag>isis</tag>
+        <tag>is</tag>
+        <tag>kalifat</tag>
+    </issue.iraq>
+    
+    <issue.ebola>
+        <tag>ebola</tag>
+    </issue.ebola>
+    
+    <issue.edathy>
+        <tag>edathy</tag>
+        <tag>kinderpornographie</tag>
+        <tag>kipo</tag>
+        <tag>pädophil</tag>
+        <tag>pädophilie</tag>
+    </issue.edathy>
 
-    <issue.test>
-        <tag>ein langer ausdruck</tag>
-        <tag>binde-strich</tag>
-        <tag>fünfe</tag>
-        <tag>achtacht</tag>
-        <tag>fehlar</tag>
-        <tag>korrektur</tag>
-    </issue.test>
+    <issue.conservative>
+        <tag>christ</tag>
+        <tag>christlich</tag>
+        <tag>christen</tag>
+        <tag>inflation</tag>
+    </issue.conservative>
 
     <issue.control>
         <tag>pillepalle</tag>

issues.xml.test

@@ -0,0 +1,29 @@
+<issuelist>
+    
+    <issue.green>
+        <tag>umwelt</tag>
+        <tag>energie</tag>
+        <tag>energiewende</tag>
+        <tag>atomkraft</tag>
+        <tag>windkraft</tag>
+        <tag>wasserkraft</tag>
+        <tag>solarstrom</tag>
+        <tag>kraftwerk</tag>
+        <tag>strom</tag>
+    </issue.green>
+
+    <issue.test>
+        <tag>ein langer ausdruck</tag>
+        <tag>binde-strich</tag>
+        <tag>fünfe</tag>
+        <tag>achtacht</tag>
+        <tag>fehlar</tag>
+        <tag>korrektur</tag>
+    </issue.test>
+
+    <issue.control>
+        <tag>pillepalle</tag>
+        <tag>schundluder</tag>
+        <tag>whatthefuck</tag>
+    </issue.control>
+</issuelist>