better pattern detection

issuecomp-analysis.R

@@ -25,7 +25,7 @@ for(d in 1:nrow(issues)) {
   
   for(t in 1:nrow(tweets_curday)){
     # Select tweet's text, make it lowercase and remove hashtag indicators (#)
-    curtext <- tolower(as.character(tweets_curday$text[t]))
+    curtext <- as.character(tweets_curday$text[t])
     curtext <- str_replace_all(curtext, "#", "")
     
     for(i in 1:length(issuelist)) {
@@ -35,6 +35,24 @@ for(d in 1:nrow(issues)) {
       tags_found <- str_detect(curtext, sprintf("%s", curtags))
       tags_found <- any(tags_found)
       
+      ######
+      
+      # Test all tags in ONE issue
+      for(t in 1:length(curtags)) {
+        curtag <- curtags[t]
+        curchars <- nchar(curtag, type = "chars")
+        
+        tags_found <- smartPatternMatch(curtext, curtag, curchars)
+        
+        if(tags_found == 1) {
+          cat("Text contains at least the tag:", curtag, "\n")
+          break
+        }
+      }
+      
+      
+      ######
+      
       if(tags_found) {
         #cat("Positive in", curissue,"from",as.character(drange[d]),"\n")
         issues[d,curissue] <- issues[d,curissue] + 1

issuecomp-functions.R

@@ -19,6 +19,27 @@ insertRow <- function(existingDF, newrow, r) {
   return(existingDF)  
 }
 
+convertLogical0 <- function(var) {
+  if(is.integer(var) && length(var) == 0) {
+    var <- 0
+  }
+  return(var)
+}
+
+smartPatternMatch <- function(string, pattern, chars) {
+  if(chars < 5) {
+    found <- agrep(pattern, string, max.distance = list(all = 0), ignore.case = TRUE)
+  }
+  if(chars > 7) {
+    found <- agrep(pattern, string, max.distance = list(all = 2), ignore.case = TRUE)
+  }
+  else {
+    found <- agrep(pattern, string, max.distance = list(all = 1), ignore.case = TRUE)
+  }
+  found <- convertLogical0(found)
+  return(found)
+}
+
 ## ERROR HANDLING
 
 # Check for empty API returns (0 or 1 or 2)