better matching, less bugs, more fun

.gitignore

@@ -1,3 +1,4 @@
 tweets_complete.csv
 current.txt
 .RData
+matched-ids

issuecomp-analysis.R

@@ -4,6 +4,8 @@ require(ggplot2)
 require(reshape2)
 require(stringr)
 
+source("issuecomp-functions.R")
+
 # Create date range
 date_start <- as.Date("2014-01-01")
 date_end <- as.Date("2014-12-31")
@@ -13,6 +15,10 @@ drange <- date_start + days(0:drange)
 
 # MATCH TWEETS ------------------------------------------------------------
 
+id_folder <- "matched-ids"
+unlink(id_folder, recursive = TRUE)
+dir.create(id_folder)
+
 issues <- data.frame(date = drange)
 issuelist <- xmlToList("issues.xml")
 issueheads <- names(issuelist)
@@ -30,22 +36,34 @@ for(d in 1:nrow(issues)) {
     # Select tweet's text, make it lowercase and remove hashtag indicators (#)
     curtext <- as.character(tweets_curday$text[t])
     curtext <- str_replace_all(curtext, "#", "")
+    curtext <- str_replace_all(curtext, "$", " ")
+    curtext <- str_replace_all(curtext, "http://.+?\\s", "URL ")
+    curid <- as.character(tweets_curday$id_str[t])
     
     # Now test each single issue (not tag!)
     for(i in 1:length(issuelist)) {
       curtags <- as.character(issuelist[[i]])
       curissue <- names(issuelist)[i]
+      curfile <- str_c(id_folder,"/",curissue,".csv")
 
       # Now test all tags of a single issue
-      for(t in 1:length(curtags)) {
+      for(s in 1:length(curtags)) {
-        curtag <- str_c("\\W", curtags[t], "\\W")
+        curtag <- curtags[s]
-        curchars <- nchar(curtag, type = "chars") - 4
+        curchars <- nchar(curtag, type = "chars")
         
+        # Check if tag is an acronym. If so, ignore.case will be deactivated in smartPatternMatch
+        if(curchars <= 4) {
+          curacro <- checkAcronym(string = curtag, chars = curchars)
+        } else {
+          curacro <- FALSE
+        }
+                 
         # 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)
+        tags_found <- smartPatternMatch(curtext, curtag, curchars, curacro)
         if(tags_found == 1) {
           #cat("Matched", curtag, "with", curtext,"\n")
           issues[d,curissue] <- issues[d,curissue] + 1
+          write(str_c(curdate,";\"",curid,"\";",curtag), curfile, append = TRUE)
           break
         }
         else {
@@ -59,44 +77,6 @@ for(d in 1:nrow(issues)) {
 
 
 
-# WEEKLY INTERVALS --------------------------------------------------------
-
-
-
-## Do not use days but week intervals
-
-wrange <- (as.integer(date_end - date_start) / 7)
-wrange <- floor(wrange) - 1
-wrange <- date_start + weeks(0:wrange)
-issues_week <- data.frame(week = wrange)
-issues_week[issueheads] <- 0
-
-
-for(w in 1:nrow(issues_week)) {
-  curweek <- issues_week$week[w]
-  currange <- curweek + days(0:6)
-  
-  day <- 1
-  
-  
-  for(d in 1:nrow(issues)) {
-    curday <- issues$date[d]
-    
-    if(curweek == curday) {
-      for(c in 2:ncol(issues)) {
-        curissue <- names(issues)[c]
-        d2 <- d + 6
-        curvalue <- sum(issues[d:d2,curissue])
-        issues_week[w, curissue] <- curvalue
-        
-      } # /for issues columns
-    } # /if day matches first day of week
-    
-  } # /for issues rows
-} # /for issues_week
-
-
-
 
 # VISUALS -----------------------------------------------------------------
 
@@ -105,10 +85,6 @@ issues_melt <- melt(issues,id="date")
 ggplot(issues_melt,aes(x=date,y=value,colour=variable,group=variable)) + geom_line(size=1)
 ggplot(issues_melt,aes(x=date,y=value,colour=variable,group=variable)) + geom_smooth(size=1,method="loess",formula = y ~ x, se=FALSE)
 
-# Level: weeks
-issues_week_melt <- melt(issues_week,id="week")
-ggplot(issues_week_melt,aes(x=week,y=value,colour=variable,group=variable)) + geom_line(size=1)
-ggplot(issues_week_melt,aes(x=week,y=value,colour=variable,group=variable)) + geom_smooth(size=1,method="loess",formula = y ~ x, se=FALSE)
 
 
 # POSSIBLY USEFUL CODE ----------------------------------------------------

issuecomp-functions.R

@@ -26,20 +26,47 @@ convertLogical0 <- function(var) {
   return(var)
 }
 
-smartPatternMatch <- function(string, pattern, chars) {
+smartPatternMatch <- function(string, pattern, chars, acronym) {
-  if(chars < 5) {
+  pattern <- str_c("\\b", pattern, "\\b")
-    found <- agrep(pattern, string, max.distance = list(all = 0), ignore.case = TRUE, fixed = FALSE)
+  
+  if(chars <= 4) {
+    found <- agrep(pattern, string, max.distance = list(all = 0), ignore.case = !acronym, fixed = FALSE)
   }
-  else if(chars > 7) {
+  else if(chars >= 8) {
-    found <- agrep(pattern, string, max.distance = list(all = 2), ignore.case = TRUE, fixed = FALSE)
+    found <- agrep(pattern, string, max.distance = list(all = 2), ignore.case = !acronym, fixed = FALSE)
   }
   else {
-    found <- agrep(pattern, string, max.distance = list(all = 1), ignore.case = TRUE, fixed = FALSE)
+    found <- agrep(pattern, string, max.distance = list(all = 1), ignore.case = !acronym, fixed = FALSE)
   }
   found <- convertLogical0(found)
   return(found)
 }
 
+viewMatchingTweets <- function(date, issue, folder) {
+  file <- str_c(folder,"/",issue,".csv")
+  df <- read.csv(file, sep = ";", colClasses="character", header = FALSE)
+  for(r in 1:nrow(df)) {
+    curdate <- as.character(df[r,1])
+    if(curdate == date) {
+      curid <- as.character(df[r,2])
+      curtag <- as.character(df[r,3])
+      cat(tweets$text[tweets$id_str == curid]," - ",curtag,"\n")
+    }
+  }
+}
+
+
+checkAcronym <- function(string, chars) {
+  curtag_up <- str_replace_all(string = curtag, pattern = "[[:lower:]]", replacement = "")
+  curchars_up <- nchar(curtag_up, type = "chars")
+  if(curchars_up == curchars) {
+    return(TRUE)
+  }
+  else {
+    return(FALSE)
+  }
+}
+
 ## ERROR HANDLING
 
 # Check for empty API returns (0 or 1 or 2)

issues.xml

@@ -62,9 +62,9 @@
     
     <issue.iraq>
         <tag>irak</tag>
-        <tag>isis</tag>
+        <tag>ISIS</tag>
-        <tag>is</tag>
+        <tag>IS</tag>
-        <tag>kalifat</tag>
+        <tag>Kalifat</tag>
     </issue.iraq>
     
     <issue.ebola>