Index: trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java
===================================================================
--- trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java	(revision 231)
+++ trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/WeblogParser.java	(revision 232)
@@ -10,7 +10,9 @@
 import java.util.Collection;
 import java.util.HashMap;
+import java.util.HashSet;
 import java.util.LinkedList;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 
 import de.ugoe.cs.eventbench.web.data.WebEvent;
@@ -69,4 +71,36 @@
 	/**
 	 * <p>
+	 * List that stores the users (identified through their cookie id) to each
+	 * sequence.
+	 * </p>
+	 */
+	private List<String> users;
+
+	/**
+	 * <p>
+	 * List that stores the frequent users (identified through their cookie id)
+	 * to each sequence.
+	 * </p>
+	 */
+	private List<String> frequentUsers;
+
+	/**
+	 * <p>
+	 * Sequences for all frequent users.
+	 * </p>
+	 */
+	private List<Collection<List<WebEvent>>> sequencesFrequentUsers;
+
+	/**
+	 * <p>
+	 * Threshold that defines how many sessions of a user are require to deem
+	 * the user frequent. Note, that only sessions whose lengths is in range if
+	 * {@link #minLength} and {@link #maxLength} are counted.
+	 * </p>
+	 */
+	private int frequentUsersThreshold = -1;
+
+	/**
+	 * <p>
 	 * Name and path of the robot filter.
 	 * </p>
@@ -164,4 +198,40 @@
 	public void setUrl(String url) {
 		this.url = url;
+	}
+
+	/**
+	 * <p>
+	 * Sets the threshold for frequent users.
+	 * </p>
+	 * 
+	 * @param threshold
+	 *            threshold value; if the value is &lt;1, the sessions of the
+	 *            frequent users will not be determined
+	 */
+	public void setFrequentUserThreshold(int threshold) {
+		this.frequentUsersThreshold = threshold;
+	}
+
+	/**
+	 * <p>
+	 * Returns the IDs of all frequent users.
+	 * </p>
+	 * 
+	 * @return IDs of the frequent users
+	 */
+	public List<String> getFrequentUsers() {
+		return frequentUsers;
+	}
+
+	/**
+	 * <p>
+	 * Returns the sequences of all frequent users.
+	 * </p>
+	 * </p>
+	 * 
+	 * @return list of the sequences of all frequent users
+	 */
+	public List<Collection<List<WebEvent>>> getFrequentUserSequences() {
+		return sequencesFrequentUsers;
 	}
 
@@ -192,4 +262,5 @@
 
 		sequences = new ArrayList<List<WebEvent>>();
+		users = new ArrayList<String>();
 
 		int lineCounter = 0;
@@ -216,5 +287,9 @@
 			if (values.length == 6) { // post vars found
 				for (String postVar : values[5].trim().split(" ")) {
-					postedVars.add(postVar);
+					// TODO manual filtering of bad variables, should be
+					// automated
+					if (!postVar.contains("and")) {
+						postedVars.add(postVar);
+					}
 				}
 			}
@@ -236,4 +311,5 @@
 						cookieSessionMap.put(cookie, sessionIds);
 						sequences.add(new LinkedList<WebEvent>());
+						users.add(cookie);
 					}
 					Integer lastSessionIndex = sessionIds
@@ -251,4 +327,5 @@
 						newSession.add(event);
 						sequences.add(newSession);
+						users.add(cookie);
 					} else {
 						lastSession.add(event);
@@ -260,15 +337,57 @@
 			}
 		}
+		Console.traceln("" + sequences.size() + " user sequences found");
 		pruneSequences();
-	}
-
-	/**
-	 * <p>
-	 * Prunes sequences shorter than {@link #minLength}.
+		Console.traceln("" + sequences.size()
+				+ " remaining after pruning of sequences shorter than "
+				+ minLength);
+		Set<String> uniqueUsers = new HashSet<String>(users);
+		Console.traceln("" + uniqueUsers.size() + " unique users");
+		if (frequentUsersThreshold > 0) {
+			generateFrequentUserSequences(uniqueUsers);
+		}
+	}
+
+	/**
+	 * <p>
+	 * Generates the frequent user sequences, according to the threshold
+	 * {@link #frequentUsersThreshold}.
+	 * </p>
+	 * 
+	 * @param uniqueUsers
+	 *            set with all user IDs
+	 */
+	private void generateFrequentUserSequences(Set<String> uniqueUsers) {
+		frequentUsers = new ArrayList<String>();
+		sequencesFrequentUsers = new ArrayList<Collection<List<WebEvent>>>();
+		for (String user : uniqueUsers) {
+			List<String> tmp = new ArrayList<String>();
+			tmp.add(user);
+			List<String> usersCopy = new LinkedList<String>(users);
+			usersCopy.retainAll(tmp);
+			int size = usersCopy.size();
+			if (size >= frequentUsersThreshold) {
+				frequentUsers.add(user);
+				Collection<List<WebEvent>> sequencesUser = new ArrayList<List<WebEvent>>();
+				for (int i = 0; i < sequences.size(); i++) {
+					if (users.get(i).equals(user)) {
+						sequencesUser.add(sequences.get(i));
+					}
+				}
+				sequencesFrequentUsers.add(sequencesUser);
+
+			}
+		}
+		Console.traceln("" + frequentUsers.size() + " users with more than "
+				+ frequentUsersThreshold + " sequences");
+	}
+
+	/**
+	 * <p>
+	 * Prunes sequences shorter than {@link #minLength} and longer than
+	 * {@link #maxLength}.
 	 * </p>
 	 */
 	private void pruneSequences() {
-		Console.traceln("" + sequences.size() + " user sequences found");
-		// prune sequences shorter than min-length and longer than maxLength
 		int i = 0;
 		while (i < sequences.size()) {
@@ -276,11 +395,10 @@
 					|| sequences.get(i).size() > maxLength) {
 				sequences.remove(i);
+				users.remove(i);
 			} else {
 				i++;
 			}
 		}
-		Console.traceln("" + sequences.size()
-				+ " remaining after pruning of sequences shorter than "
-				+ minLength);
+
 	}
 
@@ -338,5 +456,8 @@
 			for (String paramPair : paramPairs) {
 				String[] paramSplit = paramPair.split("=");
-				getVars.add(paramSplit[0]);
+				// TODO manual filtering of bad variables, should be automated
+				if (!paramSplit[0].contains("and")) {
+					getVars.add(paramSplit[0]);
+				}
 			}
 		}
Index: trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/commands/CMDloadSessionsFromClickstream.java
===================================================================
--- trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/commands/CMDloadSessionsFromClickstream.java	(revision 231)
+++ trunk/EventBenchConsole/src/de/ugoe/cs/eventbench/web/commands/CMDloadSessionsFromClickstream.java	(revision 232)
@@ -5,8 +5,10 @@
 import java.security.InvalidParameterException;
 import java.text.ParseException;
+import java.util.Collection;
 import java.util.List;
 
 import de.ugoe.cs.eventbench.data.GlobalDataContainer;
 import de.ugoe.cs.eventbench.web.WeblogParser;
+import de.ugoe.cs.eventbench.web.data.WebEvent;
 import de.ugoe.cs.util.console.Command;
 import de.ugoe.cs.util.console.Console;
@@ -38,15 +40,23 @@
 		int minLength = -1;
 		int maxLength = -1;
-		if( parameters.size()>=3 ) {
+		boolean generateFrequentUsers = false;
+		int frequentUserThreshold = 20;
+		if (parameters.size() >= 3) {
 			serverUrl = (String) parameters.get(2);
 		}
-		if (parameters.size() >= 5) {
+		if (parameters.size() >= 6) {
 			timeout = Integer.parseInt((String) parameters.get(3));
 			minLength = Integer.parseInt((String) parameters.get(4));
 			maxLength = Integer.parseInt((String) parameters.get(5));
 		}
+		if (parameters.size() >= 8) {
+			generateFrequentUsers = Boolean.parseBoolean((String) parameters
+					.get(6));
+			frequentUserThreshold = Integer
+					.parseInt((String) parameters.get(7));
+		}
 
 		WeblogParser parser = new WeblogParser();
-		if( serverUrl!=null ) {
+		if (serverUrl != null) {
 			parser.setUrl(serverUrl);
 		}
@@ -55,4 +65,7 @@
 			parser.setMinLength(minLength);
 			parser.setMaxLength(maxLength);
+		}
+		if (generateFrequentUsers) {
+			parser.setFrequentUserThreshold(frequentUserThreshold);
 		}
 		try {
@@ -71,4 +84,14 @@
 			Console.traceln("Old data \"" + sequencesName + "\" overwritten");
 		}
+		if (generateFrequentUsers) {
+			List<String> frequentUserIDs = parser.getFrequentUsers();
+			List<Collection<List<WebEvent>>> frequentUserSessions = parser
+					.getFrequentUserSequences();
+			for (int i = 0; i < frequentUserIDs.size(); i++) {
+				GlobalDataContainer.getInstance().addData(
+						sequencesName + "_" + frequentUserIDs.get(i),
+						frequentUserSessions.get(i));
+			}
+		}
 	}
 
@@ -80,5 +103,5 @@
 	@Override
 	public void help() {
-		Console.println("Usage: loadSessionsFromClickstream <filename> <sequencesName> {<serverUrl>} {<timeout> <minSessionLength> <maxSessionLength>}");
+		Console.println("Usage: loadSessionsFromClickstream <filename> <sequencesName> {<serverUrl>} {<timeout> <minSessionLength> <maxSessionLength>} {<generateFrequentUsers> <frequentUserThreshold>}");
 	}