Text Mining with RapidMiner

Ertek, G., Tapucu, D., and Arın, I., 2013. Text Mining with RapidMiner. In: Markus
Hofmann, Ralf Klinkenberg (Eds.) RapidMiner: Data Mining Use Cases and Business
Analytics Applications. Chapman & Hall/CRC Data Mining and Knowledge Discovery
Series. Chapman and Hall/CRC.
Note: This is the final draft version of this paper. Please cite this paper (or this final
draft) as above. You can download this final draft from the following websites:
http://research.sabanciuniv.edu
http://ertekprojects.com/gurdal-ertek-publications/
TEXT MINING WITH RAPIDMINER
G. Ertek, D. Tapucu, and I. Arın
Sabancı University, Istanbul, Turkey
The goal of this chapter is to introduce the text mining capabilities of RAPIDMINER through a
use case. The use case involves mining reviews for hotels at TripAdvisor.com, a popular web
portal. We will be demonstrating basic text mining in RAPIDMINER using the text mining
extension. We will present two different RAPIDMINER processes, namely Process01
andProcess02, which respectively describe how text mining can be combined with association
mining and cluster modeling. While it is possible to construct each of these processes from
scratch by inserting the appropriate operators into the process view, we will instead import these
two processes readily from existing model files. Throughout the chapter, we will at times
deliberately instruct the reader to take erroneous steps that result in undesired outcomes. We
believe that this is a very realistic way of learning to use RAPIDMINER, since in practice, the
modeling process frequently involves such steps that are later corrected.

USE CASES WITH RAPIDMINER
Working Title
Dr. Markus Hofmann
Institute of Technology Blanchardstown, Ireland
Ralf Klinkenberg
Rapid-i
A JOHN WILEY & SONS, INC., PUBLICATION

Copyright c 2012 by John Wiley & Sons, Inc. All rights reserved.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey.
Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form
or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as
permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior
written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to
the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400,
fax (978) 646-8600, or on the web at www.copyright.com. Requests to the Publisher for permission should
be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ
07030, (201) 748-6011, fax (201) 748-6008.
Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in
preparing this book, they make no representations or warranties with respect to the accuracy or
completeness of the contents of this book and specifically disclaim any implied warranties of
merchantability or fitness for a particular purpose. No warranty may be created ore extended by sales
representatives or written sales materials. The advice and strategies contained herin may not be
suitable for your situation. You should consult with a professional where appropriate. Neither the
publisher nor author shall be liable for any loss of profit or any other commercial damages, including
but not limited to special, incidental, consequential, or other damages.
For general information on our other products and services please contact our Customer Care
Department with the U.S. at 877-762-2974, outside the U.S. at 317-572-3993 or fax 317-572-4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print,
however, may not be available in electronic format.
Library of Congress Cataloging-in-Publication Data:
Use Cases with RapidMiner / Markus Hofmann. . . [et al.].
Printed in the United States of America.
10 9 8 7 6 5 4 3 2 1

CONTRIBUTORS
GURDAL ERTEK, Sabancı University, Istanbul, Turkey
DILEK TAPUCU, Sabancı University, Istanbul, Turkey
INANC ARIN, Sabancı University, Istanbul, Turkey
v

CONTENTS IN BRIEF
1 Text Mining with RapidMiner 1
vii

CONTENTS
List of Figures xi
List of Tables xiii
1 Text Mining with RapidMiner 1
1.1 Introduction 1
1.1.1 Text Mining 2
1.1.2 Data Description 2
1.1.3 Running RAPIDMINER 2
1.1.4 RapidMiner Text Processing Extension Package 2
1.1.5 Installing Text Mining Extensions 4
1.2 Association Mining of Text Document Collection (Process01) 7
1.2.1 Importing Process01 7
1.2.2 Operators in Process01 9
1.2.3 Saving Process01 13
1.3 Clustering Text Documents (Process02) 15
1.3.1 Importing Process02 15
1.3.2 Operators in Process02 15
1.3.3 Saving Process02 18
1.4 Running Process01 and Analyzing the Results 21
1.4.1 Running Process01 21
1.4.2 Empty Results for Process01 23
ix

x CONTENTS
1.4.3 Specifying the Source Data for Process01 23
1.4.4 Re-Running Process01 27
1.4.5 Process01 Results 28
1.4.6 Saving Process01 Results 35
1.5 Running Process02 and Analyzing the Results 38
1.5.1 Running Process02 38
1.5.2 Specifying the Source Data for Process02 39
1.5.3 Process02 Results 40
1.6 Conclusions 45
Glossary 47

LIST OF FIGURES
1.1 The TripAdvisor data set 3
1.2 Running RAPIDMINER as administrator 3
1.3 RAPIDMINER splash screen when no extension packages are installed 4
1.4 Managing RAPIDMINER extension packages 4
1.5 Dialog box stating that no extension packages are yet installed 5
1.6 Updating RAPIDMINER 5
1.7 Installing extension packages for text mining 5
1.8 RAPIDMINER splash screen with the three extension packages installed
for text mining 6
1.9 Importing an existing process 7
1.10 Selecting the process to import 8
1.11 Operators for Process01 and the Parameters for Process
Documents from Files operator 8
1.12 Operators within the Process Documents from Files nested operator 10
1.13 Parameters for the operators within the Process Documents from
Files operator 11
xi

xii LIST OF FIGURES
1.14 Parameters for the operators in Process01 12
1.15 Configuring LocalRepository 13
1.16 Specifying the Root directory for LocalRepository 14
1.17 Storing Process01 in LocalRepository 14
1.18 Operators in Process02 and the Parameters for Process Documents
from Files operator 15
1.19 Operators within the Process Documents from Files nested operator
and the Parameters for the Generate n-Grams (Terms) operator 16
1.20 Parameters for the operators in Process02 16
1.21 Saved processes within LocalRepository 19
1.22 Renaming a process 19
1.23 Renaming Process01 correctly 20
1.24 Corrected names for Process01 and Process02 in LocalRepository
and the corresponding files 20
1.25 Opening and running Process01 21
1.26 Dialog box alerting the switch to the result perspective 21
1.27 Result Overview for Process01 results 22
1.28 Specifying the data source text directories for Process01 24
1.29 Specifying the text directories 24
1.30 Specifying the text directories 25
1.31 Specified text directories 26
1.32 Running Process01 again 27
1.33 Running of Process01 27
1.35 WordList generated by Process01 28
1.36 Meta Data View for the ExampleSet generated by Process01 29
1.37 Data View for the ExampleSet generated by Process01 29
1.38 Table View for the AssociationRules generated by Process01 30
1.39 Graph View for the AssociationRules generated by Process01 31
1.40 Graph View for the AssociationRules, without the node labels 31
1.41 Filtering rules in the Graph View for the AssociationRules 32

LIST OF FIGURES xiii
1.42 Document Occurences of the words in the WordList 33
1.43 Saving the WordList for Process01 34
1.44 Saving the ExampleSet for Process01 34
1.45 Saving the AssociationRules incorrectly, without selecting
LocalRepository or another directory 35
1.46 Saving the AssociationRules correctly, selecting the directory to be
saved into 36
1.47 Exporting the graph visualization as an image file 37
1.48 Specifying the directory to be saved into, the file name, and the file type 37
1.49 Switching to the design view 38
1.50 Opening and running Process02 38
1.51 Error message due to not specifying the source data for Process02 39
1.52 Specifying the text directories for the source data of Process02 39
1.54 Meta Data View for the ExampleSet generated by Process02,
including the n-Grams 41
1.55 Data View for the ExampleSet generated by Process02, and the
relative occurence frequency of the word absolut in hotel 73943.txt 42
1.56 Text View for the Cluster Model generated by Process02, displaying
the number of examples in each cluster 42
1.57 Folder View for the Cluster Model generated by Process02,
displaying the examples in each cluster 43
1.58 Centroid Table for the Cluster Model generated by Process02,
displaying the average frequency of each word in each cluster 44
1.59 Final view of LocalRepository with the processes and their results saved 44

CHAPTER 1
TEXT MINING WITH RAPIDMINER
Sabancı University, Istanbul, Turkey
1.1 INTRODUCTION
The goal of this chapter is to introduce the text mining capabilities of RAPIDMINER through a
use case. The use case involves mining reviews for hotels at TripAdvisor.com, a popular web
portal. We will be demonstrating basic text mining in RAPIDMINER using the text mining
extension. We will present two different RAPIDMINER processes, namely Process01 and
Process02, which respectively describe how text mining can be combined with association
mining and cluster modeling. While it is possible to construct each of these processes from
scratch by inserting the appropriate operators into the process view, we will instead import
these two processes readily from existing model ﬁles.
Throughout the chapter, we will at times deliberately instruct the reader to take erroneous
steps that result in undesired outcomes. We believe that this is a very realistic way of learning
to use RAPIDMINER, since in practice, the modeling process frequently involves such steps
that are later corrected.
Use Cases with RapidMiner, First Edition. By Hofmann, Klinkenberg
Copyright c 2012 John Wiley & Sons, Inc.
1

2 TEXT MINING WITH RAPIDMINER
1.1.1 Text Mining
Text mining (also referred to as text data mining or knowledge discovery from textual
databases), refers to the process of discovering interesting and non-trivial knowledge from
text documents. The common practice in text mining is the analysis of the information
extracted through text processing to form new facts and new hypotheses, that can be
explored further with other data mining algorithms. Text mining applications typically deal
with large and complex data sets of textual documents that contain significant amount of
irrelevant and noisy information. Feature selection aims to remove this irrelevant and noisy
information by focusing only on relevant and informative data for use in text mining. Some
of the topics within text mining include feature extraction, text categorization, clustering,
trends analysis, association mining and visualization.
1.1.2 Data Description
The files required for this chapter, including the data and pre-built processes reside within
a folder titled LocalRepository. The data used in this chapter comes from TripAd-
visor.com, a popular web portal in the hospitality industry, and is shown in Figure 1.1.
This publicly available data set contains the reviews and ratings (1 through 5) of clients
or customers for 1850 hotels. The original data was extracted by The Database and Sys-
tems Information Laboratory at the University of Illinois at Urbana-Champaign,and is
available under http://sifaka.cs.uiuc.edu/~wang296/Data/. There are 1850 text
documents in the original data set,corresponding to the reviews of 1850 hotels. Each
document contains all the reviews for that hotel. While it is possible to run text mining
processes with the original data, we will be using a subset of the data containing only
the first 100 text documents. The data set used in this chapter may be downloaded from
http://people.sabanciuniv.edu/ertekg/papers/supp/09.zip or the short web
link http://bit.ly/Kzdn5x, as well as http://research.sabanciuniv.edu .
1.1.3 Running RAPIDMINER
When running RAPIDMINER, it is strongly recommended to right click the mouse button
on the start menu and Run as administrator, as shown in Figure 1.2, rather than simply
clicking or double clicking the executable or shortcut icon. By running as administrator, we
are granted the permissions to install extension packages. Running the software without
the administrator rights may cause errors when trying to install updates, including the
extension packages. Figure 1.3 shows the splash screen displayed when RAPIDMINER is
run for the first time or when it is run without any extensions installed. In Figure 1.3,
the highlighted region is currently blank but it will later contain the icons for the installed
extensions. When RAPIDMINER is loaded, the user starts with the welcome perspective (a
perspective is a particular combination of views), as shown in Figure 1.4.
1.1.4 RapidMiner Text Processing Extension Package
RAPIDMINER is the most popular open source software in the world for data mining,
and strongly supports text mining and other data mining techniques that are applied in
combination with text mining. The power and flexibility of RAPIDMINER is due to the
GUI-based IDE (integrated development environment) it provides for rapid prototyping

INTRODUCTION 3
Figure 1.1 The TripAdvisor data set
Figure 1.2 Running RAPIDMINER as administrator
and development of data mining models, as well as its strong support for scripting based
on XML (extensible mark-up language). The visual modeling in the RAPIDMINER IDE
is based on the deﬁning of the data mining process in terms of operators and the ﬂow of

Figure 1.3 RAPIDMINER splash screen when no extension packages are installed
process through these operators. Users specify the expected inputs, the delivered outputs,
the mandatory and optional parameters, and the core functionalities of the operators, and the
complete process is automatically executed by RAPIDMINER. Many packages are available
for RAPIDMINER, such as text processing, Weka extension, parallel processing, web mining,
reporting extension, series processing, PMML, community, and R extension packages. The
package that is needed and used for text mining is the Text Processing package, which can
be installed and updated through the Update RapidMiner menu item under the Help menu.
1.1.5 Installing Text Mining Extensions
We will initiate our text mining analysis by importing the two previously built processes.
However, even before that we have to check and make sure that the extensions required for
text mining are installed within RAPIDMINER. To manage the extensions, select the Help
menu and then manage Extensions menu item, as shown in Figure 1.4. The dialog box
that comes up, as shown in Figure 1.5, does not list any extensions. Thus, the extensions
required have to be installed.
Figure 1.4 Managing RAPIDMINER extension packages

INTRODUCTION 5
Figure 1.5 Dialog box stating that no extension packages are yet installed
Figure 1.6 Updating RAPIDMINER
Figure 1.7 Installing extension packages for text mining

Click the Close button and select Help menu and then Update RapidMiner menu item
as shown in Figure 1.6. RAPIDMINER will connect to the internet and fetch the list of
available updates, eventually displaying all the available updates, as in Figure 1.7. In this
window, the selection of an update for installation can be made only by double clicking
the check box on the left hand side of that update’s name. When an update is selected
for installation, a small green check sign appears on the check box, as in Figure 1.7. The
extension packages (available as updates) needed for text mining are Text Processing,
Web Mining, and Wordnet Extension. Select these updates as shown in Figure 1.7.
Then, click the install button. When the Terms of Use window appears, click the radio
button for I accept the terms of use, and click Ok . During the installation of updates,
the installation process can be stopped by ﬁrst clicking inside the Progress window, and
clicking Stop , but don’t do this now. When the Update Complete dialog box appears,
click Yes to restart RAPIDMINER with the newly installed updates available . When the
splash screen for RAPIDMINER is displayed, you will notice that the icons for the installed
extensions appear in the previously blank region (Figure 1.8).
Figure 1.8 RAPIDMINER splash screen with the three extension packages installed for text mining

ASSOCIATION MINING OF TEXT DOCUMENT COLLECTION (Process01) 7
1.2 ASSOCIATION MINING OF TEXT DOCUMENT COLLECTION
(Process01)
1.2.1 Importing Process01
We will now initiate the text mining analysis by importing the processes supplied with this
chapter. For this, select Files and then the Import Process menu item as in Figure 1.9. Go
to the LocalRepository folder (supplied with this chapter), then to Processes folder,
click on Process01.rmp, and click the Open button, as in Figure 1.10. The RAPIDMINER
process Process01 is now displayed in the design perspective, as shown in Figure 1.11.
Figure 1.9 Importing an existing process

Figure 1.10 Selecting the process to import
Figure 1.11 Operators for Process01 and the Parameters for Process Documents from Files
operator

1.2.2 Operators in Process01
The parameters for the operators in this process are given on the right-hand side of the
process, listed under the Parameters tab. For the first process, the parameter text di-
rectories specifies where to read the text data from. A very important parameter is the
vector creation used. In Process01, the selected vector creation method is TF-IDF
TF-IDF (Term Frequency-Inverse Document Frequency) is a term weighting method. It
gives higher weight to terms that appear frequently in the document but not many times in
other documents. However, this may yield too many words, even up to tens of thousands of
them. Too many words would make it prohibitive to carry out the successive data mining
steps due to lengthy running times for the data mining algorithms. Hence, it is a very good
idea to prune the resulting word set using a prune method, by selecting the method and
its parameters within the Parameters view.
Click on the Process Documents from Files operator, as in Figure 1.11. In Pro-
cess01, words that appear in less than 70.0 % of the documents are pruned, as can be seen
from the value of 70.0 for the prune below percent parameter. It is also possible to prune
the words that appear in too many documents, but this was not done in this example, as can
be seen from the value of 100.0 for the prune above percent parameter. In association
mining, we are interested in the items (words in text mining context) that appear in 100.0%
of the transactions (documents in our text mining context), since they can form interest-
ing frequent itemsets (word lists) and association rules, that provide actionable insights.
Thus, it is appropriate to set prune above percent to 100.0 in Process01 (Figure 1.11),
including the items (words) that appear in every document.
Process01 consists of five operators (Figure 1.11). Firstly, the Process Documents
from Files operator performs text processing which involves preparing the text data for
the application of conventional data mining techniques. Process Documents from Files
operator reads data from a collection of text files and manipulates this data using text
processing algorithms. This is a nested operator, meaning that it can contain a sub-process
consisting of a multitude of operators. Indeed, in Process01, this nested operator contains
other operators inside. Double click on this operator, and you will see the sub-process
inside it, as outlined in Figure 1.12. This sub-process consists of six operators that are
serially linked (Figure 1.12):
• Tokenize Non-letters (Tokenize)
• Tokenize Linguistic (Tokenize)
• Filter Stopwords (English)
• Filter Tokens (by Length)
• Stem (Porter)
• Transform Cases (2) (Transform Cases)

Figure 1.12 Operators within the Process Documents from Files nested operator

Figure 1.13 Parameters for the operators within the Process Documents from Files operator

The sub-process basically transforms the text data into a format that can be easily
analyzed using conventional data mining techniques such as association mining and cluster
modeling. The parameters for each of the operators in this sub-process (within the Process
Documents from Files operator) are displayed in Figure 1.13. Notice that Figure 1.13
was created manually by combining multiple snapshots.
In this sub-process, the Tokenize Non-letters (Tokenize) and Tokenize Linguistic
(Tokenize) operators are both created by selecting the Tokenize operator, but with different
parameter selections. The former operater tokenizes based on non letters whereas the latter
operater tokenizes based on the linguistic sentences within the English language. The
Filter Stopwords (English) operator removes the stop words in the English language from
the text data set. The Filter Tokens (by Length) operator removes all the words composed
of less than min chars characters and more than max chars characters. In this example,
words that have less than 2 characters or more than 25 characters are removed from the
data set. The Stem (Porter) operator performs stemming and the Transform Cases(2)
(Transform Cases) operator transforms all the characters in the text into lower case. It
should be noted that the name of this last operator is not a good name, and we, as the
modelers have forgotten to rename the operator after its name was automatically assigned
by RAPIDMINER. This mistake should be avoided in constructing the processes.
Figure 1.14 Parameters for the operators in Process01

The parameters for each of the operators in Process01 (Figure 1.11) are displayed
in Figure 1.14. Notice that Figure 1.14 was created manually by combining multiple
snapshots. The Text to Nominal operator transforms the text data into nominal (categorical)
data. The Numerical to Binomial operator then transforms the data into binominal
form. This means that each row represents a document, a few columns provide meta-
data about that document and the remaining columns represent the words appearing in
all the documents, with the cell contents telling (true or false) whether that word exist
in that document or not. FP-Growth algorithm is used for identifying the frequent item
sets. In this example, the min support parameter is 0.7 (Figure 1.14), meaning that the
operator generates a list of the frequent sets of words (itemsets) that appear in at least 70
% of the documents. Notice that, it will be computationally efficient to select the min
support value in the FP-Growth operator to be equal to prune below percent value for
the Process Documents from File operator (Figure 1.11) divided by 100. Also, the
max items parameter is 2, meaning that the generated list is limited to pairs of words
(2-itemsets), and the list will not contain frequent word sets (itemsets) with 3 or more
words in them. The final operator in Process01, namely Create Association Rules,
receives the list of frequent word sets from the FP-Growth operator, and computes the
rules that satisfy the specified constraints on selected association mining criteria. In this
example, the association rules are computed according to the the criterion of confidence,
as well as gain theta and laplace k. The specified minimal values for these 3 criteria are
0.8, 1.0 and 1.0, respectively.
1.2.3 Saving Process01
So far Process01 has been imported in to the workspace of RAPIDMINER. Now it is a
good time to keep it in the LocalRepository so that we will not have to import it again
next time we work with RAPIDMINER. Click on the Repositories view, right-click on
the LocalRepository and select Configure Repository as shown in Figure 1.15. This
is actually an initialization step before using RAPIDMINER for the first time, but we will
still go through this step to ensure that we are saving everything to a preferred folder
directory in our computer. In this chapter, we will be saving everything under the Root
directory of C:RapidMiner, as in Figure 1.16. Next, we can save Process01 in to the
LocalRepository. Right-click on the LocalRepository text and select Store Process
Here, as in Figure 1.17. When the Store Process dialog window appears, click Ok. Our
importing and saving of Process01 is now completed.

Figure 1.15 Conﬁguring LocalRepository
Figure 1.16 Specifying the Root directory for LocalRepository
Figure 1.17 Storing Process01 in LocalRepository

CLUSTERING TEXT DOCUMENTS (Process02) 15
1.3 CLUSTERING TEXT DOCUMENTS (Process02)
The second text mining process that we will introduce in this chapter is Process02, and
involves the clustering of the 100 documents in the text collection.
1.3.1 Importing Process02
Toimportthisprocess, select Filemenu, Import Processmenuitem, select Process02.rmp
within the Import Process window, and click Open.
1.3.2 Operators in Process02
Process02 is given in Figure 1.18. Similar to Process01, Process02 begins with the
Process Documents from Files operator, whose parameters are given in Figure 1.18.
In Process02, the selected vector creation method is different; it is Term Frequency.
The impact of this new selection will be illustrated later. For now, it should be noted that
this selection results in the computation of the relative frequencies of each of the words
in each of the documents in the data set. For example, if a word appears 5 times within
a document that consist of 200 words, then the relative frequency of that word will be
5/200 = 0.025. This value of 0.025 will appear under the column for that word, at the
row for that document. In Process02, the prune method is again percentual, just as in
Process01 (Figure 1.11). However, the value for the prune below percent parameter is
now different.
Figure 1.18 Operators in Process02 and the Parameters for Process Documents from Files
operator

Figure 1.19 Operators within the Process Documents from Files nested operator and the
Parameters for the Generate n-Grams (Terms) operator
Figure 1.20 Parameters for the operators in Process02
The prune below percent parameter in Process01 was 70.0, whereas it is now 20.0
in Process02. The reason for this change is due to the fact that the applied data mining
technique in Process01 is association mining, whereas in Process02 it is clustering.

The former technique is computationally much more expensive then the latter, meaning
that running association mining algorithms on a data set takes much longer running time
compared to the k-Means clustering algorithm on the same data set. Therefore, with the
same amount of time available for our data mining process, we have to work with much
smaller data sets if we are carrying out association mining, as in Process01. The values
displayed in Figures 1.11 and 1.18 have been determined through trial and error, such that
the running time for the processes do not exceed 30 seconds on a laptop with Intel i5
processor and 4GB RAM.

Figure 1.19 shows the contents of this operator. While there were six operators within this
nested operator in Process01, there are now seven operators in Process02. The newly-
added operator is Generate n-Grams (Terms). The only parameter for this operator is
max length which is set equal to 2 in our example (Figure 1.19).
The parameters for the Select Attributes and Clustering (k-Means(fast)) operators
within Process02 are displayed in Figure 1.20. The Select Attributes operator takes
the complete data set and transforms it into a new one by selecting only the columns with
numeric values, i.e. columns corresponding to the words. This transformation is required
for the next operator, which performs clustering based on numerical values. Clustering
(k-Means (fast)) operator carries out the k-Means clustering algorithm on the numerical
data set. Since each row of the data (each document in the text collection) is characterized
in terms of the occurrence frequency of words in it, this operator will place together the
documents that have a similar distribution of the word frequencies. The k value, which
denotes the number of clusters to be constructed, is set to 5 in our example.
1.3.3 Saving Process02
Now, let us save Process02 into LocalRepository: Click on Repositories view, right
click on LocalRepository view and select Store Process Here. When the store process
dialog box appears click Ok. As shown in Figure 1.21, both processes are now saved under
local repository. Since we had earlier defined the directory root for local repository
as C:RapidMiner, the saved processes will appear as .rpm (RAPIDMINER process) files
under that directory.
Unfortunately, there is a problem with the naming of the files: the file names also contain
the extension .rpm, in addition to the .rpm extension itself. For example, the correct name
for Process01rmp.rmp under C:RapidMiner should be Process01.rmp. Right click
on Process01.rmp text within RAPIDMINER, and select rename as in Figure 1.22 then,
as in Figure 1.23 remove the .rmp extension from the file name and click OK. Do the same
for renaming Process02 and you will obtain the results in Figure 1.24.

Figure 1.21 Saved processes within LocalRepository
Figure 1.22 Renaming a process

Figure 1.23 Renaming Process01 correctly
Figure 1.24 Corrected names for Process01 and Process02 in LocalRepository and the
corresponding ﬁles

RUNNING Process01 AND ANALYZING THE RESULTS 21
1.4 RUNNING Process01 AND ANALYZING THE RESULTS
Having imported and saved the processes into local repository, now we will run these
processes.
1.4.1 Running Process01
For running Process01, double click on Process01.rmp under the Repositories view.
Process01 will be loaded into the Process view. Click the run button in the tool bar, as in
Figure 1.25. You may be prompted and asked whether you would like to close all results
before starting process (Figure 1.25). Click Yes. When the process run is completed,
you will be prompted again and asked whether you would like to switch to the result
perspective (Figure 1.26). Click Yes and you will have the screen in Figure 1.27. This is
indeed the result perspective, displaying the result obtained through your process. Until
now, you worked on your process in the design perspective, where we could change your
process by adding/removing/editing operators. Now, in the result perspective, we are not
able to modify our process. In the tool bar, we can switch to the design perspective or the
result perspective by clicking their respective buttons.
Figure 1.25 Opening and running Process01
Figure 1.26 Dialog box alerting the switch to the result perspective

Figure 1.27 Result Overview for Process01 results

1.4.2 Empty Results for Process01
In result perspective, we can see the latest results at the bottom of the result overview.
To see the details click inside the bar with the most recent date and time and you will
see an expanded view as in Figure 1.27. In the extended view you will note that there
are three types of results obtained as a result of running Proces01. These three results
correspond precisely to the three result ports for Process01 which are represented by the
three half-circles on the right of the process view in Figure 1.11 and are labeled with res.
The first result is the wordlist (wor) arising from the Process Documents From Files
operator (Figure 1.11). The second result is the example set exa, which arises from the
FP-Growth operator. The third and final result is the list of association rules (rul) which
arises from the Create Association Rules.
In Figure 1.27 the word list contains 0 entries and the number of examples is 0. But
why? Was there something wrong with our processes? We will answer this question next.
1.4.3 Specifying the Source Data for Process01
Running Process01 resulted in no tangible results and the reason is very simple. We have
not yet specified the text data set that will be processed and mined in Process01. To
specify the data source, as in Figure 1.28 switch to the design perspective by clicking the
notepad button, click on the Process Documents from Files operator and click on the
box to the right of the parameter text directories.
In the dialog box click on the directory button, (see Figure 1.29) select the source
data folder, double click TripAdvisor First 100 Hotels folder, and click the open button
(Figure 1.30). Now the directory for the source text data appears in the dialog box, as in
Figure 1.31. Now click Ok.

Figure 1.28 Specifying the data source text directories for Process01
Figure 1.29 Specifying the text directories

Figure 1.30 Specifying the text directories

Figure 1.31 Speciﬁed text directories

1.4.4 Re-Running Process01
Having speciﬁed the data source, we can run Process01 again. Click on the run button
(see Figure 1.32) and click Yes when you are asked. You should click Save Process
Before Start?, and click Yes when you are asked whether to Close old results before
starting process. When the process is running you can observe which operator is currently
being used. For example, in Figure 1.33 the FP-Growth algorithm is running, as can be
understood from the small green triangle on its lower left corner. From Figure 1.33 we
can also read (from the lower left corner of the RAPIDMINER window) that Process01 has
been running for 14 seconds (14 s) and the FP-Growth operator has been running for 7
seconds (7 s).
Figure 1.32 Running Process01 again
Figure 1.33 Running of Process01

1.4.5 Process01 Results
When the new results are created, click Yes to the result perspective. In the result perspective
you will now see a see a blue bar. Click inside the bottommost blue bar and you will see an
overview of the results, as in Figure 1.34. This time the world list is not empty, it Contains
631 entries. The Number of examples in the example set is equal to 100, and there
are 250 attributes (Figure 1.34).
Figure 1.35 WordList generated by Process01

Figure 1.36 Meta Data View for the ExampleSet generated by Process01
Figure 1.37 Data View for the ExampleSet generated by Process01

Click on the world list view to display the words found by the Process Documents
from Files operator, as in Figure 1.35.
Click on the ExampleSet (Numerical to Binomial) view to display the example set.
Figure 1.36 presents the Meta Data View, which gives the metadata (information on the
data attributes) of the ExampleSet. Now click on the Data View button to display the
data itself (Figure 1.37). Each row corresponds to a text document and the first column
is an automatically generated key attribute. The extra four columns contain the label, file
name, file location and the file creation data. The remaining columns correspond to the
world list. The cells under the word attributes take the binominal value of true or false,
denoting whether the word exists in that document.
The final result is the set of association rules. Click on the AssociationRules(Create
Association Rules) tab to display the association rules (Figure 1.38). In the Table View,
table grid presents the generated association rules, with one rule in each row. For example,
the second row states “IF worth THEN holidai” with a Support level of 0.760 and
Confidence level of 0.800. This rule means that in 76 of the 100 documents, words with
stem worth and holidai appear together. Furthermore in 80% of the documents where a
word derived from the stem worth appears, at least one word derived from the stem holidai
is observed. On the left hand side of the grid we can select and Show rules matching a
particular set of words (Figure 1.38).
Figure 1.38 Table View for the AssociationRules generated by Process01

Figure 1.39 Graph View for the AssociationRules generated by Process01
Figure 1.40 Graph View for the AssociationRules, without the node labels

Figure 1.41 Filtering rules in the Graph View for the AssociationRules

These association rules can also be displayed in the form of a graph. For this, click on
the Graph View radio button as in Figure 1.39. The initial graph visualization may be
cluttered due to the Node Labels. As Click on the check box Node Labels to uncheck
that option and eliminate or reduce the clutter in (Figure 1.40). In this window, filtering
can again be carried out by selecting words from the list box on the left hand side. Once a
filtered word is selected you will want to again display the Node Labels as in Figure 1.41.
At this point, select a word from the list box and check the Node Label check box (Figure
1.41).
The final interesting result that we would like to share is displayed in Figure 1.42. Click
on WordList(Process Documents from Files) tab to return back to the word list. Click
on the attribute label Document Occurences to obtain a descending order. Then move
the vertical scroll bar to see the border line where the value changes from 100 to 99. From
Figure 1.42 we can observe that the word work appears in all 100 documents, whereas the
word big appears in 99 documents.
Figure 1.42 Document Occurences of the words in the WordList

Figure 1.43 Saving the WordList for Process01
Figure 1.44 Saving the ExampleSet for Process01

1.4.6 Saving Process01 Results
Having generated results of the Process01 we will save these results into LocalRepos-
itory, so that we can save time when we want to analyze the results next time. The three
types of results, namely the Word List, ExampleSet, AssociationRules have to be saved
individually. First, click on the save button in the upper right corner of the results, click
on LocalRepository in the appearing window, and type in an appropriate name inside the
Name text box as in Figure 1.43, then click Ok.
Next, click on the ExampleSet (Numerical to Binominal) tab, and repeat the same
steps, as in Figure 1.44.
Finally, click on the AssociationRules, as in Figure 1.45. For this last result we would
like to show you a common possible mistake. After clicking the save button, omit clicking
on LocalRepository and click Ok as Figure 1.45. Please note that in this case we did not
specify a location for the generated results. This will cause an error . When you see the
error dialog box , click Close. Then, follow the correct steps; Click on the save button,
click on LocalRepository, give an appropriate unique name and click Ok as in Figure
1.46.
Figure 1.45 Saving the AssociationRules incorrectly, without selecting LocalRepository or
another directory

Figure 1.46 Saving the AssociationRules correctly, selecting the directory to be saved into

Now that all the data for the generated results are saved let us also see how we can save
(export) the data visualizations. Click on the export button to the right of save button, as
in Figure 1.47. Then specify the directory where the visualization will be saved (exported)
and select the image file type as in Figure 1.48. In this example, select the .png image
format, and then click Ok. You can check that the visualization is now saved as an image
file, as specified.
Figure 1.47 Exporting the graph visualization as an image file
Figure 1.48 Specifying the directory to be saved into, the file name, and the file type

1.5 RUNNING Process02 AND ANALYZING THE RESULTS
In this section we will run Process02 and analyze the generated results.
1.5.1 Running Process02
To switch to the design view click on the design view button (Notepad icon) as in Figure
1.49. Then double click on Process02 inside the Repositories view, and click the run
button as in Figure 1.50. You will be prompted with an error dialog box, as in Figure 1.51.
Figure 1.49 Switching to the design view
Figure 1.50 Opening and running Process02

1.5.2 Specifying the Source Data for Process02
The reason for this error is that we have not specified the data set for Process02 to solve
this problem. To resolve this, first click on Close (Figure 1.51). Then, as in Figure 1.52,
click on Process Documents from Files operator and click on the button next to the text
directories parameter inside the Parameters view. Then, select the same data set as you
have selected for Process01, as shown in Figures 1.28, 1.29, 1.30 and 1.31. Now the data
source has been specified and we can run Process02 to obtain tangible results. Click on
the run button as in Figure 1.50, and answer Yes to the questions that are prompted.
Figure 1.51 Error message due to not specifying the source data for Process02
Figure 1.52 Specifying the text directories for the source data of Process02

1.5.3 Process02 Results
The results of Process02 are shown in Figure 1.53. There are three types of results
generated after running Process02:

Firstly, the Word List contains the ﬁltered words from the documents in our example,
the Word List generated Contains 5175 entries.
The second result is the Example Set, which contains data requiring the word compo-
sition of each of the documents in the database.
The third result of Process02 is Centroid Cluster Model. In our example ﬁve clusters
(Cluster 0, Cluster 1, Cluster2, Cluster 3, Cluster 4) have been generated, where each
cluster contains a subset of 100 documents.
The Word List is very similar to the word list of Process01 word list. However, there
is a difference due to the additional Generate n-Grams (Terms) operator (Figure 1.19).
This additional operator extracts n-grams (sequences of words), as well as single words. To
see these generated n-grams as a part of the word list you can click on the WordList view or
the Example view. Click on the Example view, as in Figure 1.54 and you will see several
of these n-grams. For example, besides the word absolut the n-gram words absolut love,
absolut perfect, absolut stai are part of the WordList, and appear as attributes in the
ExampleSet (Figure 1.54). Now, to see the data itself, click on Data View radio button
as in Figure 1.55. In this grid, each number tells the relative frequency of a word in a
document. For example, the word absolut (the word that have absolut as their stem)
constitutes 0.030 proportion (3%) of all the words in document number 13, that contains
the reviews for hotel 73943.
Figure 1.54 Meta Data View for the ExampleSet generated by Process02, including the
n-Grams

The main contribution of Process02 is the cluster model, which clusters documents,
and thus the hotels, according to the similarity of the frequency distribution of the words
contained in their TripAdvisor reviews. Now click on Cluster Model (clustering) view
to view the cluster model results, as in Figure 1.56. Then, to view the content of each of
the clusters, click on Folder View radio button, as in Figure 1.57. In our example the ﬁrst
cluster of hotels is cluster 0, which contains 28 hotels, including hotels with row numbers
47, 51, 55, 95, 100.
Figure 1.55 Data View for the ExampleSet generated by Process02, and the relative occurence
frequency of the word absolut in hotel 73943.txt
Figure 1.56 Text View for the Cluster Model generated by Process02, displaying the number
of examples in each cluster

Figure 1.57 Folder View for the Cluster Model generated by Process02, displaying the examples
in each cluster

How do these clusters differ from each other? To get an answer to this question, click
on Centroid Table radio button (Figure 1.58). In this Centroid table we can observe the
average frequency of each word in the documents of each cluster. For example the word
absolut appears much more frequently in cluster 3, compared to the other clusters: In
cluster 3, the average frequency for absolut is 0.013, whereas its at most 0.008 in the
others.
Figure 1.58 Centroid Table for the Cluster Model generated by Process02, displaying the
average frequency of each word in each cluster
Finally, save the results of Process02, just as you did for Process01. The ﬁnal picture
in the Repositories view will be as in Figure 1.59.
Figure 1.59 Final view of LocalRepository with the processes and their results saved

CONCLUSIONS 45
1.6 CONCLUSIONS
In this chapter, we have shown the basic RAPIDMINER operators for text mining, as well as
how they can be used in conjunction with other operators that implement popular conven-
tional data mining techniques. Throughout the chapter, we have discussed two processes,
Process01 and Process02, which complement text processing with association mining
and cluster modeling, respectively. We have shown how these text mining processes, which
are formed as combinations of text processing and data mining techniques, are modeled
and run in RAPIDMINER, and discussed how their results are analyzed. RAPIDMINER has an
extensive set of operators available for text processing and text mining, which can also be
used for extracting data from the web, and perform a multitude of other types of analysis.
While most of these operators were not discussed in this chapter due to space limitations,
the reader is strongly encouraged not to suffice with the contents of this chapter, and ex-
plore and experiment other text processing and mining operators and capabilities within
RAPIDMINER, its official extensions, and the processes posted by the RAPIDMINER user
community.
Acknowledgement
This work is financially supported by the UbiPOL Project within the EU Seventh Framework
Programme, under grant FP7-ICT 248010. The authors thank Assoc. Prof. Yücel Saygın
for supporting and helping the writing of this chapter. The authors also acknowledge the
earlier help of Sabancı University alumni Yücel Balıklılı and Emir Balıkçı with the design
of the first process in this chapter.

GLOSSARY
Decision Tree Decision Trees are....
Data Exploration Data Exploration is of particular importance when...
Use Cases with RapidMiner, First Edition. By Hofmann, Klinkenberg
Copyright c 2012 John Wiley & Sons, Inc.
47

Text Mining with RapidMiner

More Related Content

Similar to Text Mining with RapidMiner (20)

More from ertekg (20)

Recently uploaded (20)

Text Mining with RapidMiner