Sammon Embedding with Tensorflow

Embedding algorithms, especially word-embedding algorithms, have been one of the recurrent themes of this blog. Word2Vec has been mentioned in a few entries (see this); LDA2Vec has been covered (see this); the mathematical principle of GloVe has been elaborated (see this); I haven’t even covered Facebook’s fasttext; and I have not explained the widely used t-SNE and Kohonen’s map (self-organizing map, SOM).

I have also described the algorithm of Sammon Embedding, (see this) which attempts to capture the likeliness of pairwise Euclidean distances, and I implemented it using Theano. This blog entry is about its implementation in Tensorflow as a demonstration.

Let’s recall the formalism of Sammon Embedding, as outlined in the previous entry:

Assume there are high dimensional data described by d-dimensional vectors, X_i where i=1, 2, \ldots, N. And they will be mapped into vectors Y_i, with dimensions 2 or 3. Denote the distances to be d_{ij}^{*} = \sqrt{| X_i - X_j|^2} and d_{ij} = \sqrt{| Y_i - Y_j|^2}. In this problem, Y_i are the variables to be learned. The cost function to minimize is

E = \frac{1}{c} \sum_{i<j} \frac{(d_{ij}^{*} - d_{ij})^2}{d_{ij}^{*}},

where c = \sum_{i<j} d_{ij}^{*}.

Unlike in previous entry and original paper, I am going to optimize it using first-order gradient optimizer. If you are not familiar with Tensorflow, take a look at some online articles, for example, “Tensorflow demystified.” This demonstration can be found in this Jupyter Notebook in Github.

First of all, import all the libraries required:

import numpy as np
import matplotlib.pyplot as plt
import tensorflow as tf

Like previously, we want to use the points clustered around at the four nodes of a tetrahedron as an illustration, which is expected to give equidistant clusters. We sample points around them, as shown:

tetrahedron_points = [np.array([0., 0., 0.]), np.array([1., 0., 0.]), np.array([np.cos(np.pi/3), np.sin(np.pi/3), 0.]), np.array([0.5, 0.5/np.sqrt(3), np.sqrt(2./3.)])]

sampled_points = np.concatenate([np.random.multivariate_normal(point, np.eye(3)*0.0001, 10) for point in tetrahedron_points])

init_points = np.concatenate([np.random.multivariate_normal(point[:2], np.eye(2)*0.0001, 10) for point in tetrahedron_points])

Retrieve the number of points, N, and the resulting dimension, d:

N = sampled_points.shape[0]
d = sampled_points.shape[1]

One of the most challenging technical difficulties is to calculate the pairwise distance. Inspired by this StackOverflow thread and Travis Hoppe’s entry on Thomson’s problem, we know it can be computed. Assuming Einstein’s convention of summation over repeated indices, given vectors a_{ik}, the distance matrix is:

D_{ij} = (a_{ik}-a_{jk}) (a_{ik} - a_{jk})^T = a_{ik} a_{ik} + a_{jk} a_{jk} - 2 a_{ik} a_{jk},

where the first and last terms are simply the norms of the vectors. After computing the matrix, we will flatten it to vectors, for technical reasons omitted to avoid gradient overflow:

X = tf.placeholder('float')
Xshape = tf.shape(X)

sqX = tf.reduce_sum(X*X, 1)
sqX = tf.reshape(sqX, [-1, 1])
sqDX = sqX - 2*tf.matmul(X, tf.transpose(X)) + tf.transpose(sqX)
sqDXarray = tf.stack([sqDX[i, j] for i in range(N) for j in range(i+1, N)])
DXarray = tf.sqrt(sqDXarray)

Y = tf.Variable(init_points, dtype='float')
sqY = tf.reduce_sum(Y*Y, 1)
sqY = tf.reshape(sqY, [-1, 1])
sqDY = sqY - 2*tf.matmul(Y, tf.transpose(Y)) + tf.transpose(sqY)
sqDYarray = tf.stack([sqDY[i, j] for i in range(N) for j in range(i+1, N)])
DYarray = tf.sqrt(sqDYarray)

And DXarray and DYarray are the vectorized pairwise distances. Then we defined the cost function according to the definition:

Z = tf.reduce_sum(DXarray)*0.5
numerator = tf.reduce_sum(tf.divide(tf.square(DXarray-DYarray), DXarray))*0.5
cost = tf.divide(numerator, Z)

As we said, we used first-order gradient optimizers. For unknown reasons, the usually well-performing Adam optimizer gives overflow. I then picked Adagrad:

update_rule = tf.assign(Y, Y-0.01*grad_cost/lapl_cost)
train = tf.train.AdamOptimizer(0.01).minimize(cost)
init = tf.global_variables_initializer()

The last line initializes all variables in the Tensorflow session when it is run. Then start a Tensorflow session, and initialize all variables globally:

sess = tf.Session()

Then run the algorithm:

nbsteps = 1000
c =, feed_dict={X: sampled_points})
print "epoch: ", -1, " cost = ", c
for i in range(nbsteps):, feed_dict={X: sampled_points})
    c =, feed_dict={X: sampled_points})
    print "epoch: ", i, " cost =

Then extract the points and close the Tensorflow session:

calculated_Y =, feed_dict={X: sampled_points})

Plot it using matplotlib:

embed1, embed2 = calculated_Y.transpose()
plt.plot(embed1, embed2, 'ro')

This gives, as expected,

download (5)

This code for Sammon Embedding has been incorporated into the Python package mogu, which is a collection of numerical routines. You can install it, and call:

from mogu.embed import sammon_embedding
calculated_Y = sammon_embedding(sampled_points, init_points)

Continue reading “Sammon Embedding with Tensorflow”

Release of shorttext 0.3.3

On November 21, 2016, the Python package `shorttext’ was published. Until today, more than seven versions have been published. There have been a drastic architecture change, but the overall purpose is still the same, as summarized in the first introduction entry:

This package `shorttext‘ was designed to tackle all these problems… It contains the following features:

  • example data provided (including subject keywords and NIH RePORT);
  • text preprocessing;
  • pre-trained word-embedding support;
  • gensim topic models (LDA, LSI, Random Projections) and autoencoder;
  • topic model representation supported for supervised learning using scikit-learn;
  • cosine distance classification; and
  • neural network classification (including ConvNet, and C-LSTM).

And since the first version, there have been updates, as summarized in the documention (News):

Version 0.3.3 (Apr 19, 2017)

  • Deleted CNNEmbedVecClassifier.
  • Added script ShortTextWord2VecSimilarity.

Version 0.3.2 (Mar 28, 2017)

  • Bug fixed for gensim model I/O;
  • Console scripts update;
  • Neural networks up to Keras 2 standard (refer to this).

Version 0.3.1 (Mar 14, 2017)

  • Compact model I/O: all models are in single files;
  • Implementation of stacked generalization using logistic regression.

Version 0.2.1 (Feb 23, 2017)

  • Removal attempts of loading GloVe model, as it can be run using gensim script;
  • Confirmed compatibility of the package with tensorflow;
  • Use of spacy for tokenization, instead of nltk;
  • Use of stemming for Porter stemmer, instead of nltk;
  • Removal of nltk dependencies;
  • Simplifying the directory and module structures;
  • Module packages updated.

Although there are still additions that I would love to add, but it would not change the overall architecture. I may add some more supervised learning algorithms, but under the same network. The upcoming big additions will be generative models or seq2seq models, but I do not see them coming in the short term. I will add corpuses.

I may add tutorials if I have time.

I am thankful that there is probably some external collaboration with other Python packages. Some people have already made some useful contributions. It will be updated if more things are confirmed.

Continue reading “Release of shorttext 0.3.3”

Short Text Categorization using Deep Neural Networks and Word-Embedding Models

There are situations that we deal with short text, probably messy, without a lot of training data. In that case, we need external semantic information. Instead of using the conventional bag-of-words (BOW) model, we should employ word-embedding models, such as Word2Vec, GloVe etc.

Suppose we want to perform supervised learning, with three subjects, described by the following Python dictionary:

classdict={'mathematics': ['linear algebra',
           'variational calculus',
           'functional field',
           'real analysis',
           'complex analysis',
           'differential equation',
           'statistical optimization',
           'stochastic calculus',
           'numerical analysis',
           'differential geometry'],
          'physics': ['renormalization',
           'classical mechanics',
           'quantum mechanics',
           'statistical mechanics',
           'functional field',
           'path integral',
           'quantum field theory',
           'condensed matter',
           'particle physics',
           'topological solitons',
           'spontaneous symmetry breaking',
           'atomic molecular and optical physics',
           'quantum chaos'],
          'theology': ['divine providence',
           'Holy Trinity',
           'divine degree',
           'creedal confessionalism',

And we implemented Word2Vec here. To add external information, we use a pre-trained Word2Vec model from Google, downloaded here. We can use it with Python package gensim. To load it, enter

from gensim.models import Word2Vec
wvmodel = Word2Vec.load_word2vec_format('<path-to>/GoogleNews-vectors-negative300.bin.gz', binary=True)

How do we represent a phrase in Word2Vec? How do we do the classification? Here I wrote two classes to do it.


We can represent a sentence by summing the word-embedding representations of each word. The class, inside, is coded as follow:

from collections import defaultdict

import numpy as np
from nltk import word_tokenize
from scipy.spatial.distance import cosine

from utils import ModelNotTrainedException

class SumEmbeddedVecClassifier:
    def __init__(self, wvmodel, classdict, vecsize=300):
        self.wvmodel = wvmodel
        self.classdict = classdict
        self.vecsize = vecsize
        self.trained = False

    def train(self):
        self.addvec = defaultdict(lambda : np.zeros(self.vecsize))
        for classtype in self.classdict:
            for shorttext in self.classdict[classtype]:
                self.addvec[classtype] += self.shorttext_to_embedvec(shorttext)
            self.addvec[classtype] /= np.linalg.norm(self.addvec[classtype])
        self.addvec = dict(self.addvec)
        self.trained = True

    def shorttext_to_embedvec(self, shorttext):
        vec = np.zeros(self.vecsize)
        tokens = word_tokenize(shorttext)
        for token in tokens:
            if token in self.wvmodel:
                vec += self.wvmodel[token]
        norm = np.linalg.norm(vec)
        if norm!=0:
            vec /= np.linalg.norm(vec)
        return vec

    def score(self, shorttext):
        if not self.trained:
            raise ModelNotTrainedException()
        vec = self.shorttext_to_embedvec(shorttext)
        scoredict = {}
        for classtype in self.addvec:
                scoredict[classtype] = 1 - cosine(vec, self.addvec[classtype])
            except ValueError:
                scoredict[classtype] = np.nan
        return scoredict

Here the exception ModelNotTrainedException is just an exception raised if the model has not been trained yet, but scoring function was called by the user. (Codes listed in my Github repository.) The similarity will be calculated by cosine similarity.

Such an implementation is easy to understand and carry out. It is good enough for a lot of application. However, it has the problem that it does not take the relation between words or word order into account.

Convolutional Neural Network

To tackle the problem of word relations, we have to use deeper neural networks. Yoon Kim published a well cited paper regarding this in EMNLP in 2014, titled “Convolutional Neural Networks for Sentence Classification.” The model architecture is as follow: (taken from his paper)


Each word is represented by an embedded vector, but neighboring words are related through the convolutional matrix. And MaxPooling and a dense neural network were implemented afterwards. His paper involves multiple filters with variable window sizes / spatial extent, but for our cases of short phrases, I just use one window of size 2 (similar to dealing with bigram). While Kim implemented using Theano (see his Github repository), I implemented using keras with Theano backend. The codes, inside, are as follow:

import numpy as np
from keras.layers import Convolution1D, MaxPooling1D, Flatten, Dense
from keras.models import Sequential
from nltk import word_tokenize

from utils import ModelNotTrainedException

class CNNEmbeddedVecClassifier:
    def __init__(self,
        self.wvmodel = wvmodel
        self.classdict = classdict
        self.n_gram = n_gram
        self.vecsize = vecsize
        self.nb_filters = nb_filters
        self.maxlen = maxlen
        self.trained = False

    def convert_trainingdata_matrix(self):
        classlabels = self.classdict.keys()
        lblidx_dict = dict(zip(classlabels, range(len(classlabels))))

        # tokenize the words, and determine the word length
        phrases = []
        indices = []
        for label in classlabels:
            for shorttext in self.classdict[label]:
                category_bucket = [0]*len(classlabels)
                category_bucket[lblidx_dict[label]] = 1

        # store embedded vectors
        train_embedvec = np.zeros(shape=(len(phrases), self.maxlen, self.vecsize))
        for i in range(len(phrases)):
            for j in range(min(self.maxlen, len(phrases[i]))):
                train_embedvec[i, j] = self.word_to_embedvec(phrases[i][j])
        indices = np.array(indices,

        return classlabels, train_embedvec, indices

    def train(self):
        # convert classdict to training input vectors
        self.classlabels, train_embedvec, indices = self.convert_trainingdata_matrix()

        # build the deep neural network model
        model = Sequential()
                                input_shape=(self.maxlen, self.vecsize)))
        model.add(Dense(len(self.classlabels), activation='softmax'))
        model.compile(loss='categorical_crossentropy', optimizer='rmsprop')

        # train the model, indices)

        # flag switch
        self.model = model
        self.trained = True

    def word_to_embedvec(self, word):
        return self.wvmodel[word] if word in self.wvmodel else np.zeros(self.vecsize)

    def shorttext_to_matrix(self, shorttext):
        tokens = word_tokenize(shorttext)
        matrix = np.zeros((self.maxlen, self.vecsize))
        for i in range(min(self.maxlen, len(tokens))):
            matrix[i] = self.word_to_embedvec(tokens[i])
        return matrix

    def score(self, shorttext):
        if not self.trained:
            raise ModelNotTrainedException()

        # retrieve vector
        matrix = np.array([self.shorttext_to_matrix(shorttext)])

        # classification using the neural network
        predictions = self.model.predict(matrix)

        # wrangle output result
        scoredict = {}
        for idx, classlabel in zip(range(len(self.classlabels)), self.classlabels):
            scoredict[classlabel] = predictions[0][idx]
        return scoredict

The output is a vector of length equal to the number of class labels, 3 in our example. The elements of the output vector add up to one, indicating its score, and a nature of probability.


A simple cross-validation to the example data set does not tell a difference between the two algorithms:


However, we can test the algorithm with a few examples:

Example 1: “renormalization”

  • Average: {‘mathematics’: 0.54135105096749336, ‘physics’: 0.63665460856632494, ‘theology’: 0.31014049736087901}
  • CNN: {‘mathematics’: 0.093827009201049805, ‘physics’: 0.85451591014862061, ‘theology’: 0.051657050848007202}

As renormalization was a strong word in the training data, it gives an easy result. CNN can distinguish much more clearly.

Example 2: “salvation”

  • Average: {‘mathematics’: 0.14939650156482298, ‘physics’: 0.21692765541184023, ‘theology’: 0.5698233329716329}
  • CNN: {‘mathematics’: 0.012395491823554039, ‘physics’: 0.022725773975253105, ‘theology’: 0.96487873792648315}

“Salvation” is not found in the training data, but it is closely related to “soteriology,” which means the doctrine of salvation. So it correctly identifies it with theology.

Example 3: “coffee”

  • Average: {‘mathematics’: 0.096820211601723272, ‘physics’: 0.081567332119268032, ‘theology’: 0.15962682945135631}
  • CNN: {‘mathematics’: 0.27321341633796692, ‘physics’: 0.1950736939907074, ‘theology’: 0.53171288967132568}

Coffee is not related to all subjects. The first architecture correctly indicates the fact, but CNN, with its probabilistic nature, has to roughly equally distribute it (but not so well.)

The code can be found in my Github repository: stephenhky/PyShortTextCategorization. (This repository has been updated since this article was published. The link shows the version of the code when this appeared online.)

Continue reading “Short Text Categorization using Deep Neural Networks and Word-Embedding Models”

NSFW Image Classification

At the end of last month, Yahoo opened the sources of training a model to classify not suitable/safe for work (NSFW) images, particularly pornographic images, using convolutional neural network (CNN). It was implemented with Caffe. Users need to supply the training data, positive being the NSFW images, and negative being the suitable/safe for work (SFW) images, to train the model. The model takes an image as the input, and output a score between 0 and 1.

The codes are available on Github, with the about the installation.


Continue reading “NSFW Image Classification”

rJava: Running Java from R, and Building R Packages Wrapping a .jar

While performing exploratory analysis, R is a good tool, although we sometimes want to invoke some stable Java tools. It is what the R Package rJava is for. To install it, simply enter on the R Console:


And to load it, enter:


As a simple demonstration, we find the length of a strength. Start the JVM, enter:


Then we create an instance of a Java string, and find its length as follow:

s <- .jnew('java/lang/String', 'Hello World!')
.jcall(s, 'I', 'length')

The first line, with the function .jnew, create a Java string instance. It is safe to put the full package path of the class. The second line, with the function .jcall, call the method length() for String. The second parameter, ‘I’, indicates it returns an integer. The type has to follow the JNI notation for native types. If it is an integer double array, type ‘I[[‘. If it is not a native class like String, use its total package path.

Example: Peter Norvig’s Spell Corrector Written in Scala

What should we do if we already have a .jar file we want to wrap? I would start with a simple one. Two years ago, I implemented Peter Norvig’s spell corrector (see his article) in Scala (which is a language for Java Virtual Machine (JVM) as well, see this entry), and posted on my Github repository: stephenhky/SpellCorrector. You may check out to your Eclipse or IntelliJ IDEA, and build a .jar file. (Or you can download the .jar file here.) For the program to run, do not forget to download his corpus named big.txt. The project has a class called SpellCorrector, which only the necessary codes are listed below:

package home.kwyho.spellcheck


import scala.collection.mutable.Map

class SpellCorrector {
 var wordCounts : Map[String, Int] = Map()
 val alphabets = ('a' to 'z').toSet

 def train(trainFile : File) = {
    val lines = Source.fromFile(trainFile) mkString
    val wordREPattern = "[A-Za-z]+"
    wordREPattern.r.findAllIn(lines).foreach( txtWord => {
       val word = txtWord.toLowerCase
       if (wordCounts.keySet contains(word)) {
          wordCounts(word) = wordCounts(word)+1
       } else {
          wordCounts += (word -> 1)

// other codes here ....

 def correct(wrongSpelling: String) : String = {
    val edit0words = Set(wrongSpelling) intersect wordCounts.keySet
    if (edit0words.size>0) return edit0words.maxBy( s => wordCounts(s))
    val edit1words = getEditOneSpellings(wrongSpelling)
    if (edit1words.size>0) return edit1words.maxBy( s => wordCounts(s))
    val edit2words = getEditTwoSpellings(wrongSpelling)
    edit2words.maxBy( s => wordCounts(s))

Putting the .jar file and big.txt into the same folder. Then initialize the JVM, and add the .jar file into the classpath:


Create an instance for SpellChecker, and train the corpus big.txt. Remember to put the whole package path as the class:

corrector <- .jnew('home/kwyho/spellcheck/SpellCorrector')
bigfile <- .jnew('java/io/File', 'big.txt')
.jcall(corrector, 'V', 'train', bigfile)

The first line create a SpellChecker instance, the second line create a File instance for big.txt, and the third line call the train() method. The JNI notation ‘V’ denotes ‘void.’ Entering ‘corrector’ will give a string indicates it is a Java object:

[1] "Java-Object{home.kwyho.spellcheck.SpellCorrector@5812f9ee}"

Then we can do spell correction by designing the following function:

correct<-function(word) {
   javaStrtext <- .jnew('java/lang/String', word)
   .jcall(corrector, 'Ljava/lang/String;', 'correct', javaStrtext)

Then you can easily perform spell correction as follow:


Some people put .class file instead of .jar file. In that case, you need to put the compiled Java class into the working directory. You can refer to an entry in Darren Wilkinson’s research blog for more details.

Building an R Package

It is another matter to build an R package that wraps a .jar file. In Hilary Parker’s entry and my previous entry, there are details about building an R package with roxygen2. There is also a documentation written by Tobias Verbeke.

So to start building it, in RStudio, start a project by clicking on the button “Project: (None)” on the top right corner of RStudio, choose “New Directory,” and then “R Package.” Type in the name (“RSpellCorrection” here), and specify a directory. Then click “Create Project.” A new RStudio window will show up. From the menu bar, choose “Build” > “Configure Build Tools”. Then click on “Configure…” button. There is a dialog box coming out. Check everything, and click “OK”.


The instructions above are rather detailed. But starting from now, I will skip the procedural details. Then start a file named, say, onLoad.R under the subfolder R/, and put the following codes there:

.onLoad <- function(libname, pkgname) {
  .jpackage(pkgname, lib.loc=libname)

This is a hook function that R will call when this package is being loaded. You must include it. Then in the file named DESCRIPTION, put in the relevant information:

Package: RSpellCorrection
Type: Package
Title: Spell Correction, Scala implementation run in R
Version: 0.1.0
Author: Kwan-Yuet Ho, Ph.D.
Maintainer: Kwan-Yuet Ho, Ph.D. <>
Description: Implementation of Peter Norvig's spell corrector in Scala, wrapped in R
License: N/A
LazyData: TRUE
RoxygenNote: 5.0.1
Depends: R(>= 2.7.0), rJava (>= 0.5-0)

Note the last line (“Depends…”), which you have to include because R will parse this line, and load rJava automatically. Remember there is a space between “>=” and the version number. Do not use library function in your code.

First, create a subfolder inst/java, and put the .jar file there.

Then start a file, called correct.R under subfolder R/, and write a function:

#' Retrieve a Java instance of SpellCorrector.
#' Retrieve a Java instance of SpellCorrector, with the training file
#' specified. Language model is trained before the instance is returned.
#' The spell corrector is adapted from Peter Norvig's demonstration.
#' @param filepath Path of the corpus.
#' @return a Java instance of SpellCorrector
#' @export
getcorrector<-function(filepath='big.txt') {
    .jaddLibrary('spellchecker', 'inst/java/spellcorrector.jar')
    corrector<- .jnew('home/kwyho/spellcheck/SpellCorrector')
    bigfile<- .jnew('java/io/File', filepath)
    .jcall(corrector, 'V', 'train', bigfile)

This return a Java instance of SpellCorrector as in previous section. There is a large block of text above the function, and they are for producing manual using roxygen2. The tag “@export” is important to tell roxygen2 to make this function visible to the users.

Then add another function:

#' Correct spelling.
#' Given an instance of SpellCorrector, return the most probably
#' corrected spelling of the given word.
#' @param word A token.
#' @param corrector A Java instance of SpellCorrector, given by \code{getcorrector}.
#' @return Corrected spelling
#' @export
correct<-function(word, corrector) {
    javaStrtext <- .jnew('java/lang/String', word)
    .jcall(corrector, 'Ljava/lang/String;', 'correct', javaStrtext)

Then click “Build & Reload” button on the “Build” Tab:


Then the package will be built, and reloaded. The manual documents (*.Rd) will be produced as well. You can then play with the spell corrector again like this:


Assuming you put this into the Github repository like I did (link here), you can install the new R package like this:


Then the R package will be downloaded, and installed for use. Or another option is that if you wish to install from your local directory, just enter:

install.packages('<path-to>/RSpellCorrection', repos = NULL, type = 'source')

A complete version of this R package can be found in my Github repository: stephenhky/RSpellCorrection. You may want to add a into the repository, which you need to know the Markdown language by referring to Lei Feng’s blog entry.

Continue reading “rJava: Running Java from R, and Building R Packages Wrapping a .jar”

Developing R Packages

Because of work, I developed two R packages to host the functions that I used a lot. It did bring me a lot of convenience, such as that I don’t have to start my data analysis in a particular folder and switch later on.

To do that, you need to use RStudio. Then you have to install devtools package by calling in the R console:


and load it by simply call:


And then you have to install the roxygen2 package by calling:


There are a lot of good tutorials about writing an R package. I especially like this Youtube video clip about building an R package with RStudio and roxygen2:

And Hilary Parker’s blog entry is useful as well.

On the other hand, if you are publishing your R package onto your Github repository, it would be nice to include a README file introducing your work. You need to know the Markdown language to write the file named, and put it onto the root folder of your repository. My friend, Qianli Deng, showed me this Lei Feng’s blog entry, which I found extremely useful. Markdown is remarkably simpler than LaTeX.

Continue reading “Developing R Packages”

textmineR: a New Text Mining Package for R

Previously, I wrote an entry on text mining on R and Python, and did a comparison. However, the text mining package employed was tm for R. But it has some problems:

  1. The syntax is not natural for an experienced R users.
  2. tm uses simple_triplet_matrix from the slam library for document-term matrix (DTM) and term-occurrence matrix (TCM), which is not as widely used as dgCMatrix from the Matrix library.

Tommy Jones, a Ph.D. student in George Mason University, and a data scientist at Impact Research, developed an alternative text mining package called textmineR. He presented in a Stat Prog DC Meetup on April 27, 2016. It employed a better syntax, and dgCMatrix. All in all, it is a wrapper for a lot of existing R packages to facilitate the text mining process, like creating DTM matrices with stopwords or appropriate stemming/lemmatizing functions. Here is a sample code to create a DTM with the example from the previous entry:


texts <- c('I love Python.',
           'R is good for analytics.',
           'Mathematics is fun.')

               doc_names = c(1:length(texts)),
               ngram_window = c(1, 1),
               stopword_vec = c(tm::stopwords('english'), tm::stopwords('SMART')),
               lower = TRUE,
               remove_punctuation = TRUE,
               remove_numbers = TRUE

The DTM is a sparse matrix:

3 x 6 sparse Matrix of class &amp;quot;dgCMatrix&amp;quot;
  analytics fun mathematics good python love
1         .   .           .    .      1    1
2         1   .           .    1      .    .
3         .   1           1    .      .    .

On the other hand, it wraps text2vec, an R package that wraps the word-embedding algorithm named gloVe. And it wraps a number of topic modeling algorithms, such as latent Dirichlet allocation (LDA) and correlated topic models (CTM).

In addition, it contains a parallel computing loop function called TmParallelApply, analogous to the original R parallel loop function mclapply, but TmParallelApply works on Windows as well.

textmineR is an open-source project, with source code available on github, which contains his example codes.

Continue reading “textmineR: a New Text Mining Package for R”

Blog at

Up ↑