Applied Natural Language Processing With Python

Implementing Machine Learning And Deep Learning

Algorithms For Natural Language It Pro York
University Skillsoft Books York University Ii
download
https://ebookbell.com/product/applied-natural-language-
processing-with-python-implementing-machine-learning-and-deep-
learning-algorithms-for-natural-language-it-pro-york-university-
skillsoft-books-york-university-ii-22049616

Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.

Applied Natural Language Processing In The Enterprise Teaching

Machines To Read Write And Understand 1st Edition Ankur A Patel

https://ebookbell.com/product/applied-natural-language-processing-in-
the-enterprise-teaching-machines-to-read-write-and-understand-1st-
edition-ankur-a-patel-33946944

The Applied Ai And Natural Language Processing Workshop Learn How To

Use Powerful Natural Language Processing Techniques Within Your Own
Artificial Intelligence Applications Krishna Sankar

https://ebookbell.com/product/the-applied-ai-and-natural-language-
processing-workshop-learn-how-to-use-powerful-natural-language-
processing-techniques-within-your-own-artificial-intelligence-
applications-krishna-sankar-49589870

Applied Natural Science Environmental Issues And Global Perspectives

Goldfein

https://ebookbell.com/product/applied-natural-science-environmental-
issues-and-global-perspectives-goldfein-5672250

Applied Ecology And Natural Resource Management Mcpherson Gr

https://ebookbell.com/product/applied-ecology-and-natural-resource-
management-mcpherson-gr-2045914
Traditional And Indigenous Knowledge For The Modern Eraa Natural And
Applied Science Perspective 1st Edition David R Katerere Editor Wendy
Applequist Editor Oluwaseyi M Aboyade Editor Chamunorwa Togo Editor

https://ebookbell.com/product/traditional-and-indigenous-knowledge-
for-the-modern-eraa-natural-and-applied-science-perspective-1st-
edition-david-r-katerere-editor-wendy-applequist-editor-oluwaseyi-m-
aboyade-editor-chamunorwa-togo-editor-11911644

Applied Pharmaceutical Practice And Nutraceuticals Natural Product

Development Cristbal No Aguilar

https://ebookbell.com/product/applied-pharmaceutical-practice-and-
nutraceuticals-natural-product-development-cristbal-no-
aguilar-48737048

Applied Geomorphology For Mitigation Of Natural Hazards 1st Edition

Masahiko Oya Auth

https://ebookbell.com/product/applied-geomorphology-for-mitigation-of-
natural-hazards-1st-edition-masahiko-oya-auth-4626114

Applied Analysis Mathematical Methods In Natural Science Takasi Senba

https://ebookbell.com/product/applied-analysis-mathematical-methods-
in-natural-science-takasi-senba-997504

Applied Methods For Agriculture And Natural Resource Management A

Festschrift In Honor Of Richard E Howitt 1st Ed Siwa Msangi

https://ebookbell.com/product/applied-methods-for-agriculture-and-
natural-resource-management-a-festschrift-in-honor-of-richard-e-
howitt-1st-ed-siwa-msangi-10486152
Applied Natural
Language Processing
with Python
Implementing Machine Learning
and Deep Learning Algorithms for
Natural Language Processing
—
Taweh Beysolow II
 Applied Natural
Language Processing
with Python
Implementing Machine
Learning and Deep Learning
Algorithms for Natural
Language Processing

Taweh Beysolow II
Applied Natural Language Processing with Python
Taweh Beysolow II
San Francisco, California, USA

ISBN-13 (pbk): 978-1-4842-3732-8     ISBN-13 (electronic): 978-1-4842-3733-5

https://doi.org/10.1007/978-1-4842-3733-5
Library of Congress Control Number: 2018956300
Copyright © 2018 by Taweh Beysolow II
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or
part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way,
and transmission or information storage and retrieval, electronic adaptation, computer software,
or by similar or dissimilar methodology now known or hereafter developed.
Trademarked names, logos, and images may appear in this book. Rather than use a trademark
symbol with every occurrence of a trademarked name, logo, or image we use the names, logos,
and images only in an editorial fashion and to the benefit of the trademark owner, with no
intention of infringement of the trademark.
The use in this publication of trade names, trademarks, service marks, and similar terms, even if
they are not identified as such, is not to be taken as an expression of opinion as to whether or not
they are subject to proprietary rights.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal
responsibility for any errors or omissions that may be made. The publisher makes no warranty,
express or implied, with respect to the material contained herein.
Managing Director, Apress Media LLC: Welmoed Spahr
Acquisitions Editor: Celestin Suresh John
Development Editor: Siddhi Chavan
Coordinating Editor: Divya Modi
Cover designed by eStudioCalamar
Cover image designed by Freepik (www.freepik.com)
Distributed to the book trade worldwide by Springer Science+Business Media New York,
233 Spring Street, 6th Floor, New York, NY 10013. Phone 1-800-SPRINGER, fax (201) 348-4505,
e-mail [email protected], or visit www.springeronline.com. Apress Media, LLC is a
California LLC and the sole member (owner) is Springer Science + Business Media Finance Inc
(SSBM Finance Inc). SSBM Finance Inc is a Delaware corporation.
For information on translations, please e-mail [email protected], or visit http://www.apress.
com/rights-permissions.
Apress titles may be purchased in bulk for academic, corporate, or promotional use. eBook
versions and licenses are also available for most titles. For more information, reference our Print
and eBook Bulk Sales web page at http://www.apress.com/bulk-sales.
Any source code or other supplementary material referenced by the author in this book is
available to readers on GitHub via the book’s product page, located at www.apress.com/
978-1-4842-3732-8. For more detailed information, please visit http://www.apress.com/
source-code.
Printed on acid-free paper
To my family, friends, and colleagues for their continued
support and encouragement to do more with myself than
I often can conceive of doing
Table of Contents
About the Author���������������������������������������������������������������������������������ix

About the Technical Reviewer�������������������������������������������������������������xi

Acknowledgments�����������������������������������������������������������������������������xiii
Introduction����������������������������������������������������������������������������������������xv

Chapter 1: What Is Natural Language Processing?������������������������������1

The History of Natural Language Processing��������������������������������������������������������2
A Review of Machine Learning and Deep Learning����������������������������������������������4
NLP, Machine Learning, and Deep Learning Packages with Python����������������4
Applications of Deep Learning to NLP�����������������������������������������������������������10
Summary������������������������������������������������������������������������������������������������������������12

Chapter 2: Review of Deep Learning��������������������������������������������������13

Multilayer Perceptrons and Recurrent Neural Networks������������������������������������13
Toy Example 1: Modeling Stock Returns with the MLP Model�����������������������15
Vanishing Gradients and Why ReLU Helps to Prevent Them��������������������������27
Loss Functions and Backpropagation������������������������������������������������������������29
Recurrent Neural Networks and Long Short-Term Memory��������������������������30
Toy Example 2: Modeling Stock Returns with the RNN Model�����������������������32
Toy Example 3: Modeling Stock Returns with the LSTM Model���������������������40
Summary������������������������������������������������������������������������������������������������������������41

v
Table of Contents

Chapter 3: Working with Raw Text����������������������������������������������������43

Tokenization and Stop Words������������������������������������������������������������������������������44
The Bag-of-Words Model (BoW)��������������������������������������������������������������������������50
CountVectorizer���������������������������������������������������������������������������������������������51
Example Problem 1: Spam Detection������������������������������������������������������������53
Term Frequency Inverse Document Frequency���������������������������������������������57
Example Problem 2: Classifying Movie Reviews�������������������������������������������62
Summary������������������������������������������������������������������������������������������������������������74

Chapter 4: Topic Modeling and Word Embeddings����������������������������77

Topic Model and Latent Dirichlet Allocation (LDA)����������������������������������������������77
Topic Modeling with LDA on Movie Review Data�������������������������������������������81
Non-Negative Matrix Factorization (NMF)�����������������������������������������������������������86
Word2Vec������������������������������������������������������������������������������������������������������������90
Example Problem 4.2: Training a Word Embedding (Skip-Gram)�������������������94
Continuous Bag-of-Words (CBoW)��������������������������������������������������������������������103
Example Problem 4.2: Training a Word Embedding (CBoW)�������������������������105
Global Vectors for Word Representation (GloVe)�����������������������������������������������106
Example Problem 4.4: Using Trained Word Embeddings with LSTMs����������111
Paragraph2Vec: Distributed Memory of Paragraph Vectors (PV-DM)����������������115
Example Problem 4.5: Paragraph2Vec Example with Movie
Review Data������������������������������������������������������������������������������������������������116
Summary����������������������������������������������������������������������������������������������������������118

Chapter 5: Text Generation, Machine Translation, and Other

Recurrent Language Modeling Tasks������������������������������121
Text Generation with LSTMs�����������������������������������������������������������������������������122
Bidirectional RNNs (BRNN)��������������������������������������������������������������������������126

vi
 Table of Contents

Creating a Name Entity Recognition Tagger������������������������������������������������������128

Sequence-to-Sequence Models (Seq2Seq)������������������������������������������������������133
Question and Answer with Neural Network Models������������������������������������������134
Summary����������������������������������������������������������������������������������������������������������141
Conclusion and Final Statements���������������������������������������������������������������������142

Index�������������������������������������������������������������������������������������������������145

vii
About the Author
Taweh Beysolow II is a data scientist and
author currently based in San Francisco,
California. He has a bachelor’s degree in
economics from St. Johns University and a
master’s degree in applied statistics from
Fordham University. His professional
experience has included working at Booz
Allen Hamilton, as a consultant and in various
startups as a data scientist, specifically
focusing on machine learning. He has applied machine learning to federal
consulting, financial services, and agricultural sectors.

ix
About the Technical Reviewer
Santanu Pattanayak currently works at GE
Digital as a staff data scientist and is the author
of the deep learning book Pro Deep Learning
with TensorFlow: A Mathematical Approach
to Advanced Artificial Intelligence in Python
(Apress, 2017). He has more than eight years of
experience in the data analytics/data science
field and a background in development and
database technologies. Prior to joining GE,
Santanu worked at companies such as RBS,
Capgemini, and IBM. He graduated with a degree in electrical engineering
from Jadavpur University, Kolkata, and is an avid math enthusiast. Santanu
is currently pursuing a master’s degree in data science from the Indian
Institute of Technology (IIT), Hyderabad. He also devotes his time to data
science hackathons and Kaggle competitions, where he ranks within the
top 500 across the globe. Santanu was born and brought up in West Bengal,
India, and currently resides in Bangalore, India, with his wife.

xi
Acknowledgments
A special thanks to Santanu Pattanayak, Divya Modi, Celestin Suresh
John, and everyone at Apress for the wonderful experience. It has been a
pleasure to work with you all on this text. I couldn’t have asked for a better
team.

xiii
Introduction
Thank you for choosing Applied Natural Language Processing with Python
for your journey into natural language processing (NLP). Readers should
be aware that this text should not be considered a comprehensive study
of machine learning, deep learning, or computer programming. As such,
it is assumed that you are familiar with these techniques to some degree.
Regardless, a brief review of the concepts necessary to understand the
tasks that you will perform in the book is provided.
After the brief review, we begin by examining how to work with raw
text data, slowly working our way through how to present data to machine
learning and deep learning algorithms. After you are familiar with some
basic preprocessing algorithms, we will make our way into some of the
more advanced NLP tasks, such as training and working with trained
word embeddings, spell-check, text generation, and question-and-answer
generation.
All of the examples utilize the Python programming language and
popular deep learning and machine learning frameworks, such as scikit-­
learn, Keras, and TensorFlow. Readers can feel free to access the source
code utilized in this book on the corresponding GitHub page and/or try
their own methods for solving the various problems tackled in this book
with the datasets provided.

xv
CHAPTER 1

What Is Natural
Language
Processing?
Deep learning and machine learning continues to proliferate throughout
various industries, and has revolutionized the topic that I wish to discuss
in this book: natural language processing (NLP). NLP is a subfield of
computer science that is focused on allowing computers to understand
language in a “natural” way, as humans do. Typically, this would refer to
tasks such as understanding the sentiment of text, speech recognition, and
generating responses to questions.
NLP has become a rapidly evolving field, and one whose applications
have represented a large portion of artificial intelligence (AI)
breakthroughs. Some examples of implementations using deep learning
are chatbots that handle customer service requests, auto-spellcheck on cell
phones, and AI assistants, such as Cortana and Siri, on smartphones. For
those who have experience in machine learning and deep learning, natural
language processing is one of the most exciting areas for individuals to
apply their skills. To provide context for broader discussions, however, let’s
discuss the development of natural language processing as a field.

© Taweh Beysolow II 2018 1

T. Beysolow II, Applied Natural Language Processing with Python,
https://doi.org/10.1007/978-1-4842-3733-5_1
Chapter 1 What Is Natural Language Processing?

The History of Natural Language Processing

Natural language processing can be classified as a subset of the broader
field of speech and language processing. Because of this, NLP shares
similarities with parallel disciplines such as computational linguistics,
which is concerned with modeling language using rule-based models.
NLP’s inception can be traced back to the development of computer science
in the 1940s, moving forward along with advances in linguistics that led to
the construction of formal language theory. Briefly, formal language theory
models language on increasingly complex structures and rules to these
structures. For example, the alphabet is the simplest structure, in that it is
a collection of letters that can form strings called words. A formal language
is one that has a regular, context-free, and formal grammar. In addition to
the development of computer sciences as a whole, artificial intelligence’s
advancements also played a role in our continuing understanding of NLP.
In some sense, the single-layer perceptron (SLP) is considered to be the
inception of machine learning/AI. Figure 1-1 shows a photo of this model.

Figure 1-1. Single-layer perceptron

The SLP was designed by neurophysiologist Warren McCulloch and

logician Walter Pitt. It is the foundation of more advanced neural network
models that are heavily utilized today, such as multilayer perceptrons.

2
 Chapter 1 What Is Natural Language Processing?

The SLP model is seen to be in part due to Alan Turing’s research in the
late 1930s on computation, which inspired other scientists and researchers
to develop different concepts, such as formal language theory.
Moving forward to the second half of the twentieth century, NLP starts
to bifurcate into two distinct groups of thought: (1) those who support a
symbolic approach to language modelling, and (2) those who support a
stochastic approach. The former group was populated largely by linguists
who used simple algorithms to solve NLP problems, often utilizing pattern
recognition. The latter group was primarily composed of statisticians
and electrical engineers. Among the many approaches that were popular
with the second group was Bayesian statistics. As the twentieth century
progressed, NLP broadened as a field, including natural language
understanding (NLU) to the problem space (allowing computers to react
accurately to commands). For example, if someone spoke to a chatbot and
asked it to “find food near me,” the chatbot would use NLU to translate this
sentence into tangible actions to yield a desirable outcome.
Skip closer to the present day, and we find that NLP has experienced
a surge of interest alongside machine learning’s explosion in usage over
the past 20 years. Part of this is due to the fact that large repositories of
labeled data sets have become more available, in addition to an increase in
computing power. This increase in computing power is largely attributed
to the development of GPUs; nonetheless, it has proven vital to AI’s
development as a field. Accordingly, demand for materials to instruct
data scientists and engineers on how to utilize various AI algorithms has
increased, in part the reason for this book.
Now that you are aware of the history of NLP as it relates to the present
day, I will give a brief overview of what you should expect to learn. The
focus, however, is primarily to discuss how deep learning has impacted
NLP, and how to utilize deep learning and machine learning techniques to
solve NLP problems.

3
Chapter 1 What Is Natural Language Processing?

 Review of Machine Learning and Deep

A
Learning
You will be refreshed on important machine learning concepts,
particularly deep learning models such as multilayer perceptrons (MLPs),
recurrent neural networks (RNNs), and long short-term memory (LSTM)
networks. You will be shown in-depth models utilizing toy examples before
you tackle any specific NLP problems.

 LP, Machine Learning, and Deep Learning

N
Packages with Python
Equally important as understanding NLP theory is the ability to apply it in
a practical context. This book utilizes the Python programming language,
as well as packages written in Python. Python has become the lingua
franca for data scientists, and support of NLP, machine learning, and
deep learning libraries is plentiful. I refer to many of these packages when
solving the example problems and discussing general concepts.
It is assumed that all readers of this book have a general understanding
of Python, such that you have the ability to write software in this language.
If you are not familiar with this language, but you are familiar with others,
the concepts in this book will be portable with respect to the methodology
used to solve problems, given the same data sets. Be that as it may, this
book is not intended to instruct users on Python. Now, let’s discuss some of
the technologies that are most important to understanding deep learning.

TensorFlow
One of the groundbreaking releases in open source software, in addition
to machine learning at large, has undoubtedly been Google’s TensorFlow.
It is an open source library for deep learning that is a successor to Theano,
a similar machine learning library. Both utilize data flow graphs for

4
 Chapter 1 What Is Natural Language Processing?

computational processes. Specifically, we can think of computations as

dependent on specific individual operations. TensorFlow functionally
operates by the user first defining a graph/model, which is then operated
by a TensorFlow session that the user also creates.
The reasoning behind using a data flow graph rather than another
computational format computation is multifaceted, however one of the
more simple benefits is the ability to port models from one language to
another. Figure 1-2 illustrates a data flow graph.

Graph of Nodes, also called operations (ops)

biases

weights Add Softmax

MatMul Xent
inputs

targets

Figure 1-2. Data flow graph diagram

For example, you may be working on a project where Java is the

language that is most optimal for production software due to latency
reasons (high-frequency trading, for example); however, you would like to
utilize a neural network to make predictions in your production system.
Rather than dealing with the time-consuming task of setting up a training
framework in Java for TensorFlow graphs, something could be written in
Python relatively quickly, and then the graph/model could be restored by
loading the weights in the production system by utilizing Java. TensorFlow
code is similar to Theano code, as follows.

    #Creating weights and biases dictionaries

    weights = {'input': tf.Variable(tf.random_normal([state_
size+1, state_size])),

5
Exploring the Variety of Random
Documents with Different Content
The Project Gutenberg eBook of Antique
Works of Art from Benin
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.

Title: Antique Works of Art from Benin

Author: Augustus Henry Lane-Fox Pitt-Rivers

Release date: October 22, 2013 [eBook #44014]

Most recently updated: October 23, 2024

Language: English

Credits: E-text prepared by Henry Flower and the Online Distributed

Proofreading Team (http://www.pgdp.net) from page
images generously made available by Internet
Archive/Canadian Libraries
(http://archive.org/details/toronto)

*** START OF THE PROJECT GUTENBERG EBOOK ANTIQUE WORKS

OF ART FROM BENIN ***
The Project Gutenberg eBook, Antique Works of Art from Benin, by
Augustus Henry Lane-Fox Pitt-Rivers

Note: Images of the original pages are available through Internet

Archive/Canadian Libraries. See
http://archive.org/details/antiqueworksofar00pittuoft

Many of the illustrations can be enlarged by clicking on

them.

ANTIQUE WORKS OF ART

FROM

BENIN,
 COLLECTED BY
LIEUTENANT-GENERAL PITT RIVERS,
D.C.L., F.R.S., F.S.A.
Inspector of Ancient Monuments in Great Britain, &c.

PRINTED PRIVATELY.
1900.
 LONDON:
HARRISON AND SONS, PRINTERS IN ORDINARY TO HER MAJESTY,
ST. MARTIN’S LANE, W.C.
 WORKS OF ART FROM BENIN,
WEST AFRICA.

OBTAINED BY THE PUNITIVE EXPEDITION IN 1897, AND NOW IN

GENERAL PITT RIVERS’S MUSEUM AT FARNHAM, DORSET.
Benin is situated on the Guinea Coast, near the mouth of the Niger,
in latitude 6·12 north, and longitude 5 to 6 east.
It was discovered by the Portuguese at the end of the fourteenth or
commencement of the fifteenth centuries. The Portuguese were
followed by the Dutch and Swedes, and in 1553 the first English
expedition arrived on the coast, and established a trade with the
king, who received them willingly.
Benin at that time appears by a Dutch narrative to have been quite a
large city, surrounded by a high wall, and having a broad street
through the centre. The people were comparatively civilized. The
king possessed a number of horses which have long since
disappeared and become unknown. Faulkner, in 1825, saw three
solitary horses belonging to the king, which he says no one was bold
enough to ride.
In 1702 a Dutchman, named Nyendaeel, describes the city, and
speaks of the human sacrifices there. He says that the people were
great makers of ornamental brass work in his day, which they seem
to have learnt from the Portuguese. It was visited by Sir Richard
Burton, who went there to try to put a stop to human sacrifices, at
the time he was consul at Fernando Po. In 1892 it was visited by
Captain H. L. Galloway, who speaks of the city as possessing only
the ruins of its former greatness; the abolition of the slave trade had
put a stop to the prosperity of the place, and the king had prohibited
any intercourse with Europeans. The town had been reduced to a
collection of huts, and its trade had dwindled down to almost nil.
The houses have a sort of impluvium in the centre of the rooms,
which has led some to suppose that their style of architecture may
have been derived from the Roman colonies of North Africa.
In 1896 an expedition, consisting of some 250 men, with presents
and merchandise, left the British settlements on the coast, and
endeavoured to advance towards Benin city. The expedition was
conducted with courage and perseverance, but with the utmost
rashness. Almost unarmed, neglecting all ordinary precautions,
contrary to the advice of the neighbouring chiefs, and with the
express prohibition of the King of Benin to advance, they marched
straight into an ambuscade which had been prepared for them in the
forest on each side of the road, and as their revolvers were locked
up in their boxes at the time, they were massacred to a man with
the exception of two, Captain Boisragon and Mr. Locke, who, after
suffering the utmost hardships, escaped to the British settlements on
the coast to tell the tale.
Within five weeks after the occurrence, a punitive expedition entered
Benin, on 18th January, 1897, and took the town. The king fled, but
was afterwards brought back and made to humiliate himself before
his conquerers, and his territory annexed to the British crown.
The city was found in a terrible state of bloodshed and disorder,
saturated with the blood of human sacrifices offered up to their Juju,
or religious rites and customs, for which the place had long been
recognised as the “city of blood.”
What may be hereafter the advantages to trade resulting from this
expedition it is difficult to say, but the point of chief interest in
connection with the subject of this paper was the discovery, mostly
in the king’s compound and the Juju houses, of numerous works of
art in brass, bronze, and ivory, which, as before stated, were
mentioned by the Dutchman, Van Nyendaeel, as having been
constructed by the people of Benin in 1700.
These antiquities were brought away by the members of the punitive
expedition and sold in London and elsewhere. Little or no account of
them could be given by the natives, and as the expedition was as
usual unaccompanied by any scientific explorer charged with the
duty of making inquiries upon matters of historic and antiquarian
interest, no reliable information about them could be obtained. They
were found buried and covered with blood, some of them having
been used amongst the apparatus of their Juju sacrifices.
A good collection of these antiquities, through the agency of Mr.
Charles Read, F.S.A., has found its way into the British Museum;
others no doubt have fallen into the hands of persons whose chief
interest in them has been as relics of a sensational and bloody
episode, but their real value consists in their representing a phase of
art—and rather an advanced stage—of which there is no actual
record, although no doubt we cannot be far wrong in attributing it to
European influence, probably that of the Portuguese some time in
the sixteenth century.
A. P. R.
Rushmore, Salisbury,
April, 1900.
 DESCRIPTION OF PLATE I.

Fig. 1.—Bronze plaque, representing two warriors with broad leaf-

shaped swords in their right hands. Coral or agate head-dress. Coral
chokers, badge of rank. Leopards’ teeth necklace. Coral scarf across
shoulder. Leopards’ heads hanging on left sides. Skirts each
ornamented with a human head. Armlets, anklets, etc. Ground
ornamented with the usual foil ornament incised.
Fig. 2.—Bronze plaque, representing two figures holding plaques or
books in front. Coral chokers, badge of rank. Reticulated head-
dresses of coral or agate, similar to that represented in Plate XXI,
Fig. 121. Barbed objects of unknown use behind left shoulders,
ornamented with straight line diaper pattern. Ground ornamented
with foil ornaments incised. Guilloche on sides of plaque.
Fig. 3.—Bronze plaque, representing three warriors, two with
feathers in head-dress and trefoil leaves at top; one with pot helmet,
button on top. The latter has a coral choker, badge of rank, and all
have leopards’ teeth necklaces. The central figure has a cylindrical
case on shoulder. Two have hands on their sword-hilts. All three
have leopards’ heads on breast, and quadrangular bells hanging
from neck. Leopards’ skins and other objects hang on left sides.
Ground ornamented with foil ornaments incised.
Fig. 4.—Bronze plaque, figure of warrior with spear in right hand,
shield on left shoulder. Head-dress of coral or agate, similar to that
represented in Plate XXI, Fig. 121. Quadrangular bell hanging from
neck. Chain-like anklets. Coral choker, badge of rank, and leopards’
teeth necklace. A nude attendant on right upholds a large broad
leaf-shaped sword, with a ring attached to pommel. Another holds
two sistri or bells fastened together by a chain. Small figure on left is
blowing an elephant’s tusk trumpet. Figures above in profile are
holding up tablets or books. The dress of one of them is fastened
with tags or loops of unusual form. These figures have Roman
noses, and are evidently not negro. Ground ornamented with the
usual foil ornament incised.
 DESCRIPTION OF PLATE II.

Figs. 5 and 6.—Bronze plaque, representing a warrior in centre,

turned to his left. He has a beard and a necklace of leopards’ teeth,
but no coral choker. He has a high helmet, somewhat in the form of
a grenadier cap. Quadrangular bell on neck. Dagger in sheath on
right side, and various appurtenances hanging from his dress. He
holds a narrow leaf-shaped sword in his right hand over an enemy
who has fallen, and who has already a leaf-shaped sword thrust
through his body. The victim has a sword-sheath on left side, with
broad end, and a peculiar head-dress. His horse is represented
below with an attendant holding it by a chain and carrying barbed
darts in his left hand. On the right of the conqueror is a small figure
blowing a tusk trumpet, and on his right a larger figure carrying a
shield in his left hand and a cluster of weapons. He has a high
helmet, ornamented with representations of cowrie shells of nearly
the same form as that of the central figure. Above are two figures,
one blowing what appears to be a musical instrument and the other
carrying a barbed pointed implement, and armed with a sword in
sheath similar to that of the fallen warrior. The plaque appears to
represent a victory of some kind, and all the conquerors have the
same high helmet. The ground is ornamented with the usual foil
ornament incised.

Figs. 7 and 8.—Bronze plaque, representing a king or noble on

horseback sitting sideways, his hands upheld by attendants, one of
whom has a long thin sword in his hand in sheath. Two attendants,
with helmets or hair represented by ribs, are holding up shields to
shelter the king from the sun. The king or noble has a coral choker,
badge of rank, with a coral necklace hanging on breast. Horse’s
head-collar hung with crotals. A small attendant carries a “manilla”
in his hand. The two figures above are armed with bows and arrows.
Ground ornamented with foil ornaments incised.
De Bry, “India Orientalis,” says that in the sixteenth
century both the king and chiefs were wont to ride side-
saddle upon led horses. They were supported by
retainers, who held over their heads either shields or
umbrellas, and accompanied by a band of musicians
playing on ivory horns, gong-gongs, drums, harps, and a
kind of rattle.
 DESCRIPTION OF PLATE III.

Fig. 9.—Bronze plaque, naked figure of boy; hair in conventional

bands; three tribal marks over each eye and band on forehead. Coral
choker, badge of rank. Armlets and anklets. Four rosettes on ground
and usual foil ornaments. De Bry says that all young people went
naked until marriage.
Fig. 10.—Bronze plaque, figure of warrior with helmet or hair
represented by ribs. Leaf-shaped sword upheld in right hand. A
bundle of objects on head upheld by left hand. Object resembling a
despatch case on left side, fastened by a belt over right shoulder.
Human mask on left side. Four fishes on ground, and the usual foil
ornaments incised.
Figs. 11 and 12.—Bronze plaque, representing a figure holding a
ball, perhaps a cannon ball, in front. Coral choker, badge of rank.
Three tribal marks over each eye. Crest on head-dress, feather in
cap. Skirt wound up behind left shoulder. Skirt ornamented with a
head and hands. Four rosettes on ground, and usual foil ornaments
incised. Guilloche on sides of plaque.
 DESCRIPTION OF PLATE IV.

Fig. 13.—Bronze plaque, figure of warrior, feather in cap; broad leaf-

shaped sword in right hand. Coral choker, badge of rank. Leopards’
teeth necklace. Coral sash; ground ornamented with leaf-shaped foil,
ornaments incised.
Figs. 14 and 15.—Bronze ægis or plaque, with representations of
two figures with staves in their right hands. Coral chokers, badge of
rank. On the breasts are two Maltese crosses hanging from the
necks, which appear to be European Orders. The objects held in left
hands have been broken off. The hats are similar to that on the head
of the figure, Fig. 91, Plate XV. Ground ornamented with the usual
foil ornaments incised.
Fig. 16.—Bronze plaque, figure of warrior with pot helmet, button on
top. Coral choker, badge of rank, on neck. Leopards’ teeth necklace.
Quadrangular bell on breast. Armlets, anklets, &c. Four rosettes on
ground, and the usual foil ornaments incised.
Fig. 17.—Bronze plaque, figure of warrior with spear in right hand,
shield in left hand; pot helmet, button on top. Quadrangular bell
hanging from neck. Coral choker, badge of rank. Leopards’ teeth
necklace. Leopard’s skin dress with head to front. On the ground are
two horses’ heads below and two rosettes above. Ground
ornamented with the usual foil ornaments incised.
Fig. 18.—Bronze plaque, figure of warrior. Peculiarly ornamented
head-dress. Coral choker, badge of rank. Leopards’ teeth necklace.
Broad leaf-shaped sword in right hand. Coral sash on breast.
Leopard’s mask hanging on left side. Armlets, anklets, &c. Small
figure of boy, naked, to right, holding a metal dish with lid in form of
an ox’s head. A similar object may be seen amongst the Benin
objects in the British Museum.
 DESCRIPTION OF PLATE V.

Figs. 19, 20 and 21.—Stained ivory carving of figure on horse. Coral

choker; spear in right hand, the shaft broken. Tribal marks on
forehead incised. Chain-bridle or head-collar. Degenerate guilloche
pattern on base. Straight line diaper pattern represented in various
parts. The stand formed as a socket for a pole.
Figs. 22, 23 and 24.—Ivory carving of figure on horse, with spear in
right hand and bell on neck, and long hair. The bridle formed as a
head-collar. Degenerate guilloche pattern on base. The stand formed
as a socket for a pole ornamented with bands of interlaced pattern
and the head of an animal.
 DESCRIPTION OF PLATE VI.

Figs. 25 and 26.—Ivory carving of a human face. Eyes and bands on

forehead inlaid. Straight line diaper pattern on head-dress, above
which are conventionalised mud-fish. Four bands of coral across
forehead. Ears long and narrow. Found hidden in an oaken chest
inside the sleeping apartment of King Duboar.
Fig. 27.—Carved wooden panel, consisting of a chief in the centre;
broad leaf-shaped sword, with ring attached to pommel, upheld in
right hand, studded with copper nails, and ornamented with
representations of itself. In left hand a fan-shaped figure terminating
in two hands. Coral choker, badge of rank. Bell on neck and cross-
belts. Skirt ornamented with three heads and a guilloche pattern of
three bands with pellets. Anklets. Attendant on left holding umbrella
over chief’s head. Serpent with human arm and hand in its mouth,
head upwards; eyes of inlaid glass; body studded with copper nails.
Leopard, drawn head upwards. On right, figure with jug in left hand
and cup in right hand, standing in a trough or open vessel. Small
attendant with paddle in right hand. At top a bottle bound with
grass, and figure of some object, perhaps a stone celt bound with
grass. Brass and iron screws are used for ornamentation in this
carving. Guilloche pattern of two bands without pellets around the
edge of the panel.
Figs. 28, 29 and 30.—Ivory carved tusk, 4 feet 1 inch long from
bottom to point; traversed by five bands of interlaced strap-work.
The other ornamentation consists of:—Human figures with hands
crossed on breast; bird standing on pedestal; human figures with
hands holding sashes; trees growing downwards; a rosette; mudfish;
crocodiles with heads upwards; a serpent with sinuous body, head
downwards; two cups; a serpent, head upwards; detached human
heads. Some of the representations are so rude that it requires
experience to understand their meaning. On this tusk the interlaced
pattern is the prevailing ornament, and it passes into the guilloche
pattern. This tusk is more tastefully decorated than the other tusk,
Figs. 167 and 168, Plate XXVI, but with less variety in the carving.
These carved tusks are said to represent gods in the Ju-ju houses.
Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.

More than just a book-buying platform, we strive to be a bridge

connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.

Join us on a journey of knowledge exploration, passion nurturing, and

personal growth every day!

ebookbell.com