Software feasibility study to

International Journal of Database Management Systems ( IJDMS ) Vol.7, No.4, August 2015
DOI : 10.5121/ijdms.2015.7401 1
SOFTWARE FEASIBILITY STUDY TO
TRANSFORM COMPLEX SCIENTIFIC WRITTEN
KNOWLEDGE TO A CLEAR, RATIONALE AND
SIMPLE LANGUAGE
Manoj Khandelwal1
and Mohammad Jafarabad2
1
Faculty of Science and Technology, Federation University Australia, PO Box 663,
Ballarat, Victoria 3353, Australia
Phone: +61 3 5327 9821
2
Department of Computer Engineering, Young Researchers Club, Robatkarim Branch,
Islamic Azad University, RobatKarim, Iran.
ABSTRACT
Ali Akbar Dehkhoda, the prominent lexicographer, describes a person who has difficulty in grasping
knowledge as someone who “Cannot understand something without knowing all its details.” If the
knowledge required by somebody is in a language other than the person’s mother tongue, access to this
knowledge will surely meet special difficulties resulting from the person’s lack of mastery over the second
language. Any project that can monitor knowledge sources written in English and change them into the
user’s language by employing a simple understandable model is capable of being a knowledge-based
project with a world view regarding text simplification. This article creates a knowledge system,
investigates some algorithms for analyzing contents of complex texts, and presents solutions for changing
such texts simple and understandable ones. Texts are automatically analyzed and their ambiguous points
are identified by software, but it is the author or the human agent who makes decisions concerning
omission of the ambiguities or correction of the texts.
KEYWORDS
Written knowledge, complex texts, text simplification, ambiguous words
1. INTRODUCTION
Knowledge that is implemented and published in written form is created correspondent to the
literacy level of its author. Previous studies, the extent of proficiency in writing techniques, and
the viewpoints of the author of any knowledge, play an important role in the process of designing
and implementing documents related to the knowledge. In most cases, before written knowledge
(created by an expert in the related industry or a scientist) is published, it passes through a
process of correcting the writing of or the scientific errors in the text by the editorial board of the
journal, conference, or publications. Moreover, the educational level or the areas of study the
readers of the texts must have sometimes specified in the classification of some forms of

2
knowledge. The problems that will then arise for the readers are ambiguities and complexities in
the texts that make them difficult to understand.
Understanding knowledge written by a specialist is not an easy task for beginners because they
will need great mental activity and intelligence to adapt their level of knowledge to that of the
specialists. Difficult expressions or complex sentences containing unfamiliar words can prevent
grasp of the written knowledge or lead to misunderstanding it. Special arrangements are required
to prevent this. Moreover, if knowledge is to be written in a way as to follow all the details,
conventions, methods, and rules of formal writing, it will entail great costs for the organization
publishing this kind of knowledge. Therefore, projects for disambiguation of knowledge
containing texts must be executed in a way that most of the ambiguities present in these texts are
omitted at minimal cost.
Authors, scholars, and researchers present their personal knowledge through theoretical-applied
research articles in publications related to their field of study, or in book form, to be used by the
public. Disambiguation of educational texts can be performed through various processes.
Knowledge managers and users, or intelligent software, can discover ambiguities in texts. The
creator of the knowledge can also carry out correction of text ambiguities and complexities at
minimal cost. In this article, the iterative and augmentative method of software production is
studied and analyzed to execute the project for identifying ambiguities and literary complexities
in knowledge containing texts. The project will make it possible to implement software that can
identity ambiguous words or sentences and, in addition, play a considerable role in removing
these ambiguities through making suitable recommendations to the author of the knowledge
containing text.
2. BACKGROUND OF TEXT SIMPLIFICATION
Language is one of the most important of mental phenomena and has been in use for many years
among people living in various regions. At present, there are about 7000 languages in the world,
2400 of these are on the verge of extinction, and 840 are used only in New Guinea. Linguists
investigate languages in the domains of syntax, phonology, morphology, and semantics. Syntax
deals with sentence recognition and studies the rules by which words are combined and placed
next to each other. Phonology is concerned with the positions of chain and sub-chain phonetic
elements in the language system. Morphology is a part of language grammar that analyzes the
structure of words by recognizing their morphemes), while semantics studies meanings in human
languages.
The famous linguist Darrow has said, “There is a completely explicit, clear, and standard-based
relationship with great importance between language and its content, and knowledge, cognition,
and language are the most important constituents of learning” [9]. The necessary condition for
reducing the time it takes to understand the content of a written text, and to increase the speed of
learning, is for the text to be fluent, simple, and without ambiguities. If a text is written fluently,
understanding the text and correct thinking about the subject happen simultaneously with reading
it. If a text is complicated and difficult to grasp, only partial thinking about the subject takes place
during the reading, and the reader completes the thinking process by asking questions. The effort
required for finding answers to these questions makes the reader less inclined to receive the
knowledge (and this is inconsistent with the principles of knowledge management)

3
Due to the complexity of language structure, most text processing systems do not achieve
satisfactory success when put to use. Google Website Translation is among the most successful
text processing systems. Another site active in content analysis is the Online Summarization Site.
These sites apply their proprietary algorithms to analyze texts. So far, no successful online or
offline software has been successfully applied to turn complex texts into simple ones. However, a
series of research has been conducted by a team of researchers at Lexile Company that is
presented in the form of Lexile Framework.
Lexile offers a standard called LF that evaluates the difficulty of reading texts and the capacity of
individuals in reading them. The Lexile Level ranges from 0 to 1800, and each book is assigned
one of the numbers in this range. The difficulty of reading texts is measured based on a series of
mathematical formulae. Results of these calculations are obtained based on two features:
semantic difficulty based on words of the same family with respect to frequency of use in the
text, and syntactic complexity based on sentence length [2].
After the evaluations, the LF assigns a value to the difficulty of the text (Table 1). Based on the
Lexile indices, and considering the age and education level of each individual, it will be possible
to determine the capacity of each person in reading the text.
Table 1: Lexile standard
Rank Standard of the
difficulty
of texts
Standard of the
reading capacity of
individuals
1 200-400 Up to 300
2 300-500 140-500
3 500-700 330-700
4 650- 850 445-810
5 750-950 565-910
6 850-1050 665-1000
The LF recognizes the complexity of texts but has two main shortcomings. First, the whole text is
not processed, only one or a few sample paragraphs from the text are analyzed and, based on the
analysis, the difficulty of the whole text is shown by a score. Second, LF does not specify the
reasons for the difficulty of the text, while in the idea proposed by us the reasons for the difficulty
of the text must be specified for its author. Other research on text simplification has been carried
out as case studies and in the three areas of mathematics, computers, and literature. Among the
most important of them are various projects for identifying the role played by pronoun
antecedents in sentence complexity and the study of ambiguity in words that are spelled the same
but have different meanings.

4
3. STUDY OF THE IMPORTANCE AND THE GOAL OF SIMPLIFYING
TEXTS WRITTEN IN ENGLISH
English plays a basic and important role in the daily lives of people because it has become a
communication bridge between various societies and most international communications are
carried out in this language. Moreover, most reference books in various sciences are written in
English and specialists and students studying at various levels and in different fields need to have
sufficient knowledge of English to be able to use these references. Therefore, unfamiliarity with
this language causes problems in carrying out various duties, one of the most important of which
is the impossibility of receiving up-to-date and needed knowledge.
The main purpose in presenting this project of simplifying texts was to make reading texts in
English possible for those having university education but low levels of acquaintance with the
English language, although execution of this project can be useful, in many instances, for other
sectors of the society. Since one of the main mottos of the United Nations Organization
UNESCO is eradication of illiteracy in societies, simplification of knowledge containing texts
can also be considered one of the tools of realizing the high goals of this Organization. In this
section, two main classes of people are studied as target groups for easy reading of science
containing texts.
3.1. Those with low capacity in language learning
Research has shown more than 25% of the population in the Unites States does not acquire the
expected level of reading ability after 9 years of formal education. This figure reaches 45-50% in
several other countries. Illiteracy and inability to apply the basic vocabulary and comprehension
skills learned in reading techniques may be due to various factors such as insufficient education,
non-standard education, social problems, mild mental retardation, etc. People with such problems
are not able to read and understand applied texts available in advanced societies.
The goal in this article is to prepare the ground for faster comprehension of scientific texts by
people with university education. As expected, a native speaker of English with this level of
education is capable of sufficiently understanding such texts, but will certainly face difficulties
when reading a text in which the standard skills of the English language have not been observed.
The reason for this is that having academic education and being native speakers are not sufficient
conditions for understanding every book written in a special field. Experience and continuous
studies are other required conditions for comprehending texts, and they will help readers to read
complex texts.
3.2. Non-native speakers
Any language learned after the first one (the mother tongue) is called “second language.” Eric
Linenberg, a pioneering linguist and neurologist, called second language a language consciously
learned or spoken by a person after maturity. In most cases, people never reach a similar level of
fluency and comprehension in the second language as their mother tongues, and face problems in
finding correct words and grammatical structures. Learning a second language can be a life-long
process for many people. Psychologists have reached the conclusion the reason for this is that
there are separate centers in the brain for learning and for controlling skills in a second language.
It must also be noted that people weak in learning a new language are different from the mentally
retarded.

5
There are 1.3 billion people living in China and, in many cases, they access the required
knowledge only in sources that are in Chinese. Therefore, there is no great need to learn English
in China. Since English has been omitted from college entrance examinations, studying a
scientific text in English for graduates of the best universities in China is a difficult and costly
task [8]. Our proposed project to simplify scientific texts written in English, and to make them
understandable for people with low levels of proficiency in this language, can substantially help
this section of the world population.
Understanding written text for the blind is like learning a foreign language. These people have
limited understanding of textual metaphors. However, blind people can use easy-to-read materials
in which the Braille form and format is employed. A blind French person learns French Braille
and, hence, using easy-to-read materials helps this individual to read and understand texts written
in English Braille [1].
4. LEVELS OF SIMPLIFYING TEXTS
The need students and researchers have for reading specialized texts, and the various levels of
familiarity with the language, make it more difficult to set a defined level for simplifying texts.
There are various levels, classes, classifications, and models in each language. Considering the
level of language skills a person has depends on many factors such as age, occupation, level of
education, personal interests in the language, talent, etc., classification of people and mentioning
evidence for each various language level is difficult and, sometimes, seems to be impossible.
Therefore, examples are often not presented for each level. For example, it is not possible to say
what level of language proficiency a certain age group or all graduates of a given level of
education have reached. Therefore, complex study and research are needed to assign a level of
language proficiency to a group of people with one or more common characteristics [9].
Determining the level of English language skills in people is a very complex process. Various
models have been designed for classifying the levels of the English language. These models are
conventional and divided into the three classes of advanced, intermediate, and Beginner levels.
The advanced level is a high level of language proficiency in people possessing the capabilities of
the English language. People whose mother tongues are not English but have attained an
advanced level of proficiency in English and speak very much like native speakers are called near
native speakers. These people cannot be easily differentiated from native speakers.
People with the advanced level of skills in reading English texts can study and understand
English texts and do not need text simplification, although text simplification accelerates their
comprehension. In some cases, the presence of ambiguous and complex sentences in texts makes
their understanding difficult, even for native speakers. In these cases, only a linguistics specialist
can analyze these texts. Therefore, although text simplification will greatly help people with
intermediate or beginner level proficiency in the language, in many cases it will certainly solve
problems native speakers or people with advanced levels of proficiency in English face when
reading ambiguous or complex texts.
5. DESIGNING THE PROPOSED SOFTWARE SYSTEM
Considering various studies have been carried out for many years on correcting the grammars of
various languages, and especially that of English, with the goal of removing literary ambiguities,
all grammatical points of the language have been presented in thousands of books. Therefore, the

6
purpose of this project is not to carry out new linguistic studies, but rather previous studies are
analyzed to present an intelligent system for discovering the main factors contributing to text
complexity through changing the existing language rules into algorithms. Obviously, accurate
recognition of all factors influencing text complexity requires complex text processing systems
and, if facts regarding software and hardware infrastructure are not considered in the feasibility
study, problems will arise when implementing this project. Therefore, the whole idea behind this
project will be to use new algorithms within the scope of previous linguistic studies.
The proposed project includes producing software for analyzing texts and for finding ambiguities
in them. The software must read the text and then point out the ambiguities and complexities to
its author. The author can then edit the text and omit the challenging points by paying attention to
the recommendations made. This will help to turn the contents of books and articles into fluent
knowledge before their publication. Factors causing text complexity are now studied in the
following four groups:
5.1. Presence of a word in the text that has more than one common meaning
Presence of words with several meanings is directly related to text complexity. Depending on
their roles in sentences, some words take on special meanings. Sometimes, the presence of a
word with several meanings results in several interpretation of the texts. In daily conversations,
people naturally understand the meaning of a word with several meanings by paying attention to
the topic of the conversation. However, accurate understanding of cases where the presence of a
word with several meanings causes ambiguity and complexity is not easy in text processing by
computers [4]. Many applied programs have been developed for removing ambiguity from words
in written texts. Having insight into the domain of intertextuality substantially helps in
automatically recognizing whether a word is ambiguous for the reader or not.
In a single sentence, the meaning allocated to each word is usually determined by the two factors
of the syntactic format embedded in the word and the semantics of the words (syntactic
dependency) [3]. For example, in sentence (a) the word “absorb” must have a meaning close to
“financial affairs” because of its proximity to the word “cost” which is related to finance. On the
other hand, in sentence (b) it comes before the word “information” and, therefore, it can be
related to a meaning associated with the domain of “learning or of being informed”. In sentence
(c), the word “treated” is related to the medical field and has the meaning of “being cured,” while
in sentence (d) the same word is associated with the domain of human relations and has the
meaning of “behavior.”
a. The customer will absorb this cost. [Asset] (“pay”)
b. The customer will absorb this information. [Information] (“learn”)
c. Peter treated Mary with antibiotics. [with Medication] (“medical”)
d. Peter treated Mary with respect. [with Quality] (“human relations”)
The above sentences illustrate that determination of the domain related to each word can play a
major role in removing its ambiguity in the sentence. Of course, in this project we are only after
recognizing ambiguity and, therefore, if we find a word with several meanings in a sentence and
the domain close to this word is not clear, we need to specify the ambiguity present in the word.
The important point to be taken into account at this point is the frequency of the usage of each
meaning of the word in the text. That is, if a word has two meanings one of which is rarely used,
it should not be included among ambiguous words. However, if a phrase like “as well as” has two

7
different meanings that are used with relatively equal frequency, it may be ambiguous under
certain conditions [7].
5.2. Complexities related to pronouns
Presence of pronouns that are distanced from their antecedents in the text makes sentences
ambiguous. For example, if the antecedent of a pronoun comes a few sentences before the
pronoun, it will be very difficult for the reader to find the antecedent’s exact location. Moreover,
sometimes it is not clear in sentences exactly to what an adjective refers [5]. For example, in the
sentence “little girls and boys” it is not clear whether the writer meant “all little girls and boys” or
“little girls and little boys.” Many applied programs have been written for recognizing pronoun
antecedents in which accuracies of software performance regarding various kinds of pronouns
have been mentioned. The maximum accuracy is that of recognizing the antecedent for the
second person pronoun “you,” and the minimum those of recognizing the antecedents for the
pronouns “it” and “they.”
S. Lappin and M. McCord conducted a study only on personal, number, gender pronouns in
1990, and determined a governor for the sentence containing the pronouns [5]. For example, in
the sentence “I saw her,” the word “saw” is the governor for the words “I” and “her.” The
relationship between the pronouns in the sentence is then determined depending on the governing
literary type. In the section on pronoun recognition, it is impossible to analyze the data with
labeling it. Table 2 offers an example of classifying verbs. This classification has been employed
in data mining through thematic modeling.
5.3. Presence of several easy synonyms for a word
Simple synonyms and equivalents can be used in place of heavy and difficult ones to simplify a
text. For example, replacing “terminate” with “finish” can make understanding the text easier for
readers. It is recommended that the text be once matched with the essential database in an English
dictionary and, if suitable synonyms are found, they should replace the words in the text. This
way words can be included in the text that are familiar to most readers. Databases of essential
words for each specific field must be prepared in this stage. In thesis [10], the two sources of
everyday literature for children and search engines were used to find the essential words.
Table 2: Thematic modeling
General topics Examples of verbs
Cognition Think, analyze, judge…
Communications Tell, ask, teach …
Feelings Feel, love, fear …
Social affairs Participate, make,
establish …
5.4. The role played by passive sentences in text ambiguity
If we were to caution the author of an article to omit passive sentences, our recommendations for
the text would certainly increase. On the other hand, in some cases the author may not have the
necessary information and be forced to use passive sentences. Therefore, and considering the fact
that passive sentences are involved in text complexity, the proposed solution is to find the
percentage of passive sentences used in each paragraph. If a high percentage of passive sentences
are used in the text, the author can be informed to correct the text. Text processing software

8
recognizes the percentage of passive sentences very accurately. Research has shown passive
sentences are most commonly employed in daily conversations but used less frequently, and in
descending order, in stories, magazines, newspapers, and university textbooks. This shows
omission of passive sentences in university texts must receive special attention in order to remove
ambiguities from them.
6. FEASIBILITY OF THE PROPOSED SYSTEM
first point to consider is that it certainly will not be easy to simplify masses of articles (large-
volume data). For example, the IEEE Xplore site alone presents more than 20000 articles each
year in its research-university databases. Therefore, it is impossible to simplify the entire human
knowledge into simple texts. The only thing that can be done is to work case by case. Since we
plan to offer a system for making changes in texts to turn them into easy-to-read materials, we
will investigate the possibility of implementing some of the processes included in the system that
recommends simple reading.
Figures (eg, Figure 1) must be numbered consecutively, 1, 2, etc., from start to finish of the
paper, ignoring sections and subsections. Tables (eg, Table 1) are also numbered consecutively,
1, 2, etc., from start to finish of the paper, ignoring sections and subsections, and independently
from figures
Table 2: Feasibility of the proposed system

9

10
7. CONCLUSIONS
Google has reported that more than 85% of searches made by users have been for finding useless
information in the internet space. Moreover, according to the announcement made by the
International Publishers Association, only 60% of the books published in Europe in 2013
contained knowledge-enhancing materials, and the rest were attractive novels or were in the
entertainment or political domains, etc. The need people have for knowledge, and the fact that
people have become comfort seekers, have led to the path toward acquiring useless information
that most often lacks knowledge and is used by the public because it is cheap (or free) , rapidly
accessible, enjoyable, and easy to read. Therefore, if a project can be executed to turn knowledge-
containing complex texts into simple, explicit, expressive, and highly readable ones, it will
naturally be possible to take a long step toward making knowledge acquisition easy. In this
research, a preliminary project was studied for creating software to change complex texts into
simple ones and to remove their ambiguities.
REFERENCES
[1] C. Blake, J. Kampov, A. K. Orphanides, D. West, and C. Lown, “Query expansion, lexical
simplification and sentence selection strategies for multi-document summarization,” in Document
understanding conference (DUC-2007), 2007.
[2] J. Allanson and S. H. Fairclough, “A research agenda for computing,” Interacting with Computers,
2004.
[3] J. Medero and M. Ostendorf, “Analysis of vocabulary difficulty using wiktionary,” in Proc. SLaTE
Workshop, 2009.
[4] Korhonen, Y. Krymolowski, and Z. Marx. 2003. Clustering polysemic subcategorization frame
distributions semantically, pages64–71.
[5] Lappin and M. McCord. 1990. A syntactic filter on pronominal anaphora for slot grammar. In
Proceedings of 28th ACL, pages 135–142.
[6] Max, A. (2006). “Writing for language-impaired readers”. In Proceedings of the 7th International
Conference on Intelligent Text Processing and Computational Linguistics, pages 567–570, Mexico
City. Springer-Verlag.
[7] Petersen, S. E. and Ostendorf, M. (2007). “Text simplification for language learners: A corpus
analysis”. In Proceedings of the Speech and Language Technology for Education Workshop (SLaTE-
2007), pages 69–72, Pennsylvania, USA.
[8] S. Crossley, D. Allen, and D. McNamara, “Text simplification and comprehensible input: A case for
an intuitive approach,” Language Teaching Research, vol. 16, no. 1, pp. 89–108, 2012.
[9] S. E. Petersen and M. Ostendorf, “Text simplification for language learners: A corpus analysis,” in
Speech and Language Technology for Education workshop, 2007.
[10] Siddharthan, A. (2003). Syntactic Simplification and Text Cohesion. PhD thesis, Universityof
Cambridge.
[11] V. Hatzivassiloglou and K.R. McKeown. 1997. Predicting the semantic orientation of adjectives. In
Proceedings of ACL/EACL, pages 174–181.
[12] Witten, I., Frank, E. 2005. Data Mining: Practical Machine Learning Tools and Techniques, 2nd
Edition. Morgan Kaufmann

Software feasibility study to

More Related Content

What's hot (20)

Viewers also liked (18)

Similar to Software feasibility study to (20)

Recently uploaded (20)

Software feasibility study to