A SURVEY ON QUESTION ANSWERING SYSTEMS: THE ADVANCES OF FUZZY LOGIC

Dhinaharan Nagamalai et al. (Eds) : ACSIT, ICITE, SIPM - 2018
pp. 75–93, 2018. © CS & IT-CSCP 2018 DOI : 10.5121/csit.2018.80806
A SURVEY ON QUESTION ANSWERING
SYSTEMS: THE ADVANCES OF FUZZY
LOGIC
Eman Mohamed Nabil Alkholy1
, Mohamed Hassan Haggag1
and
Constantine’s Koutsojannis2
1
Faculty of Computers & Information, Helwan University, Cairo, Egypt
2
Health Physics & Computational Intelligence Lab, TEI of Western Greece,
Aigion
ABSTRACT
In real world computing environment with using a computer to answer questions has been a
human dream since the beginning of the digital era, Question-answering systems are referred to
as intelligent systems, that can be used to provide responses for the questions being asked by the
user based on certain facts or rules stored in the knowledge base it can generate answers of
questions asked in natural , and the first main idea of fuzzy logic was to working on the problem
of computer understanding of natural language, so this survey paper provides an overview on
what Question-Answering is and its system architecture and the possible relationship and
different with fuzzy logic, as well as the previous related research with respect to approaches
that were followed. At the end, the survey provides an analytical discussion of the proposed QA
models, along or combined with fuzzy logic and their main contributions and limitations.
KEYWORDS
Question-answering, Neutral language processing, Fuzzy Logic, Answer Extraction, Evaluation
Metrics
1. INTRODUCTION
As technology and human-computer interaction advances, there is an increased interest in
affective computing and a large amount of data is generated and made it available every day, and
for that it requires to integrate and query a huge amount of heterogeneous data, for that NLP has
been recognized as a possible solution that capable to manipulate and represent the complex
query as uncertain and complicated that are existing in them. Which leads to the generation of
QA consisting of Equation and Answer that mapping between these information [1]. However,
Question Answering (QA) is a fast-growing research area that combines the research from
Information Retrieval, Information Extraction and NLP. It can be seen as the next step in
information retrieval is to automatically generating answers to natural language questions from
humans, that allow users to pose questions in natural language and receive succinct answers.

76 Computer Science & Information Technology (CS & IT)
In other aspect (view) Meaning is essential to human language. The idea of the Fuzzy Is use fuzzy
set theory that it derived from it a form of multivalued logic to deal with approximate reasoning,
to represent and process the linguistic information and attributes of the real world [1]. This survey
provides an overview of QA system its system architecture and the possible relationship and
differentiation with fuzzy logic, our believe is to have human-friendly dealing with more
accuracy in information that needs to use fuzzy logic with NLP.
2. BACKGROUND
Over the past four decades Question Answering systems have been transitional much at par with
the whole of natural language processing. In this section, we present a previous work on
development of QA system and it's propose, the earliest system was developed in 1959 (in the
spirit of the era called The Conversation Machine), and A large number of QA system have been
developed since 1960’s One of the most memorable systems was BASEBALL developed by
(Green et al. 1961 in NL DB systems) [2]. Although, capable of answering rather complex
questions, BASEBALL was, not surprisingly, for answering questions about baseball games
played in the American league over one season, restricted to questions about baseball facts, and
most question answering systems were for a long time restricted to front-ends to structured
databases. And in 1963 they develop QA system PROTOSYNTHEX that permit user to ask a
question in English, it accept statements in (sub set of English) as input to its database and
accepts quotations as a query to the database. And for read and solve the kind of word problems
they develop (Problem-solving systems) STUDENT QAs system in 1964 That can read and solve
the kind of word problems found in high school algebra books. The program is often cited as an
early accomplishment of AI in natural language processing [3].
In early 1966 they provide QA system ELIZA that can communication with user) And This
conversation can be simulated with a psychologist. It was able to converse on any topic by
resorting to very simple rules that detected important words in the person's input, And in 1972
they develop SHRDLU that offered the possibility to operation of a robot in a toy world (the
"blocks world)to be simulated with ability to ask the robot questions about the state of the world
SCHOLAR QA system develop in 1973 it was a formal theory characterizing the variety of
plausible inferences people use to ask questions about which their knowledge is incomplete. it has
methods embed to lessons learned from such research into the SCHOLAR system [4].
In the same year 1973 they develop the first automatic question answering system (SAM). After
three years in 1976 TRIPSYS (HWIM) was the first QA that understand speech Question ,
TRIPSYS(HWIM) was developed as the context for a research project in continuous speech
understanding- it understands and answers questions about planned and taken trips, travel budgets
and their status, costs of various modes of transportation to various places it's called HWIM (for
“Hear What I Mean[2]). And the second famous QA system was (NL DB systems) Problem
solving systems LUNAR in 1977[5] for answering questions about the geological analysis of
rocks returned by the Apollo moon missions.
In 1977 develop two QA system the first one is GUS it was a dialog system for airline reservation
second one was LIFER that develop to asking questions about U.S. Navy ships [6]. This system
used a semantic grammar with domain information built within in 1978 they start to develop
system that deal with story comprehension (NL DB systems) QUALM an application that use for
story comprehension and this application responsible for scripts and plans in a very restrictive

Computer Science & Information Technology (CS & IT) 77
domain. In 1983 Salton and McGill describe question answering (QA) systems as mainly provide
direct answers to questions. Finally Kupiec (1993) employed similar but rather simpler WH
question models to build a QA system [8].
In 1991 was QA system (LILOG) text understanding system that operated on the domain of
tourism information in a German city. In 1993 they start to combined NLP with the use of an
online encyclopedia by developing MURAX QA system that combined NLP with the use of an
online encyclopedia with hand-coded annotations to sources. In subsequent developments, QAS
aimed on making linguistic analysis of the questions to capture the intended requirements in a
natural way [7]. In subsequent developments, one of QAS aimed to capture the intended
requirements in a natural way IS to making linguistic analysis of the questions.
In recent 90’s, question answering achieved a great progress due to the introduction of the Text
Retrieval Conferences (TREC) question answering track there has been great progress in open
domain Question answering (Voorhees 2001) . These systems use unrestricted text as a primary
source of knowledge. One such system, MASQUE (ANDROUTSOPOULOS et al., 1993) use
logic representation to represents natural language questions, and for retrieving intended
information from database, it use to translates the logic query into a database query. It separates
from mapping process the task of linguistic process. FAQ Finder (Burke et al., 1997) does
matching of the question list that compiled in a knowledge base with questions through statistical
similarity and semantic similarity and for syntax-based natural language understanding technique.
In 1999 was LASSO that win the question answering task. It used question classification
technique and syntaxbased natural language understanding technique. Another QAS in (2000)
was developed by Riloff and Thelen (QUARC), that has the ability to classify questions into
different types and use lexical and semantic clue to derive their expected answer [9] [10].
Later, the focus of developing QASs was shifted toward open domain QAS, TREC Evaluation
campaign which is taking place every year since 1999 to manage and query large volume of data
and represent most of research in open domain question answering from unstructured data
sources, and that lead to question-answering evaluations as a recent success started as part of the
Text Retrieval Conference (TREC). The best systems are now able to answer more than two
thirds of factual questions in this evaluation, it descripts the results and the associated evaluation
Methodology develop by Ellen Voorhees, the second paper, by Buchholz and Daelemans,
explores the requirements for answering complex questions that have compound answers or
multiple correct answers [11].
The third paper, by Lin and Pantel, describes a new algorithm to capture paraphrases that allow a
more accurate mapping from questions to potential answers. The fourth paper, by Light, Mann,
Rilo and Breck, describes experiments that systematically factor and assess question answering
into component sub-problems. The first TREC evaluation campaign provides a list of 200
questions and a document collection. The answers were known to be present in the collections.
The maximum lengths of answers were allowed to be 50 or 250 characters[12]. Systems were
asked to give 5 ranked lists of answers.
In the next campaign, TREC-9 held in 2000, the number of questions and size of document
collections were increased. In TREC-10 in 2001, a new complexity with respect to answers. The
lengths of answers were reduced to 50 words [11]. In TREC- 11, held in 2002, systems were
expected to give exact short answers to the questions. In TREC from 2002 to 2007, the list of
questions, definition questions, and factoid questions were included in the evaluation campaigns.

In TREC 2005 and TREC 2006 there was temporal question in addition to 75 topics that contains
different type of Questions. In TREC 2007, there were a collection of documents that include
collection of blogs, and progress competitions that lead to increasing documents collection
complexity, and questions complexity, and that also effect to increasing answer evaluation
strategies complexity.
In 2000 they start to develop QA system for Unix operating system Unix Consultant For
answered questions pertaining to the Unix operating system [10]. This system had ability to
accommodate various types of users by phrasing the answer, and it has its domain that contain a
comprehensive hand-crafted knowledge base. In 2001 there was two QA systems the first one
was INSIGHT question answering system which uses some surface patterns, wins the question
answering task in TREC-10, and the second one was SiteQ use the density-based extraction
method to retrieve related passages first and then extract the exact answers in them, which can
greatly improve the extraction speed.
In 2002 STARTQA system was the first web-based QA system for English and in the same year
2002 was Answer bus QA system(ODQA) they develop this system to accepts questions in
several languages (extend the answer extraction process from the local data source to the World
Wide Web, which allows them to deal with large count of questions. After one year they develop
QA system (ARANEA) it was first fully downloadable open-source Web-based factoid question
answering system, and in the same year 2003 create a QA system AQUA was a sophisticated
automatic question answering system, which combines natural language understanding technique,
ontological knowledge, logical reasoning abilities and advanced knowledge extraction techniques
[12].
In 2008 they start to develop a new QA system with different approach and it was a List questions
ask for different instances of a particular kind of information to be returned.in 2012 they develop
QA Systems that oriented to work with opinions rather than facts can also be included in this
category. In 2014 was QAKIS by Cabrio et al and FREITAS14 develop by Freitas and Curry and
also INTUI3 by Dima, in 2015 was develop three QA system first was HAKIMOV15 by Usbeck
et al [12]. and second was QASYO by Hakimov et al. last. However, QA systems have developed
over the past few decades until they reached the structure that we have nowadays.
2.1 Implementation Approaches for QA Systems
The basic aim of QA System is to provide correct and short answer with high accuracy to the user
and There are many Approaches used in Question answering system based on different purpose
[13]. This section will present an implementation approaches for various categories of QA
System such as:
2.1.1 First Approach
That relied on artificial intelligence (AI) it called Linguistic approach and it has the ability to
build QA logics by using methods that integrated natural language processing (NLP) techniques
and knowledge base. It used Linguistic techniques for formulating user’s question into a precise
query that merely extracts the respective response from the structured database, it Used to
understand natural language text, linguistic & common knowledge Linguistic techniques such as
such as (tokenization, POS tagging and parsing [14].

2.1.2 Second Approach
(rule-based approach) that rely on the (rule-based mechanism), they are built to identify question
classification features. Quarc developed by Rilloff et al., and Cqarc [15] developed by Hao et al.
Used to identify the question class by using semantic clues in question and for looking lexical
they used heuristic rules.
2.1.3 The Third Approach
(Statistical approach) gives the better results than other approaches user can use this approaches
successfully to applied the different stages of a QA system, it independent of structured query
languages and can formulate queries in natural language form. IBM’s statistical QA system was
based on the statistical model. This system utilized maximum entropy model for question/ answer
classification based on various N-gram or bag of words features. Moschitti had used Rocchio and
SVM text classifiers for question and answer categorization. Availability of huge amount of data
on internet increased the importance of statistical approaches [16].
2.1.4 The Fourth Approach
is Pattern matching that have the ability to replace the sophisticated processing involved in other
competing approaches by using the expressive power of text patterns. Some of patterns matching
QA rely on templates for response generation while most of of the patterns matching QA systems
[17].
2.1.5 The Fifth Approach
is the Surface Pattern based approach. This approach rely on an extensive list of patterns to
extracts answers from the surface structure of the retrieved documents. It’s automatically learning
based pattern or it is human crafted. [18]. However, to identify any question answer it depend on
the basis of similarity between their reflecting patterns having certain semantics as well as their
Methods, definition and Characteristics, Limitation and Aims for each QA system that was
developed before which will be helpful for new directions of research in this area (Table 1).
Table 1: List of Most Popular QA Systems

Table2: Fuzzy Logic Implementation Approaches For QA Systems
QA System Method & Definition
Quantitative Fuzzy Semantics
(L. A. ZADEH)(1971)
the first paper on fuzzy sets it can construct
fuzzy query languages for purposes of
information retrieval, and, possibly, to
implementation of fuzzy algorithms and
programs. [37]
A Fuzzy-Set-Theoretic Interpretation of Linguistic
Hedges L. A. Zadeh (1972)
represent a hedge as an operator, linguistic
hedge such as very, more or less, much,
essentially, slightly. [35]
PRUF- A Meaning Representation Language For
Natural Languages” (1978)
PRUF provide a basis for question-answering
and inference from fuzzy premises [40].
Test-Score- Semantics as A Basis For A
Computational Approach (1986)
test-score semantics provides a framework for
the representation of the meaning of
dispositions.[42]
A Computational Approach to Fuzzy Quantifiers In
Natural Languages (1983)
deal with fuzzy quantifiers (denote the
collection of quantifiers in NLP whose
representative elements are: much, not many,
very many, notvery many) that are treated as
fuzzy numbers[37]

Fuzzy Quantifiers: A Natural Language Technique
for Data Fusion 2001
intended to formalize the notion of ‘linguistic
adequacy’. then argue that the models of the
theory are plausible from a linguistic
perspective.[37]
Fuzzy Logic-Based Natural Language Processing
and Its Application to Speech Recognition
2017
it create a system that can learn from a
linguistic corpus. it use fuzzy semantic
relations to represented by words and use such
relations to process the word sequences
generated by speech recognition systems. [32]
The Fuzzy Formal Concept Analysis (FFCA)
2003
proposed a fuzzy FCA-based approach for
conceptual clustering for automatic generation
of concept hierarchy from uncertainty
information.[43]
A Hybrid Approach Using Ontology Similarity and
Fuzzy Logic
For Semantic Question Answering 2014
the user enters a source string as a question. the
first objective of the machine is to syntactically
analyze the text from the source. using fuzzifier
approach (semantic fuzzy ontology) [44]
Improving translation memory fuzzy matching by
paraphrasing
2015
present an innovative approach to match
sentences having different words but the same
meaning. presented a method that improves the
fuzzy match of similar, but not identical
sentences. [41]
Semantic parsing on freebase from question-answer
pairs
train a semantic parser that scales up to
Freebase.method : build a coarse mapping from
phrases to predicates using a knowledge base
and a large text corpus. [33]
FLINTSTONES: A fuzzy linguistic decision tools
enhancement suite based on the 2-tuple linguistic
model and extensions
Use fuzzy linguistic tools called Flintstones
that proposed to solve linguistic decision
making problems based on the 2-tuple
linguistic model in order to validate the
performance of the software suite with real
datasets[34]
A fuzzy linguistic approach for human resource
evaluation and selection in software projects
2015
present a fuzzy linguistic approach that utilizes
2-tuple fuzzy linguistic terms and supports the
selection of suitable human resources based on
their skills and the required skills for each
project task.[35]
Fuzzy logic in natural language processing
2017
they outline how model of the meaning of basic
constituents of natural language (nouns,
adjectives, adverbs, verbs) has been elaborated
in FNL[39]

3. QA SYSTEM OVERVIEW
Question answering is a process that understanding user natural language query and has the
ability to provide a correct answer and extract it from retrieving relevant documents, data, or
knowledge base .A question answering (QA) Question-answering systems are referred to as
intelligent systems ,that can be used to provide responses for the questions being asked by the
user based on certain facts or rules stored in the knowledge base it can generate answers of
questions asked in natural [28].
3.1 QA System Purpose and Types
The purpose of a QA system is to provide correct answers to user questions in both structured and
non-structured collection of data. there are three main types for the QA Depending on the target
domain and the way questions are answered it will present in table.
Table 4: Types of QA Systems
4. QA SYSTEM ARCHITECTURE & ALGORITHM
4.1 System Architecture
As shown in figure (1) a QA system contain three main part beside other supplementary
components: question classification (QUESTION PROCESSING), information retrieval
(DOCUMENT PROCESSING) and answer extraction (ANSWER PROCESSING). The user
writes a question using the user query interface. Then this query is used to extract all the possible

answers for the input question. The architecture of Question Answering system is as shown in
Figure1
Figure (1): Question Answering System Architecture
4.1 Question Answering System Algorithm
4.1.1 Step One
(Question processing): Given a question as input, user writes his question by an interface the
function of the question processing module is to process and analyses the question, and to create
some representation of the information requested module is to process and analyses the question,
and to create some representation of the information requested the question, and to create some
representation of the information requested analyses the question, and to create some
representation of the information requested. This leads to the classification of question that
include three parts: Query Interface, question analyzer and Question classification. This step help
to generate complex structured queries . and detects the expected answer type of a question e.g.
the expected answer type of " When was last time that you feel this headache?" is date or “Do
other family members have similar headaches?” Is yes/no this information helps guide the answer
extraction process [29].
4.1.2 Step Two
after Input questions -Determining the question type: is selected from question taxonomy that
system uses. and candidate answers are extracted using all the information gathered in the
previous steps, e.g., keywords, entities and relations. The remaining candidate answers are then
filtered and ranked according to different measures of similarity to filter out candidates with
incompatible types. The Document Processing Module retrieves documents from the corpus that
are likely to contain answers to the user's question. It consists of a query generation algorithm,
text search engine and information retrieval that retrieve the relevant documents based upon
important keywords appearing in the question. The query generation algorithm takes an input the

user's question and creates a query containing terms likely to appear in documents containing an
answer. This query is passed to the text in system, which uses it to retrieve answer. [30].
4.1.3 Step Three
The answer processing module is responsible for identifying, extracting and validating answers
from the set of ordered paragraphs passed to it from the Document Processing Module, takes
input from the retrieval component and tries to retrieve an exact phrase to return as an answer to
achieve this required parsing and detailed question analysis by using of answer extraction
algorithms. The identity answer extraction returns the CenterPoint of the passage, stripping words
from either end until it fits within the specified answer. Then Answer Display The result and
converted into required text which is required by the user and displayed to the user.
4.2 Evaluation Metrics
There are several parameters that are used to analyze the performance of different Question
Answering Systems. In this section we describe some of the evaluation metrics used in Question
Answering System to evaluate its performance such as: precision, recall and F-measure, a
weighted harmonic mean of precision and recall, can be defined with respect to predicates for the
purposes of QA evaluation. These precisions and recall metrics express true precision and recall,
not approximations, when coupled with an answer key in which the judgments can be reasonably
assumed to be exhaustive[31].
4.2.1 Precision
Precision- Recall is the most common metric to evaluate information retrieval system. Precision
is the ratio of retrieved documents that are relevant to all retrieved documents in the ranked list
[30].
4.2.2 Recall
Recall in information retrieval is the fraction of the documents that are relevant to the query that
are successfully retrieved.
For example: text search on a set of documents recall is the number of correct results divided by
the number of results that should have been returned In binary classification, recall is called
sensitivity. So it can be looked at as the probability that a relevant document is retrieved by the
query. It is trivial to achieve recall of 100% by returning all documents in response to any
query[36]. Therefore, recall alone is not enough but one needs to measure the number of non-
relevant documents.

4.2.3 F-Measure
A measure that combines precision and recall is the harmonic mean of precision and recall, the
traditional F-measure or balanced F-score:
known as the F1 measure, because recall and precision are evenly weighted. It is a special case of
the general Fβ measure (for non-negative real values of β ):
Two other commonly used F measures are the F2 measure, which weights recall higher than
precision, and the F0.5 measure, which puts more emphasis on precision than recall.
The F-measure was derived by van Rijsbergen (1979) so that measures the effectiveness of
retrieval with respect to a user who attaches β times as much importance to recall as
precision"[36].
It is based on van Rijsbergen's effectiveness measure (5)
Their relationship is where (6)
5. FUZZY LOGIC OVERVIEW
This section presents a quick overview of what fuzzy logic is and its system architecture and the
possible relationship with Question Answer, Fuzzy logic is a technique used for representing and
manipulating uncertain information , In the more traditional propositional logic, each fact or
proposition , the truth vale in fuzzy logic may range between completely true and completely
false that it’s a concept of partial truth that fuzzy logic used to handle it By contrast, in logic.
non-numeric values are often used to facilitate the expression of rules and facts linguistic variable
and that in fuzzy logic application such as age may accept values such as young and its
antonym old. And Because natural languages do not always contain enough value terms to
express a fuzzy value scale, it is common practice to modify linguistic values
with adjectives or adverbs[37].
So any systems that cannot be precisely described by mathematical models or devices that have
significant uncertainties or contradictory conditions, and linguistically controlled devices or
systems can use fuzzy logic. As Lotfi Zadeh once stated, fuzzy logic is not going to replace
conventional logic or methodologies, rather it will supplement them in circumstances where
conventional approaches fail to solve a problem effectively [38]. In recent years, Fuzzy logic has
proved to be particularly useful in expert system and other artificial intelligence applications and
NLP . Therefore, many of the previous researchers in academia, applications include modeling,
evaluation, optimization, decision making, control, diagnosis and information Measure the

growing interest in fuzzy logic. In particular, fuzzy logic is best suited for control-systems fields.
it has been applied in areas such as breakdown prediction of nuclear reactors in Europe,
earthquake forecasting in China, and subway control in Japan.
5.2. Fuzzy Logic General Architecture
Figure 2: Fuzzy Logic General Architecture
5.2.1-Fuzzification
receives the inputs and transforms it into fuzzy sets (by maps the given inputs to fuzzy sets and
linguistic variable). the fuzzy rules premises that are predefined in the rule base of the specified
application can be matched with defined linguistic variables [36].
5.2.2- Rule Base
includes a set of linguistic rules designed in the form (if-then rules). These rules define the
consequent of the model in terms of the given linguistic variables such as low, moderate and high.
In addition, it specifies the type and number of the used membership functions of input and
output parameters [38].
5.2.3-the fuzzy inference engine
is considered the central part of the fuzzy system it contains the Knowledge Base, that stores IF-
THEN rules provided by experts. It simulates the human reasoning process by making fuzzy
inference on the inputs and IF-THEN rules, combines rules from the fuzzy rule base and gives a
mapping from input fuzzy sets, each rule is interpreted as a fuzzy implication. In this step the
inference engine is applied to a set of rules included in the fuzzy rule base and that to produce the
intended output. This procedure involves many steps as follows: first, it matches the linguistic
variables of the input with the rules’ premises. Second, it activates the matched rules in order to
deduce the resultant of each fired rule, and finally it combines all consequents by using fuzzy set
union in order to generate the final output which is represented as fuzzy set output [37].

5.2.4- the Defuzzifier
It transforms the fuzzy set obtained by the inference engine into a crisp value, the output is
produced as a linguistic variable, which is fuzzy and can be interpreted in different ways. The
process of finding a crisp output after fuzzification and inference is called Defuzzification There-
fore, the fuzzy set output in this stage is converted to a crisp output (which results on a fuzzy
output set).
5.3 Advantages of Using Fuzzy Logic with QA System
QA Question answering system enable users to access the knowledge in a natural way by asking
questions and get back relevant correct answers, and that requires understanding of natural
language text, linguistics and common knowledge, derived from the documents and providing
accurate answers to a user’s question often expressed as uncertainty words. So many of previous
researchers on how to build QA logics, find that using artificial intelligence (AI) based methods
that integrate natural language processing (NLP) techniques and knowledge base or corpus help
to build QA logics. And they found that using fuzzy logic to modify linguistic values can retrieve
the answers by matching user’s question with the existing hierarchical ontology, and not only
provide syntactic answers, but also semantic answers based on the question terms. It can produce
result sets according to the degree of vague expressions in natural language, and vague adjectives
in particular. And also, one of the main strengths of Fuzzy logic is that it allows the semantic
partitions to overlap. In (1988) Smithson (argues that psychological explanations that permit
choice under partial and uncertain constraints can be compatible with fuzzy logic [39]. For that
Fuzzy logic is widely used in a different field such as expert systems, business and medicine. It
can measure the output of the process and takes control actions on the process continuously.
Using fuzzy logic in NLP make the possibility of constructing a theory for artificial languages
whose terms have fuzzy meaning and contribute it to a clarification of the concept of semantic
meaning. Fuzzy logic, has a direct relation to NLP and QA development , A Critical Survey on
the use of Fuzzy in Natural Language Processing start from ( 2004, Araujo) for rule learning, in
2010 was for knowledge representation by and in 2011 by Lai, Wu, Lin, & Huang,), research
was for word meaning inference in (2012,Carvalho,Batista,&Coheur)for linguistic
summarization[38]. in 2013was two search the first one was about grammatical inference ,by(
(Kazemzadeh, Lee, & Narayanan)and the second one was for emotion recognition by (Luong,
Socher, & Manning) in 2014 by (Cambria, Gastaldo, Bisio, &Zunino) for text categorization .
most of the researchers currently working in the NLP field with fuzzy logic in different aspects,
tries to find some guidelines on what could be done to and make fuzzy logic more interfere with
NLP. Therefore, many of the previous researchers relied on used fuzzy logic in NLP have many
advantage as it can represent the meaning of dispositions provide with more accuracy in
expressive with easy-to-use operators provides simpler framework that lead to increase the
accuracy of recognition system efficient. and also, fuzzy logic can use as translators or framework
to match scores higher than before across different text domains.
6. CONCLUSIONS
in this survey paper are show an overview on what Question-Answering is and its architecture
and the possible relationship with fuzzy logic and how fuzzy logic can be used as intermediate
semantic representations of vague expressions. we pin down the previous related research with
summarized and organized recent research results in a novel way that integrated and added

understanding to work in the question-answering field. It emphasized the classification of the
existing literature, developing a perspective on the area, and evaluating trends. However, because
it is impossible for a survey to include all or even most of previous research, this survey included
only the work of the top-publishing and top-cited authors in the QA field.
7. FUTURE WORK
In this paper it has been discussed some of the elementary approach, and it perform fairly well but
suffer from some limitations. This fact leads us to next step as a next step, we will try to tame our
feature seton one possibility is to development a QA system using hybrid approach in our system
that will be designed to be able to answer number of questions mainly based on fuzzy rule that is
less impacted by the serious imbalance between negative and positive instances. This should help
to increase the accuracy of Future result the architecture of Future Question Answering system
using fuzzy logic rule is as shown in Figure 3.
Figure 3. Question Answering System Using Fuzzy Logic Rule
REFERENCES
[1] Hendrix. G.G, Sacerdoti, E.D,sagalowicz. D. Slocum. J. "'Developing a natural Language interface
tocomplex data in ACM Transaction on database system. 3(2). pp. 105- 147,1978.
[2] Green BF, Wolf AK, Chomsky C, and Laughery K. Baseball: An automatic question answerer. In
Proceedings of Western omputing Conference, Vol. 19, 1961, pp. 219–224.
[3] [Cook, 2006] Cook, J. L. (2006). College students and algebra story problems: Strategies for
identifying relevant information. Reading Psychology, 27:95 – 125.
[4] Weizenbaum J. ELIZA - a computer program for the study of natural language communication
between man and machine. In Communications of the ACM, Vol. 9(1), 1966, pp. 36-45

[5] Woods W. Progress in Natural Language Understanding - An Application to Lunar Geology. In
Proceedings of AFIPS Conference, Vol. 42, 1973, pp. 441–450.
[6] Bobrow, D. G., Kaplan, R. M., Kay, M., Norman, D. A., Thompson, H., and Winograd, T. (1977).
GUS, A frame driven dialog system. Artificial Intelligence, 8, 155–173.
[7] Julian Kupiec. 1993. MURAX: A robust linguistic approach for question answering using an on-line
encyclopedia. In Robert Korfage, Edie Rasmussen, and Peter Willett, editors, Proceedings of the
Sixteenth Annual International ACM SIGIR Conference on Research and Development in
Information Retrieval, pages 181–190. Special issue of the SIGIR FORUM
[8] Lehnert, W. G. (1978). The Process of Question Answering: A Computer Simulation of Cognition.
Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
[9] Huettner, A. (2000) “Question Answering”. In: 5th Search Engine Meeting.
[10] Voorhees EM. The TREC-8 question answering track report. In Proceedings of TREC-8, 1999, pp.
77-82.
[11] K.L.Kwok,L.Grunfeld,N.Dinstl,and M.Chan.TREC-9 Cross language,web and quationt -answering
track experiments using PIRCS.In Voorhees and Harman .
[12] E.M.Voorhees and D.K. Harman,editors,Proceedings of the Ninth Text REtreival Conference (TREC-
9)
[13] Stefanie Tellex, "Pauchok: A Modular Framework for question Answering", Master Thesis Submitted
to the Department of Electrical Engineering and computer science, Maccachusetts institute of
Technology, June 2003..
[14] Michael Kaisser. Question Answering by Searching Large Corpora with Linguistic Methods.
Master’s thesis, Saarland University, Germany, 2004.
[15] Riloff E and Thelen M. A Rule-based Question Answering System for Reading Comprehension
Tests. In ANLP /NAACL Workshop on Reading Comprehension Tests as.
[16] Ittycheriah A, Franz M, Zhu WJ, Ratnaparkhi A and Mammone RJ. IBM’s statistical question
answering system. In Proceedings of the Text Retrieval Conference TREC-9, 2000.
[17] Joho, H. (1999) ”Automatic detection of descriptive phrases for Question Answering Systems: A
simple pattern matching approach”. MSc Dissertation. Department of Information Studies,
Universityof Sheffield. Sheffield, UK
[18] Deepak Ravichandran, Abraham Ittycheriah and salimroukos, “Automatic Derivation of surface text
pattern for a maximum Entropy Based question answering system”, Work done while the author was
an intern at IBM TJ Watson research center during summer 2002.
[19] Cai D, Dong Y, Lv D, Zhang G, Miao X. A Web-based Chinese question answering with answer
validation. In Proceedings of IEEE International Conference on Natural Language Processing and
Knowledge Engineering, pp. 499-502, 2005.
[20] J.-L. Binot, L. Debille, D. Sedlock, and B. Vandecapelle. Natural Language Interfaces: A New
Philosophy. SunExpert Magazine, pages 67{73, January 1991

[21] Berger A, Caruana R, Cohn D, Freitag D, and Mittal V. Bridging the lexical chasm: statistical
approaches to answer-finding. In Proceedings of the 23rd annual international ACM SIGIR
conference on Research and development in information retrieval, 2000, pp. 192-199.
[22] Bronnenberg, W.J., Bunt, H.C., Landsbergen, S.P.J., Scha, RoJ.H., Schoenmakers, W.J., van Utter,n,
E.P.C. (1979) The question answering system PHLIQAI. In L.Bolc (ed.), Natural communication
with computers, McMillan, London; Hanser Verlag, M~nehen.
[23] SACERDOTI, E.D. Language access to distributed data with error recovery. Proc. 5th Int. Joint Conf.
on Artificial Intelligence, Cambridge, Mass., Aug. 1977.
[24] Ask Jeeves. 1996. www.ask.com Site last visited in 28-march-2018.
[25] AnswerBus, Question Answering System. Website: http://answerbus.com J. Allen. 1995. Natural
Language Understanding. The Benjamin/Cummings Publishing Company, Menlo Park, CA.
[26] Vargas-Vera M. and Motta E and Domingue J. (2003a): AQUA: An Ontology-Driven Question
Answering System. AAAI Spring Symposium, New Directions in Question Answering, Stanford
University, March 24-26, 2003.
[27] Boris Katz, Sue Felshin, Deniz Yuret, Ali Ibrahim, Jimmy Lin, Gregory Marton, Alton Jerome
McFarland, and BarisTemelkuran. 2015. Omnibase: Uniform access to heterogeneous data for
question answering. In Proceedings of the 7th International Workshop on Applications of Natural
Language to Information Systems (NLDB 2015).
[28] Unmeshsasikumar, Sindhu L, "A survey of Natural Language question answering system",
international journal of computer applications(0975-8887), volume 108 -No 15. December 2014 .
[29] Anthony Fader, Luke Zettlemoyer, and Oren Etzioni. 2014. Open question answering over curated
and extracted knowledge bases. In Proceedings of the 20th ACM SIGKDD International Conference
on Knowledge Discovery and Data Mining, KDD ’14, pages 1156–1165, New York, NY, USA. ACM
[30] Sanjay K. Dwivedi and vaishalisingh “Research and reviews in question answering system”,
International Conference on Computational Intelligence: Modeling Techniques and Applications
CIMTA) 2013 Procedia Technology 10 ( 2013 ) 417 – 424.
[31] Ellen M. Voorhees and Dawn M. Tice. 1999. The TREC-8 question answering track evaluation.
TREC-8 Proceedings, pages 41–63, Nov.
[32] JIPING SUN, FAKHRI KARRAY, Fuzzy Logic-Based Natural Language Processing and Its
Application to Speech Recognition, Ontario, N2L 3G1, Canada,2017
[33] J.Berant, A. Chou, R. Frostig, and P. Liang, “Semantic parsing on freebase from question-answer
pairs,” in Proceedings of the Conference on Empirical Methods in Natural Language Processing
(EMNLP '13), vol. 2, p. 6, 2013. View
[34] F. J. Estrella, M. Espinilla, F. Herrera, and L. Martínez, “FLINTSTONES: a fuzzy linguistic decision
tools enhancement suite based on the 2-tuple linguistic model and extensions,” Information Sciences,
vol. 280, pp. 152–170, 2014
[35] V. C. Gerogiannis, E. Rapti, A. Karageorgos, and P. Fitsilis, “A fuzzy linguistic approach for human
resource evaluation and selection in software projects,” in Proceedings of the 5th International
Conference on Industrial Engineering and Operations Management (IEOM' 15), pp. 1–9, Dubai,
UAE, March 2015.

[36] VAN RIJSBERGEN, C.J. Information Retrieval, Second Edition, Butterworths, London (1979).
[37] L. A. Zadeh, "A computational approach to fuzzy quantifiers in natural languages", Comput. Math.
Appl., vol. 9, pp. 149-184, 1983.
[38] A. Niewiadomski, "A type-2 fuzzy approach to linguistic summarization of data", IEEE Trans. Fuzzy
Syst., vol. 16, no. 1, pp. 198-212, Feb. 2008.
[39] VilémNovák,Fuzzy Systems (FUZZ-IEEE), 2017 IEEE International Conference on, 10.1109/FUZZ-
IEEE.2017.8015405, Naples, Italy, 8, 2017
[40] Zadeh, L. A. (1978) ,PRUF-A Meaning Representation Language for Natural Languages, Int. Journal
Man–Manchine Studies, 10, pp. 395–460; also in Fuzzy Sets and Applications: Selected Papers by L.
A. Zadeh, John Wiley & Sons, New York, pp. 499–568 (1987).
[41] Konstantinos Chatzitheodorou, Proceedings of the Workshop on Natural Language Processing
(NLP4TM)Improving translation memory fuzzy matching by paraphrasing, pages 24–30, Hissar,
Bulgaria, Sept 2015
[42] L. Zadeh. Test-Score Semantics as a Basis for a Computational Approach to the Representation of
Meaning. Literary and Linguistic Computing, 1986.
[43] Fenza, G., V. Loia, and S. Senatore. 2008. “Concept Mining of Semantic Web Services by Means of
Extended Fuzzy Formal Concept Analysis (FFCA).” IEEE Int. Conf. on Systems, Man and
Cybernetics
[44] Monika Rani, Maybin K. Muyeba, O.P. Vyas, A hybrid approach using ontology similarity and fuzzy
logic for semantic question answering." In Advanced Computing, Networking and Informatics
Volume 1, pp. 601-609. Springer, Cham, 2014.
AUTHOR
Eman Mohamed Nabil Alkholy
Faculty of Computers & Information, Helwan University, Cairo, Egypt.

A SURVEY ON QUESTION ANSWERING SYSTEMS: THE ADVANCES OF FUZZY LOGIC

More Related Content

What's hot (20)

Similar to A SURVEY ON QUESTION ANSWERING SYSTEMS: THE ADVANCES OF FUZZY LOGIC (20)

More from cscpconf (20)

Recently uploaded (20)

A SURVEY ON QUESTION ANSWERING SYSTEMS: THE ADVANCES OF FUZZY LOGIC