Content uploaded by Ayesha Ameen
Author content
All content in this area was uploaded by Ayesha Ameen on Mar 18, 2019
Content may be subject to copyright.
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
20
Knowledge based Recommendation System in Semantic
Web - A Survey
Ayesha Ameen
Professor & Head of Department I.T.
Deccan College of Engineering and Technology
Darusalam, Hyderabad, Telangana, India
ABSTRACT
Knowledge based recommendation systems use knowledge
about users and products to make recommendations.
Knowledge-based recommendations are not dependent on the
rating, nor do they have to gather information about a
particular user to give recommendations. Knowledge
acquisition is the most important task for constructing
knowledge-based recommendation system. Acquired
knowledge must be represented in some structured machine-
readable form, e.g., as ontology to support reasoning about
what products meets the user’s requirements. In Semantic
Web, knowledge is represented in the form of ontology.
Representation of knowledge in structured form of ontology
in Semantic Web makes the application of knowledge based
recommendations system on Semantic Web very easy, as
there is no need to construct knowledge base from scratch.
Performance of knowledge based recommendations systems
can be enhanced by exploiting ontology reasoning
characteristics. This paper explores different techniques used
to generate knowledge-based recommendations highlighting
the advantages of knowledge based recommendation system
over other recommendation techniques.
General Terms
Recommendation systems
Keywords
Semantic Web, Ontologies, Reasoning, Knowledge base.
1. INTRODUCTION
World Wide Web has become a major source of information
acquisition as it contains millions of documents related to any
topic, which is of interest to users. Users often find it difficult
to extract the relevant information from the documents
returned as a result of the query posted on Web the reason
behind this is, WWW contains documents which can be
interpretable by only human but not by machine[1].
Recommendation system is used to solve this problem by
generating personalized recommendations to Web users.
Personalized recommendation in Web is no longer considered
as an option but has become a necessity because of the
movement from traditional physical stores of products or
information to virtual stores of products and information [2].
As a result of this movement customers have a wide variety of
options to choose from. Users can switch from one Website to
another in virtual store; as many Websites offer the same type
of services and products. It becomes difficult to retain
customers in virtual store. Personalized recommendations
help to solve the customer retention problem.
Recommendation systems improve the trust of customer in
business by building customer loyalty and one to one
relationship by understanding the needs of each customer.
2. RECOMMENDATION SYSTEM
The aim of recommendation system is to generate meaningful
recommendations to users for items that might be interesting
to them. Recommendation systems have been widely used by
information sources for personalizing their contents for the
users [3]. In the context of Semantic Web, widely used
recommendation approaches are content-based, collaborative
filtering and knowledge-based [4]. Content based filtering
approach analyses the contents of documents; collaborative
filtering approaches are based on the opinion of group of users
who have the same preferences; knowledge-based approaches
utilize the knowledge in a structured form to produce
personalized recommendations [5].
2.1 Content-based filtering
Content-based recommendation systems recommend items
similar to the items a particular user has liked in the past [6].
Content based recommendation systems analyse items to
identify those items that can be interesting to the users [7].
Recommendations are produced by matching up the attributes
of the object or item with the user preferences or interest
which is stored as attributes of user profile. Techniques used
for content based recommendation systems differ in the way
they analyse the items of documents or descriptions of items
to build up user profile. The following paragraphs discuss
some of the content-based filtering approaches used in
Semantic Web.
Ijntema et al. [8] used Concept Frequency-Inverse Document
Frequency (CF-IDF) which is an adaptation of Term
Frequency-Inverse Document Frequency (TF-IDF) with
semantics for domain ontology to make recommendation for
items. CF-IDF represents items in documents as weighted
vectors of key concepts instead of terms. Weights are assigned
accordingly. High weights are assigned to most discriminating
features/preferences and low weights are assigned to less
informative features/preferences, thereby avoiding noise term,
which can pollute the recommendation output. User profile
and items are represented in terms of CF-IDF.
Recommendations are generated by comparing items with
user profile using cosine similarity.
Bogdanov et al. [9] proposed a user model with input as
explicit preferences from users instead of calculating CF-IDF
as in the approach proposed by IJntema for making
recommendations. Semantic descriptors are calculated for
each input. Trained classifiers are used to obtain the class
labels of each input represented as semantic descriptors. For
each input, the classifier returns probability estimates of
classes on which it was trained. Three approaches are used for
producing recommendations. First two approaches represent
the user model in term of vector and uses weighted Pearson
correlation distance to calculate the distance between user
model vectors and item vector. The first approach calculates
the mean for the user model vector and recommends the item
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
21
nearest to the mean vector. The second approach uses all the
points in user model vector and recommends items that are
close to any point in user model vector. The third approach
represents user model as probability density preferences in
semantic space. Recommendations are generated by using
expectation maximization algorithm.
2.2 Collaborative filtering
recommendation systems
Collaborative filtering recommendation creates groups of
users according to their preferences and generates
recommendation based on items liked by the other users
belonging to the same group [10]. Collaborative filtering
system uses similarity measures to compute the similarity
between users and items recommended. Performance of
collaborative filtering recommendation systems depend on
different approaches used for similarity computation.
Collaborative filtering recommendation system when applied
to Semantic Web computes semantic similarity between items
and users to produce personalized recommendations. Cold
start problem can be reduced by recommending items that are
semantically similar to the given item in Semantic Web. Item
sparsity problem is reduced by mapping items and users to
domain ontology in semantic collaborative filtering [11].
Some of the approaches used for collaborative filtering in
Semantic Web have been discussed in the following
paragraphs.
Lee et al. [12] used semantic collaborative filtering techniques
based on personalized search Bayesian Belief network
(pEBBN) to retrieve documents that have high semantic
similarity to the given query. pEBBN uses domain knowledge
to represent concepts in concept layer which represents the
semantics of user preference queries and documents with their
corresponding concepts. To personalize the search the first
step is to find the implicit authority of document by
collaborative filtering, which finds a set of user who has the
same preferences as that of the user who has submitted the
query. In the second step a score function which performs
conceptual mapping along with the semantic similarities
between like-minded users is computed on the documents
accessed by like-minded users and the query submitted and
the preferences of the user. Top-k documents are returned as a
result of the search for each like-minded user.
Lee’s approach computes semantic similarity between the
documents and queries but does not focus on the semantic
properties that are associated with items and users. Semantic
properties can be utilized to gain more knowledge regarding
items and users in recommendations system. Semantic
properties are used in a new technique for collaborative
filtering called property-based collaborative filtering (PBCF)
in an approach proposed by López et al. [13]. PBCF
represents the domain knowledge about items and users in the
form of ontology to support reasoning about them in a formal
way. PBCF separates users and their properties as well as
items and their properties to build a matrix of values
describing the influence of one item property of item on users
with certain user property. This type of representation solves
many problems that existed in traditional collaborative
filtering recommendation systems.
The approach proposed by Lopez utilized semantic property
information for generating recommendation, whereas the
approach proposed by Fard et al. [14] computes semantic
similarity between users to find nearest neighbours. Items are
recommended to user based on the previous ratings of nearest
neighbours. Semantic similarity method computes weighted
sum between Relation Similarity, Taxonomy Similarity and
Attribute Similarity which are the measures used for
calculating semantic similarity between ontology concepts.
3. KNOWLEDGE BASED
RECOMMENDATION SYSTEMS
Content based filtering and Collaborative filtering
recommendations are suited for products that are purchased
frequently, such as, books, news, and music. In case of items,
such as, computers, cars, financial services, loans and
apartments, which are not purchased, frequently, it becomes
difficult to collect rating, and the user will not be satisfied
with recommendations produced based on old item
preferences. Therefore, using collaborative filtering and
content-based recommendation techniques will not produce
good quality recommendations. The challenges faced in the
domain of items, which are not frequently purchased, are
tackled by exploiting deep knowledge in the product domain
and user’s requirements by Knowledge-based
recommendation techniques.
Knowledge-based recommendations are produced based on
two approaches. The first approach is case based
recommendations, which find products from the case base that
are similar to the products described by the user’s
requirements. Past experiences that can be used to achieve the
goals of the system are represented by cases. The second
approach is constraint-based recommendation systems, which
recommends items, based on explicit rules specified as
constraints on knowledge base.
3.1 Case-based recommendations
Case-based recommendations are a form of content-based
recommendation that uses organized knowledge in the form of
cases to make recommendations to the users [15]. Case base
consists of collection of previous problem or cases that have
been solved. Each case consists of two parts. The first part
describes the problem at hand and is called specification part.
The second part describes the solution used to solve the
problem and is called solution part. To solve new problems
case whose specification matches up to the current problem
are retrieved and the solution is adapted to suit the current
problem. An approach for case based recommendation is
discussed in the following paragraph.
Daramola et al. [16] proposed an approach based on
ontological framework designed to produce knowledge-based
recommendation to the users in tourism domain. Ontologies
are developed for destination, restaurant and accommodation.
User preferences are explicitly collected from the user and
stored in the system. Past user preferences are stored as cases
in case base. Architecture presented in this approach produces
recommendations by matching current user preferences with
the past cases and by performing semantic match between the
ontological description of destinations, restaurant and
accommodation stored in knowledge base of the system.
3.2 Constraint-based recommendation
Constraint based recommendation systems are also called
Rule-based recommendation systems because they produce
recommendation based on explicitly specified constraint or
rules. The rules define the mapping between customer
requirements and item features [17]. Recommender
knowledge base consists of user requirements, item properties
and rules. Items that satisfy the rules with respect to a given
set of user requirement are generated as recommendation [18].
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
22
Rule based recommendation systems are suited for
applications where user is ready to spend more time and effort
to ensure that they get the best item recommended [19].
Mostly they are used where the user does not have sufficient
knowledge about items domain like e-government, production
systems, telecommunication switches, and financial services.
Advantages of using Rule based recommendations
Using Rule-based recommendations systems have many
advantages, such as,
1. Domain knowledge
2. Rule-based recommendations are based on domain
knowledge, which comprises all details about the
domain along with the restrictions that exist in the
domain.
3. Result explanation
4. Rule-based recommendation systems not only
provide high quality recommendations, but also
provide explanations for inclusion of item in
recommendation list based on the combination of
item properties, which lead to the selection of item.
Rule-based recommendation systems, when applied on
Semantic Web can benefit from the underlying ontologies,
which form the backbone of Semantic Web. Ontologies can
be used to represent the domain knowledge and impose
restrictions on the domain in a machine process able way.
Ontologies support reasoning which determine new
information automatically which is not explicitly specified in
the knowledge base. Reasoning characteristic of ontology can
be used to derive the explanation for the presence of items in
the recommended list. Therefore, Semantic Web provides a
better platform for implementing Rule-based recommendation
system as it fulfils and extends the advantages of Rule-based
recommendation systems [20].
4. RULE BASED PERSONALIZED
RECOMMENDATION IN SEMANTIC
WEB
Rule based Personalization systems produce
recommendations, which meet certain requirements
represented as rules. Rules form an integral part of Semantic
Web-layered architecture. Logic and proof layers support rule
layer in Semantic Web.
4.1 Significance of rules in Semantic Web
Rules provide a foundation for automated reasoning that
supports intelligent exploitation and manipulation of
information content [21]. Designing of rules to support the
functioning of Semantic Web application was identified as a
major design issue by Tim Berners Lee et al [22]. Listed
below is the significant issue involved in the design of
Semantic Web.
“Adding logic to Semantic Web {the means to use rules to
make inferences, choose courses of action and answer
questions} is the task before the Semantic Web community at
the moment."
Therefore, rules play a vital role in realizing the full potential
of Semantic Web inference mechanism. Although ontologies
support some basic form of Description Logic (DL)
reasoning, they cannot support range of knowledge-based
services that can be supported on Semantic Web [23]. To
improve reasoning capabilities beyond OWL, rules are used in
Semantic Web.
4.2 Rules-based personalization
Rule-based reasoning approaches can be used for
personalization of Semantic Web by writing rules for
implementing personalization logic. Rules represent
knowledge with conditions in some domain of logic, such as,
first order logic. A rule is defined as ‘If-then’ clause
containing logical functions and operations, which can be
expressed in a rule language. If-clause specifies the condition
or premises and then- clause specifies the conclusion or action
to be taken. If conditions are true in if-clause, then the
conclusion or action will be carried out in the then-clause.
Reasoning based approaches for personalization in Semantic
Web were first proposed by Antoniou et al. [24]. They
categorized approaches into Monotonic, Non-monotonic,
Evolution updates and events, and reasoning about actions.
Monotonic reasoning is static. The truth of statement does not
change when new information is added and this type of
reasoning is performed by DL reasoner which is based on
Open World Assumption which allows easy integration of
new information and the existing information truth value is
not affected with the addition of new information.
Non-Monotonic is the reverse of Monotonic where adding of
new information can affect truth value of existing
information. Defeasible reasoning and Answer set programs
are examples of non-monotonic reasoning systems. Defeasible
reasoning is a rule-based approach which works with
incomplete and inconsistent information [25]. It can represent
facts, rules, and priorities among rules. Answer Set Programs
are non-monotonic logic programs based on the Answer Set
Semantics, which use extended logic programs for reasoning
and problem solving by considering possible alternative
scenarios [26]. Evolution updates and events represent
dynamic aspects of personalization in Semantic Web [27].
This approach represents reactive behavior specifying actions
to be taken according to the situation by writing rules. Event-
Condition-Action paradigm is used to represent the reactive
behavior. An occurrence of a specific activity is an event,
when an event occurs, a condition is checked; if condition is
satisfied, an action is carried out.
Reasoning about actions and time is an example of temporal
reasoning where reasoning is performed about the
phenomenon that occurs in time using properties
characterizing dynamic behavior with truth-value depending
on changes occurring in the world.
Mu et al. [28] further refined the categorization proposed by
Antoniou and categorized rule-based recommendation
approaches into four major categories. Categories of Rule
based Personalization approaches are mentioned are as
follows.
1. Logical Languages
2. Event-Action -Rule
3. Expert systems
4. Rule based Inference Engine
The first category is based on extending the logical languages
to provide personalized recommendations. The second
category is the same as Evolution updates and events
discussed in the preceding paragraph where rules are written
for the occurrences of events and actions are carried out if
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
23
conditions are true. The third category is Expert systems
which consist of ‘if-then’ rules and knowledge base can be
used for obtaining personalized recommendations by applying
reasoning methods. The last category is of rule engine which
are application software used to derive new knowledge from
the existing knowledge. Personalization is done by applying
rules for reasoning over data.
4.2.1 Logical Languages
Logical languages can be extended for supporting
personalization in Semantic Web [29]. Description logic is a
decidable fragment of first order predicate logic which is used
to represent ontologies in Semantic Web [30]. Web Ontology
language which is W3C recommended language for
representing ontologies in Semantic Web is based on DL. DLs
have strong inference mechanism. Therefore they can be used
for Personalization by extending DL semantics. Some of the
approaches used for rule-based personalization using logical
languages are discussed in the following paragraphs.
Personalization performed by extending logical languages was
demonstrated by Mu et al. OWL DL is extended by writing
DL safe rule implementing personalization logic for
application. A finite set of DL safe rules is called a logic
program, which is at the higher conceptual level than the
imperative programs consisting of if-then statements. DL safe
rules are combination of OWL DL and function free Horn
rules. Rules can be modified depending on the user
requirements. A DL safe rule does not support negation and
disjunction operations in the rule.
Tran et al. [31] proposed an approach based on extending
Description Logic with one of its sublanguage ALC (Attribute
Language Complement) which supports negation and
disjunction along with conjunction existential limit and value
restriction, which were not supported in earlier approach by
Mu. To obtain personalized recommendation ALC (D) is used
which represents information in a specific domain. User
profile describing items of interest is obtained from the user,
which is matched with the resource profile describing all
items in the domain to produce personalized
recommendations.
Rule based personalization based on extending logical
language has advantage in that it has strong ability of
expression and decidability but they require intense
knowledge of logic programming to write the rules. It is
difficult to interpret the rules written in DL by the user
thereby abstracting the personalization logic.
4.2.2 Event Condition Action Rules
Event-condition-action (ECA) rules support the reactive
functionality of Semantic Web. ECA rules allow representing
applications reactive functionality to be defined and managed
in a single rule base rather than in diverse programs thereby
enhancing maintainability and modularity of application [32].
Approaches used for personalization in Semantic Web based
on Event Condition Action rules produce recommendations
based on the occurrences of events which cause triggering of
rule, and the corresponding action is carried if the condition
under which action must be carried out is true [33]. The
following paragraphs discuss some of the approaches used for
rule based personalization using event condition action rules
in Semantic Web.
Rule management ontology was created by Debattista et al.
[34] for representing rules. Rules are modeled on event-
condition-action pattern concepts in Rule management
ontology. Rules are transferred and stored in rule pool.
Context information consisting of set of events are collected
and stored as event set in log. Pattern matching is done on
rules in rule pool and events in event set. If they match,
corresponding actions are carried out to generate personalized
recommendation.
Pattern matching is computational expensive and consumes
time for computing match. Barla et al. [35] proposed an
approach that does not depend on pattern matching to
generate personalized recommendations. Instead, it utilizes
client and server log information to generate events for a
particular user and store it in user model. Events ontology is
created describing all events and the attributes associated with
the events. Event action rules contain all knowledge about
processing log of events and update the user model. Rules are
used to generate personalized recommendations by carrying
out mapping between user model and knowledge about
previous user preferences.
Approaches used for rule based personalization based on
Event–Condition-Action require context information to
generate personalized recommendations based on the
occurrence of events; they also use information stored in
client and server logs to generate recommendations.
Collection of context information from various sources is
time-consuming. Processing of log information to find
interesting pattern requires time and effort.
4.2.3 Expert systems
Expert systems are computer programs used for producing
recommendations or problem solving based on knowledge in
some domain [36]. Traditional Web-based Expert system does
not support the method for representing data in a format that
can be used for machine reasoning, which becomes a major
drawback for them [37]. This drawback can be overcome by
using Semantic Web instead of traditional Web where there is
representative format supporting representation of information
in a machine process able format. Some of the approaches
used for rule-based personalization in Semantic Web using
expert systems are discussed in the following paragraphs.
Garcia et al. [38] proposed an approach for producing
recommendation based on expert system using fuzzy logic
techniques. Customer is the users of this tourist
personalization system and is required to submit their
preferences information, which is converted into fuzzy sets by
an expert describing customer characteristics. Fuzzy rules are
evaluated to match the fuzzy representation of hotel with
customer characteristics. Results of the match are defuzzified
to obtain the concrete results according to the hotel ontology.
The best hotel is recommended based on the calculated
weights.
García approach did not use a structured storage, such as, an
ontological knowledge base to store the information about the
domain. Wanner et al. [39] proposed an approach that
overcomes this drawback and uses ontology based knowledge
base as the main data structure to store information about the
environmental domain used for designing an expert system
offering personalized support to the citizens in questions
related to the environmental conditions in their habitat.
Information is collected from the user, environmental data and
stored in ontological knowledge base in a uniform format.
When user formulate a request regarding environmental
conditions personalization is carried out by applying fuzzy
reasoning which provides pages related to the current
environmental conditions.
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
24
Rule-based personalization approaches based on expert
system require fuzzy logic integration with the description
logic reasoners to produce recommendations. Integration of
fuzzy logic requires transformation of ontology concepts into
fuzzy sets which is a cumbersome process
4.2.4 Rule based Inference Engine
Rule-based inference engines are always required for
inference mechanism based on free form rules. Rule based
inference engine when implemented in code, are called
reasoners which are software programs used to derive new
facts from the existing knowledge [40]. As Semantic Web
standards are becoming popular, there is a need for Rule-
based inference engine to support intelligent processing of
Semantic Web data.
An example of the rule engine based reasoning approach is
proposed by Doulaverakis et al. [41] represented as Panacea,
which is semantic framework for drug recommendations.
Ontologies are used to represent medical knowledge and
terminology in Panacea. It uses a layered reasoning approach
at first level OWL DL reasoner that is used to resolve the
inconsistencies in the ontological representation of medical
knowledge at second level RDF Rule reasoner that is used to
produce drug recommendations according to the set of
medical rules.
5. CONCLUSION
Knowledge based recommendation systems using ontologies
as the knowledge bases are explored in this paper. Knowledge
based recommendation system can be used in any domain.
Benefits of applying Knowledge based recommendation
system are emphasized in this paper. Rule based
personalization system, which is a category of Knowledge
based recommendation system is discussed in detail.
6. REFERENCES
[1] Berendt, Bettina, Andreas Hotho, Dunja Mladenic,
Maarten Van Someren, Myra Spiliopoulou, and Gerd
Stumme. “A roadmap for Web mining: From Web to
Semantic Web”. Springer Berlin Heidelberg, 2004.
[2] Nasraoui, Olfa. "World wide Web personalization."
Encyclopedia of Data Mining and Data Warehousing,
Idea Group (2005).
[3] Bobadilla, Jesús, Fernando Ortega, Antonio Hernando,
and Abraham Gutiérrez. "Recommender systems
survey." Knowledge-Based Systems 46 (2013): 109-132.
[4] Gedikli, Fatih, and Dietmar Jannach. "Recommender
Systems, Semantic-Based." Encyclopedia of Social
Network Analysis and Mining. Springer New York,
2014. 1501-1510.
[5] Felfernig, Alexander, Michael Jeran, Gerald Ninaus,
Florian Reinfrank, Stefan Reiterer, and Martin Stettinger.
"Basic approaches in recommendation systems." In
Recommendation Systems in Software Engineering, pp.
15-37. Springer Berlin Heidelberg, 2014.
[6] Pazzani, Michael J., and Daniel Billsus. "Content-based
recommendation systems." In The adaptive Web, pp.
325-341. Springer Berlin Heidelberg, 2007.
[7] Lops, Pasquale, Marco De Gemmis, and Giovanni
Semeraro. "Content-based recommender systems: State
of the art and trends." In Recommender systems
handbook, pp. 73-105. Springer US, 2011.
[8] IJntema, Wouter, Flavius Frasincar, Frederik
Hogenboom, and Uzay Kaymak. "News Personalization
using the CF-IDF Semantic Recommender." (2011).
[9] Bogdanov, Dmitry, MartíN Haro, Ferdinand Fuhrmann,
Anna Xambó, Emilia Gómez, and Perfecto Herrera.
"Semantic audio content-based music recommendation
and visualization based on user preference examples."
Information Processing & Management 49, no. 1 (2013):
13-33.
[10] Su, Xiaoyuan, and Taghi M. Khoshgoftaar. "A survey of
collaborative filtering techniques." Advances in artificial
intelligence 2009 (2009): 4.
[11] Aggarwal, Charu C., and S. Yu Philip. "Semantic based
collaborative filtering." U.S. Patent 6,487,539, issued
November 26, 2002.
[12] Lee Jae-Won, Kim Han-Joon And Lee Sang-Goo,
”Applying taxonomic knowledge and semantic
collaborative filtering to personalized search: a bayesian
belief network based approach”,in proceedings of 2010
12th International Asia-Pacific Web Conference IEEE.
(2010).
[13] López-Nores, Martín, Yolanda Blanco-Fernández, José J.
Pazos-Arias, and Alberto Gil-Solla. "Property-based
collaborative filtering for health-aware recommender
systems." Expert Systems with Applications 39, no. 8
(2012): 7451-7457.
[14] Fard, Karamollah Bagheri, Mehrbakhsh Nilashi, and
Naomie Salim. "Recommender system based on
semantic similarity." International Journal of Electrical
and Computer Engineering (IJECE) 3, no. 6 (2013): 751-
761.
[15] Smyth, Barry. "Case-based recommendation." In The
adaptive Web, pp. 342-376. Springer Berlin Heidelberg,
2007.
[16] Daramola, Olawande, Mathew Adigun, and Charles Ayo.
"Building an ontology-based framework for tourism
recommendation services." Information and
Communication Technologies in Tourism 2009 (2009):
135-147.
[17] 17.Felfernig, Alexander, and Robin Burke. "Constraint-
based recommender systems: technologies and research
issues." In Proceedings of the 10th international
conference on Electronic commerce, p. 3. ACM, 2008.
[18] Felfernig, Alexander, Gerhard Friedrich, Dietmar
Jannach, and Markus Zanker. "Developing Constraint-
based Recommenders." Recommender systems
handbook 1 (2011): 187.
[19] Felfernig, Alexander, Sarah Haas, Gerald Ninaus,
Michael Schwarz, Thomas Ulz, Martin Stettinger, Klaus
Isak, Michael Jeran, and Stefan Reiterer. "Recturk:
Constraint-based recommendation based on human
computation." In RecSys 2014 CrowdRec Workshop, pp.
1-6. 2014.
[20] Dell’Aglio, Daniele, Irene Celino, and Dario Cerizza.
"Anatomy of a Semantic Web-enabled knowledge-based
recommender system." In Proceedings of the 4th
international workshop Semantic Matchmaking and
Resource Retrieval in the Semantic Web, at the 9th
International Semantic Web Conference. 2010.
International Journal of Computer Applications (0975 – 8887)
Volume 182 – No. 43, March 2019
25
[21] Alferes, Jóse Júlio, Carlos Viegas Damásio, and Luís
Moniz Pereira. "Semantic Web logic programming
tools." In Principles and practice of Semantic Web
reasoning, pp. 16-32. Springer Berlin Heidelberg, 2003.
[22] Berners-Lee, Tim, James Hendler, and Ora Lassila. "The
Semantic Web." Scientific american 284, no. 5 (2001):
28-37.
[23] Smart, Paul R. "Rule-Based Intelligence on the Semantic
Web: Implications for Military Capabilities." (2007).
[24] Antoniou, Grigoris, Matteo Baldoni, Cristina Baroglio ,
Robert Baumgartner, François Bry, Thomas Eiter, Nicola
Henze et al.” Reasoning methods for personalization on
the Semantic Web”. na, 2004.
[25] Antoniou, G., Billington, D., Governatori, G., And
Maher, M.” Representation Results for Defeasible
Logic”. ACM Transactions on Computational Logic 2,2
(2002), 255–287.
[26] Heymans, S., And Vermeir, D. “Integrating Semantic
Web reasoning and answer set programming”. In Answer
Set Programming, Advances in Theory and
Implementation, Proc. 2nd Intl. ASP’03 Workshop,
Messina, Italy (2003), pp. 194–208
[27] Alferes, José Júlio, Antonio Brogi, João Alexandre Leite,
and Luís Moniz Pereira. "Evolving logic programs." In
Logics in Artificial Intelligence, pp. 50-62. Springer
Berlin Heidelberg, 2002.
[28] Mu, Xiangwei, Yan Chen, Yan Cao, and Yan Lin.
"Personalized Recommendation System Modeling in
Semantic Web." Advances in Information Sciences and
Service Sciences 5, no. 2 (2013): 278.
[29] Baader, Franz, Ian Horrocks, and Ulrike Sattler.
"Description logics as ontology languages for the
Semantic Web." In Mechanizing Mathematical
Reasoning, pp. 228-248. Springer Berlin Heidelberg,
2005.
[30] Krötzsch, Markus, Frantisek Simancik, and Ian
Horrocks. "Description Logics."IEEE Intelligent Systems
29, no. 1 (2014): 12-19.
[31] Tran, Thanh, Philipp Cimiano, and Anupriya Ankolekar.
"A Rule-Based Adaption Model for Ontology-Based
Personalization." In Advances in Semantic Media
Adaptation and Personalization, pp. 117-135. Springer
Berlin Heidelberg, 2008.
[32] Papamarkos, George, Alexandra Poulovassilis, and Peter
T. Wood. "Event-Condition-Action Rule Languages for
the Semantic Web." In SWDB, pp. 309-327. 2003.
[33] Moore, Philip T., and Hai V. Pham. "Personalization and
rule strategies in data-intensive intelligent context-aware
systems." The Knowledge Engineering Review 30, no.
02 (2015): 140-156.
[34] Debattista, Jeremy, Simon Scerri, Ismael Rivera, and
Siegfried Handschuh. "Ontology-based Rules for
Recommender Systems." In SeRSy, pp. 49-60. 2012.
[35] Barla, Michal, Michal Tvarožek, and Mária Bieliková.
"Rule-based user characteristics acquisition from logs
with semantics for personalized Web-based systems."
Computing and Informatics 28, no. 4 (2012): 399-428.
[36] P. Jackson, “Introduction to Expert Systems”, 3rd
edition. Boston: Addison- Wesley Longman Publishing
Co., 2001
[37] Verhodubs, Olegs, and Janis Grundspenkis. "Towards
the Semantic Web Expert System." Scientific Journal of
Riga Technical University. Computer Sciences 44, no. 1
(2011): 116-123.
[38] García-Crespo, Ángel, José Luis López-Cuadrado,
Ricardo Colomo-Palacios, Israel González-Carrasco, and
Belén Ruiz-Mezcua. "Sem-Fit: A semantic based expert
system to provide recommendations in the tourism
domain."Expert systems with applications 38, no. 10
(2011): 13310-13319.
[39] Wanner, Leo, Marco Rospocher, Stefanos Vrochidis,
Lasse Johansson, Nadjet Bouayad-Agha, Gerard
Casamayor, Ari Karppinen et al. "Ontology-centered
environmental information delivery for personalized
decision support." Expert Systems with Applications 42,
no. 12 (2015): 5032-5046.
[40] Singh, Swapna, and Ragini Karwayun. "A comparative
study of inference engines." In Information Technology:
New Generations (ITNG), 2010 Seventh International
Conference on, pp. 53-57. IEEE, 2010.
[41] Doulaverakis, Charalampos, George Nikolaidis,
Athanasios Kleontas, and Ioannis Kompatsiaris.
"Panacea, a semantic-enabled drug recommendations
discovery framework." In VDOS+ DO@ ICBO. 2013.
IJCATM : www.ijcaonline.org
