Customer Feedback System
Evolution towards Semantically-enhanced Systems
Oleksiy Khriyenko
Industrial Ontologies Group, Department of Mathematical Information Technology, University of Jyväskylä,
P.O. Box 35, FIN-40014, Jyväskylä, Finland
Keywords: Semantic Customer Feedback, Automated Customer Feedback Processing, Semantic Personalization,
Semantic User Profile.
Abstract: The digital economy requires services be created in nearly real time – while continuously listening to the
customer. Managing and analysing the data collected about products and customers become very critical.
Successful companies must collect data regarding customer behaviour in a sensible manner, understand
their customers and engage in constant interaction with them. Nowadays, having a huge data storage
capacity, everyone collects data and hopes that it will be useful someday. But, it is frustrating when you do
not know whether something useful will come out of it. It is not a problem to collect data, but it is very
difficult to analyse it. To utilize the data they collect and analyse customer feedback quickly, companies
require automation of customer feedback processing. To hear a real voice of a customer, companies are
trying to engage customer to the feedback provisioning process. Therefore, the paper reviews digitalized
customer feedback strategies, highlights challenges of a feedback gathering and further computation. As a
result, paper presents an approach for semantic enhancement of a customer feedback system.
1 INTRODUCTION
To create successful innovative business and
succeed with new product or service, company must
listen to customer feedback. Customers can help to
develop a better product, to provide a better service,
to bring purpose to the product or valuable service
offering, to tell how they really feel about a product
or service and provide the best advice. By having
customer insight, listening to customers and making
them happy, company can create strong and long-
term relationships with customers getting lifetime
revenue, because people do business with people
they like, know and trust.
There are a number of actionable strategies to get
feedback from customers. There are some old
fashion approaches such as: quarterly business
reviews, paper-based customer surveys, telephone
surveys, personal emails, etc. In some cases, these
strategies give higher survey response rate because
they are more personalized. From other side, these
methods are quite expensive, because they usually
require human involvement and further adaptation
for automated machine processing. For example,
hard copy customer feedback forms might be a good
way to be used during business meetings or
exhibitions. But, further computation and processing
of the feedbacks will require a lot of manual work
from experts, unless some more advanced tools and
techniques (e.g. Digital Pen and Paper ) are used.
We are living in a digital era when business and
social life more and more adopt solutions and tools
of digital world. Therefore, in this paper, we will
concentrate on digitalized strategies for customer
feedback gathering and analysis. Nowadays, many
companies apply digitalized methods to collect
customer feedback themselves or request
corresponding service from third-party companies.
Among various digital strategies that support
customer feedback gathering and analysis, we may
highlight such strategies as: website feedback tools,
crowdsourcing feedback tools, online survey tools,
social web based feedbacks, website activity and in-
app feedbacks, etc. Generally, feedback tools allow
creation customized feedback forms for surveys,
polls, quizzes and ratings, using multiple-choice
questions, ranking lists, votes, matrix forms and
open free-text response areas. Some of the tools (for
example UserSnap, Bugmuncher, etc.) use visual
features to provide visual feedback (highlights and
selections of webpage elements) along with
associated comments/annotations. Dashboards give
518
Khriyenko O..
Customer Feedback System - Evolution towards Semantically-enhanced Systems.
DOI: 10.5220/0005480505180525
In Proceedings of the 11th International Conference on Web Information Systems and Technologies (WEBIST-2015), pages 518-525
ISBN: 978-989-758-106-9
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
access to ratings and comments, enabling to filter the
feedback by categories, monitor usability, tweak
designs, and track the success of new iterations.
By adopting current digitalized methods and
techniques, we speed-up and increase a scale of
feedback collection process, but we do not solve
related to this problems (e.g. automated analysis of
customer feedback). There are challenges that stay
unsolved yet. Therefore, paper touches such
challenging issue as: fruitful customer engagement
into feedback provisioning process, and automation
of feedback analysis process. The next section
provides analysis of digitalized customer feedback
strategies and highlights corresponding challenges.
Section 3 presents an approach towards
Semantically-enhanced Customer Feedback System
facilitated by Semantic Feedback Framework.
2 CUSTOMER FEEDBACK
CHALLENGES
Despite the fact that there are a lot of various tools
and platforms to support customer feedback
(suggestions, ideas, etc.) gathering exist nowadays,
there are still a lot of challenges and bottlenecks
with respect to this process. The most significant of
them are: customer engagement into a process of
feedback provisioning; simplification of this process
for him/her (feedback provider) via making it more
personalized, more intuitive and unobtrusive;
minimization of human/expert involvement into a
process of feedback analysis by feedback consumer
via automation of feedback computation.
To attract customers and make them willing to
provide a feedback, companies try to apply different
approaches. Some companies maximize amount of
possible feedback provisioning channels to keep a
close contact with a customer, no matter what kind
of communication device, location or further actions
customer has. But, all this multi-channel strategies
and obtrusive behavior do not guaranty willingness
of customers to provide a feedback. Another strategy
to attract customers to provide a feedback is to offer
customers monetized compensation and discount
coupons, or participation in a lottery. Taking into
account psychology of people and possibility to get
something valuable for little time spent, this strategy
seems to be more successful. But, again, nobody can
guaranty that people honestly answer all the
questions of surveys and do not chose answer
randomly. “One click feedback” strategy does not
require more than few seconds to share own opinion
by clicking “like” or “dislike” button or clicking a
“star” to rank an associated content. But this strategy
brings a benefit only in some specific cases,
because, generally, simple “like/dislike” or “rank”
based strategies are not sufficient for companies who
require more comprehensive and specific feedback.
To get comprehensive feedback from customers,
to get some suggestions or shared ideas, we have to
deal with customers who are interested in
product/service improvement and willing to spend
time for that. One way to achieve this is to recognize
any inconvenience, trouble or difficulties that
customer experiences at particular moment and help
him/her by asking associated questions. In case you
succeed with problem detection and customer starts
“conversation”, you immediately have valuable
feedback and have a customer who will provide you
more useful information while you or your
automated support system provides him/her
necessary suggestions. Therefore, we have to supply
customer with appropriate communication channel
to help him/her to specify occurred problem.
Customer’s activity, aimed at searching of problem
solutions, might be considered as a certain form of a
customer feedback on inconvenient functionality or
design of a product/service. Since we are talking
about automated solutions, system should deal with
a product/service description (product functionality
and features), as well as problems definitions
(provided by customer), in machine readable form to
automate appropriate matching. Thus, customer
feedback supportive system should supply customer
with a tool for problem definition and provider with
a tool for product/service description, and further
transform them into machine readable form.
Nowadays, Semantic Web technology (Berners-Lee
et. al., 2001)(Semantic Web, 2001) might be
considered as one of the most promising approaches
for this purpose, enabling automated integration and
computation of data on semantic level using
corresponding domain ontologies and semantic
matching/alignment techniques (Shvaiko and
Euzenat, 2012)( Jain et.al., 2010).
What might be a good purpose (additionally to
already mentioned) that inspires customer to provide
a feedback and share own opinion about
product/service? It is a “believe”, believe of a
customer in a fact that his/her feedback (suggestion,
preferences, etc.) will be taken into account and,
someday, will be paid back. People are not willing to
spend time providing feedbacks to the
products/services that they have bought or have used
already. Because companies do not apply business
models where customer gets new improved version
CustomerFeedbackSystem-EvolutiontowardsSemantically-enhancedSystems
519
of a product (he/she has bought) for free or with
valuable discount. But, customers might be
interested in co-creation of a new product/service
that meet their expectations and preferences, as well
as be interested in improvement of existing services
they are using. Thus, we need a mechanism that
collects customer’s preferences/suggestions with a
purpose of further influence on: improvement of
existing and/or already used services; elaboration of
a new product/service that meets expectations and
needs of customers; improvement of advertisement
process towards more intelligent and personalized
one. From one hand, this mechanism should be
centralizer in a sense of focusing at particular
customer. From the other hand, it should be
distributed among various systems and services used
by him/her. Thus, such mechanism could be
considered as a distributed system that creates or
updates a personalized customer profile of a person;
provides aggregated customer profile for product
developers and service providers (based on context-
dependent target group request); behaves as a
personal assistant providing personalized
advertisements for customer.
While collecting feedbacks after product
purchase (service use), not much customers are
really willing to spend extra time after (especially if
really interesting for the customer aspects might
constitute only a little part from the scope of whole
feedback form). Therefore, assuming that customer
is interested in further extension/update of his/her
personal customer profile, feedback collection
mechanism must support proactiveness of a
customer in the feedback provision process and
provide a possibility to initiate provisioning of a
feedback at the moment considered by customer as a
right and suitable one. It means ability of a customer
through pointing at any part of visual representation
of a product, highlighting certain concept
(meaningful word) or piece of a text to access
feedback provisioning tool with respect to associated
feature/functionality of the product/service.
Let us consider some examples. Many websites,
internet shops and aggregators use automatic
localization of a customer and change not only the
language of product/service description, but also
change a currency, associated price conversion,
transportation options, etc. Sure, such intelligent
personalization avoids extra location definition by
user. But, in context of globalization and
international mobility, such automated localization
adds inconvenience for non-local users. Therefore,
instead of having corresponding issue among many
others in a separate customer feedback form, it
would be more logical to allow user to provide a
corresponding feedback exactly at the moment
he/she manually change localization. Another
example might be beneficial for product developers.
Let us imagine a person who is looking for a kettle
of particular color and with unheated surface feature.
In case, the person cannot find appropriate product,
he/she should be able to specify concrete needs and
preferences directly from the current product page
via appropriate access points: by pointing to the
colored part of the product image, be able to find
and specify a color property among a list of the
properties/features associated with selected part of
the product; by selecting corresponding piece of a
text that describes some features of the product
surface, be able to specify “unheatedness” as a
desired feature of the kettle’s surface. One more
example might be taking from on-line e-Learning
domain. It is a normal practice to ask students to
provide a feedback at the end of the online-course by
filling certain form. This form might not always
concern exact issues that are important for students.
Student might not remember all the problems related
to the content or the study process of the course at its
end. It would be reasonable to allow student to stress
the problem immediately when it appears, does not
matter whether it is unclear explanation of certain
topic or task definition, a gap of the student’s
knowledge in certain topic/subject, or a schedule for
the tasks performance, etc. So, by getting contextual
access to appropriate/associated part of the feedback
directly from the course materials, student can
dynamically provide valuable input for the course
instructor. As we can see, such approach does not
only engage customers into feedback provision
process, allowing them proactively behave to specify
personal needs, preferences and desires via more
intuitive and context-aware interface, but also uses
human intelligence for natural extension of a
feedback model in dynamic and co-creative way.
To survive in highly competitive environment
companies should understand that there are no any
other bosses than customers. It brings new
challenges because customers want to talk to
organizations in their own words at a time and place
convenient to them. One of the best ways to collect
data from the customers is to give them the
opportunity to recall/retell and share their
experiences in their own words. “Don’t box them in
by predefined questions that might not always be
appropriate,” said Shayne Paddock, Chief
Information Officer of ZDirect Company that
provides hotel management solutions and Hotel
Marketing Automation tools in particular. To
WEBIST2015-11thInternationalConferenceonWebInformationSystemsandTechnologies
520
provide better services and make more accurate
decisions, company should understand the thoughts
and feelings of a targeted group of people. With
respect to automation of review analytics, opinion
Observer system, presented in (Liu B. et. al., 2005),
helps potential buyer to compare different customer
opinions with respect to the target product(s). Other
relevant researches have been done with respect to
the sentiment classification (Dave K. et.al, 2003).
Current tools automate customer feedback
analytics in a scope of structured data. The problem
is that we collect only information that we managed
to structure before and miss the rest information that
might be crucial. Current automation of existing
customer feedback supporting systems restricts
customer with a defined set of answers or possible
options to be chosen. Allowing some restrictions and
putting some boundaries to the systems, we are able
automation of a process, but we also get restricted
scope, restricted outcome, restricted achievements…
This problem concerns not only customer feedback
domain. According to (Mcdonald J. et. al., 2012),
most computer-assisted assessment involves
students being able to recognize a correct response
rather than recall and independently generate an
answer. We automate a process, but with restrictions
caused by this automation, we do not approach the
final goal. In case of customer feedback systems,
predefined set of options means that feedback
consumer knows possible opinions of the customers
in advance and uses their feedbacks only to get a
statistics. But, many companies nowadays are
looking for more. They would like to hear actual
opinion, new suggestion and idea, new knowledge
from the customer that could not be provided via
fixed predefined questioner forms. Thus, we should
more consider meaning of free-text customer
feedbacks, assuming that other forms of feedbacks
are more or less structured already.
Automation or semi-automation of customer
feedback analysis and its further computation is
possible only if content can be understood not only
by human (expert), but by machines as well.
Semantic annotation of customer feedback becomes
very crucial for automation of its analysis.
Nowadays, innovative customer feedback systems
should adopt semantic technologies and support
semantic annotation of a content. It is very hard, if
not impossible, for any automatic technique to
achieve perfect accuracy due to the difficulty of
natural language understanding. Systems that need
near-perfect solutions require convenient user-
friendly mechanism for human involvement to
correct errors made by automatic techniques. It is
much more reasonable to ask user what (s)he meant,
rather than to ask some external expert about the
same later.
3 SEMANTICALLY-ENHANCED
CUSTOMER FEEDBACK
FRAMEWORK
Semantic Web technologies will work in a full
extend and bring benefit to the society welfare only
if information will be presented in machine readable
semantic form. To reach the Web with semantically
annotated data, we should follow Semantic Web
paradigm and not only transform existing old data
into machine readable form, but produce new data
and knowledge in a form already suitable for
automated processing, reuse and shearing by
machines (semantically oriented software). Both
cases require sophisticated semantic annotation tools
to adapt existing and create new content. Taking into
account huge amount of existing human oriented
content, a process of its further semantic annotation
could not be associated with duties of content
owners/creator only. In contrast, creation of new
semantically ready (annotated) content will be
considered mainly as a duty of content creator.
Taking into account that content creators are not
domain experts or knowledge engineers, we need
simple, effective and very handy tools to support
users to produce semantically ready content. In the
future, when people will study methods of semantic
annotation (most probably even at the school level),
we will be able to use more sophisticated and
professional methods and techniques imbedded into
all the content creation tools. For the moment, the
main tool people utilize to express their feelings,
attitudes, thoughts providing a feedback is natural
language. Thus, we have to apply natural language
processing (NLP) techniques simultaneously follow
a process of initial content (feedback) creation,
making it semantically ready in (semi)automated
way via suggestion of appropriate semantic
transformations.
Automation of unstructured free-text feedback
analysis requires advanced and intelligent
mechanisms for its semantic annotation, however,
even structured elements of customer feedbacks lack
for semantic annotation to be further integrated with
other semantically related data and services.
Semantic Text Analysis enables to "understand" the
natural language statements provided in a free-text
feedback format (Jurafsky D. and Martin J.H., 2000)
CustomerFeedbackSystem-EvolutiontowardsSemantically-enhancedSystems
521
(Allen J.F. et.al., 2008). Among the various NLP
techniques, opinion mining and sentiment analysis
(Gamon M., 2004) might be especially valuable for
customer feedback analysis. One of the solutions is
to design and develop methods that enable the
automated annotation of plain text with ontology
concepts (Cherfi H. et. al., 2008). One of such
methods (Zavitsanos E. et. al., 2010) is based on the
pre-processing of the input text with techniques that
extract semantic information from text (e.g., word
senses) using knowledge bases, like the WordNet
thesaurus, and the Wikipedia electronic
encyclopedia. There are also some unsupervised
pattern based techniques that automatically annotate
text with existing ontology concepts without using
any type of learning: (C-)PANKOW (Cimiano P. et.
al., 2005) and ONTEA platform (Laclav´ık M. et.
al., 2009). NiosTo - a software application that
implements an opinion word extraction algorithm as
well as a dictionary-based sentiment classification
(Agathangelou et. al., 2014). Initial extraction and
further annotation of instances in customer
feedbacks allow basic classification and clustering
of the feedbacks to simplify further analysis for
expert. But, the extracted data goes beyond the
entities; it includes their properties and their
relationship graph as well. It can be further analyzed
for a more precise interpretation and reasoning based
on semantic description of application domain and
particular product or service. It might build new
knowledge with respect of product functionality or
possible additional use case. Free-text form feedback
typically lacks strict structure, but in many cases it
carries structured or semi-structured information.
The state-of-the-art information extraction (IE)
techniques are usually meant for web documents or
news. Customer feedbacks are different in the sense
that they often contain references to product- or
service-specific functionality, goals, features,
usability, etc.
To minimize level of human involvement and
allow automation of customer feedbacks or
suggestions analysis, we should minimize their free-
text part and transform it into semantically enriched
machine readable form. Feedback should contain
explicitly defined semantic meaning to allow further
automated computation. In the previous section we
already discussed challenges of feedback analysis
automation and highlighted importance of text
mining/analysis to make content more structured and
meaningful. Existing text analysis software could be
considered as a good separate tool for domain
experts, but they are too far to be imbedded to any
other application/service as a part of its functionality
and be easily used by ordinary user.
Since organizations do not have the capacity to
employ professional annotators to make customer
feedbacks machine readable and ready for
automated analysis, we have to involve customer to
this process. It would be reasonable to utilize users’
capabilities (knowledge and experience) and involve
them into the process of semantic annotation of a
feedback (in other words, allow them to provide
semantically-enriched feedbacks). At the same time,
it is a non-trivial exercise for an ordinary feedback
provider as well as an average web content provider
(who is not an expert in knowledge modeling and
representation) to provide semantically annotated
content. Therefore, it is crucial to provide feedback
providers with easy to use interfaces that simplify
the annotation process, placing annotation in the
context of their feedback provisioning process.
However, not only content of customer a feedback
needs to be semantically annotated, but whole
feedback should be present in machine readable
form to be further used by other
applications/services and integrated with other data.
Thus, we have to elaborate a framework (see Figure
1) around semantically-enhanced customer
feedbacks, including: ontology as a basis for
semantic enhancement of customer feedback, tools
for feedback consumers (product/service providers)
to make digital content semantic feedback enabled
and prepare corresponding semantic feedback
template, tool for feedback provider (customer) to
provide semantically enriched feedback by
populating corresponding template with actual
inputs, and tool for further processing and
management of semantic feedbacks.
Digital Content Enhancement Tool (DCE tool)
allows digital content creator to annotate/associate
various elements (or parts of them) of the content
(image, video/audio, buttons, links, etc.) with
corresponding concepts from domain ontology.
Extended with appropriate JavaScript package and
corresponding bindings of ontology concepts and
elements of a digital content, this digital content
becomes ready to support intuitive and proactive
feedback provisioning. Whenever user clicks at
mapped part of an image, or choses certain time
frame of a video/audio file, or focuses at button or
link, and activates “feedback provisioning” function
via contextual menu, he/she will be redirected to
appropriate semantic feedback provisioning form
(SF Form). Due to the binding of selected element to
particular semantic concept of domain ontology,
system naturally presents associated concepts to the
user and allows further annotation of corresponding
WEBIST2015-11thInternationalConferenceonWebInformationSystemsandTechnologies
522
Figure 1: Semantic enhancement of Customer Feedback Framework.
properties (Concept browser). To be even more
intuitive, system filters and sorts out elements of
ontology and presents the most relevant of them.
Thus, referring to our examples from the previous
section, user will be able to specify properties of
thekettle (color, surface heating ability, etc.), to
block automated localization function of websites,
etc. In case of highlighting a free-text content or
free-text based answers of a user, system applies
NLP and automated semantic annotation techniques
(e.g. OnTeA, RDFaCE, DBpedia Spotlight,
Semantator, etc.) to find relation between the text
and concepts of domain ontology. Thus, system
builds corresponding bindings (associations) of the
text with ontological concepts on the fly, and
redirect user to the same semantic feedback
provisioning form. Additionally to semantic
enhancement of digital content, feedback consumer
is supplied with SFTC tool to create semantic enable
customer feedback form - Semantic Feedback
Template (SFT). SFT is a RDF file that contains
semantically enriched feedback template, ontology-
driven structure with possibly defined annotation
options.
Gathering all necessary tools and formats under
Semantic Feedback Framework, the framework
requires corresponding ontology for semantic
enrichment of a customer feedback. Semantic
Feedback (SF) and Semantic Feedback Template
(SFT) consist of a set of functional and non-
functional properties. Non-functional attributes
describe a purpose and target group of a feedback,
feedback requestor, time, place, etc. Functional part
is a set of feedback elements that presents a structure
of corresponding survey, poll, quiz, rating, etc. In
contrast to SFT, SF instance contains actual inputs
(values) from a feedback provider as well as an extra
non-functional properties related to feedback
provider. Structured elements of feedback are
presented by corresponding subclasses that present
multiple-choice questions with a set of predefined
options, ranking lists, votes, matrix forms, etc. In
turn, unstructured elements are text-, visual-, sound-,
video-, and emotion-based elements of customer
feedbacks. Taking into account, that any text is a set
of small meaningful pieces, text-based element is
presented as a set of such pieces - knowledge
statements (RDF triples). We did not research other
types of unstructured elements in detail, because it
was not in a scope of the presented research. Further
elaboration of this issue is left for future work.
Being created, SF is further processed by
Semantic Feedback Processing and Management
Tool (SFPM tool). Applying SF and corresponding
ontology alignment on top of semantic description of
the initial product/service, SFPM tool provides
customized, improved or suggested product
description update for product/service provider. At
the same time, SF becomes a source for personal
customer profile (PCP) update in the PCP Storage.
Storage and PCP Management System allows
management of own PCP for users and provides
aggregated customer profile (ACP) based on target
group request from product/service providers. ACP
could be used by SFPM tool to generate new
product/service that meets customer needs and
expectations. Similarly, Personalized Advertisement
Module (PAM) can push personalized
advertisements to potential customers by matching
PCPs with product/service descriptions.
CustomerFeedbackSystem-EvolutiontowardsSemantically-enhancedSystems
523
4 CONCLUSIONS
The paper tackled some of the challenges of
customer feedback gathering and its automated
processing. Nowadays, companies are looking for
new strategies and techniques to engage customer in
collaboration process making the process attractive
and friendly for them. They require automation of
customer feedback analysis and approach that allows
retrieving of new knowledge out of collected
feedback and suggestions. Therefore, paper
discussed possible steps to meet highlighted
challenges and proposed an approach of semantic
enhancement of customer feedback framework.
Involvement of customers into collaborative product
review and feedback provisioning process will
provide a deeper understanding of their needs and
increase the likelihood that the new products will
meet customer’s needs. Earlier customer
involvement into the process of customer feedback
semantic enrichment might dramatically facilitate
automation of feedback processing.
The vision of a Semantic Web has been proposed
to annotate web resources with semantic mark-up,
using knowledge representation languages, such as
RDF(S) or OWL. Analogically to WWW, we adopt
Semantic Web technologies to facilitate automated
analysis and computation of customer feedbacks.
Representation of a customer feedback in machine
readable form with appropriate semantic annotation
(especially human oriented free text part of
feedback) will not only allow machines
automatically manipulate with the content, but also
retrieve new knowledge out of it and make it
available to other systems for collaborative analysis
and unexpected results. Referring to Dr. Kenji
Takeda’s statement “What’s interesting if you
publish data and make it freely available to
everybody, so truly open, the people who use this
data are not necessarily the ones you think of”, we
make customer feedback an interoperable and
sharable piece of information.
ACKNOWLEDGEMENTS
The research is done in the Agora Center (University
of Jyvaskyla, Finland) in collaboration with Inno-W
Company under the Need4Speed program in
DIGILE SHOK (funded by TEKES and consortium
of industrial partners).
REFERENCES
Agathangelou P., Katakis I., Kokkoras F., Ntonas K.,
2014. Mining Domain-Specific Dictionaries of
Opinion Words, In 15th International Conference on
Web Information System Engineering (WISE 2014),
Thessaloniki, Greece, 12-14 October, 2014.
Allen J.F., Swift M., Beaumont W., 2008. Deep semantic
analysis of text, Proceedings of the Conference on
Semantics in Text Processing, p.343-354, September
22-24, 2008, Venice, Italy.
Berners-Lee T., Hendler J., Lassila O., 2001. “The
Semantic Web”, Scientific American 284(5), pp.34-43.
Cimiano P., Ladwig G., Staab S., 2005. Gimme' the
context: context-driven automatic semantic annotation
with c-pankow. In WWW '05, pages 332-341, NY,
USA, 2005. ACM Press. ISBN 1-59593-046-9.
Cherfi, H., Corby, O., Faron-Zucker, C., Khelif, K.,
Nguyen, M. T., 2008. Semantic Annotation of Texts
with RDF Graph Contexts. In ICCS Supplement.
Dave, K., Lawrence, S., Pennock, D., 2003. Mining the
Peanut Gallery: Opinion Extraction and Semantic
Classification of Product Reviews. WWW’03.
Gamon M., 2004. Sentiment classification on customer
feedback data: noisy data, large feature vectors, and
the role of linguistic analysis. In Proceedings of the
20th international conference on Computational
Linguistics. Association for Computational
Linguistics, 2004, p. 841.
Jain P., Hitzler P., Sheth A.P., Verma K., Yeh P.Z., 2010.
“Ontology Alignment for Linked Open Data”. In:
Proceedings of the 9th International SemanticWeb
Conference, ISWC 2010, Shanghai, China, November
7-11, 2010, Springer-Verlag, 402–417.
Jurafsky D., Martin J.H., 2000. Speech and Language
Processing: An Introduction to Natural Language
Processing, Computational Linguistics, and Speech
Recognition, Prentice Hall PTR, Upper Saddle River,
NJ, 2000.
Laclav´ık, M., Maynard, D., 2009. Motivating Intelligent
Email in Business: An Investigation Into Current
Trends for Email Processing and Communication
Research. In E3C Workshop; IEEE Conference on
Commerce and Enterprise Computing; pp. 476–482.
Liu B., Hu M., Cheng J., 2005. Opinion observer:
analyzing and comparing opinions on the web. In
proceedings of the 14th international conference on
World Wide Web, Chiba, Japan, May 10 – 14, 2005,
ACM Press New York, NY, 342-351.
Mcdonald J., Knott A., Zeng R. 2012. Free-text input vs
menu selection: exploring the difference with a tutorial
dialogue system. In proceedings of the Australasian
Language Technology Association Workshop.
Dunedin, New Zealand, December 2012, pp.97-105.
Semantic Web, 2001. URL: http://www.w3.org/2001/sw/
Shvaiko P., Euzenat J., 2012. Ontology matching: state of
the art and future challenges. IEEE Transactions on
Knowledge and Data Engineering, 2012.
Sreerama K., 1998. Murthy, Automatic Construction of
Decision Trees from Data: A Multi-Disciplinary
WEBIST2015-11thInternationalConferenceonWebInformationSystemsandTechnologies
524
Survey. Data Mining and Knowledge Discovery, v.2
n.4, p.345-389, December 1998.
Zavitsanos E., Tsatsaronis G., Varlamis I., Paliouras G.,
2010. Scalable semantic annotation of text using
lexical and web resources. In Artificial Intelligence:
Theories, Models and Applications (pp. 287-296).
Springer Berlin Heidelberg.
CustomerFeedbackSystem-EvolutiontowardsSemantically-enhancedSystems
525
