A powerful tool to track opinions in forums, blogs, ebusiness sites, etc., has become essential for companies, politicians as well as for customers, and that because of the huge amount of texts available which make the manual exploration more and more difficult and useless. In this paper, we present our approach of identification of opinions based on an ontological exploration of texts. This approach aims to study the role of domain ontologies and their contributions in the identification phase. In our approach, domain ontology and sentiments lexicon are needed as pre-requirements.
The views available on the Internet have a significant impact on
users, for example, if users have already researched opinions on
a product, they are willing to pay more for a product whose
opinion is more favorable than another, and the product will be
more marketed than another whose opinion is less favorable
[
Companies, politicians, and customers need a powerful tool to
track opinions, sentiments, judgments, and beliefs that people
can express in blogs, comments, or in the form of texts, toward a
product, a service, a person or an organization, etc. [
Generally, research works in this area can be grouped into three main categories:
Development of linguistic and cognitive models for
opinion mining where all approaches based on
dictionary or corpus are used automatically or
semiautomatically to extract opinions based on the semantic
orientations of words and phrases [
This article focuses on identification and classification of opinions in Arabic texts, which aims to calculate the semantic
To answer these questions, we propose in this paper to study the role of ontologies used in opinion mining, and more specifically, our goal is to study how domain ontology can be used to:
The paper is organized as follows: We present in Section 2, state of the art of the main approaches used in the field and the motivations of our work. We present in the next section, our approach and the general architecture of opinions identification process. 2 1.1
Overall, two main types of work are distinguished, those that are based on simple features extraction from the texts, and those who organize features into a hierarchy using taxonomies or ontologies. The extraction process mainly concerns explicit features. We can distinguish two main families: Opinion Mining
All approaches that do not use knowledge representation
models are based on the use of algorithms to discover the
different characteristics of a product or an object. Only the
expressions of opinions (adjectival and adverbial) are
extracted, then a summary is produced to show for each
characteristic, the positive and the negative opinions and the
total number of these categories [
The main limitation of these approaches is that there is a large
number of extracted features and a lack of organization. In
addition, similar concepts are not grouped (for example, in
some domains, the words “دعوم ” and “ءاقل” witch have the
same meaning “appointment”), and possible relationships
between the features of an object are not recognized
(example: “جوهق” “coffee” is a specific term of “بسش”
“drink”). Thus, analysis of polarity (positive, negative or
neutral) of the text is done by assigning the dominant polarity
of opinion words, regardless of the polarities associated with
each feature individually [
The family itself can be divided into two subfamilies:
(a) Use of Taxonomies
This kind of approaches does not seek a list of features, but
rather a hierarchical organized list by the use of taxonomies.
We recall that a taxonomy is a list of terms organized
hierarchically through a sort of “is a kind of”. In [
Generally, the use of taxonomies is coupled with a classification technique; the sentences corresponding to the leaves of the taxonomy are extracted. At the end of the process, a summary that can be more or less detailed is produced. (b) Use of Ontologies These approaches aim to organize the features using elaborated representation models. Unlike taxonomies, ontology is not restricted to a hierarchical relationship between concepts, but can describe other types of paradigmatic relations such as synonymy, or more complex relationships such as relations of composition or spatial relationships.
Generally, the extracted features correspond exclusively to
terms contained in the ontology. The feature extraction phase
is guided by a domain ontology, built manually [
Similar features are grouped together using semantic similarity measures.
Based on the conceptual model described in [
In a review, the opinion holder comments a set of features of an object or a product using opinion expressions. Each feature corresponds to a concept or a property in the ontology O.
Ontologies have also been used to support polarity mining.
For example, in [
In [
Structuring of features: Ontologies are tools that provide a lot of semantic information. They help to define concepts, relationships, and entities that describe a domain with an unlimited number of terms; Extraction of features: Relationship between concepts and lexical information can be used to extract explicit and implicit features. 3 1.3 elements:
For each studied domain, our approach requires three basic A domain ontology O, where each concept and each property is associated to a set of labels that correspond to their semantics; A lexical resource L of opinion expressions;
For example in the following comment, the EDUs are between square brackets, the EOUs are underlined, and the characteristics of the object are in bold. There is an inverse relationship between the EDUa and the EDUb, representing the review expressed in the EDUd. a[فتاه زاهج تيستشا ، سمأ موي ] b[اشاتمم فتاهلا ناك اذإ ىتح]
c[ادج طيست ميمصتلا ناف ] d[ةمعلالا يره يف لاملآل ةيخملا ءيشلا ] [Yesterday, I purchased a phone] a. [Even if the phone is excellent]b, [the design is very basic]c, [which is disappointing in this mark]d. This step aims to extract for the comment all the labels of the ontology. As each concept is an explicit feature, we simply project the lexical components of the ontology on the text to obtain, for each EDU, all the features. To extract the implicit features, ontology properties are used. We recall that these properties are to define the relationships between concepts of the ontology. For example, the property “قوسي ”,“drive” links the concepts “قئاس ”,“conductor” and “جزايس”,“car”. the used opinions expressions. For example, if our lexicon contains the concept “حعيثط”, “nature”, and sentiments lexicon contains the word “بخلا ”, “amazing”, from the EDU “حتلاخ حعيثط ”, “amazing nature”, it is easy to extract the couple (حعيثط, حتلاخ ), (nature, amazing) from the text.
Expressions: Expressions, as in the EDU “حلوثقم جئاتو”, “acceptable results”, where the opinion word “لوثقم ”, “acceptable” was not extracted in step (a) (see section 3.1). In this case, the lexicon of opinions can be automatically updated with the recovered opinion word.
Opinion Expressions: As in the EDU “حعئاز حيسطم حتاغ”, “wonderful rainforest” where the feature “حيسطم”,“rainforest” has not been extracted in step (b) (see section 3.1), in this case, the domain ontology can be updated by adding a new concept or a new property in the right place.
“It‟s slow”. This kind of EDU expresses an implicit feature. In this case, we use the ontology properties to retrieve the associated concept in the ontology. Features Only: An EDU with features alone can also be an indicator of the presence of an implicit opinion expression towards the feature as in “ أجلم تحثصأ حقيدحلا هيفسحىمل”ل, “the park became a haven for perverts”, witch express a negative opinion towards “حقيدحلا”, “the park”. 1.4
In this section, we present the general architecture of our approach and the different modules constituting our system: Sentiments Lexicon
EOUs Extracting
EOUs
EDUs Segmentation
Texts
EDUs Features and EOUs
Associating
Classification Classification Result
Domain
Ontology Features Extracting
Features
Classification Techniques
As indicated in the last figure, our system contains the following modules: Texts EDUs Segmentation: Generally, extraction of elementary discourse units (EDUs), depends on the use of delimiters such as “.” , “,”, “?” “!”; EOUs Extracting: Elementary opinions units EOUs and semantic orientations are usually extracted using a lexicon of emotions specific to domain of study; Features Extraction: Features can be extracted by a simple projection of the ontology on the elementary discourse units (EDUs);
feature should be associated to one or more elementary opinions units in order to extract its semantic orientation; Classification: The last phase of our work is to classify the identified opinions into positive or negative classes using supervised classification techniques. 4
In this paper we presented our approach based on an ontological exploration of Arabic texts. Our method is promising because the use of ontologies improves the extraction of features and facilitates the association between opinions expressions and opinionated features of the object. On the one hand, domain ontology is useful within its list of concepts which carry much semantic data in the system. The use of ontology concepts labels can recognize terms that refers to the same concepts and provides a hierarchy between these concepts. On the other hand, ontology is useful to its list of properties between concepts that can recognize the opinions expressed on the implicit features.