Elsevier

Information Sciences

Volume 476, February 2019, Pages 222-238
Information Sciences

An automatic procedure to create fuzzy ontologies from users’ opinions using sentiment analysis procedures and multi-granular fuzzy linguistic modelling methods

https://doi.org/10.1016/j.ins.2018.10.022Get rights and content

Abstract

The high amount of information that users continually provides to the Internet is unorganized and difficult to interpret. Unluckily, there is no point in having high amounts of information that we cannot work with. Therefore, there is a need of methods that sort this information and stores it in a way that can be easily accessed and processed. In this paper, a novel method that uses sentiment analysis procedures in order to automatically create fuzzy ontologies from free texts provided by users in social networks is presented. Moreover, multi-granular fuzzy linguistic modelling methods are used in order to select the best representation mean to store the information in the fuzzy ontology. Thanks to the presented method, information is transformed and presented in an organized way making it possible to properly work with it.

Introduction

Since the appearance of Web 2.0 technologies [1], [2], the amount of information that is stored in the Web has increased dramatically. This is because in the new Internet framework, the user is the main character and the source of all the information that is present on the Web. While in its beginnings, Internet was used mainly for consuming information that was posted by a small group of people, now it is a mean to share any thought, opinion and feeling that the users are experiencing. All this information is extremely valuable but difficult to interpret and make use of due to two main reasons [3]:

  • Subjective nature of the information: The information that the users provide to the Internet is more related to their own opinions and feelings than to specific and measurable facts and objects.

  • Information is not formatted: Users like to express themselves using free text. Therefore, they do not follow any formatted way of exposing their arguments when expressing themselves on the Internet.

Since computers are built to deal with numerical and formatted information, it is difficult for them to understand and interpret all these non-formatted and subjective opinions and concepts. Since most of the information that is present on the Internet is represented in a conceptual or subjective way, there is a need of methods that are capable to transform the data in a way that computational systems can understand and process. Information provided by users on the Internet is extremely valuable since if it is correctly treated and organized, other users can benefit from this overall collective knowledge. In this paper, a novel method that overcomes all these issues and allows the collective knowledge information to be represented in a manageable way is presented.

One way of interpreting users’ opinions is by the use of sentiment analysis procedures [4], [5]. Thanks to them, it is possible to analyze, in way that the computer is able to understand, the kind of imprecise information that the users habitually provide on the Internet. Basically, these procedures are able to measure the sentiment that the user is experiencing when writing an specific text and providing an specific value to the system. Thanks to sentiment analysis, the system can easily interpret and manage the information. Therefore, sentiment analysis procedures have become an indispensable mean when dealing with users subjective opinions.

Extracting the information from the Web is just the first step in order to deal with the users’ opinion information. Once that the information is extracted, there is a need to store it in a organized way. Therefore, there is a need of defining structural ways to represent the information in order for the users to access and make use of it. The selected structure must be able to deal with the natural imprecision that the retrieved information has. Fuzzy ontologies [6], [7], due that they are capable of dealing with imprecise information, are one interesting choice. They clearly have more representation capability than former ontologies which are not able to store information represented using linguistic modelling [8], [9], [10] and fuzzy sets [11]. Since the information nature that we want to represent is inherently imprecise, fuzzy sets and linguistic modelling environments provide an excellent mathematical background that the computational systems can use to deal with the data. This makes fuzzy ontologies one of the best tools to store imprecise and subjective information.

In this paper, a novel method that is capable to extract information from Internet users and store it in an organized manner on a fuzzy ontology is presented. Multi-granular fuzzy linguistic modelling methods are used in order to select the linguistic label sets that better fit the information that is being stored. Thanks to this method, it is possible to manage and work with all the non-formatted information that users provide on the Web. The novel developed method assigns an structure to the collective Internet knowledge and represents it using the fuzzy ontologies framework. Thanks to this, computational systems can deal with this complex data and anyone can retrieve information and benefit from the opinions provided by the users on the Internet.

The rest of the paper is organized as follows. In Section 2, basis of all the tools that our method uses to accomplish its goal are exposed. In Section 3, the proposed method is described in detail. In Section 4, a use case example is shown in order to ease the understanding of the proposed method. In Section 5, advantages and drawbacks of the method are discussed and compared with the ones of other similar methods. Finally, some conclusions are pointed out.

Section snippets

Preliminaries

In order to make this paper as self-contained as possible, this section will introduce several concepts and methods that will be mentioned along the paper. In Section 2.1, the procedure followed to carry out granularity transformations in linguistic label sets is exposed. In Section 2.2, basis of fuzzy ontologies are introduced. In Section 2.3, sentiment analysis procedures main structure is discussed.

Creating fuzzy ontologies from users opinions

In this section, the developed method is described in detail. By the use of sentiment analysis procedures, opinions are transformed into data that can be managed in an organized way by fuzzy ontologies. Also, multi-granular fuzzy linguistic modelling methods are used in order to express the information using the most adequate linguistic label set. The following steps are followed in order to carry out all this process:

  • 1.

    Extracting information from users’ opinions: Texts containing the information

Illustrative example

In order to enhance the comprehension of the developed method, a brief example is exposed in this section. Imagine that information provided by Internet users about wines wants to be retrieved and stored in a fuzzy ontology. Although there is a high quantity of wines and descriptions that can be analyzed, in order to present an easy to follow example, we will focus in 5 wines and two descriptions: price (pr) and acidity (ac). First of all, it is necessary to retrieve opinion texts that refer to

Discussion

In this paper, a novel method that is capable of using sentiment analysis procedures in order to extract information from subjective free texts provided by Internet users is presented. The retrieved information is stored in an organized way on a fuzzy ontology. Thanks to this, it is possible for other users to make use of the generated fuzzy ontology in order to retrieve and take advantage of the information that is stored there. The main advantages of the presented method are described below:

Conclusions

In this paper, a novel method that is capable of extracting collective knowledge from users’ opinions and represent it in a fuzzy ontology is developed. The novel developed method uses sentiment analysis procedures in order to extract the subjective information that is present in users’ opinions texts. Thanks to this, a computational system can understand and process this kind of subjective information. Once that the information is extracted, 2-tuple linguistic representation, linguistic

Acknowledgments

The authors would like to acknowledge the financial support from the FEDER funds provided in the National Spanish project TIN2016-75850-P and also the support of the RUDN University Program 5-100 (Russian Federation).

References (46)

  • J. Morente-Molinera et al.

    Solving multi-criteria group decision making problems under environments with a high number of alternatives using fuzzy ontologies and multi-granular linguistic modelling methods

    Knowl. Based Syst.

    (2017)
  • M. Giatsoglou et al.

    Sentiment analysis leveraging emotions and word embeddings

    Expert Syst. Appl.

    (2017)
  • FanZ.-P. et al.

    Product sales forecasting using online reviews and historical sales data: a method combining the bass model and sentiment analysis

    J. Bus. Res.

    (2017)
  • O. Appel et al.

    Cross-ratio uninorms as an effective aggregation mechanism in sentiment analysis

    Knowl. Based Syst.

    (2017)
  • L. Martínez et al.

    An overview on the 2-tuple linguistic model for computing with words in decision making: extensions, applications and challenges

    Inf. Sci.

    (2012)
  • J.A. Morente-Molinera et al.

    Building and managing fuzzy ontologies with heterogeneous linguistic information

    Knowl. Based Syst.

    (2015)
  • F. Ali et al.

    Fuzzy ontology-based sentiment analysis of transportation and city feature reviews for safe traveling

    Transp. Res. Part C Emerg. Technol.

    (2017)
  • J.A. Morente-Molinera et al.

    Creating knowledge databases for storing and sharing people knowledge automatically using group decision making and fuzzy ontologies

    Inf. Sci.

    (2016)
  • K. Huffman

    Web 2.0: beyond the concept practical ways to implement RSS, podcasts, and wikis

    Educ. Libr.

    (2017)
  • P.R. Humanante-Ramos et al.

    Electronic devices and web 2.0 tools: usage trends in engineering students

    Int. J. Eng. Educ.

    (2017)
  • N.D. Rodríguez et al.

    A fuzzy ontology for semantic modelling and recognition of human behaviour

    Knowl. Based Syst.

    (2014)
  • ZhangF. et al.

    Storing fuzzy description logic ontology knowledge bases in fuzzy relational databases

    Appl. Intell.

    (2018)
  • F.J. Cabrerizo et al.

    A consensus model for group decision making problems with unbalanced fuzzy linguistic information

    Int. J. Inf. Technol. Decis. Mak.

    (2009)
  • Cited by (0)

    View full text