Research and trends in the studies of Collective Intelligence from 2012 to 2015

The interaction between groups of people and machines supports the transfer of knowledge and strengthen Collective Intelligence (CI) making it robust. The CI with the support of technology progresses through platforms and computer systems including ontology, clusters, agents, Web intelligence among others. This research consists of a content analysis of recent studies 2012-2015 on CI. After an extensive search of publications on electronic databases, two hundred and fifteen papers were selected and exposed in a document's analysis following the approach of Zott. In this research we identified three categories to consider: Learning, technology and decision-making. The analysis revealed that CI is strongly related with technology, supporting the processes of training people and promoting collaborative learning as a new form of literacy.


Introduction
The scientific communication is the main practice of discussion into the research community.It allows the generation of knowledge and progress of societies.CI has been investigated for several years, is currently taking momentum with the use of technologies that impressively supports collaboration between individuals.Between 2012 and 2015 there have been published in scientific journals several papers related to CI. CI encompasses a variety of domains, interaction and collaboration within groups in order to perform a variety of works, like sharing knowledge and strengthening decisionmaking in problem solving [22].
CI has been defined with different approaches.A recent definition understands CI, as "a set of intelligent units like experts, agent systems, or simply set individuals which are autonomous in making decisions" [44].For [41] CI "is a group/shared intelligence that emerges from the collaboration and competition of many entities, either human or digital".According to [77] CI can be considered "a new emerging feature of communities in order to connecting human beings and a new contribution to the acquisition and production of knowledge."Finally, [52] defines "Collective Intelligence (CI) as "an active field of research, which capitalizes the knowledge of human collectives in order to create, to innovate and to invent." Classical definitions of CI include [53], "Collective intelligence (CI) is the capacity of human collectives to engage in intellectual cooperation in order to create, innovate and invent".According to [78] "defined very broadly as groups of individuals doing things collectively that seem intelligent." Other authors agree in indicating that CI is present in various domains.Here we reference some of them: [64] indicated CI will impact the different domains of science, also [25] pointed out that in the scientific literature raised the definition of the CI in several domains.Further, [61] referenced in [17] reported that the CI is a field multidisciplinary.Researches and studies in several domains as sociology, psychology, analysis of social networks, biology, economy and in general in behaviour of masses have been developed.

Domain education
CI has an enormous potential in the domain of education [52], [22].Some authors have investigated CI in the domain of education are presented in Table1:

Description
Author(s)

Materials
Teachers and students creating knowledge, educational materials or curricula and assessment materials among others. [66]

Improvement of the learning process
Dynamic aspects of the process of teaching and learning using technology. [23]

Examples
Use of student data to support decisionmaking and collaboration. [22] Use of social media technologies (SMTs) for learning, exploring whether SMTs lead to the production of forms of CI. [79] Expressing the interactive nature of m-learning.
[47] referenced by [30] E-learning 4.0 can be supported by the CI.[30] This interaction allows students to create and share knowledge, Moore called it CI and Kowitz and Smith indicated that shared learning is the most advanced form of instructional process. [14]

Domain computing
CI impacts on the models in computer science [45], and it emphasizes the areas of computing including data management of heterogeneous groups [49].Computing services (science and engineering) have grown in the past 10 years thanks to CI since it has improved the quality of these services [38].Table no. 2 shows some authors that refer to the domain of computing using CI:

Computational intelligence technologies
Such as consensus theory, fuzzy systems, neural systems, evolutionary computation and others can support CI. [80]

Improving process
Negotiation mechanism for dynamic scheduling based on CI, they proposed to negotiate the use of software agents that must interact and collaborate in order to improve the global schedule. [43]

Examples
Worked on the topic of the Semantic Web as an intermediate solution for analyzing source tools supported by the integration and sharing of data and knowledge. [32] Application for 3D movie production with CI can be implemented in the mobile computing environment [72] The multi-agent technology allows decision making based on a set of agents [4]; [75]; [26] Web Intelligence creates knowledge from different knowledge bases.

Domain social sciences
CI is in the social sciences [65], and it provides tools to support organizations.Commercial enterprises, government, military institutions, and civic organizations use CI processes [21].Table no.3 shows some authors that refer to the domain of Social Sciences using CI:

Organizations
The interaction among members of an organization moves from one set of people related to a highly cohesive team that the author defines as collective intelligence, seeking to improve all types of organizations and working environment. [68]

Improving process
Aims to improve the management process of the quality of IT services, CI has a mechanism for increasing human knowledge with digital knowledge from enterprise data sources, which provides a greater understanding of the company. [69]

Examples
The Wiki is an instance of online CI.The most widely implemented and largest libraries (Wikipedia) have used Wiki. [40] The PolicyGrid project research the role of Grid, Web 2.0 technologies and Semantic Web to support e-Social Science [13] The platform, developed in open source, includes conventional and unconventional data to be shared between users of a social network, allows Apps to create Intelligence Political participation via social media: A case study of deliberative quality in the public online budgeting process of Frankfurt/Main, Germany 2013.[81] The research among several groups: workers, students, friends, etc., outlines collaboration to achieve collective intelligence [51] 1.4.Others domain Other domains applying CI are: Tourism and Travel [71], agriculture projects [70], health reported by [8] who designed a system to take advantage of the CI from students, teachers, professionals, institutions and clients who are interested in this field.Additionally, the research of [25] reports that it can be useful for systems of diagnosis with the aim to create new schemes for the acquisition of knowledge.
Finally, [48] includes other examples of CI in various domains such as: Computer & science, biology, political scientists, economy, history, sociology, organizational theorists, anthropology and psychology.

Decision-Making
In the domain of education CI can be used to assess student progress and improve decision-making [52].In the decisionmaking groups, these are given by the consensus, according [39]; [50].Also, a research by [25] based on "Knowledge Acquisition for Medical Diagnosis Using Collective Intelligence"; it explains the scheme to collect information for diagnostic methods based on CI and consensus.In the investigation of [82] propose multi-agents for capture of decisions making, exploiting based modelling system and Web-based methodologies in the field of Software.In future works, they will continue investigating on new models of capture of decisions with the blended of the CI and the agents.[46] reported a prototype to supervise students in the process of brainstorming topics and to measure results and interactions in real-time.[55] reported consensus processes in large scale decision making under uncertainty groups, which incorporates the use of aggregation operators.
We hope that the current study contributes to CI and increase understanding about current trends in it; we conducted a content analysis of published papers on the subject.
The rest of this paper is organized as follows: Section 2 contains the methods, referring to the analysis of content in selected research papers; Section 3 contains the results of the selected research papers reviewed and coded according to the method; Section 4 contains some concluding remarks and road map for future research and Section 5 contains some limitations.

Methods
Analyses of content published in scientific journals allows to assess the level of development of a particular discipline and to appreciate its research trends [29], [6].According to [33], content analysis is a method that separate large amounts of information regarding specific purposes.The content analysis involves comparing, contrasting, and categorizing a set of data.[36] used a classification method to analyze collections of papers.According to [62] content analysis has proven to be an effective research method; they also referenced the following authors: [58] in distance education, [59] in educational technology and [67] in science education.Finally, [28] has used the method in the investigation into blended learning in higher education.
Content analysis was selected because its potential to classify text material [5].The literature exemplifies content analysis as a methodology for analyzing and observing the trends of published articles in journals [54].

Data collection
The selection of papers regarding to the literature of CI were done in: Science Direct, Web of Science, SpringerLink and Wiley.The searches for the CI were limited to articles in journals, full-text, all language and all sciences.The term of "collective intelligence" was used in all parts of the article, and it locates a total of 1724 papers published from 2012 to 2015.(Figure1).According to [11] the title of a scientific article describes with few words the content of this article, and the abstract can be considered a miniature version of the article.In addition, the American National Standards Institute referencing [11] states that "a well-prepared abstract enables readers to quickly identify and exactly the content of a document, determine its relevance to their interests and decide if they have to read the work in its entirety."Finally, [11] considers that keywords label the scientific article.Refining the search with the term "collective intelligence" appearing in keyword, title or abstract the previous 1724 papers were limited to a set of 215.
Next, the set of papers were refined again following the approach of [74] in order to limiting the final list of papers.Over this final list a content analysis was implemented to interpret the contents of the papers classifying and/or encoding the various elements of text in categories [20].
Following [74] the final inclusion criteria of papers are based on the fact that the issue of CI is addressed on a nontrivial manner.
 It is the focus of the paper and usually appears in the title. It appears at least in two of three: title, abstract and key words  It appears in the abstract, in the keywords, and it is clear that the paper is a contribution to IC At the end, a final list of 119 papers resulted as relevant and was analyzed for this research.Appendix A shows the list of reviewed articles.The Figure 2 shows the scheme followed for the selection of the items.

Analysis
To perform content analysis, categories must be established according to the focus of each particular research.Categories are intended to group together sets of pieces of information, which refer to the same aspect [24].The categories applied to analyze the attributes of texts in the list of papers of this research are shown in Table 4.The papers of this research have been coded according to the previous categories.

Results and discussions
The results of the research are presented in different subsections:

Domains of the research
Nowadays, CI is developing in diverse domains, all the 119 papers have been assigned to one of these areas, and when the scope of the paper was more than one, the main was selected.The Figure 3 shows the different domains discussed in the literature in the period 2012-2015.The computing domain has the highest value.This may be due in part according to [83] symbiosis between computers and humans are the key to a potential of technologies and high impact services designed to enhance the capabilities of human beings to solve problems.Combining ICT knowledge and skills can improve and solve global problems.[53] indicates that the technologies provide powerful tools for increasing cognitive processes of personal and collective way, the processes of CI can be multiplied by digital networks.Additionally, [48] indicates that we must understand that the CI depends on the integration and combination of humans and machines, organizations and networks.

Category
The frequency of the categories previously defined in table 4 and their associated words are shown in Figure

Collective Intelligence and learning
Knowledge and CI have been several investigations.[57] applied a CI framework to characterize education in the context of a web based tool for teachers, called Instructional Architect (IA).It allows teachers to find, create and share instructional activities for students using online learning resources.[25] presented a research entitled "Knowledge Acquisition for Medical Diagnosis Using Collective Intelligence", it indicates that using the wisdom of crowds can get new biomedical knowledge.These researchers exhibit a scheme to collect diagnostic information Diagnosis Decision Support Systems; they are based on consensus and CI.
[37] proposed CorpWiki, self-regulation of the wiki system, it allows the acquisition of high-quality knowledge.This will develop CI organizations make efficient use of the intelligence of its employees and the facilities provided by technology, such as Web 2.0.That way you can create and evaluate knowledge that is timely and assured quality.
Regarding to learning, [12] referenced [63] reported online communities that provide a learning space to build CI.They are communities where each member has the potential to contribute and participate in the discussions, which increases the possibility to solve complex problems.Moreover in the field of education and the use of technologies, [66] reported in his article that Web 2.0 provides a framework for education on the Web, allowing students to experience CI and creativity.
There is evidence in the literature that papers dealing with CI enable collaboration between groups, affecting all areas, especially education.Studies indicate that teachers and students are committed to CI.They can help to create, share and reuse new content or they can be consumers by displaying other content.Also, the use of software tools enhances the IC, both in the generation of knowledge and the administrative aspect of education.

Collective Intelligence and technology
In research literature linking CI and technology investigations are published in the scientific literature.[84] proposed in their research a framework of CI based narrative reasoning and natural language processing.It exhibits a hybrid model that combines the Narrative Knowledge Representation Language (NKRL) and HARMS (Humans, software Agents, Robots, Machines and Sensors).
Others researches includes the Web.[76] reported a project called "Open Geometry Textbook" whose objective was to develop a web-based platform to gain knowledge on the subject of geometry and create a textbook through CI involving Internet users.According to [28] Semantic Web aims to exchange structured information and formal knowledge to achieve CI on the Web.The Semantic Web enables the distribution of data and interconnection to provide information to users.It also allows the sharing of knowledge, collaboration and cooperation.
Technological systems with Web 2.0 evolve in e-learning and the CI is strengthens.In the investigation of [27] proposed an adaptive learning system centred on the user based on the CI of users and employs item response theory.The results indicate that students are more satisfied and learn efficiently.In the publication of [60] reported the effects of application of information technologies and communication ICT from the perspective of the CI.They refer to the CI, as the exchange of information through specific tools.
Regarding to technological tools several are described in scientific databases, including [8] entitled "Aprendis: a tool for formal learning in Health Informatics", which aims to harness the CI of professionals, students, teachers, clients and institutions interested in the area of Health Informatics, specifically the Portuguese-speaking community.
Finally, [19] indicates that CI arises from the following: 1) data, information, knowledge; 2) software, hardware and; 3) experts and stake-holders which produce knowledge through their input and feedback from them.CI must rely on systems and software tools to develop new collective knowledge.Human and computer tools must be integrated into all domains of knowledge allowing the development of CI.

Collective Intelligence and decision-making
According to several authors, CI appears in a variety of forms of collective knowledge, and it is the result of consensus decision-making in different processes [44].Investigation of [25] lets you collect diagnostic information Diagnosis Decision Support Systems methods based on consensus and CI, the objective of the research was to apply the CI to share medical knowledge and build a knowledge base on using consensus methods, achieving results make better medical diagnosis.
[19] created an information system to support the Egyptian Academy of Scientific Research and Technology, in the Millennium Project The system organizes information from experts, scientists, leaders and the general public, aided by the software.It improves decision-making, civic participation and social cohesion.
According to [52] the objective of recommendation systems is to support the decision-making process of the user.They propose in their research use the CI and recommendation systems to improve decision-making.
The objective of recommendation systems is to support the decision-making process of the user.[52] propose in their research to use the CI and recommendation systems to improve decision-making.
Research in the area of CI is opening opportunities for research in the area of decision-making.The literature indicates the need to make quick decisions, to deal with this situation, it is better to use new models and tools incorporating IC.

Conclusions
This article provided a diagnosis of CI research in journals during the period 2012-2015.Reviewed the papers and made the analysis from each article, we have the following conclusions:  Researchers can locate a variety of peer reviewed scientific documentation, taking the problem to handle a large wealth of information.To resolve this problem there are methods and techniques, for example the analysis of content [15].It appears at least in two of three: title, abstract and key words  Our study was conducted in articles published in high impact journals; the focus of our investigation was learning, technology and decision-making.We found that the CI is exhibit in works to enhance learning.The decisions of individuals and groups benefit from the use of technologies to process large amounts of information.We believe that the CI provides an alternative to the creation of knowledge, using technologies and helping decision making. The analyses method used, which are more suitable for the data in this type of study, can be used for other researchers in future studies.The repetition of this type of research allows other scientists to be better informed and will keep people updated on trends on IC.

Limitations
The study has some limitations; the sample was obtained from literature search in four databases.This work could be extended by reviewing literature from other sources.Despite of the attempt of scientific rigor, some sources may have been lost during the sample analysis and the identification of issues and their classification is subjective, we recommend that further similar research on CI are made.

Figure 2 .
Figure 2. Scheme followed for the selection of the articles

Figure 4 .
Figure 4. Category and Frequency

Table 2 .
Domain computing using CI

Table 3 .
Domain Social Sciences using CI