State of Industry 5.0—Analysis and Identiﬁcation of Current Research Trends

: The term Industry 4.0, coined to be the fourth industrial revolution, refers to a higher level of automation for operational productivity and efﬁciency by connecting virtual and physical worlds in an industry. With Industry 4.0 being unable to address and meet increased drive of personalization, the term Industry 5.0 was coined for addressing personalized manufacturing and empowering humans in manufacturing processes. The onset of the term Industry 5.0 is observed to have various views of how it is deﬁned and what constitutes the reconciliation between humans and machines. This serves as the motivation of this paper in identifying and analyzing the various themes and research trends of what Industry 5.0 is using text mining tools and techniques. Toward this, the abstracts of 196 published papers based on the keyword “Industry 5.0” search in IEEE, science direct and MDPI data bases were extracted. Data cleaning and preprocessing were performed for further analysis to apply text mining techniques of key terms extraction and frequency analysis. Further topic mining i.e., unsupervised machine learning method was used for exploring the data. It is observed that the terms artiﬁcial intelligence (AI), big data, supply chain, digital transformation, machine learning, internet of things (IoT), are among the most often used and among several enablers that have been identiﬁed by researchers to drive Industry 5.0. Five major themes of Industry 5.0 addressing, supply chain evaluation and optimization, enterprise innovation and digitization, smart and sustainable manufacturing, transformation driven by IoT, AI, and Big Data, and Human-machine connectivity were classiﬁed among the published literature, highlighting the research themes that can be further explored. It is observed that the theme of Industry 5.0 as a gateway towards human machine connectivity and co-existence is gaining more interest among the research community in the recent years.


Introduction
Today's manufacturing industry is currently experiencing a rapid transformation due to the onset of fast-growing digital technologies and Artificial Intelligence (AI)-based solutions. Manufacturers throughout the world are faced with the challenge of increasing productivity while keeping humans in loop at manufacturing industries. This task becomes even more difficult as robots become more crucial to the manufacturing process by means of emerging technologies such as brain-machine interfaces and advances in AI. These challenges can be addressed by the next industrial revolution, known as Industry 5.0. In short, the concept of Industry 5.0 refers to humans and robots working as collaborators that technology may be tailored to encourage values, and that technological innovation can be built on ethical objectives, not the other way around [6].
The overall current state of understanding of Industry 5.0 describes it as the movement to bring the human touch back to the manufacturing industry. This is driven by the consumer's desire for mass personalization. This understanding means that Industry 5.0 products provide consumers with a way of realizing their urge to express themselves, and they will pay a premium to do so [7]. To summarize, Industry 5.0 is a concept that seeks to make industry more sustainable, human centric, and resilient. Some view it as an evolutionary, incremental advancement that builds on the concepts and practices of industry 4.0 and others view Industry 5.0 as a complement to the Industry 4.0 paradigm. Table 1 contrasts objectives, systemic approaches, human factors, enabling technologies and concepts, and environmental consideration of Industry 4.0 and 5.0 [7,8]. Since Industry 5.0 is a new concept there is little agreement on how it is defined. However, it is observed that the primary trend of Industry 5.0 is the introduction of human-robot co-working environment and the creation of smart society. To understand the perspective of what Industry 5.0 is, its evolution, and the technologies and domains that enable meeting Industry 5.0, in this paper text mining tools and techniques are used to explore the published literature landscape to identify commonalities and identify future directions of research in spearheading the transformation towards Industry 5.0.
The remainder of the paper is organized as follows, Section 2 details on the data gathering process detailing on the databases from which the data is extracted and analyzed, Section 3 expands on the text mining approach used and highlights the findings on the current state of Industry 5.0 research, and Section 4 concludes the article and identified the contributions.

Data Gathering and Preprocessing
The data gathering process involved identifying and extracting data of published research articles from scholarly databases. The databases used to extract data for this study include IEEE (Institute of Electrical and Electronic Engineers), Science Direct, and MDPI. These databases have a wider variety of coverage in terms of sources than other databases which is important since publications focusing on Industry 5.0 may be outside the "top" journals in the field [9]. Further, the restricted access to the authors of the databases available is another factor in choosing the identified sources and restricting the analysis to the abstracts. To identify published articles addressing the topic of Industry 5.0, key term "Industry 5.0" was used to search the metadata and identify published articles.
This included any published articles mentioning "Industry 5.0" in the title, abstract, or the keywords. This is because Industry 5.0 is still an emerging term, and it is still unclear what other key terms and synonyms are used [9]. The keyword search produced a total of 196 documents which included 26 from IEEE, 76 from MDPI, and 94 from Science Direct. The time range for these documents is 2016-2022 indicating the earliest publication on Industry 5.0 in 2016. The data collected included the publisher, title, publication year, and abstract for each publication retrieved. The data was then sorted by the database it was retrieved from and converted for further analysis into an .xlsx file. Once sorted, the data was labeled to identify the publisher, title, and the abstract corresponding to each published article retrieved. To analyze the gathered data, "R" a statistical language widely used by statisticians and data miners was utilized to transform, visualize, and analyze the data. This included data preprocessing, transformation, key term extraction, frequency analysis, and topic modeling. Table 2 illustrates the search results of each individual data base the abstracts were extracted from. The data is converted into a. csv format to be imported to R and then cleaned for further analysis by removing unwanted characters such as white spaces, numbers, symbols, and tag from the abstracts. Once these characters are removed the next steps were to delete stop words and convert words to lower case. Several text mining tasks, facilitated by Quanteda package [10] and dpylr package in R, include removing common words from documents. Stop word removal is a process of removing commonly occurring words for conjunction and propositions such as, a, I, in, for, with, the, not, on, and several similar other words that do not usually contribute much to the meaning of a given sentence. Most text written in English language follows punctuation and use of lower case and upper-case text. Though capitalizations enable humans to differentiate between nouns and proper nouns, in text analysis, words irrespective of where they are capitalized are treated equally and thereby converting all the characters to either lower case or upper case. There is also the process of stemming and creating vectors. Stemming helps to standardize the text by prefixes, suffixes, and inappropriate pluralization's in the text document. Creating vectors involves transforming the data into a representation to act as a suitable input for text mining algorithms.

Data Analysis and Discussion
Text mining, which is also referred to intelligent text analytics, text data mining, and text knowledge discovery, is defined as the discovery of either new or previously unknown information through the extraction of information from various written resources. Text mining help to uncover new information and knowledge by identifying patterns in documents from several sources [11]. Text mining may involve several other methods such as Natural Language Progression (NLP), Information retrieval, Clustering, Document Classification, Web mining, Information Extraction, and Concept Extraction [12]. It has been widely recognized among researchers that text mining's feasibility for exploring published literature and discovering concepts and trends across a given domain has seen tremendous growth. For example, Bach et al. discuss the advantage of using text mining in the financial sector for stock market predictions [13], Aureli portrays the applicability of text mining for studying organizations' social and environmental reports [14], Namugera et al. use text mining to study the social media usage of traditional media houses in Uganda to understand the topics these media houses discuss and determine if they are positively or negatively correlated [15], and use of text mining tools and techniques to analyze the landscape of Model based Systems Engineering [16].
In this paper the text mining framework as illustrated in Figure 1 [17] is used for exploring and analyzing the published articles on Industry 5.0 to identify key terms often used and the themes into which Industry 5.0 research can be classified into using the text data extracted.

Frequently Used Terms Extraction from the Data
Term extraction and frequency analysis is focused on pinpointing the relevant terms in each collection of text. Identifying the unique terms with statistical techniques such as calculating relative term frequency among the documents in a dataset enables better understanding of the information provided from the text [18]. The Inverse Documentation Frequency (IDF) measure is a widely utilized method for determining the contribution of terms. The greatest advantage of IDF is that it can aid in determining the influence of term in a group of given documents by identifying the frequency of term in a document and the number of times it occurs in a document [18,19]. When it comes to "low" and "high" IDF values, low values represent less informative terms appearing in several documents, and high values represent more informative terms appearing in only a few documents. In this paper, frequency analysis was utilized to identify the primary terms associated with Industry 5.0. Table 3 represents the top 20 frequently observed terms in the data along with their corresponding frequency. The term "Industrial Revolution" was observed to occur the most in the database. This is indeed expected since Industry 5.0 is referred quite often to as the fifth industrial revolution. Several publications in the data gathered use this term in comparing the different industrial revolutions throughout the years. Industry 5.0, just as past industrial revolutions, is predicted to have major impacts on the dynamics of socio-economic systems more specifically to have a large impact in industrial production systems [20]. The term "Artificial Intelligence" is observed to have the second highest count. Artificial intelligence seems to be one of the central components of Industry 5.0, mostly addressed for automating manufacturing processes, furthering the primary focus of cooperation between man and machine [21]. The term with the third highest count was "Supply Chain." This is highly significant since it is believed that Industry 5.0 will influence supply chains to an unprecedented level. The trends for Industry 5.0 supply chains include the incorporation of collaborative robots (co-bots), intelligent systems, mass personalization, and mass customization [22]. The term with the fourth highest count is "Big Data." Big Data is integral to Industry 5.0. It is believed that Industry 5.0 will introduce new innovations in management framework that takes Big Data into consideration. Furthermore, Big Data will be crucial for reaping the maximum benefit of Industry 5.0 such as modern technologies and new innovations in Internet of Things (IoT) and artificial intelligence [23]. The term "Digital Transformation" has the fifth highest count. This follows the prediction that Industry 5.0 will bring about a transformation towards digital platforms and a digital economy. In all, it is predicted that there will be a digital ecosystem, an open, distributed, self-organizing, system of system. The intention of this digital transformation is to unite subsystems to provide a common information space with access to a rich set of re-usable applied services that can support resource planning and control in real time. The digital transformation for Industry 5.0 should also provide standard access to cloud resources and services and to data perceived by external smart sensor networks [24]. The use of terms "manufacture industry", "production system", and "manufacture system" are centered on the dialogue from the research community discussing the shift from Industry 4.0 to 5.0 in the manufacturing and production engineering domain for the plausibility of advanced human machine interfaces for improved integration and better automation.

Term Frequency Analysis
The terms identified from the overall data set enabled to gain understanding on what specific terms were more focused upon by the researchers addressing Industry 5.0 paradigm. A measure of frequency for the identified terms is used to plot on a line graph, the relative use of terms over the years to identify their usage trends by the researchers. Based on the terms extracted in Section 3.2, the following terms were identified for analysis (a) "twin"to understand the trend on exploring the use of digital twins in enabling Industry 5.0, (b) "data"-for exploring the trend on use of big data i.e., set of diverse actionable data to empower Industry 5.0, (c) "intelligence"-to understand the trend on the use of artificial intelligence to aid Industry 5.0, (d) "cloud"-for understanding the trend in exploring the use of cloud based technologies and cloud computing as an enabler of Industry 5.0, (e) "IoT"-for exploring the trend on identifying IoT as an enabler for Industry 5.0, and (f) "machine"-for exploring the dialogue on the use of machine learning towards Industry 5.0 transformation. Figure 2 illustrates the trends on the use of the aforementioned terms over the past years. The use of the terms starting from year 2016 in the graph indicates to the limitation of the data that was gathered for analysis, starting from year 2016 where the first peer reviewed publication on Industry 5.0 was observed. Please see Table 1. The use of term "twin" in 2016 indicates to the initial attempt at exploring the use of digital twins to address Industry 5.0. Starting year 2018 the use of terms "intelligence" "cloud", and "IoT" are observed, indicating an initial interest in exploring artificial intelligence, IoT, and cloud computing technologies as enabler of Industry 5.0, with more interest in IoT. A peak in use of actionable data sets i.e., Big Data in the year 2021 followed by IoT and machine learning indicate more interest among the research communities to explore from an integration perspective toward a well-connected, distributed, intelligent, and actionable human centric systems. Please note that the abrupt drop in the term usage reflects to the limitation on the data gathered on publications until early 2022 considered for the analysis.

Topic Analysis
Topic Analysis was used to understand the major themes around which the published literature on Industry 5.0 can be classified. Topic analysis from text mining is defined as the act of extracting topics or thematic elements from a given set of documents. Topic analysis focuses on the characterization of a topic based on distribution of terms and the mixture of each topic in a document [25]. One of the most common methods of topic analysis from text mining is "Latent Dirichlet Allocation" (LDA) [26]. LDA is an unsupervised machine learning method mostly used for applications such as opinion modeling, extracting topics from source codes, and hashtag recommendations [17]. One of the greatest advantages of LDA is its pertinence to several domains while taking three domains into consideration, documents, words, and topics. LDA enables the user to define the number of topics as a parameter that the textural data must be characterized into. It is to be noted that, if this parameter is small, the topic identified provide only few semantic contexts whereas when the parameter to too large there is a scope for the topics identified to overlap. This, after several trails the authors limit the number of topics the data is to be characterized into 5 to be more reflective and coherent to Industry 5.0. Table 4 depicts the top five topics, and ten most likely terms observed in each topic, and a representative label for each topic provided by the authors.    The top abstracts, in a descending order, that represent each individual topics (Table 4) have been summarized to explore and better understand the perspective in the context of Industry 5.0 being addressed so far through published articles Theme 1-Industry 5.0 in context of supply chain evaluation and optimization in manufacturing processes: This topic supports the exploration of how Industry 5.0 can enable supply chain evaluation and optimization in manufacturing processes. More specifically, use of a multi-objective mathematical model to design a sustainable-resilient supply chain based on strategic and tactical decision levels [27], optimization of mining methods using multilevel, multi-factor, multi-objective, and multi-index comprehensive evaluation system involving technology, economy, and safety [28], exploring Social Value Orientation theory for understanding decision making preferences for join resource allocation [29], exploring the constructs of Industry 5.0 for supporting supply chain operations [22], research directions for supply chain transformations [30], influence of industrial internet of things and emerging technologies on digital transformation capabilities of organizations [31], effectiveness indicators for enterprise resource planning systems to aid digitization of information flows [32], enabling constructs of Industry 5.0 to control and manage supply chains in emergency [33], future of supply chains in context of Industry 5.0 [34], and approaches for supply chain digitization [35]. Addressing the 13% of the data gathered, a constant interest among the researchers is observed over time on this topic. Figure 4 portrays this trend.
Theme 2-Industry 5.0 in context of enterprise management, innovation, and digitization: Composed of 27% of the data analyzed, this thematic topic addresses the construct of Industry 5.0 in context of Enterprise Management, Innovation, and Digitization. This theme addresses research on translating critical success factors of project management in relation to Industry 4.0 for sustainability in manufacturing enterprise [36], an absolute innovation management framework for addressing the importance making innovation more understandable, implementable, and part of routine in organizations [23] required for adopting to the constructs of Industry 5.0, importance of addressing the nexus of entrepreneurial leadership and product innovation through design thinking [37] a core need for moving towards the notion of Industry 5.0 in organizations, identification of how digital product and process innovations might affect profitable customer strategies in a global context [38], use of biological resources and policy to drive Industry 5.0 [39], using sustainability based metrics for digital technologies [40] to enhance production operations, identification of value drives for successful digital transformations [41], significance and adoption of global reporting initiative standards in context of technology sustainability [42], enablers and challenges of digitization [43], and the importance of technological adoption and development based on the needs and demands of society [44]. This thematic aspect is observed to have a constant interest among the researchers over time for addressing Industry 5.0. Figure 4 portrays this trend.
Theme 3-Industry 5.0 in context of smart and sustainable manufacturing: Representing 25% of the data gathered, this major thematic aspect addresses Industry 5.0 in context of enabling smart and sustainable manufacturing. This thematic aspect is also observed to have a constant interest among the researchers over time for addressing Industry 5.0. Figure 4 portrays this trend. This theme majorly addresses pathways to manufacturing systems that can adopt by exploring the drivers and barriers manufacturing systems might face when seeking a transition to smart and sustainable paradigms towards Industry 4.0 and beyond [45], impact of industrial mathematics on industrial revolutions and how it can enable smart industries to meet customer need for future uncertain business environments [2], incorporating sustainable manufacturing measures for opportunities towards smart manufacturing [46], identification of components that enable Industry 5.0 for intelligent production systems [47], applicability of cloud based decision making based on information acquired from sensors [48], enabling smart manufacturing processes [49], machine learning enabled power dispatch systems [50], digital twins for sustainable operations [51], biomimetic designs for industry 5.0 [52], and integration of software suites and digital technologies for effective manufacturing quality management systems [53].
Theme 4-Industry 5.0 transformation driven by IoT, Bigdata, and AI: This theme addressing the use of IoT, big data, and AI towards Industry 5.0 is observed gain a lot of attention recently among the research community (illustrated in Figure 4) though it encompasses only 11% of published research articles so far (illustrated in Figure 3). This topic further relates to utilizing IoT technology such as sensors and actuators into the industrial process of Industry 5.0 to aid in the mass customization of products [54]. Research in theme is addressed on, taxonomy for integration of blockchain and machine learning in an IoT environment [55], expanding technological infrastructure, provision of budgetary support based on sustainable business models, standardization, and synchronization protocols, improving stakeholders' engagement and involvement [56], taxonomy analysis to aid in implementing methods and algorithms for different IoT application [57], in identifying techniques to improve the security and efficiency of data transmission between the IoT devices [58], exploring the use of deep learning and AI for monitoring [59], use of amazon web services using IoT and cyber physical systems for equipment monitoring [60], challenges and impediments of IoT [61,62], and energy efficiency and assessment models using big data and AI [63].
Theme 5-Human machine connectivity and co-existence: This theme relates to emergence of Industry 5.0 as the concept of human-robot/human-machine coexistence. This refers to the aspect of humans and robots in loop supporting and assisting each other in manufacturing and production engineering processes [64]. Research addressed in this theme include knowledge-based tasks and automation for humans and robots to measure cycle times [65], exploring the application of social value orientation theory to human machine contexts and multiagent systems [29], scientific improvements transforming the production lines and machines in intelligent systems [47], soft robotics for industrial applications involving manipulation of fragile objects [66], achieving a balance between capital and labor welfare by deploying Industry 4.0 technologies with a worker centric approach [67], and use of AI for robust solutions in mobile robotics [68]. Further, research in this theme also addresses the need for resilient workforce for adapting to workplaces and enterprises [69], perspectives on human centricity in future smart manufacturing [70], and use of agent-based approach to explore effects of human-robot interactions [71].

Conclusions
Enabled by the capabilities of text mining techniques, in this paper an attempt to understand and classify Industry 5.0 based on the published research articles with the time frame of when the term was first coined i.e., 2016 to the year 2022 is portrayed. Addressing the objective of the paper, term extraction technique was used to identify the most often occurring terms in the abstract text data gathered on Industry 5.0 related publications. The terms artificial intelligence, big data, supply chain, digital transformation, machine learning were observed to be most referred to. This coincides with the fact that Industry 5.0 is seen to facilitate repetitive tasks with the use of artificial intelligence and machine learning technologies, parallelly assisting humans for cognitive support. Further, big data and digital transformation are foreseen to provide an information space rich with data that can be used for resource planning and control in real time. Topic analysis technique was used to identify the thematic aspects of papers published addressing Industry 5.0. Five different thematic aspects were observed across the landscape, with most of them in the context of smart and sustainable manufacturing followed by human machine connectivity and co-existence. More specifically, the theme of Industry 5.0 as a gateway towards human machine connectivity and co-existence is observed to gain a lot of interest among the research community. Further a brief description of the top ten papers from the data defining the topics is provided for the audience to understand the perspective and relevance of topics.
By examining the analysis provided future research directions can be identified and predicted on how Industry 5.0 will impact the manufacturing landscape in the years to come. The results obtained are limited to the data i.e., 196 abstracts extracted. It is to be noted that these the results are subject to change with increase in the number of abstracts used for analysis along with expanding the digital libraries used for extracting the data. As a future work, it is believed a comprehensive analysis along with integration of data crawling techniques will provide a better perspective on what Industry 5.0 is and how it is perceived among the research community.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.