Demystifying Graph Data Science: Graph algorithms, analytics methods, platforms, databases, and use cases
2: Chitkara University Research and Innovation Network (CURIN), Chitkara University, India
3: Department of Computer Science, University of Oviedo, Spain
4: Siemens Healthcare Pvt. Ltd, India
With the growing maturity and stability of digitization and edge technologies, vast numbers of digital entities, connected devices, and microservices interact purposefully to create huge sets of poly-structured digital data. Corporations are continuously seeking fresh ways to use their data to drive business innovations and disruptions to bring in real digital transformation. Data science (DS) is proving to be the one-stop solution for simplifying the process of knowledge discovery and dissemination out of massive amounts of multi-structured data.
Supported by query languages, databases, algorithms, platforms, analytics methods and machine and deep learning (ML and DL) algorithms, graphs are now emerging as a new data structure for optimally representing a variety of data and their intimate relationships.
Compared to traditional analytics methods, the connectedness of data points in graph analytics facilitates the identification of clusters of related data points based on levels of influence, association, interaction frequency and probability. Graph analytics is being empowered through a host of path-breaking analytics techniques to explore and pinpoint beneficial relationships between different entities such as organizations, people and transactions. This edited book aims to explain the various aspects and importance of graph data science. The authors from both academia and industry cover algorithms, analytics methods, platforms and databases that are intrinsically capable of creating business value by intelligently leveraging connected data.
This book will be a valuable reference for ICTs industry and academic researchers, scientists and engineers, and lecturers and advanced students in the fields of data analytics, data science, cloud/fog/edge architecture, internet of things, artificial intelligence/machine and deep learning, and related fields of applications. It will also be of interest to analytics professionals in industry and IT operations teams.
Inspec keywords: data mining; data visualisation; learning (artificial intelligence); data structures; graph theory
Other keywords: graph theory; Big Data; artificial intelligence; data mining; information systems; database management systems; data analysis; learning; data visualisation; data structures
Subjects: Database management systems (DBMS); Information networks; File organisation; Graphics techniques; Combinatorial mathematics; Data mining; Machine learning (artificial intelligence)
- Book DOI: 10.1049/PBPC048E
- Chapter DOI: 10.1049/PBPC048E
- ISBN: 9781839534881
- e-ISBN: 9781839534898
- Page count: 415
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 Toward graph data science
- + Show details - Hide details
-
p.
1
–14
(14)
In this chapter, we are focusing on how the work with graph concepts in the data science and also going to focus how advantageous is to use graphical representations in data science. We also refer the best-suited algorithm for the successful working of the project. Here we can perform the variations by using one algorithm to the other algorithm in various modes. By doing the comparison process among the various algorithms, we can expect the result in the better manner as well as more efficient manner. The author Nolan brought the concept of graphs in the computing environment for making calculation, to show the variations of various models and many more. By embedding computing with the statistics will be helpful for the better understanding and also the number of employment is rising because of merging graph techniques in machine learning. To bring up the difference or to bring up the comparison of the works, the graph techniques were very useful.
-
2 Data science: the Artificial Intelligence (AI) algorithms-inspired use cases
- + Show details - Hide details
-
p.
15
–37
(23)
The data science field is growing fast with the faster maturity and stability of its implementation technologies. We had been fiddling with traditional data analytics methods. But now, with Artificial Intelligence (AI), it is possible to embark on predictive and prescriptive insights generation in time. There are several data science (DS) use cases emerging with the wider adoption and adaptation of AI technologies and tools. This chapter is dedicated to illustrate various AI-inspired use cases.
-
3 Accelerating graph analytics
- + Show details - Hide details
-
p.
39
–56
(18)
Handling graph databases properly leads to obtaining a precise result within a fixed time. It is essential to take a fast-track in graph analytics to address the challenges faced in exploring the data, as it is an incipient form of data analysis. The analytic methodologies involved in discovering the underlying implications of data to deliver an optimized result should get drive-by emerging techniques. A graph database possesses many complex patterns which are to be quickly identified for specific outcomes. Graph analytics assist in building patterns based on the connectivity between the nodes to provide proper context. Graph analysis is transforming into a significant factor for today's data analysis, so there is a need to accelerate its medium to the next level by implanting its features with recent technologies. To boost the performance of graph analytics, Artificial Intelligence (AI) is embraced with its enhanced features and techniques. Graph-based machine learning and analytics act as a boon for data scientists to increase their productivity and representation of data which is widely expected from enterprise users. Implementing AI produces machine-generated quick and precise values and results in a complete dataset-adopting machine learning techniques helps in feeding training data to the algorithm, building relationships between disparate data points, and enabling the connection between nodes and structure graph databases. This chapter leads the path of advanced graph analytics, which will lead data analyzers to take their next move in a data-powered business.
-
4 Introduction to IoT data analytics and its use cases
- + Show details - Hide details
-
p.
57
–107
(51)
The Internet of Things (IoT) industry has gone through an extensive transformation in recent years with the adoption of more modern software methodologies. In this chapter, we will present how this transformation has affected several IoT markets and discuss general architectural approaches taken. The different layers of the architecture will be decomposed and presented with focus on Artificial Intelligence and how it applies in containerized and orchestrated-deployment models.
-
5 Demystifying digital transformation technologies in healthcare
- + Show details - Hide details
-
p.
109
–131
(23)
Digital transformation of healthcare is indeed a foundation stone toward a patient-focused system of healthcare. It can allow healthcare professionals to drive efficiency, consider whatever the patient needs, develop confidence and trust, and have enhanced user experience. Enhance the efficiency of treatment and access to medical care while reducing costs as demographics age and increase, rising life expectancy and placing public spending on healthcare on demand are some of the key issues facing healthcare institutions around the globe. Basically, all members in the vast and growing healthcare sector are substantially growing their digitization and technical advancement activities in order to address those and other challenges, but they also provide demand for growth in healthcare and substantial money in digital wellbeing. Healthcare organizations are distinctive across the world and healthcare societies are starting to move at various responsibilities, based on the locality, the legal arrangement, the policy process, the personal institution, the involvement of the health industry and the specific objectives of digitalization in each individual's life: such as improving patient centeredness in hospitals and increasing productivity to new forms of care, e.g., remote patient monitoring by utilizing fog, cloud technology, and the digital economy. Through elite elders to patients, wearable sensors have become particularly involved in monitoring physiological causes, encouraging fitness, and optimizing commitment to the procedure. In this chapter, fog computing-based wearable sensor networks (FWSNs) have been suggested in healthcare monitoring for elderly people using IoT. During regular activity, a wearable tracker for continuous - time tracking of blood pressure, breathing frequency, and motion was investigated. Furthermore, the sensor data will be transmitted to the IoT system Ethernet device, and the Approved Person uses the data to monitor the elder's health through the Internet. Furthermore, the simulation tool for the wearable interface and its utilization show how computational resource costs can be lowered while retaining health demands for access to medical stored data in a cloud and fog distribution environment. The findings of the tests demonstrate that the proposed approach is user-friendly, dependable, and cost-effective to use it on a daily basis.
-
6 Semantic knowledge graph technologies in data science
- + Show details - Hide details
-
p.
133
–153
(21)
Knowledge graphs, the representation of data as just a semantic graph, had already garnered considerable attention in both the academia and industry worlds. Their ability to provide semantically appropriate data had already made significant reasonable solutions to several tasks, such as solving problems, guidance and knowledge representation, and has been considered to be a tremendous potential to several researchers to develop the most advanced technology. Even though numerous "Big Data" applications have been already facilitated in all kinds of commercial and science domains through information graphs but limits of effectiveness have been met by the document-centric frameworks of research. Recent controversies on the growing abundance of research journals and the issue of accuracy have stressed everything. This creates an opportunity to reconsider the prevailing view of communications of document center research scholars and turn it into a flow of information based on experience by depicting and transferring knowledge via semantic-based interconnected knowledge graphs. The development and advancement of information systems create a shared understanding of knowledge exchanged among users. The methods of query and exchange of data in the learning center of the world are at the center of the knowledge-based flow of data. By incorporating these frameworks into current and new science technology services, the knowledge structures that are now latent and profoundly concealed in documentation can be completely visible and right accessible. It would have the possibility to transform data science effort, as knowledge and analysis findings can be easily interrelated and ideally suited to diverse information requirements. In comparison, experimental findings are exactly equivalent and simpler to replicate. In this chapter, the conception of a knowledge graph for data science describes the potential framework for knowledge graph technologies and initial attempts to incorporate the framework.
-
7 Why graph analytics?
- + Show details - Hide details
-
p.
155
–178
(24)
The study of relationships between entities such as clients, goods, operations, and devices is graph analytics, also called network analysis. To obtain knowledge that can be used in marketing or, for example, for analysing social networks, companies use graph models. The term 'graphical analysis' explicitly involves the study and analysis of data, which can be translated in a broad schematic. Graphical analytics are a fast-growing domain in the area of large-scale data mining and visualisation that is used in various multidisciplinary applications like network protection, finance and health care. While several methods have already addressed the study of unstructured collections of multidimensional points in the past, graph analytic technologies are a relatively new trend that presents a number of challenges. Graph analytics are a combination of mathematical, theory of graphs and techniques used to model, store, extract and performance analysis graph-structured information. The techniques recognise modules or interacting subgroups within graphs, search for sub-graphs that are similar to a particular pattern. Due to their polytrophic nature, graphs have acute importance and have widespread big data applications in the real world, e.g., information discovery, social media, search engines, network structures, etc. The main issue is the development of large-scale applications of efficient systems for storage, processing and analysis. Graph analytics are used in numerous applications to model all kinds of relationships and processes. Data scientists and business users can define and analyse complex relationships in healthcare datasets through graph analytics. Gartner Research said in a recent study, "Graph analysis is probably the single most efficient competitive differentiator for organisations that follow data-driven operations and decisions after data capture design." Since the data sources in health organisations, heterogeneously complex and highly dynamic data sources are well-known, the healthcare domain has acquired its importance through the effect of big data. While the position of large graph analytical methods, platforms and tools is realised across different domains, promising research directions are shown by their effect on healthcare organisations to introduce and produce new use cases for possible healthcare applications. The effectiveness of healthcare applications is solely dependent on the underlying nature and implementation of appropriate methods in the sense of broad graph analysis, as demonstrated in ground breaking research attempts. In this chapter, from the perspective of different stakeholders, we discuss the various methodological options available in the patient-centred healthcare system. In order to promote individual patients from diverse viewpoints, we address different architectures, benefits and repositories of each discipline that provide an integrated representation of how separate healthcare operations are carried out in the pipeline.
-
8 Graph technology: a detailed study of trending techniques and technologies of graph analytics
- + Show details - Hide details
-
p.
179
–198
(20)
Graphs are information structures to depict connections and communications between elements in complex frameworks. Graph analytics has been in use for a long time in the field of data analytics. Its main purpose is to create a database of interconnected entities and to model the relationships and processes in various information systems. In the field of data science and analytics, it is the graph analytics, an alternative method of analysis that uses the process of abstraction and this abstraction is called a graph model which helps the analyst to analyze the whole data or results in a summarized form that reduces the analytics complexities. Many organizations use the Graph model to leverage analysis in marketing or social networks. Graph storage is also an important fact along with graph analytics. The underlying structure of any database in which graph data is stored is often called graph storage as native and non-native graph storage. This chapter is going to explain graph analytics as well as graph storage in detail with the knowledge of various graph logical approaches and looks at existing graph storage and computational advances. This research provides analytical insights about the various graph analytical technologies used globally and shows a comparison between existing graph storage and computer technology. This research additionally evaluates the performance, qualities, and impediments of different graph databases and graph processing models.
-
9 A holistic analysis to identify the efficiency of data growth using a standardized method of non-functional requirements in graph applications
- + Show details - Hide details
-
p.
199
–216
(18)
In the modern era, several opportunities are provided to transfer data through graph models in which digital transformation plays a vital role. Maintaining the data using several devices will cause a processing time delay. Data collection is an important task in all data processing units, as is storing this type of information, as is providing security on this data through a database. To improve this process, the data retrieval is done using a graph data model. The proposed method is used to find the best way to store each record in a graph database rather than in another cloud or distributed database. In this, various techniques used in providing a better solution for data processing are done on graph databases without schema. To provide a good solution without any time delay, the graph analytics algorithm will help in making decisions on better results. In this method, many applications will be taken as case studies for finding the best relationship on the given graph database. In this, the collected data will be converted into graph format, an easy way of finding the duplication. The data model generated on each vertex is converted into low- and high-dimensional data forms. This chapter will go over a number of real-time Neo4j applications that are used to find optimal relationships on various datasets in an efficient manner.
-
10 Roadmap of integrated data analytics - practices, business strategies and approaches
- + Show details - Hide details
-
p.
217
–233
(17)
One part of productivity growth is creativity. Innovation can be used in processes as well as in goods. Innovation in the process helps a business to make the most effective use of its resources while innovation in products aims at creating new products or enhancing customer service. Innovative concepts have been derived from human naivety and imagination throughout history. But what if data and algorithms in some instances could help, boost or even substitute human ingenuity? Increasingly, data or analytical approaches are used to discover and develop hypotheses in vast quantities of various data. In fields including materials science, synthetic biology and life sciences, data and analysis will transform research and development. In addition, integrated data analytics helps to cleverly narrow the universe of possible combinations and leads to discoveries with the vast quantities of data to be sorted and almost limitless possible combinations of functions. With regard to this advent, this chapter provides a brief guide (i) to an era of integrated data analytics; (ii) to present a detailed roadmap of integrated, data analytics has been implemented in business strategies; (iii) to provide the approaches to construct integrated data analytics platform.
-
11 Introduction to graph analytics
- + Show details - Hide details
-
p.
235
–255
(21)
Graph is considered to be the collection of the points or vertices or nodes and the lines between the points or the edges. The structural characteristics of a data can be plotted easily using graphical representation. In this chapter, we are going to deal about the introduction of graph analytics and about graph structure data. A detailed discussion about the graph algorithms and graph databases gives an additional advantage to read this chapter. Some of the latest graph analytics tools also explained. Data science plays a major role in today's computing world. The subdivision of data science that contracts with mining information from graphs by executing analysis on them is known as "Graph Analytics." These graph analytics are mainly supported in data science for handling huge datasets such as Google, Amazon, Facebook, e-Commerce, and Finance. It can also support use cases include cybersecurity, drug interaction, reference engines, contact tracing, social networks, and supply chains.
-
12 A study of graph analytics for massive datasets
- + Show details - Hide details
-
p.
257
–272
(16)
The analysis and research of data which can be altered into a comprehensive graph is referred to as "graph analytics." Graph-based data analytics is a budding field in both data mining and data visualization and is applied for a wide variety of applications such as network protection, banking, and healthcare, both multi-disciplinary and high impact applications [5]. Despite the fact that many methods have been developed in the past to analyze unstructured collections of multidimensional objects, graph analytic technologies are a recent trend that poses several challenges, not only in terms of the output of algorithms that are related to data mining that facilitate algorithmic computational data discovery [3]. Graph analytics primarily aimed to evaluate graph oriented structured data in order to uncover answers to queries (e.g. Identify the person who is the most prominent person in a community? What are the main technology nodes for better practice and decision-making on the internet and urban networks?)
Analysis of graphs has always attracted and has always been an important topic for researchers in the history of computing; however, the rise of the uses of advanced analytics for large amounts of semi-structured or unstructured data and the revolution of big data has lately picked up the interest of the information systems community [1]. The qualitative effect of data, as well as the impact of graph analytics technology on organizations, has affected the requirements for business outcomes. Graph analytics for big data can not only recognize but also visualize crucial insights in big data. Furthermore, graph analytics may assist in identifying associations between different types of data and determining their existence and meaning within the context [2].
In this chapter, we will present the fundamentals of graph analytics and how graphs are related to big data. The chapter will also show some of the most common graph databases and discuss various big data graph analytics approaches which use the massive datasets, as well as different frameworks for each approach. In the latter part of the chapter, various issues and challenges related to big graph analytics will be addressed. A case study for implementation of graph analytics using python will also be discussed.
-
13 Demystifying graph AI
- + Show details - Hide details
-
p.
273
–307
(35)
Graphs are emerging as futuristic and flexible data structures that can fluently and fluidly model different relationships and processes over physical, biological, social, and information systems. Graph nodes or vertices represent the system's entities. Nodes are connected by edges/links, which represent relationships between those entities. Such an influencing representation helps to express and expose complex interdependencies in data.
The Artificial Intelligence (AI) domain fundamentally represents a collection of pioneering algorithms and approaches to uncover and emit out human-like intelligence from data volumes through an iterative and insightful process of building, evaluating, and optimizing AI models. As indicated above, another interesting facet gaining prominence in the recent past is the aspect of data representation through enigmatic graph structures. This transition has resulted in solving a myriad of complex business problems. Now by applying proven and potential AI procedures and processes on graph data, the task of knowledge discovery out of data mountains is becoming simpler and speedier. The prediction accuracy and performance of AI models when applied on graph data show a lot of perceptible improvements. This strategic and subtle convergence is being widely touted as the Graph AI paradigm. In this chapter, we are to discuss what, how, and why Graph AI is acquiring all the attention and how this new paradigm is bound to be a trend-setter for the ensuing era of knowledge.
-
14 Application of graph data science and graph databases in major industries
- + Show details - Hide details
-
p.
309
–323
(15)
Graph data science can be used across several key industries such as Social Media Networks, Finance, Healthcare, Risk Management, Transportation, and Supply Chain. Using Graph databases, data can be presented in a new way which may open unlimited conceivable outcomes and help to tackle complex issues. In this chapter, we will see how these Graph Models help to resolve real-world problems in an unconventional and highly efficient way.
-
15 Graph data science for cybersecurity
- + Show details - Hide details
-
p.
325
–343
(19)
Cybersecurity has turned out to be an important field of research and development. With the continuous rise of cyber-attacks on application data, business workloads, and IT services, there is a need for unearthing unbreakable and impenetrable security solutions. This chapter is to describe how data science methods come handy in producing and sustaining resilient and robust cybersecurity systems and services.
-
16 The machine learning algorithms for data science applications
- + Show details - Hide details
-
p.
345
–378
(34)
It is going to be data-driven insights and insights-driven decisions and actions for the total humanity. Data is being recognized as the new fuel for any individual, innovator, and institution to envisage and deliver smart and sophisticated services to its clients and customers. Data is being touted as a strategic asset for any enterprise to insightfully plan ahead and provide next-generation offerings and premium services with clarity and confidence. Newer products and solutions can be unearthed and deployed to assist humans in their everyday decisions, deals, and deeds. However, for data to be overwhelmingly beneficial, data getting garnered from multiple places have to be transitioned into information and knowledge. The process for enacting this strategically sound transformation is being continuously updated and upgraded for achieving the required optimization. That is, process excellence is gaining the attention of professors and professionals. Further on, there are scores of automated tools and enabling platforms for empowering this transition activity. Data analytics is being touted as the prime method to extract actionable insights out of data heaps.
In the recent past, with the flurry of artificial intelligence (AI) algorithms, frameworks, libraries, platforms, accelerators, specialized engines, and high-performance processing architectures, AI-enabled data analytics is seeing the reality. Data science is the fast-emerging and evolving domain of study and research for finding viable ways and means that can simplify and streamline the activity of emitting hidden and useful knowledge out of data volumes. In this chapter, we want to dig deeper to spell out the strategic implications of data science technologies, tools, platforms, and infrastructures. Especially how machine learning (ML) algorithms are influencing the futuristic field of data science.
-
Back Matter
- + Show details - Hide details
-
p.
(1)