Cognitive City Platform as Digital Public Infrastructure for Developing a Smart, Sustainable and Resilient City in Indonesia

The development of smart cities has evolved through several generations since the beginning. Many requirements and technologies have been devised to facilitate the implementation of the smart city concept, among which the smart city platform stands out. The implementation of smart city platforms facilitates the advancement of technology-driven services in smart cities, hence enhancing the overall well-being of their residents. Nevertheless, the existing notion of smart cities has evolved into a more comprehensive model encompassing intelligence, sustainability, and resilience. The paradigm shift in urban development necessitates fulfilling diverse needs by both smart cities and platforms. Research about the current smart city platforms necessitates adaptation to align with the criteria for developing smart, sustainable, and resilient cities. This study presents the Cognitive City Platform as a digital public infrastructure compatible with advancing intelligent, sustainable, and resilient urban areas. The adaptability of the system architecture on the cognitive city platform allows for its implementation in cities of different scales and sizes. Additionally, the platform incorporates city anomaly analytics as its initial level of cognitive functionality. This study additionally presents findings regarding the practical usage of the Cognitive City platform and empirical data about assessing its performance and scalability.


I. INTRODUCTION
In conjunction with the rapid pace of urbanization, cities across the globe are encountering several issues about environmental management, social welfare, and economic growth [1].These issues led to the development of the smart city concept.The notion of a smart city encompasses the integration of human beings as valuable social entities, engaging with infrastructure and technology to foster economic development within a conducive environment [2].The concept of a smart city is to establish a cohesive and interlinked system wherein individuals, technology, and the environment coexist harmoniously to foster economic advancement, The associate editor coordinating the review of this manuscript and approving it for publication was Edith C.-H. Ngai .environmental sustainability, and an enhanced standard of living for all inhabitants [3].
The notion of smart cities utilizes technology, data analytics, and networked systems to enhance urban functioning and deliver efficient services.The utilization of internet connectivity in urban areas and the adoption of applications and data visualization can enhance connectivity among various services and facilitate data-driven decision-making processes [4], [5].Nevertheless, implementing smart city solutions extends beyond technological advancements, as it necessitates focusing on long-term sustainability and considering crucial environmental and social concerns.Smart city implementation aligns with the imperative of establishing a city with intelligent attributes and the capacity to shift toward sustainability [6], [7].
The concept of sustainable cities has gained significant attention in the field of urban planning, as global efforts are being made to tackle environmental issues and establish urban settings that are both habitable and ecologically sound.These urban areas are planned and developed with a holistic strategy emphasizing incorporating sustainable energy sources, effective public transportation networks, ecological landscapes, and environmentally conscious infrastructure.In order to secure the alignment of technical advancements and urban development projects with long-term ecological balance, economic resilience, and societal well-being, smart cities must emphasize sustainability [8].Implementing smart cities that align with sustainability objectives enhances the overall well-being of individuals and safeguards the ecological, societal, and economic resources [9].
In addition to advancing the smart city paradigm towards a smart, sustainable city, urban areas must also strive to integrate resilient capacities [10] harmoniously.The notion of resilience pertains to the development of cities that can withstand and recover from a diverse range of natural disturbances.Resilience raises the question of how a city can establish an effective adaptation system in the face of many types of disasters.The primary emphasis of resilience is on mitigating risks and enhancing the system's capacity to withstand and rebound from diverse hazards and calamities [11].
Given the escalating frequency and intensity of many challenges, such as natural disasters, pandemics, and climate change, establishing resilient communities and infrastructure has emerged as a crucial objective worldwide.Resilience encompasses the capability to recover from disturbances and the aptitude to adjust, create, and flourish amidst persistent fluctuations and unpredictability [12].The primary determinant of a city's economic activities is its infrastructure.In disaster-induced damage to urban infrastructure, various challenges emerge, including a compromised environment, limited resource availability, and disrupted social cohesion.These burdens pose significant implications for the well-being of urban dwellers, underscoring the criticality of city resilience to uphold the stability and functionality of urban systems.
Technology is pivotal in enhancing resilience by providing inventive solutions that can enhance early warning systems, facilitate effective communication during emergencies, and support decision-making based on data.By utilizing advanced technologies such as artificial intelligence, the Internet of Things, and big data analytics, society can enhance its ability to anticipate and respond to potential threats, fortify its resilience, and ensure a safer and more sustainable future for all.Consequently, incorporating Artificial Intelligence and data analytics represents vital technological features for achieving a resilient city.Artificial Intelligence (AI) has the potential to assist in the identification and prediction of weather patterns, hence facilitating decision-making processes for disaster mitigation [13].
The advent of smart city platform technologies has become a pivotal element in contemporary urban development, as they facilitate the integration of diverse data sources and systems to enhance efficiency, maximize resource utilization, and elevate the general well-being of inhabitants.This platform functions as a fundamental infrastructure for managing and analyzing extensive volumes of data across diverse industries, including transportation, energy, healthcare, and public services.Nevertheless, notwithstanding their apparent benefits, current smart city platforms mostly concentrate on technology and analytics that are excessively tailored to individual cities.In order to address these constraints, it is imperative to transition towards cognitive and networked systems.A cognitive approach could be implemented using an integrated platform with artificial intelligence and machine learning techniques to analyze data and generate well-informed judgments in real time.This approach would facilitate the advancement of proactive, adaptive, and holistic urban management, thereby linking smart city projects with the overarching objectives of sustainability and resilience [14].
Utilizing cognitive systems encompassing machine learning and artificial intelligence has facilitated urban areas tackling intricate issues and enhancing effectiveness across many disciplines.This technology facilitates the analysis of vast quantities of data by urban areas, anticipating patterns and formulating data-centric strategies to enhance resource allocation and the quality of services provided to inhabitants.Nevertheless, despite the plethora of cognitive systems at our disposal, there is a lack of consensus over the definitive criteria defining a ''cognitive city.''The distinct setting, objectives, and difficulties of each city necessitate tailored solutions per their individual requirements and priorities.The concept of cognitive systems encompasses integrating interconnected networks of intelligent devices and sensors and advanced analytics capabilities that provide real-time monitoring and adaptive reactions to dynamic urban situations.
Furthermore, cognitive cities must foster a collaborative environment between the public and private sectors, actively involving inhabitants in order to guarantee that technology is utilized in a manner that effectively tackles their challenges and enhances their daily experiences.The continuing evolution of technology necessitates an ongoing effort to define the fundamental attributes of a cognitive city.This undertaking demands adaptable methodologies that can accommodate urban communities' dynamic requirements and aspirations [15].
Cognitive cities embody a progressive and visionary urban paradigm that harnesses cutting-edge technology, including artificial intelligence, machine learning, and the Internet of Things, to establish intelligent and adaptive urban ecosystems.In a cognitive urban environment, data from many sources, such as intelligent gadgets, sensors, and citizen interactions, is systematically gathered and promptly scrutinized to extract significant insights and facilitate well-informed 120158 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
decision-making processes.These insights can be utilized to enhance the efficiency of city services, optimize transportation systems, enhance public safety measures, and tackle environmental concerns.The significance of assistive technologies becomes crucial in this context.Cognitive cities can promote inclusivity and facilitate the full participation and benefit of individuals, irrespective of age or ability, in transforming urban technology through the development and integration of assistive technologies.The integration of assistive devices in cognitive city design can enhance the efficiency and scope of intelligent systems while promoting inclusivity and equity within urban environments [16].The presence of intelligence within a city is insufficient to address all prevailing problems and difficulties.Therefore, an additional element is required, as more than the mere incorporation of information and communication technology is required [17].Instead, a city must cultivate adaptability, proactivity, and predictiveness.Establishing an adaptive, proactive, and predictive urban environment necessitates a foundation rooted in sustainability and resilience.The objective of sustainability is to establish a city that achieves economic growth through effective resource management while enhancing its residents' well-being.
Consequently, it necessitates the capacity to mitigate, endure, and recuperate from such disruptions or calamities [18].In order to address the issues mentioned above, a proposal is put out for implementing a cognitive city platform.This platform aims to meet the unmet requirements for digital platform support within a smart city platform, enabling the realization of an adaptable, proactive, and predictive urban environment.The primary findings of this study are mentioned below.
• Proposing a novel cognitive city platform architecture for the needs of developing smart, sustainable, and resilient cities that are adaptive to various city scales • Constructing the city anomaly analytics framework as a basic cognitive owned by the cognitive city platform to help identify abnormal city situations • Evaluation, presentation, and discussion of implementation results in real-world city use cases The writing structure in this research begins with an introduction that helps direct the problem and research.Section II presents the research background that led to this research, including smart, sustainable, and resilient cities, cognitive cities, cognitive city platforms, and city anomaly analytics.
Section III details the system design we propose in this research.In Section IV, we implemented our system design, architecture, and city anomaly analysis in one of the cities included in the 100 resilient cities.In Section V, we evaluate and analyze the performance of the cognitive city platform and city anomaly analysis.The final section presents the impact of our research on the city as a result of implementation and discusses future works.

II. BACKGROUND
The background structure presented in this research attempts to identify research that leads to the proposed cognitive city platform.

A. SMART, SUSTAINABLE & RESILIENT CITIES
The phrase ''smart'' encompasses intelligence, adaptability, and efficiency, which have been incorporated into diverse facets of our daily existence [19].The current epoch of intelligent technology presents significant opportunities for addressing intricate problems, including enhancing healthcare systems, promoting sustainable resource management, and other related endeavors.Nevertheless, in light of this expeditious advancement, it is imperative to prioritize responsible advancement and confront ethical concerns, protect user privacy, and foster inclusion.As society embraces the paradigm shift toward intelligent technologies, we must direct our efforts toward effectively utilizing this intelligence to foster a more interconnected world that yields advantages for all individuals [20].
Smart cities hold significant promise in enhancing living conditions by efficiently managing and utilizing limited urban resources.However, the lack of accessible real-world data sets and testing environments for evaluating the efficacy of planned models and approaches has impeded research advancement [21].Within the dynamic field of technology informatics, ''smart'' has emerged as a prevailing and influential term, undergoing significant evolution and generating disruptive effects.''smart technology'' encompasses incorporating artificial intelligence, machine learning, and data analytics into diverse devices and systems, facilitating their capacity to adapt, acquire knowledge, and generate intelligent choices [22].Integrating technology, ranging from smartphones to smart homes and even entire smart cities, has brought about a revolutionary transformation in our interactions and utilization of technological advancements.
Moreover, with the continuous progress of smart technology, there exists the possibility of effectively tackling intricate issues about healthcare, transportation, and environmental sustainability [23].The advancement of intelligent systems is crucial for the future of technological informatics, necessitating a steadfast commitment to enhancing privacy measures, addressing ethical concerns, and promoting equitable access to the advantages derived from these advancements.In the current era of smart technology, carefully managing the interplay between innovation and responsible implementation is imperative.This approach enables society to effectively harness the benefits of these state-of-the-art advancements [24].
The emergence of the smart notion can be attributed to integrating information technology encompassing hardware, software, and network technology.This integration aims to enhance individuals' cognitive abilities in making informed business process decisions within a collective setting, such as a group or organization.Smart devices are electronic information and communication equipment with artificial intelligence, machine learning, and large-scale data processing capabilities [25].Artificial intelligence facilitates the provision of cognitive information to previously nonresponsive equipment, thereby enhancing their intelligence.Additionally, gadgets are endowed with computational capabilities and connected to the Internet, forming the Internet of Things (IoT), which enables them to adapt to unforeseen circumstances [26], [27].Many applications can be found, from small devices to asset tracking devices, including transportation, data management, libraries, trash management, and more.Smart gadgets establish a network connection through several wireless protocols, enabling interactive communication [28], [29], [30].
Sustainability has emerged as an important pillar of modern society, reflecting the urgent need to protect and preserve our planet for future generations [31], [32].At its core, sustainability is about meeting the needs of the present without compromising the ability of future generations to meet their own needs.It includes a holistic approach to environmental, social, and economic considerations [33].Sustainable living goes beyond environmental concerns and promotes social equity, fair labor practices, and responsible consumption.As individuals, businesses, and governments increasingly recognize the importance of sustainability, we are witnessing a transformative shift towards greener practices, sustainable development, and the adoption of green policies [34], [35].
Sustainable informatics focuses on optimizing energy efficiency in data centers and IT infrastructure, reducing e-waste through proper recycling and responsible disposal, and promoting using renewable energy sources to power computing processes.In addition, sustainable informatics also emphasizes the development of environmentally friendly software and applications that consume fewer resources and the adoption of cloud-based solutions to optimize resource allocation and reduce hardware requirements.By integrating sustainability into informatics, we can contribute to building a more environmentally conscious and sustainable technological urban planning [36], [37].The objective of incorporating sustainability into urban planning is to establish a city that prioritizes inclusive growth, equitable resource distribution, safeguarding the entire community's well-being, and ensuring universal accessibility [38].
Resilience extends beyond basic survival, necessitating the ability to rebound with increased strength and improved preparedness to confront forthcoming disruptions [39].By placing resilience as a top priority, urban areas can augment their capacity to safeguard the welfare of their inhabitants, foster sustainable progress, and effectively overcome adversities, thus fostering the creation of safer and more prosperous settings for all individuals [40].Furthermore, implementing economic diversification and adopting creative techniques are crucial elements of resilience since they empower urban areas to effectively recuperate and adjust to evolving economic environments [41].Through the utilization of technology to encourage community connections, these urban areas have the potential to cultivate robust social networks that provide mutual assistance and support in times of adversity.Resilient smart cities prioritize the ongoing advancement and refinement of their systems, leveraging technology as a potent instrument to bolster their capacity for long-term readiness, responsiveness, and adaptability.This approach aims to cultivate a sustainable and prosperous urban milieu that is well-equipped to meet future challenges [42].The acquisition of toughness necessitates the cultivation of a repertoire of skills that can be progressively honed.The notion of resilience pertains to the capacity to experience and adjust to challenges, alterations, bereavement, and uncertainty.The concept of resilience in urban areas pertains to establishing cities that can effectively absorb, recuperate, and proactively prepare for forthcoming disturbances.By fostering resilience, cities may facilitate sustainable development, foster prosperity, and promote inclusive progress [43].
A resilient city requires urban designer education equipped with awareness, skills, and practical experience to propose appropriate solutions to face the speed of urbanization [44].Cognitive cities epitomize the forefront of urban development, wherein sophisticated technology like artificial intelligence (AI) and big data are harnessed to establish profoundly intelligent and interconnected urban landscapes.Cognitive cities involve collecting, analyzing, and utilizing extensive data created by a range of sensors, devices, and systems to enhance people's quality of life (45).Cognitive cities utilize the abundance of information available to them to make well-informed decisions, predict future requirements, and proactively address obstacles, such as congestion or environmental issues (46).Furthermore, there is a strong emphasis on citizen involvement and empowerment, whereby residents are actively encouraged to participate in decision-making processes and contribute to the cognitive environment of the city.Cognitive cities strive to establish urban environments that are seamless, efficient, and sustainable, with the primary objective of utilizing technology to enhance the well-being of their inhabitants through constant learning and adaptation [47].
In this version, the focus is on utilizing technology to support community activities in the city.High-tech equipment and software are used in various fields, including health, road transport, parking, and government services.The first-generation smart city presents the concept of six pillars, including economy, mobility, government, environment, society, and lifestyle [48].Due to the high rate of urbanization, it is necessary to assess the contribution of platforms to creating new tools for the development of human-centered, smart, sustainable, and resilient cities.This platform is used as a tool in urban community activities [49].
Several cities in Indonesia have implemented the smart city concept in their city governance.In order to realize a Smart City in the City of Surabaya, it is led by the mayor and all government agencies of the City of Surabaya.One way to conceptualize a smart city is as an icon of a sustainable and livable city [51].Building a smart, sustainable, resilient city is similar to a complex infrastructure development megaproject in which several diverse stakeholders, such as contractors, urban planners, and policymakers, provide services integrated with advanced technologies such as artificial intelligence, big data, autonomous vehicles, and technology.advanced and others [52].As a result, smart city projects often need help to plan, implement, and operate.Understanding these challenges in integrating technical areas into comprehensive planning and management is essential.In addition, governance structures and policies that encourage various specializations in smart cities play an important role in implementing complex projects [53].Apart from Surabaya, several cities implementing the smart city concept are Denpasar and Semarang [54], [55].
The resilient city concept has also begun to be applied in several regions of Indonesia.The implementation of the Resilient City Concept in Semarang City is built on six pillars, namely sustainable water and energy: efforts to increase the supply of urban drinking water, but the reach of services still needs to be expanded.New economic opportunities: Collaboration between government, private sector, universities, and Community Social Institutions (NGOs) to increase research opportunities in new strategic economic sectors.Preparedness for disaster risks and disease outbreaks: There are various activities to increase the dissemination of information, innovation, and collaboration in preparation for disasters and disease outbreaks.Integrated mobility: improve mobility performance, improve the quality and quantity of public transportation, and develop a public transportation management system.Public information and transparent governance: maximizing Semarang's development performance through improving participatory processes and coordination systems and developing public information systems through data integration.Competitive HR: increasing the capacity of the Semarang workforce through the provision of career information centers, training, and workshops, as well as the provision of educational information [56].The resilient city concept applied in the BWK V area of Kendari City, which includes Kadia, Poasia, and Kambu Districts, is based on four aspects of infrastructure resilience, namely road networks, clean water networks, drainage networks, electricity networks, and green open spaces as follows; Road Network: application of Robustness, namely the ability to withstand extreme conditions, for example disasters while maintaining its function.Water Network: Implementing the concepts of Robustness and Resourcefulness to ensure that every residential area has clean water reserves.Drainage Network: There is development of a drainage network that is evenly distributed throughout the area with sustainable planning.Electrical Network: existence of communication technology and emergency information systems.Green Open Space: reduces the impact of flooding and maintains groundwater quality [57].The sustainable concept is applied to the Danish city of Copenhagen, which aims to create adaptation and mitigation initiatives, a city without waste, an integrated public sector, and an environmentally friendly alternative fuel transportation system to improve the population's quality of life [58].

B. COGNITIVE CITY PLATFORM (CCP)
Currently, cities have evolved from several generations of smart cities and added the characteristics of cities, namely sustainable and resilient.With changing city conditions, knowledge is needed that continues to be collected and can be analyzed according to the city context.In this context, stakeholder actions need to provide decisions based on data and information that are appropriate and adaptive to change [59].Cognitive abilities must be adopted from the smallest entities that inhabit cities, namely humans.
The cognitive concept is more than just intelligent; cognitive supports computing that operates at a higher level by connecting devices.Smart city focuses on real-time information, while cognitive city aims to be both real-time but also proactive and predictive.A cognitive city enables continuous collaboration and interaction between technology and residents.The main pillars of cognitive city have at least the following points.

• Cognitive Computing
Leveraging artificial intelligence technology to enable cities to gain insight into citizen behavior, predict potential problems, and how to take action to prevent them.
• Community Involvement Involve the community in the decision-making process by utilizing digital platforms.The digital platform allows a city to gather feedback and insight from citizens, which can be used to shape policy decisions and improve city services.
• Interconnected Systems Creating systems and services that are optimal for the community by integrating all existing services such as public transportation, public utility services, health services, emergency services, and other services • Publicly Accessible Data Making publicly available data available to citizens, businesses, and organizations enables each entity to address urban challenges.The concept of cognitive city pertains to integrating Internet of Things (IoT) technology and developing smart cities, which generate substantial amounts of data.This data is then analyzed and processed using artificial intelligence techniques.IBM Watson is a prominent example of a commercial product that has transitioned to a cognitive framework.This cognitive system encompasses a range of analytical and machine-learning services built upon the principle of dynamic learning.In other words, the system continually enhances its learning process in subsequent iterations by incorporating feedback from prior iterations.The phrase ''cognitive computing'' is employed by IBM to refer to computers that can acquire knowledge from diverse data sets, engage in reasoning processes, engage in natural language interactions with humans, and develop contextual experience [60].A cognitive city is regarded as a progression from the smart city concept.The notion of a smart city often focuses on using data sensing and processing techniques to enhance the efficiency and sustainability of urban services.These services commonly include smart lighting, smart energy management, smart water management, and several other applications.In contrast, the cognitive city concept seeks to not only attain the objective mentioned above but also foster resilience in providing services.Moreover, cognitive cities encompass not just technological information sources but also various additional sources, including cultural, educational, spatial, economic, and political information.
In the smart city concept, if a resident wants to report a water leak on the road.So, he must search several websites or apps to find the right department and then fill out a form to report the problem.The report must be processed after a while, and it can take several hours or days to fix the water pipe leak problem.However, in the Cognitive City concept, residents only need to take a picture of the water leak using a smartphone and upload it to the Citizen platform, which uses cognitive computing to analyze the image and determine its exact location and severity.The system then generates work orders and sends them directly to the repair department.Residents can track repair progress in real-time via the platform, which also updates when leaks will be fixed.Meanwhile, the traffic department can direct residents to alternative routes to avoid congestion [61].
The cognitive city platform aims to integrate various technologies and services to create an interconnected and smart urban environment and provide analysis and data-driven decision support systems to help stakeholders.The Cognitive City Platform enables the government to connect directly with its citizens.The government can monitor the infrastructure of city facilities and events in the city area in real-time so that the city government can make decisions quickly based on the data collected.To realize a smart city, a digital-based infrastructure is needed to unite data, information, and technology from various sources to enable efficient and effective management and delivery of urban services.
Platforms must do more heavy lifting and provide general application functionality such as tracking objects in global coordinate space, enabling precise device positioning, and abstracting access to distributed infrastructure [62].The Smart City Platform is built by digital infrastructure consisting of advanced technological elements that can support the sustainability of Smart City.Such as; and process large amounts of data so that stakeholders can access it.As well as other advanced technology devices.The platform must be ready to receive and manage large amounts of information so that it enters the Big Data domain.Therefore, the technology must i) have high availability, ii) be scaled up quickly, and iii) provide a unified interface for all applied technologies [63].
Cognitive city platform implemented in a city certainly has their challenges.The challenges and points in implementing the cognitive city platform continue to be pursued, especially if it can be implemented in Indonesia; these challenges include security and privacy issues [64].The existing platform is centered on government management [65] and the need for inner cities. Understand the needs of users, visitors, and city staff [66].
Artificial intelligence technology in cognitive cities can solve many of the main problems of urban mobility, the most famous of which is public transportation, which allows public transportation to receive and access dates and tracking directly, thereby increasing time and customer satisfaction.There is also traffic management to predict traffic to reduce congestion, pollution, and the number of transportation accidents [67].Smart city platforms focus on real-time information, whereas what is needed now is not just real-time but how to be proactive and predictive.Therefore, a more cognitive system is needed to answer smart city problems.Cognitive enables collaboration and continuous interaction between technology and city residents to make the platform adaptive, proactive, and predictive.The use of technology on smart city platforms is very specific.It only focuses on solving problems that generally exist in large cities, so to realize adaptive, proactive, and predictive needs to be supported by a generic system [68].
The transition from smart to cognitive cities is critical to addressing pressing urban resilience and sustainability challenges.Although smart cities have made significant progress in leveraging technology to improve various aspects of urban life, cognitive cities take these advances further by integrating real-time data analysis and AI-based decision-making [69].Existing platforms are compared using the term IoT Platform, which focuses on data from IoT devices in the comparison done by [5].IoT Data Collection, Semantic Interoperability, Event Detection and Data Analytics, and Application Support Development are some currently available features.
Data management, application runtime, sensor management, data processing, external data access, service management, software engineering tools, city model, historical data, distributed sensing and processing, resource recovery, resource management, and event management are among the functional requirements for a smart city platform according to research [89].The platform aims to be the hub for technology and people to solve city problems in the cognitive concept.
Integration of citizen engagement also plays an important role in encouraging sustainable behavior and raising awareness regarding urban issues.In essence, the evolution from smart cities to cognitive cities is a logical step forward in creating urban environments that are not only technologically advanced but also resilient and sustainable, addressing the complex challenges facing cities today and paving the way for a more livable future [70], [71].
The concept of a ''Smart, Sustainable & Resilient City'' requires cognitive solutions.Cognitive solutions are guided by cognitive methodology and combined with digital platforms and cognitive computing technologies.The solution is a method for solving practical resource management problems in smart cities [72].Complex problems can be solved using a cognitive approach considering all needs and goals [73], [74].

C. CITY ANOMALY ANALYSIS
Various kinds of research have proposed various kinds of analysis that can be implemented in cities.However, the analyses carried out are too specific to certain cities.A city can apply this analysis, but another city cannot.It is necessary to identify what analysis should be used in various cities with various scales and data.In this research, we propose city anomaly analysis as a generic analysis that can be applied in various cities.
Anomalies are an inherent element of the large volume of data generated by smart cities, and it is imperative not to disregard them.This anomaly could indicate a misallocation of resources within industrial systems, which is a matter of great importance in the avionics platform to prevent unexpected problems or to detect unusual behavior in the environment that may be linked to catastrophic events.Consequently, the ability to identify anomalies can substantially impact the system's overall efficacy under surveillance.One of the most challenging aspects of anomaly detection involves establishing the right boundaries that distinguish regular behavior from abnormal behavior.Finding abnormal observations to train the model on is a somewhat infrequent occurrence.In contrast to typical behavior, deviant behavioral patterns are exceptionally infrequent in real-world contexts.The scholarly community has developed effective methodologies for detecting anomalies in past data, doing real-time analysis, and predicting unforeseen patterns in smart cities [75].
Anomaly analysis empirically analyzes values in a data set for unexpected behavior to provide an initial baseline overview.It reveals potentially flawed data values, elements, or records.City anomaly analysis involves observing and evaluating unusual, abnormal, or inappropriate patterns in the urban environment.These anomalies can involve various aspects, including social, economic, environmental, and infrastructure.The development of communication and sensing technology is increasing rapidly, and this causes a large amount of heterogeneous data to form the basis of city anomaly or urban anomaly.City anomaly is an event that occurs in an urban environment, such as traffic jams, unexpected crowd gatherings, disasters, accidents, and traffic regulations that can pose a major threat to public security and stability if not handled promptly.On the other hand, urban anomalies are used as a detection and prediction tool to improve citizens' quality of life and minimize losses caused by urban anomalies.For example, traffic jams are the most confusing problem in big cities today.Severe traffic jams can bring a lot of economic losses and spoil people's moods.If most of the traffic jams in a city could be predicted or detected early, they could be avoided by telling people to change their travel or transportation routes.In this way, people will save time on travel and improve their quality of life [76].
In the context of smart cities and urban planning, a city anomaly can refer to an irregular or unexpected pattern or event within a city that may have implications for its smart city initiatives.Here are a few examples of what this could mean: • Traffic Congestion Anomalies: In a city, traffic data is often collected and analyzed to optimize traffic flow.An urban anomaly in this context could be a sudden, unexplained traffic jam or congestion that disrupts the usual traffic flow [77].
• Energy Usage Anomalies: Urban areas frequently strive to optimize energy consumption [78], [79].Anomalous energy usage refers to an abrupt and atypical power or water consumption surge inside a certain locality or structure.
• Public Transportation Anomalies: Within a technologically advanced urban environment boasting a highly efficient public transportation system, an anomaly can be defined as an unforeseen occurrence resulting in a delay or disturbance in the provision of transportation services [80].
• Safety and Security Anomalies: In the present context, an urban anomaly could manifest as an atypical surge in criminal activity or an abrupt alteration in accident patterns.Identifying such abnormalities can facilitate the efficient response of law enforcement and emergency services [81].
• Environmental Anomalies: Urban areas frequently engage in the surveillance of atmospheric conditions, levels of pollution, and various other environmental parameters.These anomalies may serve as indicators of environmental concerns that necessitate prompt intervention [82].
These technologies can assist municipal authorities and urban planners in adopting a proactive approach toward unexpected events while enhancing the efficiency of city services to benefit citizens.
In the multi-perspective concept, the following are the methods used to detect anomalies [83]: • Conventional Techniques The conventional method refers to the primary approach employed for anomaly detection.Statistical methodologies employ a statistical model to represent data, wherein departures from the expected patterns are regarded as anomalous occurrences.The approach can be classified into two categories: parametric and non-parametric.In the parametric category, the data distribution underlying the analysis is assumed to be known, and the distribution parameters are estimated using the available data.Parametric approaches encompass a variety of statistical techniques, such as Gaussian Models, Regression models, and combinations of Parametric Distributions.
Conversely, in the non-parametric case, the parameters of the underlying distribution are unknown; they are determined from existing data.Examples include methods based on histograms or Kernel Functions.
• Data-Driven Techniques Data-driven techniques typically pertain to learning-based approaches that rely on substantial volumes of data to compensate for the lack of a robust underlying mathematical model.These methods enable extracting valuable information through ''learning'' from the available data.Machine learning is a comprehensive academic domain encompassing numerous areas of practical implementation.Typically, it is categorized into three primary classifications: supervised learning, unsupervised learning, and reinforcement learning.Nevertheless, it is worth noting that there exist alternative approaches, such as semi-supervised learning.Furthermore, with the progression of technology, there is a concurrent rise in deep learning.Numerous machine learning methodologies frequently adopt a deep approach or are integrated with deep learning approaches.
Similarly, research [84] distinguishes three fundamental approach functions to the outlier detection problem, viz: 1. Supervised: Normality and abnormality are modeled; require labeled data for each category.Example: Use machine learning algorithms such as Decision Trees, Naïve Bayes, and rule-based classifiers.2. Unsupervised: Identify anomalies without prior knowledge of the data.Example: clustering algorithm.3. Semi-supervised: Only normality is modeled; the anomaly is identified because it is not within normal limits; it is also known as novelty detection or novelty recognition.Example: One Class Support Vector Machine and Isolation Forest.These methods can be applied in the cognitive city platform by adjusting the options according to the characteristics of the city.

III. SYSTEM DESIGN A. CONCEPTUAL FRAMEWORK
The conceptual framework of the Cognitive City Platform was built to provide a fundamental understanding of how CCP works can help develop smart, sustainable, and resilient cities.The first concept starts from how a system can perform sensing of the city.This sensing process is carried out by obtaining maximum and optimal data from various sources.In the urban landscape, understanding is the foundation for transformative progress.The ability to discern patterns from a vast ocean of information, understand subtle systems of interaction and understand the needs and aspirations of its diverse population distinguishes truly cognitive cities from other cities [85].However, understanding alone is not enough.The urgency of action becomes paramount once understanding is achieved.In an ever-evolving world where poverty, prosperity, and resilience require urgent attention, understanding must quickly translate into decisions and initiatives that drive positive change.Cognitive cities thrive not only in terms of knowledge but also in the responsive actions they trigger, ultimately shaping a better, smarter urban future.

1) SENSING
Sensing is a mechanism for sensing a system's external and internal conditions.In the context of a smart city, sensing is an activity in obtaining data provided by the environment.The sensing concept can consist of physical sensing and digital sensing.By definition, physical sensing is an activity to get data from the physical world, such as what the temperature is today, how much humidity is today, how many vehicles have passed, how many people have been detected, and so on.An electronic system such as a sensor is needed to carry out physical sensing, which can retrieve the data according to needs.Digital sensing is an activity to obtain data from the digital world, such as the number of visitors today, the number of comments, and the infrastructure traffic load today.Almost all digital sensing only uses script software to obtain data.
The hardware used for digital sensing is usually basic digital infrastructure such as computers, servers, internet networks, and mobile devices.

2) UNDERSTANDING
Understanding is a system concept that can analyze data and produce certain insights within the scope of a particular process.In understanding the concept, there is at least an analytical process consisting of descriptive, predictive, prescriptive, and diagnostic.Cognitive understanding is important in urban solution support systems, offering a 120164 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.dynamic bridge between human wisdom and technological prowess.Designing these systems involves more than just data processing.The system requires the inclusion of cognitive abilities that enable them to understand the intricacies of urban life.By leveraging machine learning, natural language translation, and advanced analytics, these systems can analyze huge data sets, recognize hidden patterns, and gain a deep understanding of urban dynamics.This cognitive understanding empowers city decision-support systems to offer real-time insights, anticipate emerging trends, and provide appropriate solutions to urban challenges.It is about creating intelligent, adaptable systems and collaborating with city administrators, citizens, and stakeholders to create more efficient, sustainable, and resilient cities.A cognitive understanding of system design drives urban decision support systems in an increasingly complex and interconnected urban landscape.

3) ACTING
The urgency to take decisive action against cities has never been more apparent than in data gathered from our urban environment.The sheer volume and speed of information cities generate today has become a clear need for urgent action.Climate change, traffic congestion, public health crises, and social inequality require rapid, data-driven interventions.However, data alone is not an explanation; the ability to sense, understand, and act on data is what truly matters.To fully exploit the potential of this insight, cities need a robust digital platform that can seamlessly integrate data flows from multiple sources, interpret them in real time, and facilitate informed decision-making.This digital platform serves as the centerpiece of modern urban governance, enabling cities to sense emerging challenges, understand their differences, and take proactive steps to build more resilient, sustainable, and liveable urban environments.Digital platforms are key to turning knowledge into impactful action in a world shaped by data [39].

B. PLATFORM ARCHITECTURE
The cognitive city platform architecture comprises several components organized in a distributed system.These components work comprehensively to fulfill the core services of the platform.

1) URBAN DATA
Urban data refers to spatio-temporal data generated by mobile devices or distributed sensors in an urban environment.Urban data refers to factual information that is documented and pertains to cities.The data is not solely derived from mobile devices or dispersed sensors.City-generated data can be derived from various sources, including data input into municipal services.Examples of such data include disaster, population, and unstructured data, such as social media and online media.These sources contribute to the contextual understanding of conversations about a city.The collection and analysis of urban statistics necessitate the establishment of specific criteria and metrics to depict the factual occurrences within a metropolis accurately.For instance, variables such as temperature measured in degrees Celsius, water level sensors calibrated in meters, or the quantification of mentions or likes received by a specific municipal post on social media.These measurements serve as a gauge for objective information or occurrences within the urban area.

2) PROACTIVE SENSING MANAGER
The current system serves as a platform for managing many systems to acquire and organize urban data.Contemporary scholarly investigations mostly center on the methodologies employed to extract data from a given source and store it within a database.In addition to this, the dispersion of urban data presents a significant challenge for municipal governments in autonomously gathering data.Consequently, the involvement of external entities, such as the corporate sector and communities, becomes imperative in the data collection.The Proactive Sensing Manager serves as a cognitive platform owned by the local government, enabling the management of individuals who express their willingness to assist the government by registering their scripts, software, and servers as components of the urban data collector.The municipal administration is capable of overseeing the entities responsible for urban data collection and delineating the types of urban data that are prohibited from being collected.In order to enhance the recognition of the private sector and local communities inside the urban setting for data collection, the government has the potential to enact suitable incentive programs.The Proactive Sensing Manager facilitates the utilization of Sensing Tools, which are created by many entities and can take the shape of software, hardware, or participatory sensing.Using a proactive sensing manager, the government can efficiently gather a substantial amount of urban data to expedite the acquisition of knowledge pertaining to city circumstances.Urban data can be defined as spatio-temporal data produced by mobile devices or dispersed sensors in an urban setting.Urban data encompasses verifiable and empirical information that is systematically recorded and specifically relates to urban areas.The data under consideration is not exclusively obtained from mobile devices or distributed sensors.City-generated data can be obtained from many sources, encompassing the data entered into municipal services.Illustrative instances of such data encompass disaster, population, and unstructured data, such as social media and online media.These materials enhance the contextual comprehension of discussions related to an urban area.Properly representing factual occurrences inside a metropolis requires adopting certain criteria and metrics for gathering and analyzing urban statistics.For example, variables such as temperature, measured in degrees Celsius, water level sensors calibrated in meters, or the quantification of mentions or likes obtained by a specific municipal post on social media can be considered.These metrics function for quantifiable data or events within the metropolitan region.

3) DATA PREPROCESSING MODULE
The data preprocessing module inside a cognitive city platform plays a vital role in managing the preparation and conversion of raw data obtained from diverse sources within a smart city setting.This module facilitates the cleansing and structuring of data, rendering it suitable for analysis.Consequently, it enhances the platform's ability to extract insights, make informed decisions, and execute diverse activities.The data preparation module of a cognitive city platform encompasses several essential tasks and processes.It contains data cleaning and data transformation.

4) COGNITIVE ANALYTICS MANAGER
Prior studies have primarily concentrated on identifying the optimal algorithm for addressing challenges within the smart city domain.While it is true that each city possesses distinct challenges that set it apart from others, The algorithm devised in prior studies may not necessarily be transferable to diverse urban contexts.The Cognitive Analytics Manager module can effectively oversee diverse data analytics algorithms, adeptly organize data flows, and ascertain crucial indicators of critical performance data.Algorithms and analytical data models created by city governments, private sector groups, or communities can be uploaded into the Cognitive Analytics Manager.This method allows for a more comprehensive examination of a city, as it benefits from the collaborative efforts of these different entities.

5) ALGORITHM SELECTOR
Several scholarly investigations about data analytics in smart cities center on enhancing an algorithm and its subsequent implementation inside a certain use case or area.The Cognitive City Platform does not possess a universally applicable algorithm capable of resolving every given data analytics issue within a smart city context.Various groups, including governmental, business, and community organizations, have the potential to make valuable contributions by sharing their algorithms that are relevant to specific situations.In order to facilitate experimentation with their data, municipal stakeholders can choose an algorithm that is deemed to yield the most optimal analysis.This selected method can then be stored in flow storage for future reference and utilization.

6) FLOW STORAGE
Consequently, the algorithm can be chosen based on the most suitable case study, allowing flow storage to effectively store data for selection and testing utilizing a specific method.The flow storage contents of two cities may vary depending on the specific urban issue they aim to address.

7) DATA KPI CONFIGURATION
Every individual datum possesses distinct properties.In order to offer notification to stakeholders, it is sufficient to establish both the upper and lower limits for data that is known with certainty.Furthermore, the KPI configuration data also encompasses information regarding the reliability of the configuration data for analysis purposes.The present features possess the potential for further development by assessing data trustworthiness, enabling the analysis of said data.Consequently, the algorithm can be chosen based on the most suitable case study, allowing for the storage of data flow and selecting and testing a specific approach.The flow storage contents of two cities may vary depending on the specific urban issue they aim to address.

8) ONLINE ANALYTICS
The capability to execute real-time data processing for applications that require interactive functionality within a constrained latency threshold [86].
120166 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

9) REAL-TIME ANALYTICS
The ability to execute flow data processing in real-time applications.Real-time applications operate inside the temporal framework as perceived by the user at the present moment.The delay must be below the stated threshold [86].

10) BATCH ANALYTICS
The present study focuses on conducting batch data analysis for applications that produce delay [86].

11) ACTION MANAGER
Once the data has been subjected to a specific form of analysis, it yields particular insights.This insight will serve as a stimulus for the action manager to allocate appropriate actions to individuals at various levels within and across different agencies.Suppose water level monitoring has surpassed its maximum threshold.In that case, a message will be promptly forwarded to the Public Works Department and the Disaster Management Agency to facilitate the dissemination of an early warning.

12) CREDENTIAL MANAGER
The credential manager oversees the management of data and access permissions on the platform.Credential Manager encompasses various subsystems, including User Management, Organization Management, Application Management, Single Sign On, and Interface Management, each serving distinct functions.The various subsystems work synergistically to effectively control and regulate access privileges to the Cognitive City Platform.

13) USER MANAGEMENT
The subsystem manages user registration, profiles, passwords, and role assignments about data and applications associated with the Cognitive City Platform.

14) ORGANIZATION MANAGEMENT
The primary function of this organization is to oversee the management of user groups within a certain department, private sector, or community.Additionally, it facilitates the integration of other systems with the cognitive city platform.

15) APPLICATION MANAGEMENT
In order to be integrated with the Cognitive City Platform, every application must undergo registration within the Application Management subsystem.The urban applications will possess the capability to monitor their login activities.The rationale behind integrating the Cognitive City Platform with other urban systems is to implement Single Sign On (SSO) functionality.The city-level access enables city inhabitants to access smart city services offered by different vendors through a unified authentication process, requiring a single set of login credentials.

16) INTERFACE MANAGEMENT
An interface is required for communication between any apps or devices associated with the Cognitive City Platform.
Implementing Interface Management functions is necessary to effectively regulate and monitor Application Programming Interface (API) rules.

17) DATA MANAGER
Managing collected and analyzed data is essential to protect each data point.Within the data management system exists a component known as the City Master Data.Every municipality globally possesses distinct datasets encompassing various aspects such as population statistics, healthcare indicators, environmental measurements, and other data.The provided data can serve as an initial profile and benchmark for use within the KPI Configuration Data subsystem.In order to enhance the compatibility of systems, the Data Manager is provided with a RESTful API Generator.This feature enables sharing any pre-existing data collection following the data access permissions established by the Data Role Manager.The Data Role Manager assigns read-only or write-only permissions to each data collection.The Data Role Manager engages with the Credential Manager for access management via a dedicated administration interface.

18) FILE STORAGE
The data obtained from urban areas encompasses several formats, including tabular and textual representations and visual media such as pictures, videos, and spreadsheets.The data might be saved in file storage in the form of files.

19) STRUCTURED DATABASE
Utilizing a structured database helps in the organization and management of the extensive volume of data generated and gathered from diverse sources inside the urban environment.The structured database functions as a fundamental framework for the storing, retrieval, analysis, and decision-making processes related to data.

20) UNSTRUCTURED DATABASE
Unstructured databases or data stores are employed to manage and analyze non-tabular or non-relational data.A need for more organization and the absence of a predetermined framework commonly characterizes such data.Unstructured data within a cognitive urban environment can comprise diverse information, such as textual content, visual imagery, audiovisual recordings, social media updates, sensory measurements, and other similar data types.

21) END USER INTERFACE
The End User Interface refers to a system established by various entities, such as the local government, commercial sector, or community, to represent data collected and processed visually.The End User Interface can manifest as a web application, mobile application, or display on an Internet of Things (IoT) device in reaction to either the Action Manager or the Data Manager.

IV. IMPLEMENTATION
Implementing the Cognitive City Platform involves a collaborative effort with the Semarang City Government, an active participant in the Rockefeller Foundation's 100 Resilient Cities initiative.The subject of discussion is Semarang City.Semarang City serves as the administrative and political center of the province of Central Java, situated in Indonesia.This urban center ranks as Indonesia's fifth most populous metropolitan area, following Jakarta, Surabaya, Bandung, and Medan.Semarang, a burgeoning urban center on Java island, exhibits a population of approximately 1,693,035 individuals as of the middle of 2023.Implementing the Cognitive City Platform in the City of Semarang is carried out through a series of steps: 1

. Focus Group Discussion
The activity uses the Proactive Sensing Module to incorporate data maturity observation into the platform.The various offices inside the city of Semarang exhibit varying levels of maturity.A total of 17 outpatient departments (OPDs) and other regional agencies adhered to the observation procedure, from which only 4 OPD samples were subsequently chosen.

Platform Development and Deployment
This activity focuses on developing a cognitive city platform that utilizes web application technology.Subsequently, proceed with the deployment of the platform onto the municipal server.

Platform operation
The Semarang city government uses the cognitive city platform operationally with assistance from the development team.This operational activity uses the official live domain of the Semarang City Government.
A. FOCUS GROUP DISCUSSION Regional Apparatus Organizations (OPD) in Semarang City participated in this discussion through an interview session.The list of OPDs participating in data readiness and integration activities is shown in Table 1.This activity aims to provide socialization on the application of the cognitive city platform with a scope adapted to the needs of the city of Semarang.In addition, each organization conducts an independent assessment to measure the level of data maturity in each work area.The following Table 2 is a data maturity scale used to carry out the assessment.
Based on the results of the independent assessment, the following results were obtained in Table 3 Based on the results of the FGD, 4 data organizations are ready to be integrated into the Cognitive City Platform.

B. PLATFORM DEVELOPMENT AND DEPLOYMENT
In general, the development of the Cognitive City Platform adopts the web application concept, where each subsystem has its framework.

1) ORGANIZATION DATA
Organization data are urban data derived from data collected by the city government.The following information was incorporated into the cognitive city platform based on the results of the FGD: Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.• Data is sourced from the integration system from birth to death, which the Department of Health owns.This data is integrated using the Database to Database method.
• Weather radar data from Meteorology, Climatology, and Regional Geophysics Agency.This data is obtained by downloading the Excel file and migrating it into the database.
• Information regarding disaster reports submitted by locals.This data is integrated in accordance with the collaboration requirements of city residents.A webscraper is used to incorporate the data.
Organizational data is integrated using existing tools as well as with the Sensing Tool implemented in the Proactive Sensing Manager.Semarang City Disaster Management Agency.This data set contains citizen-reported flood events from 2012 to 2022.The sensing tool for Regional Disaster Management Agency is implemented using Figure 4: This flow can be implemented using low code programming described by Figure 5.The program configuration are saved in JSON file.We develop an interpreter to convert JSON (.json) to Robot Framework file (.robot).Robot Framework executes.Robot file and extract the data from website based on programming flow.

3) DATA PREPROCESSING
In this module, the data that has been integrated is transformed into other forms, namely missing data handling, data aggregation, and other testing analyses to improve data quality.Regional meteorological, climatological, and geophysical agency data is Physical Data Source category data, as stated in the data source, and has the characteristics shown in Figure 6.The second data source is from the Regional Disaster Management Agency, and it is cyber data derived from citizen reports to the city government, as shown in Figure 7. Cross-domain data fusion is one method to improve information quality [87].[87].
By looking at the characteristics of the data, the same object in the cross-domain data is the date.Dates become latent objects in cross-domain data fusion processes in data preprocessing.This data fusion process tries to map weather data associated with flood disaster events based on the results of residents' reports.The fusion results can be seen in Figure 9.The data is 12,017 records with the following ratio.
• 276 records were compared between weather data and flood disaster • 11,741 records are juxtaposed between weather data and non-flood data The fusion data is then analyzed using the cognitive analytics manager.

4) CREDENTIAL MANAGER
The Credential Manager is a security layer on the cognitive city platform.This subsystem records all user access in a city, whether public, private, or government employees.A login username and password are required to access Credential Manager.After successfully logging in, the authenticated user can see which applications he can access, as shown in Figure 10.

b: APPLICATION AND INTERFACE MANAGEMENT
The application has an interface that acts as a bridge between the view and the database, so even though it is in a separate architecture, it is included in this presentation.
Application management functions are used to manage applications registered by the organization.Each organization may contain multiple organizations.When adding an organization, government personnel must enter the name, domain, organization, snippet of a screenshot, and whether or not this application is publicly accessible.Third-party sensing tools can be integrated with the Cognitive City Platform through this Application menu.
Figure 13 shows how applications registered with Application Management can be monitored based on SSO-integrated login activity.The benefit of this system is that less actively used applications can be recommended to be turned off to save resources.The less active application occurs frequently in several Indonesian cities; many applications are developed but must be managed with long-term goals.(1) Model Configuration Model configuration is implemented using a drop-down selector that leads to the model or algorithm available on the cognitive city platform as shown in Figure 14.
The model used in this configuration detects city anomalies.Anomaly detection is performed on the preprocessed integration dataset from the previous stage.This configuration enables stakeholders or city staff to select the best configuration based on the data available to them.When implementing the cognitive city platform, this can be adapted to various types of data and city scales so that analysis can still be done first to produce the best results before deciding to build a model and then registering it in the model configuration.
(2) Flow Storage Flow storage is a type of configurable storage.Data analyzed with specific models will be saved in flow storage, which can be used for as long as the data stream exists.PostgreSQL is the data storage vehicle used by Flow Storage.Any government or private organization can save this configuration for their data.As a result, the cognitive city platform will fulfill basic analysis requirements and then save it as a reference.(3) Data KPI Configuration Key Performance Indicator is a feature that is used when there is data whose performance is measured using KPIs determined by City regulations.
Each region has its own set of rules and regulations.In terms of city performance, a city with a population of around 1 million will have different city indicators than a city with a population of 20,000.The KPI Configuration menu makes it easier to differentiate between city indicators.As shown in Figure 11, staff can enter the name of the KPI, then the KPI for which data source, table, column metrics, and column datetime.

5) DATA MANAGER
Data is very important in a cognitive city.City is the continuous production and consumption of data that must be stored in the appropriate container.This database's implementation consists primarily of PostgreSQL, Hadoop, and MongoDB.Furthermore, as shown in Figure 17, the data management dashboard is built with NodeJS Strapi and ReactJS to provide optimal visualization for managers.
In data management, you must also log in using the credential manager to access the API the login is authorized to access.This mechanism facilitates the use of both cognitive city platforms and platforms developed by the private sector and the community to solve problems and make decisions using a single data set.

6) ACTION MANAGER
Data analyzed using a cognitive analytics manager must be communicated to city stakeholders [88].The action manager communicates information when an anomaly in the monitored data occurs.Depending on the needs and readiness of the city's implementation, various types of alerts can be implemented in the action manager, such as notifications in the command center, email, or other messaging such as What-sApp, Telegram, and others.In this study, we attempted to implement one type of alert, namely email via Gmail SMTP and notifications via the application via the default action list in the form of alerts, as shown in Figure 18.

7) END-USER INTERFACE
This subsystem is the cognitive city platform's final appearance.The end-user interface in this study is in the form of a web application deployed in the domain of the official city of Semarang (go.id), as shown in Figure 19.The web application displays various information based on the city's needs.The city's needs are determined by city regulations or regional leadership programs currently in place.This dashboard web application is strategically implemented as a large screen to help stakeholders make decisions.The dashboard can also be linked to other systems developed by each government agency.In addition, a mobile version of the dashboard can be created for easy access.

8) DEPLOYMENT
The developed architecture must be deployed to a server so that only one city can access it.Figure 19 depicts the deployment process, including one load balancer server and two application servers.The go.id domain is mapped to the load balancer, which is then mapped to currently available servers.Nginx is used as a web server in a load balancer.Then, each application has a Docker container that runs each subsystem and supporting systems such as databases.Each subsystem is implemented as a docker image for ease of configuration.

V. EVALUATION AND DISCUSSION
The performance of subsystems with a backend was evaluated in this study to test the backend's ability to serve requests from the user and the front end.

A. PERFORMANCE EVALUATION
The platform's performance is assessed through performance evaluation.Platforms must be able to get the best access due to changing Internet conditions.Performance testing is used to determine an application's or software's level of stability, scalability, and responsiveness to a given workload.Apache Jmeter is a popular tool for running performance tests.Apache Jmeter can perform load and stress testing on RESTful APIs and supports a variety of protocols including HTTP and HTTPS.
The components of performance testing are as follows [89].
• Request, how many requests have entered the server • Min, record the shortest or lowest time required to execute a request.
• Max, record of the longest or longest time required to execute a request.
• Avg, average time spent executing requests.
• Std dev, the spread of the data set relative to its mean.The smaller the std.dev value indicates that the data being executed is more consistent.
• Throughput, the number of requests successfully processed per time unit (seconds, minutes, hours) by the server.This time is calculated from the first request when it is executed until the last request.Throughput is the main metric that shows the number of requests that can be handled in a certain time, the greater the throughput value, the better the performance.
• Kb/s, the amount of data successfully sent or downloaded from the server during the performance testing execution process in units of kb every 1 second.Because of the state of the Internet in Indonesia, throughput metrics were included as part of the evaluation in this study as a reference.

1) DATA MANAGER PERFORMANCE
In Table 4, Data Manager Case, with a continuously added workload of 15,000 the platform answered an average response time of 40 milliseconds in all scenarios explored by the experimenter, even when the platform experienced a large number of concurrent requests.The Throughput value of the Data Manager Case Experiment, seen from the Requests that were successfully processed by the running server, is 121.8 per second.The platform's average throughput increased substantially in scenario testing.The Speed Up metric measures how performance improves as new resources are added to the system, while the improvement metric measures throughput gains.

2) CREDENTIAL MANAGER PERFORMANCE
In Table 5, the Credential Manager case, with a continuously added workload of 15,000, the platform responded with an average response time of 173 per millisecond in all scenarios investigated by the experimenter, even when the platform was experiencing a high number of concurrent requests.The Credential Manager case has a throughput value of 51.4 per second based on the Requests successfully processed by the running server.

3) COGNITIVE ANALYTICS MANAGER PERFORMANCE
In Table 6, the Cognitive Analytics Manager case, with a continuously added workload of 15,000, the platform responded with an average response time of 71 per millisecond in all scenarios investigated by the experimenter, even when the platform experienced a high volume of concurrent requests.The Request successfully processed by the running server in the Cognitive Analytics Manager case is 449.1 per second.

B. ANALYTICS EVALUATION
The analysis evaluated in this study compares the algorithm used on the Cognitive City Platform between data that has not been aggregated with community reports and data that has been aggregated with community reports to determine how much citizen collaboration influences the quality of the analysis.This study explores how combining environmental sensors with community reports can improve the ability to spot anomalies on flood sensors.Environmental sensor data alone and environmental sensor data combined with citizen reports are compared using an evaluation matrix.
The PyCaret Python library aids in data preparation, modeling experimentation, and testing the performance of data mining models.The Pycaret library was chosen for this study because it can perform many model tests quickly and easily.PyCaret is a wrapper that makes use of several machine learning libraries and frameworks such as scikitlearn, XGBoost, Microsoft LightGBM, spaCy, and others to perform tasks such as data preparation, modeling, model comparison, parameter tuning.The goal of this study is to evaluate the success of implementing an anomaly detection work architecture on datasets that will be processed by various types of data mining models, including Random Forest, k-Nearest Neighbors, Decision Trees, Extra Trees, Nave Bayes, Gradient Boosting, Logistic Regression, and Support Vector.
PyCaret is used in this study for classification using the Confusion Matrix.The confusion matrix is a predictive analytics tool for evaluating the performance of machine learning classification models.The confusion matrix displays and compares the actual and predicted values of the model used to generate the accuracy metrics.
• Accuracy: the model's ability to correctly classify data.
• Precision is the agreement between the requested data and the model's prediction results.
• Recall: the model's ability to retrieve information.
• The F-1 Score compares weighted average precision and recall.If the dataset contains a small number of False Negatives and False Positives, accuracy can be used as a reference for algorithm performance.If the numbers are not close enough, use the F1 Score.Table 7 summarizes the performance of anomaly detection analysis on data prior to fusion.
The table shows that the recall and precision performance is quite small because the data quality needs to be improved by providing human knowledge to the data so that the information in the dataset increases.
Table 8 shows that recall and precision performance have improved while remaining below accuracy.Aside from looking at anomaly performance in the Cognitive Analytics 120174 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.Manager, this comparison can also be made regarding algorithm performance.

C. DISCUSSION
Cities gain data integration functions, citizens and city-level staff management, general anomaly analysis, and action alerts related to data or information that does not comply with KPIs with the development of the cognitive city platform.
The performance of low code programming tools in sensing tools can function properly, taking data and then entering it into the database.However, performance calculations have yet to be performed for this subsystem.Furthermore, the Robot Framework's execution process generates its mance report.Then, in of data preprocessing, the data fusion method was used exclusively in this study.The data was preprocessed using open-source data analytics tools integrated with the cognitive city platform credential manager.
The analysis results communicated via alert only provide information related to the information or data being monitored, and it appears in the email of government personnel who handle this matter specifically.As part of resilience, this action manager system can be recommended as a tool to assist when a disaster occurs.Regarding long-term viability, the cognitive city platform can store city master data, which can be used as another indicator in data monitoring.Even if the internet connection is unstable, the cognitive city platform can still be accessed with many requests.
The final question is how this cognitive city platform can improve people's quality of life.Currently, evaluations can only be conducted in agencies with data related to integration.

VI. CONCLUSION
Digital platforms are one of the pillars that support smart, sustainable, and resilient cities.Digital platforms are tools that can help cities develop more effectively.In this study, a cognitive city platform was developed to assist city governments, particularly in Indonesia, in developing smart, sustainable, and resilient cities as part of the smart city transformation.The developed architecture can be adopted in various city sizes, with several architectural features that can be adapted.Collaboration between city entities is a priority in the cognitive city platform, beginning with the process of sensing, understanding, and acting so that all city entities can equally participate in improving the quality of life, preserving resources so that they can be used as well as possible for the next generation, and increasing resilience both in terms of coordination and from the perspective of information technology.
There are several improvements that can be made in the research, both in terms of implementation and evaluation.Various types of data are required for implementation in order to see the performance of each subsystem in the cognitive city platform.Data that is integrated into the cognitive city platform must be evaluated for its level of trust so that the quality of data and analysis results can be improved in the future.Then, on the evaluation side, several analysis algorithms cannot yet be included in the platform because some algorithms, such as deep learning, require a lot of computing power, necessitating deployment changes.

FIGURE 1 .
FIGURE 1. Smart city to smart, sustainable, and resilient mapping.

1 .
IoT sensor devices allow one object to send data via a real-time connection without the help of computers or humans.2.Artificial Intelligence (AI) and Machine Learning aretechnologies for imitating human intelligence, making predictions, and making decisions.3. Data Collection and Analytics technology to collect data from IoT devices, public sources, and other city systems.This data is then analyzed to gain insights and facilitate data-driven decision-making.4. Cloud Computing.Cloud-based infrastructure to store

FIGURE 2 .
FIGURE 2. Conceptual framework of cognitive city platform.

FIGURE 3 .
FIGURE 3. Architecture of cognitive city platform.

2 )
PROACTIVE SENSING MANAGER Development Flask and Javascript are used in the development of Proactive Sensing Manager and run on port 5002.Proactive Sensing Manager can be accessed by registering or being registered in Credential Manager.There are several Sensing Tools in Proactive Sensing Manager.The Sensing Tool in the Cognitive City Platform is based on Low Code Programming.Several blocks were created to pull data from various organization data sources.In the Semarang City case study, a sensing tool is developed to collect data from the

FIGURE 5 .
FIGURE 5. Low code programming in sensing tool.

FIGURE 6 .
FIGURE 6.Weather data characteristic from meteorology, climatology, and regional geophysics agency.

FIGURE 7 .
FIGURE 7.Weather data characteristic from meteorology, climatology, and regional geophysics agency.

FIGURE 9 .
FIGURE 9. Fusion result of weather data and disaster report.

FIGURE 10 .
FIGURE 10.Credential manager dashboard logged in by city-level Single Sign On (SSO).

Figure 14
Figure14shows how to keep track of the user's application access.The city government can easily manage all city users in one dashboard using this feature.

Figure 18
Figure 18 depicts the number of indicators that must be monitored and used as a reference for data-driven action.

TABLE 1 .
List of Organization in Semarang FGD.

TABLE 2 .
Measurement scale for self-assessment.

TABLE 4 .
Data manager performance test result.

TABLE 5 .
Credential manager performance test result.

TABLE 6 .
Cognitive analytics manager performance test result.

TABLE 7 .
Before data fusion.

TABLE 8 .
After data fusion.