:::

Special Cases

01
2022.9
【2022 Solutions】 Xiaoshi Intelligent Inspection uses the two swords of "AOI + AI" to combine software and hardware to open up the two channels of appearance defect detection and supervision.

Quality inspection, like a double-edged sword, has always been a favorite and painful subject for Taiwanese manufacturers When AI deep learning enters the industrial visual inspection of traditional manufacturing industries, it can not only save inspection manpower investment, solve the problem of inconsistent manual visual standards, overcome the limited visual recognition and defect detection blind spots of traditional automatic optical inspection AOI, and also enable real-time traceability Causes of quality problems The overall AIAOI visual inspection solution developed by Xiaoshi Intelligent Inspection integrates software and hardware to create efficient appearance defect detection capabilities, helping electronics OEM customers create high-efficiency products with a miss detection rate of less than 1 and an overkill rate of less than 3 Check the level Xiaoshi Intelligent Inspection was established in 2020 Although it is a new venture two years ago, it did not start from scratch Founder and CEO Hong Peijun and the core team have been deeply involved in Foxconn factories for many years and participated in countless smart factory-related solutions and process improvements , has profound AI deep learning development capabilities, and accumulated rich experience in world-class AI application implementation Seeing that AI industrial inspection must be the last mile for the manufacturing industry to move towards Industry 40, Hong Peijun resolutely decided to implement AI deep learning technology in the field of smart manufacturing with high output value, and specialized in the development of AI industrial visual inspection For the manufacturing industry, product inspection is the most important part of all quality control, but traditional industrial inspection faces two major pain points 1 Manual visual inspection Today, more than 95 of the entire manufacturing industry still relies on manual visual inspection Inspection makes it difficult for manual visual quality inspection standards to be consistent, and visual inspection of fine objects, such as passive components or highly reflective components, will cause long-term vision damage 2 Traditional AOI automatic optical inspection The product has limited visual recognition capabilities and blind spots in defect detection Among them, the detection of appearance defects such as scratches, oil stains, dirt or hair and other unexpected subtle defects has always been a problem in AOI applications Insurmountable difficulties AIAOI visual inspection overall solution is a great boon for appearance defect detection When designing the product roadmap of Xiaoshi Zhikan, customer group positioning and strengthening customer product services and value were important indicators Moreover, appearance defect detection has always been an unresolved pain in the manufacturing industry, Hong Peijun said With industrial quality inspection AI software as the core, Xiaoshi Intelligent Inspection provides an overall solution for AIAOI visual inspection It mainly promotes three major products, including "QVI-T AI deep learning inspection modeling platform software" and "AI six-sided defect inspection and screening machine" ” and “AI Industrial Quality Inspection Platform” The main customer groups served are semiconductor packaging and testing, EMS electronics foundry, small metal parts processing and other industries with high production capacity and high gross profit margin In response to customer needs, Xiaoshi Intelligent Inspection provides corresponding software and hardware services, combining self-developed AI deep learning software and hardware quality inspection equipment to reduce the manual visual burden on the production line and effectively improve the production quality of the factory In order to help equipment manufacturers and technical engineers with development capabilities accurately grasp product appearance defect detection, Xiaoshi Intelligent Inspection independently developed QVI-T deep learning detection software, which can provide customers with defect location, defect classification, defect segmentation, anomaly detection and text recognition Key functions such as this are different from the fixed detection methods of traditional software Algorithms can be refined based on different industrial detection methods and different APIs can be developed to connect devices with different lenses The software design of this platform is very lightweight It is a SaaS software built on public cloudprivate cloud It mainly involves simple image uploading, labeling, training modeling, and verification testing After completion, users can download models, SDKs, APIs, and reports Effectively help customers achieve AI inference functions Currently, most of the industrial inspection services on the market are traditional AOI software industrial inspection machines, which can only measure product contours such as the head and length of fasteners, etc, and cannot truly provide detection of subtle product surface defects such as screw head cracks and tooth damage There is a lack of such high-precision defect detection companies in the market, Hong Peijun observed Xiaoshi Intelligent Inspection developed and independently built the "AI six-sided defect detection and screening machine" from customized services in the past to providing standardized services for customers at the current stage It provides standardized testing services for fasteners in measurement and surface defects, as well as passive components High-speed surface defect detection of similar products This professional machine uses the AI deep learning AOI composite algorithm technology independently developed by Xiaoshi Intelligent Inspection Through parallel computing technology, it can achieve model inference up to 3 milliseconds per picture, and realize multiple complex defect detection on the electrodes and body of passive components This professional machine is mainly used for the inspection of fasteners, small metal parts and passive components In terms of competitiveness in the industry, the software hardware integration provided by the AI six-sided defect inspection and screening professional machine is an important core competitive advantage of Xiaoshi Intelligent Inspection It is not as simple as it sounds Hong Peijun said with emotion that this special machine is very important in the industrial inspection industry Commonly known as the highly integrated integration of optical mechanisms, electronic controls, software and algorithms, the process requires continuous optimization and iteration, and requires multiple client verifications and modifications After a long period of hard work, the technical threshold has also been raised The AI six-sided defect detection and screening professional machine will be the main product promotion direction of Xiaoshi Intelligent Inspection in the next 3-5 years It is believed that AI combined with measurement technology and surface defect detection will be an important source of core competitiveness of Xiaoshi Intelligent Inspection, Hong Peijun said AI six-sided defect detection and screening professional machine will be the main product promotion direction of Xiaoshi Intelligent Inspection in the next 3-5 years Faced with the booming development of Industry 40 in smart factories, customers often ask "Does quality inspection data have secondary use value" Hong Peijun said that the "AI Industrial Quality Inspection Platform" launched by Xiaoshi Intelligent Inspection has a machine learning mechanism , which can be used for secondary use of quality inspection data to provide customers with multiple functions including real-time monitoring and early warning of production quality, quality traceability analysis, quality factor assessment, process parameter prediction and recommendation Taking the successful introduction into the automotive parts factory as an example, through the prediction and recommendation of process parameters provided by the AI industrial quality inspection platform, when we know the product defects, we build a set of models based on the experience of past masters, coupled with the network connection data from the previous stage, After integration, we have process data, incoming material data, and quality inspection data We can predict whether these machine parameters have run out, and we can recommend whether the process parameters of certain sections should be adjusted up or down Through the AI industrial quality inspection platform, Xiaoshi Intelligent Inspection can help customers connect visual quality inspection results, process data and acceptance standards with the existing MES system of the customer's factory to improve production quality, improve efficiency and reduce costs In terms of business model, Xiaoshi Zhiqian also provides a software subscription system for the deep learning detection modeling platform software It provides public cloud customers with traffic subscription and charges based on the amount of image uploads, while private cloud customers adopt an annual license fee license charging mechanism In addition, the company also provides customers with a buyout charging mechanism for the overall solution equipment, and provides a one-year warranty, after which consumables and software update maintenance fees are charged annually Going in the opposite direction, using both hard and soft methods, with a missed detection rate of less than 1 and rapid modeling in 15 minutes Faced with various small-volume and multi-sample inspection needs in the manufacturing industry, general AI deep learning visual inspection usually requires customers to collect a large number of photos of defective products, which is time-consuming to label, and also causes customers to have difficulty in importing AI, and defective products cannot be collected The introduction cycle is long and implementation is full of risks If there are not enough bad samples, the model will be inaccurate Kosaki Chikan goes in the opposite direction and uses its product "AI Visual Inspection Model Development Tool" to train models through pictures of good products provided by customers It is relatively easy for AI to learn good products, no labeling is required, and the time can be quickly compressed to complete the modeling Take the implementation of IPC electronics industry - AAEON Technology as an example In order to reduce the manpower input of the quality inspection station in the PCBA production line and have standardized quality inspection, Xiaoshi Intelligent Inspection provides an overall solution for PCBA AI visual inspection software and hardware services, and conduct in-line inspection on the factory's highly automated assembly line, effectively saving inspection manpower investment, improving the standardization of quality inspection rates, and improving the problem of inconsistent standards caused by manual visual inspection Through the introduction of AI visual inspection software and hardware integrated solutions, we have effectively helped customers maintain an overkill rate of less than 3 in the past two years, and achieved high-efficiency performance with a missed detection rate of less than 1 In addition, this solution allows practitioners who do not understand AI to quickly operate modeling By installing the modeling tool on the device, when the customer has a new product number and needs to create a model, he only needs to provide 10 pictures of good products to scan under the device It only takes 15 minutes to quickly train the model In terms of product core strategic layout, compared with market competitors who rely solely on general software services to seize all manufacturing markets, it is not feasible to apply it to industrial inspection Hong Peijun has observed over the past 10 years and believes that only software hardware can With technical thresholds and focusing on one industry and field, only by adopting a standardized company's AI six-sided defect detection and screening special machine can it be replicated and scaled up, and the company can truly continue to move towards optimization and create product competitiveness, even if there are other competing products It’s not easy to compete for this pie, Hong Peijun said Xiaoshi Intelligent Inspection’s overall AIAOI visual inspection solution creates rapid modeling and excellent results for customers with a missed detection rate of less than 1 The most competitive AIAOI overall solution provider with global presence For new entrepreneurs, facing business expansion is a challenge every day Hong Peijun said that small companies are easily snatched away by large companies, company talents are poached by high salaries, lack of deep customer relationships, and the business team is not large enough, etc How to overcome this Hong Peijun believes that the key to success and competitiveness of a new start-up company is to be diligent in making up for mistakes, provide better services, provide more immediate feedback, and create more professional solutions to convince customers Since its establishment in 2020, Xiaoshi Intelligent Inspection has always gone against the grain in terms of product core strategic layout, surpassing the competitive market among its peers, and actively taking root in the overall solution of AI visual inspection software and hardware Hong Peijun hopes that Xiaoshi Intelligent Inspection will become the world's most competitive AIAOI overall solution provider for the electronics and semiconductor industries in the future, and provide the top AIAOI professional machines and equipment to the electronics and semiconductor industry customer base Hong Peijun said that the technical capabilities of the company's AI six-sided defect detection and screening professional machine have reached the top domestic level In order to speed up the research and development of professional machines to become more standardized and sell them to overseas markets, the company will conduct a fundraising plan at this stage, hoping to use legal persons such as the Capital Strategy Council to assist in more business connections and fundraising channels For the medium and long-term goals, Xiaoshi Intelligent Inspection will lay out the global market including mainland China and Southeast Asian countries At the same time, it will follow the international footsteps of major OEMs in global layout Under the target inspection project, it will continue to develop specialty products and spread towards the international field 「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2022-09-01
【2021 Solutions】 Exclusive Innovative AI Algorithm Miniaturization Patent: Mantuo Technology Assists in Bringing Large-Scale AI to the Cloud

With the Internet of Things, the demand for AI algorithms is becoming increasingly massive and complex The use of Edge AI, which can effectively reduce costs, enhance security, and improve execution efficiency, will become an inevitable trend Founded in 2018, Mantuo Technology's DeepMentor mainly provides miniaturized AI algorithm SaaS services It is a real solution provider on the market that can deploy AI algorithms running on cloud servers or GPU processors to the Edge Device level, maintaining an accuracy of over 99 while significantly reducing data movement and algorithm complexity Its exclusive and innovative Miniaturized Electronic Design Automation process MAT can shorten time, reduce costs, and still achieve an accuracy higher than 99, key for deploying cloud AI to Edge Devices Ten years of dedication, Mantuo Technology's miniaturization technology gains market attention After more than ten years of research and development in the laboratory, the founder of Mantuo Technology, Wu Xinyi, and his team's exclusive patent of the miniaturized electronic design automation process has resulted in over 20 international A-level papers and several awards including the Best Paper Award Mantuo Technology's technology has been recognized by the market, and in the past two years, it has won the 'Most Investment Worthy Award' at the 8th Yungu Cloud Leopard Incubation Demo Day and the 'Strongest System Innovation Award' with Chunghwa Telecom, as well as the '2021 G Camp International Link Award' and the 'YEZ International Accelerator Special Award' at the 5th 'International Innovation and Entrepreneurship Training Camp G Camp' organized by the Ministry of Economic Affairs in August 2021 滿拓科技獲得第八屆雲谷雲豹育成總決賽「最具投資價值獎」。圖左二為滿拓科技創辦人吳昕益。 AI algorithms are quite large mathematical models Mantuo Technology is focusing on future trends from the era of the Internet of Things IOT to the intelligent Internet of Things AIOT All kinds of terminal products will integrate AI to provide more powerful functions However, many AI algorithms are relatively complex and large, which cannot be integrated into existing terminal devices Therefore, Mantuo Technology developed an exclusive miniaturization technology MAT to optimize the algorithms, simplifying the complex algorithms so they can be incorporated into small embedded systems for easier interfacing with hardware equipment Indeed, what challenges do companies currently face when deploying Edge AI Mantuo Technology's Marketing Director, Yang Yuming, explains that in terms of hardware computation, the cost of GPUs is high, and due to power consumption and heat generation, they are not suitable for edge computing devices In terms of software, there is a lack of good solutions on the market, and most applications can only deploy lightweight simple AI models tiny AI, which are easily affected by climate or external environmental changes, significantly impacting accuracy Importantly, including major international manufacturers in their 2021 white papers, it is mentioned that accuracy can be immediately reduced by 5-15 during the compression process Mantuo Technology offers a complete Edge AI computing solution, with its exclusive and powerful miniaturization technology, which can reduce computational and data volume by 90 while keeping accuracy within 1, successfully deploying multiple complex professional million-parameter AI algorithms to edge devices Mantuo Technology's main customers target system manufacturers and software and hardware developers who want to upgrade IoT to AIoT Mantuo provides the DeepLogMarker software platform to help system vendors and developers quickly obtain miniaturized AI algorithms Through simple training and deployment, they can easily convert commercial IoT devices into AIoT products Yang Yuming stated that the company will launch the SaaS software platform service DeepLogMarker in the fourth quarter of 2021 The first phase will offer the nine most commonly used professional miniaturized AI algorithms, such as object recognition, posture detection, facial recognition, and age and gender algorithms, allowing customers to choose as needed Different AI functions can also be combined according to different usage requirements The platform adopts a subscription model, providing all AI developers, like engineers, startup teams, and device manufacturers, to download and use these algorithms SaaS platform services - simple subscription to use AI algorithms Specifically, customers only need to select the required algorithms on DeepLogMarker, and with a few steps, they can deploy the algorithms to the Edge AI hardware platform Customers do not need to spend a substantial amount of资金capital and time to build an AI environment, and even beginners in AI can directly access all kinds of AI applications needed on the DeepLogMarker platform Mantuo Technology provides professional AI models that have been miniaturized, allowing startup entrepreneurs and developers to subscribe to miniaturized AI algorithms and purchase the Edge AI box to interface with their IoT systems, thereby upgrading the IoT systems to AIoT systems Mantuo Technology is committed to becoming a SaaS software company in the Edge AI field By the end of 2021, it will provide a variety of Edge AI solutions and algorithms on the online platform, hoping to build a complete DeepLogMaker Edge AI user ecosystem Enterprises and cloud service providing platforms like Amazon AWS, Microsoft Azure, IC design companies, startups, and IoT device manufacturers can use the AI services from Mantuo Technology on the platform, to inspire various innovative applications, enhance AI value, and create business opportunities together 滿拓科技建構DeepLogMaker Edge AI使用者生態圈 At the same time, strategic partners with international SaaS operations experience are also welcome to join hands with Mantuo Technology Besides deepening the Taiwanese market, Mantuo Technology's services will also expand into East Asia and international markets such as the United States and Europe The ultimate aim is to significantly reduce the deployment cost of Edge AI, help in the widespread application of AI at a grassroots level, and broaden the applications of Edge AI in intelligent retail, smart manufacturing, smart home appliances, smart medical, and more Mantuo Technology believes that now is the best time for Taiwanese manufacturers to enter the Edge AI market Currently, major chip manufacturers such as NVIDIA, Intel, Qualcomm, NXP, and cloud leaders AWS, Google, Microsoft are all actively investing in this field If Taiwanese manufacturers want to break into the Edge AI market, considering the strengths of Taiwan's small and medium-sized enterprises, industrial manufacturing advantages, and government resources, unique value in software and hardware integration and technology is still the best leverage Finding the right entry point, Mantuo Technology hopes to become the key hub that connects international software giants' resources with Taiwan's hardware advantages Mantuo Technology's miniaturized algorithms enable Edge devices to perform more robust AI functions With the advent of the 5G era, the various AI intelligent application scenarios that everyone is looking forward to can truly materialize The current IoT products need to be fully upgraded 'Solving problems through system integration is solving problems from the root' Mantuo Technology's software and hardware integration solutions, including various AI silicon IP licensing and SaaS services, are expected to officially hit the market from the end of 2021 to early 2022, at which time the development of AI applications in Taiwan may show a different aspect Founder of Mantuo Technology, Wu Xinyi「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2021-08-26
【2022 Solutions】 Complete checkout in 1 second, Viscovery AI image recognition assists smart retail

Artificial intelligence AI has gradually changed the way various industries operate in recent years However, most of the work is still done by humans, with AI playing a supporting role This has led to emergence of the term "AI Copilot," which stands for "AI-driven tools or assistants" that aim to assist users in completing various tasks and improve productivity and efficiency The concept of AI Copilot comes from the role of "co-pilot" During flight, the co-pilot assists the main pilot in completing various tasks to ensure flight safety and efficiency In fact, there have been signs of various "machines" beginning to play the role of "copilot" in different fields since the Industrial Revolution, assisting humans in completing heavy physical and repetitive tasks, greatly improving factory production efficiency, and driving rapid economic development Following the advancement of computing equipment and breakthroughs in machine learning, deep learning, and image recognition technologies, the concept of AI Copilot has gradually taken shape The development of AI Copilot marks the transition from "machine-assisted to AI-assisted" Early robots could only complete preset repetitive tasks, but today's AI copilot can learn and adapt to new environments and tasks, and continuously optimize its performance in practical applications This transformation not only changes human-machine interactions, but also has a profound impact on various industries The application scope of AI copilot covers various industries, including finance, healthcare, manufacturing, education, retail, etc, and are everywhere to be seen Application of AI copilot in the retail industry AI image recognition checkout In the retail industry, the application of AI copilot has begun to show concrete results Take Viscovery's AI image recognition checkout system as an example This system is a type of AI copilot model that helps store clerks speed up checkout or assists consumers in simplifying the self-service checkout process The store clerk needs to scan the product barcodes one by one in the regular checkout method If a product does not have a barcode, such as bread and meals, the clerk needs to first visually confirm the items, and then input them into the POS checkout system one by one Based on actual measurements at a chain bakery, it takes 22 seconds for an experienced clerk from "visual recognition" to "entering product information of a plate of 6 items into the checkout system" New clerks may need even more time In addition, according to a Japanese bakery operator, it takes 1 to 2 months to train employees to become familiar with products Now with AI image recognition technology, store clerks let AI handle the "product recognition" step, and AI will play the role of copilot, quickly identifying items within 1 second, speeding up checkout to save 50 of checkout time, and optimizing customers'shopping experience The time cost of training employees to identify bread can also be effectively shortened Even for products with barcodes, AI can quickly identify multiple items in one second, which is more efficient than scanning barcodes one by one The self-checkout system "assisted" by AI image recognition allows consumers to successfully complete shopping without the help of store clerks, eliminating the trouble of swiping barcodes or searching for items on the screen, which improves the shopping experience In a time when store clerks are hard to hire due to labor shortage, this also helps stores reduce operating costs AI quickly identifies multiple checkout items in just one second Source of image Viscovery Recently, startups dedicated to developing AI image recognition checkout solutions have emerged in various countries The most lightweight solution currently known is in Taiwan It can be immediately used by installing a Viscovery lens and a tablet installed with Viscovery AI image recognition software at the checkout counter to connect to the store's existing POS checkout system There are various integration methods, including plug-and-play and API solutions integrated with the store's POS system Viscovery AI image recognition system can be painlessly integrated with the store's existing POS system Source of image Viscovery Example of AI image recognition checkout Currently, the Viscovery AI image recognition system is being used in bakery chains in Taiwan, Chinese noodle shops in Singapore, micromarkets in department stores in Sendai, Japan, and Japanese bakeries and cake shops Over 7 million transactions were completed through this AI system, which identified more than 40 million items These use cases demonstrate the extensive application of the Viscovery AI image recognition system in the retail industry In the future, the company will continue to explore the various possibilities of using Vision AI in retail and catering nbsp The Viscovery AI image recognition system is already being used in bakeries, cake shops, restaurants, and convenience stores in Japan, Singapore, and Taiwan Source of image Viscovery

2022-04-27
【2021 Solutions】 Motor Protectors: How Haobo Technology Employs AI to Understand the 'Hearts' of Machines

Motors are a critical power source for various devices in modern times, and are key components of many automated equipment, acting like the heart of machines Any malfunctions not only shorten the lifespan of the motors themselves but also affect the entire system's operation, potentially causing delays in production due to downtime Through the use of Smart Vibration Diagnostic and Monitoring Solution SVDM Solution, Haobo Technology utilizes AI to comprehend motors, becoming an indispensable tech partner in intelligent manufacturing The essential industrial motors see an almost 4 annual compound growth rate According to statistics, the market size for industrial motors was 329 billion in 2017, with an expected compound annual growth rate of 363 from 2019 to 2025 Motors have a wide variety of uses, nearly essential in all equipment operations, and represent a technology that is well-developed, with numerous suppliers and a long product lifecycle Haobo Technology, established in 2015, started by solving motor noise issues In recent years, due to the rapidly evolving AI technology, Haobo Technology has continuously innovated and researched, integrating vibration frequency with sound-based audio processing technologies and AI algorithms for noise reduction, along with their self-developed wideband low-noise vibration sensors They introduced a new generation of industry-leading 'Smart Vibration Diagnostic and Monitoring Solution' SVDM Solution, overcoming previous challenges in detecting main vibration frequencies and predicting vibration models in complex environments, making it an essential solution for accelerating industrial intelligence Haobo Technology's 'Smart Vibration Diagnostic and Monitoring Solution' has been successfully applied in PCB drilling machines, semiconductor equipment, machine tools, and medical equipment Haobo Technology's CEO Lu Hongyi states that the Smart Vibration Diagnostic and Monitoring Solution is distinguished not only by its noise reduction capability but also by the AI model's ability to instantly recognize abnormal frequencies in motor vibrations Since general motor sensors are too large to be mounted on the motors, Haobo Technology has customized thin sensors to be installed on the spindle motors, allowing AI models to be computed in real-time on edge devices, significantly reducing the cost of computational resources, with an overall potential reduction in costs by one-fourth Simultaneously, advancements in vibration detection for motor operations have exceeded past technological limits, thus allowing for the immediate detection of the minutest changes and providing early malfunction warnings, helping to extend the lifespan of the motors Haobo Technology has established a vibration data AI analysis platform that constantly monitors the health of motors Haobo Technology's self-developed AI engine is capable of training database modules from collected operational vibration data of motors and establishing a proprietary database for that equipment, which provides real-time comparisons during operation, monitoring if the production equipment is functioning normally Upon detecting any abnormalities, the system instantly issues a warning alert, enabling on-site managers to address the issue immediately to prevent large quantities of finished or semi-finished products from defects Haobo Technology's vibration data AI analysis platform can detect the most subtle vibrations in motors, thus precisely predicting their health status Lu Hongyi mentions that the company independently develops its sensors, hardware, firmware, and AI data analysis platforms, which allows for the detection of the faintest vibrations in motors, thereby accurately predicting their health status This optimization of product quality and preemptive monitoring of production equipment health aims to enhance productivity Haobo's 'Smart Vibration Diagnostic and Monitoring Solution' SVDM Solution has successfully been applied in PCB drilling machines, semiconductor equipment, machine tools, and medical devices, with future plans to actively promote across various industries and enter the international market, hoping to become a pioneer in smart vibration diagnostics Haobo Technology CEO Lu Hongyi「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

2021-10-06

Records of Solutions

這是一張圖片。 This is a picture.
【2020 Solutions】 Robot doctors perform instant diagnosis to maintain product yield and reduce the risk of shut down

Factory machines operate 24 hours a day and will cause heavy losses if they are shut down due to failure If the usage status can be monitored by machine and abnormalities immediately determined, it will help the industry maintain product yields, machine stability, reduce maintenance costs, and shorten maintenance time At present, mainstream AI mostly uses big data gathered in advance or already on hand for training and characterization However, despite the a wide variety of machinery and equipment in factories and the marketrsquos emphasis on predictive maintenance and monitoring in recent years, the amount of data accumulated is very limited How can we quickly meet the needs of factories Provides immediate diagnosis and shortens the troubleshooting process Good Tech Instruments has accumulated a huge database and can quickly help manufacturers protect machinery, equipment, and product quality Wen-Tze Hsu, founder and technical manager of Good Tech Instruments, said that the main feature of machine learning AI monitoring system products is that it quickly learns from the experience of practitioners in the original field, and then uses pre-built mathematical models to classify and manage human experience based on physical and mathematical calculations, accumulating a large amount of data through repetitive production actions At the same time, the previously learned experience data is overlapped with data accumulated over a long period of time and deviation conditions for optimization, in order to achieve the goal of optimizing the monitoring model The time series charts generated by the system are subsequently used to define the predictive maintenance time for various industries and equipment The implementation of instant diagnosis can increase production by 40 working days per year Good Tech Instruments has accumulated a huge database and can quickly help manufacturers protect machinery, equipment, and product quality Wen-Tze Hsu also pointed out that factories have a lot of data, but collecting data does not mean that the data will be useful, and collecting data for deep learning will not necessarily solve the problem of predictive monitoring Good Tech Instrument's monitoring program uses existing or learned rules for monitoring when it is first installed, in order to protect machinery and product quality In the monitoring process, the underlying algorithm will continue to collect and classify data Users can perform labeling actions in the process or use the collected and classified data groups for deep learning AI technology is mainly applied in equipment monitoring and predictive maintenance processes The VMSreg-ML Intelligent Machine Learning Monitoring System can achieve three benefits IAvoid business losses due to production line shutdowns caused by unexpected damage to equipment and molds By preventing unexpected shutdowns and time waiting for materials, production can be increased by 40 working days per year, and quality assurance manpower can be reduced by one third IIOnline monitoring of production processes serves as a defense mechanism against defective products, which can reduce customer complaints and rejected goods IIIMonitor the health status of machinery and equipment, avoid delayed maintenance or excessive maintenance, improve management of industrial production risks, and reduce waste of operating costs The target market of Good Tech Instruments is monitoring applications for stamping, such as the health of motors, crankshafts, punch rods, and springs It can also be applied to smartphone casings, automobile manufacturing body parts, 5G equipment casings, and metal fasteners, and its customer base is mainly in the semiconductor industry Wen-Tze Hsu also said that AI has been the most important trend in the manufacturing industry in recent years, and equipment monitoring is now regarded by most companies as the first step in implementing smart manufacturing systems

這是一張圖片。 This is a picture.
【2020 Solutions】 Apex Intelligence Enhances Facial Recognition Technology by Tenfold Efficiency

Facial recognition software applying deep learning and edge computing model compression technologies significantly reduces costs compared to cloud computing and has increased efficiency by more than ten times The technology has been introduced to law enforcement agencies for suspect identification and will be used in more access control systems, membership systems, and provide diverse AI services in the futureBy integrating edge computing, facial recognition time is saved by more than ten timesAs end-user demand for AI increases, most current AI applications involve sending files to the cloud, where they are processed and analyzed by high-end processors like CPUs and GPUs Future devices such as tablets, smartphones, surveillance cameras, and smart doorbells will feature AI applications, addressing issues of insufficient computing resources and decreased performance with edge computing and efficient deep learning algorithmsApex Intelligence developed AiBo facial recognition software applying deep learning and edge computing model compression technology that allows IPCams to automatically recognize every smile of children aged 0-12 through local device computation, emotion detection accuracy exceeds 98, significantly saving more than ten times the photography and recognition time compared to cloud services Photos can also be uploaded to the cloud to create smart albums using facial classification technology, allowing users to share albums with friends and familyUsing edge computing reduces expenditures by 4 million annuallyWith edge computing, there's no need to spend time sending all images back to the cloud for recognition and classification This saves more storage space and costs, and can expand services to remote areas or locations with unstable internet This technology has already been partnered with major domestic manufacturers, significantly reducing the average annual cloud storage, computer processing, and service costs by 4 million The application also utilizes facial recognition for real authentication, adding an extra layer of protection for online transactions, while physical retail stores can use security system image analysis to quickly identify customers, providing more precise service qualityApex Intelligence's product technology can perform facial detection, identification, classification, object detection, and gaze detection The main focus of its AI deep learning technology is on image processing, with existing products including facial unlock and smart photo albums For example, in collaboration with UMC, it is determined whether personnel wearing safety gear enter within safety cones This technology can be further applied in homes, family restaurants, kindergartens, and more, enhancing applications in access control systems and membership systems to provide diverse AI servicesZhongwei Chen, founder and CEO of Apex Intelligence, also pointed out that images occupy a significant proportion in the data types of the Internet of Things AI can extend the value of images, inspiring different creativity and generating enormous business opportunities in the future「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

這是一張圖片。 This is a picture.
【2020 Solutions】 Using Just One Optical Lens to Render Symptoms Unseen

In ancient times, traditional doctors needed to look, smell, ask, and touch to diagnose patients by checking their pulse, complexion, and symptoms Modern AI doctors, through an optical lens, can scan and detect physiological information such as heart rate, blood pressure, and heart rate variability, making any symptoms apparent The AI doctor 'sees' physiological signals Founded by Distinguished Professor Wu Bingfei from the Department of Electrical Engineering at National Chiao Tung University, the AI startup Julius Innovation primarily utilizes 'AI Image-Based Physiological Information Measurement' It employs ordinary camera lenses to capture continuous facial images for signal processing Through algorithms, it can discern heart rate, heart rate variability, and blood pressure This technology’s main feature is its non-contact measurement As the aging population grows, the demand for long-term care has sharply increased However, for elderly people who dislike wearing wearable devices, even the best wearable devices are of no use Furthermore, the elderly are generally less familiar with 3C products, turning such devices into potential burdens Julius Innovation, however, has addressed this issue with a new solution, using an optical lens in an imaging detection system to monitor facial features and determine measurements like blood pressure and heart rate Compared to various wearable devices, whether worn on the body or hands, the advantage of a camera lens is its simplicity and completely unobtrusive presence Real-time measurement, precise tracking Currently, the market is full of wearable physiological information measurement products, or video surveillance products that lack physiological detection capabilities This solution represents the only successfully commercialized image-based physiological information measurement system The technology highlights include 1 Marketable non-contact, continuous output of heart rate, heart rate variability, and blood pressure detection 2 Fast results in under 6 seconds 3 Measurements can be taken even while wearing glasses 4 Comparable to medical-grade instrument precision 5 Implementation of clinical trials with hospitals to collect actual physiological data and optimize the AI algorithm 6 Uses physiological information measurements for broad applications such as detecting stress, deceit, or fatigue Lateral technology creates diverse vertical applications In medical care, it primarily aids elderly cardiovascular patients by eliminating the discomfort of wearable medical instruments and allowing automatic daily health records and prediction of cardiovascular and other diseases without altering lifestyles This saves medical staff resources and offers more intelligent elderly care solutions In smart finance, facial image processing technology detects physiological information and emotional changes plus masking behaviors By installing this system in bank ATMs or counters, it enhances monitoring device functionality, observes the state of individuals withdrawing money, and issues alerts to bank staff upon detecting unusual emotions or behaviors to counteract financial crime or fraud In transportation, using Julius's technology can detect whether a driver is fatigued If the data shows the driver is fatigued, a warning is issued, advising the driver to cease driving to avoid risks In the financial field, Julius Innovation also collaborates with Shanghai Commercial Bank In newly established smart branches, they have incorporated the image-based physiological information detection system, using physiological data to enhance emotion recognition, strengthen KYC verification at banks, protect against ATM fraud, and provide VIP services, offering a novel digital banking service experience This technology has been implemented in Taichung and Hsinchu digital branches Moreover, the non-contact physiological information measurement applications are extensive, with Julius Innovation focusing on smart care, smart finance, smart transportation, and smart security as its four key sectors Originally, care, transportation, finance, and security were standalone vertical areas, but Julius specializes in lateral technology, allowing more diverse applications across these sectors 「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【解決方案】杭特電子AI創新應用，監控科技全面升級
【2020 Solutions】 Hangte Electronics AI Innovation Applications, Comprehensive Upgrade of Monitoring Technology

Hangte developed AI Cam, offering smart innovative services The development of monitoring products has undergone breakthrough development due to the advent of artificial intelligence technology traditional network cameras IP Cam have been upgraded to AI network cameras with smart image automatic analysis and response capabilities AI Cam not only meets the high-resolution demands of various platforms with image quality but also continuously expands its capabilities based on its smart application properties, thereby creating innovative services in smart imaging applications that keep evolving and being suitably applied in various environments and scenarios For example, in the use of human detection applications, this smart monitoring feature can separately identify each person in the surveillance video and delineate a specific range as a basis for monitoring decisions, conducting systematic analysis and management users can calculate the dwell time, count people traffic, and also issue immediate alerts when a designated person enters the monitored range as shown in the diagram below Smart monitoring features can individually categorize each person in the surveillance video In the aging society with a shortage of manpower, smart long-term care efficiently enhances care quality According to statistics from the Ministry of Interior, Taiwan has entered the 'aging society' structure Since 1993, the population of elderly people aged 65 and above has steadily increased from 1,491,000 to 3,312,000 by the end of March 2018 according to estimates by the National Development Council, the elderly population is expected to exceed 20 by 2026, making Taiwan parallel with Japan, South Korea, Singapore, and some European countries as a 'super-aged society' In other words, in the future, Taiwan's society may become a structure where one in every five people is elderly, significantly increasing the need for long-term care facilities and retirement centers Given the trend toward an aging population, the manpower resources for caring for the elderly will inevitably diminish over time The solution to the aforementioned shortage of manpower for long-term care needs is to optimize the operation of long-term care institutions using artificial intelligence technology, particularly by utilizing AI network cameras with smart image auto-analysis and instant response capabilities, combined with the Internet of Things IoT automatic connection features to promptly identify problems and handle them effectively, embodying a high-efficiency smart service application In the smart operation system of long-term care institutions, we can set up a people counting function to calculate the daily traffic, using this data as a reference for observing the dynamics of personnel entering and exiting a designated detection area, useful for calculating attendance or foreign visitor flow, and adjusting door threshold times or work schedules accordingly additionally, the electronic fence feature can define and secure specific areas, immediately alerting the management if an unauthorized visitor breaches the area, coupled with fall detection to set up designated areas for the care recipients prone to falls, immediately notifying the caregiver if an incident occurs within the detection range as shown in the diagram below Using AI surveillance cameras in long-term care facilities reduces the workload and enhances care quality Edge computing smart imaging, cloud IoT innovative services The application of artificial intelligence combined with the Internet of Things in monitoring scenarios is not limited to potential suspects or dangerous individuals in retirement and long-term care environments, AI surveillance can also enhance monitoring and integrated management of resources Hangte Electronics, leveraging over thirty years of industry experience and RampD capability in monitoring imaging technology, successfully combines Internet of Things technology IoT and Artificial Intelligence AI, not only making good use of cloud computing to reduce installation and management costs of surveillance environments but also greatly enhancing the overall monitoring effectiveness of security systems by utilizing edge computing to provide image preprocessing capabilities and coupling it with AI smart image analysis technology for various image recognition and instant analysis of monitored subjects' behavior, automatically linking response mechanisms to correspond to triggering events, fully dedicated to crafting innovative service solutions for smart surveillance 20 for clients Fully dedicated to crafting innovative service solutions for smart surveillance 20 for clients 「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

這是一張圖片。 This is a picture.
【2020 Solutions】 Not Just Attractiveness, But Capturing Consumer Attention: Rosetta.ai Personalized Recommendation System

When consumers don't know what to buy or are specifically interested in certain types of products, their first thought might be to shop online Why Because it's fast, convenient, offers 24-hour delivery, and they don't have a preference for specific brands, allowing them to find products from various sellers on a single platform While browsing online, you've surely seen product recommendation features like 'You may also like' or 'Others also viewed' These not only help you find the products you need more quickly but also allow for fast comparison of similar items in terms of specs and price It's a very useful feature for consumers For merchants, this data helps understand customers' habits and preferences 'Recommendation systems' basically ineffective Amazon and the globally known streaming platform Netflix see significant revenue boosts from their recommendation systems Coupled with articles teaching how to increase conversion rates, e-commerce business owners might have quietly thought, 'I've tried using similar recommendation systems, but they haven't shown significant results' Perhaps, aside from focusing on targeted marketing, we should consider each consumer as an individual with unique preferences, behaviors, and needs, creating a personalized shopping experience just for them Traditional methods, like recommending more fans after one buys a fan, no longer meet the current users' needs Modern networks use AI algorithms to dynamically calculate user behavior and update website merchandise, promptly adjusting the recommendations on sites and e-commerce platforms, such as recommended content and the content viewed, to better align with what users are seeking Why not analyze consumer profiles, allowing customers to have their 'personalized website' Data shows that e-commerce worldwide spends a lot on advertising to drive traffic, yet the actual conversion rates average only 2 - 3 Why is this Rosettaai has learned from over five years of e-commerce experience that two main factors affect customer conversion and retention Inability to grasp customer purchase preferences, motivations, and situations, including personal factors E-commerce often involves wearing many hats with insufficient resources and budget to adopt innovative technology Thus, Rosettaai has always focused on two critical business factors 'Customer Acquisition' and 'Customer Retention', emphasizing integration of all touchpoints throughout the consumer's online and offline journey, rather than just being an online recommendation tool Its personalized recommendation system suggests products that best suit each customer's taste, analyzing preferences to help brands more accurately understand their consumer profiles and recommend styles to potential customers Rosettaai also designs recommendation systems to suit different pages and KPI stages based on consumer shopping processes Users can select scenarios based on their website's needs the system also provides real-time feedback and automated engines Now, with over 30 unified API-ready free modules and 14 basic scenario combinations, your e-commerce site will no longer be just another site With AI and deep learning, the concept of homepage will be redefined, tailoring product recommendations and homepage design to meet individual consumer demands From another angle, creating a precision recommendation service for e-commerce, allowing the use of a personalized recommendation system, not only prevents customer churn and increases average order value but also enhances turnover and provides a novel shopping experience for modern consumers, resulting in a mutually beneficial cycle 「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

【解決方案】總能聽到你的聲音Aiello智能語音小管家
【2020 Solutions】 Always Hears Your Voice - Aiello Voice Assistant

The development of AI voice assistants has become increasingly mature, but there is a lack of applications specifically for the hotel industry Therefore, the AI voice assistant developed by Aiello can complete room service, equipment introduction, check-out, and stay extension, making it a considerate AI butler The introduction of AI voice assistants into the hotel industry will help improve the experience of guests Google Assistant, Apple's Siri, and Amazon's Alexa are known as the world's three major voice assistants The applications of voice assistants are becoming growingly diverse, but they are mostly used in handheld devices, personal applications and home applications Some hotel operators are optimistic about demand in the hotel industry, and have introduced AI voice assistants into hotel industry applications, leading to the rise of smart hotels Aiello was founded in October 2018 and launched the Aiello Voice Assistant that supports Chinese and English The Aiello butler, which looks like an alarm clock, can assist with room service, equipment introduction, check-out, and stay extension after receiving voice commands from hotel guests Hotel voice assistant For example, if a hotel guest needs drinking water or newspapers, he only needs to say to the Aiello Voice Assistant "Please send a bottle of mineral water to my room" or "Please send a newspaper to my room" After the Aiello butler receives the guest's voice command, the customer service manager will directly send someone to deliver mineral water or newspaper This prevents poor service or even customer complaints because counter staff missed a call In addition to receiving instructions from guests, Aiello Voice Assistant can also combine the functions of a phone and an alarm clock, and has a built-in Bluetooth speaker, so that it can automatically play online streaming music with a single command, providing guests with a better service experience Deepening service scenarios and expanding practical applications Aiello Voice Assistant uses natural language processing NLP and semantic analysis to recognize customers' voice commands and complete multiple room services, reducing unnecessary communication time and potential human error Among them, semantic analysis is used to replace the rule-based model and the exhaustive method of linguistics It can understand at least three questions at a time, which is more in line with people's conversation habits and more adaptable to different situations, thereby deepening the service scenario Hsien-Hsien Liao, co-founder of Aiello, also believes that deepening the service scenario is the only way for consumer experience reach a certain level BERT proposed by Google also uses image analysis to replace the rule-based model and the exhaustive method of linguistics This technology has gradually become mainstream in the market The introduction of AI voice assistants into the hotel industry to provide services will effectively replace manpower to optimize and improve the quality and efficiency of room service The current version of Aiello is already online and will initially target the Chinese market The company is currently negotiating with hotel groups and system service providers in first-tier cities, and has two models monthly rental and buyout In the future, more services will be integrated to expand the business model, including calling taxis, purchasing tickets for entertainment, and booking travel packages In the future, the company will further design a new type of in-room service system specifically for hotels It transforms the in-room voice portal and the hotel's SaaS platform into a new retail channel to maximize benefits for hotel operators The current challenge is that it is difficult to find news and music content providers in Taiwan who are willing to cooperate in providing services However, as the system becomes more mature and the products continue to be promoted towards the hotel industry, more content manufacturers will be attracted to work with the company

RelaJet 洞見未來協助聲障者輕鬆「聽說」
【2020 Solutions】 RelaJet Envisions the Future to Aid Hearing-Impaired with Easy 'Listen & Speak'

For the hearing-impaired, being able to enjoy a low-noise, good listening experience at a relatively affordable price is perhaps the greatest happiness in life Chen Bai-ru, the founder of Future Insight Tech and a hearing-impaired individual, 'heard' the voices of the hearing-impaired Understanding the struggles of not being able to hear clearly, he utilized AI technology to develop a 'Multi-person Voice Separation Engine' to help solve their challenges The Opportunity to Hear Again According to statistics, there are about 99,535 people with hearing disabilities in Taiwan, accounting for about 10 of those with disabilities These nearly 100,000 hearing-impaired individuals live in a world where 'listening and speaking freely' is challenging For the hearing-impaired, the two main problems faced are One, hearing aids are expensive among the six major global brands, the average price is around NT60,000, with high-end models even reaching NT150,000, which is not feasible for an average middle-class family Two, the performance of traditional hearing aids is not sufficient when the surrounding environment is too noisy or the volume too high, it becomes very difficult to clearly hear the speaker's voice Due to their size, conventional noise cancellation methods involving several microphones cannot be used in typical Bluetooth headsets In noisy environments with many people talking, such as restaurants, gyms, and supermarkets, the quality of noise reduction during calls is not ideal Therefore, Chen Bai-ru, also a hearing-impaired individual and the founder of Future Insight Tech, uses AI deep learning technology to achieve noise removal and output clean human voices with just a single microphone Feature Recognition Completed in 10 Milliseconds with No Latency in Speaking and Hearing Being able to complete all feature recognition calculations within 10 milliseconds is the greatest advantage of the RelaJet Multi-person Voice Separation Engine Why 10 milliseconds Because if the processing time for speech by hearing aids exceeds this limit, it can lead to a delay that causes dizziness in the individual Therefore, any hearing aids classified as medical devices are required to complete all processing steps within 10 milliseconds The US Food and Drug Administration FDA is set to allow the sale of non-prescription Over-the-Counter, OTC hearing aids in 2020, which will significantly reduce the costs of experimentation and certification and make hearing aids more affordable Additionally, the purchasing channels will be more open, eliminating the cumbersome fitting process Future Insight is seizing this business opportunity, actively establishing partnerships with the top six global hearing aid brands, and also entering the Bluetooth headset market to benefit more hearing-impaired individuals Future Insight Tech's specific approach to incorporating AI involves a customized model with an algorithm chip, achieving a noise reduction of 20Db and power consumption below 9Ma This noise reduction model requires only a single microphone to remove noise and output clean human voices, significantly enhancing the call quality of Bluetooth headsets「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

這是一張圖片。 This is a picture.
【2020 Solutions】 Changing Your Perception of Chatbots: Engaging in Meaningful Conversations

The influence of Artificial Intelligence AI on human life is growing daily, whether for businesses, factories, or individuals, everyone is starting to long for a more convenient lifestyle through technology Whether it's the virtual characters in movies or figments of our imagination, they reflect our desires for a futuristic world an entity that can instantly respond to all your needs and handle everything meticulously for you Consequently, Apple brought us 'Siri', Amazon presented 'Alexa', and various tech giants have followed with their intelligent assistantsTalk to anything you wantAccording to a Gartner survey, conversational AI's Natural Language Processing NLP technology is now among the top three AI technologies How NLP is applied to enhance consumer experience is a domain actively being studied and progressed by all conversational AI providers Despite human language being ambiguous and unstructured for machines, with the advent of NLP, we can parse patterns within these large unstructured datasets, enabling better understanding of the embedded messages NLP also aids in addressing business challenges, especially for frequently asked predictable questions or routine continuous work, which are increasingly being managed by AI-based Chatbots due to advancements in machine learning and computational power However, for Chatbot developers, finding the right applications is just the first step designing an engaging experience is crucial for retaining users Generally, the impression of Chatbots remains stuck on the stereotype of customer service, often appearing clueless or only capable of responding with a few programmed answers, which makes them seem unintelligent and disappointing Although many factors influencing the user experience design of chatbots warrant further investigation, enhancing consumer experience and altering perceptions require more advanced Natural Language Understanding NLU technologies for semantic analysis, sentiment analysis, and advanced conversational applications, making Chatbots smarter and more attuned to human preferences Asia Pacific Intelligence is one of the few companies in Taiwan focusing on machine intelligence, dedicating themselves to improving human life through proprietary development of NLU technology and specializing in Chatbot applications via the Opentalk platform Established less than three years ago, it is already the only technological partner in Taiwan for the global top-five industrial AI company 'iFLYTEK'Asia Pacific Intelligence constructs 'Multi-Turn Dialogue Querying Capability', enhancing the understanding abilities of ChatbotsThrough its internally developed Natural Language Understanding technology, Asia Pacific Intelligence has developed a rapid deployment platform for conversational robots, now equipped with multi-turn dialogue querying capabilities By integrating a domain knowledge graph, these customer service bots can resolve 70-80 of issues at the first point of contact More complex and diverse queries still rely on human customer service However, the 80 accuracy rate is sufficient for customer service staff to handle more complex customer demands Additionally, multi-turn dialogue robots can also be applied in factory settings When customers encounter issues with machinery, they can describe the problem and the machine's condition to the robot through multi-turn dialogues, the robot can determine the correct problem within a limited scope, effectively pinpoint the issue using the knowledge graph, and notify technical staff for repairs The machine's preliminary judgment and reporting can generally resolve issues within a day The underlying data structure of multi-turn dialogue Chatbots relies on interfacing with corporate websites or databases, categorizing data like user preferences, user question and answer data, and user personas However, apart from requiring a large amount of data, the most crucial aspect is the continuous 'feeding' of domain knowledge to make robots increasingly smarter Additionally, Bo-Han Wu, the founder of Asia Pacific Intelligence, believes 'AI learning should be data-driven thus, dealing with extensive dialogue content cannot be managed with small data Although various inference technologies are rapidly developing, ultimately, it is still humans who make decisions based on understanding'Speaking Out Infinite Possibilities for the FutureVoice technology plays a crucial role in penetrating daily life For example, Google Assistant announced the launch of its Traditional Chinese version in 2018, actively expanding its voice market in Taiwan Developers can now upload their voice skills on Action on Google for use by others Asia Pacific Intelligence APMIC, leveraging its semantic understanding technology, has listed a voice skill named 'Bus Helper' where users can activate it by saying 'I want to talk to Bus Helper' to their phone or smart speaker This voice-enabled service can save the hassle of opening an app and typing during the busy morning rush, simply by using voice instead of hands to check bus statuses The development of voice capabilities requires advanced NLU technology to accurately determine user intent Major firms like Google, Amazon, and Microsoft are also actively participating in NLU technology research, suggesting that the future of voice skill applications will introduce more eye-catching features, providing more thoughtful and intelligent user experiences, making life more convenient「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

「知識圖譜」打造衛教小天使，回答孕期遇到的大小問題
【2020 Solutions】 'Knowledge Graphs' create health education angels, answering all questions faced during pregnancy

Artificial Intelligence AI is evolving rapidly Traditional scripted response smart customer services are no longer sufficient Emerging from this is the use of Knowledge Graphs in smart assistants Through knowledge graphs, these smart assistants can provide exhaustive, tailored customer service, helping consumers efficiently and effortlessly find their desired products The most common AI applications in daily life are smart customer services or smart assistants For smart customer services, the aim is mainly to address customer inquiries, but they often face the issue of providing too generic and rigid responses, lacking flexibility Utilizing knowledge graphs to establish smart assistants To develop a smooth-questioning smart assistant, it's crucial to optimize technologies such as natural language processing, semantic analysis engines, multi-turn dialogues, context awareness, emotion recognition, and personalization Supported by the Economic Ministry's Technology Department, the Institute for Information Industry’s Service Innovation Research Institute has developed tools for domain knowledge construction and management These tools assist in transforming corporate websites, databases, and documents into corporate knowledge graphs using AI recognition models When used in smart customer service, if the customer query is vague, it can proactively ask follow-up questions based on the knowledge graph and provide precise answers through multiple dialogue rounds, saving on professional labor costs and response time, and delivering better consumer service experiences Chen Bing-Yi, head of the Institute, points out that simple smart assistants operate by connecting to fixed APIs and responding based on predefined data formats, which is straightforward for queries like tracking shipments through Line or Facebook Messenger However, for more flexible customer issues such as cancer insurance clauses, credit card application terms, or legal advice, more complex technologies are required Technically, AI must convert unstructured data from documents, websites, or even social media into patterns that computers can understand, ie, establishing corporate knowledge graphs In simple terms, knowledge graphs are primarily built by extracting all significant texts and marking the relationships between them, automatically creating relational graphs among all individuals, events, and objects For example, the computer can extract and determine the relation between Nantou and Oolong Tea as 'Origin' from social data, and link 50 Lan's products to Oolong Tea as 'Product' Once a consumer wishes to know 'Where does 50 Lan’s Oolong tea come from', the system can infer using the knowledge graph and provide 'Nantou' as the response, making customer service smarter and more professional Additionally, smart assistants no longer passively receive queries they can actively notify and remind consumers, for instance, notifying pregnant women of important prenatal tests they need to undergo through Line and Facebook Messenger, ensuring no tests are missed and enhancing safety The smart assistants developed by the Institute are being progressively applied across retail, finance, and medical sectors including applications such as querying credit card benefits, product specifications, and medical education advice, with several companies already having successfully implemented these practices Plans are in place to integrate external social data or open up knowledge graphs for future development, continuing to collaborate with partners to advance the application of knowledge graphs in various new forms of AI services Applications of Knowledge Graphs Health Education Angel Chatbot New mothers often face a multitude of questions during pregnancy, such as morning sickness, diet, fetal health, and prenatal check-up considerations, affecting their sleep, personal health, and potentially impacting fetal health The Service Innovation Research Institute, in collaboration with Hua Xin Health Technology, uses AI technology to process health education manuals from the Ministry of Health and Welfare and community discussion articles This initiative aims to build a Health Education Angel using natural language processing and knowledge graph technologies specifically for addressing various issues encountered during pregnancy The institute and Hua Xin Health Technology have collaborated to develop the Health Education Angel chatbot to answer various questions during pregnancy If there are questions related to prenatal examinations, they can be inquired through 'Prenatal Examination ABC', where the system will provide information about whether certain prenatal tests are chargeable and the diseases they can screen for If there are concerns regarding dietary and health status during pregnancy, 'Pregnancy Encyclopedia' can be utilized By posing questions in natural language, the system will offer the most relevant expert answers Additionally, information about valuable pregnancy resources and welfare assistance is available for pregnant mothers to consult The Health Education Angel can pose questions in natural language, and the system will provide the most relevant expert responses「Translated content is generated by ChatGPT and is for reference only Translation date：2024-05-19」

rows
Rows：119, 14 pages