AI-Powered Emotion Recognition System for Customer Satisfaction / Campaign Measurement and Analysis in Telecommunication Retailers

Abstract

The invention relates to a system used in telecommunications, physical retail network management, data-centric regional marketing strategy, pricing optimization, and customer behavior simulation that models and simulates a telecommunications operator's physical retail network and the recorded customer behavior associated with that network in a completely synthetic digital environment, without any real risk to personal data.

Description

1 TARIFF Customer Emotion Recognition with AI Powered in Telecommunications Retailers Satisfaction / Campaign Measurement and Analysis System Technical Area 5 The invention has implications for telecommunications retail, customer experience management, and anonymized sentiment analytics. Used in the fields of AI-powered facial expression recognition and dealer performance management, Customer service at telecommunication retailers using AI-powered emotion recognition. It relates to the satisfaction / campaign measurement and analysis system. State of the Art Measuring customer satisfaction at telecommunications retailers is largely the norm today. It is based on surveys. These methods and current technologies have serious shortcomings. The shortcomings and technical problems that arose during this process are listed below:  Low Participation and Unreliable Surveys: The vast majority of customers did not participate in the survey. 15 They don't fill it out. Those who do fill it out are usually either very satisfied or very dissatisfied. These are the results. This leads to an unhealthy sample size and an inaccurate representation of the true level of satisfaction. This leads to inaccurate measurement.  Delayed and Reactive Intervention: Survey results are usually available weeks or months later. It becomes analyzable. A satisfaction problem at a dealership can only be resolved through a large number of complaints. It becomes noticeable when it accumulates, and intervention is delayed.  Face Recognition and Privacy Breach Risk: Traditional camera systems for face recognition When used in this way, it violates customer privacy and goes against laws such as KVKK / GDPR. They are incompatible. Therefore, they cannot be widely used in retail.  Lack of Analysis at Micro (Dealer) and Macro (Nation) Scale: Existing systems, a single 25 from measuring customer satisfaction at the dealership, to analyzing this data at the provincial, district, regional, and national levels. It lacks the ability to aggregate and transform into meaningful insights.  Subjective Evaluation of Campaign Effectiveness: Campaigns conducted at dealerships Success is often measured solely by sales figures. The campaign's customer... Its impact on customer satisfaction and brand perception cannot be objectively assessed. 30  Lack of Objective Feedback for Staff Motivation: Dealer staff, customer They cannot receive immediate and objective feedback about the impact it has on them. Performance Evaluations are generally based on management observations or sales figures. In conclusion, due to the negative aspects described above and the current solutions being the subject of discussion... Due to its inadequacy, it has become necessary to make improvements in the relevant technical field. 35 2 Purpose of the Invention The invention was created by drawing inspiration from existing situations and overcoming the aforementioned drawbacks. It aims to solve the problem. The main purpose of the invention is to provide AI-powered emotion recognition in telecommunication stores. The goal is to provide a customer satisfaction / campaign measurement and analysis system. Another goal of the invention is to enable privacy-focused Edge AI cameras to perform anonymous operations based on deep learning. sentiment analysis, geographic information systems integration, and a centralized decision support platform. The goal is to provide a comprehensive system that uses 10. Another purpose of the invention is to combine hardware (special camera) and software (analytical engine, control panel) The goal is to provide an end-to-end platform that works together; traditional surveys have low participation rates and By eliminating reliability issues, anonymous data is obtained from every customer interaction. To provide an objective and continuous measurement of satisfaction using sentiment data; across 15 provinces in Türkiye. Dynamic heat maps showing customer satisfaction levels at the district and even neighborhood level. The goal is to visualize regional challenges and successes by creating monthly / quarterly / annual reports for each dealer. By calculating an "Emotion-Based Performance Score," we provide sales and audit teams with an objective assessment. The aim is to provide evaluation criteria; before, during and after the campaigns implemented at dealerships. By analyzing the changes in customer sentiment scores, the campaigns target customer 20 The goal is to measure the real impact on customer satisfaction; when a customer smiles, the dealer staff's reaction... By displaying instant positive feedback and ratings on the screen, it can increase employee motivation and The goal is to reinforce customer-focused behavior; a sudden and sustained increase in the customer satisfaction index at a dealership. When a drop in performance is detected, the system automatically sends alerts to the central monitoring teams, reporting problems. The goal is to intervene in growth; AI 25, built-in and distributed to dealers. By using specialized cameras with processors, we never send the raw footage to the center, only... Anonymized emotion metadata (a human face was detected, emotion category, trust) The goal is to generate a score (timestamp); a deep learning-based system that runs inside the camera. Using a model that employs convolutional neural networks (CNNs), we can detect faces in incoming images and The aim is to classify into one of seven basic emotion categories; the camera, 30 in a customer When the word "Happy" is detected, this information is immediately sent to the tablet used by the dealer staff or The goal is to transmit anonymous sentiment data from all dealers as a notification on the screen; dealers' geographical coordinates (latitude, longitude) and administrative hierarchy information (province, district, region) The goal is to collect data by labeling and gathering it in a central cloud database; from the entire dealer network. It provides a web-based management panel that visualizes and analyzes data; 35 in a dealer when a sudden and statistically significant decrease in happiness score is detected (e.g., 7 3 (daily moving average falling below 2 standard deviations), central bank, sales and control. To provide an algorithm that sends automated email / SMS alerts to teams; to a specific dealer When a campaign is launched, the system calculates sentiment scores before, during, and after the campaign. The goal is to calculate the "Emotion-Based ROI" of the campaign by comparing the results; the calculated Dealer performance scores and sentiment data are automatically processed by the Sales Facility Management System (SFA 5). and to ensure that audit teams have APIs that will transmit this information to their mobile applications. Explanation of the Figures Figure 1 is a schematic view of the system described in the invention. Explanation of Part References 1. Privacy-Focused Edge AI Camera Hardware 2. Anonymous Emotion Analysis and Facial Expression Recognition Algorithm 3. In-Dealer Real-Time Feedback System 4. Geographic Information System (GIS) Integrated Data Collection and Aggregation Layer 15 5. Central Sentiment Analytics and Decision Support Platform (Management Panel) 6. Anomaly and Trend Detection Module 7. Campaign Performance Analytics Module and API Integration Layer Detailed Description of the Invention 20 In this detailed explanation, the preferred configurations of the invention are not merely for better understanding the subject. This is intended to facilitate understanding and will not impose any limiting effects. The invention, Customer service at telecommunication retailers using AI-powered emotion recognition. It is a satisfaction / campaign measurement and analysis system, physically placed at dealerships, It has an AI processor, never sends the raw image externally, only to anonymous 25 It generates sentiment metadata by performing real-time processing thanks to built-in algorithms. Privacy-focused edge AI camera that reduces central load and guarantees GDPR / DPR compliance. hardware (1), faces with deep learning-based CNN model working inside the camera. without knowing them, simply classifying their expressions (happy, sad, angry, surprised, neutral), for each customer. An anonymous emotion analysis and facial expression recognition algorithm that calculates an anonymous "emotion score" (2), 30 When the camera detects a positive expression (happy / smiling) on ​​a customer, it instantly transmits this information. Instant notifications sent to dealership personnel's tablets / screens for employee motivation. Real-time in-dealer feedback system (3) that provides positive reinforcement from all dealers incoming anonymous sentiment data, dealers' geographical coordinates and administrative hierarchy information (province, district) by labeling and collecting the data in a central database, 35 a unifying geographic information system (GIS) integrated data collection and aggregation layer (4), all 4 It is a web-based management panel that visualizes data from the dealer network, and reflects Turkish sentiment. map, dealer performance scorecard, anomaly detection, and campaign performance analytics Central sentiment analytics and decision support platform (5) which includes modules, dealer-based sentiment performing time series analysis of scores using statistical methods (moving average, standard deviation) (deviation) detects sudden drops or rises, the happiness score at a dealer is at a critical level of 5 When it falls below a certain level, the center automatically sends an anomaly alert to the sales and inspection teams. Trend detection module (6), before, during and the campaigns carried out in dealers subsequently analyzing changes in customer sentiment scores, campaigns targeting customer It objectively measures its impact on satisfaction and calculates the "Emotion-Based ROI". Calculated dealer performance scores, sentiment data and alerts, sales management system 10 (SFA) enables audit teams to automatically use mobile applications and other corporate systems. Campaign performance analytics module and API integration including shared API interface. It contains layer (7). The system works on the principle that eight basic elements, a telecommunications company, thousands of 15 collecting, processing, and transforming anonymous sentiment data from its distributors into meaningful insights. And it is built on enabling action to be taken based on these insights. The process is at each dealership. It starts with the privacy-focused edge AI camera hardware (1) found in the store. These cameras monitor those entering the store. It constantly analyzes customers' facial expressions. Anonymous emotion analysis within the camera and The facial expression recognition algorithm (2) recognizes faces without recognizing expressions (happy / smiling, sad, 20 It classifies emotions as angry / frowning, surprised, neutral and assigns an anonymous "emotion score" to each customer. accounts. When a customer expresses happiness, real-time in-store feedback is provided. The system (3) is activated and a positive notification and score is immediately displayed on the dealer personnel's screen. It sends and motivates staff. All anonymous sentiment data, dealer's geographic location, and Timestamp included, geographic information system (GIS) integrated data collection and aggregation 25 The data is transmitted to the central cloud platform via layer (4). Here the data is processed and centralized. It is visualized by sentiment analytics and decision support platform (5). In this panel, province and district An updated sentiment heat map of Türkiye, based on data, allows all dealers to assess "Sensitive-Based Performance". It includes a scoreboard showing the "score" and campaign analyses. Anomaly and trend detection. module (6) continuously analyzes sentiment scores on a dealer basis. A dealer has a happiness score of 30 When a sudden and sustained decline is detected, the center automatically sends alerts to sales and inspection teams. sends. Campaign performance analytics module and API integration layer (7), to dealers It measures the impact of the implemented campaigns on customer sentiment. Finally, all of this... Data and alerts are transmitted to sales management system and audit teams' mobile devices via the relevant API. This information is automatically transmitted to their applications, allowing field teams to take proactive action. 35

Claims

REQUESTS 1. Customers at telecommunication retailers using AI-powered emotion recognition. It is a satisfaction / campaign measurement and analysis system, and its features include:  Physically placed in dealer locations, equipped with an AI processor, that captures the raw image never sends anything out, only produces anonymous emotion metadata, built-in 5 By performing real-time processing thanks to algorithms, it reduces the workload on the processing center and Privacy-focused edge AI camera hardware that guarantees KVKK / GDPR compliance (1),  Face recognition using a deep learning-based CNN model that operates within the camera, It only categorizes expressions (happy, sad, angry, surprised, neutral), for each customer. Anonymous emotion analysis and facial expression recognition that calculates an anonymous "emotion score" 10 algorithm (2),  When the camera detects a positive expression (happy / smiling) on ​​a customer, it transmits this information. instantly transmitting a notification to the dealer's staff member's tablet / screen. Real-time in-dealer feedback that provides immediate positive reinforcement for motivation. system (3), 15  Anonymous sentiment data from all dealers, dealers' geographical coordinates and administrative data It collects data in a central database by labeling it with hierarchical information (province, district), Geographic information system (GIS) integrated data that aggregates data based on time and space. collection and aggregation layer (4),  It is a web-based management panel that visualizes data from the entire dealer network, 20 Türkiye sentiment heat map, dealer performance scorecard, anomaly detection and campaign. Centralized sentiment analytics and decision support including performance analytics modules platform (5),  Performing time series analysis of sentiment scores on a dealer basis using statistical methods (Moving average, standard deviation) is a 25-bit indicator that detects sudden drops or rises. When the satisfaction score at the dealership falls below a critical level, headquarters, sales and control departments are alerted. Anomaly and trend detection module that sends automatic alerts to its teams (6),  Customer sentiment before, during, and after campaigns conducted at dealerships analyzing changes in scores and the impact of campaigns on customer satisfaction 30 dealers that objectively measure the impact and calculate "Emotion-Based ROI". performance scores, sentiment data and alerts, sales management system (SFA), audit teams automatically with mobile applications and other corporate systems Campaign performance analytics module and API including shared API interface. integration layer (7) It includes. 35

Images