Method and system for user emotional matching based on sentiment similarity

Abstract

An emotion similarity-based user emotion matching system may include a memory and a processor for storing instructions. When the instructions are executed by the processor, the system may control the system to collect user-specific emotion record data, the collected emotion record data includes emotion labels, emotion scores, and time information, vectorize emotion profiles based on emotion data for a recent period for each user from the collected emotion record data, generate a target set of users among a plurality of users, calculate cosine similarity between vectorized emotion profiles, extract matching candidates between users if the calculated cosine similarity is above a certain threshold, and propose at least one connection among empathy-based report exchange, character sharing, or anonymous conversation to the user for the extracted matching candidates.

Description

Technology Field

[0001] The present invention relates to an artificial intelligence-based emotion processing system, and more specifically, to a method and system capable of providing empathy-based communication and emotional support by vectorizing user emotion data and automatically matching users exhibiting similar emotional flows or states through similarity analysis. Furthermore, the present invention can be applied to fields such as emotion computing, empathetic social networks, emotion-based recommendation systems, digital well-being, and psychological support platforms. Background Technology

[0003] Existing user matching systems have primarily formed connections between users based on superficial data such as demographic information, interests, and geographical location. This traditional matching method may have limitations in that it fails to adequately consider users' inner states or emotional needs. Since connections in social networks are mainly centered around existing relationships or shared interests, there may be constraints on forming connections based on actual emotional resonance or psychological similarity.

[0004] While current emotion recognition technology focuses on identifying the emotional state of individual users, systems that systematically utilize this for building emotional connections or empathy networks with other users are not yet sufficiently developed. In particular, technical approaches that vectorize emotion data and apply mathematical models such as cosine similarity to identify and connect users with similar emotional patterns have been implemented only to a limited extent.

[0005] Furthermore, existing systems tend to focus only on the transient state of emotions, resulting in a lack of matching mechanisms that consider patterns of emotional change over time or similarities in emotional rhythms.

[0006] Furthermore, the absence of safe, empathy-based communication channels capable of forming meaningful emotional connections while maintaining anonymity and privacy is cited as a factor exacerbating the emotional isolation and disconnection of modern people. Consequently, there is a growing need for technological solutions that enable meaningful connections based on emotional similarity while protecting user privacy. The problem to be solved

[0008] The emotion similarity-based user emotion matching method and system according to the present invention aims to provide a structure capable of automatically identifying and connecting users exhibiting similar emotional patterns or states by systematically collecting and vectorizing user emotion data. Through this, a network capable of communication and support based on emotional empathy can be established.

[0009] The present invention aims to provide a technical method capable of effectively mapping multidimensional emotional data, such as emotion scores, emotion types, and temporal change patterns of emotions, into a vector space, and quantitatively measuring emotional similarity between users by utilizing a cosine similarity-based algorithm. This enables more meaningful connections based on internal emotional states rather than superficial characteristics.

[0010] The present invention aims to implement a secure communication channel that enables anonymous, empathy-based conversations or the exchange of emotional support while maintaining privacy and data security. Through this, users can form emotional connections without concerns about the exposure of personal information, while sharing their vulnerable emotional states.

[0011] The present invention aims to form a more stable and sustainable emotional network based on the commonality of continuous emotional patterns, rather than temporary emotional states, by constructing a long-term emotional matching system based on the similarity of emotional rhythms over time. Furthermore, the present invention aims to provide an scalable platform that supports various forms of emotional interaction, such as sharing emotion reports, exchanging emotional characters, and recommending personalized content. Through this, it is possible to enable various forms of empathy-based communication tailored to the emotional needs and preferences of users. means of solving the problem

[0013] The emotion similarity-based user emotion matching system according to the present document may include a memory and a processor for storing instructions. When the instructions are executed by the processor, the system may control the system to collect user-specific emotion record data, the collected emotion record data includes emotion labels, emotion scores, and time information, vectorize emotion profiles based on emotion data for a recent period for each user from the collected emotion record data, generate a target user set among a plurality of users, calculate cosine similarity between vectorized emotion profiles, extract matching candidates between users when the calculated cosine similarity is above a certain threshold, and propose at least one connection among empathy-based report exchange, character sharing, or anonymous conversation to the user for the extracted matching candidates. Effects of the invention

[0015] The emotion similarity-based user emotion matching method and system according to this document can form connections based on the similarity of inner emotional states and patterns, rather than superficial characteristics, through the vectorization of user emotion data and a cosine similarity-based matching algorithm. This has the effect of enabling meaningful interactions between users based on a deeper level of empathy and understanding.

[0016] The emotion similarity-based user emotion matching method and system according to this document can accurately capture subtle differences and similarities in emotion patterns by utilizing advanced machine learning techniques in the process of converting emotion data into multidimensional vectors and calculating similarity.

[0017] The emotion similarity-based user emotion matching method and system according to this document can provide a safe communication environment capable of emotional empathy and support while maintaining anonymity and data security. This provides users with a sense of psychological safety that allows them to freely express and share their vulnerable emotional states, which can have the effect of improving the quality of emotional communication in a digital environment.

[0018] The emotion similarity-based user emotion matching method and system according to this document can establish a connection based on commonalities of long-term emotional tendencies rather than temporary emotional states by performing matching based on the similarity of emotional rhythms and patterns over time. Brief explanation of the drawing

[0020] FIG. 1 is a diagram illustrating the overall structure of an artificial intelligence-based system according to one embodiment. FIG. 2 is a diagram illustrating the learning of a neural network according to one embodiment. FIG. 3 is a diagram illustrating the configuration of an artificial intelligence model according to one embodiment. Figure 4 is a block diagram showing the configuration of an emotion similarity-based user emotion matching system according to one embodiment. Figure 5 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment. Figure 6 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment. Figure 7 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment. Specific details for implementing the invention

[0021] Hereinafter, embodiments are described in detail with reference to the attached drawings. However, various modifications may be made to the embodiments, and thus the scope of the patent application is not limited or restricted by these embodiments. It should be understood that all modifications, equivalents, and substitutions to the embodiments are included within the scope of the rights.

[0022] Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only and may be modified and implemented in various forms. Accordingly, the embodiments are not limited to the specific disclosed forms, and the scope of this specification includes modifications, equivalents, or substitutions that fall within the technical concept.

[0023] Terms such as "first" or "second" may be used to describe various components, but these terms should be interpreted solely for the purpose of distinguishing one component from another. For example, the first component may be named the second component, and similarly, the second component may be named the first component.

[0024] When it is stated that a component is "connected" to another component, it should be understood that it may be directly connected to or joined to that other component, or that there may be other components in between.

[0025] The terms used in the embodiments are for illustrative purposes only and should not be interpreted as intended to be limiting. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as "comprising" or "having" are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0026] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the embodiments pertain. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0027] In addition, when describing with reference to the attached drawings, identical components are assigned the same reference numeral regardless of drawing symbols, and redundant descriptions thereof are omitted. In describing the embodiments, if it is determined that a detailed description of related prior art could unnecessarily obscure the essence of the embodiments, such detailed description is omitted.

[0028] The embodiments can be implemented in various forms of products such as personal computers, laptop computers, tablet computers, smartphones, televisions, smart home appliances, intelligent automobiles, kiosks, and wearable devices.

[0029] Artificial Intelligence (AI) systems are computer systems that implement human-level intelligence; unlike existing rule-based smart systems, they are systems in which machines learn and make decisions autonomously. As AI systems improve in recognition accuracy and gain a more accurate understanding of user preferences with continued use, existing rule-based smart systems are gradually being replaced by deep learning-based AI systems.

[0030] Artificial intelligence technology consists of machine learning and component technologies utilizing machine learning. Machine learning is an algorithmic technology that autonomously classifies and learns the characteristics of input data, while component technologies are technologies that mimic the cognitive and judgmental functions of the human brain by utilizing machine learning algorithms such as deep learning, and are comprised of technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control.

[0031] The various fields where artificial intelligence technology is applied are as follows. Linguistic understanding refers to technologies that recognize, apply, and process human language and text, including natural language processing, machine translation, dialogue systems, question answering, and speech recognition / synthesis. Visual understanding refers to technologies that perceive and process objects like human vision, including object recognition, object tracking, image search, people recognition, scene understanding, spatial understanding, and image enhancement. Inference and prediction refers to technologies that logically reason and predict by judging information, including knowledge / probability-based inference, optimization prediction, preference-based planning, and recommendation. Knowledge representation refers to technologies that automatically process human experiential information into knowledge data, including knowledge construction (data generation / classification) and knowledge management (data utilization). Motion control refers to technologies that control the autonomous driving of vehicles and the movement of robots, including motion control (navigation, collision, driving) and manipulation control (behavior control).

[0032] Generally, to apply machine learning algorithms to real-world situations, training is performed using a trial-and-error method due to the inherent characteristics of the fundamental methodologies. In particular, deep learning requires hundreds of thousands of iterations. Since it is impossible to execute this in a real physical external environment, training is instead performed through simulations that virtually recreate the actual physical environment on a computer.

[0033] In the present invention, Artificial Intelligence (AI) refers to a technology that imitates human learning ability, reasoning ability, and perceptual ability, and implements them on a computer, and may include concepts such as machine learning and symbolic logic. Machine Learning (ML) is an algorithmic technology that classifies or learns the characteristics of input data on its own. AI technology can analyze input data as a machine learning algorithm, learn from the results of the analysis, and make judgments or predictions based on the results of the learning. Furthermore, technologies that mimic the functions of the human brain, such as cognition and judgment, by utilizing machine learning algorithms can also be understood as falling within the category of AI. For example, technological fields such as linguistic understanding, visual understanding, reasoning / prediction, knowledge representation, and motion control may be included.

[0034] Machine learning can refer to the process of training neural network models using experience in processing data. Through machine learning, computer software can improve its own data processing capabilities. A neural network model is constructed by modeling the correlations between data, and these correlations can be expressed by multiple parameters. A neural network model extracts and analyzes features from given data to derive correlations between them; machine learning can be defined as the process of optimizing the model's parameters by repeating this process. For example, a neural network model can learn the mapping (correlation) between inputs and outputs for data given as input-output pairs. Alternatively, even when only input data is provided, a neural network model can derive regularities between the given data and learn those relationships.

[0035] An artificial intelligence learning model or neural network model can be designed to implement the structure of the human brain on a computer and may include multiple network nodes that have weights and simulate neurons of a human neural network. The multiple network nodes may have interconnected relationships by simulating the synaptic activity of neurons, where neurons exchange signals through synapses. In an artificial intelligence learning model, multiple network nodes may be located in layers of different depths and exchange data according to convolutional connections. The artificial intelligence learning model may be, for example, an Artificial Neural Network (ANN) or a Convolutional Neural Network (CNN). As an embodiment, the artificial intelligence learning model may be machine learned according to methods such as supervised learning, unsupervised learning, and reinforcement learning. Machine learning algorithms for performing machine learning may include Decision Tree, Bayesian Network, Support Vector Machine, Artificial Neural Network, Ada-boost, Perceptron, Genetic Programming, and Clustering.

[0036] Among these, CNNs are a type of multilayer perceptron designed to use minimal preprocessing. CNNs consist of one or more convolutional layers and standard artificial neural network layers stacked on top, additionally utilizing weights and pooling layers. Thanks to this structure, CNNs can fully utilize two-dimensional input data. Compared to other deep learning architectures, CNNs demonstrate good performance in both image and audio fields. CNNs can also be trained using standard backpropagation. CNNs have the advantage of being easier to train than other feedforward artificial neural network techniques and using a small number of parameters.

[0037] Convolutional networks are neural networks comprising sets of nodes with bounded parameters. Many computer vision tasks have been significantly improved, driven by the increased size of available training data and the availability of computational power, combined with algorithmic advancements such as discriminative linear units and dropout training. In the case of massive datasets, such as those available for many tasks today, outfitting is not critical, and increasing the network size improves test accuracy. Optimal utilization of computing resources becomes a limiting factor. To address this, distributed, scalable implementations of deep neural networks can be employed.

[0039] FIG. 1 is a diagram illustrating the overall structure of an artificial intelligence-based system according to one embodiment.

[0040] As illustrated in FIG. 1, an artificial intelligence-based system (400) may include a plurality of user terminals (110-1, 110-n), a server (120), and a database (130). This system adopts a distributed architecture and operates based on a client-server model, with each component playing a unique role and contributing to increasing the efficiency and scalability of the entire system. According to one embodiment, the database (130) is depicted as being configured separately from the server (120), but is not limited thereto; depending on system design and operational efficiency, the database (130) may be integrated within the server (120). This integrated configuration has the advantage of improving data access speed and reducing system complexity. For example, the server (120) may include a plurality of artificial intelligence models and processing units for performing machine learning algorithms, and these may implement various types of deep learning and machine learning technologies (e.g., CNN, RNN, Transformer, reinforcement learning, etc.) to respond to user requests and provide intelligent services. According to one embodiment, a plurality of user terminals (110-1, 110-n), a server (120), and a database (130) can be connected to communicate with each other through a network (N), which enables real-time data exchange and smooth interaction between system components.

[0041] According to one embodiment, the network (N) may perform wireless or wired communication between a plurality of user terminals (110-1, 110-n), a server (120), a database (130), etc., and is located in the center of FIG. 1 and serves as a hub connecting all components. For example, the network may perform wireless communication according to methods such as 5G, LTE (Long-Term Evolution), LTE-A (LTE Advanced), CDMA (Code Division Multiple Access), WCDMA (Wideband CDMA), WiBro (Wireless BroadBand), WiFi (Wireless Fidelity), Bluetooth, NFC (Near Field Communication), GPS (Global Positioning System), or GNSS (Global Navigation Satellite System). The 5G network supports high-speed data transmission (up to 20Gbps), ultra-low latency (1ms or less), and large-scale connectivity, making it suitable for real-time processing of large-capacity AI models. For example, the network (N) may be configured to perform wired communication using methods such as USB (Universal Serial Bus), HDMI (High Definition Multimedia Interface), RS-232 (Recommended Standard 232), Ethernet, fiber optic cable, or POTS (Plain Old Telephone Service). In particular, in data center environments, high-performance network technologies such as 400Gbps Ethernet or InfiniBand can be utilized for communication between large-scale AI computing clusters.

[0042] According to one embodiment, user terminals (110-1, 110-n) are various client devices that access the system and can be implemented in various forms such as smartphones, tablets, desktop computers, wearable devices, and IoT devices. These terminals are responsible for transmitting service requests through a user interface and receiving and displaying results processed by a server (120). Each terminal may also run a lightweight AI model locally, which can reduce network latency and enhance privacy protection. Direct connection or peer-to-peer (P2P) communication between terminals may be possible, as indicated by the dotted line, which is particularly useful in distributed learning or edge computing scenarios.

[0043] According to one embodiment, the server (120) is the central processing unit of the system, responsible for processing user requests, executing artificial intelligence models, and managing data. The server is equipped with a high-performance processor (CPU, GPU, TPU, NPU, etc.) for large-scale computation, large-capacity memory, and a stable operating system, and may be operated in a virtualized manner on a cloud infrastructure. The server performs training and inference functions for complex deep learning models and transmits the results to user terminals or stores them in a database. In addition, the stability and reliability of the system can be ensured through functions such as load balancing, error recovery, and security management.

[0044] According to one embodiment, a database (130) is a storage facility for storing and managing various data, and is depicted in a cylindrical shape in FIG. 1. Data stored in the database (130) is data acquired, processed, or used by at least one component of a plurality of user terminals (110-1, 110-n) or a server (120), and may include software (e.g., programs), user profiles, training datasets, model weights, log records, etc. Structured relational data may be stored in a SQL-based database (MySQL, PostgreSQL, etc.), and unstructured large-scale data may be stored in a NoSQL database (MongoDB, Cassandra, etc.) or a distributed file system (Hadoop HDFS, etc.). The database (130) may include volatile and / or non-volatile memory, where volatile memory (RAM) is used for caching or temporary data processing requiring fast data access, and non-volatile memory (SSD, HDD, tape drive, etc.) is used for permanent data storage. Databases also ensure the security and integrity of data through data encryption, access control, and backup and recovery mechanisms.

[0045] This artificial intelligence-based system (400) can be utilized in various application fields and can provide services such as user behavior analysis, recommendation systems, natural language processing, image and video recognition, and predictive analysis. In particular, the distributed architecture and scalable design of the system enable stable performance to be maintained even as the number of users and data volume increase.

[0047] FIG. 2 is a diagram illustrating the learning of a neural network according to one embodiment.

[0048] As illustrated in FIG. 2, the learning device can train a neural network (123) to classify review responses received from multiple user terminals (110-1, ���) by item. Additionally, the learning device can train a neural network (123) to extract user stay history from user movement path information. The neural network (123) is also called an artificial neural network and is a computational model designed inspired by the neural structure of the human brain, specialized in recognizing and learning complex patterns within data. According to one embodiment, the learning device may be a separate entity from the server (120), but it may also be implemented integrated into the same system, so it is not limited thereto.

[0049] According to one embodiment, the neural network (123) includes an input layer (121) into which training samples are input and an output layer (125) that outputs training outputs, and can be learned based on the difference between the training outputs and labels (i.e., actual correct data). Here, the labels are defined based on items corresponding to review responses (e.g., service quality, price satisfaction, cleanliness, etc.) and can be defined based on user dwell history (place visited, time spent, frequency of visit, etc.) corresponding to movement path information. The neural network (123) is connected as a group of multiple nodes and is defined by weights between the connected nodes and an activation function that activates the nodes. Various forms of the activation function may be used, such as sigmoid, hyperbolic tangent (tanh), and ReLU, which enables the network to learn non-linear patterns.

[0050] According to one embodiment, the learning device can train a neural network (123) using a Gradient Descent (GD) technique or a Stochastic Gradient Descent (SGD) technique. The GD technique is a method of updating weights all at once by calculating gradients based on the entire dataset, while the SGD technique is a method of increasing computational efficiency and the possibility of escaping a local optimum by updating weights more frequently using only a randomly selected portion of data (mini-batch). The learning device can use a loss function designed by the outputs and labels of the neural network. The loss function is an important element that quantifies the difference between the model's predicted value and the actual correct answer to suggest the direction of learning.

[0051] The learning device can calculate the training error using a predefined loss function. The loss function can be predefined with labels, outputs, and parameters as input variables, where the parameters can be set by weights within the neural network (123). For example, the loss function can be designed in the form of Mean Square Error (MSE), entropy, etc. MSE is calculated as the squared mean of the difference between the predicted value and the actual value and is mainly used for regression problems, while cross-entropy is a function that measures the difference between the predicted probability distribution and the actual distribution and is suitable for classification problems. In addition, loss functions that are less sensitive to outliers, such as Huber Loss, can be utilized, and various techniques or methods can be employed in the embodiments where the loss function is designed.

[0052] According to one embodiment, the learning device can identify weights that influence the training error using a backpropagation technique. Backpropagation is a process of calculating the degree to which each weight contributes to the final error while propagating the error calculated in the output layer toward the input layer, and is performed through differential calculations using the chain rule. Here, the weights are relationships between nodes within the neural network (123). The learning device may use an SGD technique using labels and outputs to optimize the weights found through the backpropagation technique. For example, the learning device may update the weights of a loss function defined based on labels, outputs, and weights using an SGD technique. This process operates by calculating the gradient (the derivative of the loss function with respect to the weights) and adjusting the weights in the direction of this gradient to gradually decrease the value of the loss function.

[0053] According to one embodiment, the learning device extracts first objects from a review response, obtains first labels which are items corresponding to the first objects, applies the first objects to a first neural network to generate first training outputs corresponding to the first objects, and can train the first neural network based on the first training outputs and the first labels. At this time, the first objects may include key keywords, sentence structures, sentiment expressions, etc. extracted from the review text, and thereby learn the ability to identify the core content of the review and classify it into the corresponding items.

[0054] According to one embodiment, the learning device extracts second objects from movement path information, obtains second labels which are user dwell history corresponding to the second objects, applies the second objects to a second neural network to generate second training outputs corresponding to the second objects, and can train the second neural network based on the second training outputs and the second labels. The second objects may consist of data such as the user's GPS coordinate sequence, movement speed, stopping point, and movement pattern, thereby learning the ability to determine where and for how long the user stayed.

[0055] According to one embodiment, a learning device can generate first training feature vectors based on constituent features (e.g., sentence structure, word frequency, part-of-speech distribution), positional features (e.g., location of important words, location of key sentences), and pattern features (e.g., repetitive expressions, usage patterns of specific phrases) of a review response. These feature vectors are generated through a process of converting raw text data into a numerical form that can be processed by a neural network, and various methods such as TF-IDF, Word2Vec, and BERT may be employed to extract features.

[0056] According to one embodiment, the learning device can generate second training feature vectors based on constituent features of movement path information (e.g., shape of the path, estimation of the means of movement), length features (e.g., total travel distance, travel time, distance between each point), and pattern features (e.g., repeated visit patterns, movement patterns by day of the week / time of the day). Since location data has continuous values ​​over time, time-series data processing techniques or spatial data analysis methodologies may be applied, and various methods may be employed to extract features.

[0057] According to one embodiment, the learning device can obtain training outputs by applying first training feature vectors to the neural network (123). This process is also called feedforward and is a process in which input data passes through each layer of the network to generate a final prediction value. The learning device can train the review item extraction algorithm of the neural network (123) based on the training outputs and first labels. The learning device can train the review item extraction algorithm of the neural network (123) by calculating training errors corresponding to the training outputs and optimizing the connection relationships of nodes within the neural network (123) to minimize the training errors. This process is repeated over several epochs, and in each epoch, the entire training dataset is processed and weights are updated. The server (120) can automatically extract and classify items from new review responses using the first neural network that has been trained.

[0058] According to one embodiment, the learning device can obtain training outputs by applying second training feature vectors to the neural network (123). The learning device can train the user stay history acquisition algorithm of the neural network (123) based on the training outputs and second labels. The learning device can train the user stay history acquisition algorithm of the neural network (123) by calculating training errors corresponding to the training outputs and optimizing the connection relationships of nodes within the neural network (123) to minimize the training errors. Since movement path data has temporal continuity, a recurrent neural network (RNN) structure such as LSTM (Long Short-Term Memory) or GRU (Gated Recurrent Unit) can be applied, which enables effective learning of the patterns of the sequence data. The server (120) can obtain user stay history from movement path information using the second neural network after training is complete, and the information thus obtained can be utilized in various application fields such as improving location-based services, building customized recommendation systems, and analyzing traffic patterns.

[0060] FIG. 3 is a diagram illustrating the configuration of an artificial intelligence model according to one embodiment.

[0061] An artificial intelligence model according to one embodiment may include an input layer, a hidden layer, and an output layer. This multi-layer structure is effective for learning complex patterns and solving non-linear problems, and is utilized as a core structure, particularly in deep learning models. The input layer is a layer related to input values ​​input into the artificial intelligence model and serves to receive data coming from the outside. As illustrated in FIG. 3, the input layer consists of three nodes, and each node may represent a specific feature or attribute of the data. The number of nodes in the input layer may vary depending on the dimension or characteristics of the data to be processed, and may be determined by the number of pixels in the case of image data, the dimension of the word vector in the case of text data, etc.

[0062] According to one embodiment, a feature map can be output by performing a MAC (multiply-accumulate) operation and an activation operation on the input values ​​in the hidden layer. In FIG. 3, the hidden layer consists of four nodes, and these nodes are connected to all nodes of the input layer to form a fully connected structure. The MAC operation may be an operation that multiplies the input values ​​by their corresponding weights and sums the multiplied values, and this acts as the basic unit of information processing in a neural network.

[0063] According to one embodiment, the activation operation may be an operation that inputs the result of a MAC operation into an activation function and outputs a result value. The activation function is a key element that enables a model to learn complex patterns by introducing non-linearity into the result of a linear operation. The activation function may be of various types. For example, the activation function may include a sigmoid function (which outputs a value between 0 and 1 and is useful for expressing probability), a tangent function (which outputs a value between -1 and 1 and is suitable for balanced data), a ReLU function (which outputs 0 for negative inputs and outputs it as is for positive inputs, mitigating the vanishing gradient problem), a Leaky ReLU function (which applies a linear function with a small gradient to negative inputs), a MaxOut function (which selects the maximum value among several linear functions), and / or an ELU function (which is an exponential linear unit function that provides a smooth saturation curve for negative inputs), but there are no limitations on the types thereof. Recently, new activation functions such as Swish and Mish are also being studied and are contributing to the improvement of model performance.

[0064] According to one embodiment, the hidden layer may be composed of at least one layer, and the more layers there are, the more complex patterns can be learned, resulting in a deep learning structure. For example, when the hidden layer is composed of a first hidden layer and a second hidden layer, the first hidden layer performs MAC operations and activation operations based on the input value of the input layer to output a feature map, and the feature map, which is the result value from the first hidden layer, can become the input value for the second hidden layer. The feature map is an intermediate representation representing features extracted from input data; in image processing, it can represent visual features such as edges, textures, and shapes, and in text processing, it can represent semantic patterns or syntactic structures. The second hidden layer can perform MAC operations and activation operations based on the feature map, which is the result value of the first hidden layer, and through this hierarchical processing, increasingly abstract and high-dimensional features can be extracted.

[0065] According to one embodiment, the output layer may be a layer associated with the result of an operation performed in the hidden layer, and in FIG. 3, it consists of three nodes. The number of nodes in the output layer varies depending on the type of problem to be solved; for binary classification problems, it may be one node, for multi-class classification problems, it may be equal to the number of classes, for regression problems, it may be determined by the dimension of the value to be predicted, and for sequence generation problems, it may be determined by the length of the output sequence. In the output layer, a special activation function such as a softmax function or a sigmoid function is generally used to generate a final predicted value or a probability distribution.

[0066] In one embodiment, the learning model learns syllable (character) patterns that are frequently combined and used in a given corpus to automatically learn the boundaries of compound words and named entities, integrates object information from a first UI source with object information rendered in a browser to create a learning object information file, and uses the learning object information file to generate training data for training a deep learning network. In this process, natural language processing (NLP) technology is utilized, and architectures such as word embedding, recurrent neural networks (RNN), and Transformers may be applied. Additionally, the learning model receives data from various domains of a support system, standardizes the data from the various domains into an integrated format based on at least one standardization method corresponding to each of the various domains, learns and infers data from a specific domain, determines information to be transmitted for standardization from the data of the specific domain, and performs post-processing on the data from the various domains.

[0067] According to one embodiment, the first UI source includes an XML file, and the training object information file includes an input JSON file for feature learning and an output JSON file that serves as label data during training. This structured data format allows the model to process it easily and is particularly useful for analyzing UI elements of web-based applications. The output JSON file includes a file containing HTML DOM Tree information implemented in compliance with web standards, and the various domains include at least one of a RAN (radio access network), a transport, or a core. A RAN refers to a wireless access network where a mobile device connects to a cellular network; a transport refers to a network layer responsible for data transmission; and a core refers to the central part of a network responsible for central processing and routing. Post-processing may include a correlation function, which plays an important role in analyzing relationships between data collected from various domains and discovering meaningful patterns. Through correlation analysis, the model can identify the causes of anomalies or performance degradation occurring in complex systems and improve prediction accuracy.

[0069] Figure 4 is a block diagram showing the configuration of an emotion similarity-based user emotion matching system according to one embodiment.

[0070] A system (400) according to one embodiment may include a processor (420) and memory (430), and some of the illustrated components may be omitted or substituted. Although the system diagram briefly depicts only the core components, the actual implementation may include various additional components such as an input / output controller, a system bus, a graphics processing unit (GPU), a network interface card (NIC), and a storage device controller. A system (400) according to one embodiment may be a server or a terminal; if implemented as a server, it may be operated in a virtualized environment as part of a cloud infrastructure or implemented as a physical hardware server, and if implemented as a terminal, it may be implemented in various forms such as a desktop computer, laptop, tablet, smartphone, embedded system, or IoT device. According to one embodiment, the processor (420) is a component capable of performing operations or data processing regarding the control and / or communication of each component of the system (400), and may be composed of one or more processors. In modern processor architectures, multi-core designs are common, which improve parallel processing performance by integrating multiple independent processor cores within a single chip. For example, configurations such as dual-core, quad-core, and octa-core are possible, and each core has independent cache memory (L1, L2), and some cache (L3) can be shared among cores.

[0071] The memory (430) can store information related to the method described above or a program in which the method described above is implemented. The memory subsystem is designed with a hierarchical structure and is composed of several levels according to access speed and capacity. The memory (430) may be volatile memory or non-volatile memory. Volatile memory is memory in which stored data is lost when the power is turned off, and is mainly implemented in the form of DRAM (Dynamic RAM) or SRAM (Static RAM). DRAM stores each bit in the form of charge in a capacitor and requires periodic refreshing, whereas SRAM uses flip-flop circuits to maintain data, so it does not require refreshing but has higher cost and power consumption due to the use of more transistors. The memory (430) can store various file data, and the stored file data can be updated according to the operation of the processor (420). The memory management unit (MMU) is responsible for converting virtual memory addresses to physical memory addresses, managing page tables, and maintaining memory protection and cache consistency.

[0072] According to one embodiment, the processor (420) can execute a program and control the device (400). The processor can operate through basic pipeline stages of instruction fetch, decode, execute, and write-back. The code of the program executed by the processor (420) can be stored in memory (430). In addition to the application, the operating system, device driver, middleware, system services, etc. are loaded into memory and executed, and these provide basic system functions such as hardware resource management, process scheduling, interrupt handling, and file system management. The operations of the processor (420) can be performed by loading instructions stored in memory (430). In this process, the program counter (PC) points to the address of the next instruction to be executed, the instruction register (IR) stores the instruction currently being executed, and various general-purpose registers store operation data and intermediate results. The system (400) can be connected to an external device (e.g., a personal computer or a network) through an input / output device (not shown in the drawing) and exchange data. The input / output system can operate on an interrupt-based or polling-based basis and can transfer data directly between memory and the input / output device without the intervention of a processor through a Direct Memory Access (DMA) controller.

[0073] According to one embodiment, data exchange between memory (430) and processor (420) is performed via a system bus, which consists of an address bus, a data bus, and a control bus. The address bus specifies the memory location that the processor intends to access, the data bus transmits actual data, and the control bus transmits control information such as read / write signals. The system (400) may be operated under an operating system that supports advanced features such as multitasking, virtual memory, memory protection, and separation of privileges, and if a real-time operating system (RTOS) is used, it may satisfy deterministic response times and strict time constraints. For artificial intelligence applications, the memory (430) stores a neural network model, weight parameters, a training dataset, intermediate calculation results, etc., and the processor (420) may perform inference and learning by executing forward propagation and backpropagation algorithms. The system (400) can ensure high availability and fault tolerance by applying technologies such as clustering, load balancing, and failover mechanisms for scalability, and can implement technologies such as data encryption, access control, secure booting, and memory protection for security.

[0075] Figure 5 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment.

[0076] The operations described through FIG. 5 can be implemented based on instructions that can be stored in a computer recording medium or memory (e.g., memory (430) of FIG. 4). The order of each operation of FIG. 5 may be changed, some operations may be omitted, and some operations may be performed simultaneously.

[0077] In operation 510, the system (400) can collect user-specific emotion record data. In this process, the system (400) can obtain data including each user's emotion label, emotion score, and time information. The collected emotion data can serve as a basis for tracking the user's emotional state over time. The system (400) can collect this data by utilizing natural language processing technology or emotion analysis models such as HuggingFace Transformers.

[0078] In operation 520, the system (400) can vectorize an emotion profile based on emotion data for a recent period of time for each user from the collected emotion record data. This vectorization process may be a task of converting emotion data into a form that can be quantitatively analyzed. The system (400) may calculate an average intensity score for each emotion label and convert it into a multidimensional vector, which may be a method of numerically representing the user's emotional state.

[0079] In operation 530, the system (400) generates a target set of users from among multiple users, calculates cosine similarity between vectorized sentiment profiles, and can extract matching candidates between users if the calculated cosine similarity is above a certain threshold. Additionally, the system (400) can suggest at least one connection to the extracted matching candidates to the user, such as empathy-based report exchange, character sharing, or anonymous conversation. This process can be implemented using an algorithm such as cosine_similarity from sklearn.metrics.pairwise.

[0080] According to one embodiment, the system (400) collects emotion record data for each user, and the collected emotion record data includes emotion labels, emotion scores, and time information. It vectorizes an emotion profile based on emotion data for each user over a recent period from the collected emotion record data, generates a target user set among a plurality of users, calculates cosine similarity between vectorized emotion profiles, extracts matching candidates between users when the calculated cosine similarity is above a certain threshold, and can propose at least one connection among empathy-based report exchange, character sharing, or anonymous conversation to the user for the extracted matching candidates.

[0081] According to one embodiment, the system (400) can collect user-specific emotion record data. This emotion record data is a basic unit that quantitatively expresses the user's emotional state and, specifically, may include emotion labels such as 'joy', 'sadness', and 'anger', an emotion score indicating the intensity of these emotions, and time information in which the emotions were recorded. The collection of emotion data can be implemented by analyzing the user's text input using natural language processing technology, and various emotional states can be extracted simultaneously by utilizing a multi-label emotion analysis model such as HuggingFace Transformers.

[0082] According to one embodiment, the system (400) can vectorize an emotion profile based on emotion data for a recent period for each user from the collected emotion record data. Emotion profile vectorization is a process of converting qualitative emotion data into a numerical form that can be processed by a machine learning algorithm, and can be expressed as a fixed-length multidimensional vector by calculating an average intensity score for each emotion label. In this process, characteristics of emotional change over time can also be captured based on emotion data from the last N days, which can serve as a basis for more accurately determining emotional similarity.

[0083] According to one embodiment, the system (400) can generate a target user set among a plurality of users and calculate cosine similarity between vectorized emotion profiles. Cosine similarity is a method of measuring the cosine value of the angle between vectors, and can quantify how similar the directions of two vectors are as a value between -1 and 1. This calculation can be performed efficiently by utilizing the cosine_similarity function of the sklearn.metrics.pairwise module, and a value closer to 1 may indicate that the emotion profiles of two users are similar.

[0084] According to one embodiment, the system (400) can extract matching candidates between users when the calculated cosine similarity is greater than or equal to a preset threshold. This threshold can be appropriately set considering the operational purpose of the system and user experience, and can serve as an important parameter to ensure the quality level of the matching. The extraction of matching candidates can be implemented by assigning priorities based on similarity scores, so that users with the most similar emotional profiles can be selected as top candidates.

[0085] According to one embodiment, the system (400) may propose to the user at least one connection among empathy-based report exchange, character sharing, or anonymous conversation to the extracted matching candidate. Such connection proposals may have the purpose of promoting emotional empathy and communication between emotionally similar users. Empathy-based reports may be implemented in the form of sharing anonymized emotional data of another user in a similar emotional state, and character sharing may be a method of sharing a virtual avatar that expresses emotional data as visual elements. Anonymous conversation may be performed through a secure channel that allows only emotional data and related context to be exchanged without personal identification information.

[0087] Figure 6 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment.

[0088] The operations described through FIG. 6 can be implemented based on instructions that can be stored in a computer recording medium or memory (e.g., memory (430) of FIG. 4). The order of each operation of FIG. 6 may be changed, some operations may be omitted, and some operations may be performed simultaneously.

[0089] In operation 610, the system (400) can utilize a natural language processing-based sentiment analysis model to extract multiple sentiment labels, including joy, sadness, and anger, and intensity scores for each sentiment from the user's text input. In this process, the system (400) can utilize a deep learning model, such as HuggingFace Transformers, to accurately identify and quantify various emotional states embedded in the text. This sentiment analysis can serve as a basis for objectively evaluating the user's emotional state.

[0090] In operation 620, the system (400) can store the extracted emotion data and the corresponding time information together in a time series form in a database. Additionally, the system (400) can convert the user-specific emotion profile into a multidimensional vector form by calculating the average intensity score for each emotion label based on the emotion data over a certain period during the vectorization process. This multidimensional vector can be stored in the database as a fixed-length vector composed of the intensity values ​​of each emotion type as elements. The system (400) can implement this data storage by utilizing a database system such as SQLite or Redis.

[0091] In operation 630, the system (400) can calculate the cosine value of the angle between vectors by utilizing a similarity measurement function of a machine learning library during the cosine similarity calculation process. The system (400) can determine that the closer this cosine value is to 1, the higher the similarity, and the closer it is to 0, the lower the similarity. Additionally, the system (400) can assign matching priority based on the similarity score by identifying user pairs that exceed a preset similarity threshold during the matching candidate extraction process. This process can be efficiently implemented by utilizing the cosine_similarity function of sklearn.metrics.pairwise.

[0092] In operation 640, the system (400) may present a guidance message containing matching information based on emotional rhythm similarity and interaction options during the process of providing matching suggestions to the user. Depending on the user's choice, the system (400) may activate a selected connection function among creating an anonymous conversation channel, exchanging emotional reports, or sharing emotional characters. This process can serve to enhance the user experience and promote the formation of emotional empathy. These functions can be implemented through a FastAPI-based matching recommendation API.

[0093] According to one embodiment, the system (400) utilizes a natural language processing-based emotion analysis model in the process of collecting user-specific emotion record data to extract a plurality of emotion labels, including joy, sadness, and anger, and an intensity score for each emotion from the user's text input, stores the extracted emotion data and corresponding time information together in a time series form in a database, converts the user-specific emotion profile into a multidimensional vector form by calculating an average intensity score for each emotion label based on emotion data over a certain period, and stores the multidimensional vector in the database as a fixed-length vector composed of an intensity value of each emotion type as an element, calculates the cosine value of the angle between vectors using a similarity measurement function of a machine learning library in the cosine similarity calculation process, determines that the closer the cosine value is to 1, the higher the similarity, and the closer it is to 0, the lower the similarity, identifies user pairs that exceed a preset similarity threshold in the matching candidate extraction process to assign a matching priority according to the similarity score, presents a guidance message including matching information based on emotional rhythm similarity and interaction options in the process of providing a matching proposal to the user, and selects a connection among the creation of an anonymous chat channel, exchange of emotion reports, or sharing of emotion characters according to the user's choice. You can activate the function.

[0094] According to one embodiment, the system (400) may utilize a natural language processing-based sentiment analysis model in the process of collecting user-specific sentiment record data. This model can recognize various emotional states from a user's text input based on deep learning technologies such as Transformer Architecture or BERT. It can capture multifaceted emotional states by simultaneously identifying multiple emotion labels, such as joy, sadness, and anger, in the text and quantifying the intensity of each emotion into a score between 0 and 1. This multi-emotion labeling approach may have the advantage of being able to more accurately model complex human emotional states.

[0095] According to one embodiment, the system (400) may store the extracted emotion data and the corresponding time information together in a time series format in a database. This time series data structure may be essential for tracking the temporal changes and patterns of emotions and can be efficiently managed through a database system such as SQLite or Redis. Through the time series storage method, time-dependent characteristics such as the periodicity, duration, and transition speed of emotions can be analyzed, which can provide information beyond simple emotional states.

[0096] According to one embodiment, the system (400) can, in the process of vectorizing user-specific emotion profiles, calculate an average intensity score for each emotion label based on emotion data over a certain period (e.g., the last 7 or 30 days) and convert it into a multidimensional vector form. This multidimensional vector is a fixed-length vector composed of elements of intensity values ​​for each emotion type (joy, sadness, anger, anxiety, etc.), and can provide a statistical summary of the emotional state. Such a vector representation can be efficiently stored in a database and can serve as the basis for calculating emotion similarity by enabling mathematical operations in a high-dimensional space.

[0097] According to one embodiment, the system (400) can calculate the cosine value of the angle between vectors by utilizing a similarity metric function of a machine learning library such as sklearn.metrics.pairwise in the process of calculating cosine similarity. This cosine value quantifies the directional similarity of the two vectors as a value between -1 and 1, and may indicate that the closer to 1, the more similar the emotional profiles of the two users are, and the closer to 0, the more different they are. Since cosine similarity focuses on the direction rather than the magnitude of the vectors, it may be suitable for matching that prioritizes the relative distribution pattern rather than the absolute intensity of emotional expression.

[0098] According to one embodiment, the system (400) can assign matching priorities based on similarity scores by identifying pairs of users that exceed a preset similarity threshold (e.g., 0.7 or 0.8) during the matching candidate extraction process. This threshold is an important parameter that ensures the quality of the matching and the accuracy of the connection proposal, and can be optimized based on system operation data and user feedback. By assigning priority to pairs of users with high similarity scores, connections with a high likelihood of emotional empathy can be suggested preferentially.

[0099] According to one embodiment, the system (400) may present a guidance message containing matching information based on emotional rhythm similarity and various interaction options during the process of providing a matching proposal to a user. This message may induce user participation by expressing the emotional similarity information that serves as the basis for the matching in a form that is easy for the user to understand and by clearly guiding possible methods of interaction. Depending on the user's choice, the system (400) may activate connection functions such as creating an anonymous conversation channel, exchanging emotional reports, or sharing emotional characters. The anonymous conversation channel may be implemented as a messaging system with filtering technology applied for privacy protection, the exchange of emotional reports may be implemented by sharing each user's emotional statistics and insights, and the sharing of emotional characters may be implemented by expressing emotional states as visual characters to facilitate interaction.

[0101] Figure 7 is a flowchart illustrating a user emotion matching method based on emotion similarity according to one embodiment.

[0102] The operations described through FIG. 7 can be implemented based on instructions that can be stored in a computer recording medium or memory (e.g., memory (430) of FIG. 4). The order of each operation of FIG. 7 may be changed, some operations may be omitted, and some operations may be performed simultaneously.

[0103] In operation 710, the system (400) can generate a multidimensional emotion profile vector by using the user's emotion score data as a basic vector element in the emotion profile vectorization process and calculating an emotion rhythm periodicity indicator as a secondary vector element through the emotion transition frequency and periodicity measurement unit. This multidimensional emotion profile vector may include information quantifying the pattern of emotion change over time along with the basic emotion intensity value. The system (400) can effectively capture the periodicity and pattern of emotions by utilizing a time series analysis tool.

[0104] In operation 720, the system (400) can dynamically adjust the matching algorithm according to the user's current major emotional state by applying weights to the similarity of a specific emotion category along with the overall emotion profile similarity when extracting matching candidates. This application of weights can improve the appropriateness of the matching by giving greater importance to the user's current emotional state. The system (400) can provide more accurate matching results by optimizing these weights through a machine learning algorithm.

[0105] In operation 730, the system (400) calculates an emotional resonance index between users as a normalized value based on the cosine similarity calculation result, and can visually display this emotional resonance index on a user interface through a display processing unit. Visualization of the emotional resonance index can allow the user to intuitively understand the basis and degree of the matching, which can contribute to increasing the transparency and reliability of the matching system.

[0106] In operation 740, the system (400) can create a two-way anonymous messaging channel when the user accepts the match. In this channel, a privacy protection mechanism can be applied through a personal information filtering module to share only emotional data and related context without exposing user identification information. This guarantee of anonymity can create an environment where users can express their emotions more freely and safely. This privacy protection mechanism can be implemented by utilizing API security layers such as JWT authentication and authorization verification.

[0107] According to one embodiment, the system (400) generates a multidimensional emotional profile vector by using the user's emotional score data as a basic vector element in the emotional profile vectorization process and calculating an emotional rhythm periodicity indicator as an auxiliary vector element through an emotional transition frequency and periodicity measurement unit, wherein the multidimensional emotional profile vector includes information quantifying the emotional change pattern over time along with a basic emotional intensity value, and dynamically adjusts the matching algorithm according to the user's current major emotional state by applying weights to the similarity of a specific emotional category along with the overall emotional profile similarity when extracting matching candidates, calculates an emotional resonance index between users as a normalized value according to the cosine similarity calculation result, and visually displays the emotional resonance index on a user interface through a display processing unit, and if the user accepts the matching, creates a two-way anonymous messaging channel, and in the channel, applies a privacy protection mechanism that allows only emotional data and related context to be shared without exposing user identification information through a personal information filtering module.

[0108] According to one embodiment, the system (400) may use the user's emotion score data as a basic vector element in the emotion profile vectorization process. This basic vector element may consist of intensity values ​​for each emotion category (joy, sadness, anger, etc.). The system (400) may also calculate an emotion rhythm periodicity indicator as an auxiliary vector element through an emotion transition frequency and periodicity measurement unit. The emotion rhythm periodicity indicator is a numerical representation of the frequency and regularity of the user's emotions changing in a specific pattern, and can be calculated using a time-series analysis tool such as Prophet. This periodicity indicator can enable more accurate emotional matching by providing temporal pattern information beyond simple emotional states.

[0109] According to one embodiment, the system (400) can generate a multidimensional emotion profile vector by combining these basic elements and auxiliary elements. This multidimensional vector may include information quantifying not only basic emotion intensity values ​​but also patterns of emotion change over time. For example, the speed of emotion transition, the duration of emotion, and the probability of transitioning between specific emotions may be included in this vector, which can more comprehensively represent the user's emotional characteristics.

[0110] According to one embodiment, the system (400) may apply weights to the similarity of a specific emotion category along with the overall emotion profile similarity when extracting matching candidates. This is a method of dynamically adjusting the matching algorithm by prioritizing the user's current major emotional state; for example, if the user is currently experiencing strong sadness, a higher weight may be given to the similarity of the sadness emotion. This dynamic application of weights can enhance emotional empathy and support effects by providing a match that is more suitable for the user's current emotional state.

[0111] According to one embodiment, the system (400) can calculate an emotional resonance index between users as a normalized value based on the cosine similarity calculation result. The emotional resonance index is an indicator that expresses how harmoniously the emotional profiles of two users resonate as a standardized score between 0 and 100, and may be a form in which the complex cosine similarity calculation result is converted into a form that the user can intuitively understand. The system (400) can display this emotional resonance index on a user interface as a visual element, such as a graph, gauge, or color gradient, through a display processing unit. Such visualization can clearly convey the basis and strength of the matching to the user, thereby increasing credibility and engagement.

[0112] According to one embodiment, the system (400) can create a two-way anonymous messaging channel when a user accepts a match. This channel is a platform capable of real-time or asynchronous message exchange and can be implemented using security technologies such as JSON Web Token (JWT)-based authentication. The system (400) can automatically detect and remove or mask personally identifiable information, such as names, locations, and contacts, from message content by utilizing a filtering module and natural language processing technology. At the same time, by applying a privacy protection mechanism that selectively shares only emotional data and related context, a safe environment can be provided where users can engage in deep emotional exchange while maintaining anonymity.

[0114] According to one embodiment, the system (400) continuously collects interaction data between matched users and analyzes the data to calculate a matching quality feedback indicator including conversation duration, message exchange frequency, emotional responsiveness, and user satisfaction evaluation; inputs the calculated matching quality feedback indicator into a machine learning algorithm to automatically optimize similarity thresholds, vector weights, and matching priority determination rules, thereby continuously improving the accuracy of the matching algorithm; updates the existing emotion profile vector in real time and readjusts the matching results when a change exceeding a set standard of the user's emotional state is detected; processes the emotion report or conversation content shared during the emotional exchange between matched users using an emotion analysis algorithm to measure the level of mutual empathy and emotional support effects; and quantifies the strength of the emotional connection between users according to the measured level of empathy and emotional support effects to provide a customized suggestion that promotes continuous interaction.

[0115] According to one embodiment, the system (400) can continuously collect interaction data between matched users. This interaction data may include information such as message timestamps, conversation content, and response times, thereby allowing for the quantification of various aspects such as conversation duration, frequency of message exchange, emotional responsiveness, and user satisfaction evaluation. The system (400) can synthesize this data to calculate a matching quality feedback indicator, which can be used as a standard for objectively evaluating the effectiveness of the matching and the user experience. The collection of interaction data can be implemented through a log system and a user feedback interface of the messaging platform.

[0116] According to one embodiment, the system (400) can automatically optimize key parameters of the matching system by inputting the calculated matching quality feedback indicators into a machine learning algorithm, such as supervised learning or reinforcement learning. Specifically, the accuracy of the matching algorithm can be continuously improved by adjusting the similarity threshold, vector weights by sentiment category, and matching priority determination rules based on data. This self-optimization process enables the system to learn from user feedback and interaction patterns to provide more effective matching, and can cause the quality of the service to gradually improve over time.

[0117] According to one embodiment, the system (400) can update the existing emotion profile vector in real time and readjust the matching result when a change exceeding a set criterion for the user's emotional state (e.g., a change of 30% or more in a specific emotion score) is detected. This is intended to provide timely matching by always accurately reflecting the user's current emotional state, thereby ensuring the agility of the system to adaptively respond to rapid changes in the emotional state. Real-time updates can be implemented through efficient linkage between the emotion analysis module and the vector database.

[0118] According to one embodiment, the system (400) can process emotion reports or conversation content shared during the process of emotional exchange between matched users using a natural language processing-based emotion analysis algorithm. Through this, the level of mutual empathy (how well they understand and respond to each other's emotions) and the effect of emotional support (how much the emotional state improves through conversation) can be quantitatively measured. The measurement of the empathy level can be achieved by analyzing the appropriate response rate to emotional expressions, the degree of agreement of emotion words, and the frequency of positive support expressions.

[0119] According to one embodiment, the system (400) can quantify the strength of the emotional connection between users into a score between 0 and 100 based on the measured level of empathy and the effect of emotional support. This emotional connection strength indicator can be used as a comprehensive indicator representing the depth of emotional harmony and mutual support relationships between two users. Based on this indicator, the system (400) can promote continuous interaction between users by providing customized suggestions, such as specific activity suggestions, conversation topic recommendations, or shared content recommendations. These customized suggestions can further strengthen the emotional connection between users and contribute to the formation of a long-term emotional support network.

[0121] According to one embodiment, the system (400) implements a consent-based access control system for the protection of personal information during the process of collecting user emotional data, provides an authorization management interface that allows the user to set the range and level of emotional data that is allowed to be shared in detail, applies differential privacy techniques to the collected raw emotional data to convert it into a form that cannot identify individuals, performs a vectorization process, implements a consent-based information exchange protocol that selectively shares only mutually agreed-upon information during user matching, operates a natural language processing-based monitoring system to detect malicious use or inappropriate content in an anonymous conversation channel between matched users, executes immediate intervention and protective measures when a risk factor is detected, and can enhance data security by applying encrypted storage and access restrictions to all information including user emotional data and matching history.

[0122] According to one embodiment, the system (400) may implement a consent-based access control system for the protection of personal information during the process of collecting user emotion data. According to one embodiment, the system (400) may provide a permission management interface that allows the user to set the scope and level of emotion data that is permitted for sharing in detail. This interface may be implemented in the form of an intuitive control panel that enables detailed permission settings by emotion type (joy, sadness, etc.), intensity (high, medium, low), and sharing target (all users, specific groups, specific users). Such detailed settings allow the user to adjust the scope of emotion data sharing according to their privacy preferences, thereby improving trust in the system and user experience.

[0123] According to one embodiment, the system (400) may apply a differential privacy technique to collected raw sentiment data to convert it into a form where personal identification is impossible, and then perform a vectorization process. Differential privacy is a technique that adds statistical noise to the data to hide accurate individual data while maintaining the statistical utility of the entire dataset, thereby enabling accurate sentiment matching while protecting the user's personal information. This technique can be implemented by utilizing a mathematical framework such as a privacy model or a Laplace mechanism.

[0124] According to one embodiment, the system (400) may implement a consent-based information exchange protocol that selectively shares only mutually agreed-upon information during user matching. This protocol includes a process in which two users explicitly approve the scope of information to be disclosed to each other, and can ensure mutuality and transparency of information sharing through a two-way verification process. Such a protocol may be implemented in a manner similar to a REST API-based request-response model or an OAuth consent flow, and can contribute to building trust between users.

[0125] According to one embodiment, the system (400) may operate a natural language processing-based monitoring system to detect malicious use or inappropriate content in an anonymous conversation channel between matched users. This monitoring system can utilize advanced language models such as BERT or GPT to analyze conversation content in real time and identify risk factors such as hate speech, harassment, exposure of personal information, and fraud attempts. High-accuracy risk detection can be achieved by combining text classification algorithms and pattern recognition technology.

[0126] According to one embodiment, the system (400) can execute immediate intervention and protective measures upon detection of a risk factor. These measures may include phased responses such as displaying a warning message, filtering risk messages, temporarily suspending conversation, and sending an administrator notification, and an appropriate level of intervention may be performed depending on the severity of the risk. This protection mechanism can enhance the reliability and sustainability of the system by ensuring that users can engage in emotional interaction in a safe environment.

[0127] According to one embodiment, the system (400) can enhance data security by applying access restriction mechanisms, such as role-based access control (RBAC), and encryption storage using an industry-standard encryption algorithm such as AES-256 to all information including user sentiment data and matching history. These security measures can be applied to both data in transit and data in storage, and can be continuously monitored and improved through regular security audits and vulnerability scans. This comprehensive security approach can enhance the reliability of the entire system by ensuring the confidentiality, integrity, and availability of user data.

[0129] According to one embodiment, the system (400) generates a location-based emotion map by combining the user's location information and emotion data, and enables real-time connection between emotionally similar users within a set distance range by implementing a matching algorithm that combines geographical proximity and emotional similarity, and when it detects that the user's emotional state remains in a negative area for a set period, it prioritizes matching users who have successfully transitioned from similar emotional patterns to a positive state, groups multiple emotionally similar users using a similarity clustering algorithm to form an emotional support community, analyzes the collective emotion data of group members to recommend group activities or content that promotes shared emotional well-being, analyzes the change pattern between the user's past emotion data and current emotion state to predict future emotion states, and can provide pre-prepared matching and support services tailored to expected emotional changes.

[0130] According to one embodiment, the system (400) can generate a location-based emotion map by combining a user's location information and emotion data. This emotion map can be implemented in a visualized form that links geographical coordinates with emotional states and can be utilized to identify collective emotional trends in a specific region or place. Additionally, the system (400) can capture the influence of location context on emotions by integrating and processing GPS data and emotion analysis results. Furthermore, the system (400) can enable real-time connections between emotionally similar users within a set distance range (e.g., within 5 km) by implementing a hybrid matching algorithm that combines geographical proximity and emotional similarity. This location-based matching can enhance the depth and continuity of emotional support by providing the possibility of leading to potential offline meetings beyond online connections.

[0131] According to one embodiment, if the system (400) detects that a user's emotional state remains in a negative area for a set period (e.g., 7 consecutive days), it may prioritize matching users who have successfully transitioned to a positive state from a similar emotional pattern. The system (400) may utilize an emotional trajectory analysis algorithm to identify emotional recovery patterns and reflect this in determining the matching priority.

[0132] According to one embodiment, the system (400) can form an emotional support community by grouping a number of emotionally similar users using a similarity clustering algorithm such as K-means or hierarchical clustering. This grouping can create a rich support environment where diverse perspectives and experiences can be shared, going beyond the limitations of one-to-one matching between individuals. The system (400) can apply an algorithm that models group dynamics to determine the optimal group size and composition, which can contribute to the formation of a sustainable emotional support network.

[0133] According to one embodiment, the system (400) can analyze the collective emotional data of group members to recommend group activities or content that promote collective emotional well-being. Such recommendations can be generated through a combination of collaborative filtering and content-based filtering techniques that consider the group's major emotional states, interests, and past interaction patterns. The system (400) can evaluate the effectiveness of recommendations by tracking changes in emotional states after participation in group activities and continuously improve the recommendation algorithm based on this.

[0134] According to one embodiment, the system (400) can predict future emotional states by analyzing the pattern of change between a user's past emotional data and current emotional state using time series analysis techniques. This prediction model can be implemented using advanced time series prediction algorithms such as ARIMA, Prophet, or recurrent neural networks (RNN), and can capture the periodicity and trends of individual emotional changes. Based on these predictions, the system (400) can proactively provide pre-prepared matching and support services tailored to expected emotional changes. This proactive approach can significantly enhance the preventive value of the system by preventing potential emotional crises and enabling continuous management of emotional well-being.

[0136] According to one embodiment, the system (400) collects and analyzes the content consumption patterns of users matched with the user's emotional profile through a content analysis module, identifies media content, literary works, music, or art works that users in similar emotional states have responded positively to, and provides them as a personalized recommendation list through a personalized recommendation generation unit, identifies users who exhibit similar emotional responses under specific environmental conditions through a context matching engine by analyzing the correlation between the user's emotional data and external environmental factors including weather, season, and social events, calculates a linguistic compatibility index by analyzing the user's emotional expression language and communication style using natural language processing technology, improves the accuracy of matching between users with enhanced communication effectiveness by combining the linguistic compatibility index with emotional similarity, tracks emotional changes occurring during the interaction process between matched users in real time, and quantitatively evaluates the impact of the interaction on each user's emotional state to measure and record the effect of promoting emotional well-being.

[0137] According to one embodiment, the system (400) can collect and analyze the content consumption patterns of users matched with the user's emotional profile through a content analysis module. In this process, the system (400) can extract data such as the user's media consumption history, ratings, and reactions, and identify correlations with the emotional state. Content analysis can be performed through a hybrid approach combining natural language processing and metadata analysis, and can be utilized to classify the emotional characteristics of the content. Subsequently, the system (400) can identify media content, literary works, music, or art works to which users in similar emotional states have responded positively, and provide them as a personalized recommendation list through a personalized recommendation generation unit. Such emotion-based recommendations can be implemented as a hybrid recommendation system combining collaborative filtering and content-based filtering techniques, and can help discover content that has a therapeutic effect in a specific emotional state.

[0138] According to one embodiment, the system (400) can analyze the correlation between the user's emotional data and external environmental factors, including weather, seasons, and social events. This analysis can be performed by temporally mapping the user's emotional data with external data, such as weather data, calendar information, and social media trends collected through a public API. Through this, the system (400) can utilize a context matching engine to identify users who exhibit similar emotional responses under specific environmental conditions (e.g., rainy days, public holidays, or the occurrence of specific social issues). The similarity of emotional response patterns to environmental factors provides another dimension of emotional resonance, which may enable more sophisticated matching.

[0139] According to one embodiment, the system (400) can calculate a linguistic compatibility index by analyzing the user's emotional expression language and communication style using natural language processing technology such as BERT or GPT. This index can be calculated by quantifying various linguistic characteristics such as vocabulary selection, sentence structure, directness / indirectness of expression, frequency of metaphor use, and density of emotional words. Furthermore, the system (400) can improve the accuracy of matching between users with enhanced communication effectiveness by combining the calculated linguistic compatibility index with emotional similarity. Matching that considers linguistic compatibility can create a communication environment where users can more easily understand and empathize with each other, going beyond simple similarity of emotional state.

[0140] According to one embodiment, the system (400) can track emotional changes occurring during the interaction process between matched users in real time. This tracking can be performed by collecting data such as emotional analysis of conversation content, response time, changes in message length, and frequency of emotional expression. In addition, the system (400) can measure and record the effect of promoting emotional well-being by quantitatively evaluating the impact of the interaction on each user's emotional state through comparison of pre- and post-emotion scores, the rate of increase in positive emotional expression, and the rate of decrease in negative emotional expression. Such measurement of effects can provide data that objectively proves the therapeutic value of the system and can motivate users to continue participating by visualizing the positive impact of the interaction.

[0142] According to one embodiment, the system (400) can collect biosignal data including heart rate variability, skin conductivity, breathing patterns, and sleep patterns from a user's wearable device. Additionally, the system (400) can generate a more accurate emotional state vector using a multimodal emotion fusion method that integrates text-based emotion data and biosignal data. This multimodal approach can overcome the limitations of a single modality by combining the advantages of various data sources and provide a more complete picture of the emotional state.

[0143] According to one embodiment, the system (400) can detect potential emotional states that are not explicitly expressed in text by learning the correlation between biosignal patterns and emotional states through a deep learning-based time series analysis model. This potential emotion detection function can enable more accurate emotion profiling by capturing emotional states that the user has not consciously recognized or expressed. Additionally, the system (400) can remove environmental fluctuation factors unrelated to emotion, such as physical activity, ambient temperature, and drug effects, from the biosignals using statistical methods and machine learning techniques. This noise removal process can improve the accuracy of the analysis by increasing the purity of emotion-related signals.

[0144] According to one embodiment, the system (400) can build a personalized emotion-biosignal interpretation model by learning a standard biosignal pattern for each user through long-term data collection. This personalized model can enable more accurate interpretation of emotional states by taking into account individual differences that may appear differently for each user even with the same biosignal pattern. Additionally, the system (400) can calculate an authenticity index of emotional expression between 0 and 1 by calculating the degree of agreement between emotions expressed in text and emotions detected by biosignals through statistical correlation analysis. This authenticity index is an indicator that quantifies the degree of agreement between the emotions expressed by the user and the emotions actually experienced physiologically, and can be used as a measure of emotional self-awareness and honesty.

[0145] According to one embodiment, the system (400) can utilize the authenticity index as an additional weighting factor of the matching algorithm to prioritize matching between users whose expressed emotions and actual physiological responses match. This authenticity-based matching can promote more meaningful and sustainable emotional exchanges by connecting users who value the truthfulness and depth of emotional communication. Consequently, the system (400) can promote the formation of a deeper level of emotional empathy through the fusion of biosignals and text emotion data, and can contribute to building connections based on emotional self-awareness and genuine empathy beyond simple matching of emotional states.

[0147] According to one embodiment, the system (400) can calculate an individual emotional resilience index by tracking the pattern, speed, and frequency of a user transitioning from a negative emotional state to a positive emotional state. This emotional resilience index can be calculated through an algorithm that identifies inflection points in time-series emotional data and quantifies the transition speed, and can be expressed as a standardized score between 0 and 100. Additionally, the system (400) can predict each user's emotional vulnerability and the resulting emotional risk situations. This prediction can be achieved by analyzing situational factors that triggered negative emotional spikes in past emotional data and identifying patterns, and can be utilized as an early warning system.

[0148] According to one embodiment, the system (400) can identify users who exhibit complementary emotional resilience in different situations and form a mutually complementary emotional support relationship. For example, by connecting a user who shows strong resilience to occupational stress but is vulnerable to interpersonal problems with a user who has the opposite characteristics, a relationship can be established that complements each other's weaknesses. Additionally, the system (400) can digitize the emotional coping patterns of users with high resilience and convert them into customized coping strategies for users with low resilience. This may include a process of adapting effective coping methods extracted through text mining and behavioral pattern analysis to suit the user's characteristics and situation.

[0149] According to one embodiment, the system (400) may configure a multi-layered support network that is automatically activated when an emotional crisis situation is predicted. This network may be designed with a support structure at multiple levels, such as a first level (matched users), a second level (groups of similar emotions), and a third level (professional support resources), and may include a response mechanism that expands stepwise according to the severity of the crisis. Additionally, the system (400) may quantitatively evaluate the therapeutic effect of the system by monitoring the impact of emotional interactions between users on the development of each individual's emotional resilience over a long period. This evaluation may be measured through indicators such as changes in the resilience index before and after interaction, a decrease in the duration of negative emotional states, and an increase in the frequency of positive transitions.

[0150] According to one embodiment, the system (400) can continuously adjust the parameters of the resilience-based matching algorithm based on the evaluation results. This can be implemented by utilizing a reinforcement learning approach to strengthen matching patterns associated with successful emotional recovery and weaken patterns with low effectiveness. Through this continuous optimization process, the system (400) can implement a matching system optimized for enhancing emotional well-being and psychological resilience. This may signify an evolution beyond simple emotion-similarity-based matching into a therapeutic system that actively supports the user's emotional growth and resilience development.

[0152] According to one embodiment, the system (400) may perform a multidimensional emotion profile vectorization process to analyze the user's emotion data more precisely. In this process, the system (400) can more comprehensively identify the user's emotional characteristics by analyzing not only basic emotion intensity data but also patterns of emotion change over time. The system (400) can quantify the frequency and intensity of emotion transitions by sampling emotion data at specified time intervals and calculating the difference in emotion scores between two consecutive time points, and this information can serve as an important indicator of how sensitively the user reacts to emotional changes. Additionally, the system (400) can analyze the periodicity of emotion changes by applying a Fourier transform algorithm to convert emotion data in the time domain into the frequency domain, thereby confirming whether the user has an emotion pattern that repeats at a specific period. This periodicity information can help in understanding emotional changes associated with the user's biological rhythm or lifestyle patterns.

[0153] According to one embodiment, the system (400) may utilize a weighted cosine similarity calculation method to give greater consideration to the user's current emotional state during the matching process. In this method, the system (400) may perform matching that reflects the user's immediate emotional needs by giving higher weight to the strongest emotional category at the current point in time, which may have the effect of allowing the user to receive more appropriate empathy and support for the major emotions currently being experienced. The system (400) may convert the calculated cosine similarity value into a normalized emotional resonance index between 0 and 1 and provide it in a form that the user can intuitively understand, which can clearly convey the degree of emotional similarity with the matching partner through visual elements such as a percentage bar graph or a pie chart.

[0154] According to one embodiment, the system (400) can establish a messaging environment that enables meaningful communication while ensuring anonymity when a connection between users is approved. To this end, the system (400) can utilize a regular expression-based pattern matching technology to detect personally identifiable information, such as names, phone numbers, email addresses, and addresses, in real time and automatically mask it, while simultaneously protecting privacy by using a hash function to convert specific information into an unrecoverable form. This mechanism can provide a safe communication environment that allows users to freely share contextual information related to their emotional experiences without worrying about the exposure of personal information.

[0155] According to one embodiment, the system (400) can collect and analyze various data generated from interactions between users to continuously improve the quality of the match. The system (400) can determine how long users maintain communication by measuring the duration of conversation sessions and can evaluate the activity of the conversation by analyzing the frequency of message exchanges per unit of time. Additionally, the system (400) can determine the immediacy and engagement of communication by measuring the time a user takes to respond to a message from another party, and can evaluate emotional responsiveness by analyzing the ratio of positive or negative emotional expressions appearing in the message content. The system (400) can collect subjective satisfaction information through star ratings or satisfaction survey responses provided directly by users, and these complex indicators can serve as criteria for evaluating the success of the match from various angles.

[0156] According to one embodiment, the system (400) can continuously optimize the performance of the matching system by utilizing collected matching quality feedback indicators as training data for a machine learning algorithm. The system (400) can learn conditions leading to successful matching through a reinforcement learning model or identify the relationship between specific feedback patterns and matching results through a supervised learning model, and based on this, can dynamically adjust the threshold of cosine similarity or optimize the weights for each element of the sentiment profile vector. Additionally, the system (400) can improve its ability to suggest a more suitable matching partner to the user by improving the rules for determining the priority among matching candidates.

[0157] According to one embodiment, the system (400) may provide a dynamic profile update function to respond to situations where the user's emotional state changes in real time. The system (400) may continuously monitor the difference between the user's current emotional data and the existing profile and immediately update the profile when a change exceeding a preset range of change or change criteria is detected, thereby providing a matching service that reflects the user's latest emotional state. The system (400) may analyze text content appearing during conversations or the exchange of emotional reports between users using natural language processing techniques to measure the level of mutual empathy and the effectiveness of emotional support, and based on this information, may quantify the strength of the emotional connection between users to suggest additional interaction activities or relationship deepening programs.

[0158] According to one embodiment, the system (400) can establish a comprehensive privacy protection system to protect the user's personal information. The system (400) can perform a procedure to clearly explain the purpose of data collection, the method of use, and the retention period before collecting emotion data, and to obtain specific and explicit consent from the user, and can provide a setting interface that allows the user to finely adjust which parts of their emotion data will be shared and to what extent. The system (400) can apply differential privacy techniques to the collected raw emotion data to enable collective pattern analysis without directly exposing the individual user's actual data, and this can be implemented by adding statistical noise or aggregating data in groups.

[0159] According to one embodiment, the system (400) may apply a strict consent-based protocol even during the process of exchanging information between users. The system (400) may allow matched users to pre-set the items and scope of information to be disclosed to each other, and may control the selective sharing of only information that both parties have agreed to, thereby providing a balance point that allows users to form meaningful connections while protecting their privacy. The system (400) may operate a monitoring system utilizing natural language processing technology to detect in real-time malicious use or inappropriate content that may occur in an anonymous conversation channel, and may execute immediate protective measures such as sending a warning message, restricting message transmission, or forcibly terminating the conversation channel when profanity, slander, or threatening remarks are detected.

[0160] According to one embodiment, the system (400) can apply a strong security system to all of the user's personal data. The system (400) can apply an advanced encryption algorithm of AES 256 bits or higher to all personally identifiable information, such as emotion data, emotion profile vectors, and matching history, and store it in a database, and can control access to the data within the necessary scope only for authorized system components according to a predefined access restriction policy. These security measures can constitute a multi-layered security system that protects user data from unauthorized external access or internal abuse of authority.

[0161] According to one embodiment, the system (400) can provide a location-based service that combines a user's physical location and emotional state. The system (400) can generate an emotion map that visualizes the overall emotional atmosphere of a region by analyzing anonymized emotion data collected within a specific geographical area, thereby helping users understand the emotional environment of the region where they are located and to seek appropriate response measures. The system (400) can facilitate connections between users who are physically close and emotionally similar through a hybrid matching algorithm that simultaneously considers geographical proximity and emotional similarity, which can provide the possibility of leading to actual meetings beyond online communication.

[0162] According to one embodiment, the system (400) can implement a selective matching function to provide special support to users in a negative emotional state. The system (400) can continuously monitor the user's emotional data to detect situations where the intensity of negative emotions remains above a certain level or the ratio of negative emotions to positive emotions reaches a concerning level, and in such cases, it can prioritize matching users who have experienced similar difficulties in the past but have now recovered to a positive emotional state. This can create an environment where more effective emotional support can be provided through empathy and advice based on actual experiences.

[0163] According to one embodiment, the system (400) can build a collective emotional support system beyond individual matching. The system (400) can form a community that can share common interests or experiences by grouping multiple emotionally similar users using advanced clustering algorithms such as K-means clustering or DBSCAN, and can recommend customized activities or content that can enhance the emotional well-being of the entire group by analyzing the collective emotional data of the members within this community. The system (400) can predict future emotional changes by analyzing the user's past emotional patterns and current state through a time-series analysis-based prediction model such as an LSTM network, and based on this, can provide customized services such as preventive support services, comforting messages, and referrals to expert counseling in a timely manner.

[0164] According to one embodiment, the system (400) can provide a personalized recommendation service by analyzing content consumption patterns related to the user's emotional state. The system (400) can identify various content such as movies, dramas, webtoons, books, music, and digital artwork that users in similar emotional states have positively evaluated or consumed for a long time, and based on this information, can provide a customized list of content that can help with emotional regulation or mood change by recommending content suitable for the user's current emotional state according to priority. The system (400) can perform context analysis to analyze the statistical correlation between the user's real-time emotional data and external environmental factors such as weather, seasons, and major social events, and through this, can identify users who show similar emotional responses under specific environmental conditions and utilize this environmental context for matching or content recommendation.

[0165] According to one embodiment, the system (400) can analyze linguistic compatibility to enhance the effectiveness of communication between users. The system (400) can identify each user's unique communication style by analyzing the vocabulary diversity of emotional expressions appearing in text input by the user, the complexity of sentence structure, and the frequency and intensity of positive or negative expressions using natural language processing technology, and can calculate a linguistic compatibility index based on this. The system (400) can improve the matching accuracy between users expected to show high effectiveness in actual communication beyond simple emotional similarity by comprehensively considering the linguistic compatibility index calculated in this way with the cosine similarity based on the emotion profile.

[0166] According to one embodiment, the system (400) can measure and evaluate the effects of interactions between users in real time. The system (400) can track changes in each user's emotional state by continuously analyzing the emotional value of messages exchanged between matched users, and can evaluate the actual impact of the interaction on each user's emotional well-being through quantitative indicators such as changes in emotional scores before and after the interaction, an increase in positive emotional expressions, and a decrease in the stress index. These measurement results are recorded within the system and can be used as important feedback data for future improvements to the matching algorithm or service quality, and can ultimately contribute to providing more effective emotional support services to users.

[0167] According to one embodiment, the system (400) implements an emotional resonance amplification mechanism so that emotional exchange between matched users can generate an synergistic empathy effect beyond simple sharing of emotions.

[0168] According to one embodiment, the system (400) can simultaneously analyze emotional patterns at various time scales, ranging from microscopic emotional fluctuations in 10-second intervals to macroscopic emotional cycles in seasonal intervals, through multiscale emotional rhythm analysis. This analysis utilizes a time-frequency decomposition technique based on a wavelet transform to separate multiple overlapping periodicities in emotional time-series data and extract characteristic patterns at each scale. The system (400) can register the extracted multiscale rhythm patterns as emotional fingerprints and promote sustainable emotional harmony and long-term well-being by matching users with complementary rhythm patterns.

[0169] According to one embodiment, the system (400) can predict a user's potential emotional response to a specific stimulus or situation in thousands of possible scenarios through emotional response Monte Carlo simulation. This simulation can generate a distribution of emotional changes in various situations by probabilistically synthesizing the response patterns of other users with profiles similar to the user's past emotional data. Based on these simulation results, the system (400) can recommend matching with the user with the highest emotional resilience in a specific situation or suggest the optimal timing and method of intervention to prevent a potential emotional crisis. Emotional response simulation can enable proactive and preventive emotional management by exploring a virtual future emotional trajectory beyond simple pattern recognition.

[0171] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0172] The embodiments described above may be implemented as hardware components, software components, and / or combinations of hardware and software components. For example, the devices, methods, and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include multiple processing elements and / or multiple types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0173] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave in order to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0174] Although the embodiments have been described above with reference to the limited drawings, those skilled in the art can apply various technical modifications and variations based on the above. For example, suitable results may be achieved even if the described techniques are performed in a different order than described, and / or if the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

Claims

Claim 1 An emotion similarity-based user emotion matching system comprises a memory and a processor for storing instructions, wherein when the instructions are executed by the processor, the system collects user-specific emotion record data including emotion labels, emotion scores, and time information, vectorizes an emotion profile based on emotion data for a recent period for each user from the collected emotion record data, generates a target set of users among multiple users, calculates cosine similarity between vectorized emotion profiles, extracts matching candidates between users when the calculated cosine similarity is above a certain threshold, performs at least one action among empathy-based report exchange, character sharing, or anonymous conversation for the extracted matching candidates, continuously collects interaction data between matched users, and analyzes the collected interaction data to calculate a plurality of matching quality feedback indicators including the duration of a conversation session, the frequency of message exchange per unit time, the response time to the other party's message, emotional responsiveness including the ratio of positive / negative emotion expressions, and a user satisfaction evaluation including at least one of a star rating provided by a user or a satisfaction survey response provided by a user, and the calculated matching quality feedback indicators By using it as an input value for a machine learning algorithm including a reinforcement learning model or a supervised learning model, and by optimizing rules for determining the cosine similarity threshold, weights for each element of the multidimensional emotion profile vector, and priorities among matching candidates, the accuracy of the matching algorithm is continuously improved; and when it is detected that a user's emotional state has changed beyond a preset range of change in emotion intensity or a standard for change in emotion type, the user's existing emotion profile vector is updated in real time, and the matching result is readjusted based on the updated emotion profile vector.A system that processes text included in emotional reports or conversation content shared during the process of emotional exchange between the matched users using a natural language processing-based sentiment analysis model to measure indicators representing the level of mutual empathy and the effect of emotional support, and quantifies the strength of the emotional connection between users into preset steps according to the measured level of empathy and the effect of emotional support, thereby controlling the provision of customized suggestions including suggestions for additional interaction activities or guidance on relationship deepening programs. Claim 2 In claim 1, the instructions are characterized in that, when executed by the processor, the system utilizes a natural language processing-based sentiment analysis model during the process of collecting user-specific sentiment record data to extract multiple sentiment labels, including joy, sadness, and anger, and intensity scores for each sentiment from the user's text input; stores the extracted sentiment data and corresponding time information together in a time-series form in a database; converts the user-specific sentiment profile into a multidimensional vector form by calculating the average intensity score for each sentiment label based on sentiment data over a certain period during the process of vectorizing the user-specific sentiment profile; and stores the multidimensional vector in the database as a fixed-length vector composed of intensity values ​​of each sentiment type as elements; calculates the cosine value of the angle between vectors using a similarity measurement function of a machine learning library; determines that similarity is higher when the cosine value is close to 1 and lower when it is close to 0; assigns matching priorities according to similarity scores by identifying user pairs that exceed a preset similarity threshold during the matching candidate extraction process; displays a guidance message including matching information based on sentiment rhythm similarity and interaction options; and, depending on the user's selection, creates an anonymous chat channel, exchanges sentiment reports, or shares sentiment characters. A system that controls the activation of selected connection functions. Claim 3 In claim 1, when the instructions are executed by the processor, the system uses the user's emotion score data as a basic vector element in the emotion profile vectorization process, quantifies the emotion transition frequency by sampling emotion data at specified time intervals and calculating the amount of change between consecutive emotion data, applies a Fourier transform algorithm to analyze the periodicity of emotion changes in the frequency domain and calculates a periodicity index to generate a multidimensional emotion profile vector using this as an auxiliary vector element, the multidimensional emotion profile vector includes information quantifying the slope, variance, and periodicity of emotion changes over time along with the basic emotion intensity value, extracts matching candidates by dynamically adjusting the matching algorithm according to the user's current major emotional state using a weighted cosine similarity calculation method that applies a specified weight to the emotion category with the highest current emotion intensity when calculating the overall emotion profile similarity, calculates an emotion resonance index by converting the cosine similarity calculation result into a normalized value between 0 and 1, visually displays the emotion resonance index as a percentage bar graph on the user interface, creates a two-way anonymous messaging channel when the user accepts the match, and in the channel A system that controls the application of a privacy protection mechanism that automatically filters personally identifiable information, including names, phone numbers, and addresses, through regular expression-based pattern matching and data masking using hash functions, while still being able to share sentiment data and related context. Claim 4 delete Claim 5 In claim 1, the instructions, when executed by the processor, perform a consent-based access control procedure in which the system specifies data collection items and purposes of use and obtains explicit consent from the user during the process of collecting the user's emotion data; provide a user interface that allows the user to set in detail the scope and level of sharing of emotion data to be shared; perform the emotion profile vectorization process after converting the collected raw emotion data into a form that makes personal identification impossible or difficult by applying at least one differential privacy technique, such as adding noise to the data or grouping the data to prevent individual user information from being revealed; control information exchange according to a consent-based information exchange protocol in which, during user matching, information is selectively shared only when both matched users mutually agree to their respective information disclosure items; monitor conversation content in real time using natural language processing technology to detect inappropriate content according to predefined criteria in an anonymous conversation channel between the matched users; execute immediate intervention and protective measures, including sending a warning message to the user, temporarily restricting message transmission, or forcibly terminating the conversation channel when a risk factor is detected; and all personal identification including the user's emotion data, emotion profile vector, and matching history. A system that controls data security to enhance data security by applying a pre-configured encryption method, including an AES (Advanced Encryption Standard) 256-bit or higher encryption algorithm, to available information stored in a database, and applying a pre-configured access restriction policy. Claim 6 In claim 1, when the instructions are executed by the processor, the system combines location information and emotion data collected from the user to generate a location-based emotion map that visualizes the emotion distribution of users within a specific geographical area, and provides the location-based emotion map to the user to enable them to grasp the overall emotional atmosphere of the surroundings; performs real-time connections between emotionally similar users within a distance range set by the user by using a matching algorithm that combines geographical proximity based on the user's current location and the similarity of emotion profiles according to preset weights; and if it detects that the intensity of a preset negative emotion among the user's emotional states persists above a certain level or that the ratio of negative emotion to positive emotion exceeds a threshold for a specified period, it prioritizes matching users who have experienced similar negative emotion patterns in the past but are confirmed through analysis of emotion record data to have transitioned to a positive emotional state, and groups multiple emotionally similar users according to preset clustering criteria including K-means clustering based on the similarity between emotion profile vectors or the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm, thereby [identifying] common emotional experiences or interests It supports the formation of a shared emotional support community, analyzes the collective emotional data of members of the formed community or group to recommend group activities or related positive psychology content—such as online group meditation sessions, the opening of thematic discussion forums, or collaborative goal-achieving activities—that can help promote the group's collective emotional well-being, and analyzes the trends of change between the time-series patterns of the user's past emotional data and the current emotional state,A system that predicts future emotional states over a specified period through a time-series analysis-based prediction model including a Long Short-Term Memory (LSTM) network, and controls the provision of support services—such as sending pre-prepared customized comforting messages, guiding users to connect with expert counseling, or recommending mood-boosting content—in accordance with predicted emotional changes. Claim 7 In claim 1, when the instructions are executed by the processor, the system performs a personalized recommendation generation function by collecting and analyzing the user's emotion profile and the content consumption history of other matched users in an anonymized form, identifying media content including movies, dramas, and webtoons that users in a specific emotional state have shown a positive response to, literary works including books or poems on a specific theme, and art works including music or digital artwork of a specific genre or mood, and providing them to the user in the form of a personalized recommendation list, performing context analysis that analyzes the statistical correlation between the user's real-time emotion data and external environmental factors including weather information collected from the user's current location, the current season, and major social issues or anniversaries to identify users who exhibit similar emotional responses under specific external environmental conditions, and utilizing this information for matching weight adjustment or content recommendation, and calculating a linguistic compatibility index by analyzing communication styles including the lexical diversity of emotional expressions, the complexity of sentence structures, and the frequency and intensity of positive / negative expressions appearing in text data input by the user through natural language processing technology, and comprehensively considering the calculated linguistic compatibility index together with the emotion profile-based cosine similarity to the communication of A system that controls matching to improve the matching accuracy between users expected to be highly effective, analyzes in real time the emotional value of message content exchanged during the interaction process between the matched users to track changes in each user's emotional state, evaluates the impact of the interaction on quantitative indicators such as changes in each user's emotional score, changes in the frequency of positive emotional expression, or reduction in stress index, measures the effect of promoting emotional well-being through the interaction between the two, and controls recording the results in the memory.

Images