A user behavior prediction method and device, computer equipment and a storage medium

Abstract

The application discloses a user behavior prediction method and device, computer equipment and a storage medium, and relates to the technical field of artificial intelligence and the technical field of financial technology. The application classifies historical short messages based on reading states to obtain historical read short messages and historical unread short messages, filters the historical unread short messages to obtain effective short messages in the historical unread short messages, constructs a training data set based on the historical read short messages and the effective short messages, and trains a user behavior prediction model based on the training data set and a decision tree algorithm, wherein the decision tree algorithm is a LightGBM algorithm, short message data of a user to be predicted is input into the user behavior prediction model, and a behavior prediction result of the user to be predicted is obtained. The application also relates to the technical field of blockchains, and the short message data is stored on a blockchain node. The application can fully combine historical unread short messages to train a user behavior prediction model, reduces waste of short message resources, and improves the accuracy of user behavior prediction.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence technology and financial technology, specifically to a user behavior prediction method, device, computer equipment, and storage medium. Background Technology

[0002] SMS is the lowest-cost way for operators to reach users per message. However, existing user data is insufficient to help operators identify a sufficiently accurate target audience. For example, many SMS messages are sent but users don't click on them, resulting in a significant waste of SMS fees. Under the overarching policy of cost reduction and efficiency improvement, how to save SMS resources while increasing the efficiency of reaching users has become a pressing pain point. Take insurance sales SMS as an example. Insurance companies use SMS to promote insurance products, hoping to reach potential customers, encourage them to click on the message, learn about product details, and make a purchase. By collecting this user behavior data, they aim to predict user preferences. However, because many SMS messages are not clicked or viewed, the available data is limited, making it impossible to accurately predict user preferences and resulting in wasted SMS resources. Summary of the Invention

[0003] The purpose of this application is to propose a user behavior prediction method, apparatus, computer device, and storage medium to solve the technical problems of existing SMS-based user preference prediction schemes, which suffer from limited available data, inability to accurately predict user preferences, and waste of SMS resources.

[0004] To address the aforementioned technical problems, this application provides a user behavior prediction method, which employs the following technical solution: A user behavior prediction method, comprising: Obtain pre-collected historical SMS messages and determine the reading status of the historical SMS messages; Based on the reading status of the historical SMS messages, the historical SMS messages are classified to obtain historical read SMS messages and historical unread SMS messages; Filter the historical unread text messages to obtain the valid text messages from the historical unread text messages; A training dataset is constructed based on the historical read SMS messages and the valid SMS messages, and a user behavior prediction model is trained based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The system receives a user behavior prediction instruction, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the behavior prediction result of the user to be predicted.

[0005] Furthermore, the step of filtering the historical unread text messages to obtain valid text messages from the historical unread text messages specifically includes: Calculate the time difference between the historical read SMS messages and the historical unread SMS messages; The time difference is compared with a preset time threshold, and the unread text messages in the history are filtered out based on the time difference comparison result. Semantic recognition is performed on the historical read SMS messages and the SMS messages that have not expired, and the semantic similarity between the historical read SMS messages and the SMS messages that have not expired is calculated. The semantic similarity is compared with a preset similarity threshold, and the valid SMS messages among the SMS messages that have not expired are selected based on the semantic similarity comparison results.

[0006] Furthermore, the step of calculating the time difference between the historical read SMS messages and the historical unread SMS messages specifically includes: Identify the historical read text message that is closest to the aforementioned historical unread text message to obtain the target historical read text message; Calculate the time difference between the historical unread SMS messages and the target historical read SMS messages.

[0007] Furthermore, the step of performing semantic recognition on the historical read SMS messages and the non-timeout SMS messages respectively, and calculating the semantic similarity between the historical read SMS messages and the non-timeout SMS messages, specifically includes: Semantic recognition is performed on the historical read text messages to obtain a first semantic result; Semantic recognition is performed on the non-timeout SMS messages to obtain a second semantic result; The semantic similarity between the first semantic result and the second semantic result is calculated based on the cosine similarity algorithm.

[0008] Furthermore, the step of comparing the time difference with a preset time threshold and filtering out unread text messages that have not timed out based on the time difference comparison result specifically includes: When the time difference is less than or equal to the time threshold, the historical unread SMS messages are determined to be SMS messages that have not timed out. When the time difference is greater than the time threshold, the historical unread SMS message is determined to be a timed-out SMS message; Filter the historical unread text messages to obtain the text messages that have not expired from the historical unread text messages; The step of comparing the semantic similarity with a preset similarity threshold and filtering valid SMS messages from the non-timeout SMS messages based on the semantic similarity comparison result specifically includes: When the semantic similarity is less than or equal to the similarity threshold, the SMS message that has not expired is determined to be a valid SMS message; When the semantic similarity is greater than the similarity threshold, the SMS message that has not expired is determined to be an invalid SMS message; The non-timeout SMS messages are filtered to obtain the valid SMS messages.

[0009] Furthermore, the training dataset includes a training set and a validation set. The step of constructing a training dataset based on the historical read SMS messages and the valid SMS messages, and training the user behavior prediction model based on the training dataset and a preset decision tree algorithm, specifically includes: The text messages in the training set are sorted in chronological order. The training set, sorted by completion time, is subjected to feature engineering to obtain the user preference features corresponding to the training set. A gradient decision tree is constructed based on the LightGBM algorithm and the user preference features corresponding to the training set. Combine the gradient decision trees to generate a user behavior prediction model; The user behavior prediction model is validated using the validation set, and a user behavior prediction model that passes the performance validation is output.

[0010] Furthermore, the steps of receiving user behavior prediction instructions, reading SMS data of the user to be predicted, inputting the SMS data into the user behavior prediction model, and obtaining the behavior prediction result of the user to be predicted specifically include: Responding to user behavior prediction commands, check SMS reading permissions; When the SMS reading permission check is passed, the SMS mailbox of the user to be predicted is read to obtain the SMS data of the user to be predicted; Identify read and unread SMS messages in the SMS data of the user to be predicted, and filter out valid SMS messages from the unread SMS messages; The valid text messages from the read and unread text messages are input into the user behavior prediction model to obtain the behavior prediction result of the user to be predicted.

[0011] To address the aforementioned technical problems, this application also provides a user behavior prediction device, which employs the following technical solution: A user behavior prediction device, comprising: The SMS collection module is used to acquire pre-collected historical SMS messages and determine the reading status of the historical SMS messages; The SMS classification module is used to classify the historical SMS messages based on their reading status to obtain historical read SMS messages and historical unread SMS messages; The SMS filtering module is used to filter the historical unread SMS messages to obtain the valid SMS messages from the historical unread SMS messages; The model training module is used to construct a training dataset based on the historical read SMS messages and the valid SMS messages, and to train a user behavior prediction model based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The behavior prediction module is used to receive user behavior prediction instructions, read SMS data of the user to be predicted, input the SMS data into the user behavior prediction model, and obtain the behavior prediction result of the user to be predicted.

[0012] To address the aforementioned technical problems, this application also provides a computer device that employs the following technical solution: A computer device includes a memory and a processor, the memory storing computer-readable instructions, the processor executing the computer-readable instructions to implement the steps of the user behavior prediction method as described in any of the preceding claims.

[0013] To address the aforementioned technical problems, this application also provides a computer-readable storage medium, employing the technical solution described below: A computer-readable storage medium storing computer-readable instructions that, when executed by a processor, implement the steps of the user behavior prediction method as described in any one of the preceding descriptions.

[0014] Compared with the prior art, the embodiments of this application have the following main advantages: This application discloses a user behavior prediction method, apparatus, computer device, and storage medium, relating to the fields of artificial intelligence and fintech. This application acquires pre-collected historical SMS messages and determines their reading status. Based on the reading status, the historical SMS messages are classified into read and unread messages. Unread messages are then filtered to identify valid messages. A training dataset is constructed based on the read and valid messages. A user behavior prediction model is trained using this training dataset and a pre-defined decision tree algorithm, specifically the LightGBM algorithm. The model receives user behavior prediction instructions, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the predicted behavior result for the user. By filtering unread messages to identify valid messages, constructing a training dataset using read and valid messages, and combining this with the LightGBM algorithm to train the user behavior prediction model, this application can fully utilize the characteristics of unread messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction. Attached Figure Description

[0015] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 An exemplary system architecture diagram is shown, in which this application can be applied; Figure 2 A flowchart of one embodiment of the user behavior prediction method according to this application is shown; Figure 3 A schematic diagram of the structure of one embodiment of the user behavior prediction device according to this application is shown; Figure 4 A schematic diagram of the structure of one embodiment of a computer device according to this application is shown. Detailed Implementation

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0018] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0020] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0021] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0022] Terminal devices 101, 102, and 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 players (Moving Picture Experts Group Audio Layer IV), laptops, and desktop computers, etc.

[0023] Server 105 can be a server that provides various services, such as a backend server that supports the pages displayed on terminal devices 101, 102, and 103. The server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0024] It should be noted that the user behavior prediction method provided in this application embodiment is generally executed by a server, and correspondingly, the user behavior prediction device is generally set in the server.

[0025] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0026] Continue to refer to Figure 2 The diagram illustrates a flowchart of an embodiment of the user behavior prediction method according to this application. Embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0027] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0028] Many current SMS-based user behavior prediction solutions use training data from SMS messages that users have already clicked and read. However, many SMS messages are sent but users don't click on them, resulting in a significant waste of SMS resources. Under the overarching policy of cost reduction and efficiency improvement, how to save SMS resources while simultaneously increasing user reach efficiency has become a pressing pain point. Take insurance sales SMS messages as an example. Insurance companies use SMS to promote insurance products, hoping to reach potential customers, encourage them to click on the message, learn about product details, and make a purchase. By collecting this user behavior data, they aim to predict user preferences. However, because many SMS messages are not clicked or viewed, the available data is limited, making it impossible to accurately predict user preferences and resulting in wasted SMS resources.

[0029] To address the aforementioned technical problems, this application discloses a user behavior prediction method, apparatus, computer device, and storage medium, relating to the fields of artificial intelligence and fintech. The main idea of ​​this application is to filter unread SMS messages to obtain valid SMS messages, combine read SMS messages and valid SMS messages to construct a training dataset, and then use the LightGBM algorithm to train a user behavior prediction model. This approach can fully incorporate the features of unread SMS messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction.

[0030] The user behavior prediction method includes the following steps: S201, retrieve the pre-collected historical SMS messages and determine the reading status of the historical SMS messages.

[0031] In this embodiment, the server first needs to collect some historical SMS messages for training the model and store these historical SMS messages in the database. When collecting SMS messages, it is necessary to determine the reading status of each SMS message, which includes read and unread.

[0032] S202, classify historical SMS messages based on their reading status to obtain historical read SMS messages and historical unread SMS messages.

[0033] In this embodiment, historical SMS messages are divided into read SMS messages and unread SMS messages based on their reading status, namely, historical read SMS messages and historical unread SMS messages.

[0034] S203, filter the historical unread text messages to obtain the valid text messages from the historical unread text messages.

[0035] In this embodiment, some unread text messages may contain invalid messages (such as spam messages), requiring filtering to obtain valid unread messages. The server categorizes historical text messages and filters historical unread messages to obtain valid messages from the historical unread messages.

[0036] Further, the steps for filtering unread text messages to obtain valid messages from the unread text messages include: Calculate the time difference between historical read SMS messages and historical unread SMS messages; The time difference is compared with a preset time threshold, and the unread text messages that have not expired are filtered out based on the time difference comparison results. Semantic recognition is performed on historical read SMS messages and SMS messages that have not expired, and the semantic similarity between historical read SMS messages and SMS messages that have not expired is calculated. The semantic similarity is compared with a preset similarity threshold, and the valid SMS messages among the SMS messages that have not expired are selected based on the semantic similarity comparison results.

[0037] In this embodiment, the time difference between historical read SMS messages and historical unread SMS messages is first calculated, i.e., the time interval between read and unread SMS messages. The calculated time difference is compared with a preset time threshold, and unread SMS messages whose time difference does not exceed the threshold are selected, i.e., messages that have not timed out. Semantic recognition is then performed on the selected historical read SMS messages and messages that have not timed out, i.e., the SMS content is analyzed and understood to extract key information and features. Semantic analysis is then performed on historical read SMS messages and messages that have not timed out to calculate their semantic similarity, i.e., to determine the degree of similarity between the SMS content. The calculated semantic similarity is compared with a preset similarity threshold, and messages that have not timed out and whose semantic similarity exceeds the threshold are selected, i.e., valid SMS messages.

[0038] Through the above steps, historical unread SMS messages were filtered twice: time difference filtering and semantic filtering, to obtain the valid SMS messages. This process mainly relies on the calculation and comparison of time difference and semantic similarity to determine which unread SMS messages are valid and can be further used to train user behavior prediction models.

[0039] Furthermore, the steps for calculating the time difference between historical read SMS messages and historical unread SMS messages specifically include: Identify the historical read text message that is closest to the historical unread text message and obtain the target historical read text message; Calculate the time difference between historical unread text messages and target historical read text messages.

[0040] In this embodiment, the first step is to find the historical read SMS message that is closest to the unread SMS message in terms of time. This can be achieved by comparing the timestamps of the SMS messages and selecting the read SMS message with the closest time as the target historical read SMS message. Once the target historical read SMS message is determined, the time difference between the historical unread SMS message and the target historical read SMS message can be calculated. This can be achieved by subtracting the two timestamps to obtain the time difference.

[0041] By following the steps above, the time difference between historical read SMS messages and historical unread SMS messages can be calculated. This process mainly relies on comparing and subtracting timestamps to determine the time difference between an unread SMS message and the closest read SMS message.

[0042] Furthermore, the steps of performing semantic recognition on historical read SMS messages and non-timeout SMS messages respectively, and calculating the semantic similarity between historical read SMS messages and non-timeout SMS messages, specifically include: Semantic recognition is performed on historical read text messages to obtain the first semantic result; Semantic recognition is performed on SMS messages that have not expired to obtain a second semantic result; The semantic similarity between the first and second semantic results is calculated based on the cosine similarity algorithm.

[0043] In this embodiment, the first step is to perform semantic recognition on the content of historical read SMS messages, i.e., analyze and understand the text to extract key information and features. This can be achieved using Natural Language Processing (NLP) techniques, such as text segmentation, part-of-speech tagging, and named entity recognition. The final result is the first semantic representation of the historical read SMS messages. Similarly, semantic recognition is needed on the content of SMS messages that have not expired to obtain a second semantic representation. The purpose of this step is to compare the unexpired SMS messages with the historical read SMS messages to determine their semantic similarity. After obtaining the semantic representations of the historical read SMS messages and the unexpired SMS messages, the cosine similarity algorithm can be used to calculate their semantic similarity.

[0044] Cosine similarity is a commonly used method to measure the similarity between two vectors. It can be used to compare the degree of similarity between two semantic representations. The closer the calculated semantic similarity value is to 1, the more similar the two texts are semantically; the closer the value is to 0, the greater the semantic difference.

[0045] By following the steps above, the semantic similarity between historical read SMS messages and SMS messages that have not expired can be calculated. This process mainly relies on semantic recognition and cosine similarity calculation to determine the degree of semantic similarity between the two texts.

[0046] Furthermore, the step of comparing the time difference with a preset time threshold and filtering out unread SMS messages that have not yet timed out based on the time difference comparison result specifically includes: When the time difference is less than or equal to the time threshold, unread text messages in the past are identified as text messages that have not timed out. When the time difference is greater than the time threshold, unread text messages in the past will be identified as timed-out text messages. Filter the historical unread text messages to get the text messages that have not expired; In this embodiment, the calculated time difference is compared with a preset time threshold in this step. If the time difference is less than or equal to the time threshold, it means that the historical unread SMS is within the threshold range, i.e., it has not timed out. Therefore, these historical unread SMS will be identified as non-timeout SMS. Conversely, if the time difference is greater than the time threshold, it means that the historical unread SMS has exceeded the preset time threshold, i.e., it has timed out. Therefore, these historical unread SMS will be identified as timeout SMS. Based on the above comparison results, the historical unread SMS are filtered, and those historical unread SMS identified as non-timeout SMS are selected as the final result.

[0047] By following the steps above, we can filter out unread SMS messages that have not expired from the historical unread SMS messages by comparing the time difference with the time threshold. This process mainly relies on comparing the time difference and judging the comparison results to determine which unread SMS messages have not expired.

[0048] Furthermore, the step of comparing semantic similarity with a preset similarity threshold and filtering valid SMS messages from those that have not expired based on the semantic similarity comparison results specifically includes: When the semantic similarity is less than or equal to the similarity threshold, the SMS message that has not expired is determined as a valid SMS message; When the semantic similarity is greater than the similarity threshold, the SMS message that has not expired will be identified as an invalid SMS message; Filter the SMS messages that have not expired to obtain the valid ones.

[0049] In this embodiment, the calculated semantic similarity is compared with a preset similarity threshold in this step. If the semantic similarity is less than or equal to the similarity threshold, it means that the semantic difference between the non-timeout SMS and the historical read SMS is large, i.e., the semantic similarity between the two is low. Therefore, these non-timeout SMS will be determined as invalid SMS. Conversely, if the semantic similarity is greater than the similarity threshold, it means that the semantic difference between the non-timeout SMS and the historical read SMS is small, i.e., the semantic similarity between the two is high. Therefore, these non-timeout SMS will be determined as valid SMS. Based on the above comparison results, the non-timeout SMS are filtered, and the non-timeout SMS determined as valid SMS are selected as the final result.

[0050] By following the steps above, valid SMS messages can be filtered out from those that have not expired by comparing semantic similarity with a similarity threshold. This process mainly relies on comparing semantic similarity and judging the comparison results to determine which of the valid SMS messages are valid.

[0051] S204: Construct a training dataset based on historical read SMS messages and valid SMS messages, and train a user behavior prediction model based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm.

[0052] In this embodiment, historical read SMS messages and valid SMS messages are used as training datasets to train the user behavior prediction model. A preset decision tree algorithm (LightGBM algorithm) is used to train the training dataset to build the user behavior prediction model.

[0053] LightGBM (Light Gradient Boosting Machine) is a machine learning algorithm based on the gradient boosting framework. It is a fast and efficient gradient boosting decision tree algorithm. LightGBM is designed to provide an efficient gradient boosting framework capable of handling large-scale datasets and achieving a good balance between accuracy and training speed. LightGBM has been widely used in machine learning competitions and practical applications, especially suitable for handling large-scale datasets and scenarios requiring efficient training. It can be used for regression and classification problems and supports various extended functions, such as handling missing values ​​and categorical features.

[0054] Furthermore, the training dataset includes a training set and a validation set. The training dataset is constructed based on historical read SMS messages and valid SMS messages. The steps for training the user behavior prediction model based on the training dataset and a pre-defined decision tree algorithm specifically include: Sort the text messages in the training set in chronological order; Feature engineering is performed on the training set after completion time sorting to obtain the user preference features corresponding to the training set. A gradient decision tree is constructed based on the LightGBM algorithm and the user preference features corresponding to the training set. Combine gradient decision trees to generate a user behavior prediction model; The user behavior prediction model is validated using a validation set, and the user behavior prediction model that passes the performance validation is output.

[0055] In this embodiment, the SMS messages in the training set are sorted by completion time to ensure the correct chronological order of the training dataset. Feature engineering is then performed on the sorted training set. Feature engineering refers to constructing new features based on the original data to improve model performance. In this approach, user preference features, such as the user's preference for a certain type of SMS message, can be constructed based on information such as SMS content and sender in the training set. Multiple gradient decision trees are constructed using the LightGBM algorithm and the corresponding user preference features from the training set. These multiple gradient decision trees are then combined to generate a user behavior prediction model. Combining multiple decision trees improves the model's accuracy and robustness. The trained user behavior prediction model is then validated using a validation set. By comparing the model's predictions on the validation set with the actual results, the model's accuracy and generalization ability are evaluated. If the model passes the performance validation, meaning the predictions match the actual results, then the model can be output and applied to real-world scenarios.

[0056] Through the above steps, a training dataset can be built based on historical read SMS messages and valid SMS messages, and a user behavior prediction model can be trained using a preset decision tree algorithm. This process mainly relies on feature engineering and decision tree algorithm training to generate a user behavior prediction model with high accuracy and strong generalization ability.

[0057] In one specific embodiment of this application, firstly, a dataset for training the model is collected and prepared. The dataset should include relevant information on historical read SMS messages and valid SMS messages, as well as corresponding target variables representing user behavioral preferences (e.g., 1 for liking, 0 for disliking). Then, the data is preprocessed, including handling missing values, outlier handling, and feature normalization. If missing values ​​exist, they can be filled or deleted. Furthermore, features can be normalized or standardized as needed to ensure they are within a similar numerical range. Next, feature engineering is performed on the dataset to extract features related to user behavioral preferences, including textual features of SMS content (e.g., bag-of-words model, TF-IDF encoding, word embeddings, etc.), time features, sender features, SMS length, etc. The goal of feature engineering is to extract informative features from the raw data that describe user preferences. The dataset is divided into a training set and a validation set. The training set is used for model training, and the validation set is used for model evaluation and tuning. Typically, 70% to 80% of the data is used as the training set, and the remaining data as the validation set. The LightGBM algorithm is used for model training. During training, model performance can be optimized by adjusting hyperparameters (such as tree depth, learning rate, and number of leaf nodes). The trained model is evaluated using a validation set, and further tuning is performed based on the evaluation results. Metrics such as accuracy, precision, recall, and F1 score can be calculated to evaluate model performance. If the model performance is unsatisfactory, strategies such as feature selection and hyperparameter tuning can be adjusted as needed. Finally, the trained model is used to predict behavior on new SMS data. The SMS data to be predicted is input into the model, and the model outputs the corresponding behavior prediction result, representing the probability of the user's preference for the SMS message.

[0058] S205: Receive user behavior prediction instruction, read SMS data of the user to be predicted, input the SMS data into the user behavior prediction model, and obtain the behavior prediction result of the user to be predicted.

[0059] In this embodiment, a user behavior prediction instruction is received, and the SMS data of the user to be predicted is read. The SMS data of the user to be predicted is then input into a trained user behavior prediction model, and the user's behavior is predicted by the model.

[0060] Furthermore, the steps of receiving user behavior prediction instructions, reading SMS data of the user to be predicted, inputting the SMS data into the user behavior prediction model, and obtaining the behavior prediction result of the user to be predicted specifically include: Responding to user behavior prediction commands, check SMS reading permissions; When the SMS reading permission check is passed, the SMS mailbox of the user to be predicted is read to obtain the SMS data of the user to be predicted; Identify read and unread SMS messages in the SMS data of the user to be predicted, and filter out the valid SMS messages from the unread SMS messages; The valid text messages from read and unread text messages are input into the user behavior prediction model to obtain the behavior prediction results for the user to be predicted.

[0061] In this embodiment, during this step, the server needs to check whether the user has permission to read SMS messages. If the user does not have permission, the system cannot read the SMS data of the user to be predicted, and therefore cannot perform behavior prediction. After the permission test is passed, the server reads the SMS inbox of the user to be predicted and obtains the SMS data of the user to be predicted. This SMS data will be used as input data for behavior prediction. The SMS data of the user to be predicted is then identified and filtered. First, the system needs to identify read and unread SMS messages. Then, the system needs to filter unread SMS messages, selecting only valid SMS messages as input data. Valid SMS messages can be filtered according to preset conditions, and the same processing method as the historical SMS filtering scheme can be used for the SMS data of the user to be predicted. The valid SMS messages from the read and unread SMS messages are used as input data and input into the trained user behavior prediction model. The model will calculate and analyze based on the input data to obtain the behavior prediction result for the user to be predicted. This result can represent the probability of the user's preference for SMS messages.

[0062] In summary, based on user behavior prediction instructions, the system can obtain corresponding behavior prediction results by reading the SMS data of the user to be predicted and inputting it into a trained model. This result can help the system understand the user's SMS preferences, thereby enabling personalized recommendations or other related operations.

[0063] In one specific embodiment of this application, useful information is extracted from unread and unviewed SMS messages by filtering out valid messages. In an insurance sales scenario, valid messages can include information such as the user's level of interest in the insurance product, the number of times the message was clicked, and the time it was read. Filtering valid messages reduces the waste of SMS resources and avoids sending invalid messages to users. Then, valid messages from read and unread SMS messages are combined to construct a training dataset. Read messages can include those that the user has clicked and viewed, and the information in these messages can reflect the user's behavioral preferences and purchase intentions. Valid messages from unread SMS messages can provide more information about the user's behavior and interests. By combining valid messages from read and unread SMS messages, a more comprehensive understanding of user behavioral characteristics can be achieved, improving the accuracy of user behavior prediction. Finally, the LightGBM algorithm is used to train the training dataset to obtain a user behavior prediction model. Training the user behavior prediction model using the LightGBM algorithm can better capture user behavior patterns and preferences, thereby improving the efficiency of reaching users.

[0064] In the scenario of promoting insurance products via SMS, by filtering valid SMS messages and using the LightGBM algorithm for model training, SMS resources can be saved while improving the efficiency of reaching users. By making full use of the effective information in unread SMS messages, user behavioral preferences can be predicted more accurately, thereby improving the sales effect of insurance products.

[0065] In the above embodiments, this application discloses a user behavior prediction method, relating to the fields of artificial intelligence and fintech. This application acquires pre-collected historical SMS messages and determines their reading status. Based on the reading status, the historical SMS messages are classified into read and unread messages. Unread messages are then filtered to obtain valid messages. A training dataset is constructed based on the read and valid messages. A user behavior prediction model is trained using the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The model receives user behavior prediction instructions, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the predicted behavior result for the user. By filtering unread messages to obtain valid messages, constructing a training dataset using read and valid messages, and combining it with the LightGBM algorithm to train the user behavior prediction model, this application can fully utilize the characteristics of unread messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction.

[0066] In this embodiment, the user behavior prediction method runs on an electronic device (e.g., Figure 1The server shown can acquire data or receive commands via wired or wireless connections. It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wideband) connections, and other currently known or future wireless connection methods.

[0067] It should be emphasized that, to further ensure the privacy and security of the aforementioned SMS data, the SMS data can also be stored in a blockchain node.

[0068] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0069] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When executed, the computer-readable instructions can include the processes of the embodiments of the above methods. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0070] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0071] Further reference Figure 3 As a response to the above Figure 2 The present application provides an embodiment of a user behavior prediction device to implement the method shown, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0072] like Figure 3 As shown, the user behavior prediction device 300 described in this embodiment includes: SMS collection module 301 is used to acquire pre-collected historical SMS messages and determine the reading status of historical SMS messages; SMS classification module 302 is used to classify historical SMS messages based on their reading status to obtain historical read SMS messages and historical unread SMS messages; SMS filtering module 303 is used to filter historical unread SMS messages to obtain valid SMS messages from the historical unread SMS messages; The model training module 304 is used to construct a training dataset based on historical read SMS messages and valid SMS messages, and to train a user behavior prediction model based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The behavior prediction module 305 is used to receive user behavior prediction instructions, read the SMS data of the user to be predicted, input the SMS data into the user behavior prediction model, and obtain the behavior prediction result of the user to be predicted.

[0073] Furthermore, the SMS filtering module 303 specifically includes: The time difference calculation unit is used to calculate the time difference between historical read SMS messages and historical unread SMS messages; The time difference comparison unit is used to compare the time difference with a preset time threshold and filter out unread SMS messages that have not expired based on the time difference comparison result. The semantic similarity calculation unit is used to perform semantic recognition on historical read SMS messages and SMS messages that have not expired, and to calculate the semantic similarity between historical read SMS messages and SMS messages that have not expired. The semantic similarity comparison unit is used to compare semantic similarity with a preset similarity threshold, and to filter valid SMS messages from those that have not expired based on the semantic similarity comparison results.

[0074] Furthermore, the time difference calculation unit specifically includes: The target SMS identification subunit is used to identify the historical read SMS that is closest to the historical unread SMS, and to obtain the target historical read SMS. The time difference calculation subunit is used to calculate the time difference between historical unread SMS messages and target historical read SMS messages.

[0075] Furthermore, the semantic similarity calculation unit specifically includes: The first semantic recognition subunit is used to perform semantic recognition on historical read SMS messages to obtain the first semantic result; The second semantic recognition subunit is used to perform semantic recognition on SMS messages that have not expired, and obtain the second semantic result. The semantic similarity calculation subunit is used to calculate the semantic similarity between the first semantic result and the second semantic result based on the cosine similarity algorithm.

[0076] Furthermore, the time difference comparison unit specifically includes: The first time difference comparison result sub-unit is used to determine historical unread SMS messages as SMS messages that have not timed out when the time difference is less than or equal to the time threshold. The second time difference comparison result subunit is used to determine historical unread SMS messages as timed-out SMS messages when the time difference is greater than the time threshold. The first SMS filtering subunit is used to filter historical unread SMS messages to obtain SMS messages that have not expired from the historical unread SMS messages; Furthermore, the semantic similarity comparison unit specifically includes: The first semantic similarity comparison result subunit is used to determine the SMS that has not expired as a valid SMS when the semantic similarity is less than or equal to the similarity threshold. The second semantic similarity comparison result subunit is used to determine the SMS that has not expired as invalid SMS when the semantic similarity is greater than the similarity threshold; The second SMS filtering subunit is used to filter SMS messages that have not expired, and obtain the valid SMS messages among the SMS messages that have not expired.

[0077] Furthermore, the training dataset includes a training set and a validation set, and the model training module 304 specifically includes: The SMS sorting unit is used to sort the SMS messages in the training set according to time order. The feature engineering unit is used to perform feature engineering on the training set after it is sorted by completion time to obtain the user preference features corresponding to the training set. Gradient decision tree building unit, used to build gradient decision trees based on the LightGBM algorithm and the user preference features corresponding to the training set; Decision tree model building unit, used to combine gradient decision trees to generate user behavior prediction models; The performance verification unit is used to perform performance verification on the user behavior prediction model using a verification set, and outputs a user behavior prediction model that passes the performance verification.

[0078] Furthermore, the behavior prediction module 305 specifically includes: The permission confirmation unit is used to respond to user behavior prediction commands and detect SMS reading permissions; The SMS reading unit is used to read the SMS mailbox of the user to be predicted and obtain the SMS data of the user to be predicted when the SMS reading permission detection is passed; The SMS classification unit is used to identify read and unread SMS messages in the SMS data of the user to be predicted, and to filter out valid SMS messages from the unread SMS messages; The behavior prediction unit is used to input valid text messages from read and unread text messages into the user behavior prediction model to obtain the behavior prediction results of the user to be predicted.

[0079] In the above embodiments, this application discloses a user behavior prediction device, relating to the fields of artificial intelligence and fintech. This application acquires pre-collected historical SMS messages and determines their reading status. Based on the reading status, the historical SMS messages are classified into read and unread messages. Unread messages are then filtered to obtain valid messages. A training dataset is constructed based on the read and valid messages. A user behavior prediction model is trained using the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The device receives user behavior prediction instructions, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the predicted behavior result for the user. By filtering unread messages to obtain valid messages, constructing a training dataset using read and valid messages, and combining it with the LightGBM algorithm to train the user behavior prediction model, this application can fully utilize the characteristics of unread messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction.

[0080] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0081] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0082] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0083] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 4. Of course, the memory 41 may also include both the internal storage unit and its external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for user behavior prediction methods. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0084] In some embodiments, the processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions stored in the memory 41 or to process data, such as executing computer-readable instructions for the user behavior prediction method.

[0085] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.

[0086] In the above embodiments, this application discloses a computer device relating to the fields of artificial intelligence and fintech. This application acquires pre-collected historical SMS messages and determines their reading status. Based on the reading status, the historical SMS messages are categorized into read and unread messages. Unread messages are then filtered to identify valid messages. A training dataset is constructed based on the read and valid messages. A user behavior prediction model is trained using this training dataset and a preset decision tree algorithm, specifically the LightGBM algorithm. The model receives user behavior prediction instructions, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the predicted behavior result for the user. By filtering unread messages to identify valid messages, constructing a training dataset using read and valid messages, and combining this with the LightGBM algorithm to train the user behavior prediction model, this application can fully utilize the characteristics of unread messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction.

[0087] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the user behavior prediction method described above.

[0088] In the above embodiments, this application discloses a computer-readable storage medium, relating to the fields of artificial intelligence and fintech. This application acquires pre-collected historical SMS messages and determines their reading status. Based on the reading status, the historical SMS messages are classified into read and unread messages. Unread messages are then filtered to obtain valid messages. A training dataset is constructed based on the read and valid messages. A user behavior prediction model is trained using the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The model receives user behavior prediction instructions, reads the SMS data of the user to be predicted, inputs the SMS data into the user behavior prediction model, and obtains the predicted behavior result for the user. By filtering unread messages to obtain valid messages, constructing a training dataset using read and valid messages, and combining it with the LightGBM algorithm to train the user behavior prediction model, this application can fully utilize the characteristics of unread messages to train the user behavior prediction model, reducing SMS resource waste and improving the accuracy of user behavior prediction.

[0089] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0090] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0091] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A user behavior prediction method, characterized by, include: Obtain pre-collected historical SMS messages and determine the reading status of the historical SMS messages; Based on the reading status of the historical SMS messages, the historical SMS messages are classified to obtain historical read SMS messages and historical unread SMS messages; Filter the historical unread text messages to obtain the valid text messages from the historical unread text messages; A training dataset is constructed based on the historical read SMS messages and the valid SMS messages, and a user behavior prediction model is trained based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. Receive user behavior prediction instructions, read SMS data of the user to be predicted, input the SMS data into the user behavior prediction model, and obtain the behavior prediction result of the user to be predicted. The step of filtering the historical unread text messages to obtain the valid text messages from the historical unread text messages specifically includes: Calculate the time difference between the historical read SMS messages and the historical unread SMS messages; The time difference is compared with a preset time threshold, and the unread text messages in the history are filtered out based on the time difference comparison result. Semantic recognition is performed on the historical read SMS messages and the SMS messages that have not expired, and the semantic similarity between the historical read SMS messages and the SMS messages that have not expired is calculated. The semantic similarity is compared with a preset similarity threshold, and the valid SMS messages among the SMS messages that have not expired are selected based on the semantic similarity comparison results; The step of comparing the semantic similarity with a preset similarity threshold and filtering valid SMS messages from the non-timeout SMS messages based on the semantic similarity comparison result specifically includes: When the semantic similarity is greater than the similarity threshold, the SMS message that has not expired is determined to be a valid SMS message; When the semantic similarity is less than or equal to the similarity threshold, the SMS message that has not expired is determined to be an invalid SMS message; The non-timeout SMS messages are filtered to obtain the valid SMS messages.

2. The user behavior prediction method of claim 1, wherein, The step of calculating the time difference between the historical read SMS messages and the historical unread SMS messages specifically includes: Identify the historical read text message that is closest to the aforementioned historical unread text message to obtain the target historical read text message; Calculate the time difference between the historical unread SMS messages and the target historical read SMS messages.

3. The user behavior prediction method of claim 1, wherein, The step of performing semantic recognition on the historical read SMS messages and the SMS messages that have not expired, and calculating the semantic similarity between the historical read SMS messages and the SMS messages that have not expired, specifically includes: Semantic recognition is performed on the historical read text messages to obtain a first semantic result; Semantic recognition is performed on the non-timeout SMS messages to obtain a second semantic result; The semantic similarity between the first semantic result and the second semantic result is calculated based on the cosine similarity algorithm.

4. The user behavior prediction method of claim 1, wherein, The step of comparing the time difference with a preset time threshold and filtering out unread SMS messages that have not timed out based on the time difference comparison result specifically includes: When the time difference is less than or equal to the time threshold, the historical unread SMS messages are determined to be SMS messages that have not timed out. When the time difference is greater than the time threshold, the historical unread SMS message is determined to be a timed-out SMS message; The historical unread text messages are filtered to obtain the text messages that have not expired from the historical unread text messages.

5. The user behavior prediction method as described in any one of claims 1 to 4, characterized in that, The training dataset includes a training set and a validation set. The step of constructing the training dataset based on the historical read SMS messages and the valid SMS messages, and training the user behavior prediction model based on the training dataset and a preset decision tree algorithm, specifically includes: The text messages in the training set are sorted in chronological order. The training set, sorted by completion time, is subjected to feature engineering to obtain the user preference features corresponding to the training set. A gradient decision tree is constructed based on the LightGBM algorithm and the user preference features corresponding to the training set. Combine the gradient decision trees to generate a user behavior prediction model; The user behavior prediction model is validated using the validation set, and a user behavior prediction model that passes the performance validation is output.

6. The user behavior prediction method as described in claim 5, characterized in that, The steps of receiving user behavior prediction instructions, reading SMS data of the user to be predicted, inputting the SMS data into the user behavior prediction model, and obtaining the behavior prediction result of the user to be predicted specifically include: Responding to user behavior prediction commands, check SMS reading permissions; When the SMS reading permission check is passed, the SMS mailbox of the user to be predicted is read to obtain the SMS data of the user to be predicted; Identify read and unread SMS messages in the SMS data of the user to be predicted, and filter out valid SMS messages from the unread SMS messages; The valid text messages from the read and unread text messages are input into the user behavior prediction model to obtain the behavior prediction result of the user to be predicted.

7. A user behavior prediction device, characterized in that, The user behavior prediction device implements the steps of the user behavior prediction method as described in any one of claims 1 to 6, and the user behavior prediction device comprises: The SMS collection module is used to acquire pre-collected historical SMS messages and determine the reading status of the historical SMS messages; The SMS classification module is used to classify the historical SMS messages based on their reading status to obtain historical read SMS messages and historical unread SMS messages; The SMS filtering module is used to filter the historical unread SMS messages to obtain the valid SMS messages from the historical unread SMS messages; The model training module is used to construct a training dataset based on the historical read SMS messages and the valid SMS messages, and to train a user behavior prediction model based on the training dataset and a preset decision tree algorithm, wherein the preset decision tree algorithm is the LightGBM algorithm. The behavior prediction module is used to receive user behavior prediction instructions, read SMS data of the user to be predicted, input the SMS data into the user behavior prediction model, and obtain the behavior prediction result of the user to be predicted.

8. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the user behavior prediction method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the user behavior prediction method as described in any one of claims 1 to 6.

Images