Data processing method and device, equipment, storage medium and product

Abstract

The invention belongs to the technical field of cloud computing, and discloses a data processing method and device, equipment, a storage medium and a product. According to the method, intention recognition and entity extraction are performed on the user input content to obtain the recognition result and the target extraction result, then the space-time triple is constructed according to the base station zipper access table, and then the user dynamic representation and the region dynamic representation are determined according to the space-time triple. And outputting chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result. According to the method, the identification result and the target extraction result corresponding to the user input content can be obtained, and then the dynamic representation of the user and the region is accurately captured, so that the problems of insufficient spatio-temporal data processing and limitation of field application are effectively solved; and the chart data is output according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result, so that visual chart data display can be provided for the user, and the user experience is enhanced.

Description

Technical Field

[0001] The present application relates to the field of cloud computing technology, and in particular to a data processing method, device, equipment, storage medium and product. Background Art

[0002] At present, in the field of intelligent dialogue systems, the shortcomings can be mainly attributed to the following points: (1) Insufficient spatiotemporal data processing: When processing queries involving spatiotemporal characteristics, there is a lack of in-depth understanding and analysis capabilities. (2) Domain limitation: It is mainly designed for general dialogue scenarios and lacks professional support and optimization for specific fields. (3) Interaction mode limitation: It is mainly designed for telephone channels and has limited processing capabilities for spatiotemporal queries of non-voice input. (4) Real-time requirements: The real-time requirements of telephone interactions may limit the in-depth analysis and processing of complex spatiotemporal queries. (5) Lack of context fusion: There are limitations in fusing spatiotemporal context information. (6) Insufficient scalability: It is mainly aimed at tracking a single dialogue state and lacks the ability to adapt and expand to changing application scenarios. Therefore, how to effectively solve the challenges of spatiotemporal data processing, provide users with intuitive chart data display, and enhance user experience has become an urgent problem to be solved. Summary of the invention

[0003] The main purpose of this application is to provide a data processing method, device, equipment, storage medium and product, aiming to solve the technical problem of how to effectively solve the challenges of spatiotemporal data processing, provide users with intuitive chart data display, and enhance user experience.

[0004] To achieve the above purpose, the present application provides a data processing method, which comprises the following steps: Perform intent recognition and entity extraction on user input content to obtain recognition results and target extraction results; Constructing space-time triples based on the base station zipper entry and exit table; Determine a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet; Outputting chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

[0005] Optionally, the step of performing intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results specifically includes: Input the user input content into the pre-trained model to obtain a pre-trained vector; Inputting the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result; The pre-trained vector is input into the DIET model to perform entity extraction on the user input content to obtain a target extraction result.

[0006] Optionally, the step of inputting the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain a target extraction result specifically includes: Input the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain an initial extraction result, wherein the initial extraction result includes at least one of location, user portrait, time, point of interest, and indicator; The initial extraction result is standardized to obtain a target extraction result.

[0007] Optionally, the step of constructing a time-space triplet according to the base station zipper entry-exit table specifically includes: Extract key fields and original base station data corresponding to the key fields from the base station zipper access table; Performing data cleaning on the original base station data to obtain cleaned base station data; Eliminating outliers from the cleaned data to obtain base station data after elimination; A space-time triplet is constructed according to the data format corresponding to the removed base station data.

[0008] Optionally, the step of determining the user dynamic representation and the region dynamic representation according to the spatiotemporal triplet specifically includes: Training the initial neural network model according to the spatiotemporal triplet to obtain a target network model, and outputting an initial user dynamic representation and an initial region dynamic representation; The initial user dynamic representation and the initial region dynamic representation are updated to obtain a target user dynamic representation and a target region dynamic representation.

[0009] Optionally, the step of updating the initial user dynamic representation and the initial region dynamic representation to obtain the target user dynamic representation and the target region dynamic representation specifically includes: Obtain new space-time triples corresponding to each moment in a preset time period; Inputting each new spatiotemporal triple into the target network model to obtain a new user dynamic representation and a new region dynamic representation; Determine the optimal time according to the new user dynamic representation and the new area dynamic representation; The initial user dynamic representation and the initial region dynamic representation are updated according to the optimal user dynamic representation and the optimal region dynamic representation corresponding to the optimal time, so as to obtain the target user dynamic representation and the target region dynamic representation.

[0010] In addition, to achieve the above-mentioned purpose, the present application also provides a data processing device, the data processing device comprising: The content processing module is used to perform intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results; A space-time triplet construction module, used to construct a space-time triplet according to a base station zipper entry and exit table; A dynamic representation determination module, used to determine a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet; The chart data determination module is used to output chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

[0011] In addition, to achieve the above-mentioned purpose, the present application also proposes a data processing device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the data processing method described above.

[0012] In addition, to achieve the above objectives, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the data processing method described above are implemented.

[0013] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the data processing method described above are implemented.

[0014] This application performs intent recognition and entity extraction on user input content to obtain recognition results and target extraction results, then constructs a spatiotemporal triplet based on the base station zipper entry and exit table, then determines the user dynamic representation and regional dynamic representation based on the spatiotemporal triplet, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, regional dynamic representation, the recognition result, and the target extraction result. This application performs intent recognition and entity extraction on user content to obtain recognition results and target extraction results corresponding to the user input content, then determines the user dynamic representation and regional dynamic representation based on the spatiotemporal triplet, and can accurately capture the dynamic representation of users and regions, effectively solving the deficiencies in spatiotemporal data processing and the restrictive problems of field applications, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, regional dynamic representation, recognition results, and target extraction results, which can provide users with intuitive chart data display and enhance user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 This is a flow chart of the first embodiment of the data processing method of the present application; Figure 2 This is a flow chart of determining a spatiotemporal triplet in an embodiment of a data processing method of the present application; Figure 3 This is a flow chart of the second embodiment of the data processing method of the present application; Figure 4 This is a DIET model architecture diagram of an embodiment of the data processing method of this application; Figure 5 A flowchart of intent recognition and entity extraction of an embodiment of a data processing method of the present application; Figure 6 This is a flow chart of the third embodiment of the data processing method of the present application; Figure 7 A schematic diagram of the process of pre-training a spatiotemporal dynamic graph according to an embodiment of the data processing method of the present application; Figure 8 A schematic diagram of the process of spatiotemporal dynamic graph reasoning of an embodiment of the data processing method of the present application; Fig. 9 A schematic diagram of a user dynamic representation and a region dynamic representation of an embodiment of a data processing method of the present application; Fig.10 A schematic diagram of the chart data of Zhixunke in accordance with an embodiment of the data processing method of the present application; Fig.11 A schematic diagram of chart data of smart addressing according to an embodiment of the data processing method of the present application; Fig.12 A schematic diagram of chart data for identifying rows in an embodiment of a data processing method of the present application; Fig.13 This is a structural block diagram of the first embodiment of the data processing device of the present application; Fig.14 It is a structural diagram of a data processing device of a hardware operating environment involved in an embodiment of the present application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0019] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0020] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0021] The main solution of the embodiment of the present application is: to perform intent recognition and entity extraction on user input content to obtain recognition results and target extraction results; to construct a space-time triplet based on the base station zipper entry and exit table; to determine the user dynamic representation and the region dynamic representation based on the space-time triplet; and to output the chart data corresponding to the user input content based on the user dynamic representation, the region dynamic representation, the recognition results and the target extraction results.

[0022] At present, in the field of intelligent dialogue systems, the shortcomings can be mainly attributed to the following points: (1) Insufficient spatiotemporal data processing: When processing queries involving spatiotemporal characteristics, there is a lack of in-depth understanding and analysis capabilities. (2) Domain limitation: It is mainly designed for general dialogue scenarios and lacks professional support and optimization for specific fields. (3) Interaction mode limitation: It is mainly designed for telephone channels and has limited processing capabilities for spatiotemporal queries of non-voice input. (4) Real-time requirements: The real-time requirements of telephone interactions may limit the in-depth analysis and processing of complex spatiotemporal queries. (5) Lack of context fusion: There are limitations in fusing spatiotemporal context information. (6) Insufficient scalability: It is mainly designed for single dialogue state tracking and lacks the ability to adapt and expand to changing application scenarios.

[0023] This application performs intent recognition and entity extraction on user input content to obtain recognition results and target extraction results, then constructs a spatiotemporal triplet based on the base station zipper entry and exit table, then determines the user dynamic representation and regional dynamic representation based on the spatiotemporal triplet, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, regional dynamic representation, recognition results, and target extraction results. This application performs intent recognition and entity extraction on user content to obtain recognition results and target extraction results corresponding to the user input content, then determines the user dynamic representation and regional dynamic representation based on the spatiotemporal triplet, and can accurately capture the dynamic representation of users and regions, effectively solving the deficiencies in spatiotemporal data processing and the restrictive problems of field applications, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, regional dynamic representation, recognition results, and target extraction results, which can provide users with intuitive chart data display and enhance user experience.

[0024] It should be noted that the execution subject of the present application may be a computing service device with data processing, network communication and program running functions, such as a computer, or an electronic device capable of implementing the above functions, a data processing device, etc. The following takes a data processing device as an example to illustrate this embodiment and the following embodiments.

[0025] Based on this, the present application embodiment provides a data processing method, referring to Figure 1 , Figure 1 This is a flowchart of the first embodiment of the data processing method of the present application.

[0026] In this embodiment, the data processing method includes the following steps: Step S10: Perform intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results.

[0027] It is understandable that the user input content can be a time-space related question that the user wants to query, such as "I want to know the average length of stay of Beijing online car-hailing drivers at charging stations during the day". The user input content is subjected to intent recognition to obtain a recognition result, and the recognition result corresponding to the above user input content is Zhixunke. The user input content can also be subjected to entity extraction to obtain a target extraction result, and the target extraction result corresponding to the above user input content is Beijing, online car-hailing drivers, daytime, charging stations, and average length of stay distribution.

[0028] Step S20: constructing a spatiotemporal triplet according to the base station zipper entry and exit table.

[0029] It should be understood that the base station zipper access table is a table used in data warehouse design, mainly used to record the historical status and latest status of base station equipment. The zipper table can restore the status of historical data by recording the start time, end time and change type (addition, modification, deletion) of each data item. The time-space triplet can be data related to the user's unique identifier, space identifier, and time in the base station zipper access table.

[0030] Furthermore, in order to effectively construct a space-time triplet, in this embodiment, the step S20 includes: extracting key fields and original base station data corresponding to the key fields from the base station zipper entry and exit table; performing data cleaning on the original base station data to obtain cleaned base station data; performing outlier removal on the cleaned data to obtain removed base station data; and constructing a space-time triplet according to the data format corresponding to the removed base station data.

[0031] It is understandable that key fields such as [user unique identifier, base station unique identifier, base station longitude and latitude, time] can be extracted from the base station zipper input table, and the original base station data corresponding to the above key fields can also be obtained from the base station zipper input table. Then, the original base station data is cleaned, which may include desensitization processing and time format conversion to obtain cleaned base station data, that is, converting the original base station data into a more universal and secure format [user unique identifier, space unique identifier, space longitude and latitude, time].

[0032] It should be understood that the cleaned data can also be subjected to outlier removal, which may include outlier removal and trajectory reconstruction to ensure the accuracy and reliability of the base station data after removal. Through this process, the data format can be converted into [user unique identifier, spatial longitude and latitude, time]. This embodiment carefully sorts the data according to user and time to ensure that the spatial position of each user at each time point is accurately recorded and indexed. In this way, this embodiment forms the final spatiotemporal triple [user unique identifier, spatial identifier, time], which provides a structured, orderly and information-rich data foundation for constructing a spatiotemporal dynamic graph.

[0033] In the specific implementation, refer to Figure 2 , Figure 2 This is a flow chart of determining a spatiotemporal triplet in an embodiment of a data processing method of the present application, as shown in FIG. Figure 2 As shown, key fields can be extracted from the base station zipper entry and exit table, including user: IMSI, space: LAC_CELL, lon, lat, time: precedenceStartTime. After desensitization / ID mapping, milliseconds are converted to seconds to obtain user: UID, space: CID, lon, lat, time: startTime. After exception processing and trajectory reconstruction, user: UID, space: lon, lat, time: startTime is obtained. After gridding and sorting in chronological order, user: UID, space: GID, time: startTime is obtained. Finally, a spatiotemporal triplet can be obtained. Figure 2 The spatiotemporal triples in the data include [7231058,19422,1677625200], etc. Each column represents the user's unique ID, spatial ID, and time, respectively. The corresponding spatial range: Beijing, 250-meter grid 40,000+, time range: March 2023, user range: more than 30 million users, data scale: 160 billion triple data.

[0034] Step S30: determining a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet.

[0035] It is understandable that the user dynamic representation and the region dynamic representation can be determined based on the spatiotemporal triplet, and can be determined specifically through a neural network model, and the user dynamic representation and the region dynamic representation can represent the complex dynamics of users and regions changing over time. After obtaining the user dynamic representation and the region dynamic representation, a spatiotemporal dynamic graph can also be constructed based on the above representations.

[0036] Step S40: outputting chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

[0037] It should be understood that graphical data corresponding to the user input content can be obtained based on the user dynamic representation, area dynamic representation, recognition results and target extraction results. For example, the user input content is "During the lunch period, which areas do Beijing deliverymen most frequently visit for delivery?" The graphical data may include tables, heat maps, etc.

[0038] This embodiment performs intent recognition and entity extraction on user input content to obtain recognition results and target extraction results, then constructs a spatiotemporal triplet based on the base station zipper entry and exit table, then determines the user dynamic representation and the regional dynamic representation based on the spatiotemporal triplet, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, the regional dynamic representation, the recognition result, and the target extraction result. This embodiment performs intent recognition and entity extraction on user content to obtain recognition results and target extraction results corresponding to the user input content, then determines the user dynamic representation and the regional dynamic representation based on the spatiotemporal triplet, and can accurately capture the dynamic representation of the user and the region, effectively solving the deficiencies in spatiotemporal data processing and the restrictive problems of field applications, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, the regional dynamic representation, the recognition result, and the target extraction result, which can provide users with intuitive chart data display and enhance user experience.

[0039] refer to Figure 3 , Figure 3 This is a flow chart of the second embodiment of the data processing method of the present application.

[0040] Based on the above first embodiment, in this embodiment, step S10 includes: Step S101: input the user input content into the pre-trained model to obtain a pre-trained vector.

[0041] It is understandable that the pre-trained model can be a bidirectional encoder based on Transformer (Bidirectional Encoder Representations from Transformer, BERT), a lightweight, efficient and accurate dialogue system model based on Transformer (ConveRT), a global vector model for unsupervised learning word representation (Global Vectors for Word Representation, GloVe), etc. The user input content is input into the above pre-trained model to obtain a pre-trained vector.

[0042] Step S102: Input the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result.

[0043] It should be understood that the DIET model (Dual Intent and Entity Transformer) is a multi-task transformer architecture that is mainly used to simultaneously process intent classification and entity recognition tasks. The DIET model can simultaneously process intent classification and entity recognition, thereby improving the accuracy of natural language understanding in dialogue systems.

[0044] In the specific implementation, refer to Figure 4 , Figure 4 This is the DIET model architecture diagram of an embodiment of the data processing method of the present application. The user input content can be a sentence or text, and then 85% of the data after semantic information extraction is input into the fully connected layer, and then the intention recognition is performed through a two-layer Transformer encoder to obtain the recognition result.

[0045] Step S103: Input the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain a target extraction result.

[0046] Furthermore, in this embodiment, the step S103 includes: inputting the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain an initial extraction result, wherein the initial extraction result includes at least one of location, user portrait, time, point of interest, and indicator; and standardizing the initial extraction result to obtain a target extraction result.

[0047] It is understandable that the DIET model can be used to extract entities from user input content, that is, Figure 4 The CRF layer in is used to extract entities and obtain the initial extraction results. Figure 5 , Figure 5 This is a flow chart of intent recognition and entity extraction in accordance with an embodiment of the data processing method of this application. Figure 5 In the example, the user input is: the number of people traveling in Guangzhou yesterday. After the ConveRT / BERT / GloVe pre-trained vector is input into the DIET model, the recognition result of intent recognition is data query, and the initial extraction result of entity extraction is time: yesterday; location: Guangzhou; indicator: number of people traveling.

[0048] It should be understood that for the standardized processing of the initial extraction results, for example, the user input content is: I want to know the average length of stay of Beijing online car-hailing drivers at charging stations during the day. The initial extraction results are: "time": ['daytime'], "location": ['Beijing'], "indicator": ['average length of stay distribution'], "persona": ['online car-hailing driver'], "poi": ['charging station']. For time word processing, automatic time word parsing and standardized processing can be used. This embodiment ensures the accuracy and consistency of time information and improves the automation level of time series processing. For location word processing, location word parsing and geographic information construction can be used. This embodiment provides strong support for applications that require geographic information and ensures the accuracy of location information. For indicator processing, indicator matching and fuzzy matching technology can be used. This embodiment improves the flexibility and accuracy of indicator processing and reduces errors caused by inconsistent indicator names. For user portrait processing, user portrait matching and spelling error tolerance technology can be used. This embodiment allows the system to accurately identify and respond to user needs in the face of uncertainty in user input. For the processing of interest points, POI matching and similarity calculation can be performed. This embodiment improves the accuracy and efficiency of interest point processing and helps the system understand and respond to the user's interest in a specific place or service. After the above standardization processing, the target extraction result can be obtained.

[0049] This embodiment inputs the user input content into a pre-trained model to obtain a pre-trained vector, then inputs the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result, and then inputs the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain a target extraction result. This embodiment inputs the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result, and then inputs the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain a target extraction result. The DIET model can accurately perform intent recognition and entity extraction on the user input content, thereby improving the accuracy of natural language understanding in the dialogue system.

[0050] refer to Figure 6 , Figure 6 This is a flowchart of the third embodiment of the data processing method of the present application.

[0051] Based on the above embodiments, in this embodiment, step S30 includes: Step S301: training the initial neural network model according to the spatiotemporal triplet to obtain a target network model, and outputting an initial user dynamic representation and an initial region dynamic representation.

[0052] Understandably, referring to Figure 7 , Figure 7 The flowchart of the pre-training of the spatiotemporal dynamic graph of the first embodiment of the data processing method of the present application is shown in the figure. For the pre-training of the spatiotemporal dynamic graph, the specific process is as follows: (1) Randomly initialize the initial neural network model parameters: In the initial stage of pre-training, all model parameters are assigned random values, which is the starting point of the training process. Assign initial values ​​to the user dynamic representation and the region dynamic representation, which can be zero vectors or other randomly generated vectors. (2) Receiving spatiotemporal triples: The initial neural network model starts to receive input data, that is, spatiotemporal triples, which include user identifiers, spatial identifiers, and timestamps. At the first breakpoint (breakpoint 1), the model receives the first spatiotemporal triplet. (3) Update model parameters: Based on the received spatiotemporal triplet data, the model parameters are adjusted to better reflect the dynamic characteristics of users and regions. The model updates the dynamic representation of users and regions through a certain learning algorithm (such as gradient descent), thereby obtaining updated user dynamic representation 1 and region dynamic representation 1. (4) Iterative update process: As more spatiotemporal triplet data is input, the model continues to learn and update to capture the dynamic characteristics of users and regions that change over time. At each new breakpoint (e.g., breakpoint 2, breakpoint N, etc.), the model receives new spatiotemporal triplets and then updates its parameters again. (5) Continuously optimize representation: Through continuous iteration, the model's representation is gradually optimized to more accurately reflect the spatiotemporal characteristics of users and regions. At each breakpoint, the model generates a new version of the user dynamic representation and the regional dynamic representation until the predetermined training cycle or performance standard is reached. (6) Model convergence: After multiple iterations, the model eventually converges to a set of optimal parameters that can best describe the spatiotemporal dynamic characteristics of users and regions. At the last breakpoint (breakpoint N), the model completes the last parameter update and obtains the final user dynamic representation N and regional dynamic representation N, that is, the initial user dynamic representation and the initial regional dynamic representation.

[0053] Step S302: updating the initial user dynamic representation and the initial region dynamic representation to obtain a target user dynamic representation and a target region dynamic representation.

[0054] It should be understood that for the reasoning of spatiotemporal dynamic graphs, the dynamic representations of users and regions are continuously iterated as time progresses. This process does not involve the update of model parameters until the breakpoint with the best effect is reached. Subsequently, the representations continue to be updated until the model converges to the final representation state of the user and region, that is, the dynamic representation of the target user and the dynamic representation of the target region.

[0055] Furthermore, in order to effectively update the initial user dynamic representation and the initial region dynamic representation, in this embodiment, the step S303 includes: obtaining new space-time triples corresponding to each moment within a preset time period; inputting each new space-time triplet into the target network model to obtain a new user dynamic representation and a new region dynamic representation; determining the optimal moment according to the new user dynamic representation and the new region dynamic representation; updating the initial user dynamic representation and the initial region dynamic representation according to the optimal user dynamic representation and the optimal region dynamic representation corresponding to the optimal moment to obtain the target user dynamic representation and the target region dynamic representation.

[0056] Understandably, referring to Figure 8 , Figure 8 This is a flow chart of the spatiotemporal dynamic graph reasoning of an embodiment of the data processing method of the present application, such as Figure 8As shown in the figure, (1) Loading pre-trained model parameters: Before the inference starts, first load the final parameters of the pre-trained user and region dynamic representations. Load the user dynamic representation N and the region dynamic representation N, which are the optimal parameters obtained during the pre-training process. (2) Receiving new spatiotemporal triple data: During the inference process, the model will receive new spatiotemporal triple data, which contains user ID, space ID and timestamp. In the first step of inference, the model receives spatiotemporal triple N+1. (3) Dynamic representation update (without updating model parameters): Although the model parameters are not updated, the dynamic representations of users and regions will be adjusted according to the new spatiotemporal triple data. Using the data of spatiotemporal triple N+1, the user dynamic representation and the region dynamic representation are updated according to the logic or algorithm obtained by pre-training to obtain the representation at time N+1. (4) Continuous inference process: As more new spatiotemporal triple data are input, the model continues to reason to capture the real-time dynamics of users and regions. For each new spatiotemporal triple (such as N+2, N+M, etc.), the model repeats (3) to update the dynamic representations of users and regions. (5) Find the best breakpoint: During the reasoning process, find a breakpoint with the best effect, that is, the moment when the user and regional dynamic representations are optimal or most consistent with the current data. Evaluate the performance impact of each updated representation on the current task and determine the best breakpoint, such as user dynamic representation N+2 and regional dynamic representation N+2. (6) Maintain the best representation: Once the best breakpoint is found, the model will maintain the user and regional dynamic representation at this moment for subsequent analysis or decision-making. At the best breakpoint, stop further updates and retain user dynamic representation N+2 and regional dynamic representation N+2 as the reasoning results, that is, as the target user dynamic representation and target regional dynamic representation. (7) Apply reasoning results: Apply the user and regional dynamic representations obtained by reasoning to actual spatiotemporal analysis tasks, such as real-time recommendation, dynamic monitoring, etc. Use the representations obtained by reasoning for further analysis or decision support to achieve real-time response to user behavior and regional characteristics.

[0057] It should be understood that reference Fig. 9 , Fig. 9 This is a schematic diagram of a user dynamic representation and a region dynamic representation according to an embodiment of a data processing method of the present application, which can be specifically represented by vectors.

[0058] In a specific implementation, this embodiment can output chart data corresponding to the user input content according to the user dynamic representation, regional dynamic representation, recognition results, and target extraction results. For example, the "Smart Customer Search" process: Take the user input "At which charging stations do online car-hailing drivers stay the longest during the day?" as an example: 1. Question input and analysis: Receive and analyze the user's specific query request. The system receives the user's query, such as "At which charging stations do online car-hailing drivers stay the longest during the day?", and uses the DIET analysis results to extract key parameters, including indicators, locations, character portraits, points of interest, and time. 2. Query charging station location information: Determine the specific geographical location of all charging stations (based on the representation vectors obtained in the pre-training stage of the spatiotemporal dynamic graph, these vectors are labeled with regional types through downstream classification tasks). The system queries the longitude and latitude positions of all charging stations according to the POI table. 3. Obtain the base station ID corresponding to the charging station: Associate the location of the charging station with the base station ID for further analysis. The system obtains the base station ID corresponding to each charging station through the POI and base station mapping table. 4. Draw a grid and obtain the base station ID: Create a grid system centered on the POI to analyze the distribution of base stations. The system draws a 200*200 square grid with each POI as the center, and traverses to obtain the base station ID within the grid. 5. Query user portraits and residence time: obtain the activity data of a specific user portrait at a specific time and place (the user portrait is based on the user representation vector of the spatiotemporal dynamic graph, and the user label is classified through the downstream classification task). The system queries the user portrait table based on the base station ID, start and end time, and online car-hailing driver portrait to obtain the number of visitors and the TOP residence time of each charging station. 6. Handle access frequency issues: For queries without specific POIs, such as "Where do deliverymen usually deliver food?", provide corresponding data analysis. The system queries the user portrait table by base station ID group, and obtains the total number of deliverymen visiting each administrative area and the TOP residence time of each base station. 7. Generate heat maps: Visually display the activity intensity of users in different locations. The system generates heat maps based on the query results to show the stay time of online car-hailing drivers or deliverymen in different locations. 8. Return query results: Provide query results to users. The system returns the query results, including the number of deliverymen in the administrative area and the address of the base station with the longest stay time. 9. Text result output: Provide a summary of the query results in text form. The system outputs text results, such as "Beijing's online car-hailing drivers spend the longest time at POI1, POI2, POI3, etc. between 8 am and 8 pm." The above process shows how the Zhixunke system goes through a series of steps, from receiving user queries to providing detailed data analysis results, including location stay time analysis and visit frequency analysis, and finally presents the results to users in the form of heat maps and text. For example, input: During the lunch period, which areas do Beijing deliverymen most frequently go to for delivery? Output: Related chart data, including tables, heat maps, etc., refer to Fig.10 , Fig.10 This is a schematic diagram of Zhixunke’s chart data in accordance with an embodiment of the data processing method of the present application.

[0059] In addition, (II) "Smart Site Selection" process: 1. Start: Start the site selection recommendation process. The user enters the system and proposes a site selection requirement. 2. User question: Obtain the user's query about the area where a specific type of point of interest (POI) is opened. For example: The user asks the system "the area that is most suitable for opening a certain type of POI". 3. Default output: Provide default recommendations based on the system's current data and algorithm. The system outputs the 5 grid areas that are most suitable for opening a certain type of POI by default. 4. Current area recommendation: Provide recommendations based on current data. Based on the existing data and reasoning model, the system recommends 5 grid areas as candidate locations for opening a certain type of POI. 5. User selection area: Allow users to participate in the site selection process to improve the accuracy of recommendations and user satisfaction. The user selects at least 10 points on the map as input, which represent areas that the user believes may be suitable for opening a certain type of POI. 6. Spatiotemporal dynamic graph reasoning: Use the area selected by the user for in-depth analysis. The spatiotemporal dynamic graph uses a large-scale reasoning model to perform in-depth analysis based on the 10 points selected by the user to determine the grid area suitable for opening a grass POI. 7. Reasoning based on user selection: Provide customized recommendations based on user selection and the system's reasoning capabilities. Based on the points selected by the user, the system infers grid areas suitable for opening grass POIs. 8. Get recommendation results: Provide final site recommendations. The system outputs 5 grid areas suitable for opening grass POIs as final recommendations. The entire process above embodies an interactive intelligent site recommendation system, which not only relies on the data analysis and reasoning models within the system, but also combines the user's direct input and preferences to provide more accurate and personalized site recommendations. In this way, the smart site selection system can help users make more informed decisions in complex geographical and business environments. For example, input: I want to open a convenience store in Beijing, what good locations are recommended? Output: A graphical display of recommended locations, refer to Fig.11 , Fig.11 This is a schematic diagram of chart data for smart addressing according to an embodiment of the data processing method of the present application.

[0060] In addition, the "Travel Identification" process: 1. Start: Initialize the congestion analysis process. The system starts and is ready to receive and process user queries about traffic congestion. 2. User questions: Collect user traffic congestion questions about specific locations and times. For example, the user asks "Why is Beijing's North Third Ring Road so congested during the morning rush hour?" 3. Model parsing parameters: Extract key information from user questions to prepare for analysis. DIET parses the location, travel purpose, road information and time in the question. 4. Read the road mapping grid table: Get the grid information associated with a specific road ID. The system reads the road mapping grid table according to the road ID and obtains all the grids of the road. 5. Determine the number of grids: Determine whether further user operation is required. The system determines whether the number of grids on the road is greater than a certain threshold. If it is, the user is required to set the start and end points. 6. User set the start and end points: Let the user specify the specific area for analysis. If the number of grids is greater than the threshold, the system prompts the user to set the start and end points on the map, and returns the points with the smallest and largest grid numbers as the latitude and longitude of the start and end points to the backend. 7. Calculate the grids between the start and end points: determine the specific road section to be analyzed. Based on the start and end points set by the user, the system calculates all grids between the start and end points of the road. 8. Query and identify the travel table: collect and analyze traffic flow data at specific times and sections. According to the start and end times, the system queries the identified travel table (the full-link travel table inferred from the spatiotemporal dynamic graph based on user representation and regional representation), and counts the passenger flow, month-on-month growth rate, and the number and proportion of similar travel modes for each travel mode. 9. Return analysis results: Provide users with the results of congestion analysis. Action: The system organizes the analysis results and returns them to the front end for users to view. For example, input: Why is Beijing North Third Ring Road so congested today? Output: Traffic information of the relevant sections including charts, refer to Fig.12 , Fig.12 This is a schematic diagram of chart data for identifying rows according to an embodiment of the data processing method of the present application.

[0061] This embodiment trains the initial neural network model according to the spatiotemporal triples to obtain the target network model, and outputs the initial user dynamic representation and the initial region dynamic representation, and then updates the initial user dynamic representation and the initial region dynamic representation to obtain the target user dynamic representation and the target region dynamic representation. This embodiment first trains the initial neural network model according to the spatiotemporal triples to obtain the trained target network model, and then updates the initial user dynamic representation and the initial region dynamic representation, which can improve the accuracy of the target user dynamic representation and the target region dynamic representation.

[0062] Reference Fig.13 , Fig.13 This is a structural block diagram of the first embodiment of the data processing device of the present application.

[0063] like Fig.13As shown, the data processing device proposed in the embodiment of the present application includes: The content processing module 10 is used to perform intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results; A space-time triplet construction module 20, used to construct a space-time triplet according to a base station zipper entry and exit table; A dynamic representation determination module 30, configured to determine a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet; The chart data determination module 40 is used to output chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

[0064] This embodiment performs intent recognition and entity extraction on user input content to obtain recognition results and target extraction results, then constructs a spatiotemporal triplet based on the base station zipper entry and exit table, then determines the user dynamic representation and the regional dynamic representation based on the spatiotemporal triplet, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, the regional dynamic representation, the recognition result, and the target extraction result. This embodiment performs intent recognition and entity extraction on user content to obtain recognition results and target extraction results corresponding to the user input content, then determines the user dynamic representation and the regional dynamic representation based on the spatiotemporal triplet, and can accurately capture the dynamic representation of the user and the region, effectively solving the deficiencies in spatiotemporal data processing and the restrictive problems of field applications, and then outputs the chart data corresponding to the user input content based on the user dynamic representation, the regional dynamic representation, the recognition result, and the target extraction result, which can provide users with intuitive chart data display and enhance user experience.

[0065] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present application. In practical applications, technicians in this field can select part or all of it according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0066] In addition, for technical details not fully described in this embodiment, please refer to the data processing method provided in any embodiment of the present application, and will not be repeated here.

[0067] Based on the above-mentioned first embodiment of the data processing device of the present application, a second embodiment of the data processing device of the present application is proposed.

[0068] In this embodiment, the content processing module 10 is also used to input the user input content into a pre-trained model to obtain a pre-trained vector; input the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result; input the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain a target extraction result.

[0069] Furthermore, the content processing module 10 is also used to input the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain an initial extraction result, wherein the initial extraction result includes at least one of location, user portrait, time, point of interest, and indicator; and standardize the initial extraction result to obtain a target extraction result.

[0070] Furthermore, the space-time triplet construction module 20 is also used to extract key fields and original base station data corresponding to the key fields from the base station zipper entry and exit table; perform data cleaning on the original base station data to obtain cleaned base station data; perform outlier removal on the cleaned data to obtain removed base station data; and construct a space-time triplet according to the data format corresponding to the removed base station data.

[0071] Furthermore, the dynamic representation determination module 30 is also used to train the initial neural network model according to the spatiotemporal triplet to obtain the target network model, and output the initial user dynamic representation and the initial region dynamic representation; update the initial user dynamic representation and the initial region dynamic representation to obtain the target user dynamic representation and the target region dynamic representation.

[0072] Furthermore, the dynamic representation determination module 30 is also used to obtain new space-time triples corresponding to each moment within a preset time period; input each new space-time triplet into the target network model to obtain a new user dynamic representation and a new area dynamic representation; determine the optimal moment according to the new user dynamic representation and the new area dynamic representation; update the initial user dynamic representation and the initial area dynamic representation according to the optimal user dynamic representation and the optimal area dynamic representation corresponding to the optimal moment, and obtain the target user dynamic representation and the target area dynamic representation.

[0073] Other embodiments or specific implementations of the data processing device of the present application can refer to the above-mentioned method embodiments and will not be described in detail here.

[0074] The present application provides a data processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the data processing method in the above-mentioned embodiment 1.

[0075] Reference below Fig.14 , which shows a schematic diagram of the structure of a data processing device suitable for implementing an embodiment of the present application. The data processing device in the embodiment of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Fig.14 The data processing device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0076] like Fig.14 As shown, the data processing device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the data processing device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the data processing device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a data processing device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0077] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0078] The data processing device provided by the present application adopts the data processing method in the above embodiment, which can solve the technical problem of how to effectively solve the challenge of spatiotemporal data processing, provide users with intuitive chart data display, and enhance user experience. Compared with the prior art, the beneficial effects of the data processing device provided by the present application are the same as the beneficial effects of the data processing method provided by the above embodiment, and the other technical features in the data processing device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0079] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0080] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0081] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the data processing method in the above-mentioned embodiment.

[0082] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0083] The computer-readable storage medium may be included in a data processing device, or may exist independently without being incorporated into a data processing device.

[0084] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by a data processing device, the data processing device: performs intent recognition and entity extraction on user input content to obtain recognition results and target extraction results; constructs a spatiotemporal triple according to the base station zipper entry and exit table; determines the user dynamic representation and the regional dynamic representation according to the spatiotemporal triple; and outputs the chart data corresponding to the user input content according to the user dynamic representation, the regional dynamic representation, the recognition result and the target extraction result.

[0085] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0086] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0087] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0088] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned data processing method, and can solve the technical problem of how to effectively solve the challenges of spatiotemporal data processing, provide users with intuitive chart data display, and enhance user experience. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the data processing method provided in the above-mentioned embodiment, and will not be repeated here.

[0089] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned data processing method when executed by a processor.

[0090] The computer program product provided by this application can solve the technical problem of how to effectively solve the challenges of spatiotemporal data processing, provide users with intuitive chart data display, and enhance user experience. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the data processing method provided by the above embodiment, and will not be repeated here.

[0091] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A data processing method, characterized in that: The data processing method comprises the following steps: Perform intent recognition and entity extraction on user input content to obtain recognition results and target extraction results; Constructing space-time triples based on the base station zipper entry and exit table; Determine a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet; Outputting chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

2. The data processing method according to claim 1, characterized in that: The step of performing intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results specifically includes: Input the user input content into the pre-trained model to obtain a pre-trained vector; Inputting the pre-trained vector into the DIET model to perform intent recognition on the user input content to obtain a recognition result; The pre-trained vector is input into the DIET model to perform entity extraction on the user input content to obtain a target extraction result.

3. The data processing method according to claim 2, characterized in that: The step of inputting the pre-trained vector into the DIET model to extract entities from the user input content and obtain a target extraction result specifically includes: Input the pre-trained vector into the DIET model to perform entity extraction on the user input content to obtain an initial extraction result, wherein the initial extraction result includes at least one of location, user portrait, time, point of interest, and indicator; The initial extraction result is standardized to obtain a target extraction result.

4. The data processing method according to claim 1, characterized in that: The step of constructing a spatiotemporal triplet according to the base station zipper entry and exit table specifically includes: Extract key fields and original base station data corresponding to the key fields from the base station zipper access table; Performing data cleaning on the original base station data to obtain cleaned base station data; Eliminating outliers from the cleaned data to obtain base station data after elimination; A space-time triplet is constructed according to the data format corresponding to the removed base station data.

5. The data processing method according to claim 1, characterized in that: The step of determining the user dynamic representation and the region dynamic representation according to the spatiotemporal triplet specifically includes: Training the initial neural network model according to the spatiotemporal triplet to obtain a target network model, and outputting an initial user dynamic representation and an initial region dynamic representation; The initial user dynamic representation and the initial region dynamic representation are updated to obtain a target user dynamic representation and a target region dynamic representation.

6. The data processing method according to claim 5, characterized in that: The step of updating the initial user dynamic representation and the initial region dynamic representation to obtain the target user dynamic representation and the target region dynamic representation specifically includes: Obtain new space-time triples corresponding to each moment in a preset time period; Inputting each new spatiotemporal triple into the target network model to obtain a new user dynamic representation and a new region dynamic representation; Determine the optimal time according to the new user dynamic representation and the new area dynamic representation; The initial user dynamic representation and the initial region dynamic representation are updated according to the optimal user dynamic representation and the optimal region dynamic representation corresponding to the optimal time, so as to obtain the target user dynamic representation and the target region dynamic representation.

7. A data processing device, characterized in that: The data processing device comprises: The content processing module is used to perform intent recognition and entity extraction on the user input content to obtain recognition results and target extraction results; A space-time triplet construction module, used to construct a space-time triplet according to a base station zipper entry and exit table; A dynamic representation determination module, used to determine a user dynamic representation and a region dynamic representation according to the spatiotemporal triplet; The chart data determination module is used to output chart data corresponding to the user input content according to the user dynamic representation, the region dynamic representation, the recognition result and the target extraction result.

8. A data processing device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the data processing method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the data processing method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the data processing method according to any one of claims 1 to 6 are implemented.

Images