A behavior data processing method and device, computer equipment and a storage medium

Abstract

The present application relates to the technical field of data processing, and discloses a behavior data processing method and device, computer equipment and a storage medium, wherein the method comprises: obtaining an original sequence of agent event behavior data; performing aggregation processing on a plurality of operations included in the original sequence to obtain a summarized behavior description; dividing a plurality of original sequences in a set range into a plurality of behavior subsequences based on the summarized behavior description; and generating a behavior summary for the behavior subsequences, wherein the behavior summary comprises set key information for visually displaying the agent event. Thus, the agent event behavior data is automatically and reasonably analyzed, and the generated behavior summary can more intuitively and accurately display the agent event behavior data, effectively solving the problem that the agent event behavior data is not effectively analyzed, resulting in cumbersome viewing operations and low reliability.

Description

Technical Field

[0001] This invention relates to data processing technology in the field of artificial intelligence, specifically to a behavioral data processing method, apparatus, computer equipment, and storage medium. Background Technology

[0002] In the field of artificial intelligence, especially in autonomous systems based on LLM (Large Language Model), the analysis and visualization of agent behavior, such as log recording and post-event analysis, is a crucial research area. Currently, the main method for tracking agent behavior and its decision-making process is manual review of text output logs. While this provides detailed agent information, manually reviewing text output logs demands a high level of user expertise, and the sheer volume of logs makes manual review inefficient and prone to errors.

[0003] Related technologies use panoramic visualization tools to observe the simulation process of LLM-based autonomous systems, visualizing agent behavior in a time-series format, allowing users to view the agent's location and behavior within the environment. However, they lack the ability to intuitively view and analyze detailed agent behavior information, and also lack interactivity and dynamic update capabilities.

[0004] In response, some projects have adopted panoramic video as an observation method, such as... Figure 1 As shown, the interface of a panoramic visualization tool is presented, which provides detailed maps and agent locations. However, users need to frequently switch and replay videos during the analysis process to track the behavior of specific actions, which is cumbersome and unreliable. Summary of the Invention

[0005] In view of this, the present invention provides a behavioral data processing method, apparatus, computer equipment, and storage medium to solve the problem that the lack of effective analysis of behavioral data of proxy events in large language models leads to cumbersome operation and low reliability in viewing proxy event behavioral data.

[0006] In a first aspect, the present invention provides a behavior data processing method, the method comprising: acquiring the original sequence of proxy event behavior data; aggregating multiple operations included in the original sequence to obtain a general behavior description; dividing multiple original sequences within a set range into multiple behavior subsequences based on the general behavior description; generating a behavior summary for the behavior subsequences, the behavior summary including key information for visualizing the proxy event.

[0007] The behavioral data processing method provided in this invention obtains the original sequence of proxy event behavioral data, aggregates multiple operations included in the original sequence, and divides the multiple original sequences within a set range into multiple behavioral subsequences based on the summarized behavioral description obtained from the aggregation process. Furthermore, it generates behavioral summaries for the behavioral subsequences, wherein the behavioral summaries include key information for visualizing the proxy event. Thus, the proxy event behavioral data is automatically and rationally analyzed, and the generated behavioral summaries can more intuitively and accurately display the proxy event behavioral data, effectively solving the problem of cumbersome viewing operations and low reliability caused by ineffective analysis of proxy event behavioral data.

[0008] In one optional implementation, the behavioral data is log data of the proxy event; correspondingly, obtaining the original sequence of the proxy event behavioral data includes: parsing and / or transforming the log data to obtain an original sequence in a set format used to characterize the operation performed at a set time point.

[0009] The behavioral data processing method provided in this embodiment of the invention obtains a raw sequence in a set format that represents the operation performed at a set time point by parsing and transforming log data. It fully parses the proxy event behavioral data to ensure the comprehensiveness and accuracy of the raw data for behavioral data processing, thereby effectively ensuring the rationality of subsequent operations.

[0010] In one optional implementation, based on the generalized behavioral description, multiple original sequences within a defined range are divided into multiple behavioral subsequences, including: converting the generalized behavioral descriptions corresponding to the multiple original sequences within the defined range into numerical vectors; performing change point detection on the numerical vectors to obtain at least one mutation point; and dividing the numerical vectors according to the at least one mutation point to obtain behavioral subsequences.

[0011] The behavioral data processing method provided in this invention first converts the generalized behavioral descriptions corresponding to multiple original sequences within a defined range into numerical vectors. Then, it detects change points on the numerical vectors to obtain at least one mutation point. Based on this mutation point, the numerical vectors are divided to obtain behavioral subsequences. Thus, by fully exploiting the similarity of multiple proxy events through high-dimensional numerical vectors, the original sequences of proxy events are reasonably segmented, resulting in behavioral subsequences that can characterize significant behaviors within different event segments.

[0012] In one alternative implementation, the change point detection of the numerical vector includes: detecting abrupt changes in the numerical vector using a sliding window method.

[0013] The behavioral data processing method provided in this embodiment of the invention detects abrupt changes in numerical vectors using a sliding window method, and fully utilizes the high-dimensional spatial data features of the original sequence of proxy event behavioral data to divide the original sequence of proxy event behavioral data into multiple behavioral subsequences that are more significant.

[0014] In one alternative implementation, generating a behavioral summary for a behavioral subsequence includes: employing a summary generation algorithm to generate a behavioral summary for the behavioral subsequence.

[0015] The behavioral data processing method provided in this embodiment of the invention employs a summary generation algorithm to generate behavioral summaries for behavioral subsequences, significantly improving the user's visualization efficiency and understanding of agent behavioral data.

[0016] In one alternative implementation, after generating a behavioral summary for a behavioral subsequence, the method further includes: when a set trigger condition is detected, displaying at least the behavioral summary in a set format.

[0017] The behavioral data processing method provided in this embodiment of the invention detects a set trigger condition and displays at least a behavioral summary in a set format. Thus, the behavioral trajectory of the agent can be displayed in a set format such as a hierarchical and continuously scalable time axis, enabling users to quickly identify the patterns and trends of the agent's behavior at different time scales.

[0018] In one alternative implementation, the behavior summary is updated when the set range and / or the original sequence changes.

[0019] The behavior data processing method provided in this embodiment of the invention dynamically updates the behavior summary when the set range and / or the original sequence changes, thereby ensuring that as agent behavior occurs in real time, the user can always view the latest agent event behavior data based on the behavior summary.

[0020] Secondly, the present invention provides a behavior data processing apparatus, comprising: an acquisition module for acquiring the original sequence of proxy event behavior data; a summarization module for aggregating multiple operations included in the original sequence to obtain a summarized behavior description; a segmentation module for dividing multiple original sequences within a set range into multiple behavior subsequences based on the summarized behavior description; and a summarization module for generating a behavior summary for the behavior subsequences, wherein the behavior summary includes key information for visualizing the proxy event.

[0021] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the behavioral data processing method described in the first aspect or any corresponding embodiment thereof.

[0022] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the behavioral data processing method described in the first aspect or any corresponding embodiment thereof. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the interface of a panoramic visualization tool in related technologies;

[0025] Figure 2 This is a flowchart illustrating a behavioral data processing method according to an embodiment of the present invention;

[0026] Figure 3 This is a flowchart illustrating another behavioral data processing method according to an embodiment of the present invention;

[0027] Figure 4 This is a schematic diagram illustrating the implementation process of a specific application example of behavioral data processing provided in this embodiment of the invention;

[0028] Figure 5 This is a flowchart illustrating another behavioral data processing method according to an embodiment of the present invention;

[0029] Figure 6 This is a schematic diagram of the BVV function provided in an embodiment of the present invention;

[0030] Figure 7 This is a schematic diagram of the agent spatial location layout in BVV provided in an embodiment of the present invention;

[0031] Figure 8 This is a structural block diagram of a behavioral data processing apparatus according to an embodiment of the present invention;

[0032] Figure 9 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] like Figure 1 The interface of the panoramic visualization tool displays the main message: "This is a pre-computed replay of a simulation that accompanies the paper entitled 'Generative agents: Interactive Simulation of Human Behavior.' It is for demonstration purposes only." Users can play, pause, or stop the behavioral data video of the agent events by clicking the "Play" and "Pause" buttons. The specific actions of clicking, double-clicking, and long-pressing the "Play" and "Pause" buttons can be customized. The playback interface displays Current Time and the behavior of specific objects, such as John Lin's Current Action, Location, and Current Conversation. While it displays relatively detailed agent event behavioral data, it requires users to frequently switch and replay videos to track specific behavioral patterns, making the operation cumbersome, unreliable, and lacking interactivity and dynamic update capabilities.

[0035] In order to solve such Figure 1 The related technologies illustrated utilize panoramic visualization tools to observe the problems existing in the simulation process of LLM-based autonomous systems, enabling more efficient processing and intuitive display of agent event behavior data. This invention first processes the raw sequences of agent event behavior data in an LLM-based autonomous system (LLMAS), and then visualizes the processing results.

[0036] According to an embodiment of the present invention, a behavioral data processing method is provided. It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system, such as a set of computer-executable instructions. The computer system can be a computer device such as a computer, a mobile terminal such as a mobile phone or PDA, or other suitable computer systems; the present invention does not limit this. Furthermore, although a logical order is shown in the flowcharts, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0037] This embodiment provides a behavioral data processing method that can be used in a computer system. Figure 2 This is a flowchart of a behavioral data processing method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps:

[0038] Step S201: Obtain the original sequence of proxy event behavior data.

[0039] In this embodiment of the invention, the proxy event behavior data can be data such as LLMAS logs. The log data can be text in different formats, and the original sequence of the behavior data can show the temporal variation characteristics of the proxy event behavior. Obtaining the original sequence of the proxy event behavior data here aims to convert the original log data into a structured data format, such as JSON, to facilitate faster and more effective analysis of the log data.

[0040] In this embodiment of the invention, log data can be parsed and transformed using log parsers or natural language processing (NLP) technologies.

[0041] It should be noted that multiple agent events may be executed within any single minimum time unit. The minimum time unit can be set according to actual needs, for example, to 1 second. Agent events may include Position (location confirmation), Perceive (environmental awareness), Think (decision-making), and Act (action execution), etc. A timeline includes multiple minimum time units; in this invention, the minimum time unit is used as a point in time, and multiple consecutive time units are considered as a time period. Other locations in the document may describe the agent event behavior data on the timeline using concepts such as point in time and time period.

[0042] Step S202: Aggregate the multiple operations included in the original sequence to obtain a generalized behavioral description.

[0043] To reduce information redundancy and improve the readability of proxy event behavior data, in this embodiment of the invention, a clustering algorithm is used to merge consecutive similar operations in the proxy event behavior data into a higher-level generalized behavior description.

[0044] Specifically, statistical analysis can be performed on the original sequence to determine when a new behavior begins and ends. Proxy event behavior data is essentially text data, and clustering algorithms commonly used in this field can be employed to cluster the text data. The criteria for judging similar operations can be set according to actual needs. For example, for two consecutive proxy events, if the similarity of the proxy event behavior data is greater than a set similarity threshold, the two proxy events are judged as similar operations.

[0045] One method for judging the similarity of proxy event behavior data is to take data from two consecutive proxy event behavior data at predetermined positions for similarity assessment. For example, based on the format of the original sequence, a sequence of proxy event behavior data can be judged by taking a predetermined number of positions forward from a predetermined start marker, or by taking a predetermined number of positions backward from a predetermined end marker.

[0046] Step S203: Based on the generalized behavioral description, the multiple original sequences within the defined range are divided into multiple behavioral subsequences.

[0047] In this embodiment of the invention, agent event behavior data based on a single point in time can be parsed or converted into a raw sequence. The aggregation processing of multiple operations included in the raw sequence described above can be the operations performed on the agent event behavior sequence based on a single point in time. Multiple raw sequences within a defined range can be a collection of raw sequences of multiple agent event behavior data within a defined time period. Based on the generalized behavior description obtained in step S202 above, the multiple raw sequences within the defined range can be segmented as a sequence set, dividing the multiple raw sequences into multiple behavior subsequences. These behavior subsequences are used to characterize the significant behaviors of the agent in different time periods.

[0048] Step S204: Generate a behavior summary for the behavior subsequence, the behavior summary including key setting information for visualizing the proxy events.

[0049] In this embodiment of the invention, a large language model-based summarization algorithm can be used to generate behavioral descriptions for behavioral subsequences. A behavioral description is a concise description of the significant behaviors represented by each behavioral subsequence within the corresponding time period. This can include key textual descriptions of behaviors such as online learning, eating, and sleeping, as well as visual element descriptions such as emoticons. This provides more intuitive descriptive information when the behavioral data needs to be visualized. Summarization can employ generative summarization algorithms such as a sequence-to-sequence and attention-based model, or an Encoder-Decoder structure model, as well as extractive summarization algorithms such as MMR (Maximal Marginal Relevant), Text Rank (a keyword extraction algorithm based on random walks), and LDA (Latent Dirichlet Allocation). Alternatively, a combination of multiple algorithms can be used.

[0050] The behavioral data processing method provided in this embodiment obtains the original sequence of proxy event behavioral data, aggregates multiple operations included in the original sequence, and divides the multiple original sequences within a defined range into multiple behavioral subsequences based on the summarized behavioral description obtained from the aggregation process. Furthermore, it generates behavioral summaries for the behavioral subsequences, wherein the behavioral summaries include key information for visualizing the proxy event. Thus, the proxy event behavioral data is automatically and rationally analyzed, and the generated behavioral summaries can more intuitively and accurately display the proxy event behavioral data, effectively solving the problem of cumbersome and unreliable behavior data viewing operations caused by ineffective analysis of proxy event behavioral data.

[0051] This embodiment provides a behavioral data processing method that can be used in a computer system. Figure 3 This is a flowchart of a behavioral data processing method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps:

[0052] Step S301: Obtain log data of the proxy event, and parse and / or transform the log data to obtain the original sequence in a set format that represents the operation performed at a set time point, thereby obtaining the original sequence of proxy event behavior data.

[0053] Here, the behavioral data is the log data of the proxy events.

[0054] In one alternative implementation, a log parser can be used to parse and / or transform log data to obtain a raw sequence in a set format that represents the operation performed at a set time point, thereby obtaining the raw sequence of proxy event behavior data.

[0055] Specifically, for a single agent, multiple agent events such as Think, Act, and Position can be executed at a given point in time. This can be achieved by parsing system logs, converting the raw log data into a raw sequence of agent event behavior data. For example, LLMAS log files can be parsed into structured JSON format, making log data processing and analysis easier. The specific parsing process may include the following operations:

[0056] In operation a1, configure the log parser in the LLMAS runtime code, and define the format and fields of the log output when an agent event occurs, such as: agent name, the time when the original event occurred, event type, event content, etc.

[0057] Operation a2 uses the log parser to parse the LLMAS system log file and obtain the parsed log object or structure.

[0058] Operation a3 uses a JSON library to convert the parsed log object into a JSON format string, thus converting proxy event behavior data into the original sequence.

[0059] For example, see reference. Figure 4 The diagram illustrates the implementation process of a specific application example of behavioral data processing provided in this embodiment of the invention. Figure 4 The "Event Sequence Processing" of the agent events shown along the timeline is the same as the "Acquisition of the Raw Sequence of Agent Event Behavior Data" operation in this step. Here, the raw sequence of agent event behavior data obtained for the agent name "Isabella Rodrignez" includes:

[0060] “Position:Dorm Oak Hill College;

[0061] Perceive:kitchen sink is idle.Cooking area is idle.

[0062] Think:John Lin is conversing about Maria's research paper.

[0063] Act:attending an online class.Completing the class assignment.

[0064] ..."

[0065] It should be noted that the original sequence of proxy events here is the original sequence of proxy event behavior data for a single point in time.

[0066] In another alternative implementation, Natural Language Processing (NLP) technology can be directly combined to analyze the log data of proxy events. By converting the log data into structured information, the system directly outputs behavioral subsequences and behavioral summaries of these subsequences. This fully utilizes the language processing capabilities of NLP to extract and summarize key information about proxy behavior, effectively processing the proxy event behavior data to obtain re-divided behavioral subsequences and corresponding behavioral summaries. This allows the processed proxy event behavior data to be visualized more intuitively, facilitating user viewing and processing.

[0067] The behavioral data processing method provided in this embodiment obtains a raw sequence in a set format that represents the operation performed at a set time point by parsing and transforming log data. It fully parses the proxy event behavioral data to ensure the comprehensiveness and accuracy of the raw data for behavioral data processing, thereby effectively ensuring the rationality of subsequent operations.

[0068] Step S302: Aggregate the multiple operations included in the original sequence to obtain a generalized behavioral description.

[0069] For example, please refer to the following: Figure 4 It can perform behavior aggregation based on Prompt (prompt word) to aggregate multiple operations included in the original sequence of proxy event behavior data. It determines that the proxy events Position, Perceive, Think, and Act are consecutive and similar operations. After behavior aggregation, a general behavior description is obtained: "Isabella Rodriguez is attending an online class. Completing the class assignment, and observing that the kitchen sink and cooking area are idle." This general behavior description is also referred to as "Response."

[0070] The aggregation operation here can employ clustering analysis algorithms commonly used in this field, such as partitioning methods, hierarchical methods, density-based methods, grid-based methods, and model-based methods.

[0071] For other detailed implementation information, please refer to Figure 1 Step S202 of the illustrated embodiment will not be described again here.

[0072] Step S303: Convert the generalized behavioral descriptions corresponding to multiple original sequences within a defined range into numerical vectors.

[0073] In one alternative implementation, text embedding algorithms can be used to capture the latent semantics within the summarized behavioral descriptions. Specifically, the summarized behavioral descriptions can be converted into numerical vectors that represent the semantic content of the agent event behavior data and the similarity between behaviors. Here, the text embedding algorithm can be one of various implementation methods, such as the bag-of-words model, which can convert the text summarizing the behavioral descriptions into higher-dimensional numerical vectors that are easier for computers to understand and process. While preserving the semantic information of the text, it converts the text into high-dimensional numerical features, facilitating text classification and other processing.

[0074] For example, please refer to the following: Figure 4 After aggregating the behavioral data of the proxy events at each time point, the resulting Response can be processed using text embedding algorithms such as Embedding to obtain a numerical vector. For example... Figure 4 The Chinese text embedding section displays numerical vectors corresponding to the original sequence of proxy event behavior data, based on a timeline composed of multiple consecutive time points.

[0075] Step S304: Detect abrupt changes in the numerical vector using the sliding window method.

[0076] By using the sliding window method (WIN, Window-based change point detection), numerical vectors can be detected to obtain at least one abrupt change point in the behavioral sequence. For example, methods such as cosine similarity and geometric distance similarity using the sliding window method can all achieve change point detection of numerical vectors. In a preferred embodiment of the present invention, cosine similarity is used to detect change points in numerical vectors.

[0077] For example, please refer to the following: Figure 4The sliding window method is used to detect abrupt changes in the numerical vector, yielding the detection results shown in the "Change Point Detection" section. Specifically, according to... Figure 4 In the timeline from top to bottom, in the "Change Point Detection" column, there is no abrupt change between the 1st and 2nd numerical vectors, there is an abrupt change between the 2nd and 3rd numerical vectors, there is no abrupt change between the 4th to 7th numerical vectors, and there is an abrupt change between the 7th and 8th numerical vectors.

[0078] The behavioral data processing method provided in this embodiment of the invention detects abrupt changes in numerical vectors using a sliding window method, and fully utilizes the high-dimensional spatial data features of the original sequence of proxy event behavioral data to divide the original sequence of proxy event behavioral data into multiple behavioral subsequences that are more significant.

[0079] Step S305: Divide the numerical vector according to at least one mutation point to obtain behavioral subsequences.

[0080] When identifying a mutation point in a numerical vector, it indicates that the agent event behavior data before and after the mutation point are used to characterize different behavior types of the agent event. Therefore, based on at least one mutation point obtained by detecting the numerical vector, the behavior sequence can be segmented into meaningful behavior subsequences, so that the behavior subsequences represent the significant behaviors of the agent in different time periods.

[0081] For example, please refer to the following: Figure 4 , can be according to Figure 4 In the "Change Point Detection" section, the timeline is divided into numerical vectors from top to bottom. Since there is no abrupt change between the 1st and 2nd numerical vectors, the original sequences corresponding to these two vectors are grouped into a single subsequence. If there is an abrupt change between the 2nd and 3rd numerical vectors, they are split. Similarly, since there is no abrupt change between the 4th to 7th numerical vectors, the original sequences corresponding to these vectors are grouped into a single subsequence. Finally, since there is an abrupt change between the 7th and 8th numerical vectors, they are split, and the original sequence corresponding to the 8th numerical vector is grouped into a separate subsequence.

[0082] The behavioral data processing method provided in this invention first converts the generalized behavioral descriptions corresponding to multiple original sequences within a defined range into numerical vectors. Then, it detects change points on the numerical vectors to obtain at least one mutation point. Based on this mutation point, the numerical vectors are divided to obtain behavioral subsequences. Thus, by fully exploiting the similarity of multiple proxy events through high-dimensional numerical vectors, the original sequences of proxy events are reasonably segmented, resulting in behavioral subsequences that can characterize significant behaviors within different event segments.

[0083] Step S306: Use a summary generation algorithm to generate a behavior summary for the behavior subsequence. The behavior summary includes key information for visualizing the proxy event.

[0084] For example, please refer to the following: Figure 4 It can be based on the method described in step 305. Figure 4 The three behavioral subsequences shown in the "Change Point Detection" column generate behavioral descriptions such as "sleeping in the Dorm for OakHill College / sleep", "attending an online class / e-learn", and "studying in the library / study" as shown in the "Behavioral Description Generation" column. Among them, the behavioral summaries such as sleep, e-learn, and study include not only the key textual information such as sleep, e-learn, and study, but also visual elements such as sleeping expressions, computer icons to indicate online learning, and book icons to indicate learning status, which can be used to more intuitively and vividly represent the proxy event behavioral data.

[0085] The behavioral data processing method provided in this embodiment of the invention automatically processes proxy event behavioral data based on the above steps, quickly identifies key features of proxy event behavioral data, and generates behavioral summaries for behavioral subsequences based on the above operations using a summary generation algorithm, without the need for manual review and summarization. This significantly improves the visualization efficiency and understanding of proxy behavioral data for users and simplifies the process of understanding proxy event behavioral data.

[0086] This embodiment provides a behavioral data processing method that can be used in a computer system. Figure 5 This is a flowchart of a behavioral data processing method according to an embodiment of the present invention, such as... Figure 5 As shown, the process includes the following steps:

[0087] Step 501: Obtain the raw sequence of proxy event behavior data. For detailed explanation, please refer to the relevant descriptions in the above embodiments; they will not be repeated here.

[0088] Step 502: Aggregate the multiple operations included in the original sequence to obtain a generalized behavioral description. For detailed explanation, please refer to the relevant descriptions in the above embodiments; they will not be repeated here.

[0089] Step 503: Based on the generalized behavioral description, the multiple original sequences within the defined range are divided into multiple behavioral subsequences. For detailed explanations, please refer to the relevant descriptions in the above embodiments; they will not be repeated here.

[0090] Step 504: Generate a behavior summary for the behavior subsequence. The behavior summary includes key setting information for visualizing the proxy events. For detailed explanation, please refer to the relevant descriptions in the above embodiments; they will not be repeated here.

[0091] Step 505: When the set trigger condition is detected, at least a summary of the behavior is displayed in the set format.

[0092] First, it should be noted that in the above embodiments, a minimum time unit was first determined, and the proxy event behavior data within this minimum time unit was used as a raw sequence for behavior data processing within a defined range. In practical applications, proxy event behavior data can be analyzed at multiple levels. For example, behavior data processing can begin with 1 second as the minimum time unit, followed by further processing with 5 seconds, 10 seconds, 30 seconds, or 1 minute as minimum time units. Using the same processing method, multi-level behavior sub-sequences and corresponding behavior summaries can be obtained.

[0093] Therefore, the results of behavioral data processing can be displayed in a multi-layered BVV (Behavioral Visualization View) format. For details, please refer to... Figure 6 The diagram illustrates the BVV (Browser-Vehicle-View) functionality. Users can click on specific points in time on the timeline to view more detailed agent event behavior data, such as agent location, agent behavior trajectory, and analysis results of agent interactions. Thus, by displaying agent behavior trajectories through a hierarchical, continuously zoomable timeline, users can quickly identify patterns and trends in agent behavior at different time scales, effectively improving the efficiency and understanding of agent behavior data visualization. This allows users to gain a deeper understanding of agent behavior and significantly enhances the interactivity of viewing agent event behavior data.

[0094] It should be noted that Figure 6 The “behavioral event summary” shown is the “behavioral summary” described above.

[0095] Specifically, BVV may include the following components:

[0096] Hierarchical Timeline Component: The timeline is the foundation of BVV, enabling the arrangement of agent behavior events in chronological order. In this embodiment, BVV uses a hierarchical horizontal timeline to display the agent's behavioral trajectory. The timeline can be scaled according to user needs, from macro-scales such as a day or week to micro-scales such as minutes or seconds. Users can continuously adjust the zoom level of the timeline to view agent event behavior data at different time scales.

[0097] Multi-agent agent behavior trajectory display component: On the timeline, the behavior of each agent is represented as a curve, and the color and shape of the curve may change according to the agent's behavior type or state.

[0098] Interactive controls: Users can zoom in or out of the timeline using interactive controls such as sliders and buttons to view agent event behavior data at different time scales. Users can also obtain more information by clicking or hovering over the agent's behavior trajectory, such as viewing detailed behavior summaries of behavior subsequences obtained after processing multiple agent event behavior data.

[0099] Behavior Event Summary Control: At specific points on the timeline, users can view behavior summaries of multiple behavior subsequences obtained after processing proxy event behavior data. These behavior summaries can be generated based on the behavior data processing method provided in the above embodiments. Specifically, they can be combined with... Figure 7 The diagram shows the spatial layout of agents in BVV, allowing you to view the spatial mapping of agent events. For example, the agent's position in the virtual environment can be displayed using highlighting, color settings, and other customizable effects within the layout where the agent's curve is located. Figure 7 The striped rectangles in the background represent the agent's current location in the environment. As the agent moves, its curve passes through different striped rectangles, forming the agent's behavioral trajectory. BVV marks specific behavioral events (such as behavioral event summaries, dialogues with other agents, etc.) on the agent curve, using different colors and shapes to help users identify agent behavior at specific points in time.

[0100] The behavioral data processing method provided in this embodiment, upon detecting a set trigger condition, displays at least a behavioral summary in a set format. This allows for the display of the agent's behavioral trajectory through hierarchical and continuously zoomable timelines, enabling users to quickly identify patterns and trends in agent behavior at different time scales. This effectively avoids the limitations of fixed-level time-series display methods in providing a panoramic visualization of agent event behavioral data. Users can flexibly adjust the level of detail in viewing agent event behavioral data based on the timeline according to their analytical needs, thereby supporting more comprehensive agent behavior analysis and further understanding the motivations behind agent event behavior, leading to more accurate decision-making.

[0101] It should be noted that the above is only one specific implementation of BVV. In practical applications, depending on the actual needs, a graphical approach can be used to display agent event behavior data instead of a timeline. Specifically, the agent's behavior trajectory is represented directly in two-dimensional or three-dimensional space in the form of graphical elements such as points, lines, and regions. This allows for a more intuitive display of agent behavior, enabling users to understand the agent event's behavior by observing the position and changes of the graphical elements.

[0102] Step 506: Update the behavior summary when the set range and / or the original sequence changes.

[0103] As agent actions occur in real time, BVV can dynamically update the agent's location and action events to ensure that users can always view the latest agent event action data.

[0104] The behavior data processing method provided in this embodiment of the invention dynamically updates the behavior summary when the set range and / or the original sequence changes, thereby ensuring that as agent behavior occurs in real time, the user can always view the latest agent event behavior data based on the behavior summary.

[0105] This embodiment also provides a behavioral data processing apparatus for implementing the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0106] This embodiment provides a behavioral data processing device, such as... Figure 8 As shown, it includes:

[0107] The acquisition module 801 is used to acquire the raw sequence of proxy event behavior data.

[0108] The summarization module 802 is used to aggregate multiple operations included in the original sequence to obtain a summarized behavior description.

[0109] The segmentation module 803 is used to divide multiple original sequences within a defined range into multiple behavioral subsequences based on a generalized behavioral description.

[0110] Summary module 804 is used to generate a behavior summary for a behavior subsequence. The behavior summary includes key setting information for visualizing the proxy event.

[0111] In some alternative implementations, the behavioral data is log data of the proxy events.

[0112] Accordingly, the acquisition module 801 includes:

[0113] The log processing unit is used to parse and / or transform log data to obtain a raw sequence in a set format that represents the operation performed at a set time point.

[0114] In some alternative implementations, the segmentation module 803 may include:

[0115] The transformation unit is used to convert the generalized behavioral descriptions corresponding to multiple original sequences within a defined range into numerical vectors.

[0116] The change point detection unit is used to detect change points in the numerical vector and obtain at least one change point.

[0117] A partitioning unit is used to divide a numerical vector based on at least one mutation point to obtain a behavioral subsequence.

[0118] In one alternative implementation, the change point detection unit includes a sliding window subunit for detecting abrupt changes in the numerical vector using a sliding window method.

[0119] In one alternative implementation, the summarization module 804 includes a generation unit for generating a behavioral summary for a behavioral subsequence using a summarization generation algorithm.

[0120] In an optional embodiment, the apparatus further includes a display module, configured to display at least the behavior summary in a set format when a set trigger condition is detected after generating a behavior summary for the behavior subsequence.

[0121] In an optional embodiment, the above apparatus further includes an update module for updating the behavior summary when the set range and / or the original sequence changes.

[0122] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0123] In this embodiment, the behavioral data processing device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0124] This invention also provides a computer device having the above-described features. Figure 8 The behavior data processing device shown.

[0125] Please see Figure 9 , Figure 9 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 9 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 9 Take a processor 10 as an example.

[0126] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0127] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0128] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0129] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0130] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0131] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0132] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A behavioral data processing method, characterized in that, The method includes: Obtain the raw sequence of proxy event behavior data; The multiple operations included in the original sequence are aggregated to obtain a generalized behavioral description; Based on the generalized behavioral description, multiple original sequences within a defined range are divided into multiple behavioral subsequences; Generate a behavior summary for the behavior subsequence, the behavior summary including key information for visualizing the proxy event; Specifically, the step of dividing multiple original sequences within a defined range into multiple behavioral subsequences based on the generalized behavioral description includes: Convert the generalized behavioral descriptions corresponding to multiple original sequences within the defined range into numerical vectors; Change point detection is performed on the numerical vector to obtain at least one abrupt change point; The numerical vector is divided according to the at least one mutation point to obtain the behavioral subsequence.

2. The method according to claim 1, characterized in that, The behavioral data is the log data of the proxy event; correspondingly, The original sequence for obtaining proxy event behavior data includes: The log data is parsed and / or transformed to obtain a raw sequence in a set format that represents the operation performed at a set time point.

3. The method according to claim 1, characterized in that, The step of detecting change points in the numerical vector includes: The abrupt change points of the numerical vector are detected using the sliding window method.

4. The method according to claim 1, characterized in that, The generation of a behavioral summary for the behavioral subsequence includes: A summary generation algorithm is used to generate behavioral summaries for behavioral subsequences.

5. The method according to any one of claims 1 to 4, characterized in that, After generating the behavioral summary for the behavioral subsequence, the method further includes: When a set trigger condition is detected, at least a summary of the behavior is displayed in a set format.

6. The method according to any one of claims 1 to 4, characterized in that, The behavior summary is updated when the set range and / or the original sequence change.

7. A behavioral data processing device, characterized in that, The device includes: The acquisition module is used to acquire the raw sequence of proxy event behavior data; The summarization module is used to aggregate multiple operations included in the original sequence to obtain a summarized behavior description; The segmentation module is used to divide multiple original sequences within a defined range into multiple behavioral subsequences based on the generalized behavioral description. The summary module is used to generate a behavior summary for the behavior subsequence, the behavior summary including key information for visualizing the proxy event; The segmentation module includes: A conversion unit is used to convert the generalized behavioral descriptions corresponding to multiple original sequences within the defined range into numerical vectors; A change point detection unit is used to detect change points in the numerical vector to obtain at least one abrupt change point; A partitioning unit is used to partition the numerical vector according to the at least one mutation point to obtain the behavioral subsequence.

8. A computer device, characterized in that, include: A memory and a processor are communicatively connected, the memory storing computer instructions, and the processor executing the computer instructions to perform the behavioral data processing method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to perform the behavioral data processing method according to any one of claims 1 to 6.

Images