A system and method for behavioral data collection based on time non-computable windows and one-way non-playback constrained execution graphs

Abstract

This invention discloses a behavioral data acquisition system and method based on a time-incalculable window and a one-way non-replayable constraint execution graph. The system includes a behavioral execution module, a behavioral acquisition module, a one-way non-replayable write cache module, a time-incalculable window module, a behavioral data reconstruction module, and a data output module. The time-incalculable window module uses a PCB / TCB non-creation mechanism to ensure that the reconstruction task remains in an uninstantiated state at the scheduling layer within a preset time window T. The one-way non-replayable write cache module uses a non-remapping storage strategy to prevent the data physical address from entering the read path mapping table. Behavioral event data is transmitted according to NREG, implemented collaboratively by the task scheduling module and the storage subsystem. This invention achieves non-repeatable consumption and time-isolated sampling of behavioral data through three layers of physical constraints: the OS layer, the storage layer, and the scheduling layer.

Description

Technical Field

[0001] This invention relates to the field of computer data acquisition and human-computer interaction technology, specifically to a behavioral data acquisition system and method based on a time-incomputable window and a one-way non-replay constrained execution graph (NREG) structure. Background Technology

[0002] Existing behavioral data acquisition systems mainly include message queues (MQ), log systems (WAL / Append-onlyLog), and event streaming systems.

[0003] The above-mentioned systems typically have the following characteristics:

[0004] Data can be consumed multiple times (multi-consumer semantics)

[0005] Supports historical event replay mechanism.

[0006] Supports state recovery based on offset or cursor

[0007] The system allows for the reconstruction of execution status through logs (event sourcing).

[0008] Therefore, the existing technology is essentially a replayable event computing system.

[0009] However, in behavioral sampling and human-computer interaction analysis scenarios, such systems have a structural problem: the “collection process” and the “computation process” of behavioral data are coupled in time, and the data semantics allow for repeated interpretation and reconsumption, thereby introducing behavioral bias. Summary of the Invention

[0010] (a) Technical problems to be solved

[0011] Provide a behavioral data acquisition system that enables behavioral data to meet the following requirements:

[0012] Time-isolated sampling

[0013] One-way non-replayable structure

[0014] Consumption-free semantic data stream

[0015] Posterior structural reconstruction capability

[0016] (II) Technical Solution

[0017] 1. System Structure Composition

[0018] This system includes:

[0019] Behavior Execution Module

[0020] Behavior collection module

[0021] One-way non-replayable write cache module

[0022] Time-incalculable window module

[0023] Behavioral data reconstruction module

[0024] Data output module

[0025] 2. One-way non-playback write structure (core creative point)

[0026] The behavioral data flow path is as follows:

[0027]

[0028] This cache module has the following structural features:

[0029] Only append-only write is supported.

[0030] No consumption confirmation mechanism is provided (no ACK semantics)

[0031] No subscription semantics provided.

[0032] No offset-based access mechanism is provided.

[0033] No replay interface is supported.

[0034] Its essence is: Non-Replayable Behavioral Log Structure.

[0035] Storage layer non-remapping constraint:

[0036] The storage subsystem implements a non-remapped storage policy for written data blocks, ensuring that their physical storage addresses are not entered into the read path mapping table after writing, thereby implementing a non-replay constraint at the storage layer. This policy ensures that written data cannot be addressed by subsequent read operations at the system level, cutting off the replay path from the perspective of physical storage resource allocation.

[0037] Differences from existing WAL / MQ systems:

[0038] Does it support playback? support Not supported Does consumer semantics exist? exist Does not exist Allow repeated reads? allow Not allowed Does it support state recovery? support Not supported Is the storage address remapable? yes No (cannot be remapped)

[0039] 3. Time-incalculable window structure (core creative point)

[0040] Let the set of behavioral events be:

[0041] Define the refactoring function:

[0042] Within the time window T, the following conditions must be met:

[0043] f(E,t) is in an "uninstantiated state" at the system scheduling layer.

[0044] Right now:

[0045] When t∈T, there is no instance of an executable refactoring task.

[0046] When t ∉ T, the system allows refactoring tasks to be instantiated and executed.

[0047] Key differences:

[0048] It's not "prohibited from execution" (rule / logic prohibition).

[0049] The issue is "the scheduling layer has not generated an execution instance" (structural constraint).

[0050] OS layer PCB / TCB mechanism implementation:

[0051] The refactoring task is in an uninstantiated state at the operating system scheduling layer within the time window T, specifically achieved through a mechanism that prevents the creation of Process Control Blocks (PCBs) and Thread Control Blocks (TCBs). Within this time window, the operating system does not have a task descriptor corresponding to the refactoring task, and the scheduler cannot recognize the existence of the task, thus failing to trigger CPU scheduling execution.

[0052] This mechanism guarantees the non-executability of the refactoring task within the time window T at the operating system kernel level, rather than relying on application-layer logic to prohibit it.

[0053] 4. Unidirectional Non-Replayable Execution Graph (NREG) – Collaborative Engineering Implementation Version

[0054] definition:

[0055]

[0056] in:

[0057] V: Set of behavioral nodes

[0058] E: Set of directed execution edges

[0059] Constraints:

[0060] (1) One-way constraint

[0061] (2) No loop constraint means there is no arbitrary closed-loop path.

[0062] (3) Single write constraint: Each node is allowed only one write operation.

[0063] (4) Once a non-replay constraint node enters the execution path, it is not allowed to be scheduled into the historical execution chain again (i.e., it cannot be reached again).

[0064] Project implementation method:

[0065] The unidirectional non-replayable execution graph (NREG) is implemented in collaboration between the computer processor's task scheduling module and the storage subsystem.

[0066] Specifically:

[0067] The task scheduling module is responsible for controlling the instantiation timing of refactoring tasks. It ensures that refactoring tasks are in an uninstantiated state within the time window T by using a PCB / TCB non-creation mechanism.

[0068] The storage subsystem is responsible for controlling the non-replayability of data writes, ensuring through a non-remapping storage policy that the physical address of data written to a file does not enter the read path mapping table.

[0069] Through the coordinated execution of the scheduling unit and the storage access control mechanism, NREG implements a one-way non-replay constraint on behavioral data at the engineering level, covering three physical constraints: OS layer (PCB / TCB), storage layer (non-remapping), and scheduling layer (uninstantiated).

[0070] 5. Data processing order constraints

[0071] Only allowed path:

[0072]

[0073] And satisfy:

[0074] No writeback path

[0075] No consumption cycle

[0076] No historical access path

[0077] (III) System Structure Description

[0078] Behavior Execution Module Generate behavioral events Application layer Behavior collection module Asynchronous collection of behavioral data Application layer One-way non-replayable write cache module Append-only, non-remapped storage storage layer Time-incalculable window module PCB / TCB not created, scheduler has no task descriptor OS layer Behavioral data reconstruction module Structural restoration is performed only after the window ends. Application layer Data output module Output reconstruction results Application layer

[0079] (iv) Core structural features

[0080] Time-uncalculable window The scheduling layer did not instantiate the execution task. OS layer (PCB / TCB) One-way non-replayable structure Non-consumer log structure Storage layer (non-remapping) NREG Execution Chart Acyclic unidirectional scheduling dependency structure Scheduling layer + storage layer collaboration

[0081] (v) Technical effects

[0082] This invention achieves:

[0083] Behavioral data cannot be consumed repeatedly.

[0084] Behavioral data cannot be replayed

[0085] Behavioral sampling and computation processes are time-isolated.

[0086] Reduce feedback interference in behavior analysis

[0087] Provide a stable posterior structural reconstruction path

[0088] Three layers of physical constraints guarantee:

[0089] OS layer: PCB / TCB not created, scheduler has no task descriptor.

[0090] Storage layer: Non-remapped storage policy, physical addresses do not enter the read path.

[0091] Scheduling layer: Controlling the timing of task instantiation Attached Figure Description

[0092] Figure 1 System overall structure diagram It displays the unidirectional transmission path of behavioral event data from the execution module to the output module, as well as the uninstantiated state of the reconstruction module within the time window T. Figure 2 NREG One-Way Loopless Execution Graph Structure Diagram This demonstrates the unidirectional non-loop constraint and unreachable characteristics of behavior nodes. Figure 3 Time-incalculable window scheduling state diagram The display shows the uninstantiated state of the reconstruction task within the time window T at the scheduling layer and the instantiated state after the window ends. Figure 4 Behavioral data lifecycle diagram It demonstrates the complete lifecycle of behavioral events from generation, collection, writing, time window freezing to post-reconstruction and output, as well as the dual cut-off paths of the storage layer and the OS layer. Figure 5 Collaborative architecture diagram of task scheduling module and storage subsystem This demonstrates the engineering mechanism by which the two work together to achieve NREG constraints. Figure 6 Comparison diagram of traditional MQ / WAL structure and the structure of this invention This invention demonstrates structural differences in replay mechanism, consumption semantics, offset support, and other aspects.

[0093] Appendix Number Reference Table

[0094] 101 Behavior Execution Module 102 Behavior collection module 103 One-way non-replayable write cache module 104 Time window T 105 Behavioral data reconstruction module 106 Data output module 201-204 NREG constraints 301-307 Time window scheduling status 401-408 Lifecycle Nodes 501-511 Task scheduling module and storage subsystem 601-616 Comparison chart marking

Claims

1. A behavioral data acquisition system, characterized in that, include: Behavior execution module; Behavior collection module; The one-way non-replayable write cache module is used to append behavioral event data to storage, and does not provide a consumption confirmation mechanism, a subscription mechanism, or a historical replay access interface; the storage subsystem executes a non-remapped storage strategy on the written data blocks, so that their physical storage addresses do not enter the read path mapping table after being written. The time-uncalculable window module is used to control the scheduling state of the reconstruction task, so that the behavior data reconstruction task is in an uninstantiated state at the scheduling layer within a preset time window T; the reconstruction task is implemented within the time window T through a process control block (PCB) and thread control block (TCB) non-creation mechanism, so that the operating system scheduler does not have a corresponding task descriptor. The behavioral data reconstruction module is used to reconstruct the behavioral event data in a structured manner after the end of the time window T. Data output module; in: The behavioral event data is transmitted in the system according to a one-way non-replayable execution graph (NREG); The unidirectional non-replayable execution graph (NREG) is implemented in collaboration between the computer processor's task scheduling module and the storage subsystem. NREG satisfies the one-way non-loop constraint and the single-write constraint. Once a node enters the execution path, it is not allowed to be scheduled back into the historical execution chain. Furthermore, there is no write-back path or state reference path from the data output module to the behavior execution module in the system.

2. The system according to claim 1, characterized in that: The preset time window T is a fixed or dynamic range between 30 seconds and 30 minutes.

3. The system according to claim 1, characterized in that: The behavioral event data includes mouse input data, keyboard input data, touch input data, and system interaction event data.

4. The system according to claim 1, characterized in that: The one-way non-replayable write cache module is implemented based on an append write queue.

5. A method for collecting behavioral data, characterized in that, include: S1: Execute user interaction behaviors and generate behavior event data; S2: Asynchronously collect behavioral event data; S3: Store the behavior event data in the one-way non-replayable write cache module in an append-only manner, and execute a non-remapping storage strategy to prevent the physical storage address from entering the read path mapping table. S4: By using the time-incalculable window module, the behavioral data reconstruction task is kept in an uninstantiated state at the scheduling layer within the preset time window T through the PCB / TCB non-creation mechanism. S5: Instantiate the reconstruction task after the time window T ends and perform structured reconstruction on the behavioral event data; S6: Output the reconstructed behavioral data.

6. The method according to claim 5, characterized in that: The preset time window T is 30 seconds to 30 minutes.

7. The method according to claim 5, characterized in that: The structured reconstruction restores behavioral events based on timestamp order.

8. A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method of claim 5.

9. The system according to claim 1, characterized in that: The one-way non-replayable write cache module does not provide a consumption confirmation mechanism, a subscription mechanism, or a historical replay access interface.

10. The system according to claim 1, characterized in that: There is no data write-back path between the behavior execution module and the data output module in the system.

11. The system according to claim 1, characterized in that: The processing of the behavior data reconstruction module does not generate runtime input to the behavior execution module.

Images