CN117350524B

CN117350524B - Novel base mapping mobile acquisition collaborative operation method and system

Info

Publication number: CN117350524B
Application number: CN202311653795.1A
Authority: CN
Inventors: 吴顺民; 莫孝周; 张潇; 张梦云; 肖观长
Original assignee: Guangdong Xinhedao Information Technology Co ltd
Current assignee: Guangdong Xinhedao Information Technology Co ltd
Priority date: 2023-12-05
Filing date: 2023-12-05
Publication date: 2024-03-26
Anticipated expiration: 2043-12-05
Also published as: CN117350524A

Abstract

According to the novel basic mapping mobile acquisition collaborative operation method and system provided by the embodiment of the invention, the task control strategy type of the collaborative operation task instruction to be processed is decided through the continuous text semantics, so that the task control strategy type of the collaborative operation task instruction to be processed can be determined, differentiated control strategies are issued and executed according to the task control strategy type, and the safe and stable operation of the novel basic mapping mobile acquisition system is ensured.

Description

Novel base mapping mobile acquisition collaborative operation method and system

Technical Field

The invention relates to the technical field of data processing, in particular to a novel base mapping mobile acquisition collaborative operation method and system.

Background

The novel basic surveying and mapping mobile acquisition collaborative operation technology is a technology for realizing the on-site surveying and mapping data acquisition of a plurality of surveying and mapping personnel and real-time collaborative operation by utilizing advanced mobile equipment and network communication technology and a collaborative work platform. The technology generally comprises the characteristics of mobile equipment, network communication, a cooperative work platform, real-time cooperative work and the like. The novel base mapping mobile acquisition collaborative operation technology has wide application range, and comprises the fields of land mapping, geographic Information System (GIS) data acquisition, city planning, facility management and the like. In the practical application process, as the intelligent communication equipment related by the technology is more, how to ensure the safe and stable operation of the novel basic mapping mobile acquisition system is one of the technical problems.

Disclosure of Invention

In order to improve the technical problems in the related art, the invention provides a novel base mapping mobile acquisition collaborative operation method and system.

In a first aspect, an embodiment of the present invention provides a novel base mapping mobile acquisition collaborative operation method, applied to an AI collaborative operation auxiliary decision-making system, the method including:

obtaining a first past collaborative task instruction; the first past collaborative job task instruction comprises a plurality of staged task texts, and the staged task texts with different task description fine granularity belong to different task description fine granularity levels;

analyzing the plurality of staged task texts through a natural language processing model to obtain a first staged task text and a second staged task text in the plurality of staged task texts; the natural language processing model comprises a local focusing variable, and the natural language processing model determines that the eccentric weight of the text of the first stage task is greater than that of the text of the second stage task according to the local focusing variable;

configuring a first discrimination annotation for the first staged task text according to the eccentric weight of the natural language processing model to the first staged task text, and configuring a second discrimination annotation for the second staged task text according to the eccentric weight of the natural language processing model to the second staged task text; the first discrimination annotation characterizes that the first staged task text belongs to a first task description fine-grained level discrimination result; the second discrimination annotation characterizes that the second stage task text belongs to a second task description fine-grained level discrimination result; the task description granularity represented by the first task description fine-grained level discrimination result is larger than the task description granularity represented by the second task description fine-grained level discrimination result;

Obtaining a first prior annotation of the first staged task text and a second prior annotation of the second staged task text; wherein the first prior annotation characterizes the first staged task text as belonging to a first prior task description fines class; the second priori annotations characterize that the second stage task text belongs to a second priori task description fine level;

optimizing the local focus variable based on a difference between the first discrimination annotation and the first prior annotation and a difference between the second discrimination annotation and the second prior annotation; the natural language processing model is used for extracting continuous text semantics of the to-be-processed collaborative work task instruction according to the optimized local focusing variable, and the continuous text semantics are used for making a decision on the task control strategy type of the to-be-processed collaborative work task instruction.

Preferably, the parsing the plurality of staged task texts through a natural language processing model to obtain a first staged task text and a second staged task text in the plurality of staged task texts includes:

determining a focusing coefficient of each staged task text in the staged task texts according to the local focusing variable through the natural language processing model; the focusing coefficient of any staged task text represents the eccentric weight of the natural language processing model to the any staged task text;

Taking the staged task text with the focusing coefficient in the first coefficient interval in the staged task texts as the first staged task text, and taking the staged task text with the focusing coefficient in the second coefficient interval in the staged task texts as the second staged task text; wherein coefficients in the first coefficient interval are greater than coefficients in the second coefficient interval.

Preferably, the natural language processing model is a sub-model of a task control strategy decision model, the optimized local focusing variable is in a fixed state in the task control strategy decision model, and the task control strategy decision model further comprises a control strategy decision branch; the method further comprises the steps of:

obtaining a second past collaborative task instruction; the second past collaborative task instruction is a collaborative task instruction obtained by identifying a past task control strategy, and the second past collaborative task instruction carries a control strategy annotation which characterizes the authenticated type of the past task control strategy;

performing feature mapping on the second past collaborative task instruction according to the optimized local focusing variable through the natural language processing model to generate past continuity text semantics of the second past collaborative task instruction;

Carrying out decision analysis on the past task control strategy according to the past continuity text semantics through the control strategy decision branch to obtain a category decision viewpoint of the past task control strategy;

and optimizing model variables except the optimized local focusing variables in a fixed state in the task control strategy decision model according to the difference between the authenticated type and the type decision viewpoint to obtain a task control strategy decision model for completing debugging.

Preferably, the number of the second past collaborative task instructions is a collaborative task instruction obtained by identifying the past task control policy from a plurality of feature dimensions, and one feature dimension is used for identifying one second past collaborative task instruction of the past task control policy; the feature mapping is performed on the second past collaborative task instruction according to the optimized local focusing variable through the natural language processing model, and the generating of the past continuity text semantics of the second past collaborative task instruction includes:

respectively carrying out feature mapping on each second past collaborative task instruction according to the optimized local focusing variable through the natural language processing model to generate a continuous text semantic vector of each second past collaborative task instruction;

And carrying out semantic stitching on the continuous text semantic vectors of the plurality of second past collaborative task instructions to generate the past continuous text semantics.

Preferably, the semantic stitching is performed on the continuous text semantic vectors of the plurality of second past collaborative task instructions, and generating the past continuous text semantic includes:

combining the continuous text semantic vectors of the plurality of second past collaborative task instructions to generate the past continuous text semantic;

or summing the continuous text semantic vectors of the second past collaborative task instructions to generate the past continuous text semantic.

Preferably, the task control strategy decision model for completing debugging comprises a natural language processing model for completing debugging and a control strategy decision branch for completing debugging; the method further comprises the steps of:

obtaining a cooperative job task instruction to be processed; the cooperative job task instruction to be processed is a cooperative job task instruction obtained by identifying a target task control strategy;

performing feature mapping on the to-be-processed collaborative operation task instruction through the debugging-completed natural language processing model, and generating target continuity text semantics of the to-be-processed collaborative operation task instruction;

And carrying out decision analysis on the target task control strategy according to the target continuity text semantics through the debugging-completed control strategy decision branch to obtain the target control strategy type of the target task control strategy.

Preferably, the number of the cooperative job task instructions to be processed is a cooperative job task instruction obtained by identifying the target task control policy from a plurality of feature dimensions, and one feature dimension is used for identifying one cooperative job task instruction to be processed of the target task control policy; the feature mapping is performed on the to-be-processed collaborative job task instruction through the debug-completed natural language processing model, and the generating of the target continuity text semantics of the to-be-processed collaborative job task instruction includes:

respectively carrying out feature mapping on each to-be-processed collaborative operation task instruction through the natural language processing model for completing debugging, and generating a continuous text semantic vector of each to-be-processed collaborative operation task instruction;

and carrying out semantic splicing on the continuous text semantic vectors of the plurality of to-be-processed collaborative job task instructions to generate the target continuous text semantic.

Preferably, the obtaining the first prior annotation of the first staged task text and the second prior annotation of the second staged task text includes:

decision analysis is carried out on the task description fine granularity of a first staged task text through a task description fine granularity judging model for completing debugging, so that the first priori task description fine granularity of the first staged task text is obtained;

performing decision analysis on the task description fine granularity of a second stage task text through the task description fine granularity judging model for completing debugging to obtain the second priori task description fine granularity of the second stage task text;

and configuring the first priori annotation for the first stage task text according to the first priori task description fine level, and configuring the second priori annotation for the second stage task text according to the second priori task description fine level.

Preferably, the method further comprises:

obtaining a third past collaborative job task instruction and a task description fine-granularity judging model to be debugged; the third past collaborative task instruction carries a task description fine-granularity annotation, and the task description fine-granularity annotation characterizes the authenticated task description fine-granularity level of the third past collaborative task instruction;

Performing decision analysis on the task description fine grain level of the third past collaborative operation task instruction through the task description fine grain discrimination model to be debugged to obtain a task description fine grain level discrimination result of the third past collaborative operation task instruction;

and optimizing model variables of the task description fine grain discrimination model to be debugged according to the difference between the authenticated task description fine grain level and the task description fine grain level discrimination result to obtain the task description fine grain discrimination model for completing debugging.

performing task text disassembly processing on the first past cooperative task instruction through a text disassembly model for completing debugging to obtain a plurality of cooperative task sub-tasks of the first past cooperative task instruction, wherein one cooperative task sub-task corresponds to a task description fine-grained level;

taking a collaborative job subtask with the highest overlapping index with the first periodic task text in the plurality of collaborative job subtasks as a first fine-granularity associated task of the first periodic task text, and configuring the first priori annotation for the first periodic task text according to a task description fine-granularity level corresponding to the first fine-granularity associated task;

Taking a collaborative job subtask with the highest overlapping index with the second stage task text in the plurality of collaborative job subtasks as a second fine-granularity associated task of the second stage task text, and configuring the second priori annotation for the second stage task text according to a task description fine-granularity level corresponding to the second fine-granularity associated task;

the first priori task description fine grain level is a task description fine grain level corresponding to the first fine grain associated task, and the second priori task description fine grain level is a task description fine grain level corresponding to the second fine grain associated task.

In a second aspect, the invention also provides an AI collaborative operation auxiliary decision-making system, which comprises a processor and a memory; the processor is in communication with the memory, and the processor is configured to read and execute a computer program from the memory to implement the method described above.

In a third aspect, the present invention also provides a computer-readable storage medium having stored thereon a program which, when executed by a processor, implements the method described above.

The invention can obtain the first past collaborative task instruction; the first past collaborative job task instruction comprises a plurality of staged task texts, and the staged task texts with different task description fine granularity belong to different task description fine granularity levels; analyzing the plurality of staged task texts through a natural language processing model to obtain a first staged task text and a second staged task text in the plurality of staged task texts; the natural language processing model comprises local focusing variables, and the natural language processing model determines that the eccentric weight of the text of the first stage task is greater than that of the text of the second stage task according to the local focusing variables; furthermore, a first discrimination annotation can be configured for the first staged task text according to the eccentric weight of the natural language processing model to the first staged task text, and a second discrimination annotation can be configured for the second staged task text according to the eccentric weight of the natural language processing model to the second staged task text; the first discrimination annotation characterizes that the first staged task text belongs to a discrimination result of the first task description fine-grained level; the second discrimination annotation characterizes the text of the second stage task as belonging to the discrimination result of the second task description fine-grained level; the task description granularity represented by the first task description fine particle level discrimination result is larger than the task description granularity represented by the second task description fine particle level discrimination result; a first prior annotation of the first staged task text and a second prior annotation of the second staged task text may also be obtained; the first priori annotation characterizes that the first staged task text belongs to a first priori task description fine level; the second priori annotations characterize that the second stage task text belongs to a second priori task description fine level; therefore, the local focusing variable can be optimized according to the difference between the first discrimination annotation and the first priori annotation and the difference between the second discrimination annotation and the second priori annotation; the natural language processing model is used for extracting continuous text semantics of the cooperative job task instruction to be processed according to the optimized local focusing variable, and the continuous text semantics are used for making decisions on task control strategy types of the cooperative job task instruction to be processed. Therefore, the design thought provided by the invention can utilize the natural language processing model to configure discrimination comments for the staged task text of the first past cooperative task instruction by utilizing the eccentric weight of the staged task text in the first past cooperative task instruction, and the task description fine granularity of the task description fine granularity level represented by the discrimination comments configured by the staged task text with higher eccentric weight can be higher, so that the distinction between the authenticated comments (such as priori comments) and the discrimination comments (such as predicted results) of the staged task text can be utilized to optimize the local focusing variable, the natural language processing model can utilize the optimized local focusing variable to configure higher eccentric weight for the staged task text with higher granularity in the cooperative task instruction to be processed, and configure lower eccentric weight for the staged task text with lower granularity in the cooperative task instruction to be processed, so that the text semantic meaning of the staged task text with higher granularity in the cooperative task instruction to be processed can be mined, the text class of the stage task description fine granularity in the cooperative task instruction to be processed can be mined, the continuous task instruction with high semantic meaning can be precisely analyzed, and the cooperative task class can be precisely controlled by utilizing the text instruction to be processed, and the cooperative task class can be precisely controlled by utilizing the cooperative task instruction to be processed.

Drawings

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

Fig. 1 is a schematic flow chart of a novel base mapping mobile acquisition collaborative operation method provided by an embodiment of the invention.

Detailed Description

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples do not represent all implementations consistent with the invention. Rather, they are merely examples of apparatus and methods consistent with aspects of the invention.

It should be noted that the terms "first," "second," and the like in the description of the present invention and the above-described drawings are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order.

The method embodiment provided by the embodiment of the invention can be executed in an AI collaborative operation auxiliary decision-making system, computer equipment or similar operation device. Taking as an example the operation on an AI collaborative job assistance decision making system, the AI collaborative job assistance decision making system may comprise one or more processors (which may include, but is not limited to, a microprocessor MCU or a processing means such as a programmable logic device FPGA) and a memory for storing data, optionally the AI collaborative job assistance decision making system may further comprise a transmission means for communication functions. It will be appreciated by those of ordinary skill in the art that the above-described structure is merely illustrative and is not intended to limit the structure of the above-described AI collaborative work aid decision-making system. For example, the AI collaborative job assistance decision making system can also include more or fewer components than shown above, or have a different configuration than shown above.

The memory may be used to store a computer program, for example, a software program of application software and a module, for example, a computer program corresponding to a novel base mapping mobile acquisition collaborative operation method in an embodiment of the present invention, and the processor executes the computer program stored in the memory, thereby executing various functional applications and data processing, that is, implementing the method described above. The memory may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid state memory. In some examples, the memory may further include memory remotely located with respect to the processor, which may be connected to the AI collaborative job assistance decision making system through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission means is used for receiving or transmitting data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of an AI collaborative job assistance decision-making system. In one example, the transmission means comprises a network adapter (Network Interface Controller, simply referred to as NIC) that can be connected to other network devices via a base station to communicate with the internet. In one example, the transmission device may be a Radio Frequency (RF) module, which is used to communicate with the internet wirelessly.

Based on this, referring to fig. 1, fig. 1 is a flow chart of a novel basic mapping mobile acquisition collaborative operation method provided in an embodiment of the present invention, where the method is applied to an AI collaborative operation auxiliary decision-making system, and further includes steps 101-105.

Step 101, obtaining a first past collaborative job task instruction.

The first past collaborative job task instruction comprises a plurality of staged task texts, and the staged task texts with different task description fine granularity belong to different task description fine granularity levels. The first past collaborative task instruction may be understood as a first collaborative task instruction sample, which is a collaborative task instruction for the novel basic survey mobile acquisition system, for implementing collaborative task control and scheduling of the novel basic survey mobile acquisition system.

Step 101 is to obtain a first past collaborative task instruction, i.e., an instruction containing a plurality of staged task texts. These task texts describe staged tasks of different levels of detail, as do the task descriptions to which each text pertains. The instruction sample is used for controlling and scheduling collaborative operations of the novel basic mapping mobile acquisition system. Briefly, step 101 is to obtain a task instruction about a collaborative job, where a plurality of task texts are included, and these texts describe staged tasks at different levels. This instruction sample is used to control and schedule collaborative operations for the new basic mapping mobile acquisition system.

An example of step 101 may be obtaining a collaborative job task instruction for a mapping task. The instruction contains the following staged task text: staged task text 1: performing a terrain survey in a designated area, including measuring ground elevation, collecting ground feature information, and establishing a fiducial point; staged task text 2: performing geographic feature recognition and image classification to determine the type and attribute of the ground object in the region; staged task text 3: performing data processing and quality control, including verifying the accuracy and consistency of the acquired data; staged task text 4: map products are generated, including digital maps, image maps, and three-dimensional models. In this example, the first past collaborative job task instruction is for a new basic survey mobile acquisition system for collaborative job control and scheduling of survey tasks. The instruction is parsed into a plurality of staged task texts, each text corresponding to a particular task stage. These tasks describe fine granularity differences, ranging from terrain surveying to terrain identification and classification, to data processing and map generation. This example illustrates the process of obtaining the first past collaborative job task instruction in step 101 and shows the content of the different staged task text.

And 102, analyzing the plurality of staged task texts through a natural language processing model to obtain a first staged task text and a second staged task text in the plurality of staged task texts.

Wherein the natural language processing model contains local focus variables (such as can be understood as attention variables), and the natural language processing model determines that the eccentric weight (such as can be understood as bias degree) of the first stage task text is greater than the eccentric weight of the second stage task text according to the local focus variables. The staged task text is part of a first past collaborative task instruction, based on which the first past collaborative task instruction may be understood as a collaborative task instruction in text form.

Step 102 is to parse the staged task text through the natural language processing model and extract the first staged task text and the second staged task text.

For example, assume that there are two staged task text: staged task text 1: performing a terrain survey in a designated area, including measuring ground elevation and collecting ground feature information; staged task text 2: image processing and feature extraction are performed to identify the clutter type and attribute.

A natural language processing model is used that determines whether the decentration weight for the first staged task text is greater than the decentration weight for the second staged task text based on local focus variables (attention variables). The eccentric weight can be understood as the degree of attention or importance of the model to a certain task text in the parsing process. In this example, if the model determines that the decentration weight of the first staged task text is higher than the decentration weight of the second staged task text, the first staged task text will be extracted as an important staged task. Specifically, the first stage task text may include contents such as "topographic survey", "ground elevation measurement" and "ground feature information collection", and the second stage task text may involve tasks such as "image processing" and "feature extraction". This example illustrates parsing the staged task text with a natural language processing model and determining the importance of the task text based on the local focus variables in step 102.

Step 103, configuring a first discrimination annotation for the first staged task text according to the eccentric weight of the natural language processing model to the first staged task text, and configuring a second discrimination annotation for the second staged task text according to the eccentric weight of the natural language processing model to the second staged task text.

The first discrimination annotation characterizes that the first staged task text belongs to a first task description fine-grained level discrimination result; the second discrimination annotation characterizes that the second stage task text belongs to a second task description fine-grained level discrimination result; the task description granularity represented by the first task description fine-grain level discrimination result is larger than the task description granularity represented by the second task description fine-grain level discrimination result. Discrimination annotations can also be understood as predictive results.

Step 103 is to configure discrimination comments for the first staged task text and the second staged task text according to the results of the natural language processing model. For example, assume that the natural language processing model identifies a first staged task text as belonging to a first task description granular level and a second staged task text as belonging to a second task description granular level. Discrimination annotations can be understood as the result of predictions of the fine granularity level to which the task text belongs. In this example, the first staged task text is labeled as a first discriminant annotation indicating that the text belongs to a higher level of task description granularity. And the second stage task text is labeled as a second discrimination annotation indicating that the text belongs to a lower level of task description granularity. Such discriminative annotations may aid in subsequent task control and decision making processes. Step 103, comparing the discrimination results of the natural language processing model on the first stage task text and the second stage task text, and respectively configuring corresponding discrimination comments for the first stage task text and the second stage task text so as to characterize the fine granularity level of the task description to which the first stage task text and the second stage task text belong.

Step 104, obtaining a first priori annotation of the first periodic task text and a second priori annotation of the second periodic task text.

Wherein the first prior annotation characterizes the first staged task text as belonging to a first prior task description fines class; the second prior annotation characterizes the second periodic task text as belonging to a second prior task description fines class. The a priori annotations can be understood as authenticated annotations, i.e. correct training debug tags.

Step 104 is to obtain a first a priori annotation of the first staged task text and a second a priori annotation of the second staged task text. For example, assume that the following a priori notes have been made: first a priori annotation: determining that the first staged task text belongs to high-level task description granularity according to task requirements and previous experience; second prior annotation: based on task requirements and previous experience, it is determined that the second-stage task text belongs to a low-level fine granularity of task descriptions. In this example, the first staged task text is labeled as a first a priori annotation, indicating that it belongs to a high level of task description granularity. And the second stage task text is labeled as a second prior comment indicating that it belongs to a low level of task description granularity. These prior annotations are based on predictions and labels made of prior knowledge and experience of the task. A priori annotations of the first and second staged task text may be obtained, via step 104, which may be used in subsequent task control and scheduling processes to better understand and process the text at different task description fine granularity levels.

Step 105, optimizing the local focus variable according to the difference between the first discrimination annotation and the first prior annotation and the difference between the second discrimination annotation and the second prior annotation.

The natural language processing model is used for extracting continuous text semantics of the to-be-processed collaborative job task instruction according to the optimized local focusing variable, and the continuous text semantics are used for deciding the type of the task control strategy of the to-be-processed collaborative job task instruction. Further, the continuous text semantics can be understood as embedded text semantics, and the task control policy types of the collaborative job task instructions to be processed can be determined by deciding the task control policy types of the collaborative job task instructions to be processed through the continuous text semantics, so that differentiated control policy issuing and executing can be performed according to the task control policy types, and safe and stable operation of the novel basic surveying and mapping mobile acquisition system is ensured.

In step 105, by optimizing the local focus variables, the system may extract the continuous text semantics of the collaborative job task instruction to be processed. This means that the system will focus on critical information that is critical to the task execution and make decisions based on that information. The process of optimizing local focus variables involves using natural language processing models to adjust these variables to increase the focus on different staged task text. Specifically, the system determines whether the eccentricity for the first staged task text is higher than the eccentricity for the second staged task text based on the attention weights (local focus variables) in the natural language processing model. The optimized local focusing variables enable the system to capture important information in task instructions, such as keywords, action requirements, time constraints and the like, more accurately. These continuity text semantics help determine the task control policy category to which the collaborative job task instruction to be processed belongs. According to the decision result of the task control strategy type, the system can adopt corresponding differentiated control strategies to process different types of tasks. This includes strategies to decide the order of execution of tasks, resource allocation, personnel scheduling, etc. Through issuing and executing the differentiated control strategy, the novel basic mapping mobile acquisition system can better cope with different task demands and ensure safe and stable operation of the system. Thus, the detailed objective of step 105 is to extract the continuous text semantics by optimizing the local focus variables to decide on task control strategy categories and implement differentiated control strategy delivery and execution to improve the performance and operational quality of the new basic mapping mobile acquisition system.

Applying steps 101-105, first, a first past collaborative job task instruction is obtained, which includes staged task text describing tasks of different fine granularity. These texts are then parsed using a natural language processing model to extract task texts of the first and second stages. Next, it is determined that the first stage task text is weighted higher than the second stage task text based on local focus variables in the natural language processing model. And using the weight information to configure a first discrimination annotation for the first-stage task text and a second discrimination annotation for the second-stage task text. The discrimination annotation characterizes the fine granularity level of the task description to which the task text belongs. A priori annotations of the first-stage task text and the second-stage task text are then obtained, the annotations representing a priori task description fine granularity levels of the task text. The local focus variable is optimized based on the differences between the first discrimination annotation and the first prior annotation and the second discrimination annotation and the second prior annotation. By doing so, the attention of the natural language processing model can be adjusted, so that the continuous text semantics of the task instructions to be processed can be better extracted. Finally, extracting continuous text semantics through the optimized local focusing variables, and deciding a task control strategy of a task instruction to be processed according to the semantics. Therefore, the type of the task control strategy to which the task instruction belongs can be determined, and differentiated control strategies are executed based on the type, so that safe and stable operation of the novel basic mapping mobile acquisition system is ensured. In the whole, the process optimizes the analysis of the task instruction and the task control strategy by using the natural language processing model and the priori information, improves the efficiency and the accuracy of the collaborative task, and ensures the safe and stable operation of the system.

In some optional embodiments, the parsing the plurality of staged task texts by the natural language processing model in step 102 obtains a first staged task text and a second staged task text in the plurality of staged task texts, including steps 1021-1022.

And 1021, determining a focusing coefficient of each staged task text in the staged task texts according to the local focusing variable through the natural language processing model.

The focusing coefficient of any staged task text characterizes the eccentric weight of the natural language processing model to the any staged task text.

Step 1022, using the staged task text with the focusing coefficient in the first coefficient interval in the staged task texts as the first staged task text, and using the staged task text with the focusing coefficient in the second coefficient interval in the staged task texts as the second staged task text.

Wherein coefficients in the first coefficient interval are greater than coefficients in the second coefficient interval.

Assume that there is a series of staged task text, each text representing a different staged task. The text is parsed using a natural language processing model, and a focus factor for each staged task text is determined from the local focus variables. The focus factor characterizes the decentration weight of the model for each stepwise task text, i.e. the degree to which the model focuses on a certain task. Illustrating: assume that there are three staged task texts, "task 1", "task 2" and "task 3", respectively. The focusing coefficients for each task text may be determined by a natural language processing model. Assume that the focusing factor of "task 1" is 0.8, the focusing factor of "task 2" is 0.5, and the focusing factor of "task 3" is 0.6. These coefficients represent the degree of attention of the model to the different tasks.

In step 1022, the staged task text will be divided into a first staged task text and a second staged task text according to the focus factor. Two coefficient sections are set, and coefficients in the first coefficient section are larger than those in the second coefficient section. Illustrating: according to the above example of the focusing coefficient, the stepwise task text in the first coefficient section has a focusing coefficient of 0.7 or more, that is, the focusing coefficient of "task 1" is 0.8. Thus, "task 1" will be categorized as first staged task text. And (3) in the second coefficient interval, the focusing coefficients of the stepwise task text are smaller than 0.7, namely, the focusing coefficients of the task 2 and the task 3 are respectively 0.5 and 0.6. They will be categorized as second stage task text.

Through these steps, the staged task text can be parsed using a natural language processing model and divided into different staged tasks according to focusing coefficients. This has the advantage that the priority or importance of the task can be determined according to the degree of interest of the model. Organizing task text in this manner may help one better understand and manage complex multi-stage tasks and improve efficiency and accuracy of task execution.

In some examples, the natural language processing model is a sub-model of a task control policy decision model, the optimized local focus variable is in a fixed state (i.e., parameter lock) in the task control policy decision model, and the task control policy decision model further includes a control policy decision branch. Based on this, the method further comprises steps 201-204.

Step 201, obtaining a second past collaborative job task instruction.

The second past collaborative task instruction is a collaborative task instruction obtained by identifying a past task control policy, the second past collaborative task instruction carries a control policy annotation, the control policy annotation characterizes an authenticated type of the past task control policy, and the control policy annotation can be further understood as a control policy type tag.

Based on step 201, a second past collaborative job task instruction is obtained. This means that a new instance of the previously identified collaborative job task instruction has been collected. Unlike the previous instruction, this instruction is accompanied by a control strategy annotation. The control policy annotation provides information about the previous task control policy, including its authenticated category or label.

And 202, performing feature mapping on the second past collaborative task instruction according to the optimized local focusing variable through the natural language processing model, and generating past continuity text semantics of the second past collaborative task instruction.

Wherein the feature mapping may be an embedding process.

Based on step 202, a feature map is performed on the second past collaborative job task instruction using the natural language processing model and the optimized local focus variables. The feature mapping may be implemented by way of an embedding process. Through this process, the second instruction is converted into a continuous text semantic representation so that the model can better understand the meaning and content of the instruction.

And 203, performing decision analysis on the past task control strategy according to the past continuity text semantics through the control strategy decision branch to obtain a category decision viewpoint of the past task control strategy.

The decision analysis may be to classify and predict past task control strategies to obtain a classification prediction viewpoint (classification decision viewpoint).

Based on step 203, a control strategy decision branch is used to make a decision analysis of past task control strategies. Based on the continuous text semantics of the past instructions, a decision analysis is performed in the model to determine which class of task control strategy the instruction belongs to. This decision analysis process can be seen as a classification prediction process on past task control strategies to get a classification decision perspective about the instruction.

And 204, optimizing model variables except the optimized local focusing variables in a fixed state in the task control strategy decision model according to the difference between the authenticated type and the type decision point to obtain a task control strategy decision model for completing debugging.

Based on step 204, other model variables in the task control strategy decision model, except for the local focus variable, are optimized according to the differences between the authenticated category and the category decision perspective. By comparing the authenticated species with the model's class predictions, it can be determined which model variables may need to be adjusted or optimized to more accurately match the authenticated species. Through the optimization process, a debugged task control strategy decision model can be obtained, the performance is better, and task instructions can be accurately identified and classified.

In general, the steps combine the information of past task instructions, control strategy annotation and model optimization techniques, so that a task control strategy decision model can better understand and interpret task instructions, and the accuracy and performance of the model are improved.

Further, the feature mapping (embedding process) has the following benefits in the above method: semantic representation extraction: the second past collaborative job task instruction is converted to a continuous text semantic representation through feature mapping. This semantic representation captures key information and meaning in the instruction, enabling the model to better understand and infer the intent and requirements of the instruction; context continuity modeling: since the textual semantic representation generated by the feature map is continuous, it helps the model establish contextual continuity in the instructions. This means that the model can better understand the association and dependency of the various parts in the instruction, thereby interpreting the meaning of the whole instruction more accurately; abstract representation learning: through the embedding process, the feature map extracts high-dimensional semantic features from the original text. These features are encoded in a low dimensional space to form a dense vector representation. The abstract representation learning enables the model to better capture semantic similarity and relevance in the instruction, so that classification and decision analysis can be performed more accurately; generalization capability is improved: the embedding process may convert different types of instructions into a unified semantic space representation. This enables the model to be better generalized to new, unseen instruction instances. By learning the shared semantic features, the model can use prior knowledge and experience to interpret new instructions and perform accurate decision analysis.

In summary, feature mapping (embedding process) advantageously provides semantic representation extraction, context continuity modeling, abstract representation learning, and generalization capability promotion. These effects help the task control strategy decision model to better understand task instructions and based thereon make more accurate decisions and predictions.

In some examples, the number of second past collaborative task instructions is a collaborative task instruction obtained by identifying the past task control policy from a number of feature dimensions, one feature dimension being used to identify one second past collaborative task instruction of the past task control policy. Based on this, the feature mapping is performed on the second past collaborative task instruction according to the optimized local focus variable by the natural language processing model in step 202, so as to generate the past continuity text semantics of the second past collaborative task instruction, which includes step 2021 and step 2022.

Step 2021, performing feature mapping on each second past collaborative task instruction according to the optimized local focus variable through the natural language processing model, so as to generate a continuous text semantic vector of each second past collaborative task instruction.

Step 2022, performing semantic stitching on the continuous text semantic vectors of the plurality of second past collaborative task instructions, and generating the past continuous text semantics.

In step 2021, feature mapping is performed on each second past collaborative task instruction by the natural language processing model and the optimized local focus variables. This step can be further broken down into the following examples: for example, assume that there are two second past collaborative job task instructions, instruction A and instruction B, respectively. And performing feature mapping on the instruction A by using the natural language processing model and the optimized local focusing variable to generate a continuous text semantic vector of the instruction A. Similarly, feature mapping is performed on the instruction B, and continuous text semantic vectors of the instruction B are generated.

In step 2022, the continuous text semantic vectors of the second plurality of past collaborative task instructions are semantically stitched to generate past continuous text semantics. The following is one example: assume that there are two consecutive text semantic vectors of the second past collaborative job instruction, vector a and vector B, respectively. In step 2022, these vectors are semantically stitched, which may be simple vector join or other stitching, to form past continuous text semantics.

By way of the above example explanation of steps 2021 and 2022, the following benefits may be obtained: each step 2021 of feature mapping generates a continuous text semantic vector for each second past collaborative job task instruction. This has the advantage that key semantic features can be extracted from each instruction and expressed in vector form. This allows the model to better understand the meaning and content of each instruction. The semantic stitching in step 2022 combines the successive text semantic vectors of the plurality of instructions together to generate past successive text semantics. By stitching these vectors, correlations and dependencies between multiple instructions can be comprehensively considered, helping the model to better understand the context information of the overall instruction.

Collectively, steps 2021 and 2022 provide a rich, continuous text semantic representation that enables the model to more accurately understand and process past collaborative task instructions. The semantic representation of the continuity is beneficial to better capturing the relevance and the dependence among instructions by the model, and improving the accuracy and the performance of the task control strategy decision model. Step 2021 and step 2022 combine feature mapping and semantic stitching to provide rich continuous text semantic representations that help the model better understand and process multiple past collaborative job task instructions, thereby improving performance of the task control strategy decision model.

In some examples, the semantic stitching of the continuous text semantic vector of the number of second past collaborative task instructions in step 2022 generates the past continuous text semantic, including step 20221 or step 20222.

Step 20221, combining the continuous text semantic vectors of the plurality of second past collaborative task instructions to generate the past continuous text semantic.

Step 20222, summing the continuous text semantic vectors of the plurality of second past collaborative task instructions to generate the past continuous text semantic.

In step 20221, the continuous text semantic vectors of the second plurality of past collaborative task instructions are combined to generate past continuous text semantics. The following is one example: assume that there are three consecutive text semantic vectors of the second past collaborative job instruction, vector a, vector B, and vector C, respectively. In step 20221, these vectors are combined, and the past continuous text semantics can be formed by vector stitching, average pooling, or the like. For example, the past continuous text semantic vector is obtained by stitching vector a, vector B, and vector C.

In step 20222, the continuous text semantic vectors of the second plurality of past collaborative task instructions are summed to generate past continuous text semantics. The following is one example: assume that there are two consecutive text semantic vectors of the second past collaborative job instruction, vector a and vector B, respectively. In step 20222, these vectors are summed to obtain the past continuous text semantic vector. For example, the past continuous text semantic vector is obtained by adding the vector A and the vector B element by element.

By way of the above example explanation of step 20221 and step 20222, the following benefits may be obtained: (1) combining semantic representations: the combination of step 20221 and the summation of step 20222 may aggregate the continuous text semantic vectors of the plurality of instructions to form a more comprehensive and comprehensive past continuous text semantic. Such a combined semantic representation can better capture relevant information and overall features between instructions; (2) contextual modeling: by combining or summing successive text semantic vectors of multiple instructions, the model can better understand and model the contextual relationships between instructions. The context modeling is helpful for understanding the meaning and the requirement of the instruction from the overall view, and the accuracy and the robustness of the task control strategy decision model are improved; (3) information fusion: the combining or summing operation enables the model to comprehensively consider contributions of multiple instructions, avoiding information loss or deviation. The continuous text semantic vectors of different instructions are fused to generate past continuous text semantic, the model can more comprehensively understand the instructions, and the decision capability of a task control strategy is improved. In summary, steps 20221 and 20222 provide a comprehensive, global past continuity text semantic representation through a combination or summation operation. The semantic representation can better capture the associated information between instructions, model the context and promote information fusion, so that the performance and effect of the task control strategy decision model are improved.

In some examples, the task control policy decision model to complete debugging includes a natural language processing model to complete debugging and a control policy decision branch to complete debugging. Based on this, the method further comprises steps 301-303.

Step 301, obtaining a cooperative job task instruction to be processed.

The cooperative job task instruction to be processed is a cooperative job task instruction obtained by identifying a target task control strategy.

And 302, performing feature mapping on the to-be-processed collaborative job task instruction through the natural language processing model for completing debugging, and generating target continuity text semantics of the to-be-processed collaborative job task instruction.

And 303, performing decision analysis on the target task control strategy according to the target continuity text semantics through the control strategy decision branch for completing debugging to obtain the target control strategy type of the target task control strategy.

Assume that for collaborative operation of a novel basic mapping mobile acquisition system, the following collaborative operation task instructions to be processed exist: performing a terrain survey and recording data; carrying out geographic feature recognition and image classification, and labeling elements such as roads, buildings and the like; processing and quality control are carried out on the acquired data, so that the accuracy and consistency of the data are ensured; map products are generated, including digital maps, elevation models, three-dimensional visualizations, and the like. These instructions cover different task steps and operational requirements.

Through the completion of the debugged natural language processing model, feature mapping may be performed on the collaborative job task instructions to be processed in step 302 to generate the target continuity text semantics. For example: "performing a terrain survey and recording data" may be mapped to target continuity text semantics, expressed as corresponding vector or semantic representations. The target continuity text semantics can be obtained through mapping by carrying out geographic feature recognition and image classification and labeling elements such as roads, buildings and the like. Likewise, corresponding target continuity text semantics may be generated for both data processing and quality control as well as instructions to generate map products.

Finally, in step 303, the control policy decision branch that completes the debugging performs decision analysis according to these target continuity text semantics to obtain a suitable target control policy class. For example: for instructions for terrain surveying and data recording, the target control strategy category may involve deployment and parameter setting of the acquisition instrument. The instructions for geographic feature identification and image classification may require selection and optimization of related algorithms and models. Instructions for data processing and quality control may require the creation of control strategies in terms of data pipelines, writing data processing scripts, quality assessment methods, and the like. The instructions for generating the map product may take into account control strategies in terms of data rendering, product output format, etc.

Through the above example, steps 301 to 303 can process complex collaborative job task instructions to be processed, and provide more accurate and effective text semantic representation of target continuity and target control strategy decision with the help of completing the debugged natural language processing model and control strategy decision branch, which is helpful for realizing efficient collaborative operation of the basic mapping mobile acquisition system.

Under some possible design ideas, the to-be-processed cooperative job task instructions are a plurality of, and the to-be-processed cooperative job task instructions are cooperative job task instructions obtained by identifying the target task control strategy from a plurality of feature dimensions, wherein one feature dimension is used for identifying one to-be-processed cooperative job task instruction of the target task control strategy. Based on this, the feature mapping is performed on the to-be-processed collaborative job task instruction by the debug-completed natural language processing model in step 302, so as to generate the target continuity text semantics of the to-be-processed collaborative job task instruction, which includes steps 3021 to 3022.

And 3021, performing feature mapping on each to-be-processed collaborative job task instruction through the debug-completed natural language processing model, and generating a continuous text semantic vector of each to-be-processed collaborative job task instruction.

And 3022, performing semantic stitching on the continuous text semantic vectors of the plurality of to-be-processed collaborative job task instructions to generate the target continuous text semantic.

In step 3021, feature mapping is performed on each of the to-be-processed collaborative task instructions through the natural language processing model that completes the debugging, so as to generate a continuous text semantic vector of each of the to-be-processed collaborative task instructions. The following is one example: assume that there are two to-be-processed collaborative job task instructions: detecting and repairing a fault of the topographic survey equipment; and (5) running an image classification algorithm to identify roads and buildings.

Each instruction can be mapped into a corresponding continuous text semantic vector by completing the debugged natural language processing model. For example, for the first instruction, the generated continuity text semantic vector may represent the meaning of the aspects of equipment maintenance and troubleshooting. For the second instruction, the generated continuous text semantic vector may relate to meaning in terms of image processing, machine learning algorithms, and object recognition.

In step 3022, semantic stitching is performed on the continuous text semantic vectors of the plurality of collaborative job task instructions to be processed, so as to generate a target continuous text semantic. The following is one example: in step 3021, two continuous text semantic vectors are generated for two collaborative job task instructions to be processed. In step 3022, the two vectors may be semantically concatenated to generate a vector representing the semantics of the target continuity text.

For example, the continuous text semantic vector of the first instruction and the continuous text semantic vector of the second instruction are spliced to obtain a target continuous text semantic vector comprising aspects of equipment maintenance, fault investigation, image processing, machine learning algorithm, target recognition and the like.

As can be seen, step 3021 performs feature mapping on each of the collaborative job task instructions to be processed by the natural language processing model that completes the debugging, converting them into continuous text semantic vectors. Doing so helps capture semantic information and contextual meaning of the instruction. Step 3022, performing semantic stitching on the continuous text semantic vectors of the plurality of collaborative job task instructions to be processed, so as to generate a target continuous text semantic vector. By concatenating semantic information of different instructions, a more comprehensive and comprehensive semantic representation of the target continuity text may be formed. In summary, by feature mapping and semantic stitching of steps 3021 and 3022, a more accurate and comprehensive text semantic representation of the target continuity can be generated. This helps to enhance the understanding and decision process of collaborative job task instructions and provides a more efficient basis for subsequent target control strategy selection.

In some exemplary embodiments, the obtaining of the first prior annotation of the first staged task text and the second prior annotation of the second staged task text in step 104 comprises steps 1041-1043.

Step 1041, performing decision analysis on the task description fine granularity of a first staged task text through a task description fine granularity discriminating model for completing debugging, so as to obtain the first priori task description fine granularity of the first staged task text.

Step 1042, performing decision analysis on the task description fine granularity of the second stage task text through the task description fine granularity discriminating model for completing debugging, so as to obtain the second priori task description fine granularity level of the second stage task text.

Step 1043, configuring the first priori annotation for the first stage task text according to the first priori task description fine grain level, and configuring the second priori annotation for the second stage task text according to the second priori task description fine grain level.

With respect to step 1041: and carrying out decision analysis on the task description fine granularity of the first staged task text through a task description fine granularity judging model for completing debugging to obtain a first priori task description fine granularity level of the first staged task text. The following is one example: assume that the first periodic task text is "conduct a terrain survey and record data". And through completing the debugged task description fine grain discrimination model, decision analysis can be carried out on the task description to obtain a first priori task description fine grain grade. In this example, the first priori task description that may be available has a fine-grained level of "high granularity," meaning that the task description provides detailed and comprehensive information.

In step 1042, decision analysis is performed on the task description granularity of the second stage task text by completing the debugged task description granularity discrimination model to obtain a second priori task description granularity level of the second stage task text. The following is one example: assuming that the second phase task text is "run image classification algorithm, roads and buildings are identified. And through completing the debugged task description fine grain discrimination model, decision analysis can be carried out on the task description to obtain a second priori task description fine grain grade. In this example, a second priori task description may be available with a fine granularity level of "medium granularity," indicating that the task description provides a degree of detail and guidance.

With respect to step 1043, a first prior annotation is configured for a first staged task text based on a first prior task description fines level and a second prior annotation is configured for a second staged task text based on a second prior task description fines level. The following is one example: according to the above example, in case the first priori task description fine grain level is "high granularity", the relevant first priori annotations, such as "detailed recording of topographic survey data", may be configured for the first periodic task text, ensuring accuracy and consistency. In the case where the second prior task description has a fine-grained level of "medium granularity", a related second prior annotation may be configured for the second stage task text, such as "identify roads and buildings using an appropriate image classification algorithm, extract accurate features".

By implementing the steps 1041 to 1043, the following advantages can be obtained: (1) a priori task description fine granularity level: step 1041 and step 1042 utilize the fine-granularity discrimination model of the task description after completing the debugging to perform decision analysis on the first stage task text and the second stage task text, and obtain the corresponding fine-granularity level of the task description. This helps to understand the details and requirements of the task; (2) a priori annotation configuration: step 1043 configures a first priori annotation for the first staged task text based on the first priori task description granularity level and configures a second priori annotation for the second staged task text based on the second priori task description granularity level. This helps to provide corresponding instructions and prompts for each task to meet the task needs and achieve the desired results.

In summary, through implementation of steps 1041 to 1043, it is possible to analyze and make decisions on fine granularity of task descriptions, and provide appropriate prior comments for each staged task, so as to improve task understanding and execution effects in the collaborative process.

The method further comprises, among other possible design considerations, steps 401-403.

Step 401, obtaining a third past collaborative job task instruction and a task description fine granularity discrimination model to be debugged.

The third past collaborative task instruction carries a task description fine-granularity annotation, and the task description fine-granularity annotation characterizes the authenticated task description fine-granularity level of the third past collaborative task instruction.

Under the application scene of the novel basic mapping mobile acquisition system, two previous collaborative operation task instructions are recorded, wherein the two previous collaborative operation task instructions are a first past collaborative operation task instruction and a second past collaborative operation task instruction respectively. A third past collaborative job task instruction is now received and a model for fine-grained discrimination of task descriptions is prepared.

And step 402, performing decision analysis on the task description fine grain level of the third past collaborative operation task instruction through the task description fine grain discrimination model to be debugged, and obtaining a task description fine grain level discrimination result of the third past collaborative operation task instruction.

And inputting a third past collaborative job task instruction into a task description fine granularity discrimination model to be debugged. The model analyzes the task description and evaluates the task description according to a predefined standard, so as to obtain a task description fine-grained level discrimination result of the third past collaborative work task instruction.

Step 403, optimizing model variables of the task description fine grain discrimination model to be debugged according to the difference between the authenticated task description fine grain level and the task description fine grain level discrimination result to obtain the task description fine grain discrimination model for completing debugging.

And comparing the judging result of the task description fine grain judging model to be debugged with the authenticated task description fine grain grade. By analyzing the differences between them, model variables in the model that need to be adjusted and optimized can be determined. For example, if the model deviates in discriminating the third past collaborative task instruction, the weight of the model may be modified or more training data may be added to improve the performance of the model. The model variables are continuously optimized, so that a task description fine-granularity judging model for completing debugging can be finally obtained, and the task description detail level of the third past collaborative work task instruction can be accurately judged.

In summary, the design concept of steps 401-403 is beneficial to fine-grained discrimination of task descriptions in the novel basic mapping mobile acquisition system. Through analysis and optimization of past collaborative job task instructions, the system can more accurately understand and evaluate details of task descriptions, so that collaborative job efficiency and quality are improved.

In some further alternative embodiments, the obtaining of the first prior annotation of the first staged task text and the second prior annotation of the second staged task text in step 104 includes steps 104 a-104 c.

Step 104a, performing task text disassembly processing on the first past cooperative task instruction through a text disassembly model for completing debugging to obtain a plurality of cooperative task sub-tasks of the first past cooperative task instruction, wherein one cooperative task sub-task corresponds to one task description fine-grained level.

Assume that a text disassembly model for completing debugging is provided for disassembling the first past collaborative task instruction. The model is applied to a first past collaborative job task instruction, and a plurality of collaborative job subtasks are obtained after processing, wherein each subtask represents different task description fine particle levels.

Step 104b, using a collaborative job subtask with the highest overlapping index with the first periodic task text in the plurality of collaborative job subtasks as a first fine-granularity associated task of the first periodic task text, and configuring the first priori annotation for the first periodic task text according to a task description fine-granularity level corresponding to the first fine-granularity associated task.

And comparing the similarity of the plurality of cooperative job subtasks and the first periodic task text, and selecting the subtask with the highest overlapping index with the first periodic task text as a first fine-granularity associated task. Then, a first priori annotation is configured for the first staged task text according to the task description fine level corresponding to the associated task. The associated task description fine-grained level information may thus be applied to the first staged task text, providing a priori task description guidance.

And 104c, using a collaborative job subtask with the highest overlapping index with the second stage task text in the plurality of collaborative job subtasks as a second fine granularity associated task of the second stage task text, and configuring the second priori annotation for the second stage task text according to the task description fine granularity level corresponding to the second fine granularity associated task.

And comparing the similarity of the plurality of cooperative job subtasks and the second periodic task text, and selecting the subtask with the highest overlapping index with the second periodic task text as a second fine-granularity associated task. And then, configuring a second priori annotation for the second stage task text according to the task description fine-grained level corresponding to the associated task. This may utilize descriptive fine-grained level information of the associated task to assist in understanding and processing the second-stage task text.

By using the text disassembly model with complete debugging and the calculation of the overlap index, the complex past collaborative task instruction can be decomposed into a plurality of collaborative task subtasks and associated with the staged task text. This helps to better understand and process complex task instructions; by configuring the first and second prior annotations for the staged task text, information from the relevant task description fine-grained level may be introduced into the staged task, providing prior task description guidance. This facilitates accurate understanding and execution of the staged tasks; by means of task disassembly and association, the task processing flow can be optimized, so that more effective division cooperation is realized, and accurate task description guidance is obtained in tasks in different stages.

In summary, the steps 104 a-104 c provide an effective task processing method through task disassembly, association and prior annotation, and fine granularity division and prior guidance introduction of task description are realized through task disassembly, association and prior annotation, so that task understanding accuracy is improved, and information redundancy and confusion are reduced.

Further, there is also provided a computer-readable storage medium having stored thereon a program which, when executed by a processor, implements the above-described method.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus and method embodiments described above are merely illustrative, for example, flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

In addition, functional modules in the embodiments of the present invention may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part.

The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a network device, or the like) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes. It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A novel basic mapping mobile acquisition collaborative operation method, which is characterized by being applied to an AI collaborative operation auxiliary decision-making system, the method comprising:

Optimizing the local focus variable based on a difference between the first discrimination annotation and the first prior annotation and a difference between the second discrimination annotation and the second prior annotation; the natural language processing model is used for extracting continuous text semantics of the to-be-processed collaborative operation task instruction according to the optimized local focusing variable, and the continuous text semantics are used for making a decision on the task control strategy type of the to-be-processed collaborative operation task instruction;

the natural language processing model is a sub-model of a task control strategy decision model, the optimized local focusing variable is in a fixed state in the task control strategy decision model, and the task control strategy decision model also comprises a control strategy decision branch; the method further comprises the steps of:

2. The method according to claim 1, wherein parsing the plurality of staged task texts through a natural language processing model to obtain a first staged task text and a second staged task text in the plurality of staged task texts includes:

3. The method of claim 1, wherein the number of second past collaborative task instructions is a number of collaborative task instructions identified from a number of feature dimensions for the past task control policy, one feature dimension being used to identify one second past collaborative task instruction of the past task control policy; the feature mapping is performed on the second past collaborative task instruction according to the optimized local focusing variable through the natural language processing model, and the generating of the past continuity text semantics of the second past collaborative task instruction includes:

4. The method of claim 3, wherein said semantically concatenating the continuous text semantic vectors of the number of second past collaborative task instructions to generate the past continuous text semantics comprises:

5. The method of claim 1, wherein the task control policy decision model to complete debugging includes a natural language processing model to complete debugging and a control policy decision branch to complete debugging; the method further comprises the steps of:

the decision analysis is carried out on the target task control strategy according to the target continuity text semantics through the control strategy decision branch which completes debugging, so as to obtain the target control strategy type of the target task control strategy;

the method comprises the steps of determining a target task control strategy, wherein a plurality of to-be-processed cooperative job task instructions are provided, the to-be-processed cooperative job task instructions are cooperative job task instructions obtained by identifying the target task control strategy from a plurality of characteristic dimensions, and one characteristic dimension is used for identifying one to-be-processed cooperative job task instruction of the target task control strategy; the feature mapping is performed on the to-be-processed collaborative job task instruction through the debug-completed natural language processing model, and the generating of the target continuity text semantics of the to-be-processed collaborative job task instruction includes:

6. The method of claim 1, wherein the obtaining a first a priori annotation of the first staged task text and a second a priori annotation of the second staged task text comprises:

configuring the first priori annotations for the first periodic task text according to the first priori task description fine-grained level, and configuring the second priori annotations for the second periodic task text according to the second priori task description fine-grained level;

wherein the method further comprises:

7. The method of claim 1, wherein the obtaining a first a priori annotation of the first staged task text and a second a priori annotation of the second staged task text comprises:

8. An AI collaborative operation auxiliary decision-making system is characterized by comprising a processor and a memory; the processor is communicatively connected to the memory, the processor being configured to read a computer program from the memory and execute the computer program to implement the method of any of claims 1-7.

9. A computer readable storage medium, characterized in that a program is stored thereon, which program, when being executed by a processor, implements the method of any of claims 1-7.