CN116388838A - Multi-scene multi-user multi-service intention translation method for satellite communication network - Google Patents
Multi-scene multi-user multi-service intention translation method for satellite communication network Download PDFInfo
- Publication number
- CN116388838A CN116388838A CN202310249795.9A CN202310249795A CN116388838A CN 116388838 A CN116388838 A CN 116388838A CN 202310249795 A CN202310249795 A CN 202310249795A CN 116388838 A CN116388838 A CN 116388838A
- Authority
- CN
- China
- Prior art keywords
- intention
- user
- network
- service
- end user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004891 communication Methods 0.000 title claims abstract description 78
- 238000013519 translation Methods 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000000605 extraction Methods 0.000 claims abstract description 59
- 238000013145 classification model Methods 0.000 claims abstract description 22
- 238000007781 pre-processing Methods 0.000 claims abstract description 16
- 238000013507 mapping Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 13
- 238000011049 filling Methods 0.000 claims description 12
- 238000003058 natural language processing Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 230000005540 biological transmission Effects 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000012937 correction Methods 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000007635 classification algorithm Methods 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 2
- 238000012790 confirmation Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 230000002452 interceptive effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012958 reprocessing Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 239000005437 stratosphere Substances 0.000 description 1
- 230000026676 system process Effects 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B7/00—Radio transmission systems, i.e. using radiation field
- H04B7/14—Relay systems
- H04B7/15—Active relay systems
- H04B7/185—Space-based or airborne stations; Stations for satellite systems
- H04B7/1851—Systems using a satellite or space-based relay
- H04B7/18519—Operations control, administration or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
- G06F40/295—Named entity recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Biomedical Technology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Aviation & Aerospace Engineering (AREA)
- Astronomy & Astrophysics (AREA)
- Databases & Information Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Radio Relay Systems (AREA)
Abstract
The invention discloses an intention translation method for multiple scenes, multiple users and multiple services of a satellite communication network. The method solves the problem of low resource utilization rate caused by large limitation and various types of user input. The realization includes that an intention recognition module is constructed; front-end user business intention input and preprocessing; the scene, the user and the service are identified to obtain a classification result; extracting to obtain an intention information extraction result; storing original business intention, and dynamically learning a model; enabling intent translation for a front-end user; intent translation for multi-scenario, multi-user and multi-service in a satellite communication network is achieved. According to the invention, the intention classification model is added in the intention recognition module, network strategies meeting different service quality requirements can be provided for front-end users through coarse-granularity and fine-granularity network strategies, and the satisfaction degree and experience of the users are improved. The coarse granularity fills the default value of the fine granularity policy, guaranteeing the stability and reliability of intent translation. Intent translation for multi-scenario, multi-user, multi-service in satellite communication networks.
Description
Technical Field
The invention belongs to the technical field of satellite communication networks, mainly relates to user intention translation, and particularly discloses a multi-scene multi-user multi-service intention translation method for a satellite communication network, which is used for driving the satellite network communication field.
Background
The satellite network is a networked system which takes a space platform (such as a geosynchronous satellite or a medium-low orbit satellite, a stratosphere balloon, an airplane and the like) as a carrier and combines ground network nodes to finish the tasks of acquisition, pretreatment, transmission and reprocessing of the satellite, and can support deep space exploration upwards, earth observation downwards and the like. As one of the basic facilities of the national key development, the satellite plays a key role in military, has great influence on the production and life style of human beings, more and more users and more complex services are accessed into a satellite network, the satellite network enters the era of reliable transmission and diversified information, the satellite network communication effectively overcomes the defect of ground communication, and plays a role in television broadcasting, global communication, offshore rescue, telemedicine and the like.
However, the satellite network facing global service needs to provide various types of services for different users, the various services have different requirements on network resources, and meanwhile, corresponding network management is more and more complicated, and the traditional network management has high manual participation and high error rate.
Secondly, the existing satellite network constructed according to specific task demands adopts mutually isolated and independent network systems to meet different task demands; for example, earth observation satellite systems and long-range communication satellite systems cannot be used with each other; even for the same type of task, is implemented by multiple sub-networks (e.g., an ambient satellite system and a resource satellite system) that are isolated from each other. Meanwhile, at present, satellites in China generally adopt a mode of overhead transmission, and data are difficult to transmit in real time. In summary, the closed partition of network resources, repeated laying of network facilities and data overhead transmission make it difficult to efficiently share limited network resources.
In summary, the challenges of the existing satellite network resource management are:
(1) The satellite network architecture is various, the network management is complex, and the error rate of the traditional manual-centered network management mode is high;
(2) Compared with limited satellite network resources, traffic information traffic increases dramatically, and the problem of satellite network resources scarcity makes it difficult to meet the demands of current tasks for resource diversity.
The advent of the intent driven network (Intent Driven Network, IDN) provides a new idea to address the challenges described above. The intention driven network is an automatic network integrating the functions of applying intention deep mining capability, network state global perception capability and network configuration real-time optimization capability, and is programmable and customizable. The intent is a declarative description of the system state. It abstracts the objects and capabilities of the network from the perspective of demand and can translate into high-level policies. In the intention driven network, the target network state can be realized by automatic conversion, verification, deployment, configuration and optimization according to the intention of an operator. Meanwhile, the abnormal event is automatically solved by means of network holographic sensing and feedback optimization closed loop, and the reliability of the network is guaranteed. The intention driven network plays a vital role in the evolution of satellite networks and the like.
The current satellite network resource management and control mainly depends on operation and maintenance personnel to specify network strategies through low-level languages, and common users without network technology cannot effectively manage network states. The intent translation system provides network operators and general users with a way to convert the task intent expressed by the user in natural language into a standard intent expression recognizable by the network, which is the first ring to implement an intent driven satellite network.
The intention translation method solves the problem of multi-intention input of users under the condition of multi-user multi-service, and is suitable for the intention translation of users with network related expertise and non-professional users. Currently, in operator networks, there have been related studies on intent translation methods, however, they are all studies on specific types of intent under specific scenarios, specific users, and specific services, and do not conduct intent translation studies on multi-scenario, multi-user, and multi-service scenarios. The prior method for intent translation is studied as follows
In the prior art, the user intention is acquired through an interactive interface, and related information about network bandwidth, time delay, endpoints and the like in the user intention is directly acquired by using named entity recognition in natural language processing. However, the method has the defects that all users and services are viewed simultaneously, the different requirements of different users and services are not considered, and the specific service requirements of specific users cannot be flexibly processed.
The second prior art is to provide a set of templates to the user through a graphical user interface to specify his user intent based on network scope and intent, the templates guiding the user in filling out information by displaying which network attributes are necessary to set, which are conditional, and which are optional. The disadvantage is that the use of graphical interfaces and template tools may limit the user to specify some possible options that are not provided, reducing flexibility while consuming more time
The third prior art identifies user intent by deploying a set of APIs (applications) in the northbound interface of an SDN (software defined network) environment, the scope of this intent emphasizes mainly connection class intent, supporting point-to-point or point-to-multipoint connections. But it is mainly directed to application developers, with a small audience and intended to cover a single area.
The prior art has more limitation on user input, and makes all users and services at the same time, does not consider the difference requirements of different users and services, cannot flexibly process the specific service requirements of specific users in specific scenes, and cannot realize reasonable allocation of resources.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a method for translating intention of a satellite communication network multi-scene multi-user multi-service with less limitation on the form of user input intention and finer division of scenes and services.
The invention relates to a method for translating intention of multiple scenes, multiple users and multiple services in a satellite communication network, which is characterized in that the intention of original service of a user is received at the front end through a webpage interface, and the intention of the original service of the user is processed at the rear end through a plurality of modules; the system comprises a plurality of modules, a user module and a user module, wherein the modules comprise a text preprocessing module, an intention recognition module and a strategy mapping module, a network strategy is finally obtained, and the network strategy is used for translating the intention of a user in a satellite communication network; the intent translation method for the multi-scene multi-user multi-service of the satellite communication network comprises the following steps:
step 1, constructing an intention recognition module; the built intention recognition module receives the original business intention input by the front-end user and sequentially inputs the original business intention to the intention classification model and the intention extraction model, the two model recognition results are output as a whole, and the correct recognition result is output; the specifications that the front-end user business intention input into the model needs to meet are: intercepting or filling the intention text with a specified length and converting the intention text into a digital index; the intention classification model is constructed based on a text classification algorithm in natural language processing and is used for identifying the scene, the user and the service type of the front-end user in multiple scenes, multiple users and multiple services; the intention extraction model is constructed based on an entity recognition algorithm in natural language processing and is used for extracting key network parameter information in front-end user business intention;
step 2, inputting and preprocessing the business intention of the front-end user; the front-end user inputs the original business intention in a natural language form through a webpage interface, carries out preprocessing operation on the original business intention text input by the front-end user, and specifically comprises error detection, error correction, filling or truncation operation of the original business intention text and conversion into a digital index, so that the business intention text originally input by the front-end user accords with the input specification of the intention recognition module and is input into the intention recognition module;
step 3, identifying scenes, users and services to obtain classification results; simultaneously identifying the scene to which the front-end user business intention belongs, the user and the specific type of the business by using an intention classification model in the intention identification module to obtain an identification result; judging the identification result to realize the specific service processing of the specific user in the specific scene; the identification result is that the non-specific scene, the user and the service are processed according to a default mode or a user-specified mode;
step 4, extracting the business intention information of the front-end user to obtain an intention information extraction result; after classifying all front-end user service intentions, extracting key network parameter information by using an intent extraction model, wherein the key network parameter information comprises bandwidth, time delay, duration, spatial position, jitter and network transmission speed parameters, and feeding back a service intention classification result and an intention information extraction result to a webpage interface for the front-end user to confirm or modify so as to obtain a correct identification result confirmed by the front-end user;
step 5, storing the original business intention of the front-end user so as to facilitate the dynamic learning of the intention classification and intention extraction model; storing the original input business intention of the front-end user and the correct recognition result confirmed by the user into a database so as to facilitate the dynamic learning of the intention classification and extraction model;
step 6, realizing intent translation of the front-end user; the correct identification result confirmed by the front-end user is fed back to the front-end webpage interface for the front-end user to store, and meanwhile, the correct identification result is input to a strategy mapping module constructed based on a finite state machine to generate a network strategy, so that the network strategy mapping is completed; the network strategy is used for the intention translation aiming at the front-end user in the satellite communication network, and the original service intention translation of the front-end user in the multi-scene multi-user multi-service of the satellite communication network is realized.
The invention also provides a method for realizing the multi-scenario multi-user multi-service intention translation of the satellite communication network according to any one of claims 1 to 5, which is characterized in that aiming at the problems of scarce resources, weak dynamic adjustment capability of resources, various user types, service demands and network resource mismatch and deficiency of the satellite communication network, the method for realizing the multi-scenario multi-service intention translation of the satellite communication network can identify the communication scenario, the user type and the service type to which the service belongs, so as to meet the demands of different users on different services under different scenarios and realize the efficient utilization of limited satellite network resources.
The invention solves the problem that the prior art has larger limit on user input, and is suitable for professional users and non-professional users; secondly, the invention provides a method for combining coarse and fine granularity strategies, which solves the problem of low resource utilization rate caused by a plurality of scenes, users and service types in a satellite communication network, improves the resource utilization rate and improves user experience.
Compared with the prior art, the invention has the following advantages:
less restrictions on user input: the intention translation method provided by the invention adds the intention classification model for the first time in the intention recognition part, so that a front-end user does not need to describe the service intention in detail when inputting, namely, the scene, the user and the service type can be recognized by the intention recognition model without embodying service quality index parameters such as bandwidth, time delay, duration and the like, the specific service processing of the specific user in the specific scene is realized, and the coarse-granularity strategy is generated by mapping by the strategy mapping module.
The method has stronger mobility and can be quickly migrated to other types of networks: the intention classification and the intention extraction model in the intention recognition module are both based on the pre-training language model Alber fine-tuning of the transfer learning, so that the method has stronger mobility and can be quickly transferred to other types of networks.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following description will briefly explain the drawings needed in the embodiments of the present application, and it is obvious that the drawings described below are only some embodiments of the present application, and that other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart diagram of a multi-scenario, multi-user, multi-service intent translation method for a satellite communication network of the present invention;
FIG. 2 is a schematic diagram of a system framework for multi-scenario, multi-user, multi-service intent translation for a satellite communication network according to the present invention.
FIG. 3 is a block diagram of a multi-tasking module architecture employed by the intent classification model of the present invention.
Detailed Description
Example 1
In order to make the objects, technical schemes and advantages of the present invention more clear, the following describes the multi-scenario multi-user multi-service intention translation method of the satellite communication network. It should be understood that the detailed description is intended to illustrate the invention and not to limit the invention.
The prior art has more limitation on user input, and makes all users and services at the same time, does not consider the difference requirements of different users and services, cannot flexibly process the specific service requirements of specific users in specific scenes, and cannot realize reasonable allocation of resources.
In view of the problems existing in the prior art, the present invention provides a method for multi-user and multi-service intent translation in a satellite communication network, and the present invention is described in detail below with reference to the accompanying drawings.
The invention relates to a method for translating intent of multiple scenes, multiple users and multiple services in a satellite communication network, referring to fig. 1, fig. 1 is a flow chart of the method for translating intent of multiple scenes, multiple users and multiple services in the satellite communication network, wherein the front end of a translation system receives original service intent of a user through a web page interface, and the rear end of the translation system processes the original service intent of the user through a plurality of modules; the modules comprise a text preprocessing module, an intention recognition module and a strategy mapping module, network strategies are finally obtained, and the network strategies are used for translating user intention in a satellite communication network. The intention translation method of the satellite communication network multi-scene multi-user multi-service comprises the following steps:
step 1, constructing an intention recognition module; the built intention recognition module receives the original business intention input by the front-end user and sequentially inputs the original business intention to the intention classification model and the intention extraction model, the two model recognition results are output as a whole, and the correct recognition result is output; the specifications that the front-end user business intention input into the model needs to meet are: intercepting or filling the intention text with a specified length and converting the intention text into a digital index; the intention classification model is constructed based on a text classification algorithm in natural language processing and is used for identifying the scene, the user and the service type of the front-end user in multiple scenes, multiple users and multiple services; the intention extraction model is constructed based on an entity recognition algorithm in natural language processing and is used for extracting key network parameter information in front-end user business intention.
Step 2, inputting and preprocessing the business intention of the front-end user; the front-end user inputs the original business intention in the form of natural language through a webpage interface, carries out preprocessing operation on the text of the original business intention input by the front-end user, and specifically comprises error detection, error correction, filling or truncation operation of the text of the original business intention, and converts the text into a digital index, so that the text of the business intention originally input by the front-end user accords with the input specification of the intention recognition module and inputs the text of the business intention into the intention recognition module.
Step 3, identifying scenes, users and services by using the constructed intention classification model to obtain classification results; simultaneously identifying the scene to which the front-end user business intention belongs, the user and the specific type of the business by using an intention classification model in the intention identification module to obtain an identification result; judging the identification result to realize the specific service processing of the specific user in the specific scene; the identification result is that the non-specific scene, the user and the service are processed according to a default mode or a user-specified mode; and obtaining the classification results of the intention of the user, the specific type of the service and the scene to which the user belongs after being processed by the specific service or processed by the appointed mode.
Step 4, extracting the business intention information of the front-end user by using the constructed intention extraction model to obtain an intention information extraction result; after classifying all front-end user service intentions, extracting key network parameter information by using an intent extraction model, specifically including bandwidth, time delay, duration, spatial position, jitter and network transmission speed parameters, and feeding back a service intention classification result and an intention information extraction result to a webpage interface for the front-end user to confirm or modify so as to obtain a correct identification result confirmed by the front-end user.
Step 5, storing the original business intention of the front-end user so as to facilitate the dynamic learning of the intention classification and intention extraction model; and storing the original input business intention of the front-end user and the correct recognition result confirmed by the user into a database so as to dynamically learn the intention classification and extraction model and improve the recognition accuracy of the intention classification and the intention extraction model.
Step 6, realizing intent translation of the front-end user; the correct identification result confirmed by the front-end user is fed back to the front-end webpage interface for the front-end user to store, and meanwhile, the correct identification result is input to a strategy mapping module constructed based on a finite state machine to generate a network strategy, so that the network strategy mapping is completed; the network strategy is used for the intention translation aiming at the front-end user in the satellite communication network, and the original service intention translation of the front-end user in the multi-scene multi-user multi-service of the satellite communication network is realized.
The invention has the following thought: firstly, constructing an intention recognition module, wherein the module consists of an intention classification model and an intention extraction model and is used for recognizing the scene, the user and the service type of a front-end user and extracting key network parameter information in the service intention; secondly, preprocessing operation is carried out on the original business intention input by the front-end user, including text error detection, error correction, filling or cutting, and the original business intention is converted into a digital index so as to enable the original business intention to accord with the input specification of the intention recognition module; after preprocessing is completed, the intention of the front-end user business is identified by using the intention classification model, and the specific business processing of the specific user in the specific scene is carried out according to the identification result; after intention classification, extracting key network parameter information from all classified service intents by using an intention extraction model, and feeding back a classification result and an information extraction result to a front-end user for confirmation or modification; after the intention extraction is completed, the original business intention and the correct recognition result of the front-end user are stored, so that the intention classification and the intention extraction model are subjected to dynamic learning, and the recognition precision is improved; finally, the intention translation of the front-end user is realized, and the correct identification result is input to the strategy mapping module to generate a network strategy for the original business intention translation of the front-end user in the satellite communication network.
The technical scheme and the means of the invention are as follows: the intent recognition module is constructed using an open source deep learning framework PyTorch and based on text classification algorithms and entity extraction algorithms in natural language processing. The system framework is constructed using a DJango framework, where the front end is built using HTML and the back end stores user intent using MySQL data.
The invention has the technical effects that: and realizing the automatic identification of the service intention of the front-end user. The method comprises the steps that an intention recognition module is built, an original business intention text input by a front-end user is converted into a digital index, and scene, user type and business type of the user are recognized through an intention classification model and an intention extraction model, so that automatic recognition of the business intention of the front-end user is realized; the automatic translation of the business intention of the front-end user is realized. By constructing the strategy mapping module, the service intention classification result output by the intention recognition module is mapped to a corresponding network strategy, so that the automatic translation of the front-end user service intention is realized, and the communication efficiency and the user experience in the satellite communication network are improved.
Example 2
The method for translating intention of multi-user and multi-service in satellite communication network is the same as that in embodiment 1, in the step 3 of the invention, scene, user and service are identified to obtain classification results, and the intention classification model in the intention identification module adopts a multi-task model based on a large-scale pre-training model Albert and a one-dimensional convolution model textCNN as a sharing layer, so that the scene, user and service types are identified simultaneously to obtain classification results. The invention has high recognition accuracy by using a large-scale pre-training language model Albert, which is a powerful natural language processing model, has a large amount of training data and deep neural network structure, can provide more accurate prediction results, and the textCNN can extract key features through a convolution layer, thereby being beneficial to recognizing scenes, users and service types; the reusability and the expandability of the model are improved, based on the implementation mode of the shared layer multitasking model, the identification problems of scenes, users and service types can be regarded as a whole, the problems are processed simultaneously, the parameters and the calculated amount of the model are reduced, the reusability and the expandability of the model are improved, and the model is applicable to different application scenes such as multiple scenes, multiple users, multiple services and the like.
Example 3
The method for multi-user multi-service intention translation of the satellite communication network is the same as that of the embodiment 1-2, and the intention extraction model in the step 4 is constructed based on a large-scale pre-training language model Albert and a conditional random field CRF (Conditional Random Field, CRF); in the extraction process, the user intention is represented by using a quadruple of < field, operation, object and result >; wherein < field > includes: communication scene, user type, service type; the < operation > includes: uploading and downloading, deleting and adding; the < object > includes network nodes, traffic flows, network resources; the < result > includes performance metrics, expected states, space-time constraints. The invention uses a large-scale pre-training language model Albert and a conditional random field CRF, so that the intent extraction is more accurate and fine. Meanwhile, the user intention is represented by using the quadruple of the < field, operation, object and result >, and the requirement and intention of the user can be better understood. The accuracy and the efficiency of the intention translation are improved, the user experience and the satisfaction degree can be improved, and the more intelligent intention translation of the satellite communication network multi-user and multi-service is realized.
Example 4
The method for multi-user multi-service intention translation of the satellite communication network is the same as that of the embodiments 1-3, and the method for multi-user multi-service intention translation of the satellite communication network is characterized in that the intention information of the front-end user is extracted in the step 4 of the invention to obtain an intention information extraction result, wherein a user intention text is marked by using a named entity recognition algorithm based on a BIOES marking method in intention extraction, and user service intention key information is accurately extracted. The method utilizes the BIOES labeling method and combines a named entity recognition algorithm to label the user intention text more accurately, and improves the accuracy of intention extraction.
Example 5
In the method for implementing intent translation of front-end users in the multi-user multi-service of satellite communication network according to the embodiment 1-4, in step 6 of the present invention, the policy mapping module generates network policies, specifically, the intent classification result is used for mapping to generate coarse-grained network policies, which are a set of network policies provided by the system for meeting the basic service quality requirements of front-end users. The intention extraction result is used for mapping and generating a fine-granularity network strategy, and is generated according to the original business intention of a front-end user, when a certain index of the fine-granularity network strategy has a default value, filling and complementing are carried out by a corresponding item of the coarse-granularity network strategy, so that the translation of the original business intention of the user in the satellite communication network multi-scene multi-user multi-business is completed. The invention can provide network strategies meeting different service quality demands for front-end users by combining coarse-granularity and fine-granularity network strategies, thereby improving the satisfaction degree and experience of the users. Secondly, the coarse-granularity network strategy is used for filling default values in the fine-granularity network strategy, so that the situation that the fine-granularity network strategy cannot be generated due to the fact that a certain index is missing is avoided, and the stability and the reliability of intent translation are guaranteed.
EXAMPLE 6
The method for translating the intention of the satellite communication network multi-scene multi-user multi-service aims at the problems of scarce resources, weak dynamic adjustment capability of resources, various user types, service demands and network resource mismatch and deficiency of the satellite communication network multi-scene multi-service intention translation method is similar to embodiments 1-5, and the method for translating the intention of the satellite communication network multi-user multi-service intention can identify the communication scene, the user type and the service type to which the service belongs so as to meet the demands of different users on different services under different scenes and realize the efficient utilization of limited satellite network resources. The multi-user multi-service intention translation method of the satellite communication network can identify the communication scene, the user type and the service type to which the service belongs, realize the specific processing of specific scenes, users and specific services and avoid 'thousands of people' side. Aiming at the problems of scarce resources, weak dynamic adjustment capability, various user types, service demands and network resource mismatch and deficiency of the satellite communication network, the invention meets the demands of different users on different services under different scenes by identifying the communication scene, the user type and the service type to which the service belongs, realizes the efficient utilization of limited satellite network resources and improves the user experience.
The invention belongs to the technical field of satellite communication networks, and discloses an intention translation method for multiple scenes, multiple users and multiple services of a satellite communication network. The user inputs the original business intention through the front-end webpage interface, and at the same time, the original business intention text is preprocessed and sent to the rear end for further processing; scene, user and business identification is carried out by sending the business intention text to the intention classification model; inputting the user business intention into an intention extraction model to extract key information in the user intention; feeding back the intention classification result and the intention extraction result to user confirmation or modification, and storing the user confirmation and identification correct result to a database for subsequent model dynamic learning; and finally, feeding the identification result back to the webpage interface, and transmitting the identification result to the strategy mapping module for parameter mapping in a JSON file format to generate a network strategy. The invention provides an efficient intention translation method by deeply analyzing the air interface characteristics, service types, service priorities and service quality requirements of users and various satellite stations on the network, constructing scenes, users and service models, and laying a foundation for an intention driving network technology in a satellite communication network.
Example 7
The method for multi-user multi-service intention translation in a satellite communication network is similar to embodiments 1-6, as shown in fig. 1, fig. 1 is a flow chart of the method for multi-scenario multi-user multi-service intention translation in a satellite communication network of the present invention. The invention provides a satellite communication network multi-scene multi-user multi-service intention translation method, which comprises the following steps:
s101: front-end user business intention input and preprocessing; the front-end user inputs the original business intention in a natural language form through a webpage interface, carries out preprocessing operation on the original business intention text input by the front-end user, and specifically comprises error detection, error correction, filling or truncation operation of the original business intention text and conversion into a digital index, so that the business intention text originally input by the front-end user accords with the input specification of the intention recognition module and is input into the intention recognition module;
s102: identifying scenes, users and services to obtain classification results; simultaneously identifying the scene to which the front-end user business intention belongs, the user and the specific type of the business by using an intention classification model in the intention identification module to obtain an identification result; judging the identification result to realize the specific service processing of the specific user in the specific scene; the identification result is that the non-specific scene, the user and the service are processed according to a default mode or a user-specified mode;
s103: extracting front-end user business intention information to obtain an intention information extraction result; after classifying all front-end user service intentions, extracting key network parameter information by using an intent extraction model, wherein the key network parameter information comprises bandwidth, time delay, duration, spatial position, jitter and network transmission speed parameters, and feeding back a service intention classification result and an intention information extraction result to a webpage interface for the front-end user to confirm or modify so as to obtain a correct identification result confirmed by the front-end user;
s104: storing front-end user original business intents so as to facilitate intention classification and intention extraction model dynamic learning; storing the original input business intention of the front-end user and the correct recognition result confirmed by the user into a database so as to facilitate the dynamic learning of the intention classification and extraction model;
s105: enabling intent translation for a front-end user; the correct identification result confirmed by the front-end user is fed back to the front-end webpage interface for the front-end user to store, and meanwhile, the correct identification result is input to a strategy mapping module constructed based on a finite state machine to generate a network strategy, so that the network strategy mapping is completed; the network strategy is used for the intention translation aiming at the front-end user in the satellite communication network, and the original service intention translation of the front-end user in the multi-scene multi-user multi-service of the satellite communication network is realized.
The application principle of each module of the present invention is described in further detail below with reference to the accompanying drawings.
As shown in fig. 2, fig. 2 is a schematic diagram of a system frame for multi-scenario, multi-user and multi-service intent translation in a satellite communication network according to the present invention, and the method for multi-scenario, multi-user and multi-service intent translation in a satellite communication network provided by the present invention includes:
the front-end user webpage interface is used for acquiring the original task intention of the user and transmitting the intention to the back-end for processing, the interactive interface comprises a user input example and an intention input text box, and the user can input the task intention in the text box according to the prompt of the input example in the interactive interface;
the data preprocessing module is used for mainly converting the user intention text into a digital index which can be identified by the intention identification model, and intercepting the text exceeding the specified input length of the model or filling the text less than the specified input length of the model;
the intention classifying model is used for classifying the input user text intention through natural language processing technologies such as a text classifying algorithm and the like to realize specific business processing of specific users in specific scenes, the intention classifying model adopts an Albert-based multi-task classifying model, a model framework is shown in fig. 3, fig. 3 is a multi-task module building block diagram adopted by the intention classifying model, in the embodiment, the intention classifying model in the intention identifying module adopts a hard parameter sharing mode, namely, the model bottom layer sharing parameters, and the upper layer tasks are independent and are generally applicable to processing tasks with strong relevance. The identification of the communication scenario, user type, and business type to which the service belongs is three independent tasks, and the multitasking model reduces data processing, model training time, maintenance costs, etc. as compared to using a single task model to process each task separately. Meanwhile, a plurality of task results can be obtained by only requesting the model once, so that the online reasoning time can be reduced, and the generalization capability of the model is enhanced.
The intention extraction model is used for extracting intention key information from input user intention texts through natural language processing technologies such as naming recognition and the like, and constructing an intention quadruple < field, operation, object and result > to represent the user intention. For example, in the invention, < field > includes: communication scenario, user type, service type, etc.; the < operation > includes: uploading and downloading, deleting and adding, etc.; the < object > includes network nodes, traffic flows, network resources, etc.; the < result > includes performance metrics, expected states, space-time constraints, etc. The named entity recognition is mainly an entity extraction algorithm based on an Albert-CRF named entity recognition model. The model uses Albert in combination with a conditional random field (Conditional Random Field, CRF). And feeding back the intention classification result and the intention extraction result to the user for confirmation or modification.
And the intention storage module stores the original intention of the user and the recognition result confirmed by the user into a database so as to facilitate the subsequent dynamic learning of the recognition model and improve the recognition precision.
And the strategy mapping module guides the network to generate executable strategies through the intention recognition result, and issues the formed specific network configuration information to network elements of the bottom layer network for execution.
In summary, the invention belongs to the technical field of satellite communication networks, and discloses an intention translation method for multiple scenes, multiple users and multiple services of a satellite communication network. The problem that the prior art has larger limitation on user input is solved, and the method is suitable for professional users and non-professional users; secondly, the invention provides a method for combining coarse and fine granularity strategies, which solves the problem of low resource utilization rate caused by a plurality of scenes, users and service types in a satellite communication network, improves the resource utilization rate and improves user experience. The user inputs the original business intention through the front-end webpage interface, and at the same time, the original business intention text is preprocessed and sent to the rear end for further processing; scene, user and business identification is carried out by sending the business intention text to the intention classification model; inputting the user business intention into an intention extraction model to extract key information in the user intention; feeding back the intention classification result and the intention extraction result to user confirmation or modification, and storing the user confirmation and identification correct result to a database for subsequent model dynamic learning; and finally, feeding the identification result back to the webpage interface, and transmitting the identification result to the strategy mapping module for parameter mapping in a JSON file format to generate a network strategy. Briefly, the implementation of the present invention includes constructing an intent recognition module; front-end user business intention input and preprocessing; identifying scenes, users and services to obtain classification results; extracting front-end user business intention information to obtain an intention information extraction result; storing front-end user original business intents so as to facilitate intention classification and intention extraction model dynamic learning; enabling intent translation for a front-end user; intent translation for multi-scenario, multi-user and multi-service in a satellite communication network is achieved. The invention provides an efficient intention translation method by deeply analyzing the air interface characteristics, service types, service priorities and service quality requirements of users and various satellite stations on the network, constructing scenes, users and service models, and laying a foundation for an intention driving network technology in a satellite communication network.
According to the invention, the intention classification model is added in the intention recognition module, a coarse granularity strategy is formed by recognizing scenes, users and services, the intention extraction model is used for extracting the key information of the users to form a fine granularity strategy, network strategies meeting different service quality requirements can be provided for front-end users through the coarse granularity and fine granularity network strategies, and the satisfaction degree and experience of the users are improved. And secondly, the default value in the fine-grained network policy is filled by using the coarse-grained network policy, so that the situation that the fine-grained network policy cannot be generated due to the lack of a certain index is avoided, and the stability and reliability of intent translation are ensured.
Claims (5)
1. A method for translating intention of multiple scenes, multiple users and multiple services in a satellite communication network is characterized in that the intention translation method receives original service intention of a user at the front end through a webpage interface and processes the original service intention of the user at the rear end through a plurality of modules; the system comprises a plurality of modules, a user module and a user module, wherein the modules comprise a text preprocessing module, an intention recognition module and a strategy mapping module, a network strategy is finally obtained, and the network strategy is used for translating the intention of a user in a satellite communication network; the intent translation method for the multi-scene multi-user multi-service of the satellite communication network comprises the following steps:
step 1, constructing an intention recognition module; the built intention recognition module receives the original business intention input by the front-end user and sequentially inputs the original business intention to the intention classification model and the intention extraction model, the two model recognition results are output as a whole, and the correct recognition result is output; the specifications that the front-end user business intention input into the model needs to meet are: intercepting or filling the intention text with a specified length and converting the intention text into a digital index; the intention classification model is constructed based on a text classification algorithm in natural language processing and is used for identifying the scene, the user and the service type of the front-end user in multiple scenes, multiple users and multiple services; the intention extraction model is constructed based on an entity recognition algorithm in natural language processing and is used for extracting key network parameter information in front-end user business intention;
step 2, inputting and preprocessing the business intention of the front-end user; the front-end user inputs the original business intention in a natural language form through a webpage interface, carries out preprocessing operation on the original business intention text input by the front-end user, and specifically comprises error detection, error correction, filling or truncation operation of the original business intention text and conversion into a digital index, so that the business intention text originally input by the front-end user accords with the input specification of the intention recognition module and is input into the intention recognition module;
step 3, identifying scenes, users and services to obtain classification results; simultaneously identifying the scene to which the front-end user business intention belongs, the user and the specific type of the business by using an intention classification model in the intention identification module to obtain an identification result; judging the identification result to realize the specific service processing of the specific user in the specific scene; the identification result is that the non-specific scene, the user and the service are processed according to a default mode or a user-specified mode;
step 4, extracting the business intention information of the front-end user to obtain an intention information extraction result; after classifying all front-end user service intentions, extracting key network parameter information by using an intent extraction model, wherein the key network parameter information comprises bandwidth, time delay, duration, spatial position, jitter and network transmission speed parameters, and feeding back a service intention classification result and an intention information extraction result to a webpage interface for the front-end user to confirm or modify so as to obtain a correct identification result confirmed by the front-end user;
step 5, storing the original business intention of the front-end user so as to facilitate the dynamic learning of the intention classification and intention extraction model; storing the original input business intention of the front-end user and the correct recognition result confirmed by the user into a database so as to facilitate the dynamic learning of the intention classification and extraction model;
step 6, realizing intent translation of the front-end user; the correct identification result confirmed by the front-end user is fed back to the front-end webpage interface for the front-end user to store, and meanwhile, the correct identification result is input to a strategy mapping module constructed based on a finite state machine to generate a network strategy, so that the network strategy mapping is completed; the network strategy is used for the intention translation aiming at the front-end user in the satellite communication network, and the original service intention translation of the front-end user in the multi-scene multi-user multi-service of the satellite communication network is realized.
2. The method for intent translation of multiple scenes, multiple users and multiple services in a satellite communication network according to claim 1, wherein in step 3, scene, user and service are identified to obtain classification results, and intent classification models in the intent identification module are based on a large-scale pretraining model Albert and a one-dimensional convolution model TextCNN as shared layer multitasking models, and scene, user and service types are identified simultaneously to obtain classification results.
3. The method for intent translation for multiple scenarios, multiple users and multiple services in a satellite communication network according to claim 1, wherein the intent extraction model in step 4 is constructed based on a large-scale pre-training language model Albert and a conditional random field CRF; in the extraction process, the user intention is represented by using a quadruple of < field, operation, object and result >; wherein < field > includes: communication scene, user type, service type; the < operation > includes: uploading and downloading, deleting and adding; the < object > includes network nodes, traffic flows, network resources; the < result > includes performance metrics, expected states, space-time constraints.
4. The method for intent translation in a satellite communication network according to claim 1, wherein in step 4, the intent information of the front-end user is extracted to obtain an intent information extraction result, and a named entity recognition algorithm is used to label the user intent text based on a BIOES labeling method in the intent extraction.
5. The method according to claim 1, wherein in step 6, in implementing intent translation of the front-end user, the policy mapping module generates network policies, specifically, intent classification results are used for mapping to generate coarse-grained network policies, and the coarse-grained network policies are a set of network policies provided by the system for meeting the basic service quality requirements of the front-end user; the intention extraction result is used for mapping and generating a fine-granularity network policy, is generated according to the original business intention of a front-end user, and is filled and completed by a coarse-granularity network policy corresponding item when a default value appears in a certain index of the fine-granularity network policy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310249795.9A CN116388838A (en) | 2023-03-15 | 2023-03-15 | Multi-scene multi-user multi-service intention translation method for satellite communication network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310249795.9A CN116388838A (en) | 2023-03-15 | 2023-03-15 | Multi-scene multi-user multi-service intention translation method for satellite communication network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116388838A true CN116388838A (en) | 2023-07-04 |
Family
ID=86960701
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310249795.9A Pending CN116388838A (en) | 2023-03-15 | 2023-03-15 | Multi-scene multi-user multi-service intention translation method for satellite communication network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116388838A (en) |
-
2023
- 2023-03-15 CN CN202310249795.9A patent/CN116388838A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110888722B (en) | Task processing method and device, electronic equipment and computer readable storage medium | |
| Zhang et al. | Toward wisdom-evolutionary and primitive-concise 6G: A new paradigm of semantic communication networks | |
| CN109709811B (en) | Intelligent manufacturing-oriented edge calculation controller and working method thereof | |
| CN101267452B (en) | A conversion method and application server for WEB service mixing scheme | |
| CN109347950B (en) | Kaa Project-based Internet of things intelligent service system | |
| CN101719933B (en) | Combination method of manufacturing grid resource services orienting whole life cycle and supporting semantemes | |
| CN111562972A (en) | Ubiquitous operating system for crowd sensing | |
| CN111600643A (en) | Satellite communication resource scheduling method and device storage medium | |
| CN109492536A (en) | A kind of face identification method and system based on 5G framework | |
| CN102968307B (en) | A kind of web based on java develops middleware | |
| CN113037802B (en) | Cloud-side data cooperation method for power Internet of things | |
| CN117008923B (en) | Code generation and compiling deployment method, platform and equipment based on AI large model | |
| CN114912433A (en) | Text level multi-label classification method and device, electronic equipment and storage medium | |
| CN108306804A (en) | A kind of Ethercat main station controllers and its communication means and system | |
| CN109816114A (en) | A kind of generation method of machine learning model, device | |
| US11838367B2 (en) | Information acquiring method, apparatus, and system | |
| AlShahwan et al. | Mobile cloud computing for providing complex mobile web services | |
| Taft | Architectural basis for highly distributed transactive power grids: Frameworks, networks, and grid codes | |
| Zhang et al. | Generative AI for space-air-ground integrated networks (sagin) | |
| CN116388838A (en) | Multi-scene multi-user multi-service intention translation method for satellite communication network | |
| CN110991279B (en) | Document Image Analysis and Recognition Method and System | |
| CN113010296A (en) | Task analysis and resource allocation method and system based on formalized model | |
| CN117313837A (en) | Large model prompt learning method and device based on federal learning | |
| CN104778254A (en) | Distributing type system for non-parameter topic automatic identifying and identifying method | |
| CN112528083B (en) | Message customization method based on distributed semantic template distribution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |