WO2020181599A1 - 一种模型应用方法、管理方法、系统及服务器 - Google Patents

一种模型应用方法、管理方法、系统及服务器 Download PDF

Info

Publication number
WO2020181599A1
WO2020181599A1 PCT/CN2019/081536 CN2019081536W WO2020181599A1 WO 2020181599 A1 WO2020181599 A1 WO 2020181599A1 CN 2019081536 W CN2019081536 W CN 2019081536W WO 2020181599 A1 WO2020181599 A1 WO 2020181599A1
Authority
WO
WIPO (PCT)
Prior art keywords
model
target
rule
scheduling
request
Prior art date
Application number
PCT/CN2019/081536
Other languages
English (en)
French (fr)
Inventor
杨绳春
李金锋
Original Assignee
网宿科技股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 网宿科技股份有限公司 filed Critical 网宿科技股份有限公司
Priority to EP19797974.3A priority Critical patent/EP3731161A1/en
Priority to US16/686,026 priority patent/US20200286012A1/en
Publication of WO2020181599A1 publication Critical patent/WO2020181599A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063Operations research, analysis or management
    • G06Q10/0631Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06316Sequencing of tasks or work
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/02Neural networks
    • G06N3/08Learning methods
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00Computing arrangements based on biological models
    • G06N3/12Computing arrangements based on biological models using genetic models
    • G06N3/126Evolutionary algorithms, e.g. genetic algorithms or genetic programming
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/04Inference or reasoning models
    • G06N5/046Forward inferencing; Production systems

Definitions

  • This application relates to the field of artificial intelligence technology, in particular to a model application method, management method, system and server.
  • AI Artificial Intelligence
  • technologies such as robotics, language recognition, image recognition, natural language processing, and expert systems have been widely used in applications such as intelligent speech, face recognition, and intelligent assistants.
  • a model based on a simulation algorithm can be used to simulate the content requested by the user to obtain a result corresponding to the requested content.
  • a business scenario usually needs to apply one or more models.
  • the models that need to be applied in a business scenario can be packaged in order.
  • the business scenario package can be used The model is processed, and the processing result is obtained.
  • the purpose of this application is to provide a model application method, management method, system and server, which can save computer resources and model maintenance costs.
  • the present application provides a model application method on the one hand, which includes: receiving request information sent by a user terminal; determining a target business scenario corresponding to the request information; and determining a relationship with the target business based on preset configuration rules
  • a model management method including: providing a configuration server for storing configuration rules and models, the configuration rules including: a preset correspondence between business scenarios and scheduling rules;
  • the scheduling rules include: a scheduling model name and a scheduling sequence; acquiring a target business scenario, determining a target scheduling rule corresponding to the target business scenario according to the target business scenario and the stored configuration rules, and feeding back the target scheduling rule.
  • model application system including:
  • the user terminal is used to send request information to the model application server and receive the request result corresponding to the request information;
  • the request information includes: service interface information and request content;
  • the model application server is used to receive the request information sent by the user terminal, determine the target business scenario according to the request information, and determine the target scheduling rule corresponding to the target business scenario based on preset configuration rules, and according to the target
  • the target sequence in the scheduling rule dispatches the target model to perform information processing, obtains the request result corresponding to the request information, and sends the request result to the user terminal;
  • the model application server is also used to store the model.
  • another aspect of the present application also provides a model management server, including: a storage unit, a message receiving unit, and a scheduling rule determination unit; wherein,
  • the storage unit is used to store configuration rules and models
  • the message receiving unit is configured to receive information; the received information includes: target business scenario information;
  • the scheduling rule determining unit is configured to determine a target scheduling rule corresponding to the target business scenario from the configuration rules stored in the storage unit according to the target business scenario received by the message receiving unit.
  • the management server includes a memory and a processor, the memory is used to store a computer program, and when the computer program is executed by the processor, the foregoing execution method is implemented.
  • each algorithm model used to simulate business scenarios only needs to be stored once, and the configuration rules can be used to pre-configure corresponding model scheduling rules for different business scenarios.
  • the configuration rules can be used to pre-configure corresponding model scheduling rules for different business scenarios.
  • a model needs to be modified only The stored model needs to be modified once, instead of separately modifying the encapsulated models in multiple business scenarios. Therefore, the technical solution provided in this application not only saves computer resources, but also reduces the cost of maintaining the model.
  • the configuration rules need to be updated, which is quick to operate and can improve the execution efficiency of developers .
  • Figure 1 is a flowchart of a model management method in an embodiment of this specification
  • Figure 2 is a flowchart of a model application method in an embodiment of this specification
  • Figure 3 is a schematic diagram of the composition of a model application system in an embodiment of this specification.
  • FIG. 4 is a schematic diagram of the unit composition of a model application server in an embodiment of this specification.
  • Figure 5 is a schematic diagram of the unit composition of a model management server in an embodiment of this specification.
  • FIG. 6 is a schematic diagram of a structure of a server in an embodiment of this specification.
  • Fig. 7 is a schematic diagram of a structure of a computer terminal in an embodiment of this specification.
  • This application provides a model management method, which can be applied to the management of models in artificial intelligence technology.
  • the model management method provided by the embodiment of the present application may include the following steps.
  • S11 Provide a configuration server for storing configuration rules and models.
  • a configuration server may be provided.
  • the configuration server can be one server or a server cluster composed of multiple servers.
  • the configuration server may store configuration rules and models.
  • the model can be used for information processing to realize artificial intelligence information processing.
  • the model may be an algorithm model applied to artificial intelligence information processing.
  • it can be an artificial neural network model, a genetic algorithm model, etc.
  • the configuration rules may be pre-stored.
  • the configuration rule may include: a preset correspondence relationship between business scenario information and scheduling rules.
  • the business scene information may be the scene name corresponding to the service item.
  • the business scenario information may be intelligent question and answer, etc.
  • the scheduling rule may be used to indicate the models that need to be called sequentially when simulating a scene.
  • the scheduling rule may include: a scheduling model name and a scheduling sequence.
  • the sequential scheduling model may include a model for analyzing questions and a model for querying answers.
  • the problem-analyzing model can analyze the content of the user's problem and extract the keywords of the problem.
  • the model for querying answers can find answers based on keywords.
  • the model can realize data interaction through a model interaction interface.
  • the model interaction interface of each model stored in the configuration server may be inconsistent. That is, the data format adopted by the interactive interface of each model may be inconsistent. Then, before calling a model for information processing, you can first convert the data format of the input information into a data format suitable for the model, then call the model for information processing, and output the information processing result of the model data format.
  • model A and model B need to be called sequentially.
  • the data format of the output "output information 1" is a 16-bit data format
  • the applicable data format of model B is a 32-bit data format. Then, first convert the data format of "output information 1" to a 32-bit data format, then call model B for processing, and output "output information 2", which is a 32-bit data format.
  • the model stored in the configuration server may adopt a unified preset model interaction interface.
  • the preset model interaction interface may adopt a preset standard data format. Then, the result obtained by one model can be directly called by another model, without the need to perform data format conversion according to the data format of the previous module and the data format of the current module before calling. It can ensure that the models are directly interconnected, which improves the efficiency of information processing.
  • model A and model B need to be called successively. Both model A and model B use the preset model interaction interface, and the output information obtained after calling model A is also data in standard data format. You can directly call B to process the output information.
  • each model can use the preset standard data format for information processing.
  • the input information in the preset standard data format can be directly processed to obtain output information in the preset standard data format.
  • all models use a preset standard data format for information processing, which can improve the efficiency of data processing.
  • each model can use different standard data formats for information processing. Then, when the model is called for data processing through the above-mentioned preset model interaction interface, the model can convert the input of the standard data format into a data format suitable for the model and process it to obtain the output of the data format suitable for the model Information, and then convert the output information of the data format applicable to the model into the output information of the standard data format. For example, in a scheduling rule corresponding to an application scenario, model A and model B need to be called successively.
  • model A and model B use the preset model interaction interface
  • model A uses the first data format for information processing
  • model B uses the second data Format for information sorting
  • the input information in the standard data format is converted into the first data format input information for processing, and output information A in the first data format is obtained, and then the output in the first data format is output
  • the information is converted into output information A in the standard data format; when model B is called to process the output information A in the standard data format, the output information A in the standard data format is converted into output information A in the second data format for processing , Obtain output information B in the second data format, convert the output information B in the second data format into output information B in the standard data format and output.
  • each model can make full use of existing models that use different data formats for information processing, which improves resource utilization.
  • Each model only needs to perform data conversion between the data format applicable to the model and the standard data format, and does not need to convert various different data formats, which reduces the complexity of data conversion.
  • S12 Obtain a target business scenario, determine a target dispatch rule corresponding to the target business scenario according to the target business scenario and the stored configuration rule, and feed back the target dispatch rule.
  • the configuration server can obtain the target business scenario.
  • the target business scenario may be determined according to the request information sent by the user terminal.
  • the request information may include: service interface information and request content.
  • the service interface information can be used to characterize the service item corresponding to the requested content.
  • the service interface information can be represented by characters.
  • the service interface information can be the text "Q&A", or the number "01".
  • the requested content may be questions, instructions, etc. issued by the user.
  • the target business scenario may be determined according to the business interface information in the request information.
  • the service interface information and the service scenario may have a one-to-one correspondence.
  • the corresponding relationship between the service interface information and the service scenario may be preset.
  • the business interface information in the request information can be "question and answer", and the content of the request can be "how to start”.
  • the business interface information "question and answer” can indicate that the service item is a smart question and answer. It is assumed that the preset business interface information and the business scenario In the correspondence relationship, the business scenario corresponding to the business interface information "Q&A" is "smart question and answer", then the target business scenario corresponding to the request information can be determined as "smart question and answer”.
  • the target scheduling rule can be filtered from the stored configuration rules according to the target business scenario.
  • the target scheduling rule may be a scheduling rule corresponding to the target business scenario.
  • the target scheduling rule may include: the name of the target model and the target sequence. Invoking the target model to perform information processing according to the target sequence can realize the simulation of the target business scenario.
  • the target dispatch rule After the target dispatch rule is determined, the target dispatch rule can be fed back to the server for finding the dispatch rule.
  • the model management method may further include: receiving a request to update a configuration rule, and processing the stored configuration rule according to a rule update operation and rule update content in the request to update the configuration rule.
  • the request to update the configuration rule may include: rule update operation and rule update content.
  • the rule update operation may include: adding configuration rules, modifying configuration rules, and/or deleting configuration rules.
  • the operation of adding a configuration rule may be used to add a correspondence between a business scenario and a scheduling rule in the stored configuration rule.
  • the received request to update configuration rules can be "new configuration rules, business scenario 1, model A, model C, model D", then, according to the request to update configuration rules, you can add to the stored configuration rules A correspondence relationship between a business scenario and a scheduling rule.
  • the business scenario in the relationship is "Business Scenario 1”
  • the scheduling rule is "Model A, Model C, and Model D”.
  • the operation of deleting the configuration rule can be used to delete the correspondence between the business scenario and the scheduling rule from the stored configuration rules.
  • the operation of modifying the configuration rule can be used to modify the correspondence between the business scenario and the scheduling rule in the stored configuration rule, including: modifying the name of the scheduling model and/or modifying the scheduling sequence.
  • the scheduling rule corresponding to "Business Scenario 2" is "Model A, Model B”.
  • the scheduling rule corresponding to "Business Scenario 2” can be modified to "Model A, Model B, Model D”.
  • the scheduling rule corresponding to "business scenario 2” can be modified to "model B, model A”.
  • the model management method may further include: receiving a request to update a model, and processing the stored model according to a model update operation in the request to update the model.
  • the request to update the model may include: model update operation and model update content.
  • the model update operation may include: adding a model, modifying a model, and/or deleting a model.
  • each algorithm model used to simulate business scenarios only needs to be stored once, and the configuration rules can be used to pre-configure corresponding model scheduling rules for different business scenarios.
  • the configuration rules can be used to pre-configure corresponding model scheduling rules for different business scenarios.
  • the embodiment of the present application also provides a model application method.
  • the model application method may include the following steps.
  • S21 Receive request information sent by the user terminal.
  • the server can receive the request information sent by the user terminal.
  • the request information may indicate the content of the service required by the user in a business scenario.
  • the request information may include: service interface information and request content.
  • the service interface information can be used to characterize the service item corresponding to the requested content.
  • the service interface information can be represented by characters.
  • the requested content may be questions, instructions, etc. issued by the user.
  • the target business scenario may be determined according to the business interface information in the request information.
  • the service interface information and the service scenario may have a one-to-one correspondence.
  • the corresponding relationship between the service interface information and the service scenario may be preset.
  • the business interface information in the request information can be "question and answer”. Assuming that the preset correspondence between the business interface information and the business scenario, the business scenario corresponding to the business interface information "question and answer" is “smart question and answer”, then it can be determined The target business scenario corresponding to the request information is "smart question and answer”.
  • S23 Determine a target scheduling rule corresponding to the target business scenario based on the preset configuration rule.
  • the server may pre-store configuration rules.
  • the configuration rule may include: a preset correspondence relationship between business scenario information and scheduling rules.
  • the business scenario information may be a business scenario name.
  • the business scenario information may be used to indicate the scenario corresponding to the service item.
  • the business scenario information can be "smart question and answer" and so on.
  • the scheduling rule may be used to indicate the models that need to be called sequentially when simulating a scene.
  • the scheduling rule may include: a scheduling model name and a scheduling sequence.
  • the model may be an algorithm model applied to artificial intelligence information processing.
  • it can be an artificial neural network model, a genetic algorithm model, etc.
  • human operations in business scenarios can be simulated to achieve artificial intelligence information processing.
  • an artificial neural network algorithm can be used to process voice information and analyze the emotional information of the voice.
  • the determining the target scheduling rule corresponding to the target business scenario based on the preset configuration rule may include: based on the preset correspondence relationship between the business scenario information in the preset configuration rule and the scheduling rule, from the storage
  • the target scheduling rules are filtered out of the configuration rules.
  • the target scheduling rule may be a scheduling rule corresponding to the target business scenario.
  • the target scheduling rule may include: the name of the target model and the target sequence. Invoking the target model according to the target sequence can realize the simulation of the target business scenario.
  • multiple models may be stored in the server.
  • the multiple stored models may adopt different model interaction interfaces.
  • the multiple stored models may also adopt a unified preset model interaction interface.
  • the target model can be scheduled according to the target sequence in the target scheduling rule for information processing, so as to obtain the request result.
  • the request content in the request information is "X”
  • the target scheduling rule corresponding to the request information is: sequential scheduling model 1 and model 2. Then, take “X” as the input information and use Model 1 for information processing to get the output result "A”, then use “A” as the input information and use Model 2 for information processing to get the output result "B”. B” is output as the request result.
  • the server may feed back the request result obtained by information processing using the target model to the user terminal.
  • the model application system includes: a user terminal and a model application server.
  • the model application server may be one server or a server cluster composed of multiple servers.
  • the user terminal may be used to send request information to the model application server and receive a request result corresponding to the request information.
  • the request information may include: service interface information and request content.
  • the model application server may be used to receive request information sent by the user terminal, determine a target business scenario based on the request information, and determine a target scheduling rule corresponding to the target business scenario based on preset configuration rules, according to The target sequence scheduling target model in the target scheduling rule performs information processing, obtains a request result corresponding to the request information, and sends the request result to the user terminal.
  • the model application server is also used to store models.
  • the model application server may include: a configuration unit, a scheduling unit, a model and a configuration storage unit.
  • the model and configuration storage unit can be used to store configuration rules and models.
  • the configuration unit may be used to determine a target scheduling rule corresponding to the target business scenario according to the target business scenario and the configuration rules stored in the model and configuration storage unit.
  • the scheduling unit may be configured to perform information processing from the model and the configuration storage unit scheduling target model according to the target sequence in the target scheduling rule determined by the configuration unit to obtain a request result corresponding to the request information.
  • the model management server may include: a storage unit, a message receiving unit, and a scheduling rule determination unit.
  • the storage unit may be used to store configuration rules and models.
  • the message receiving unit may be used to receive information.
  • the received information may include target business scenario information.
  • the scheduling rule determining unit may be configured to determine the target scheduling rule corresponding to the target business scenario from the configuration rules stored in the storage unit according to the target business scenario received by the message receiving unit.
  • the message receiving unit may also be used to receive a request for updating configuration rules.
  • the request to update the configuration rule may include: rule update operation and rule update content.
  • the storage unit may also be used to process the stored configuration rules according to the rule update operation and rule update content in the request to update the configuration rules.
  • the rule update operation may include: adding a configuration rule, modifying a configuration rule, and/or deleting a configuration rule.
  • the message receiving unit may also be used to receive a request to update the model.
  • the request to update the model may include: model update operation and model update content.
  • the model update operation may include: adding a model, modifying a model, and/or deleting a model.
  • the storage unit may also be used to perform information processing on the stored model according to the model update operation in the request to update the model.
  • the present application also provides a server, the server includes a memory and a processor, the memory is used to store a computer program, when the computer program is executed by the processor, can achieve the implementation of the above-mentioned embodiment method.
  • the computer terminal 10 may include one or more (only one is shown in the figure) processor 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a processor for storing data
  • the memory 104 and the transmission module 106 for communication functions are only for illustration, and does not limit the structure of the above electronic device.
  • the computer terminal 7 may also include more or fewer components than those shown in FIG. 7, or have a configuration different from that shown in FIG.
  • the memory 104 may be used to store software programs and modules of application software.
  • the processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104.
  • the memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory.
  • the memory 104 may further include a memory remotely provided with respect to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
  • the foregoing server deployment method may be stored as a computer program in the foregoing memory 104, and the memory 104 may be coupled with the processor 102, then when the processor 102 executes the computer in the memory 104
  • the program can implement the steps in the server deployment method described above.
  • the transmission device 106 is used to receive or send data via a network.
  • the above-mentioned specific examples of the network may include a wireless network provided by the communication provider of the computer terminal 10.
  • the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station to communicate with the Internet.
  • the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet in a wireless manner.
  • RF radio frequency
  • the technical solution provided by this application provides a model application method and a model management method based on configuration rules.
  • the configuration rules are used to pre-configure the corresponding model scheduling rules for different business scenarios.
  • the model scheduling rules of the model call the stored models urgently need information processing, and each algorithm model only needs to be stored once, which not only saves computer resources, but also reduces the cost of maintaining the model. .
  • each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware.
  • the above technical solutions can be embodied in the form of software products, which can be stored in computer-readable storage media, such as ROM/RAM, magnetic A disc, an optical disc, etc., include a number of instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiment.

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • Educational Administration (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Stored Programmes (AREA)

Abstract

本申请公开了一种模型应用方法、管理方法、系统及服务器,其中,所述模型应用方法包括:接收用户终端发出的请求信息(S21);确定与所述请求信息对应的目标业务场景(S22);基于预设配置规则,确定与所述目标业务场景对应的目标调度规则;所述目标调度规则包括:目标模型的名称及目标顺序(S23);根据所述目标调度规则中的目标顺序从存储的模型中调度目标模型进行信息处理,得到与所述请求信息对应的请求结果(S24);将所述请求结果反馈给所述用户终端(S25)。本申请提供的技术方案,能够节省计算机资源和模型维护成本。

Description

一种模型应用方法、管理方法、系统及服务器
交叉引用
本申请引用于2019年3月8日递交的名称为“一种模型应用方法、管理方法、系统及服务器”的第201910176811.X号中国专利申请,其通过引用被全部并入本申请。
技术领域
本申请涉及人工智能技术领域,特别涉及一种模型应用方法、管理方法、系统及服务器。
背景技术
人工智能(Artificial Intelligence,AI)技术是研究和开发用于模拟、延伸和扩展人的智能的技术。随着计算机科学的高速发展,人工智能技术也被越来越多地应用到人们的生活中。例如机器人、语言识别、图像识别、自然语言处理、专家系统等技术已经被广泛应用到智能语音、人脸识别、智能助手等应用领域。
人工智能技术处理用户请求信息时,可以采用基于模拟算法建立的模型对用户请求的内容进行模拟处理,得到与请求内容对应的结果。目前,模型应用时,通常一个业务场景常常需要应用一个或者多个模型,可以将一个业务场景需要应用的模型按照顺序封装起来,当该业务场景接收到用户请求信息时,可以利用该业务场景封装的模型进行处理,得到处理结果。
由于,不同的业务场景通常需要应用不同的模型,因此,当同一模型应用于不同的业务场景中时,也需要对应不同的业务场景对模型分别进行封装,模型应用不灵活且浪费计算机资源。由于一个模型被多次封装,当该模型需要 改动时,则需要对多个封装的模型分别进行改动,导致模型维护成本较高。因此,目前亟需一种灵活性较高的模型应用方法。
发明内容
本申请的目的在于提供一种模型应用方法、管理方法、系统及服务器,能够节省计算机资源和模型维护成本。
为实现上述目的,本申请一方面提供一种模型应用方法,包括:接收用户终端发出的请求信息;确定与所述请求信息对应的目标业务场景;基于预设配置规则,确定与所述目标业务场景对应的目标调度规则;所述目标调度规则包括:目标模型的名称及目标顺序;根据所述目标调度规则中的目标顺序从存储的模型中调度目标模型进行信息处理,得到与所述请求信息对应的请求结果;将所述请求结果反馈给所述用户终端。
为实现上述目的,本申请另一方面提供一种模型管理方法,包括:提供一配置服务器,用于存储配置规则与模型,所述配置规则包括:业务场景与调度规则的预设对应关系;所述调度规则包括:调度模型名称以及调度顺序;获取目标业务场景,根据所述目标业务场景和所述存储的配置规则,确定与所述目标业务场景对应的目标调度规则,反馈目标调度规则。
为实现上述目的,本申请另一方面提供一种模型应用系统,包括:
用户终端,用于向模型应用服务器发送请求信息,以及接收与所述请求信息对应的请求结果;所述请求信息包括:业务接口信息和请求内容;
模型应用服务器,用于接收所述用户终端发来的请求信息,根据所述请求信息确定目标业务场景,以及基于预设配置规则确定与所述目标业务场景对应的目标调度规则,根据所述目标调度规则中的目标顺序调度目标模型进行信息处理,得到与所述请求信息对应的请求结果,将所述请求结果发送给所述用户终端;所述模型应用服务器还用于存储模型。
为实现上述目的,本申请另一方面还提供一种模型管理服务器,包括: 存储单元、消息接收单元和调度规则确定单元;其中,
所述存储单元,用于存储配置规则及模型;
所述消息接收单元,用于接收信息;所述接收的信息包括:目标业务场景信息;
所述调度规则确定单元,用于根据所述消息接收单元接收的目标业务场景,从所述存储单元存储的配置规则中确定与所述目标业务场景对应的目标调度规则。
为实现上述目的,本申请另一方面还提供所述管理服务器包括存储器和处理器,所述存储器用于存储计算机程序,所述计算机程序被所述处理器执行时,实现上述执行的方法。
由上可见,本申请提供的技术方案中,用于实现业务场景模拟的各算法模型只需要存储一次,可以利用配置规则为不同的业务场景预先配置好对应的模型调度规则,当模拟不同业务场景时,只需要查找到业务场景对应的调度规则,再根据调度规则按序调用模型进行处理,而不需要在不同的业务场景下对一个模型进行多次封装,当一个模型需要修改时,也只需对存储的该模型进行一次修改即可,而不需要对多个业务场景下封装的模型分别进行修改,因此,本申请提供的技术方案既节省了计算机资源,也降低了维护模型的成本。同时,本申请提供的技术方案中,当有新的业务场景或已有的业务场景对应的调度规则发生变化时,只需要对配置规则进行更新即可,操作快捷,可以提高开发人员的执行效率。
附图说明
为了更清楚地说明本申请实施例中的技术方案,下面将对实施例描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1是本说明书实施例中一种模型管理方法的流程图;
图2是本说明书实施例中一种模型应用方法的流程图;
图3是本说明书实施例中一个模型应用系统的组成示意图;
图4是本说明书实施例中一个模型应用服务器的单元组成示意图;
图5是本说明书实施例中一个模型管理服务器的单元组成示意图;
图6是本说明书实施例中服务器的一种结构示意图;
图7是本说明书实施例中计算机终端的一种结构示意图。
具体实施例
为使本申请的目的、技术方案和优点更加清楚,下面将结合附图对本申请实施例作进一步地详细描述。
本申请提供一种模型管理方法,该方法可以应用于人工智能技术中对模型的管理。
请参阅图1,本申请实施例提供的模型管理方法可以包括以下步骤。
S11:提供一配置服务器,用于存储配置规则与模型。
在一个实施例中,可以提供一配置服务器。所述配置服务器可以是一个服务器,也可以是由多个服务器组成的服务器集群。
在一个实施例中,所述配置服务器可以存储配置规则与模型。
所述模型可以用于进行信息处理,以实现人工智能信息处理。在一个实施例中,所述模型可以是应用于人工智能信息处理的算法模型。例如,可以是人工神经网络模型、遗传算法模型等。
所述配置规则可以是预先存储的。
在一个实施例中,所述配置规则可以包括:业务场景信息与调度规则的预设对应关系。
所述业务场景信息可以是服务项目所对应的场景名称。例如,业务场景信息可以是智能问答等。
所述调度规则可以用于表示模拟一个场景时依次需要调用的模型。在一个实施例中,所述调度规则可以包括:调度模型名称以及调度顺序。例如,对于智能问答这一场景,依次调度的模型可以包括分析问题的模型和查询答案的模型。分析问题的模型可以分析用户提出问题的内容,提取问题的关键词。查询答案的模型可以根据关键词来查找答案。
所述模型可以通过模型交互接口实现数据交互。
在一个实施例中,所述配置服务器中存储的各模型的模型交互接口可以不一致。即各模型交互接口采用的数据格式可以不一致。那么,当调用一个模型进行信息处理前,可以先将输入信息的数据格式转换为该模型适用的数据格式,再调用该模型进行信息处理,并输出该模型数据格式的信息处理结果。
例如,在一个应用场景对应的调度规则中,需要先后调用模型A和模型B。假设调用模型A进行信息处理后,输出的“输出信息1”的数据格式为16位的数据格式,而模型B适用的数据格式为32位的数据格式。那么,先将“输出信息1”的数据格式转换为32位的数据格式,再调用模型B进行处理,并输出“输出信息2”,该“输出信息2”为32位的数据格式。
在另一个实施例中,所述配置服务器中存储的模型可以采用统一的预设模型交互接口。所述预设模型交互接口可以采用预设的标准数据格式。那么,一个模型处理得到的结果可以直接被另一个模型调用,而不需要在调用前根据前一个模块的数据格式和当前模块的数据格式进行数据格式转换。可以保证各模型之间直接进行互接,提高了信息处理的效率。
例如,一个应用场景对应的调度规则中,需要先后调用模型A和模型B,模型A和模型B都采用预设模型交互接口,则调用模型A后处理得到的输出信息也是标准数据格式的数据,可以直接调用B对该输出信息进行处理。
在一个实施例中,各模型可以采用所述预设的标准数据格式进行信息处理。例如,模型A采用所述预设的标准数据格式进行信息处理,则调用模型A进行信息处理时,可以直接对预设标准数据格式的输入信息进行处理,得到预 设标准数据格式的输出信息。在该实施例中,所有模型采用预设标准数据格式进行信息处理,可以提高数据处理的效率。
在另一个实施例中,各模型可以采用不同的标准数据格式进行信息处理。那么,通过上述预设模型交互接口调用所述模型进行数据处理时,所述模型可以将标准数据格式的输入进行转换为该模型适用的数据格式并进行处理,得到该模型适用的数据格式的输出信息,再将该模型适用的数据格式的输出信息转换为标准数据格式的输出信息。例如,一个应用场景对应的调度规则中,需要先后调用模型A和模型B,模型A和模型B都采用预设模型交互接口,模型A采用第一数据格式进行信息处理,模型B采用第二数据格式进行信息梳理,则调用模型A进行信息处理时,将标准数据格式的输入信息转换为第一数据格式输入信息进行处理,得到第一数据格式的输出信息A,再将第一数据格式的输出信息转换为标准数据格式的输出信息A;调用模型B对所述标准数据格式的输出信息A进行处理时,将所述标准数据格式的输出信息A转换为第二数据格式的输出信息A进行处理,得到第二数据格式的输出信息B,将所述第二数据格式的输出信息B转换为标准数据格式的输出信息B并进行输出。在该实施例中,各模型可以充分利用现有的采用不同数据格式进行信息处理的模型,提高了资源利用率。各模型只需进行该模型适用的数据格式和标准数据格式之间的数据转换,不需要对各种不同数据格式进行转换,降低了数据转换的复杂度。
S12:获取目标业务场景,根据所述目标业务场景和所述存储的配置规则,确定与所述目标业务场景对应的目标调度规则,反馈目标调度规则。
所述配置服务器可以获取目标业务场景。所述目标业务场景可以根据用户终端发出的请求信息来确定。
在一个实施例中,所述请求信息可以包括:业务接口信息和请求内容。其中,所述业务接口信息可以用于表征与所述请求内容对应的服务项目。所述业务接口信息可以用字符来表示。例如,所述业务接口信息可以为文字“问答”,也可以为数字“01”等。所述请求内容可以是用户发出的问题、指令等。
所述目标业务场景可以根据所述请求信息中的业务接口信息来确定。所述业务接口信息与业务场景可以具有一一对应关系。所述业务接口信息与业务场景的对应关系可以是预先设定的。
例如,请求信息中的业务接口信息可以为“问答”,请求内容可以是“如何启动”,该业务接口信息“问答”可以表示服务项目为智能问答,假设预设的业务接口信息与业务场景的对应关系中,业务接口信息“问答”对应的业务场景为“智能问答”,那么,可以确定该请求信息对应的目标业务场景为“智能问答”。
在一个实施例中,根据所述目标业务场景可以从所述存储的配置规则中筛选出目标调度规则。所述目标调度规则可以是与所述目标业务场景对应的调度规则。
在一个实施例中,所述目标调度规则可以包括:目标模型的名称和目标顺序。根据所述目标顺序调用所述目标模型进行信息处理,可以实现对所述目标业务场景的模拟。
确定目标调度规则后,可以将所述目标调度规则反馈给用于查找调度规则的服务器。
在另一个实施例中,所述模型管理方法还可以包括:接收更新配置规则的请求,根据所述更新配置规则的请求中的规则更新操作和规则更新内容对所述存储的配置规则进行处理。
在一个实施例中,所述更新配置规则的请求可以包括:规则更新操作和规则更新内容。
在一个实施例中,所述规则更新操作可以包括:新增配置规则、修改配置规则和/或删除配置规则。
所述新增配置规则的操作可以用于在已存储的配置规则中新增业务场景与调度规则的对应关系。例如,接收的更新配置规则的请求可以为“新增配置规则,业务场景1,模型A、模型C、模型D”,那么,根据该更新配置规则的请求,可以在已存储的配置规则中增加一条业务场景与调度规则的对应关系,该 关系中的业务场景为“业务场景1”,调度规则为“模型A、模型C、模型D”。
所述删除配置规则的操作可以用于从已存储的配置规则中删除业务场景与调度规则的对应关系。
所述修改配置规则的操作可以用于对已存储的配置规则中的业务场景与调度规则的对应关系进行修改,包括:修改调度模型的名称和/或修改调度顺序。
例如,“业务场景2”对应的调度规则为“模型A、模型B”。当接收的更新配置规则的请求为“修改配置规则,业务场景2,模型A、模型B、模型D”时,可以将“业务场景2”对应的调度规则修改为“模型A、模型B、模型D”。当接收的更新配置规则的请求为“修改配置规则,业务场景2,模型B、模型A”时,可以将“业务场景2”对应的调度规则修改为“模型B、模型A”。
在一个实施例中,所述模型管理方法还可以包括:接收更新模型的请求,根据所述更新模型的请求中的模型更新操作对所述存储的模型进行处理。所述更新模型的请求可以包括:模型更新操作和模型更新内容。所述模型更新操作可以包括:新增模型、修改模型和/或删除模型。对存储的模型执行更新操作后,所有调用该模型的业务场景均可以相应地应用新的模型来模拟业务场景,而不需要对每个应用场景下封装的模型分别进行更新,既节省来计算机资源,也降低来模型的维护成本。
因此,上述实施例提供的模型管理方法中,用于实现业务场景模拟的各算法模型只需要存储一次,可以利用配置规则为不同的业务场景预先配置好对应的模型调度规则,如此,当模拟不同业务场景时,只需要根据调度规则按序调用模型进行处理,而不需要在不同的业务场景下对一个模型进行多次封装,可以节省计算机资源。同时,当有新的业务场景或已有的业务场景对应的调度规则发生变化时,只需要对配置规则进行更新即可,操作快捷,可以提高开发人员的执行效率。当模型需要修改时,对存储的模型进行一次修改即可,降低了维护模型的成本。
本申请实施例还提供一种模型应用方法。请参阅图2,所述模型应用方法可以包括以下步骤。
S21:接收用户终端发出的请求信息。
服务器可以接收用户终端发出的请求信息。所述请求信息可以表示用户在一个业务场景下所需服务的内容。
在一个实施例中,所述请求信息可以包括:业务接口信息和请求内容。其中,所述业务接口信息可以用于表征与所述请求内容对应的服务项目。所述业务接口信息可以用字符来表示。所述请求内容可以是用户发出的问题、指令等。
S22:确定与所述请求信息对应的目标业务场景。
在一个实施例中,可以根据所述请求信息中的业务接口信息来确定所述目标业务场景。所述业务接口信息与业务场景可以具有一一对应关系。所述业务接口信息与业务场景的对应关系可以是预先设定的。
例如,请求信息中的业务接口信息可以为“问答”,假设预设的业务接口信息与业务场景的对应关系中,业务接口信息“问答”对应的业务场景为“智能问答”,那么,可以确定该请求信息对应的目标业务场景为“智能问答”。
S23:基于预设配置规则,确定与所述目标业务场景对应的目标调度规则。
所述服务器可以预先存储有配置规则。
在一个实施例中,所述配置规则可以包括:业务场景信息与调度规则的预设对应关系。
其中,所述业务场景信息可以是业务场景名称。所述业务场景信息可以用于表示服务项目所对应的场景。例如,业务场景信息可以是“智能问答”等。
所述调度规则可以用于表示模拟一个场景时依次需要调用的模型。在一个实施例中,所述调度规则可以包括:调度模型名称以及调度顺序。
在一个实施例中,所述模型可以是应用于人工智能信息处理的算法模型。例如,可以是人工神经网络模型、遗传算法模型等。通过利用所述模型进行信 息处理,可以模拟实现业务场景下人的操作,从而实现人工智能信息处理。例如,可以利用人工神经网络算法对语音信息进行处理,分析得到语音的情绪信息。
在一个实施例中,所述基于预设配置规则确定与所述目标业务场景对应的目标调度规则可以包括:基于预设配置规则中业务场景信息与调度规则的预设对应关系,从所述存储的配置规则中筛选出目标调度规则。所述目标调度规则可以是与所述目标业务场景对应的调度规则。
在一个实施例中,所述目标调度规则可以包括:目标模型的名称和目标顺序。根据所述目标顺序调用所述目标模型,可以实现对所述目标业务场景的模拟。
S24:根据所述目标调度规则中的目标顺序从存储的模型中调度目标模型进行信息处理,得到与所述请求信息对应的请求结果。
在一个实施例中,所述服务器中可以存储有多个模型。所述存储的多个模型可以采用不同的模型交互接口。所述存储的多个模型也可以采用统一的预设模型交互接口。
可以根据所述目标调度规则中的目标顺序调度目标模型进行信息处理,从而得到请求结果。
例如,请求信息中的请求内容为“X”,该请求信息对应的目标调度规则为:按序调度模型1、模型2。那么,将“X”作为输入信息,利用模型1进行信息处理,得到输出结果“A”,再将“A”作为输入信息,利用模型2进行信息处理,得到输出结果“B”,可以将“B”作为请求结果进行输出。
S25:将所述请求结果反馈给所述用户终端。
所述服务器可以将利用所述目标模型进行信息处理得到的请求结果反馈给所述用户终端。
本申请实施例还提供一种模型应用系统。请参阅图3,所述模型应用系统 包括:用户终端和模型应用服务器。所述模型应用服务器可以为一个服务器,也可以是由多个服务器组成的服务器集群。
所述用户终端,可以用于向所述模型应用服务器发送请求信息,以及接收与所述请求信息对应的请求结果。所述请求信息可以包括:业务接口信息和请求内容。
所述模型应用服务器,可以用于接收所述用户终端发来的请求信息,根据所述请求信息确定目标业务场景,以及基于预设配置规则确定与所述目标业务场景对应的目标调度规则,根据所述目标调度规则中的目标顺序调度目标模型进行信息处理,得到与所述请求信息对应的请求结果,将所述请求结果发送给所述用户终端。所述模型应用服务器还用于存储模型。
请参阅图4,在一个实施例中,所述模型应用服务器可以包括:配置单元、调度单元和模型及配置存储单元。
所述模型及配置存储单元,可以用于存储配置规则与模型。
所述配置单元,可以用于根据目标业务场景和所述模型及配置存储单元存储的配置规则,确定与所述目标业务场景对应的目标调度规则。
所述调度单元,可以用于根据所述配置单元确定的目标调度规则中的目标顺序从所述模型及配置存储单元调度目标模型进行信息处理,得到与所述请求信息对应的请求结果。
本申请实施例还提供一种模型管理服务器。请参阅图5,所述模型管理服务器可以包括:存储单元、消息接收单元和调度规则确定单元。
所述存储单元,可以用于存储配置规则及模型。
所述消息接收单元,可以用于接收信息。所述接收的信息可以包括:目标业务场景信息。
所述调度规则确定单元,可以用于根据所述消息接收单元接收的目标业务场景,从所述存储单元存储的配置规则中确定与所述目标业务场景对应的目 标调度规则。
在一个实施例中,所述消息接收单元还可以用于接收更新配置规则的请求。所述更新配置规则的请求可以包括:规则更新操作和规则更新内容。在该实施例中,所述存储单元还可以用于根据所述更新配置规则的请求中的规则更新操作和规则更新内容对所述存储的配置规则进行处理。其中,所述规则更新操作可以包括:新增配置规则、修改配置规则和/或删除配置规则。
在一个实施例中,所述消息接收单元还可以用于接收更新模型的请求。所述更新模型的请求可以包括:模型更新操作和模型更新内容。所述模型更新操作可以包括:新增模型、修改模型和/或删除模型。在该实施例中,所述存储单元还可以用于根据所述更新模型的请求中的模型更新操作对所述存储的模型进行信息处理。
请参阅图6,本申请还提供一种服务器,所述服务器包括存储器和处理器,所述存储器用于存储计算机程序,所述计算机程序被所述处理器执行时,可以实现上述实施例执行的方法。
请参阅图7,在本申请中,上述实施例中的技术方案可以应用于如图7所示的计算机终端10上。计算机终端10可以包括一个或多个(图中仅示出一个)处理器102(处理器102可以包括但不限于微处理器MCU或可编程逻辑器件FPGA等的处理装置)、用于存储数据的存储器104、以及用于通信功能的传输模块106。本领域普通技术人员可以理解,图7所示的结构仅为示意,其并不对上述电子装置的结构造成限定。例如,计算机终端7还可包括比图7中所示更多或者更少的组件,或者具有与图7所示不同的配置。
存储器104可用于存储应用软件的软件程序以及模块,处理器102通过运行存储在存储器104内的软件程序以及模块,从而执行各种功能应用以及数据处理。存储器104可包括高速随机存储器,还可包括非易失性存储器,如一个或者多个磁性存储装置、闪存、或者其他非易失性固态存储器。在一些实例 中,存储器104可进一步包括相对于处理器102远程设置的存储器,这些远程存储器可以通过网络连接至计算机终端10。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
具体地,在本申请中,上述的服务器的部署方法可以作为计算机程序存储于上述的存储器104中,所述存储器104可以与处理器102耦合,那么当处理器102执行所述存储器104中的计算机程序时,便可以实现上述的服务器的部署方法中的各个步骤。
传输装置106用于经由一个网络接收或者发送数据。上述的网络具体实例可包括计算机终端10的通信供应商提供的无线网络。在一个实例中,传输装置106包括一个网络适配器(Network Interface Controller,NIC),其可通过基站与其他网络设备相连从而可与互联网进行通讯。在一个实例中,传输装置106可以为射频(Radio Frequency,RF)模块,其用于通过无线方式与互联网进行通讯。
由上可见,本申请提供的技术方案中,提供了基于配置规则的模型应用方法和模型管理方法,利用配置规则为不同的业务场景预先配置好对应的模型调度规则,模拟业务场景时,根据对应的模型调度规则调用已存储的模型亟需信息处理,各算法模型只需要存储一次,既节省了计算机资源,也降低了维护模型的成本。。
通过以上的实施例的描述,本领域的技术人员可以清楚地了解到各实施例可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件来实现。基于这样的理解,上述技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来,该计算机软件产品可以存储在计算机可读存储介质中,如ROM/RAM、磁碟、光盘等,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行各个实施例或者实施例的某些部分所述的方法。
以上所述仅为本申请的较佳实施例,并不用以限制本申请,凡在本申请的精神和原则之内,所作的任何修改、等同替换、改进等,均应包含在本申请的保护范围之内。

Claims (15)

  1. 一种模型应用方法,其中,包括:
    接收用户终端发出的请求信息;
    确定与所述请求信息对应的目标业务场景;
    基于预设配置规则,确定与所述目标业务场景对应的目标调度规则;所述目标调度规则包括:目标模型的名称及目标顺序;
    根据所述目标调度规则中的目标顺序从存储的模型中调度目标模型进行信息处理,得到与所述请求信息对应的请求结果;
    将所述请求结果反馈给所述用户终端。
  2. 根据权利要求1所述的方法,其中,所述确定与所述请求信息对应的目标业务场景,包括:根据所述请求信息中的业务接口信息来确定所述目标业务场景;所述业务接口信息与业务场景具有一一对应关系。
  3. 根据权利要求1所述的方法,其中,所述基于预设配置规则确定与所述目标业务场景对应的目标调度规则包括:基于预设配置规则中业务场景信息与调度规则的预设对应关系,确定与所述目标业务场景对应的目标调度规则。
  4. 根据权利要求3所述的方法,其中,所述调度规则包括:调度模型名称以及调度顺序。
  5. 根据权利要求1所述的方法,其中,所述存储的模型采用统一的预设模型交互接口。
  6. 一种模型管理方法,其中,包括:
    提供一配置服务器,用于存储配置规则与模型,所述配置规则包括:业务 场景与调度规则的预设对应关系;所述调度规则包括:调度模型名称以及调度顺序;
    获取目标业务场景,根据所述目标业务场景和所述存储的配置规则,确定与所述目标业务场景对应的目标调度规则,反馈目标调度规则。
  7. 根据权利要求6所述的方法,其中,所述存储的模型采用统一的预设模型交互接口。
  8. 根据权利要求6所述的方法,其中,还包括:接收更新配置规则的请求,根据所述更新配置规则的请求中的规则更新操作和规则更新内容对所述存储的配置规则进行处理;所述规则更新操作包括:新增配置规则、修改配置规则和/或删除配置规则。
  9. 根据权利要求6所述的方法,其中,还包括:接收更新模型的请求,根据所述更新模型的请求中的模型更新操作对所述存储的模型进行处理,所述模型更新操作包括:新增模型、修改模型和/或删除模型。
  10. 一种模型应用系统,其中,包括:
    用户终端,用于向模型应用服务器发送请求信息,以及接收与所述请求信息对应的请求结果;所述请求信息包括:业务接口信息和请求内容;
    模型应用服务器,用于接收所述用户终端发来的请求信息,根据所述请求信息确定目标业务场景,以及基于预设配置规则确定与所述目标业务场景对应的目标调度规则,根据所述目标调度规则中的目标顺序调度目标模型进行信息处理,得到与所述请求信息对应的请求结果,将所述请求结果发送给所述用户终端;所述模型应用服务器还用于存储模型。
  11. 根据权利要求10所述的系统,其中,所述模型应用服务器包括:配置 单元、调度单元和模型及配置存储单元;其中
    所述模型及配置存储单元,用于存储配置规则与模型;
    所述配置单元,用于根据目标业务场景和所述模型及配置存储单元存储的配置规则,确定与所述目标业务场景对应的目标调度规则;
    所述调度单元,用于根据所述配置单元确定的目标调度规则中的目标顺序从所述模型及配置存储单元调度目标模型进行信息处理,得到与所述请求信息对应的请求结果。
  12. 一种模型管理服务器,其中,包括:存储单元、消息接收单元和调度规则确定单元;其中,
    所述存储单元,用于存储配置规则及模型;
    所述消息接收单元,用于接收信息;所述接收的信息包括:目标业务场景信息;
    所述调度规则确定单元,用于根据所述消息接收单元接收的目标业务场景,从所述存储单元存储的配置规则中确定与所述目标业务场景对应的目标调度规则。
  13. 根据权利要求12所述的服务器,其中,所述消息接收单元还用于接收更新配置规则的请求;
    所述存储单元还用于根据所述更新配置规则的请求中的规则更新操作和规则更新内容对所述存储的配置规则进行处理;所述规则更新操作包括:新增配置规则、修改配置规则和/或删除配置规则。
  14. 根据权利要求12所述的服务器,其中,所述消息接收单元还用于接收更新模型的请求;所述存储单元还用于根据所述更新模型的请求中的模型更新操作对所述存储的模型进行信息处理,所述模型更新操作包括:新增模型、修 改模型和/或删除模型。
  15. 一种服务器,其中,所述服务器包括存储器和处理器,所述存储器用于存储计算机程序,所述计算机程序被所述处理器执行时,实现如权利要求1至10中任一权利要求所述的方法。
PCT/CN2019/081536 2019-03-08 2019-04-04 一种模型应用方法、管理方法、系统及服务器 WO2020181599A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP19797974.3A EP3731161A1 (en) 2019-03-08 2019-04-04 Model application method and system, and model management method and server
US16/686,026 US20200286012A1 (en) 2019-03-08 2019-11-15 Model application method, management method, system and server

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910176811.XA CN109858607A (zh) 2019-03-08 2019-03-08 一种模型应用方法、管理方法、系统及服务器
CN201910176811.X 2019-03-08

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/686,026 Continuation US20200286012A1 (en) 2019-03-08 2019-11-15 Model application method, management method, system and server

Publications (1)

Publication Number Publication Date
WO2020181599A1 true WO2020181599A1 (zh) 2020-09-17

Family

ID=66900225

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/081536 WO2020181599A1 (zh) 2019-03-08 2019-04-04 一种模型应用方法、管理方法、系统及服务器

Country Status (3)

Country Link
EP (1) EP3731161A1 (zh)
CN (1) CN109858607A (zh)
WO (1) WO2020181599A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112783978A (zh) * 2021-01-28 2021-05-11 北京东方优播网络科技有限公司 数据传输方法、装置、计算机系统和存储介质
CN113645304A (zh) * 2021-08-13 2021-11-12 恒生电子股份有限公司 数据服务处理方法及相关设备
CN113923170A (zh) * 2021-09-30 2022-01-11 深信服科技股份有限公司 一种应用识别管理方法及系统
CN114125147A (zh) * 2021-11-15 2022-03-01 青岛海尔科技有限公司 设备场景功能的校验方法、场景引擎及场景平台
CN114257504A (zh) * 2021-12-15 2022-03-29 烽火通信科技股份有限公司 业务场景切换方法及系统
CN114765552A (zh) * 2021-01-04 2022-07-19 航天信息股份有限公司 数据处理方法、中台系统、存储介质和电子设备
CN114826715A (zh) * 2022-04-15 2022-07-29 咪咕文化科技有限公司 网络防护方法、装置、设备及存储介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110874244A (zh) * 2019-10-16 2020-03-10 贝壳技术有限公司 适用于不同业务场景的过滤方法、装置及存储介质
CN111159391B (zh) * 2020-01-02 2024-02-23 中国银行股份有限公司 业务场景数据筛选装置及方法
CN111427677B (zh) * 2020-03-20 2024-03-01 网易(杭州)网络有限公司 人工智能产品的生成方法、装置和服务器
CN111880864B (zh) * 2020-07-30 2023-12-26 平安国际智慧城市科技股份有限公司 基于http的模型调用方法、系统、计算机设备和存储介质
CN113900828B (zh) * 2021-12-08 2022-03-04 深圳致星科技有限公司 联邦学习专用处理器、联邦学习处理芯片及芯片

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101277304A (zh) * 2008-03-21 2008-10-01 北京航空航天大学 基于规则的Web服务运行环境管理系统及管理方法
US20120123764A1 (en) * 2009-08-07 2012-05-17 Yasuhiro Ito Computer System, Program, and Method for Assigning Computational Resource to be Used in Simulation
CN105335313A (zh) * 2014-08-07 2016-02-17 阿里巴巴集团控股有限公司 一种基础数据的传输方法及装置

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103186579A (zh) * 2011-12-29 2013-07-03 北京亿阳信通科技有限公司 一种基于数据库函数的规则实现及应用的方法及装置
US10395181B2 (en) * 2015-06-05 2019-08-27 Facebook, Inc. Machine learning system flow processing
CN105100117B (zh) * 2015-08-28 2018-07-13 北京奇艺世纪科技有限公司 一种用于访问组合服务的方法及装置
WO2017039623A1 (en) * 2015-08-31 2017-03-09 Hewlett Packard Enterprise Development Lp Generating a combined entity workflow
EP3287930A1 (en) * 2016-08-26 2018-02-28 Entit Software LLC Javascript security testing
CN108268319A (zh) * 2016-12-31 2018-07-10 中国移动通信集团河北有限公司 任务调度方法、装置及系统
CN107918548A (zh) * 2017-07-25 2018-04-17 平安科技(深圳)有限公司 业务功能实现的方法、装置、计算机设备及存储介质
CN107562421A (zh) * 2017-09-28 2018-01-09 北京神州泰岳软件股份有限公司 一种自然语言处理方法及处理平台
CN108958728A (zh) * 2018-05-18 2018-12-07 深圳壹账通智能科技有限公司 贷款产品开发方法、装置、设备和计算机存储介质

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101277304A (zh) * 2008-03-21 2008-10-01 北京航空航天大学 基于规则的Web服务运行环境管理系统及管理方法
US20120123764A1 (en) * 2009-08-07 2012-05-17 Yasuhiro Ito Computer System, Program, and Method for Assigning Computational Resource to be Used in Simulation
CN105335313A (zh) * 2014-08-07 2016-02-17 阿里巴巴集团控股有限公司 一种基础数据的传输方法及装置

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3731161A4 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114765552A (zh) * 2021-01-04 2022-07-19 航天信息股份有限公司 数据处理方法、中台系统、存储介质和电子设备
CN114765552B (zh) * 2021-01-04 2023-11-07 航天信息股份有限公司 数据处理方法、中台系统、存储介质和电子设备
CN112783978A (zh) * 2021-01-28 2021-05-11 北京东方优播网络科技有限公司 数据传输方法、装置、计算机系统和存储介质
CN113645304A (zh) * 2021-08-13 2021-11-12 恒生电子股份有限公司 数据服务处理方法及相关设备
CN113645304B (zh) * 2021-08-13 2023-06-16 恒生电子股份有限公司 数据服务处理方法及相关设备
CN113923170A (zh) * 2021-09-30 2022-01-11 深信服科技股份有限公司 一种应用识别管理方法及系统
CN114125147A (zh) * 2021-11-15 2022-03-01 青岛海尔科技有限公司 设备场景功能的校验方法、场景引擎及场景平台
CN114257504A (zh) * 2021-12-15 2022-03-29 烽火通信科技股份有限公司 业务场景切换方法及系统
CN114257504B (zh) * 2021-12-15 2023-05-23 烽火通信科技股份有限公司 业务场景切换方法及系统
CN114826715A (zh) * 2022-04-15 2022-07-29 咪咕文化科技有限公司 网络防护方法、装置、设备及存储介质
CN114826715B (zh) * 2022-04-15 2024-03-22 咪咕文化科技有限公司 网络防护方法、装置、设备及存储介质

Also Published As

Publication number Publication date
EP3731161A4 (en) 2020-10-28
CN109858607A (zh) 2019-06-07
EP3731161A1 (en) 2020-10-28

Similar Documents

Publication Publication Date Title
WO2020181599A1 (zh) 一种模型应用方法、管理方法、系统及服务器
CN112527528A (zh) 基于消息队列的数据传输方法、装置及存储介质
CN107450899B (zh) 终端控制脚本的生成方法和装置
WO2020006827A1 (zh) 一种智能应答的方法和装置
CN104702675B (zh) 用户事件的响应方法及装置
CN108306804A (zh) 一种Ethercat主站控制器及其通信方法和系统
CN110781180B (zh) 一种数据筛选方法和数据筛选装置
CN115022304B (zh) 基于rpa和ai实现ia的跨平台文件处理方法、装置和系统
CN116755844B (zh) 一种仿真引擎的数据处理方法、装置、设备及存储介质
CN110825807A (zh) 基于人工智能的数据交互转换方法、装置、设备及介质
CN116431316B (zh) 任务处理方法、系统、平台及自动问答方法
CN108900627B (zh) 一种网络请求方法、终端装置及存储介质
US20200286012A1 (en) Model application method, management method, system and server
CN109117127A (zh) 程序代码生成方法、装置及计算机可读存储介质
CN106027365A (zh) 一种信息交互的方法及装置
CN110502645A (zh) 信息查询方法及装置
CN104636249A (zh) 一种应用程序调试信息生成系统和方法
CN116700907B (zh) 服务调用需求文档存储方法、装置、电子设备和可读介质
CN111813529B (zh) 数据处理方法、装置、电子设备及存储介质
CN115277613A (zh) 一种基于业务的机器人聊天系统
CN115328997A (zh) 数据同步方法、系统、设备及存储介质
CN114329128A (zh) 标注数据的获取方法、装置、计算机设备和存储介质
CN112445811A (zh) 基于sql配置的数据服务方法、装置、存储介质及组件
CN109960490B (zh) 生成智能问答系统的方法、装置、设备、介质及问答系统
CN107291455B (zh) 一种基于工厂模式实现转账业务的方法及系统

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2019797974

Country of ref document: EP

Effective date: 20191111

ENP Entry into the national phase

Ref document number: 2019797974

Country of ref document: EP

Effective date: 20191113

NENP Non-entry into the national phase

Ref country code: DE