CN-121981223-A - Method, device, equipment and storage medium for generating reply information

Abstract

The application discloses a method, a device, equipment and a storage medium for generating reply information, which belong to the field of artificial intelligence and are used for accurately generating user reply information. The method comprises the steps of obtaining problem information input by a target user, carrying out text vectorization processing on the problem information to obtain a first feature vector, carrying out scene recognition based on the first feature vector, determining a corresponding first preset model and a corresponding second preset model based on recognition results, wherein the first preset model and the second preset model are used for generating reply information in different reply information generation stages, inputting the first feature vector into the first preset model to obtain a first output result, the first output result is a reply information first draft generated aiming at the problem information, inputting the first output result into the second preset model to obtain a second output result, and determining user reply information based on the second output result, wherein the second output result is a reply information final draft generated aiming at the first output result.

Inventors

• LI TAO
• WANG KANG

Assignees

• 郑州阿帕斯数云信息科技有限公司

Dates

Publication Date

20260505

Application Date

20251230

Claims (10)

1. 1. The method for generating the reply information is characterized by comprising the following steps: Acquiring problem information input by a target user, and performing text vectorization processing on the problem information to obtain a first feature vector; Scene recognition is carried out based on the first feature vector, and a corresponding first preset model and a corresponding second preset model are determined based on a recognition result, wherein the first preset model and the second preset model are used for generating reply information in different reply information generation stages; inputting the first feature vector into the first preset model to obtain a first output result, wherein the first output result is a reply information manuscript generated aiming at the problem information; And inputting the first output result into the second preset model to obtain a second output result, so as to determine user reply information based on the second output result, wherein the second output result is a reply information final draft generated for the first output result.
2. 2. The method of claim 1, wherein determining the corresponding first and second predetermined models based on the recognition result comprises: Determining the first preset model, the second preset model and a third preset model based on the identification result, wherein the third preset model is used for carrying out structural adjustment on the generated reply information; The determining the user reply information based on the second output result includes: And inputting the second output result into the third preset model to obtain a third output result, determining the third output result as the user reply information, wherein the third output result is the reply information after the structure of the second output result is adjusted.
3. 3. The method of claim 1, wherein after the determining the user reply message based on the second output result, the method further comprises: acquiring a preset knowledge base, wherein the preset knowledge base comprises knowledge of a plurality of different scene fields; carrying out knowledge verification on the user reply information based on the preset knowledge base; and when the knowledge verification is passed, the user reply information is sent to the terminal equipment of the target user.
4. 4. The method according to claim 1, wherein the scene recognition based on the first feature vector and determining the first preset model and the second preset model based on the recognition result comprises: performing scene recognition based on the first feature vector, and determining the scene field of the first feature vector; obtaining a plurality of fourth preset models, determining the fourth preset models corresponding to the scene fields in the fourth preset models as the second preset models, wherein different fourth preset models are used for processing user problem information in different scene fields.
5. 5. The method according to claim 4, wherein the method further comprises: Receiving a user operation instruction sent by an operator, wherein the user operation instruction comprises a first operation instruction and a second operation instruction, the first operation instruction is used for adding or deleting the fourth preset model, and the second operation instruction is used for adjusting information generation strategies of the first preset model and the fourth preset model; And executing the operation corresponding to the user operation instruction.
6. 6. The method according to claim 1, wherein the method further comprises: Acquiring user history dialogue information, and storing the user history dialogue information into a first preset database, wherein the user history dialogue information is generated by dialogue with the target user, and the first preset database is used for storing weight data; acquiring user attribute information, and storing the user attribute information into a second preset database, wherein the user attribute information is related to the target user, and the second preset database is used for storing lightweight data; And obtaining user token data, and storing the user token data into a third preset database, wherein the user token data are token data related to the target user, and the third preset database is used for storing data needing to be stored in an encrypted mode.
7. 7. The method of claim 1, wherein after the determining the user reply message based on the second output result, the method further comprises: Generating a plurality of continuous user reply sub-messages based on the second output result, wherein the user reply sub-messages are part of the user reply messages; And transmitting the user reply sub-information to the terminal equipment of the target user one by one.
8. 8. A reply message generation apparatus, comprising: The first acquisition module is used for acquiring problem information input by a target user, and performing text vectorization processing on the problem information to obtain a first feature vector; The first determining module is used for carrying out scene recognition based on the first feature vector, and determining a corresponding first preset model and a corresponding second preset model based on a recognition result, wherein the first preset model and the second preset model are used for generating reply information in different reply information generation stages; the first output module is used for inputting the first feature vector into the first preset model to obtain a first output result, and the first output result is a reply information manuscript generated aiming at the problem information; the second output module is used for inputting the first output result into the second preset model to obtain a second output result so as to determine user reply information based on the second output result, wherein the second output result is a reply information final draft generated for the first output result.
9. 9. An electronic device, the device comprising: processor, and A memory arranged to store computer executable instructions configured to be executed by the processor, the executable instructions comprising a method for performing the reply information generation of any one of claims 1 to 7.
10. 10. A storage medium storing computer-executable instructions for causing a computer to perform the reply information generation method according to any one of claims 1 to 7.

Description

Method, device, equipment and storage medium for generating reply information Technical Field The application belongs to the field of artificial intelligence, and particularly relates to a method, a device, equipment and a storage medium for generating reply information. Background With the rapid development of large-model technology, various AI applications are greatly developed, but the prior art still has the following problems that most AI applications only support a single dialogue scene, for example ChatGPT only provide general questions and answers, a vertical field model only provides single field services, and multiple fields cannot be deeply integrated in one mobile application, so that the problem of low accuracy of generated reply information can occur. Therefore, a method capable of accurately generating reply information is required. Disclosure of Invention The embodiment of the application provides a method for generating reply information, which can accurately generate the reply information. The embodiment of the application provides a reply information generation method, which comprises the steps of obtaining problem information input by a target user, carrying out text vectorization processing on the problem information to obtain a first feature vector, carrying out scene recognition based on the first feature vector, determining a corresponding first preset model and a corresponding second preset model based on recognition results, wherein the first preset model and the second preset model are used for generating reply information in different reply information generation stages, inputting the first feature vector into the first preset model to obtain a first output result, wherein the first output result is a reply information manuscript generated for the problem information, inputting the first output result into the second preset model to obtain a second output result, and determining user reply information based on the second output result, wherein the second output result is a reply information final manuscript generated for the first output result. The embodiment of the application provides a reply information generating device, which comprises a first acquisition module, a first determination module, a first output module and a second output module, wherein the first acquisition module is used for acquiring problem information input by a target user and carrying out text vectorization processing on the problem information to obtain a first feature vector, the first determination module is used for carrying out scene recognition based on the first feature vector and determining a corresponding first preset model and a corresponding second preset model based on recognition results, the first preset model and the second preset model are used for generating reply information in different reply information generating stages, the first output module is used for inputting the first feature vector into the first preset model to obtain a first output result, the first output result is a reply information initial draft generated for the problem information, the second output module is used for inputting the first output result into the second preset model to obtain a second output result, and the second output result is used for determining user reply information based on the second output result and is a reply information final draft generated for the first output result. In a third aspect, an embodiment of the present application provides an electronic device, including a processor, a memory, and a program or instruction stored on the memory and executable on the processor, the program or instruction implementing the steps of the method according to the first aspect when executed by the processor. In a fourth aspect, embodiments of the present application provide a readable storage medium having stored thereon a program or instructions which when executed by a processor perform the steps of the method according to the first aspect. In a fifth aspect, an embodiment of the present application provides a chip, where the chip includes a processor and a communication interface, where the communication interface is coupled to the processor, and where the processor is configured to execute a program or instructions to implement a method according to the first aspect. According to the method and the device for generating the reply information, the problem information input by the target user is firstly obtained, text vectorization processing is conducted on the problem information to obtain a first feature vector, scene recognition is conducted on the basis of the first feature vector, a corresponding first preset model and a corresponding second preset model are determined on the basis of recognition results, the first preset model and the second preset model are used for generating the reply information in different reply information generation stages, the first feature vector is input into the