Large model information processing method, device, electronic device, storage medium and computer program product

Abstract

This disclosure provides a large-scale model information processing method, apparatus, electronic device, and storage medium, relating to the field of data processing technology. A specific implementation scheme includes: a serverless service component receives a Hypertext Transfer Protocol (HTTP) request; based on the HTTP request, the serverless service component determines a target service component to be called from a collection of service components in the large-scale model, and determines one or more target search engines from a search engine cluster in the large-scale model; and the serverless service component interacts with the target search engine and target service component to process the HTTP request.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a large-scale model information processing method, device, electronic device, and storage medium. Background Art

[0002] As a foundational application for large models, knowledge bases can effectively eliminate illusions and provide untrained knowledge to large models, reducing learning time and training costs. However, the construction of knowledge bases is complex, making deployment difficult. Summary of the Invention

[0003] The present disclosure provides an information processing method, device, electronic device and storage medium for large models.

[0004] According to one aspect of the present disclosure, a method for processing information of a large model is provided, including: a serverless service component receiving a Hypertext Transfer Protocol (HTTP) request; the Serverless service component determining, based on the HTTP request, a target service component to be called from a set of service components of the large model, and determining one or more target search engines from a search engine cluster of the large model; the Serverless service component interacting with the target search engine and the target service component to process the HTTP request.

[0005] According to another aspect of the present disclosure, a large model information processing device is provided, including: a receiving module, configured for a serverless service component to receive a Hypertext Transfer Protocol (HTTP) request; a determining module, configured for the Serverless service component to determine, based on the HTTP request, a target service component to be called from a set of service components of the large model, and to determine one or more target search engines from a search engine cluster of the large model; and a processing module, configured for the Serverless service component to interact with the target search engine and target service component, and to process the HTTP request.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the large model information processing method described in the above-mentioned embodiment.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, on which a computer program / instruction is stored. The computer instructions are used to enable the computer to execute the large model information processing method described in the embodiment of the above aspect.

[0008] According to another aspect of the present disclosure, a computer program product is provided, including a computer program / instruction, which, when executed by a processor, implements the large model information processing method described in the embodiment of the first aspect.

[0009] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are used to better understand the present invention and do not constitute a limitation of the present invention.

[0011] Figure 1 A flowchart of a large model information processing method provided by an embodiment of the present disclosure;

[0012] Figure 2 A schematic diagram of a service component set of a large model provided by an embodiment of the present disclosure;

[0013] Figure 3 A flowchart of another large model information processing method provided by an embodiment of the present disclosure;

[0014] Figure 4 A flowchart of another large model information processing method provided by an embodiment of the present disclosure;

[0015] Figure 5 A flowchart of another large model information processing method provided by an embodiment of the present disclosure;

[0016] Figure 6 A flowchart of another large model information processing method provided by an embodiment of the present disclosure;

[0017] Figure 7 A flowchart of another large model information processing method provided by an embodiment of the present disclosure;

[0018] Figure 8 A schematic structural diagram of a large-scale information processing device provided in an embodiment of the present disclosure;

[0019] Figure 9 The present invention is a block diagram of an electronic device for implementing the information processing method of the large model according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0020] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0021] The information processing method of the large model provided by the embodiment of the present disclosure can be applied in the fields of business document question answering, customer service systems, internal knowledge retrieval systems, expert systems, etc.

[0022] The following describes the information processing method, device, and electronic device of a large model according to embodiments of the present disclosure with reference to the accompanying drawings.

[0023] Artificial Intelligence (AI) is the study of how computers can simulate certain human thought processes and intelligent behaviors (such as learning, reasoning, thinking, and planning). This discipline encompasses both hardware and software technologies. AI hardware technologies generally include computer vision, speech recognition, natural language processing, as well as deep learning / learning, big data processing, and knowledge graphs.

[0024] Natural Language Processing (NLP) is a key area of ​​research in computer science and artificial intelligence. It studies the theories and methods that enable effective communication between humans and computers using natural language. NLP integrates linguistics, computer science, and mathematics. It is primarily used in machine translation, public opinion monitoring, automatic summarization, opinion extraction, text classification, question answering, text semantic comparison, and speech recognition.

[0025] Deep learning (DL) is a new research direction in machine learning (ML). It was introduced to bring ML closer to its original goal: artificial intelligence. Deep learning studies the inherent laws and representational hierarchies of sample data. The information gained from this learning process is highly helpful in interpreting data such as text, images, and sound. Its ultimate goal is to enable machines to acquire human-like analytical learning capabilities and recognize data such as text, images, and sound. Deep learning is a complex machine learning algorithm that has achieved results in speech and image recognition that far surpass previous technologies.

[0026] Large models, as used in machine learning and artificial intelligence, are characterized by large numbers of parameters and complex structures. These models typically require extensive computing resources to train and deploy, and are capable of processing and understanding larger and more complex amounts of data. Large models can be applied to a variety of applications, including automated writing, chatbots, virtual assistants, voice assistants, and automatic translation.

[0027] Figure 1 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0028] like Figure 1 As shown, the information processing method of the large model may include:

[0029] S101: The serverless service component receives a Hypertext Transfer Protocol (HTTP) request.

[0030] It should be noted that the execution entity of the large-scale model information processing method in the embodiments of the present disclosure may be a hardware device with information processing capabilities and / or the necessary software to drive the operation of the hardware device. Optionally, the execution entity may include an in-vehicle terminal, a user terminal, and other intelligent devices. User terminals include but are not limited to mobile phones, computers, intelligent voice interaction devices, etc. This is not specifically limited in the embodiments of the present disclosure.

[0031] As you can understand, serverless is a cloud computing model that enables the deployment and operation of applications without the need to manage servers. Serverless service components are the tools and frameworks used to build and manage serverless applications.

[0032] In some implementations, a serverless service component can operate by receiving and responding to Hypertext Transfer Protocol (HTTP) requests. Alternatively, the HTTP request can instruct the serverless service component to complete various requests, such as building a knowledge base or retrieving documents.

[0033] For example, Figure 2 The diagram shows a collection of service components in the large model. Serverless service components can receive HTTP requests based on HTTP triggers.

[0034] S102: Based on the HTTP request, the Serverless service component determines the target service component to be called from the service component set of the large model, and determines one or more target search engines from the search engine cluster of the large model.

[0035] In some implementations, the service component set of the large model includes: Serverless service component, word vector generation component, extraction service component, word vector storage component, cluster service component, migration service component, non-relational database component, cloud storage component, etc.

[0036] In some implementations, the Serverless service component determines the type of HTTP request and, based on different types of HTTP requests, calls different components from the service component collection as target service components, and determines one or more available target search engines from the search engine cluster.

[0037] Optionally, a correspondence between an HTTP request and a service component may be pre-established. After the HTTP request is determined, the target service component corresponding to the HTTP request may be determined by querying the relationship.

[0038] Optionally, the target search engine can be used to construct a knowledge base, and documents can be stored in the target search engine corresponding to the knowledge base, and then the target search engine can be searched.

[0039] Optionally, the HTTP request includes: a knowledge base construction request, a document storage request, a document retrieval request, and a knowledge base migration request.

[0040] For example, Figure 2 The diagram shows a large model service component set, which includes: Serverless service components (online services), word vector generation components (embedding services), extraction services (extract services), word vector storage components (storage services), cluster services (cluster services), migration services, non-relational database components (Mongo services), and cloud storage components (bos services). The online services, embedding services, extract services, storage services, cluster services, and migration services can be deployed using serverless deployment.

[0041] When the online service receives an HTTP request based on an HTTP trigger, it can determine the target service component to be called from the above service components and determine one or more target search engines from the search engine cluster (elasticsearch cluster).

[0042] S103: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

[0043] In some implementations, the Serverless service component interacts with the target service component and target search engine corresponding to the HTTP request based on different HTTP requests. That is, the Serverless service component can call the target service component and target search engine to implement different requests indicated by the HTTP request.

[0044] For example, if the HTTP request is a document storage request, the Serverless service component can store the document in the target search engine by calling the target service component to implement document storage.

[0045] According to the information processing method of the large model provided by the embodiment of the present disclosure, the large model receives HTTP requests based on the Serverless service component, and can determine the target service component to be called from the service component set of the large model, as well as the target search engine. The Serverless service component interacts with the target service component and the target search engine to process the HTTP request, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the large model, and ensure the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0046] Figure 3 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0047] like Figure 3 As shown, the information processing method of the large model may include:

[0048] S301, the serverless service component receives a Hypertext Transfer Protocol (HTTP) request.

[0049] The relevant contents of step S301 can be found in the above embodiment and will not be repeated here.

[0050] S302, if the HTTP request is a request to build a knowledge base of a large model, the Serverless service component determines the cluster service component and the non-relational database component as the target service component from the service component set based on the knowledge base construction request.

[0051] In some implementations, after receiving an HTTP request, the Serverless service component can determine the type of the request based on the HTTP request. When it is determined that the HTTP request is a request to build a knowledge base for a large model, the service component required to build the knowledge base can be determined from the service component set as the target service component.

[0052] Optionally, a cluster service component and a non-relational database component can be used as target service components, wherein the cluster service component is used to obtain a search engine, and the non-relational database component is used to store and manage unstructured data.

[0053] S303: The Serverless service component calls the cluster service component, and the cluster service component determines an available search engine from the search engine cluster as the target search engine.

[0054] In some implementations, the serverless service component calls the cluster service component, reads the search engine cluster, and determines an available search engine from the search engine cluster as the target search engine. An available search engine indicates that the search engine is operational, meaning that the target search engine can provide effective and reliable search services.

[0055] S304: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

[0056] In some implementations, the target service component includes a cluster service component and a non-relational database component. The Serverless service component can build a knowledge base and store knowledge base information by interacting with the target search engine and the target service component.

[0057] Optionally, the cluster service component obtains the target search engine's first metadata and feeds it back to the serverless service component. The serverless service component then stores the first metadata in a non-relational database component, enabling asynchronous storage of the first metadata to improve the performance of large models. This first metadata includes index information, search engine cluster information, target search engine information, and other information.

[0058] Optionally, the index information can be generated by the cluster service component, thereby realizing the construction of the knowledge base. The cluster service component generates a first index for the target search engine and determines the first index as a meta-information of the target search engine. The first index can serve as a knowledge base.

[0059] For example, Figure 2 A schematic diagram of the service component set of the large model shown, Figure 2In the target service component, the cluster service component refers to the cluster service, the serverless service component refers to the online service, and the non-relational database component refers to the Mongo service. After receiving the HTTP request, the online service determines that it is a knowledge base build request, selects the cluster service and the Mongo service as target service components, and determines the target search engine from the Elasticsearch cluster.

[0060] The online service then calls the cluster service to read the Elasticsearch cluster, identifies an available search engine as the target search engine, and generates a first index for the target search engine as metadata. The cluster service obtains the first index, Elasticsearch cluster information, and target search engine information as first metadata, and feeds the first metadata back to the online service, which then stores the first metadata in the Mongo service.

[0061] According to the information processing method of the large model provided by the embodiment of the present disclosure, the large model receives an HTTP request based on the Serverless service component, determines that it is a knowledge base construction request, and can determine the cluster service component and the non-relational database component as the target service component from the service component set of the large model, and determine the available search engine as the target search engine. The Serverless service component interacts with the target service component and the target search engine to build a knowledge base, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the large model, and ensure the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0062] Figure 4 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0063] like Figure 4 As shown, the information processing method of the large model may include:

[0064] S401: The serverless service component receives a Hypertext Transfer Protocol (HTTP) request.

[0065] The relevant contents of step S401 can be found in the above embodiment and will not be repeated here.

[0066] S402, if the HTTP request is a document storage request, based on the document storage request, determine the extraction service component, cloud storage component, word vector storage component, non-relational database component and word vector generation component from the service component set as the target service component.

[0067] In some implementations, after receiving an HTTP request, the Serverless service component can determine the type of the request based on the HTTP request. When it is determined that the HTTP request is a document storage request, the service component required for document storage can be determined from the service component collection as the target service component.

[0068] Optionally, the target service components include an extraction service component, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component. The extraction service component is used to parse documents, the word vector generation component is used to convert document characters into word vectors, the cloud storage component is used to store documents, the word vector storage component is used to store word vectors, and the non-relational database component is used to store and manage unstructured data.

[0069] S403: The Serverless service component determines a target search engine from the search engine cluster based on the second index carried in the document storage request.

[0070] In some implementations, the serverless service component can obtain a second index based on the document storage request and determine the target search engine based on the second index. Optionally, the second index includes target search engine identification information, and the serverless service component can determine the target search engine from the search engine cluster based on the identification information.

[0071] S404: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

[0072] In some implementations, the target service component includes an extraction service component, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component. The serverless service component interacts with the target search engine and the target service component to store the original document to be stored in the target search engine.

[0073] Optionally, the Serverless service component sends the original document to the extraction service component, which parses the original document to obtain a parsed document and feeds the parsed document back to the Serverless service component.

[0074] Optionally, the Serverless service component can also send the original document to the cloud storage component, which then stores the original document.

[0075] Furthermore, the Serverless service component calls the word vector generation component, which converts the string in the parsed document to obtain a first word vector and feeds the first word vector back to the Serverless service component. The Serverless service component queries the non-relational database component based on the second index to obtain the second metadata of the target search engine. This second metadata may include information such as the target search engine's address and access password. Based on the second metadata, the Serverless service component sends the word vector to the word vector storage component, which then stores the word vector in the target search engine.

[0076] Optionally, after storing the first word vector in the word vector storage component, the Serverless service component can call the non-relational database component and modify the word vector storage status in the non-relational database component. Storage statuses include storing, storage failed, and storage successful. For example, after successfully storing the word vector in the word vector storage component, the storage status in the non-relational database component can be modified to storage successful.

[0077] In some implementations, to cope with increased load and improve the performance, responsiveness, and availability of large models, the service components of the large model can be scaled up. Alternatively, the service components within the service component set can be scaled up based on the elasticity mechanisms of serverless deployments. Elasticity refers to the automatic adjustment of computing resources based on actual load conditions to ensure the system can quickly respond to user requests and maintain high availability.

[0078] For example, Serverless service components, word vector generation components, extraction service components, word vector storage components, cluster service components, and migration service components are all deployed based on serverless, and the number of the above service components can be automatically increased to achieve capacity expansion.

[0079] Optionally, you can also access the cluster service component in the service component collection through the Serverless service component and expand the search engine cluster by registering a new search engine in the cluster service component.

[0080] For example, Figure 2 A schematic diagram of the service component set of the large model shown, Figure 2 In the target service component, the extraction service component refers to the extract service, the cloud storage component refers to the boss service, the word vector storage component refers to the storage service, the non-relational database component refers to the mongo service, the word vector generation component refers to the embedding service, and the serverless service component refers to the online service.

[0081] After receiving the HTTP request, the online service determines it's a document storage request and sends the original document to the extract service. The extract service parses the original document to generate a parsed document, stores it in the boss service, and then feeds the parsed document back to the online service. The online service then calls the embedding service to convert the parsed document into a first word vector and feeds the first word vector back to the online service. Based on the second index in the document storage request, the online service queries the Mongo service to obtain the second metadata of the target search engine. Based on this second metadata, the online service then sends the first word vector to the storage service, which then stores it in the target search engine.

[0082] Table 1 shows the fields in the search engine, as well as their types and uses.

[0083] Table 1

[0084] Field Name Field Type Field Purpose vector Vector Type Storing word vectors doc_id String Marking Documents segment_id String Mark document paragraphs enabled Boolean type Mark whether the paragraph is enabled

[0085] As can be seen from Table 1, the first word vector can be stored in the target search engine based on the vector field.

[0086] According to the information processing method of the large model provided by the embodiment of the present disclosure, the large model receives an HTTP request based on the Serverless service component, determines that it is a document storage request, and can determine the extraction service component, cloud storage component, word vector storage component, non-relational database component and word vector generation component from the service component set of the large model as the target service component, and determines the target search engine based on the second index carried by the document storage request. The Serverless service component interacts with the target service component and the target search engine to store the document in the target search engine, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the large model, and ensure the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base, provide a basis for processing large-scale traffic, and at the same time, based on the serverless elastic mechanism, the service components can be expanded to cope with the increase in load and improve the performance of the large model.

[0087] Figure 5 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0088] like Figure 5 As shown, the information processing method of the large model may include:

[0089] S501, the serverless service component receives a Hypertext Transfer Protocol (HTTP) request.

[0090] The relevant contents of step S501 can be found in the above embodiment and will not be repeated here.

[0091] S502: If the HTTP request is a document retrieval request, the Serverless service component determines the cloud storage component, the word vector storage component, the non-relational database component, and the word vector generation component from the service component set as the target service component based on the document retrieval request.

[0092] In some implementations, after receiving an HTTP request, a serverless service component can determine the type of the request based on the HTTP request. If the HTTP request is determined to be a document retrieval request, the serverless service component can be selected from a collection of service components as the target service component. The document retrieval request includes a search term.

[0093] Optionally, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component can be used as target service components. The word vector generation component is used to convert document characters into word vectors, the cloud storage component is used to store documents, the word vector storage component is used to store word vectors, and the non-relational database component is used to store and manage unstructured data.

[0094] S503: Determine at least one target search engine from the search engine cluster based on the at least one third index carried in the document retrieval request.

[0095] In some implementations, the serverless service component can obtain a third index based on a document retrieval request. Because multiple search engines can be searched during a document retrieval, the document retrieval request can include multiple third indexes, and the target search engine can be determined based on the third indexes. Optionally, the third index includes target search engine identification information, and the serverless service component can determine the target search engine from a search engine cluster based on the identification information.

[0096] S504: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

[0097] In some implementations, the target service component includes a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component. The serverless service component interacts with the target search engine and the target service component to retrieve documents in the target search engine.

[0098] Optionally, the serverless service component calls a word vector generator component, which converts the search terms in the document retrieval request into search word vectors and feeds the search word vectors back to the serverless service component. The serverless service component then queries the non-relational database component based on the third index to obtain the third-party metadata of the target search engine. This third-party metadata may include information such as the target search engine's address and access password.

[0099] Furthermore, the serverless service component searches the target search engine's associated knowledge base based on the third-party information and the search term vector, obtains the query results, and feeds them back to the serverless service component. Based on the query results, the serverless service component accesses the cloud storage component, obtains the paragraphs that match the search term, and feeds them back to the client.

[0100] Optionally, the similarity between the search term vector and the first term vector can be calculated to determine the score of the first term vector matching the search term, thereby determining the query result. The query result may include the score of the first term vector, identification information of the first term vector, paragraph identification information, etc.

[0101] Furthermore, the Serverless service component can filter out the first word vector with the highest matching degree from the query results, sort the scores in descending order, and select the top first word vector to access the cloud storage component.

[0102] Optionally, based on the paragraph identification information corresponding to the identification information of the first word vector in the query result, the cloud storage component can obtain the paragraph corresponding to the identification information as the paragraph matching the search term and feed it back to the client.

[0103] For example, Figure 2 A schematic diagram of the service component set of the large model shown, Figure 2 In the target service component, the cloud storage component refers to the boss service, the word vector storage component refers to the storage service, the non-relational database component refers to the mongo service, the word vector generation component refers to the embedding service, and the serverless service component refers to the online service.

[0104] After receiving the HTTP request, the online service identifies it as a document retrieval request and calls the embedding service to convert the search terms in the document retrieval request into a search term vector. The vector is then fed back to the online service. The online service then queries the Mongo service based on the third-party index to obtain the target search engine's third-party metadata. Based on this third-party metadata and the search term vector, the online service searches the target search engine's associated knowledge base, obtains the query results, and feeds them back to the online service. Based on the query results, the online service then accesses the BOS service to retrieve the paragraphs matching the search terms and returns them to the client.

[0105] According to the information processing method of the big model provided by the embodiment of the present disclosure, the big model receives an HTTP request based on the Serverless service component, determines that it is a document retrieval storage request, and can use the cloud storage component, word vector storage component, non-relational database component and word vector generation component from the service component set of the big model as the target service component, and determines the target search engine based on the third index carried by the document retrieval request. The Serverless service component interacts with the target service component and the target search engine to realize the retrieval of documents based on the search terms, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the big model, and ensure the availability of the big model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0106] Figure 6 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0107] like Figure 6 As shown, the information processing method of the large model may include:

[0108] S601: The serverless service component receives a Hypertext Transfer Protocol (HTTP) request.

[0109] The relevant contents of step S601 can be found in the above embodiment and will not be repeated here.

[0110] S602, if the HTTP request is a knowledge base migration request, the Serverless service component determines the non-relational database component and the migration service component from the service component set as the target service components based on the migration request.

[0111] In some implementations, after receiving an HTTP request, the Serverless service component can determine the type of the request based on the HTTP request. When it is determined that the HTTP request is a request to migrate the knowledge base, the service component required to migrate the knowledge base can be determined from the service component set as the target service component.

[0112] Optionally, a non-relational database component and a migration service component can be used as target service components, wherein the migration service component is used to migrate data from a search engine to another search engine, and the non-relational database component is used to store and manage unstructured data.

[0113] S603: The Serverless service component selects a search engine to be migrated out and a search engine to be migrated into from the search engine cluster as a target search engine based on the fourth index and one or more fifth indexes carried in the migration request.

[0114] In some implementations, because the knowledge base migration involves migrating data from one search engine to one or more other search engines, the serverless service component can obtain a fourth index based on the migration request and identify the search engine to be migrated out of from the search engine cluster. The serverless service component can also obtain one or more fifth indices based on the migration request, identify the search engine to be migrated into from the search engine cluster, and select the search engines to be migrated out of and into as target search engines.

[0115] Optionally, the fourth index and the fifth index include target search engine identification information, and the Serverless service component can determine the target search engine from the search engine cluster based on the identification information.

[0116] S604: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

[0117] In some implementations, the target service component includes a non-relational database component and a migration service component. The serverless service component can implement knowledge base migration by interacting with the target search engine and the target service component.

[0118] Optionally, the serverless service component queries the non-relational database component based on the fourth index to obtain fourth metadata of the search engine to be migrated out of, and queries the non-relational database component based on the fifth index to obtain fifth metadata of the search engine to be migrated into, and then feeds the fourth and fifth metadata back to the migration service component. The fourth metadata may include information such as the address and access password of the search engine to be migrated out of, and the fifth metadata may include information such as the address and access password of the search engine to be migrated into.

[0119] Furthermore, the migration service component accesses the search engine to be migrated based on the fourth metadata to obtain the first word vector to be migrated from the search engine to be migrated. The migration service component accesses the search engine to be migrated to based on the fifth metadata and migrates the first word vector to be migrated to the search engine to be migrated.

[0120] Optionally, if there are multiple search engines to be migrated, the migration service component will split the first word vector to be migrated out to obtain the split first word vector, and migrate the split first word vectors to the corresponding search engines to be migrated. This can enable multiple search engines to work independently, reduce interference, and improve the stability of the large model.

[0121] For example, Figure 2 A schematic diagram of the service component set of the large model shown, Figure 2 In the target service component, the non-relational database component refers to the Mongo service, the migration service component refers to the migration service, and the Serverless service component refers to the online service.

[0122] After receiving the HTTP request, the online service determines it's a knowledge base migration request and queries the Mongo service based on the fourth and fifth indices to obtain the fourth metadata of the outgoing search engine and the fifth metadata of the incoming search engine. The migration service then accesses the outgoing search engine based on the fourth metadata to obtain the first word vector to be migrated out. It then accesses the incoming search engine based on the fifth metadata to migrate the first word vector to the incoming search engine.

[0123] According to the information processing method of the large model provided by the embodiment of the present disclosure, the large model receives an HTTP request based on the Serverless service component, determines that it is a migration request for the knowledge base, and can determine the non-relational database component and the migration service component as the target service component, and based on the fourth index and the fifth index carried by the migration request, determine the search engine to be migrated out and the search engine to be migrated in as the target search engine. The Serverless service component interacts with the target service component and the target search engine, migrates the word vectors in the search engine to be migrated out to the search engine to be migrated in, realizes the migration of the knowledge base, and can realize migration expansion. While reducing resource consumption, it effectively supports large-scale traffic, improves the performance and response speed of the large model, and ensures the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0124] Figure 7 A flowchart of a large model information processing method provided in an embodiment of the present disclosure.

[0125] like Figure 7 As shown, the information processing method of the large model may include:

[0126] S701: The Serverless service component receives an HTTP request.

[0127] S702, determining that the HTTP request is a knowledge base construction request for a large model, and the Serverless service component determines the cluster service component and the non-relational database component as the target service component from the service component set based on the knowledge base construction request.

[0128] S703: The Serverless service component calls the cluster service component, and the cluster service component determines an available search engine from the search engine cluster as the target search engine.

[0129] S704: The Serverless service component interacts with the target search engine and the target service component to build a knowledge base of the large model.

[0130] S705: The Serverless service component receives the HTTP request.

[0131] S706, determine that the HTTP request is a document storage request, and based on the document storage request, determine the extraction service component, cloud storage component, word vector storage component, non-relational database component and word vector generation component from the service component set as the target service component.

[0132] S707: The Serverless service component determines a target search engine from the search engine cluster based on the second index carried in the document storage request.

[0133] S708: The Serverless service component interacts with the target search engine and the target service component to store the document.

[0134] S709: The Serverless service component receives the HTTP request.

[0135] S710, determining that the HTTP request is a document retrieval request. Based on the document retrieval request, the Serverless service component determines a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from a service component set as target service components.

[0136] S711: Determine at least one target search engine from a search engine cluster based on at least one third index carried in the document retrieval request.

[0137] S712: The Serverless service component interacts with the target search engine and the target service component to retrieve documents.

[0138] S713: The Serverless service component receives the HTTP request.

[0139] S714, determining that the HTTP request is a migration request for the knowledge base, and the Serverless service component determines the non-relational database component and the migration service component from the service component set as the target service components based on the migration request.

[0140] S715: The Serverless service component selects a search engine to be migrated out and a search engine to be migrated into from the search engine cluster as a target search engine based on the fourth index and one or more fifth indexes carried in the migration request.

[0141] S716, the Serverless service component interacts with the target search engine and the target service component to achieve the migration of the knowledge base.

[0142] According to the information processing method of the large model provided by the embodiment of the present disclosure, the large model receives HTTP requests based on the Serverless service component, and can determine the target service component to be called from the service component set of the large model, as well as the target search engine. The Serverless service component interacts with the target service component and the target search engine to process the HTTP request, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the large model, and ensure the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0143] Corresponding to the information processing methods for large models provided in the above-mentioned embodiments, an embodiment of the present disclosure also provides an information processing device for a large model. Since the information processing device for a large model provided in the embodiment of the present disclosure corresponds to the information processing methods for large models provided in the above-mentioned embodiments, the implementation methods of the above-mentioned information processing methods for large models are also applicable to the information processing device for large models provided in the embodiment of the present disclosure, and will not be described in detail in the following embodiments.

[0144] Figure 8 A schematic structural diagram of a large-scale information processing device provided in an embodiment of the present disclosure.

[0145] like Figure 8 As shown, the information processing device 800 of the large model of the embodiment of the present disclosure includes a receiving module 801, a determining module 802 and a processing module 803.

[0146] The receiving module 801 is used for the serverless service component to receive a Hypertext Transfer Protocol (HTTP) request.

[0147] Determination module 802 is used for the Serverless service component to determine the target service component to be called from the service component set of the large model based on the HTTP request, and to determine one or more target search engines from the search engine cluster of the large model.

[0148] The processing module 803 is used for the Serverless service component to interact with the target search engine and the target service component to process the HTTP request.

[0149] In one embodiment of the present disclosure, the determination module 802 is further used to: if the HTTP request is a knowledge base construction request for the large model, the Serverless service component determines, based on the knowledge base construction request, a cluster service component and a non-relational database component from the service component set as the target service component; the Serverless service component calls the cluster service component, and the cluster service component determines an available search engine from the search engine cluster as the target search engine.

[0150] In one embodiment of the present disclosure, the processing module 803 is further used for: the cluster service component obtains first metadata of the target search engine and feeds the first metadata back to the Serverless service component; the Serverless service component stores the first metadata in the non-relational database component.

[0151] In one embodiment of the present disclosure, the processing module 803 is further configured to: enable the cluster service component to generate a first index for the target search engine, and determine the first index as meta information of the target search engine.

[0152] In one embodiment of the present disclosure, the determination module 802 is further used to: if the HTTP request is a document storage request, based on the document storage request, determine an extraction service component, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; the Serverless service component determines the target search engine from the search engine cluster based on the second index carried by the document storage request.

[0153] In one embodiment of the present disclosure, the processing module 803 is further used for: the Serverless service component sends the original document to the extraction service component, the extraction service component parses the original document to obtain a parsed document, and feeds the parsed document back to the Serverless service component; the Serverless service component calls the word vector generation component, the word vector generation component converts the string in the parsed document to obtain a first word vector, and feeds the first word vector back to the Serverless service component; the Serverless service component queries the non-relational database component according to the second index to obtain the second metadata of the target search engine; the Serverless service component sends the word vector to the word vector storage component based on the second metadata, and the word vector storage component stores the word vector in the target search engine.

[0154] In one embodiment of the present disclosure, the determination module 802 is further used to: if the HTTP request is a document retrieval request, the Serverless service component determines, based on the document retrieval request, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; and determines, based on at least one third index carried by the document retrieval request, at least one target search engine from the search engine cluster.

[0155] In one embodiment of the present disclosure, the processing module 803 is further used for: the Serverless service component calls the word vector generation component, and the word vector generation component converts the search term in the document retrieval request to obtain a search term vector, and feeds back the search term vector to the Serverless service component; the Serverless service component queries the non-relational database component according to the third index to obtain the third metadata of the target search engine; the Serverless service component searches the knowledge base associated with the target search engine based on the third metadata and the search term vector to obtain the query result, and feeds it back to the Serverless service component; the Serverless service component accesses the cloud storage component based on the query result, obtains the paragraph matching the search term and feeds it back to the client.

[0156] In one embodiment of the present disclosure, the determination module 802 is further configured to: if the HTTP request is a knowledge base migration request, the Serverless service component determines, based on the migration request, a non-relational database component and a migration service component from the service component set as the target service component; and the Serverless service component determines, based on the fourth index and one or more fifth indexes carried by the migration request, a search engine to be migrated out and a search engine to be migrated into from the search engine cluster as the target search engine.

[0157] In one embodiment of the present disclosure, the processing module 803 is further configured to: the Serverless service component queries the non-relational database component based on the fourth index to obtain fourth metadata of the search engine to be migrated out, and queries the non-relational database component based on the fifth index to obtain fifth metadata of the search engine to be migrated in, and feeds the fourth metadata and the fifth metadata back to the migration service component; the migration service component accesses the search engine to be migrated out based on the fourth metadata to obtain the first word vector to be migrated out of the search engine to be migrated out; the migration service component accesses the search engine to be migrated in based on the fifth metadata, and migrates the first word vector to be migrated out to the search engine to be migrated in.

[0158] In one embodiment of the present disclosure, the processing module 803 is further used to: if there are multiple search engines to be migrated, the migration service component will split the first word vector to be migrated out to obtain split first word vectors, and migrate the split first word vectors to the corresponding search engines to be migrated.

[0159] In one embodiment of the present disclosure, the processing module 803 is further used to: expand the service components in the service component set based on the elastic mechanism of serverless deployment; or access the cluster service component in the service component set through the Serverless service component, and expand the search engine cluster by registering a new search engine in the cluster service component.

[0160] According to the information processing device of the large model provided by the embodiment of the present disclosure, the large model receives HTTP requests based on the Serverless service component, and can determine the target service component to be called from the service component set of the large model, as well as the target search engine. The Serverless service component interacts with the target service component and the target search engine to process the HTTP request, which can effectively support large-scale traffic while reducing resource consumption, improve the performance and response speed of the large model, and ensure the availability of the large model. Serverless deployment of service components can simplify the development and deployment of the knowledge base and provide a basis for processing large-scale traffic.

[0161] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0162] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0163] Figure 9 A schematic block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0164] like Figure 9 As shown, the device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to computer programs / instructions stored in a read-only memory (ROM) 902 or computer programs / instructions loaded from a storage unit 906 into a random access memory (RAM) 903. Various programs and data required for the operation of the device 900 can also be stored in the RAM 903. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0165] Various components in the device 900 are connected to the I / O interface 905, including an input unit 906 such as a keyboard, a mouse, etc.; an output unit 907 such as various types of displays, speakers, etc.; a storage unit 908 such as a magnetic disk, an optical disk, etc.; and a communication unit 909 such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0166] The computing unit 901 can be a variety of general and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 901 performs the various methods and processes described above, such as the information processing method of the large model. For example, in some embodiments, the information processing method of the large model can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 906. In some embodiments, part or all of the computer program / instructions can be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program / instructions are loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the information processing method of the large model described above can be performed. Alternatively, in other embodiments, the computing unit 901 can be configured to perform the information processing method of the large model by any other appropriate means (e.g., by means of firmware).

[0167] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs / instructions that are executable and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0168] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0169] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0170] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0171] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0172] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. The client-server relationship arises through computer programs / instructions running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0173] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0174] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for processing information of a large model, wherein: The method comprises: The serverless service component receives the Hypertext Transfer Protocol (HTTP) request. The Serverless service component determines, based on the HTTP request, a target service component to be called from a set of service components in the large model, and determines one or more target search engines from a search engine cluster in the large model; The Serverless service component interacts with the target search engine and the target service component to process the HTTP request.

2. The method according to claim 1, wherein The Serverless service component determines, based on the HTTP request, a target service component to be called from a set of service components in the large model, and determines one or more target search engines from a search engine cluster in the large model, including: If the HTTP request is a request to build a knowledge base of the large model, the Serverless service component determines, based on the knowledge base construction request, a cluster service component and a non-relational database component from the service component set as the target service component; The Serverless service component calls the cluster service component, and the cluster service component determines an available search engine from the search engine cluster as the target search engine.

3. The method according to claim 2, wherein: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request, including: The cluster service component obtains first metadata of the target search engine and feeds the first metadata back to the Serverless service component; The Serverless service component stores the first metadata in the non-relational database component.

4. The method according to claim 2, wherein: The method further comprises: The cluster service component generates a first index for the target search engine, and determines the first index as meta information of the target search engine.

5. The method according to claim 1, wherein The Serverless service component determines, based on the HTTP request, a target service component to be called from a set of service components in the large model, and determines one or more target search engines from a search engine cluster in the large model, including: If the HTTP request is a document storage request, based on the document storage request, determining an extraction service component, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; The Serverless service component determines the target search engine from the search engine cluster based on the second index carried in the document storage request.

6. The method according to claim 5, wherein: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request, including: The Serverless service component sends the original document to the extraction service component, which parses the original document to obtain a parsed document and feeds the parsed document back to the Serverless service component. The Serverless service component calls the word vector generation component, which converts the string in the parsed document to obtain a first word vector and feeds the first word vector back to the Serverless service component. The Serverless service component queries the non-relational database component according to the second index to obtain second metadata of the target search engine; The Serverless service component sends the word vector to the word vector storage component based on the second metadata, and the word vector storage component stores the word vector in the target search engine.

7. The method according to claim 1, wherein The Serverless service component determines, based on the HTTP request, a target service component to be called from a set of service components in the large model, and determines one or more target search engines from a search engine cluster in the large model, including: If the HTTP request is a document retrieval request, the Serverless service component determines, based on the document retrieval request, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; At least one target search engine is determined from the search engine cluster based on at least one third index carried in the document retrieval request.

8. The method according to claim 7, wherein: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request, including: The Serverless service component calls the word vector generation component, which converts the search term in the document search request to obtain a search word vector, and feeds the search word vector back to the Serverless service component. The Serverless service component queries the non-relational database component according to the third index to obtain the third metadata of the target search engine; The Serverless service component searches the knowledge base associated with the target search engine based on the third metadata and the search term vector to obtain a query result, and feeds the result back to the Serverless service component. The Serverless service component accesses the cloud storage component based on the query result, obtains the paragraphs matching the search term and feeds them back to the client.

9. The method according to claim 1, wherein The Serverless service component determines, based on the HTTP request, a target service component to be called from a set of service components in the large model, and determines one or more target search engines from a search engine cluster in the large model, including: If the HTTP request is a knowledge base migration request, the Serverless service component determines, based on the migration request, a non-relational database component and a migration service component from the service component set as the target service component; The Serverless service component determines a search engine to be migrated out and a search engine to be migrated in from the search engine cluster as the target search engine based on the fourth index and one or more fifth indexes carried in the migration request.

10. The method according to claim 9, wherein: The Serverless service component interacts with the target search engine and the target service component to process the HTTP request, including: The serverless service component queries the non-relational database component based on the fourth index to obtain fourth metadata of the search engine to be migrated out, and queries the non-relational database component based on the fifth index to obtain fifth metadata of the search engine to be migrated in, and feeds the fourth metadata and the fifth metadata back to the migration service component. The migration service component accesses the search engine to be migrated based on the fourth metadata, and obtains the first word vector to be migrated in the search engine to be migrated; The migration service component accesses the search engine to be migrated based on the fifth metadata, and migrates the first word vector to be migrated out to the search engine to be migrated in.

11. The method according to claim 10, wherein: The method further comprises: If there are multiple search engines to be migrated into, the migration service component splits the first word vector to be migrated out to obtain split first word vectors, and migrates the split first word vectors to the corresponding search engines to be migrated into.

12. The method according to claim 1, wherein The method further comprises: Based on the elastic mechanism of serverless deployment, the service components in the service component set are expanded; or, The cluster service component in the service component set is accessed through the Serverless service component, and the search engine cluster is expanded by registering a new search engine in the cluster service component.

13. A large-scale information processing device, wherein: The device comprises: The receiving module is used by the serverless service component to receive Hypertext Transfer Protocol (HTTP) requests. a determination module, configured for the Serverless service component to determine, based on the HTTP request, a target service component to be called from a set of service components in a large model, and to determine one or more target search engines from a search engine cluster in the large model; A processing module is used for the Serverless service component to interact with the target search engine and the target service component and process the HTTP request.

14. The device according to claim 13, wherein The determining module is further configured to: If the HTTP request is a request to build a knowledge base of the large model, the Serverless service component determines, based on the knowledge base construction request, a cluster service component and a non-relational database component from the service component set as the target service component; The Serverless service component calls the cluster service component, and the cluster service component determines an available search engine from the search engine cluster as the target search engine.

15. The device according to claim 14, wherein The processing module is further configured to: The cluster service component obtains first metadata of the target search engine and feeds the first metadata back to the Serverless service component; The Serverless service component stores the first metadata in the non-relational database component.

16. The device according to claim 14, wherein The processing module is further configured to: The cluster service component generates a first index for the target search engine, and determines the first index as meta information of the target search engine.

17. The device according to claim 13, wherein The determining module is further configured to: If the HTTP request is a document storage request, based on the document storage request, determining an extraction service component, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; The Serverless service component determines the target search engine from the search engine cluster based on the second index carried in the document storage request.

18. The device according to claim 17, wherein The processing module is further configured to: The Serverless service component sends the original document to the extraction service component, which parses the original document to obtain a parsed document and feeds the parsed document back to the Serverless service component. The Serverless service component calls the word vector generation component, which converts the string in the parsed document to obtain a first word vector and feeds the first word vector back to the Serverless service component. The Serverless service component queries the non-relational database component according to the second index to obtain second metadata of the target search engine; The Serverless service component sends the word vector to the word vector storage component based on the second metadata, and the word vector storage component stores the word vector in the target search engine.

19. The device according to claim 13, wherein The determining module is further configured to: If the HTTP request is a document retrieval request, the Serverless service component determines, based on the document retrieval request, a cloud storage component, a word vector storage component, a non-relational database component, and a word vector generation component from the service component set as the target service component; At least one target search engine is determined from the search engine cluster based on at least one third index carried in the document retrieval request.

20. The device according to claim 19, wherein The processing module is further configured to: The Serverless service component calls the word vector generation component, which converts the search term in the document search request to obtain a search word vector, and feeds the search word vector back to the Serverless service component. The Serverless service component queries the non-relational database component according to the third index to obtain the third metadata of the target search engine; The Serverless service component searches the knowledge base associated with the target search engine based on the third metadata and the search term vector to obtain a query result, and feeds the result back to the Serverless service component. The Serverless service component accesses the cloud storage component based on the query result, obtains the paragraphs matching the search term and feeds them back to the client.

21. The apparatus according to claim 13, wherein The determining module is further configured to: If the HTTP request is a knowledge base migration request, the Serverless service component determines, based on the migration request, a non-relational database component and a migration service component from the service component set as the target service component; The Serverless service component determines a search engine to be migrated out and a search engine to be migrated in from the search engine cluster as the target search engine based on the fourth index and one or more fifth indexes carried in the migration request.

22. The device according to claim 21, wherein The processing module is further configured to: The serverless service component queries the non-relational database component based on the fourth index to obtain fourth metadata of the search engine to be migrated out, and queries the non-relational database component based on the fifth index to obtain fifth metadata of the search engine to be migrated in, and feeds the fourth metadata and the fifth metadata back to the migration service component. The migration service component accesses the search engine to be migrated based on the fourth metadata, and obtains the first word vector to be migrated in the search engine to be migrated; The migration service component accesses the search engine to be migrated based on the fifth metadata, and migrates the first word vector to be migrated out to the search engine to be migrated in.

23. The device according to claim 22, wherein The processing module is further configured to: If there are multiple search engines to be migrated into, the migration service component splits the first word vector to be migrated out to obtain split first word vectors, and migrates the split first word vectors to the corresponding search engines to be migrated into.

24. The apparatus according to claim 13, wherein The processing module is further configured to: Based on the elastic mechanism of serverless deployment, the service components in the service component set are expanded; or, The cluster service component in the service component set is accessed through the Serverless service component, and the search engine cluster is expanded by registering a new search engine in the cluster service component.

25. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 12.

26. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-12.

27. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 12 is implemented.

Images