End-cloud collaborative information processing method and device

Abstract

The invention provides an end-cloud collaborative information processing method and device. The end-cloud collaborative information processing method comprises the following steps: acquiring first input information corresponding to user operation; performing first processing on the first input information by utilizing a first model to obtain a first processing result; wherein the first processing at least comprises the construction of index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information; performing second processing on the first processing result by using a second model to obtain a second processing result; wherein the first model is deployed on the end side, and the second model is deployed on the cloud side.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to an edge-cloud collaborative information processing method and apparatus. Background Technology

[0002] In current software development, software for editable text, such as AI programming assistants, is widely used for tasks like automatic code completion, comment generation, and refactoring suggestions. These tools are primarily based on large language models deployed in the cloud. By uploading contextual code from the user's editor, the cloud model performs inference and completion. However, current cloud-based model completion capabilities have limitations. The cloud model's completion function relies on limited contextual information in the editor window; it can only infer from these local code snippets and cannot fully understand the structure of the entire project. This means its completion suggestions may not be accurate or comprehensive enough for some complex or large projects. Furthermore, uploading code to the cloud can pose various security risks, including data leaks, unauthorized access, and malicious code injection. Summary of the Invention

[0003] In view of the above-mentioned technical problems existing in the prior art, this application provides an edge-cloud collaborative information processing method and apparatus.

[0004] This application provides an edge-cloud collaborative information processing method, including: Obtain the first input information corresponding to the user's operation; Using a first model, the first input information is processed to obtain a first processing result; wherein, the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information; Using the second model, the first processing result is processed a second time to obtain the second processing result; The first model is deployed on the device side, and the second model is deployed in the cloud.

[0005] In some embodiments, the first processing of the first input information to obtain a first processing result specifically includes: Based on the first input information and the target database deployed on the edge, determine the index information related to the first input information; The first input information and the index information are concatenated to obtain the first processing result.

[0006] In some embodiments, the second processing of the first processing result to obtain a second processing result specifically includes: The second processing on the first processing result includes completion processing to generate a second processing result containing the completed content; wherein, the second processing result includes tagging information, the tagging information being used to indicate the reference information of the completed content.

[0007] In some embodiments, the method further includes: Store the mapping relationship between the first input information and the second processing result; Obtain the second input information corresponding to the first input information; The second processing result is determined based on the correspondence between the second input information and the first input information.

[0008] In some embodiments, before performing a second processing on the first processing result to obtain a second processing result, the method further includes: The structured information corresponding to the index information in the first processing result is uploaded to the second model, so that the second model generates the second processing result based on the structured information of the index information.

[0009] In some embodiments, obtaining the first input information corresponding to the user operation specifically includes: Obtain the user's operation instructions; The first input information is determined to include at least the context content related to the operation instruction.

[0010] In some embodiments, before performing a second processing on the first processing result to obtain a second processing result, the method further includes: The first processing result is subjected to privacy processing to obtain a privacy-processed first processing result, and the second model generates the second processing result based on the privacy-processed first processing result.

[0011] In some embodiments, the target database deployed on the edge is constructed based at least on a semantic vectorization model.

[0012] In some embodiments, the first process utilizes at least a cross-file semantic relationship graph to construct the index information.

[0013] This application also provides an edge-cloud collaborative information processing device, including: The acquisition module is used to acquire the first input information corresponding to the user operation. A processing module is configured to perform a first processing on the first input information using a first model to obtain a first processing result; wherein the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information; and The first processing result is processed using the second model to obtain the second processing result; wherein the first model is deployed on the edge and the second model is deployed in the cloud. Attached Figure Description

[0014] In drawings that are not necessarily drawn to scale, the same reference numerals may describe similar parts in different views. The drawings generally illustrate various embodiments by way of example rather than limitation and are used, together with the description and claims, to illustrate the disclosed embodiments. Where appropriate, the same reference numerals are used in all drawings to refer to the same or similar parts. Such embodiments are illustrative and not intended to be exhaustive or exclusive embodiments of the apparatus or method.

[0015] Figure 1 This is a first flowchart of the information processing method for end-to-cloud collaboration according to an embodiment of this application; Figure 2 This is a second flowchart of the information processing method for end-to-cloud collaboration according to an embodiment of this application; Figure 3 This is a third flowchart of the information processing method for end-to-cloud collaboration according to an embodiment of this application; Figure 4 This is the fourth flowchart of the end-to-cloud collaborative information processing method according to an embodiment of this application; Figure 5 This is the fifth flowchart of the end-to-cloud collaborative information processing method according to an embodiment of this application; Figure 6 This is a structural block diagram of the end-to-cloud collaborative information processing device according to an embodiment of this application. Detailed Implementation

[0016] It should be understood that various modifications can be made to the embodiments described herein. Therefore, the above description should not be considered as limiting, but merely as an example of embodiments. Other modifications within the scope and spirit of this application will be apparent to those skilled in the art.

[0017] The accompanying drawings, which are included in and form part of this specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0018] These and other features of this application will become apparent from the following description of preferred forms of embodiments given as non-limiting examples, with reference to the accompanying drawings.

[0019] It should also be understood that although this application has been described with reference to some specific examples, those skilled in the art can certainly implement many other equivalent forms of this application.

[0020] The above and other aspects, features and advantages of this application will become more apparent when taken in conjunction with the accompanying drawings and in view of the following detailed description.

[0021] Specific embodiments of this application are described thereafter with reference to the accompanying drawings; however, it should be understood that the embodiments are merely examples of this application, which may be implemented in various ways. Well-known and / or repeated functions and structures are not described in detail to avoid unnecessary or redundant details that could obscure the application. Therefore, the specific structural and functional details herein are not intended to be limiting, but merely serve as the basis and representative basis for the claims to teach those skilled in the art to use this application in a variety of substantially any suitable detailed structures.

[0022] This specification may use the phrases “in one embodiment,” “in another embodiment,” “in yet another embodiment,” or “in other embodiments,” all of which may refer to one or more of the same or different embodiments according to this application.

[0023] This application provides an edge-cloud collaborative information processing method, which can be applied to edge-cloud collaborative information processing platforms, such as AI programming assistant platforms, to provide developers with more intelligent code generation services. It can also be applied to other text editing platforms, such as document collaboration assistants. The following description uses the application of the edge-cloud collaborative information processing method to an AI programming assistant platform as an example.

[0024] like Figure 1 As shown, the information processing method for edge-cloud collaboration includes steps S101 to S103.

[0025] Step S101: Obtain the first input information corresponding to the user operation.

[0026] The aforementioned first input information can be understood as information related to editable text, specifically including text, code, markup language, or other information, intended to provide a foundation for subsequent processing or analysis. For example, for an AI programming assistant platform, the first input information can be understood as the context information of the code to be inserted, which may include the location of the current code segment, related function or class definitions, variable declarations, existing comments, possible code logic, etc. This information is helpful for generating accurate code completion, repair suggestions, or optimization solutions.

[0027] The first input information mentioned above can be information entered by the user, or information not directly entered by the user, but rather information generated based on user operations.

[0028] For example, after receiving the user's operation to specify the text insertion position, contextual content for that position can be generated as the first input information. In other examples, the user can input contextual content corresponding to the first input information, and the subsequent first model can perform first processing on that contextual content.

[0029] Step S102: Using the first model, perform a first processing on the first input information to obtain a first processing result; wherein, the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information.

[0030] The aforementioned index information can be understood as information determined by the first model deployed on the edge, related to the first input information, and capable of interdepending on the first input information to jointly perform tasks or achieve functions. When both the first input information and the index information are configured as code information, the correlation between them is related to the core interrelationship logic between the codes, such as functional succession, data dependency, and logical complementarity. For example, one piece of code may provide data or conditions for another piece of code to process, or the output of one piece of code may serve as the input of another piece of code. They collaborate closely through function calls, data transfer, or state changes to form a complete operational flow.

[0031] For example, in the scenario of an AI programming assistant platform, the first input information mentioned above can be understood as code segment A and code segment B corresponding to the user's operation. The position where the user specifies to insert text is between code segment A and code segment B. The index information can be at least one code segment C related to code segment A and code segment B determined in the code database. Code segment C can work with code segment A and code segment B to perform a certain task or function. For example, code segment A is used to initiate an asynchronous request to obtain external data, code segment C is used to process the data after receiving it, and code segment B is used to start an asynchronous operation to update the interface or continue other tasks after code segment C has finished processing.

[0032] There is at least a logical relationship between the index information and the first input information. The first input information may include at least two code segments. The code segments contained in the index information may be inserted into the execution order in a coordinated manner with the at least two code segments of the first input information. That is, the code segments of the first input information are executed first, and then the code segments of the index information and other code segments contained in the first input information are executed in sequence.

[0033] The aforementioned enhanced context can be understood as supplementary information or explanations dynamically generated through the first processing by analyzing and understanding the context of the current content or task contained in the first input information. Enhanced context is not limited to code; it can also include the generation of text, images, audio, and other content. The purpose of enhanced context is to supplement, expand, or improve the content; it is more guiding, explicit, or semantically accurate, and can guide the model to generate outputs that better meet expectations. In the context of an AI programming assistant platform, enhanced context can be understood as code that enhances semantic understanding.

[0034] After obtaining the first processing result, the first model deployed on the edge can send the key information contained in the first processing result to the second model deployed in the cloud. For example, it can upload necessary structured information (such as function signatures, call chains, etc.) to the cloud to avoid redundant transmission and to avoid overloading of the second model deployed in the cloud, thereby improving response efficiency and reducing bandwidth consumption.

[0035] Step S103: Using the second model, perform a second processing on the first processing result to obtain a second processing result. The first model is deployed on the device side, and the second model is deployed in the cloud.

[0036] The first and second processes described above can be understood as different sequential processing of information according to logic. The first and second processes differ not only in their execution endpoints. Specifically, the first process utilizes the computing power of the edge device, while the second process utilizes the computing power of the cloud, which is capable of performing complex model reasoning, handling more complex contextual relationships, understanding implicit semantics, and generating more coherent and in-depth results.

[0037] The first processing result can be understood as the preliminary processing of the first input information, such as providing structured and simplified processing results. The second processing result can be understood as the result obtained by the second model using cloud computing power to further process the first processing result, such as completing and generating the first processing result. That is, by using the powerful computing resources and training data in the cloud, the first processing result generated on the edge is analyzed and reasoned in depth to output a more coherent and logical second processing result.

[0038] After generating the second processing result, the second model will feed the second processing result back to the edge-cloud collaborative information processing platform, so that the second processing result can be presented to the user through the edge-cloud collaborative information processing platform.

[0039] This application utilizes a first model deployed on the edge to perform first processing on first input information to obtain a first processing result. The first processing includes index information, and the first processing result is an enhanced context obtained based on the index information. A second model deployed in the cloud can perform second processing on the first processing result to obtain a second processing result. This achieves collaborative processing between the edge and the cloud without relying entirely on a large cloud model to directly process the original first input information. Instead, the first model on the edge generates the enhanced context first, and then the second model in the cloud performs the second processing. This collaborative mechanism can effectively improve the understanding and processing accuracy of the first input information, while reducing the computational pressure on the cloud, achieving a balance between performance and energy consumption, and improving overall processing efficiency.

[0040] In some embodiments, such as Figure 2 As shown, step S102, which involves performing a first processing on the first input information to obtain a first processing result, specifically includes steps S201 to S202.

[0041] Step S201: Determine index information related to the first input information based on the first input information and the target database deployed on the edge.

[0042] Step S202: Concatenate the first input information and the index information to obtain the first processing result.

[0043] In this way, index information can be determined in the target database deployed on the edge, and by concatenating the first input and the index information, a more accurate enhanced context, i.e., the first processing result, can be obtained. In particular, it can ensure that the first model can quickly and accurately obtain index information that is highly relevant to the first input information. This method of generating enhanced context not only improves processing efficiency, but also enables precise positioning with the help of index information, resulting in a more accurate and targeted first processing result.

[0044] The aforementioned target database can be stored on the edge device, and the first module can use the local target database to determine index information. Specifically, the target database can be a local vector database, which provides fast retrieval and similarity search for high-dimensional vectors (such as embedding vectors or feature vectors) through efficient algorithms and data structures.

[0045] For the AI ​​programming assistant platform scenario, the aforementioned target database stores vectorized representations of each code segment converted through machine learning models or other algorithms. By vectorizing the code segments and storing them in the local target database, similarity searches and queries can be performed quickly to determine the index information most similar to the first input information.

[0046] The target database mentioned above can support incremental update functionality. That is, when the project files are changed, the system will only re-index the affected parts instead of reprocessing the entire database, thus updating only the relevant parts. This can greatly improve update efficiency and reduce unnecessary computation and resource consumption.

[0047] The above splicing process can be understood as the processing function of the first process, which can splice based on the logical or semantic relationship between the first input information and the index information, with the aim of making the first processing result obtained after splicing complete and logical.

[0048] When both the initial input information and the index information are configured as code information, concatenating the initial input information and the index information can be understood as merging them logically or semantically, so that the model can treat the concatenated result as a whole when understanding or executing it. It's important to understand that this concatenation is not simply linking the initial input information and the index information together, but rather based on vectorized representations or semantic features. For example, the initial input information and the index information are first converted into vectors by an encoder, and then these two vectors are concatenated along a specific dimension to form a new vector representation, i.e., the first processing result. In this way, the first processing result simultaneously contains the semantic information of both the initial input information and the index information, and can be used for tasks such as similarity search, code completion, or function retrieval.

[0049] In some embodiments, such as Figure 3 As shown, step S103, which involves performing a second processing on the first processing result to obtain a second processing result, specifically includes step S301. Step S301: Performing the second processing on the first processing result includes completion processing to generate a second processing result containing completed content; wherein, the second processing result includes tagging information, the tagging information being used to indicate the reference information of the completed content.

[0050] In this way, the second processing result can be obtained by completing the first processing result using a second model in the cloud. This fully utilizes the powerful computing resources and more advanced model capabilities of the cloud, thereby improving the accuracy and completeness of the processing result. Furthermore, the completed second processing result contains identification information that indicates the reference information of the completed content, increasing the credibility of the second processing result and ensuring the traceability of the completed content.

[0051] The above completion process can be understood as supplementing, correcting, or expanding the first processing result through a second model in the cloud, making it more complete and accurate.

[0052] The reference information included in the above-mentioned marking information can be understood as the file, line number, or code block where the completed content is located, in order to help developers trace back to the specific location.

[0053] In some embodiments, such as Figure 4 As shown, the method further includes steps S401 to S403.

[0054] Step S401: Store the mapping relationship between the first input information and the second processing result.

[0055] Step S402: Obtain the second input information corresponding to the first input information.

[0056] Step S403: Determine the output of the second processing result based on the correspondence between the second input information and the first input information.

[0057] In this way, after obtaining the second input information corresponding to the first input information, there is no need to repeat the calculation process. Based on the correspondence between the second input information and the first input information, the second processing result corresponding to the first input information can be output, realizing the reuse of processing results and significantly reducing bandwidth consumption and processing latency. The second input information corresponding to the first input information can be understood as having a similar relationship between the first input information and the second input information, specifically in terms of structure, function, or logic.

[0058] For example, both the first and second input information are code information. They are syntactically very similar, using the same language constructs, functions, or statement forms. Alternatively, they may have similar functions or tasks, or similar execution flows and algorithm steps. The first model can directly call the second processing result for this second input information, without needing to repeatedly use both the first and second models for calculations on the second input information; it can directly use the second processing result corresponding to the first input information for output.

[0059] The aforementioned stored mapping relationship can be understood as establishing a correspondence between the first input information and the second processing result, so that when a second input information similar to the first input information is received subsequently, the previous output result can be called quickly and accurately, that is, the second processing result can be called, thereby improving the output efficiency of the output result.

[0060] In some embodiments, before performing a second processing on the first processing result in step S103 to obtain a second processing result, the method further includes: uploading the structured information corresponding to the index information in the first processing result to the second model, so that the second model generates the second processing result based on the structured information of the index information.

[0061] In this way, by uploading only the necessary structured information without uploading the entire first processing result to the cloud, the amount of data transmission and latency can be significantly reduced, the processing efficiency of the second model in the cloud can be improved, and redundant data can be processed repeatedly.

[0062] The structured information mentioned above includes at least one of the following: application programming interface information and data type constraint information.

[0063] In some embodiments, such as Figure 5 As shown, step S101, which involves obtaining the first input information corresponding to the user operation, specifically includes steps S501 to S502.

[0064] Step S501: Obtain the user's operation instructions.

[0065] Step S502: Determine that the first input information includes at least the context content related to the operation instruction.

[0066] In this way, after obtaining the user's operation instructions, the context of the first input information can be determined. Ensuring that the context is clear can reduce misunderstandings and errors, and improve the accuracy and response speed of the interaction.

[0067] The user's operation instructions can be understood as the user specifying the location to insert text, such as the cursor insertion position selected by the user on the display interface of the cloud-edge collaborative information processing platform. After obtaining the cursor insertion operation instructions, the context content of the cursor position can be determined. The first model can perform first processing on the context content to obtain a first processing result including enhanced context.

[0068] The aforementioned contextual content can be used to automatically remove redundant fragments unrelated to the completion position based on a mechanism that combines rules and learning methods, retaining only key information, thereby optimizing the processing efficiency and accuracy of the first model.

[0069] The aforementioned key information includes at least one of the following: call chain, function signature, key comments, and interface declaration. Key information helps enhance code readability and maintainability, ensures that developers can quickly understand and apply the completed content, and improve work efficiency.

[0070] In some embodiments, such as Figure 3 As shown, before performing a second processing on the first processing result to obtain a second processing result, the method further includes step S601. Step S601: Perform privacy processing on the first processing result to obtain a privacy-processed first processing result, and then enable the second model to generate the second processing result based on the privacy-processed first processing result.

[0071] In this way, by performing privacy processing on the initial processing results before uploading them to the cloud, user data security and privacy can be effectively protected, and the risk of sensitive information leakage can be reduced.

[0072] Before uploading the first processing result output by the first model to the second model in the cloud, a confirmation upload message can be generated. This confirmation upload message can be understood as requiring the user to select whether the first processing result can be uploaded to the second model. After the user confirms that the first processing result can be uploaded, the first processing result is then uploaded to the second model. This authorization mechanism ensures that the user has the right to choose whether to upload the first processing result, preventing unauthorized data leakage or misuse.

[0073] The aforementioned privacy measures can specifically involve classifying code based on sensitivity, parsing core algorithms and key-type fragments only on the client side, and performing processes such as variable name anonymization, sensitive string replacement, and comment filtering on the uploaded portion to ensure that the content uploaded to the cloud is minimized and has no risk of leakage. Privacy measures can also employ TLS encryption and optional two-way authentication to ensure secure transmission.

[0074] The aforementioned index information can be uploaded along with the first processing result to the second model deployed in the cloud. This allows the second model to incorporate the index information when generating the second processing result, ensuring the accuracy of the result. At this point, privacy processing can be applied to the index information, and the privacy-processed index information can be uploaded to the second model deployed in the cloud.

[0075] After the second processing result is fed back to the first model on the edge, privacy processing can be performed locally on the edge to place the control of data privacy protection on the edge, thereby improving data security and privacy.

[0076] In some embodiments, the target database deployed on the edge is constructed based at least on a semantic vectorization model.

[0077] Thus, the design and storage structure of the target database utilizes a semantic vectorization model to encode or represent data in the form of vectors, making the data in the target database not just raw characters or numbers, but vectors with deep semantic meaning.

[0078] In some embodiments, the first process utilizes at least a cross-file semantic relationship graph to construct the index information.

[0079] In this way, the semantic relationship graph of cross-file components on the local side can quickly retrieve index information during indexing, which not only improves the relevance of queries, but also enhances the processing capability in complex data environments and significantly improves the accuracy of global completion.

[0080] The above cross-file semantic relationship graph is constructed using at least one of the following analysis methods: abstract syntax tree analysis, dependency graph analysis, type inference analysis, and call chain analysis.

[0081] This application also provides an edge-cloud collaborative information processing device 100. For example... Figure 6 As shown, the edge-cloud collaborative information processing device 100 includes an acquisition module 101 and a processing module 102. The acquisition module 101 is used to acquire first input information corresponding to a user operation. The processing module 102 is used to perform a first processing on the first input information using a first model to obtain a first processing result; wherein the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information; and to perform a second processing on the first processing result using a second model to obtain a second processing result; wherein the first model is deployed on the edge, and the second model is deployed in the cloud.

[0082] This application utilizes a first model deployed on the edge to perform first processing on first input information to obtain a first processing result. The first processing includes index information, and the first processing result is an enhanced context obtained based on the index information. A second model deployed in the cloud can perform second processing on the first processing result to obtain a second processing result. This achieves collaborative processing between the edge and the cloud without relying entirely on a large cloud model to directly process the original first input information. Instead, the first model on the edge generates the enhanced context first, and then the second model in the cloud performs the second processing. This collaborative mechanism can effectively improve the understanding and processing accuracy of the first input information, while reducing the computational pressure on the cloud, achieving a balance between performance and energy consumption, and improving overall processing efficiency.

[0083] In some embodiments, the processing module 102 is further configured to: determine index information related to the first input information based on the first input information and the target database deployed on the edge; and perform concatenation processing on the first input information and the index information to obtain the first processing result.

[0084] In some embodiments, the processing module 102 is further configured to: perform the second processing, including completion processing, on the first processing result to generate a second processing result containing completed content; wherein the second processing result contains tagging information, the tagging information being used to indicate reference information of the completed content.

[0085] In some embodiments, the method further includes a storage module. The storage module is used to store the mapping relationship between the first input information and the second processing result. The acquisition module 101 is further used to acquire second input information corresponding to the first input information. The processing module 102 is further used to determine the output of the second processing result based on the correspondence between the second input information and the first input information.

[0086] In some embodiments, the processing module 102 is further configured to: upload the structured information corresponding to the index information in the first processing result to the second model before performing a second processing on the first processing result to obtain a second processing result, so that the second model generates the second processing result based on the structured information of the index information.

[0087] In some embodiments, the acquisition module 101 is further configured to: acquire the user's operation instructions; and determine that the first input information includes at least contextual content related to the operation instructions.

[0088] In some embodiments, the processing module 102 is further configured to: perform privacy processing on the first processing result before performing a second processing on the first processing result to obtain a privacy-processed first processing result, and then enable the second model to generate the second processing result based on the privacy-processed first processing result.

[0089] In some embodiments, the target database deployed on the edge is constructed based at least on a semantic vectorization model.

[0090] In some embodiments, the first process utilizes at least a cross-file semantic relationship graph to construct the index information.

[0091] Note that the various units in the embodiments of this application can be implemented as computer-executable instructions stored in memory, which, when executed by a processor, can perform corresponding steps; they can also be implemented as hardware with corresponding logical computing capabilities; or they can be implemented as a combination of software and hardware (firmware). In some embodiments, the processor can be implemented as any of an FPGA, ASIC, DSP chip, SOC (System-on-a-Chip), MPU (e.g., but not limited to Cortex), etc. The processor can be communicatively coupled to the memory and configured to execute computer-executable instructions stored therein. The memory can include read-only memory (ROM), flash memory, random access memory (RAM), dynamic random access memory (DRAM) such as synchronous DRAM (SDRAM) or Rambus DRAM, static memory (e.g., flash memory, static random access memory), etc., on which computer-executable instructions are stored in any format. The computer-executable instructions can be accessed by the processor, read from the ROM or any other suitable storage location, and loaded into the RAM for the processor to execute, to implement the wireless communication methods according to the embodiments of this application.

[0092] It should be noted that in the system of this application, the components are logically divided according to the functions they are to perform. However, this application is not limited to this and can re-divide or combine the components as needed. For example, some components can be combined into a single component, or some components can be further decomposed into more sub-components.

[0093] The various component embodiments of this application can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some or all of the components in the system according to the embodiments of this application. This application can also be implemented as a device or apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such a program implementing this application can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form. Furthermore, this application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means can be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0094] Furthermore, although exemplary embodiments have been described herein, their scope includes any and all embodiments based on this application that have equivalent elements, modifications, omissions, combinations (e.g., schemes involving intersections of various embodiments), adaptations, or alterations. Elements in the claims will be interpreted broadly based on the language used in the claims and are not limited to the examples described in this specification or during the implementation of this application, and such examples will be interpreted as non-exclusive.

[0095] The above description is intended to be illustrative and not restrictive. For example, the above examples (or one or more of them) can be used in combination with each other. Other embodiments can be used by those skilled in the art when reading the above description. Furthermore, in the above detailed description, various features may be grouped together to simplify the application. This should not be construed as an intention that a disclosed feature not claimed is necessary for any claim. Rather, the subject matter of the application may be less than all the features of a particular disclosed embodiment. Thus, the claims are incorporated herein by reference as examples or embodiments, wherein each claim is an independent, separate embodiment, and these embodiments are contemplated as being able to be combined with each other in various combinations or arrangements. The scope of this application should be determined by reference to the appended claims and the full scope of their equivalents.

[0096] The above embodiments are merely exemplary embodiments of this application and are not intended to limit this application. The scope of protection of this application is defined by the claims. Those skilled in the art can make various modifications or equivalent substitutions to this application within its substance and scope of protection, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of this application.

Claims

1. An edge-cloud collaborative information processing method, comprising: Obtain the first input information corresponding to the user's operation; Using a first model, the first input information is processed to obtain a first processing result; wherein, the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information; Using the second model, the first processing result is processed a second time to obtain the second processing result; The first model is deployed on the device side, and the second model is deployed in the cloud.

2. The information processing method for end-to-cloud collaboration according to claim 1, wherein the first processing of the first input information to obtain the first processing result specifically includes: Based on the first input information and the target database deployed on the edge, determine the index information related to the first input information; The first input information and the index information are concatenated to obtain the first processing result.

3. The information processing method for end-to-cloud collaboration according to claim 1, wherein the second processing of the first processing result to obtain the second processing result specifically includes: The second processing on the first processing result includes completion processing to generate a second processing result containing the completed content; wherein, the second processing result includes tagging information, the tagging information being used to indicate the reference information of the completed content.

4. The information processing method for edge-cloud collaboration according to claim 1, the method further includes: Store the mapping relationship between the first input information and the second processing result; Obtain the second input information corresponding to the first input information; The second processing result is determined based on the correspondence between the second input information and the first input information.

5. The information processing method for edge-cloud collaboration according to claim 2, before performing a second processing on the first processing result to obtain a second processing result, the method further includes: The structured information corresponding to the index information in the first processing result is uploaded to the second model, so that the second model generates the second processing result based on the structured information of the index information.

6. The information processing method for end-to-cloud collaboration according to claim 1, wherein obtaining the first input information corresponding to the user operation specifically includes: Obtain the user's operation instructions; The first input information is determined to include at least the context content related to the operation instruction.

7. The information processing method for edge-cloud collaboration according to claim 1, before performing a second processing on the first processing result to obtain a second processing result, the method further includes: The first processing result is subjected to privacy processing to obtain a privacy-processed first processing result, and the second model generates the second processing result based on the privacy-processed first processing result.

8. The edge-cloud collaborative information processing method according to claim 2, wherein the target database deployed on the edge side is constructed based on at least a semantic vectorization model.

9. The information processing method for end-to-cloud collaboration according to claim 1, wherein the first processing at least utilizes a cross-file semantic relationship graph to construct the index information.

10. An edge-cloud collaborative information processing device, comprising: The acquisition module is used to acquire the first input information corresponding to the user operation. A processing module is configured to use a first model to perform a first processing on the first input information to obtain a first processing result; wherein the first processing includes at least constructing index information related to the first input information, and the first processing result is an enhanced context obtained at least based on the index information. as well as The first processing result is processed using the second model to obtain the second processing result; wherein the first model is deployed on the edge and the second model is deployed in the cloud.

Images