Interaction method and device, electronic equipment, storage medium and program product

Abstract

The invention provides an interaction method and device, electronic equipment, a storage medium and a program product, and relates to the technical field of artificial intelligence, in particular to the technical fields of deep learning, large models, intelligent questions and answers and the like. The interaction method comprises the steps of displaying first corpus content in received corpus data; in response to an interaction operation of the target object for the first corpus content, updating the first corpus content and a second corpus content having a semantic dependency relationship with the first corpus content in the corpus data to obtain target corpus data; and determining feedback information related to the demand intention of the target object based on the target corpus data.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to the fields of deep learning, large models, and intelligent question answering. Background Technology

[0002] Corpus data can include text content based on natural language expressions. For example, corpus data can be text content related to a specified topic, such as press releases or academic papers. Corpus data can also include structured information such as structured tables and charts to facilitate users' quick understanding of the information. Users can edit the corpus data using electronic devices such as smartphones and computers to obtain information that meets their specific needs. Summary of the Invention

[0003] This disclosure provides an interaction method, apparatus, electronic device, storage medium, and program product.

[0004] According to one aspect of this disclosure, an interaction method is provided, comprising: displaying first corpus content in received corpus data; updating the first corpus content and second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content in response to an interaction operation of a target object on the first corpus content, thereby obtaining target corpus data; and determining feedback information related to the target object's demand intention based on the target corpus data.

[0005] According to another aspect of this disclosure, an interactive device is provided, comprising: a first display module for displaying first corpus content in received corpus data; a target corpus data acquisition module for updating the first corpus content and second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content in response to an interactive operation of a target object on the first corpus content, thereby obtaining target corpus data; and a feedback information determination module for determining feedback information related to the target object's need intention based on the target corpus data.

[0006] According to another aspect of this 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, the instructions being executed by the at least one processor to enable the at least one processor to perform the methods provided in embodiments of this disclosure.

[0007] According to another aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform the methods provided in embodiments of this disclosure.

[0008] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the methods provided in embodiments of this disclosure.

[0009] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0010] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0011] Figure 1 This illustration schematically shows an exemplary system architecture to which interactive methods and apparatus can be applied according to embodiments of the present disclosure;

[0012] Figure 2 A flowchart illustrating an interaction method according to an embodiment of the present disclosure is shown schematically;

[0013] Figure 3 The diagram illustrates an application scenario of the corpus data generation method based on a large model according to an embodiment of the present disclosure.

[0014] Figure 4 This schematically illustrates a flowchart of an interactive operation based on an embodiment of the present disclosure, which utilizes a specified large model to perform a corpus content generation task based on the contextual corpus content.

[0015] Figure 5 A block diagram of an interactive device according to an embodiment of the present disclosure is schematically shown; and

[0016] Figure 6 A schematic block diagram of an example electronic device is shown that can be used to implement the interactive methods of embodiments of the present disclosure. Detailed Implementation

[0017] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0018] In the technical solution disclosed herein, the acquisition, storage, and application of user personal information comply with the provisions of relevant laws and regulations, necessary confidentiality measures have been taken, and there is no violation of public order and good morals.

[0019] The inventors discovered that editing corpora such as news releases, academic papers, and scripts requires relatively complex interactive operations. The annotation of corpora for specific applications such as large-scale language model training necessitates even more complex interactive processes, which can easily lead to errors in the generated corpora, reducing their quality and making them unsuitable for the practical needs of training language models. Furthermore, the interactive methods used for corpora negatively impact the efficiency of corpus generation, failing to meet user requirements.

[0020] Embodiments of this disclosure provide an interaction method, apparatus, electronic device, storage medium, and program product. The interaction method includes: displaying first corpus content in received corpus data; updating the first corpus content and second corpus content in the corpus data that has a semantic dependency on the first corpus content in response to an interaction operation by a target object on the first corpus content, thereby obtaining target corpus data; and determining feedback information related to the target object's desired intent based on the target corpus data.

[0021] According to embodiments of this disclosure, by responding to an interactive operation by a target object on the first corpus content in the corpus data to be edited, the first corpus content and the second corpus content in the corpus content that have a semantic dependency on the first corpus content are updated. This can automatically and conveniently update the content in the corpus data that the target object wants to modify based on the editing intent of the corpus data represented by the interactive operation of the target object, thereby meeting the user's editing and modification needs for the corpus data, reducing the complexity of interactive operations for editing the corpus data, and avoiding omissions in the editing process of the corpus data by automatically updating the content that the target object wants to modify, thereby improving the quality level of the corpus content in the feedback information and improving the efficiency of updating the corpus data.

[0022] Figure 1 The illustration schematically depicts an exemplary system architecture to which interactive methods and apparatus can be applied according to embodiments of the present disclosure.

[0023] like Figure 1 As shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0024] Users can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 via the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (for example only).

[0025] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, laptops, and desktop computers.

[0026] Server 105 can be a server that provides various services, such as a backend management server that supports the content browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0027] It should be noted that the interaction method provided in the embodiments of this disclosure can generally be executed by the first terminal device 101, the second terminal device 102, or the third terminal device 103. Correspondingly, the interaction device provided in the embodiments of this disclosure can also be disposed in the first terminal device 101, the second terminal device 102, or the third terminal device 103.

[0028] Alternatively, the interaction method provided in this embodiment can generally be executed by server 105. Correspondingly, the interaction device provided in this embodiment can generally be located in server 105. The interaction method provided in this embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the interaction device provided in this embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0029] For example, the large model can be deployed on server 105, or it can be deployed on a server or server cluster that is different from server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or server 105.

[0030] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included. For ease of explanation of the interaction methods in the embodiments of this disclosure, a server can be used as the execution entity of the methods provided in the embodiments of this disclosure.

[0031] Figure 2 A flowchart illustrating an interaction method according to an embodiment of this disclosure is shown schematically.

[0032] like Figure 2 As shown, the interaction method includes operations S210~S230.

[0033] In operation S210, the content of the first corpus in the received corpus data is displayed.

[0034] In operation S220, in response to the target object's interactive operation on the first corpus content, the first corpus content and the second corpus content in the corpus data that have a semantic dependency relationship with the first corpus content are updated to obtain the target corpus data.

[0035] In operation S230, feedback information related to the target audience's needs and intentions is determined based on the target corpus data.

[0036] According to embodiments of this disclosure, the corpus data may include any type of text data based on natural language expression, such as news releases to be edited, academic papers, and scripts. The corpus data may include multiple corpus contents, which may be text from multiple paragraphs or sentences. In some embodiments, the corpus content may be text content from different topics in academic papers or work reports. Embodiments of this disclosure do not limit the specific type of corpus content.

[0037] According to embodiments of this disclosure, semantic dependencies between multiple corpus contents can indicate the degree of correlation of semantic attributes such as semantic similarity and semantic logical relationship between the multiple corpus contents. For example, in work reports as corpus data, the logistics reports, sales performance reports, and after-sales performance reports related to "Model A vehicle" in the work reports can have semantic dependencies related to "Model A vehicle".

[0038] In some embodiments, multiple corpus contents from the corpus data can be displayed on the interactive interface of computing devices such as smartphones and computers. Interactive operations on the first corpus content can be any type of operation used to edit the corpus content, such as text modification or formatting updates, on paragraph content, sentence content, etc. For example, an interactive operation can modify a specified sentence in the abstract section of the corpus data to obtain a new abstract section. In response to the interactive operation on the abstract section, the example section and summary section, which have semantic dependencies on the modified sentence in the abstract section, can be modified, and an updated academic paper can be generated as the target corpus data.

[0039] In some embodiments, semantic dependencies between multiple corpus contents can be determined based on the semantic similarity between the topics of multiple corpus contents in the corpus data. Alternatively, semantic dependencies between multiple corpus contents can be constructed based on manual annotation. For another example, a specified large model can be used to process multiple corpus contents in the corpus data to obtain semantic dependencies between them. The embodiments of this disclosure do not limit the specific method for determining semantic dependencies between multiple corpus contents.

[0040] In some embodiments, determining feedback information based on target corpus data may include pushing the target corpus data as feedback information to the target object.

[0041] In some embodiments, determining feedback information related to the target object's need intent based on target corpus data may further include: determining target information pairs based on target corpus content in the target corpus data and operation information of interactive operations related to the target corpus content; and determining feedback information based on the target information pairs.

[0042] According to embodiments of this disclosure, the target information pair may include target corpus content and operation information related to interactive operations on the target corpus content, as well as the mapping relationship between the target corpus content and the operation information. Thus, the target information pair can represent the target object's update intention and modification results for multiple corpus contents in the corpus data. This allows relevant personnel in specific scenarios such as corpus data approval and report content modification to clearly understand the entire modification process, reducing omissions in corpus data modifications in feedback information and improving the operational efficiency of relevant personnel on the target corpus data in specific scenarios.

[0043] In some embodiments, feedback information can be used to train the language model to be trained. The language model can learn negative examples that do not match the needs and intentions of the target object by understanding the target corpus data and interactive operation information in the target information pair of the feedback information, as well as the interactive operation process for the corpus content that serves as a negative example. Based on the target corpus data, the model can perform semantic understanding of the needs and intentions of the target object, thereby improving the model performance of the trained language model and increasing the training efficiency of the language model.

[0044] It should be noted that the interactive operation information can include any type of interactive operation information related to the corpus content, such as information input, deletion, and format modification.

[0045] The language models or large models involved in the embodiments of this disclosure can be built based on large language models. A large language model (LLM) is a deep learning-based artificial intelligence model that can be used to understand input demand information to generate text content that meets the user's needs and intentions.

[0046] In some embodiments, updating the first corpus content and the second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content may include: updating the first corpus content according to an interactive operation to obtain first intermediate content; updating the second corpus content based on the first intermediate content using a specified large model according to the semantic dependency relationship to obtain second intermediate content; and updating the corpus data by using a specified large model to perform semantic understanding on the first intermediate content and the second intermediate content to obtain target corpus data.

[0047] In one example, a work report document serves as the corpus data. The target object can modify the annual sales amount and sales quantity of vehicle model A in the work report document, obtaining the modified sales amount content as the first intermediate content. A specified large model can process the first intermediate content and multiple corpus contents in the work report document based on semantic dependencies. It can also modify the warehousing and logistics costs and related information of sold vehicles in the warehousing and logistics process report content of the work report document to obtain the second intermediate content. Simultaneously, it can modify the total sales amount and total profit data of multiple models throughout the year in the work report document to achieve automated updates. The resulting second intermediate content can also include the modified vehicle warehousing and logistics costs, as well as the total sales amount and total profit data of multiple models throughout the year. Thus, the large model integrates the first and second intermediate contents, along with other corpus content, to obtain the updated work report document as the target corpus data.

[0048] According to embodiments of this disclosure, based on semantic dependencies as prompting information, a specified large model can be used to determine second corpus content that satisfies the semantic dependency condition of the first corpus content from multiple corpus contents. For example, the semantic similarity between the second corpus content and the first corpus content can meet a preset similarity threshold. By using the specified large model to perform semantic understanding on the semantic differences between the first intermediate content and the first corpus content, the target object's intention to perform an interactive operation to update the first corpus content can be determined. Therefore, the second corpus content can be updated based on the intention to update the first corpus content, so that the updated second intermediate content can match the target object's intention. Furthermore, by using the specified large model to perform semantic understanding on the first and second intermediate content, the corpus data's language expression methods and content attributes such as the paragraph sorting method can be updated. This allows multiple corpus contents related to interactive operations in the corpus data to be updated synchronously, and by using the specified large model to fuse multiple intermediate contents with semantic dependencies, the target corpus data maintains logical coherence and semantic integrity, satisfies the target object's intention, and improves the data quality of the target corpus data.

[0049] In some embodiments, semantic dependencies are determined based on the following operation: using a specified large model to perform semantic understanding on multiple corpus contents in the corpus data based on at least one of the first corpus content and the first intermediate content, to obtain semantic dependencies.

[0050] In one embodiment, a designated large model can be used to perform semantic understanding on the first intermediate content and multiple corpus contents. This allows the designated large model to clearly understand the semantic correlation between the updated first intermediate content and multiple corpus contents in the initial corpus data, thus identifying multiple corpus contents with a high semantic correlation to the updated first intermediate content as the second corpus content. Therefore, based on the first intermediate content obtained from the target object's interactive operations and the powerful natural language understanding capabilities of the designated large model, semantic dependencies can be determined more accurately, and the multiple corpus contents that the target object needs to update synchronously can be accurately understood. This avoids errors in determining semantic dependencies based on interactive operations that could lead to a decrease in the quality of the target corpus data. Consequently, the updated second intermediate content can more accurately represent the actual needs and intentions of the target object, improving the matching degree between the target corpus data and the target object's needs and intentions, and enhancing the quality of feedback information.

[0051] In one embodiment, a designated large model can be used to perform semantic understanding on the first intermediate content, the first corpus content, and multiple corpus contents. This allows the designated large model to clearly understand the semantic differences between the updated first intermediate content and the first corpus content, as well as the semantic correlation between the first intermediate content and multiple corpus contents in the initial corpus data. This enables the determined semantic dependencies to clearly represent the semantic correlation between the first corpus content and other corpus contents, and the content attributes of the corpus content that needs to be updated as indicated by the interaction operation performed by the target object. Therefore, based on the semantic dependencies, the corresponding second corpus content can be updated more accurately to avoid omissions or errors in updates, improve the data quality of the target corpus data, and ensure that the feedback information matches the actual interaction intent of the target object.

[0052] In some embodiments, the interaction method may further include: displaying the corpus content topology; and determining a sub-topology from the corpus content topology in response to a selection operation on the corpus content topology.

[0053] According to embodiments of this disclosure, the corpus content topology includes corpus node elements representing corpus content and edge elements representing semantic dependencies between multiple corpora. By displaying multiple corpus node elements and edge elements in the corpus content topology, a structured display method based on the corpus content topology allows the target audience to clearly understand the semantic dependencies between multiple corpus contents.

[0054] In some embodiments, the target object can select multiple corpus node elements and edge elements that meet its requirements from the corpus content topology by performing any type of selection operation, such as box selection or clicking, to form a sub-topology. Based on the multiple corpus node elements and edge elements in the sub-topology, the distribution range of the second corpus that the target object needs to modify synchronously with the first corpus content can be determined in the corpus data.

[0055] According to embodiments of this disclosure, the second corpus content is determined based on semantic dependencies from the corpus content corresponding to each of the multiple candidate corpus node elements in the sub-topology, thereby enabling...

[0056] By selecting the distribution range of the second corpus content that needs to be updated synchronously with the first corpus content, the second corpus content is updated synchronously according to the target object's needs after the first corpus content is updated to obtain the first intermediate content. This enables interactive operations on the first corpus content to accurately meet the target object's needs for updating multiple corpus contents that need to be updated in the corpus data, thereby reducing the frequency of interactive operations for the target object and improving the efficiency of generating feedback information.

[0057] Figure 3The diagram illustrates an application scenario of the large-model-based corpus data generation method according to an embodiment of the present disclosure.

[0058] like Figure 3 As shown, the first interactive interface 300 displays a corpus content topology 310 related to the academic paper corpus data. The corpus content topology 310 may include multiple corpus node elements and edge elements. The first corpus node element may represent the abstract corpus content; the second corpus node element represents the historical background research corpus content for topic A; the third corpus node element represents the introduction of the research content and viewpoints of this paper regarding topic A; the fourth corpus node element represents the detailed discussion of the research content and viewpoints; the fifth node element represents the corpus content comparing and discussing the historical background research corpus content and the detailed discussion of the research content and viewpoints; and the sixth corpus node element represents the corpus content suggesting future research directions for topic A.

[0059] The target object can perform a bounding box operation on the second, third, and fourth corpus node elements in the corpus content topology 310 to obtain sub-topology 311. When the target object performs an interactive operation on the corpus content corresponding to the second node element, based on the semantic dependencies represented by the edge elements in sub-topology 311, it can determine that the corpus content corresponding to the third and fourth corpus node elements can be the second corpus content. By updating the operation information of the interactive operation, the historical background research corpus content of topic A is updated to obtain the updated first intermediate content. A specified large model can be used to perform semantic understanding on the academic paper corpus data based on the updated first intermediate content to update the second corpus content corresponding to the third and fourth corpus node elements, thus obtaining the second intermediate content. By using the specified large model to perform semantic fusion on the first and second intermediate content, the target corpus data is obtained.

[0060] In one embodiment, the corpus data can be dialogue data generated by multiple role-based large models for a preset topic. The dialogue corpus data can include dialogue content from 10 rounds, which can serve as the corpus content within the dialogue corpus data. The target object, based on a sub-topology determined by a selection operation, can represent, for example, the dialogue content from rounds 2 to 4 during the dialogue process. By performing an interaction operation on the dialogue content from round 2 as the first corpus content, the updated dialogue content from round 2 can be determined. A specified large model is then used to process the operation information of the interaction operation and the multi-round dialogue content from the dialogue corpus data to perform a corpus content generation task on the dialogue content from rounds 3 to 4 as the second corpus content, resulting in the updated dialogue content from rounds 3 to 4 as the second intermediate content. Thus, the specified large model can be used to fuse the first and second intermediate content to obtain the target corpus data. Feedback information is determined based on the target corpus data and the interaction operation information related to the target corpus data.

[0061] In some embodiments, updating the first corpus content and the second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content may further include: based on the operation information of the interactive operation, using a specified large model to perform a corpus content generation task according to the context corpus content to obtain target corpus content related to the first corpus content or the second corpus content; and determining the target corpus data according to the target corpus content.

[0062] According to embodiments of this disclosure, the operation information for interactive operations can be used to indicate modification information for modifying the content of the first corpus. For example, the operation information can be a statement added to the content of the first corpus, or it can indicate a font change to the first corpus data. The contextual corpus content can be corpus content in the corpus data that is semantically related to the content of the first corpus or the content of the second corpus.

[0063] For example, regarding the first corpus content, by utilizing a designated large model to perform a corpus content generation task based on the context corpus content, the powerful semantic understanding capabilities of the designated large model can be leveraged to understand the semantic relevance between the context corpus content and the modification information indicated by the operational information. This allows the generated first target corpus content to meet the actual needs of modifying the first corpus content based on the modification information carried by the target object's operational information. Furthermore, it ensures that the first target corpus content maintains logical coherence, semantic relevance, and corpus fluency—requirements for corpus quality—with the context corpus content. Therefore, the first target corpus content maintains logical coherence and semantic fluency with the context corpus content, thereby improving the corpus data quality.

[0064] For example, the first target corpus content, obtained by updating the first corpus content, and other corpus content in the corpus data can be used as contextual and contextual corpus content related to the second corpus content. A designated large model is used to perform a corpus content generation task on the second corpus data by processing the contextual corpus content and operational information, resulting in the updated second target corpus content. This ensures that the updated second target corpus content maintains logical coherence, semantic relevance, and fluency—requirements for corpus quality—with its corresponding contextual corpus content. Therefore, determining the target corpus data based on the first and second target corpus contents can further improve corpus quality, avoiding the simultaneous impact of typos or other interactive errors on the semantic logic and fluency of other second corpus data due to user interaction, reducing reliance on user interaction to update corpus content, and improving user experience.

[0065] In some embodiments, contextual corpus content related to the first or second corpus content can be determined from the corpus data based on semantic dependencies. Alternatively, contextual corpus content related to the first or second corpus content can be determined based on the interaction operations of the target object with multiple corpus contents in the corpus data.

[0066] In one embodiment, the contextual corpus content related to the first corpus content or the second corpus content can be all corpus content in the corpus data except for the content related to the first corpus content or the second corpus content.

[0067] Figure 4 The illustration shows a flowchart of an interactive operation based on an embodiment of the present disclosure, in which a specified large model is used to perform a corpus content generation task based on the contextual corpus content.

[0068] like Figure 4 As shown, based on the interactive operation information, the specified large model is used to perform the corpus content generation task according to the context corpus content, including operations S401~S403.

[0069] In operation S401, based on the operation information of the interactive operation, the specified large model is used to perform the task of generating corpus content according to the context corpus content, and multiple candidate corpus contents are obtained.

[0070] In operation S402, in response to a target operation targeting sub-contents in the candidate corpus content, initial intermediate sub-contents are determined from the candidate corpus content.

[0071] In operation S403, semantic fusion is performed on multiple initial sub-contents to obtain the target corpus content.

[0072] In some embodiments, a specified large model may include multiple specified large models, each performing a corpus content generation task to output candidate corpus content corresponding to each specified large model. This allows for the selection of rich corpus content for the target object performing the interactive operation by using multiple candidate corpus contents, avoiding the situation where the corpus content generated by a single large model is insufficient to meet the actual update needs of the target object, thus preventing the large model from repeatedly performing the corpus content generation task and improving the efficiency of determining the target corpus content.

[0073] In some embodiments, sub-content can be a portion of the candidate corpus content. For example, if the candidate corpus content is paragraph text, the sub-content can be a sentence within that paragraph. The target object can determine the initial intermediate sub-content to be adopted from the candidate corpus content by performing any type of target operation, such as a selection box operation. The initial intermediate sub-content can be understood as sentences or keywords that meet the target object's quality requirements and update intent requirements.

[0074] According to embodiments of this disclosure, multiple initial sub-contents can be concatenated to determine the target corpus content. Alternatively, a large language model can be used to perform semantic understanding on multiple initial sub-contents and one or more corpus contents from the corpus data as context, generating target corpus content that meets quality requirements. Thus, the semantic understanding capabilities of the large language model can be used to maintain the semantic logical coherence and linguistic fluency of the target corpus content, and to maintain semantic consistency and coherence between the target corpus content and other statements in the target corpus data. This achieves the goal of improving the data quality of the target corpus data and the data quality of feedback information by utilizing the language generation capabilities of different specified large models.

[0075] In some embodiments, a selection operation can be performed by a target object on a plurality of predefined specified large models, including model names, model versions, and model description information, to determine a plurality of specified large models for performing the corpus content generation task.

[0076] In some embodiments, the corpus content generation task includes at least one tool invocation task, in which a specified large model invokes a target tool to perform a specified task by executing the tool invocation task, thereby obtaining intermediate results for generating candidate corpus content or target corpus content.

[0077] According to embodiments of this disclosure, the target tool can be a tool resource for performing a specific task. For example, the tool resource could be an information search engine, a code execution tool, an image recognition tool, etc. A specified large model can be used to invoke the target tool by executing a tool invocation task, allowing the target tool to perform the specified task according to the tool invocation parameters output by the specified large model, obtaining the tool invocation execution result of the tool resource performing the specified task as an intermediate result.

[0078] Tools and resources may include, for example, document retrieval tools, web search tools, image processing tools, and language translation tools.

[0079] The tools include: a document retrieval tool for extracting document summaries and question-and-answer data, a database query tool for parsing query results, a web search tool for searching pages and retrieving relevant page content, a code execution and debugging tool for running program scripts in a specified language and returning the script execution results, an image processing tool for understanding image content and providing question-and-answer responses, natural language plotting, image text recognition and extraction tools, language translation tools, and multimodal data fusion tools.

[0080] In one embodiment, determining feedback information related to the target object's need intent based on target corpus data may include: determining feedback information based on target corpus data and task description information of tool invocation tasks related to the target corpus data.

[0081] In some embodiments, the task description information related to a tool invocation task includes tool task-related information. This tool task-related information may include the tool description information of the target tool being invoked, the task execution parameters of the tool invocation task, and so on. This tool task-related information can be represented in a structured manner to ensure that the task description information meets the actual needs of feedback information in a specific scenario.

[0082] For example, the task description information related to the tool invocation task may also include: the name and version identifier of the target tool, the target tool invocation parameter template, the result verification rules for the intermediate results output by the target tool, the timeout retry strategy required for invoking the target tool to execute the specified task, the execution conditions for invoking the target tool to execute the specified task, and so on.

[0083] According to embodiments of this disclosure, since the specified large model can specify the corpus content generation task based on the thinking process of thinking chains, thinking trees, etc., the corpus content generation task can include multiple target tasks connected by dependencies, and the target tasks can include tool invocation tasks. Subsequent target tasks that are dependent on the tool invocation task can use the execution result of the tool invocation task as an intermediate result, and input the intermediate result into the specified large model to execute the subsequent target task, until multiple target tasks are completed based on dependencies, resulting in candidate corpus content or target corpus content. Since the feedback information includes task description information related to the tool invocation task, the target corpus content and task description information related to the tool invocation task in the feedback information can be used to prompt the language model to be trained to learn more accurately the model's ability to perform content generation tasks by invoking tool resources, based on the relevant tool description information, tool parameter range, and intermediate result examples in the task description information. This allows the feedback information to be adapted to the scenarios of training and testing the language model, improving the quality of the target corpus data.

[0084] In one embodiment, the interactive method provided in this disclosure can be executed based on a corpus data annotation platform. The corpus data annotation platform in this embodiment will be described illustratively below.

[0085] The corpus data annotation platform includes a dialogue structure editing module, a version management module, a content verification module, and a verification and export module.

[0086] The dialog structure editing module is used to perform the following functions:

[0087] Editing of multi-turn dialogue content used as corpus data is supported, allowing editing between multiple turns of dialogue or within multiple paragraphs of the same dialogue. During dialogues between multiple large role models on a preset topic, the dialogue output by the large role models can be edited before the dialogue flow is complete.

[0088] Interactive elements are provided, including up and down arrows, insertion markers, and delete buttons for the corpus data. This allows annotators to perform structured editing operations at any location. Furthermore, after interactive operations are performed on multi-turn dialogues, other dialogue content is not cleared; updates are only made to the second corpus content with semantic dependencies. During the editing process, each dialogue is automatically saved as a draft at any time, but only when the user clicks the "submit" interactive element, ensuring the continuity of the annotator's editing process and data security.

[0089] The version management module is used to perform the following functions:

[0090] Incremental storage function: The target corpus content obtained after each update is stored using a difference algorithm. For example, when performing difference calculations on the dialogue content of version V1 and version V1.1, only the difference calculated is written, avoiding the duplication of the entire data and improving storage and retrieval efficiency.

[0091] Multi-version backtracking function: Through the patch chain, the content of any historical version of the corpus can be backtracked at any time, and the historical version can be exported as a complete corpus data, which makes it convenient for approvers to compare and restore versions.

[0092] Permissions and Collaboration: Supports multi-user collaborative editing and approval workflows. Each round of modifications records the operator, timestamp, and modification summary. Users can also perform "approval," "rejection," or "secondary editing" operations during the approval process.

[0093] The content verification module is used to verify the structural integrity, semantic coherence, format, and length of the target corpus data.

[0094] Structural integrity can automatically check for missing user questions and answers or model responses in a document and indicate "Reply missing in round N".

[0095] Semantic coherence can be assessed, for example, by using a lightweight semantic embedding model to score the topic consistency between adjacent dialogue content. If the score is below a threshold, a "Semantic jump may exist" warning will be displayed in the sidebar of the interactive interface, along with the specific keyword differences.

[0096] Format and length checks can be used to verify the number of characters, paragraphs, and special symbols in each paragraph of the dialog content to ensure that it meets the required conditions.

[0097] The system provides real-time feedback and actionable suggestions. Verification results are displayed as a "red, yellow, green light" indicator on the left side of each paragraph: green indicates no issues, yellow indicates a minor warning (suggesting the annotator check), and red indicates a serious error (suggesting the annotator correct it). Clicking the warning icon expands the list of specific issues and provides a "one-click fix" button: for example, "auto-complete placeholders for missing replies" or "auto-complete required parameters for tool call examples."

[0098] Figure 5 A block diagram of an interactive device according to an embodiment of the present disclosure is shown schematically.

[0099] like Figure 5 As shown, the interactive device 500 includes: a display module 510, a target corpus data acquisition module 520, and a feedback information determination module 530.

[0100] The first display module 510 is used to display the content of the first corpus in the received corpus data.

[0101] The target corpus data acquisition module 520 is used to update the first corpus content and the second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content in response to the interactive operation of the target object on the first corpus content, so as to obtain the target corpus data.

[0102] The feedback information determination module 530 is used to determine feedback information related to the needs and intentions of the target object based on the target corpus data.

[0103] According to embodiments of this disclosure, the target corpus data acquisition module includes: a first acquisition unit, a second acquisition unit, and a target corpus data acquisition unit.

[0104] The first acquisition unit is used to update the content of the first corpus based on the interactive operation to obtain the first intermediate content.

[0105] The second acquisition unit is used to update the content of the second corpus based on the first intermediate content using a specified large model, based on semantic dependencies, to obtain the second intermediate content.

[0106] The target corpus data acquisition unit is used to update the corpus data by performing semantic understanding on the first and second intermediate content using a specified large model, thereby obtaining the target corpus data.

[0107] According to embodiments of this disclosure, the interactive device further includes a second display module and a sub-topology determination module.

[0108] The second display module is used to display the corpus content topology, which includes corpus node elements representing the corpus content and edge elements representing the semantic dependencies between multiple corpora.

[0109] The subtopology determination module is used to determine a subtopology from the corpus content topology in response to a selection operation on the corpus content topology; wherein, the second corpus content is determined based on semantic dependencies from the corpus content corresponding to each of the multiple candidate corpus node elements in the subtopology.

[0110] According to embodiments of this disclosure, semantic dependencies are determined based on the following operation: using a specified large model to perform semantic understanding on multiple corpus contents in the corpus data based on at least one of the first corpus content and the first intermediate content, to obtain semantic dependencies.

[0111] According to embodiments of this disclosure, the target corpus data acquisition module includes: a target corpus content determination unit and a target corpus data determination unit.

[0112] The target corpus content determination unit is used to perform a corpus content generation task based on the operation information of the interactive operation and the specified large model according to the context corpus content, so as to obtain the target corpus content related to the first corpus content or the second corpus content. The context corpus content is semantically related to the first corpus content or the second corpus content, and the corpus data includes the context corpus content.

[0113] The target corpus data determination unit is used to determine the target corpus data based on the content of the target corpus.

[0114] According to embodiments of this disclosure, the target corpus content determination unit includes: a candidate corpus content acquisition subunit, an initial intermediate sub-content acquisition subunit, and a target corpus content acquisition subunit.

[0115] The candidate corpus content acquisition subunit is used to generate corpus content based on interactive operation information. It utilizes a specified large model to perform the corpus content generation task according to the context corpus content, and obtains multiple candidate corpus contents.

[0116] Initial intermediate sub-content acquisition sub-units are used to determine initial intermediate sub-content from candidate corpus content in response to target operations on sub-content in candidate corpus content.

[0117] The target corpus content acquisition subunit is used to semantically fuse multiple initial sub-contents to obtain the target corpus content.

[0118] According to embodiments of this disclosure, the corpus content generation task includes at least one tool invocation task, wherein a specified large model invokes a target tool to perform a specified task by executing the tool invocation task, thereby obtaining intermediate results for generating candidate corpus content or target corpus content; wherein the feedback information determination module includes a first determination unit.

[0119] The first determining unit is used to determine feedback information based on the target corpus data and the task description information of the tool calling task related to the target corpus data.

[0120] According to embodiments of this disclosure, the feedback information determination module includes a second determination unit and a third determination unit.

[0121] The second determining unit is used to determine the target information pair based on the target corpus content in the target corpus data and the operation information of the interactive operation related to the target corpus content.

[0122] The third determining unit is used to determine feedback information based on target information.

[0123] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0124] According to an embodiment of the present disclosure, an electronic device includes: 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, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.

[0125] According to embodiments of the present disclosure, a non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions are used to cause a computer to perform the method described above.

[0126] According to an embodiment of this disclosure, a computer program product includes a computer program that, when executed by a processor, implements the method described above.

[0127] Figure 6 A schematic block diagram of an example electronic device that can be used to implement the interactive methods of 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 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0128] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 602 or a computer program loaded into random access memory (RAM) 603 from storage unit 608. RAM 603 may also store various programs and data required for the operation of device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0129] Multiple components in device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of monitors, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0130] The computing unit 601 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as interactive methods. For example, in some embodiments, the interactive method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the interactive method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform interactive methods by any other suitable means (e.g., by means of firmware).

[0131] Various embodiments of the systems and techniques described above 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), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0133] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0135] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0136] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, distributed system servers, or servers incorporating blockchain technology.

[0137] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0138] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An interaction method, comprising: Displays the content of the first corpus in the received corpus data; In response to the interactive operation of the target object on the content of the first corpus, the content of the first corpus and the content of the second corpus in the corpus data that have a semantic dependency relationship with the content of the first corpus are updated to obtain the target corpus data; as well as Based on the target corpus data, feedback information related to the needs and intentions of the target object is determined.

2. The method according to claim 1, wherein, The step of updating the first corpus content and the second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content includes: The content of the first corpus is updated according to the interactive operation to obtain the first intermediate content; Based on the semantic dependencies, the second corpus content is updated using a specified large model according to the first intermediate content to obtain the second intermediate content; and The target corpus data is obtained by using the specified large model to perform semantic understanding on the first intermediate content and the second intermediate content.

3. The method according to claim 1 or 2, wherein, The method further includes: The corpus content topology is displayed, which includes corpus node elements representing corpus content and edge elements representing semantic dependencies between multiple corpora. In response to a selection operation for the corpus content topology, a sub-topology is determined from the corpus content topology; wherein the second corpus content is determined based on the semantic dependency relationship from the corpus content corresponding to each of the plurality of candidate corpus node elements in the sub-topology.

4. The method according to claim 2, wherein, The semantic dependencies are determined based on the following operations: Using a specified large model, semantic understanding is performed on multiple corpus contents in the corpus data based on at least one of the first corpus content and the first intermediate content to obtain the semantic dependency relationship.

5. The method according to claim 1, wherein, Updating the content of the first corpus and the content of the second corpus in the corpus data that has a semantic dependency on the content of the first corpus includes: Based on the operation information of the interactive operation, a corpus content generation task is performed using a specified large model according to the context corpus content to obtain target corpus content related to the first corpus content or the second corpus content. The context corpus content is semantically related to the first corpus content or the second corpus content, and the corpus data includes the context corpus content; and The target corpus data is determined based on the content of the target corpus.

6. The method according to claim 5, wherein, The operation information based on the interactive operation, utilizing a specified large model to perform a corpus content generation task based on the contextual corpus content, includes: Based on the operation information of the interactive operation, a specified large model is used to perform a corpus content generation task according to the context corpus content to obtain multiple candidate corpus contents. In response to a target operation on a sub-content within the candidate corpus content, initial intermediate sub-content is determined from the candidate corpus content; Semantic fusion is performed on multiple initial sub-contents to obtain the target corpus content.

7. The method according to claim 5 or 6, wherein, The corpus content generation task includes at least one tool invocation task. The specified large model invokes the target tool to perform the specified task by executing the tool invocation task, and obtains intermediate results for generating candidate corpus content or target corpus content. The step of determining feedback information related to the target object's need intent based on the target corpus data includes: The feedback information is determined based on the target corpus data and the task description information of the tool invocation task related to the target corpus data.

8. The method according to claim 1, wherein, The step of determining feedback information related to the target object's need intent based on the target corpus data includes: Based on the target corpus content in the target corpus data and the operation information of the interactive operations related to the target corpus content, target information pairs are determined; and The feedback information is determined based on the target information.

9. An interactive device, comprising: The first display module is used to display the content of the first corpus in the received corpus data; The target corpus data acquisition module is used to update the first corpus content and the second corpus content in the corpus data that has a semantic dependency relationship with the first corpus content in response to the interactive operation of the target object on the first corpus content, so as to obtain the target corpus data. as well as The feedback information determination module is used to determine feedback information related to the needs and intentions of the target object based on the target corpus data.

10. 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 to enable the at least one processor to perform the method of any one of claims 1-8.

11. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-8.

12. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-8.

Images