Information processing method and device, computer program product and electronic equipment

Abstract

The invention relates to the technical field of computers, in particular to an information processing method and device, a computer program product and electronic equipment. The information processing method comprises the steps that original keywords are extracted according to user input information, and a target scene type corresponding to the user input information is determined; a target cue word template corresponding to the target scene type is obtained from a pre-constructed template set, the template set comprises a plurality of cue word templates, and each cue word template has the corresponding scene type; and generating a target cue word according to the target cue word template and the original keyword, and inputting the target cue word into the large model to obtain an information processing result. According to the method, the cue word generation accuracy can be improved, and a large model can directly realize a customized task.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and more specifically, to an information processing method, an information processing device, a computer program product, and an electronic device. Background Art

[0002] With the development of computer and artificial intelligence technology, large models have been widely used in various fields (such as natural language processing). Large models refer to neural network models with ultra-large-scale parameters (usually more than one billion). Prompt words are keywords or phrases used to guide the model to understand and generate outputs during the training and use of large models. The accuracy of prompt words plays an important role in the large model's ability to output accurate processing results.

[0003] At present, the big model only provides general chat capabilities to the outside world. When reusing the big model's capabilities in different scenarios, it is necessary to ask questions to the big model by designing targeted prompts to obtain the target results. It is difficult for non-professionals to design reasonable and efficient prompts, which affects the accuracy of the big model's information processing and makes it impossible for the big model to accurately implement customized tasks.

[0004] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0005] The purpose of the present disclosure is to provide an information processing method and device, a computer program product and an electronic device, thereby improving the accuracy of generating prompt words at least to a certain extent, so that a large model can directly realize customized tasks.

[0006] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0007] According to one aspect of the present disclosure, there is provided an information processing method, comprising: extracting original keywords according to user input information, and determining a target scene type corresponding to the user input information; acquiring a target prompt word template corresponding to the target scene type from a pre-constructed template set, wherein the template set comprises a plurality of prompt word templates, each of the prompt word templates having a corresponding scene type; generating a target prompt word according to the target prompt word template and the original keywords, so as to input the target prompt word into a large model to obtain an information processing result.

[0008] In an exemplary embodiment of the present disclosure, the determining the target scene type corresponding to the user input information includes: identifying a target scene identifier from the user input information, and determining the target scene type according to the target scene identifier.

[0009] In an exemplary embodiment of the present disclosure, before extracting original keywords based on user input information and determining the target scene type corresponding to the user input information, the method also includes: in response to a prompt word generation operation, presenting candidate scene identifiers of the scene type corresponding to each of the prompt word templates in the template set; determining the user input information includes: in response to a scene selection operation performed in the candidate scene identifier, generating the target scene identifier in the user input information; or, in response to an input operation based on the candidate scene identifier, generating the target scene identifier in the user input information according to the input candidate scene identifier.

[0010] In an exemplary embodiment of the present disclosure, the user input information includes multiple input keywords, and extracting original keywords based on the user input information includes: extracting the multiple input keywords in sequence according to the arrangement order of the multiple input keywords in the user input information; determining the multiple input keywords as the original keywords, wherein the position identifiers of the original keywords are determined according to the order in which the multiple input keywords are extracted.

[0011] In an exemplary embodiment of the present disclosure, generating a target prompt word according to the target prompt word template and the original keyword includes: determining a filling position of the original keyword in the target prompt word template according to a position identifier corresponding to the original keyword; and filling the original keyword into the target prompt word template based on the filling position to generate the target prompt word.

[0012] In an exemplary embodiment of the present disclosure, before extracting original keywords based on user input information and determining the target scene type corresponding to the user input information, the method also includes: in response to a prompt word generation operation, presenting reference keyword information of the scene type corresponding to each of the prompt word templates in the template set; wherein the reference keyword information of each of the scene types at least includes reference keywords and the order of the reference keywords.

[0013] In an exemplary embodiment of the present disclosure, the acquiring the target prompt word template corresponding to the target scene type from a pre-constructed template set includes: acquiring a reference prompt word template from the template set according to the target scene type, the reference prompt word template having associated reference keywords; and verifying the reference prompt word template based on the reference keywords associated with the reference prompt word template and the original keywords, so as to determine the target prompt word template according to the verification result.

[0014] In an exemplary embodiment of the present disclosure, the verification of the reference prompt word template based on the reference keyword associated with the reference prompt word template and the original keyword to determine the target prompt word template according to the verification result includes: keyword matching the reference keyword associated with the reference prompt word template with the original keyword; if the reference keyword associated with the reference prompt word template matches the original keyword, determining the reference prompt word template as the target prompt word template; if the reference keyword associated with the reference prompt word template does not match the original keyword, presenting unmatched prompt information to respond to the template adjustment operation based on the unmatched prompt information to obtain the target prompt word template.

[0015] In an exemplary embodiment of the present disclosure, the process of constructing the template set includes: obtaining configuration information corresponding to the scene type, and determining the reference prompt words of the scene type according to the configuration information; removing target keywords from the reference prompt words according to the slot setting rules corresponding to the scene type to obtain a prompt word template corresponding to the scene type; and associating the prompt word template, the scene type and the removed keywords.

[0016] In an exemplary embodiment of the present disclosure, the step of inputting the target prompt word into a large model to obtain an information processing result includes: acquiring scene description information of the target scene type; inputting the scene description information and the target prompt word into the large model to obtain the information processing result; wherein the large model executes the task corresponding to the target prompt word based on the scene description information.

[0017] According to one aspect of the present disclosure, there is provided an information processing device, comprising: an information extraction module, for extracting original keywords according to user input information, and determining a target scene type corresponding to the user input information; a template acquisition module, for acquiring a target prompt word template corresponding to the target scene type from a pre-constructed template set, wherein the template set comprises a plurality of prompt word templates, each of which has a corresponding scene type; and an information processing module, for generating a target prompt word according to the target prompt word template and the original keywords, so as to input the target prompt word into a large model to obtain an information processing result.

[0018] According to one aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, any one of the methods described above is implemented.

[0019] According to one aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute any one of the above-mentioned methods by executing the executable instructions.

[0020] The information processing method in the exemplary embodiment of the present disclosure, on the one hand, extracts the original keywords according to the user input information, determines the target scene type corresponding to the user input information, obtains the target prompt word corresponding to the target scene type from the pre-built template set, and then generates the target prompt word according to the target prompt word template and the original keywords. This process obtains the target prompt word template that meets the scene requirements through the matching of the scene type. The user only needs to pay attention to the keywords in the specific scene, and the target prompt word corresponding to the specific scene can be generated according to the keywords input by the user and the target prompt word template, thereby improving the accuracy and professionalism of the target prompt word. On the other hand, the target prompt word that meets the target scene type is used as the input of the large model, so that the large model can realize information processing in the specified scene, that is, directly realize the customized task, improve the reusability of the large model, and reduce the learning and use cost of the large model.

[0021] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the following detailed description with reference to the accompanying drawings, in which several embodiments of the present disclosure are shown in an exemplary and non-limiting manner.

[0023] Figure 1 An application environment according to an exemplary embodiment of the present disclosure is shown.

[0024] Figure 2 A flowchart of an information processing method according to an exemplary embodiment of the present disclosure is shown.

[0025] Figure 3 A schematic diagram of an interactive interface for obtaining a target scene identification in user input information according to an exemplary embodiment of the present disclosure is shown.

[0026] Figure 4 A flowchart of an implementation method for extracting original keywords according to an exemplary embodiment of the present disclosure is shown.

[0027] Figure 5 A flow chart of an implementation method for generating a target prompt word according to an exemplary embodiment of the present disclosure is shown.

[0028] Figure 6 A schematic diagram of acquiring a target prompt word according to an exemplary embodiment of the present disclosure is shown.

[0029] Figure 7 An interactive interface diagram for presenting reference keyword information of a scene type corresponding to a prompt word template in a template set according to an exemplary embodiment of the present disclosure is shown.

[0030] Figure 8 A flow chart of an implementation method for acquiring a target prompt word template according to an exemplary embodiment of the present disclosure is shown.

[0031] Fig. 9 A flowchart of an implementation method for constructing a template set according to an exemplary embodiment of the present disclosure is shown.

[0032] Fig.10 A complete flowchart of an information processing method according to an exemplary embodiment of the present disclosure is shown.

[0033] Fig.11 A schematic diagram showing the composition of an information processing device according to an exemplary embodiment of the present disclosure is shown.

[0034] Fig.12 A block diagram of an electronic device according to an exemplary embodiment of the present disclosure is shown.

[0035] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0036] The exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that the present disclosure will be more comprehensive and complete and the concepts of the exemplary embodiments will be fully conveyed to those skilled in the art. The same reference numerals in the figures represent the same or similar structures, and thus their detailed description will be omitted.

[0037] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known structures, methods, devices, implementations or operations are not shown or described in detail to avoid blurring the various aspects of the present disclosure.

[0038] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or these functional entities or parts of functional entities may be implemented in one or more software hardened modules, or these functional entities may be implemented in different networks and / or processor devices and / or microcontroller devices.

[0039] The information processing method provided by the exemplary embodiments of the present disclosure can be applied to Figure 1 In the application environment shown, the terminal 101 communicates with the server 102 via a network. The data storage system can store data that the server 102 needs to process. The data storage system can be integrated on the server 102, or placed on the cloud or other network servers.

[0040] In an exemplary embodiment, the information processing method provided by the exemplary embodiment of the present disclosure may be executed by the server 102, and the corresponding information processing device is disposed in the server 102. Correspondingly, in this mode of execution by the server 102, the server 102 may start to execute the steps in the technical solution of the exemplary embodiment of the present disclosure in response to a trigger command, wherein the trigger command may be sent by a terminal used by a user, or may be triggered locally by the server in response to some automation events.

[0041] The server 102 can extract the original keywords based on the user input information, and determine the target scene type corresponding to the user input information, and then obtain the target prompt word template corresponding to the target scene type from the pre-built template set, and finally generate the target prompt word based on the target prompt word template and the original keywords, so as to input the target prompt word into the large model to obtain the information processing result.

[0042] The server 102 may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server 102 may perform background tasks.

[0043] Furthermore, in another exemplary embodiment, the terminal 101 may also have similar functions to the server 102 , thereby executing the information processing method provided by the exemplary embodiment of the present disclosure.

[0044] The terminal 101 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, an IoT device, and a portable wearable device. The IoT device may be a smart speaker, a smart TV, a smart air conditioner, and a smart vehicle-mounted device, etc. The portable wearable device may be a smart watch, a smart bracelet, a head-mounted device, etc. The terminal 101 may also be referred to as a mobile terminal, a terminal device, a mobile device, etc. The exemplary embodiments of the present disclosure do not limit the type of the terminal 101.

[0045] In addition, the technical solution of the exemplary embodiment of the present disclosure can also be executed by the terminal 101 and the server 102 in collaboration. In this mode of collaborative execution by the terminal 101 and the server 102, some steps in the technical solution provided by the exemplary embodiment of the present disclosure are executed by the terminal 101, while other steps are executed by the server 102. For example, the terminal 101 extracts the original keywords according to the user input information, and determines the target scene type corresponding to the user input information; obtains the target prompt word template corresponding to the target scene type from the pre-built template set; generates the target prompt word according to the target prompt word template and the original keywords, and sends the obtained target prompt word to the server 102, and the server 102 inputs the target prompt word into the large model to obtain the information processing result.

[0046] It should be noted that in this mode of collaborative execution by the terminal 101 and the server 102, the steps respectively executed by the terminal 101 and the server 102 can be dynamically adjusted according to actual conditions, and there is no special restriction on this.

[0047] The terminal 101 and the server 102 may be connected directly or indirectly via wireless communication, and the exemplary embodiments of the present disclosure do not impose any special limitation thereto.

[0048] In related technologies, big models generally only provide general chat capabilities to the outside world. When each business scenario needs to reuse the big model capabilities, it is necessary to design targeted prompt words to ask the big model questions to obtain the target results. However, in different scenarios, big models may require completely different types of tasks. For example, the translation scenario requires the big model to complete the text translation task, and the summary scenario requires the big model to complete the task of summarizing and summarizing a paragraph, and so on.

[0049] However, the current large models cannot directly implement customized tasks, and it is difficult for non-professional teams to design reasonable and efficient prompt words, resulting in the inability of large models to accurately perform corresponding tasks.

[0050] In view of this, an exemplary embodiment of the present disclosure provides an information processing method, which obtains a target prompt word template that meets the scene requirements through scene matching, so as to generate a target prompt word based on the target prompt word template and the original keywords, thereby reducing the difficulty of generating specialized prompt words that meet the scene. Users do not need to consider the details of the interaction with the big model, but only need to pay attention to the keywords in the specific scene to realize information processing of the specified scene, thereby reducing the learning cost and usage cost of the big model and improving the reusability of the big model.

[0051] Figure 2 1 is a flow chart showing an information processing method of an exemplary embodiment of the present disclosure. Figure 2 As shown, the information processing method includes steps S210 to S230, which are specifically as follows:

[0052] In step S210, original keywords are extracted according to the user input information, and the target scene type corresponding to the user input information is determined.

[0053] In an exemplary embodiment of the present disclosure, the user input information is information provided by the user for generating prompt words, including original keywords and scene information, and the original keywords and target scene types can be extracted according to the user input information. The scene information can be, for example, a scene identifier for indicating the target scene type.

[0054] As an example, the user input information may be "original keyword 1, original keyword 2, scene information". The original keyword and target scene type may be directly obtained based on the user input information. Of course, as an example, the user input information may also be semantic information with complete semantics, from which the original keyword and target scene type may be determined by performing steps such as word segmentation and entity recognition on the semantic information.

[0055] It should be noted that the keywords described in the exemplary embodiments of the present disclosure can be words, phrases or a text, which can be flexibly adjusted according to actual needs without limitation. For example, any one of "Chinese", "English" or "Abstract" can be used as a keyword, and "I am a programmer from country A and I want to learn about the culture of country B" can also be used as a keyword.

[0056] In the exemplary embodiments of the present disclosure, there may be multiple ways to divide scene types, which can be selected according to actual needs and are not limited to this.

[0057] As an example, based on the business type, the scenario types can include character profile scenarios, idiom correction scenarios, translation scenarios, intention question scenarios, and conversation summary scenarios. Different scenario types require large models to perform corresponding tasks. For example, the translation scenario requires a large model to complete the text translation task, and the summary scenario requires a large model to complete the task of summarizing and summarizing a paragraph. I will not list them one by one here.

[0058] The target scene type is selected from any one of the above scene types and is determined according to user input information.

[0059] In step S220, a target prompt word template corresponding to the target scene type is obtained from a pre-built template set.

[0060] In an exemplary embodiment of the present disclosure, the pre-built template set includes multiple prompt word templates, each prompt word template has its own corresponding scene type. The prompt word template is a template that is convenient for generating prompt word construction. By filling in the prompt word template, prompt words that meet the scene type can be quickly obtained.

[0061] In an optional embodiment, the prompt word templates in the template library can be the result of a large number of manual experiments and template fine-tuning for each scene, so that the prompt word templates meet the scene requirements. For example, the results of a large number of experiments and fine-tuning by experts related to each scene can be obtained to obtain the prompt word templates, making the prompt word templates more professional.

[0062] In an optional embodiment, the prompt word templates in the template library can also be obtained through a pre-trained template generation model. Specifically, when training the template generation model, the sample set {prompt word template samples, scene type samples} can be used as samples to train the template generation model until the loss function of the template generation model converges, thereby obtaining a trained template generation model. When in use, the scene type information can be input into the trained template generation model, so that a prompt word template that meets the scene requirements can be quickly and efficiently generated.

[0063] In step S230, a target prompt word is generated according to the target prompt word template and the original keyword, so as to input the target prompt word into the large model to obtain an information processing result.

[0064] In an exemplary embodiment of the present disclosure, after obtaining the target prompt word template and the original keyword, the target prompt word template may be filled with the original keyword to obtain the target prompt word.

[0065] Among them, the large model can be a large model of the billion level. LLM (Large Language Model) refers to a deep learning model trained with a large amount of data, which can generate natural language text or understand the meaning of language text. Alternatively, the large model can be constructed using a lightweight neural network, such as CNN (Convolutional Neural Network), RNN (Recurrent Neural Networks), DNN (Deep Neural Networks), etc. Compared with lightweight neural networks, a large model with a more complex network structure can also be used, for example, a heavyweight neural network is constructed, such as BERT (Bidirectional Encoder Representation from Transformers, feature encoder of bidirectional Transformers), Transformer (a language processing model), and any heavyweight neural network that can be used for information processing. The exemplary embodiment of the present disclosure is illustrated by taking the large model as a large language pre-training model as an example.

[0066] The information processing method in the exemplary embodiment of the present disclosure, on the one hand, extracts the original keywords according to the user input information, determines the target scene type corresponding to the user input information, obtains the target prompt word corresponding to the target scene type from the pre-built template set, and then generates the target prompt word according to the target prompt word template and the original keywords. This process obtains the target prompt word template that meets the scene requirements through the matching of the scene type. The user only needs to pay attention to the keywords in the specific scene, and the target prompt word corresponding to the specific scene can be generated according to the keywords input by the user and the target prompt word template, thereby improving the accuracy and professionalism of the target prompt word. On the other hand, the target prompt word that meets the target scene type is used as the input of the large model, so that the large model can realize information processing in the specified scene, that is, directly realize the customized task, improve the reusability of the large model, and reduce the learning and use cost of the large model.

[0067] In an exemplary embodiment, determining the target scene type corresponding to the user input information may include: identifying a target scene identifier from the user input information, and determining the target scene type according to the target scene identifier.

[0068] The scene identifier is the identification information that uniquely identifies the scene type. When constructing the prompt word template in the template set, a unique scene identifier is assigned to it. The target scene type can be determined by obtaining the target scene identifier in the user input information, which facilitates the subsequent rapid acquisition of the target prompt word template corresponding to the target scene type. In this process, the user does not need to consider the interaction details with the large model, but only needs to clarify the scene type, which reduces the difficulty and operation threshold of generating prompt word templates.

[0069] In an exemplary embodiment, determining the target scene type in the user input information may be performed in response to a human operation. That is, the process of obtaining the target scene identification in the user input information is implemented based on human operation, and the candidate scene identifications of the scene types corresponding to each prompt word template in the template set may be presented in response to the prompt word generation operation. Furthermore, the user can select or input operations based on the presented candidate scene identifications of each scene type to obtain the target scene identification in the user input information.

[0070] Figure 3 A schematic diagram of an interactive interface for obtaining a target scene identifier in user input information according to an exemplary embodiment of the present disclosure is shown, which can be obtained by, for example, Figure 3 The interactive interface presents candidate scene identifiers of scene types corresponding to each prompt word template in the template set.

[0071] It should be noted that each prompt word template in the pre-built template set and the candidate scene identifiers of the corresponding scene type can be configured in the interactive interface and viewed and modified by means of a slider. Figure 3 The possible configurations of the interactive interface are only illustrated for exemplary purposes, and the present disclosure does not limit this.

[0072] By presenting candidate scene identifiers of scene types corresponding to each prompt word template in the template set in advance, a reference can be provided for the user when inputting information, thereby improving the accuracy of the user input information.

[0073] As an example, based on the aforementioned exemplary embodiments, determining the user input information may include: generating a target scene identifier in the user input information in response to a scene selection operation performed in the candidate scene identifiers.

[0074] Continue to see Figure 3As shown, the user can select a scene in the interactive interface, and then respond to the user's scene selection operation and add the target scene identifier corresponding to the scene selection operation to the user input information.

[0075] As an example, based on the aforementioned exemplary embodiments, determining the user input information may include: responding to an input operation based on the candidate scene identifier, generating a target scene identifier in the user input information according to the input candidate scene identifier.

[0076] You can continue to see Figure 3 Based on the candidate scene identifiers provided by the interactive interface, the user can perform an input operation, select a target scene identifier from the candidate scene identifiers and input according to its content to add the target scene identifier to the user input information.

[0077] By providing candidate scene identifications of the scene types corresponding to each prompt word template in the template set in a visual manner, a reference is provided for users to input user information, thereby improving the accuracy of the target scene identification in the user input information and providing accurate scene type information for subsequent determination of the target prompt word template.

[0078] In an exemplary embodiment, a method for extracting original keywords is provided, wherein the user input information includes multiple input keywords. Figure 4 As shown, extracting original keywords according to user input information may include:

[0079] Step S410: extracting multiple input keywords in sequence according to the arrangement order of multiple input keywords in the user input information.

[0080] The multiple input keywords in the user input information have a sequence, such as {keyword 1, keyword 2, keyword 3}, and the multiple input keywords are obtained in sequence.

[0081] Step S420: determining the multiple input keywords as original keywords, wherein the position identifiers of the original keywords are determined according to the order in which the multiple input keywords are extracted.

[0082] After extracting a plurality of input keywords, these input keywords are determined as original keywords. The original keywords have corresponding position identifiers, and the position identifiers are determined according to the order in which the input keywords are extracted.

[0083] As an example, for example, the input keywords extracted in sequence are {Chinese; I am a programmer from country A, and I want to understand the culture of country B}, then the position identifier of "country A" is "001", and the position identifier of "I am a programmer from country A, and I want to understand the culture of country B" is "002". In other words, the size relationship of the position identifiers can be directly related to the sequence in which the input keywords are extracted. The position identifier of the original keyword corresponding to the input keyword extracted first is numerically smaller than the position identifier of the original keyword corresponding to the input keyword extracted later. Of course, the position identifier of the corresponding original keyword can be determined according to the sequence in which the input keywords are extracted, or by other means, for example, establishing a correspondence between the extraction sequence and the position identifier, and determining the position identifier of the corresponding original keyword based on the correspondence and the sequence in which the input keywords are actually extracted. The exemplary embodiments of the present disclosure do not limit this.

[0084] In an exemplary embodiment, a method for generating a target prompt word is provided. Figure 5 As shown, generating a target prompt word according to a target prompt word template and an original keyword may include:

[0085] Step S510: determining the filling position of the original keyword in the target prompt word template according to the position identifier corresponding to the original keyword.

[0086] Step S520: filling the original keyword into the target prompt word template based on the filling position to generate the target prompt word.

[0087] The position identifier corresponding to the original keyword is determined based on the order of the extracted input keywords, and the filling position of the original keyword in the target prompt word template can be determined based on the position identifier.

[0088] Specifically, the position to be filled in the target prompt word can be obtained by eliminating keywords. After eliminating keywords, the position to be filled in the target prompt word can have corresponding identification information, and then the position identifier corresponding to the original keyword is matched with the identification information of the position to be filled to determine the position of the original keyword in the target prompt word template.

[0089] As an example, see Figure 6 The figure shows a schematic diagram of obtaining a target prompt word. Taking the translation scenario as an example, the target prompt word template is {You are now a (1) expert. Please translate the following content into (1). Content: (2)}, the original keyword 1 is "English", the original keyword 2 is "I am a programmer from country A, I want to understand the culture of country B", and the corresponding position of "English" is marked as "1", and the position of "I am a programmer from country A, I want to understand the culture of country B" is marked as "2".

[0090] First, taking the original keyword 1 as an example, the position identifier "1" of the original keyword 1 is matched with the identifier information of the position to be filled, and "(1)" is matched, then the filling position of the original keyword 1 in the target prompt word template is determined to be the position corresponding to all "(1)". Similarly, the filling position of the original keyword 2 in the prompt word template is determined to be the position corresponding to all "(2)".

[0091] Then, according to the determined filling position, the original keyword 1 is filled into the position corresponding to "(1)", and the original keyword 2 is filled into the position corresponding to "(2)", and the target prompt word is obtained as {You are now an English expert, please translate the following content into English. Content: I am a programmer from country A, and I want to understand the culture of country B}.

[0092] It should be noted that Figure 6 This is only an example of generating a target prompt word. The exemplary embodiment of the present disclosure does not impose specific restrictions on the position identifier of the original keyword, the identification information of the position to be filled, etc., and can be flexibly adjusted according to actual needs.

[0093] By matching and filling the position identifier of the original keyword with the identifier of the position to be filled in the target prompt word template, the original keyword can be accurately filled into the target prompt word template. In the process of generating the target prompt word, the user does not need to consider the details of the interaction with the large model. He only needs to focus on the keywords in the specific scenario to achieve the specified task in the specific scenario.

[0094] Based on the aforementioned exemplary embodiment, in order to accurately match the original keywords extracted from the user input information with the target prompt word template, before extracting the original keywords according to the user input information and determining the target scene type corresponding to the user input information, the following may also be included:

[0095] In response to the prompt word generation operation, reference keyword information of the scene type corresponding to each prompt word template in the template set is presented; wherein the reference keyword information of each scene type at least includes reference keywords and a sequence between the reference keywords.

[0096] Specifically, when the user provides user input information, the reference keyword information of the scene type corresponding to each prompt word template presented in the template set may be used as an example, so that the user can provide the user input information with reference to the reference keyword information.

[0097] like Figure 7 The figure shows an interactive interface diagram for presenting reference keyword information of scene types corresponding to prompt word templates in a template set, which can be viewed by means of a slider. Figure 7The reference keyword information of "Candidate Scenario 1" is: "Keyword 1: English; Keyword 2: Content to be translated (e.g. I am a programmer from country A, I want to understand the culture of country B, the picture is omitted)", and the reference keyword information of "Candidate Scenario 2" is: "Keyword 1: Content to be summarized (e.g. I am a programmer from country A, I want to understand the culture of country B, ..., I have the opportunity to study abroad, the picture is omitted)". Other scenario types are not listed one by one, and the content and number of keywords can be flexibly adjusted according to actual needs.

[0098] pass Figure 7 , users can intuitively see the reference keywords under various scene types and the order between the reference keywords, so as to quickly and accurately provide user input information under the required scene type. Figure 7 The possible configurations of the interactive interface are only illustrated for exemplary purposes, and the present disclosure does not limit this.

[0099] In an exemplary embodiment, a method for obtaining a target prompt word template is provided. Figure 8 As shown, obtaining a target prompt word template corresponding to the target scene type from a pre-built template set may include:

[0100] Step S810: obtaining a reference prompt word template from a template set according to the target scene type, wherein the reference prompt word template has an associated reference keyword.

[0101] Step S820: verifying the reference prompt word template based on the reference keywords and original keywords associated with the reference prompt word template, so as to determine the target prompt word template according to the verification result.

[0102] Reference keywords refer to keywords that can be used to fill in the reference prompt word template, and may be keywords that are removed when the reference prompt word template is pre-built. The reference prompt word template can be verified by obtaining the reference keywords and the original keywords.

[0103] The reference keyword associated with the reference prompt word template can be matched with the original keyword. Keyword matching includes at least one of keyword quantity matching, keyword attribute matching (such as part of speech, entity type, etc.), etc., which can be set according to needs.

[0104] Taking keyword quantity matching as an example, the number of reference keywords associated with the reference prompt word template can be matched with the number of original keywords. If the reference keywords associated with the reference prompt word template match the original keywords, the reference prompt word template is determined as the target prompt word template. In other words, since the target scene type provided by the user is determined in the user input information, if the number of reference keywords matches that of the original keywords, it indicates to a certain extent that the scene type provided by the user is accurate. On the contrary, if the number of reference keywords does not match that of the original keywords, it indicates that the actual number of keywords required for the target scene type is different from the number of keywords provided by the user, and it is necessary to check whether the scene information in the user input information is accurate, that is, the reference prompt word template may not be accurate.

[0105] Based on this, the unmatched prompt information can be presented to respond to the template adjustment operation based on the unmatched prompt information to obtain the target prompt word template. Of course, the adjusted reference prompt word template also needs to be verified to ensure that the correct reference prompt word template is finally determined as the target prompt word template.

[0106] By verifying the reference prompt word template with reference keywords and original keywords associated with the reference prompt word template, it can be ensured that the final target prompt word template is accurate and can be accurately filled with original keywords, so as to improve the accuracy of information processing based on the target prompt word by the subsequent large model.

[0107] In an exemplary embodiment, a method for constructing a template set is provided. Fig. 9 A schematic diagram of constructing a template set is shown. The process of constructing the template set may include:

[0108] Step 910: Acquire configuration information corresponding to the scene type, and determine a reference prompt word for the scene type according to the configuration information.

[0109] The configuration information corresponding to the scene type is information used to determine the reference prompt words of the scene type. Professionals under each scene type can obtain the configuration information corresponding to each scene type through a large number of experiments and fine-tuning techniques for specific scenes, and then obtain the reference prompt words of the scene type. For example, "You are now an English expert. Please translate the following content into English. Content: I am a programmer from country A, and I want to understand the culture of country B."

[0110] Step 920: according to the slot setting rules corresponding to the scene type, the target keyword in the reference prompt word is removed to obtain a prompt word template corresponding to the scene type.

[0111] The slot setting rule refers to the rule for removing keywords from the reference prompt words of each scene type, which can be flexibly adjusted according to needs and is not restricted.

[0112] As an example, "English" and "I am a programmer from country A and I want to learn about the culture of country B" can be eliminated to obtain the prompt word template "You are now an expert in (1). Please translate the following content into (1); content: (2)".

[0113] Step S930: Associating the prompt word template, the scene type, and the eliminated keywords.

[0114] After obtaining the prompt word template, the prompt word template, the scene type and the eliminated key may be associated to complete the definition of a scene type, that is, to obtain a prompt word template corresponding to a scene type.

[0115] Among them, the scene type can be represented by a scene identifier that uniquely identifies the scene, and there is no restriction on this.

[0116] By defining a set of prompt word templates for asking questions to the large model in advance for each scene type, the extracted original keywords can be filled into the corresponding positions of the prompt word templates to complete the assembly of the prompt words, so as to quickly and accurately obtain the target prompt words.

[0117] In an exemplary embodiment, inputting the target prompt word into the large model to obtain the information processing result may include:

[0118] Obtaining scene description information of the target scene type, and inputting the scene description information and the target prompt word into the large model to obtain an information processing result;

[0119] Among them, the large model performs the task corresponding to the target prompt word based on the scene description information.

[0120] Among them, the scene description information can be a natural language description of the target scene type, and the target prompt word is a prompt word that conforms to the target scene type. At the same time, by adding scene description information for the target scene type, the large model can be guided to fully understand the scene requirements when performing the task corresponding to the target prompt word, and then obtain information processing results that conform to the scene.

[0121] Fig.10 A complete flow chart of an information processing method is shown below. Fig.10 An information processing method of an exemplary embodiment of the present disclosure is described.

[0122] First, the original keywords are extracted according to the user input information, and the target scene type corresponding to the user input information is determined.

[0123] The target scene identifier can be identified from the user input information, and the target scene type can be determined according to the target scene identifier. According to the arrangement order of multiple input keywords in the user input information, multiple input keywords are extracted in sequence, and the multiple input keywords are determined as original keywords, wherein the position identifier of the original keyword is determined according to the order in which the multiple input keywords are extracted.

[0124] Secondly, the target prompt word template corresponding to the target scene type is obtained from the pre-built template set.

[0125] Among them, according to the target scene type, a reference prompt word template is obtained from the template set, the reference prompt word template has an associated reference keyword, and based on the reference keyword and original keyword associated with the reference prompt word template, the reference prompt word template is verified to determine the target prompt word template according to the verification result.

[0126] Furthermore, a target prompt word is generated according to the target prompt word template and the original keyword.

[0127] The filling position of the original keyword in the target prompt word template is determined according to the position identifier corresponding to the original keyword, and based on the filling position, the original keyword is filled into the target prompt word template to generate the target prompt word.

[0128] Finally, the target prompt word is input into the large model to obtain the information processing result.

[0129] In this case, scene description information of the target scene type may also be obtained, and the scene description information and the target prompt word may be input into the large model to obtain an information processing result.

[0130] It should be noted that Fig.10 The corresponding contents of each step have been recorded in the above exemplary embodiments and will not be repeated here.

[0131] The information processing method in the exemplary embodiment of the present disclosure, on the one hand, extracts the original keywords according to the user input information, determines the target scene type corresponding to the user input information, obtains the target prompt word corresponding to the target scene type from the pre-built template set, and then generates the target prompt word according to the target prompt word template and the original keywords. This process obtains the target prompt word template that meets the scene requirements through the matching of the scene type. The user only needs to pay attention to the keywords in the specific scene, and the target prompt word corresponding to the specific scene can be generated according to the keywords input by the user and the target prompt word template, thereby improving the accuracy and professionalism of the target prompt word. On the other hand, the target prompt word that meets the target scene type is used as the input of the large model, so that the large model can realize information processing in the specified scene, that is, directly realize the customized task, improve the reusability of the large model, and reduce the learning and use cost of the large model.

[0132] In an exemplary embodiment of the present disclosure, an information processing device is also provided. Fig.11 As shown, the information processing device 1100 may include an information extraction module 1110, a template acquisition module 1120 and an information processing module 1130. Specifically:

[0133] The information extraction module 1110 is used to extract original keywords according to the user input information and determine the target scene type corresponding to the user input information;

[0134] The template acquisition module 1120 is used to acquire a target prompt word template corresponding to the target scene type from a pre-built template set, wherein the template set includes a plurality of prompt word templates, each of which has a corresponding scene type;

[0135] The information processing module 1130 is used to generate a target prompt word according to the target prompt word template and the original keyword, so as to input the target prompt word into the large model to obtain an information processing result.

[0136] In an exemplary embodiment of the present disclosure, the information extraction module 1110 is configured to execute: identifying a target scene identifier from the user input information, and determining the target scene type according to the target scene identifier.

[0137] In an exemplary embodiment of the present disclosure, the information extraction module 1110 is further configured to perform: in response to a prompt word generation operation, presenting a candidate scene identifier of a scene type corresponding to each of the prompt word templates in the template set; in response to a scene selection operation performed in the candidate scene identifier, generating the target scene identifier in the user input information; or, in response to an input operation based on the candidate scene identifier, generating the target scene identifier in the user input information according to the input candidate scene identifier.

[0138] In an exemplary embodiment of the present disclosure, the user input information includes multiple input keywords, and the information extraction module 1110 is configured to perform: extracting the multiple input keywords in sequence according to the arrangement order of the multiple input keywords in the user input information; determining the multiple input keywords as the original keywords, wherein the position identifiers of the original keywords are determined according to the order in which the multiple input keywords are extracted.

[0139] In an exemplary embodiment of the present disclosure, the information processing module 1130 is configured to execute: determining the filling position of the original keyword in the target prompt word template according to the position identifier corresponding to the original keyword; and filling the original keyword into the target prompt word template based on the filling position to generate the target prompt word.

[0140] In an exemplary embodiment of the present disclosure, the information extraction module 1110 is further configured to perform: in response to a prompt word generation operation, present reference keyword information of the scene type corresponding to each of the prompt word templates in the template set; wherein the reference keyword information of each of the scene types at least includes reference keywords and the order of the reference keywords.

[0141] In an exemplary embodiment of the present disclosure, the template acquisition module 1120 is configured to perform: according to the target scene type, obtaining a reference prompt word template from the template set, the reference prompt word template having associated reference keywords; based on the reference keywords associated with the reference prompt word template and the original keywords, verifying the reference prompt word template to determine the target prompt word template according to the verification result.

[0142] In an exemplary embodiment of the present disclosure, the template acquisition module 1120 is configured to execute: keyword matching the reference keyword associated with the reference prompt word template with the original keyword; if the reference keyword associated with the reference prompt word template matches the original keyword, determining the reference prompt word template as the target prompt word template; if the reference keyword associated with the reference prompt word template does not match the original keyword, presenting unmatched prompt information in response to a template adjustment operation based on the unmatched prompt information, and acquiring the target prompt word template.

[0143] In an exemplary embodiment of the present disclosure, the process of constructing the template set includes: obtaining configuration information corresponding to the scene type, and determining the reference prompt words of the scene type according to the configuration information; removing target keywords from the reference prompt words according to the slot setting rules corresponding to the scene type to obtain a prompt word template corresponding to the scene type; and associating the prompt word template, the scene type and the removed keywords.

[0144] In an exemplary embodiment of the present disclosure, the information processing module 1130 is configured to perform: obtaining scene description information of the target scene type; inputting the scene description information and the target prompt word into the big model to obtain the information processing result; wherein the big model executes the task corresponding to the target prompt word based on the scene description information.

[0145] Since the details of the various functional modules of the information processing device of the exemplary embodiment of the present disclosure have been described in the exemplary embodiment of the above-mentioned information processing method, they will not be repeated here.

[0146] It should be noted that although several modules or units of the information processing device are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0147] The exemplary embodiments of the present disclosure also provide a computer program product, which includes a computer program, and when the computer program is executed by a processor, the above-mentioned information processing method is implemented.

[0148] In one embodiment, the computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing a computer program. The readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid-state drive (SSD), and the like. Exemplarily, the computer program product may be implemented as a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory, and the like.

[0149] In one embodiment, the computer program product may be an intangible product including a computer program. Exemplarily, the computer program product may be implemented as a virtual digital product, such as a digital file storing an executable file, an installation package, etc. of the computer program.

[0150] The code of the computer program can be written in one or more programming languages. Programming languages ​​such as C language, Java, C++, etc. The program code can be executed entirely on the user computing device, or partially on the user computing device, or as a separate software package, or partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., an Internet connection provided by an operator).

[0151] The computer program may be carried or transmitted via electrical, magnetic, optical, electromagnetic, infrared, or other signals. The electronic device may convert the signal carrying the computer program into a digital signal, thereby running the computer program. When the computer program is run on an electronic device, its code is used to enable the electronic device to execute (more specifically, the processor of the electronic device may execute) the method steps of various exemplary embodiments of the present disclosure.

[0152] In addition, in an exemplary embodiment of the present disclosure, an electronic device capable of implementing the above method is also provided. It will be appreciated by those skilled in the art that various aspects of the present disclosure may be implemented as a system, method, or program product. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: a complete hardware embodiment, a complete software embodiment (including firmware, microcode, etc.), or an embodiment combining hardware and software aspects, which may be collectively referred to herein as a "circuit", "module", or "system".

[0153] Refer to the following Fig.12 1200 according to this embodiment of the present disclosure is described. Fig.12 The electronic device 1200 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0154] like Fig.12 As shown, the electronic device 1200 is in the form of a general computing device. The components of the electronic device 1200 may include, but are not limited to: the at least one processing unit 1210, the at least one storage unit 1220, a bus 1230 connecting different system components (including the storage unit 1220 and the processing unit 1210), and a display unit 1240.

[0155] The storage unit stores program codes, which can be executed by the processing unit 1210, so that the processing unit 1210 executes the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification.

[0156] The storage unit 1220 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 1221 and / or a cache memory unit 1222 , and may further include a read-only memory unit (ROM) 1223 .

[0157] The storage unit 1220 may also include a program / utility 1224 having a set (at least one) of program modules 1225, such program modules 1225 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0158] Bus 1230 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0159] The electronic device 1200 may also communicate with one or more external devices 1300 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 1200, and / or communicate with any device that enables the electronic device 1200 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 1250. In addition, the electronic device 1200 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 1260. As shown, the network adapter 1260 communicates with other modules of the electronic device 1200 via a bus 1230. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 1200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0160] Through the description of the above embodiments, it is easy for those skilled in the art to understand that the example embodiments described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the embodiment of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, and includes several instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to execute the method according to the embodiment of the present disclosure.

[0161] In addition, the above-mentioned figures are only schematic illustrations of the processes included in the method according to the exemplary embodiments of the present disclosure, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0162] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. An information processing method, characterized in that: include: Extracting original keywords according to user input information, and determining a target scene type corresponding to the user input information; Acquire a target prompt word template corresponding to the target scene type from a pre-built template set, wherein the template set includes a plurality of prompt word templates, each of which has a corresponding scene type; A target prompt word is generated according to the target prompt word template and the original keyword, so as to input the target prompt word into the large model to obtain an information processing result.

2. The method according to claim 1, characterized in that The determining the target scene type corresponding to the user input information includes: A target scene identifier is identified from the user input information, and the target scene type is determined according to the target scene identifier.

3. The method according to claim 2, characterized in that Before extracting original keywords according to the user input information and determining the target scene type corresponding to the user input information, the method further includes: In response to the prompt word generation operation, presenting candidate scene identifiers of scene types corresponding to each of the prompt word templates in the template set; Determining the user input information includes: In response to a scene selection operation performed in the candidate scene identifier, generating the target scene identifier in the user input information; Alternatively, in response to an input operation based on the candidate scene identifier, the target scene identifier in the user input information is generated according to the input candidate scene identifier.

4. The method according to claim 1, characterized in that: The user input information includes a plurality of input keywords, and the extracting original keywords according to the user input information includes: Extracting the multiple input keywords in sequence according to the arrangement order of the multiple input keywords in the user input information; The multiple input keywords are determined as the original keywords, wherein the position identifiers of the original keywords are determined according to the order in which the multiple input keywords are extracted.

5. The method according to claim 4, characterized in that The step of generating a target prompt word according to the target prompt word template and the original keyword includes: Determining the filling position of the original keyword in the target prompt word template according to the position identifier corresponding to the original keyword; Based on the filling position, the original keyword is filled into the target prompt word template to generate the target prompt word.

6. The method according to claim 4, characterized in that Before extracting original keywords according to the user input information and determining the target scene type corresponding to the user input information, the method further includes: In response to the prompt word generation operation, presenting reference keyword information of the scene type corresponding to each of the prompt word templates in the template set; The reference keyword information of each scene type at least includes reference keywords and the sequence between the reference keywords.

7. The method according to claim 1, characterized in that The step of acquiring the target prompt word template corresponding to the target scene type from the pre-built template set includes: According to the target scene type, a reference prompt word template is acquired from the template set, wherein the reference prompt word template has an associated reference keyword; The reference prompt word template is verified based on the reference keyword associated with the reference prompt word template and the original keyword, so as to determine the target prompt word template according to the verification result.

8. The method according to claim 7, characterized in that The verifying the reference prompt word template based on the reference keyword and the original keyword associated with the reference prompt word template to determine the target prompt word template according to the verification result includes: Perform keyword matching on the reference keyword associated with the reference prompt word template and the original keyword; If the reference keyword associated with the reference prompt word template matches the original keyword, the reference prompt word template is determined as the target prompt word template; If the reference keyword associated with the reference prompt word template does not match the original keyword, then unmatched prompt information is presented to respond to the template adjustment operation based on the unmatched prompt information and obtain the target prompt word template.

9. The method according to any one of claims 1 to 8, characterized in that The process of constructing the template set includes: Acquire configuration information corresponding to the scene type, and determine a reference prompt word for the scene type according to the configuration information; According to the slot setting rule corresponding to the scene type, the target keyword in the reference prompt word is removed to obtain a prompt word template corresponding to the scene type; The prompt word template, the scene type and the eliminated keywords are associated.

10. The method according to any one of claims 1 to 8, characterized in that The step of inputting the target prompt word into the large model to obtain an information processing result includes: Acquire scene description information of the target scene type; Inputting the scene description information and the target prompt word into the large model to obtain the information processing result; The large model executes the task corresponding to the target prompt word based on the scene description information.

11. An information processing device, characterized in that: The device comprises: An information extraction module, used to extract original keywords according to user input information, and determine the target scene type corresponding to the user input information; A template acquisition module, used to acquire a target prompt word template corresponding to the target scene type from a pre-built template set, wherein the template set includes a plurality of prompt word templates, each of which has a corresponding scene type; The information processing module is used to generate a target prompt word according to the target prompt word template and the original keyword, so as to input the target prompt word into the large model to obtain an information processing result.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

13. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 10 by executing the executable instructions.

Images