Question answering method and device, equipment, storage medium and program product

Abstract

The invention provides a question answering method and device, electronic equipment, a computer readable storage medium and a computer program product, and relates to the technical field of artificial intelligence such as question answering service, database technology and natural language processing. A specific embodiment of the method comprises the steps of determining a target user classification of a user based on a user portrait of the user in response to received question information sent by the user; querying target rewriting answer information corresponding to the question information and the target user classification from a target knowledge base, wherein preset question information, standard answer information corresponding to the preset question information and rewriting answer information which is obtained by rewriting the standard answer information and respectively corresponds to different user classifications are maintained in the target knowledge base; and in response to the queried target rewriting answer information, providing the target rewriting answer information to the user. Therefore, the readability of the answer information can be improved, and the reading risk of the answer information to the user due to improper word use and the like is avoided.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to the fields of question-answering services, database technology, natural language processing and other artificial intelligence technologies, and particularly to methods, apparatuses, electronic devices, computer-readable storage media and computer program products for answering questions. Background Technology

[0002] With the development of computer technology, intelligent online question-and-answer service systems have become important tools for improving user experience and information retrieval efficiency. With the rapid development of internet technology and the continuous optimization of artificial intelligence algorithms, intelligent question-and-answer systems based on technologies such as natural language processing, machine learning, and deep learning have gradually matured and are widely used in various industries. These systems can automatically process user-input text, understand and analyze questions through intelligent algorithms, and provide corresponding reference answers.

[0003] Against this backdrop, how to further improve the quality of responses during the question-and-answer process and reduce or avoid service risks in this process is a matter of concern and urgent need. Summary of the Invention

[0004] This disclosure provides a method, apparatus, electronic device, computer-readable storage medium, and computer program product for answering questions.

[0005] In a first aspect, embodiments of this disclosure propose a method for answering questions, comprising: in response to receiving question information sent by a user, determining a target user category based on the user's user profile; querying target rewritten answer information corresponding to the question information and the target user category from a target knowledge base, wherein the target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; and providing the target rewritten answer information to the user in response to finding the target rewritten answer information.

[0006] Secondly, embodiments of this disclosure provide a question-answering apparatus, comprising: a user classification determination unit configured to, in response to receiving question information sent by a user, determine a target user classification based on the user's user profile; a knowledge base query unit configured to query target rewritten answer information corresponding to the question information and the target user classification from a target knowledge base, wherein the target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user classifications obtained by rewriting the standard answer information; and a rewritten answer providing unit configured to, in response to querying the target rewritten answer information, provide the target rewritten answer information to the user.

[0007] Thirdly, embodiments of this disclosure provide an electronic device 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 implement a method for answering a question as described in any implementation of the first aspect.

[0008] Fourthly, embodiments of this disclosure provide a non-transitory computer-readable storage medium storing computer instructions that, when executed by a computer, enable a method for answering a question as described in any implementation of the first aspect.

[0009] Fifthly, embodiments of this disclosure provide a computer program product including a computer program that, when executed by a processor, enables a method for answering questions as described in any implementation of the first aspect.

[0010] The method, apparatus, electronic device, computer-readable storage medium, and computer program product for answering questions provided in this disclosure, in response to receiving question information sent by a user, determine the user's target user category based on the user's user profile; query target rewritten answer information corresponding to the question information and target user category from a target knowledge base, wherein the target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; and in response to finding the target rewritten answer information, provide the target rewritten answer information to the user.

[0011] Based on this approach, this disclosure can improve the readability of answer information and avoid reading risks for users due to inappropriate wording or other reasons.

[0012] 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

[0013] Other features, objects, and advantages of this disclosure will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is an exemplary system architecture to which this disclosure can be applied; Figure 2 A flowchart illustrating a question-and-answer process provided in this embodiment of the disclosure; Figure 3 A flowchart illustrating another question-answering process provided in this disclosure embodiment; Figure 4 A flowchart illustrating another question-answering process provided in this disclosure embodiment; Figure 5 A flowchart illustrating the process of answering questions in an application scenario provided by an embodiment of this disclosure; Figure 6 A structural block diagram of a question-answering device provided in an embodiment of this disclosure; Figure 7 This is a schematic diagram of the structure of an electronic device suitable for performing a question-answering method, provided as an embodiment of the present disclosure. Detailed Implementation

[0014] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding; these 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. It should be noted that, unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0015] Furthermore, the acquisition, storage, use, processing, transportation, provision, and disclosure of user personal information (such as issue information, context information, directly obtained user profiles, or materials used to generate user profiles, etc.) involved in the technical solutions disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0016] Figure 1 An exemplary system architecture 100 is shown, illustrating embodiments of methods, apparatuses, electronic devices, and computer-readable storage media for answering questions using this disclosure.

[0017] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0018] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various applications for enabling information communication between the terminal devices 101, 102, and 103 and server 105 can be installed. These applications include knowledge search applications, intelligent assistant applications, and instant messaging applications.

[0019] Terminal devices 101, 102, and 103 and server 105 can be either hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with displays, including but not limited to smartphones, tablets, laptops, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the aforementioned electronic devices, and can be implemented as multiple software programs or software modules, or as a single software program or software module; no specific limitation is made here. When server 105 is hardware, it can be implemented as a distributed server cluster composed of multiple servers, or as a single server. When server 105 is software, it can be implemented as multiple software programs or software modules, or as a single software program or software module; no specific limitation is made here.

[0020] Server 105 can provide various services through its built-in applications. Taking a knowledge search application that provides online question-and-answer services as an example, when running this application, server 105 can achieve the following: If it receives a question from a user via network 104 from terminal devices 101, 102, and 103, it can respond by determining the user's target user category based on the user's profile. Then, it queries the target knowledge base for the target rewritten answer information corresponding to the question and the target user category. The target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information. Finally, if server 105 finds the target rewritten answer information, it can respond by providing the target rewritten answer information to the user (e.g., terminal devices 101, 102, and 103).

[0021] Typically, maintaining a target knowledge base and performing queries within it may require significant computing resources and capabilities. Therefore, the question-answering methods provided in the subsequent embodiments of this disclosure are generally executed by a server 105 with strong computing power and abundant resources. Correspondingly, the question-answering device is also typically located within the server 105. However, it should also be noted that when terminal devices 101, 102, and 103 also possess sufficient computing power and resources, they can also perform the aforementioned calculations performed by the server 105 through their installed knowledge search applications, thereby outputting the same results as the server 105. Especially when multiple terminal devices with different computing capabilities exist simultaneously, but the knowledge search application determines that the terminal device it is using has strong computing power and abundant remaining computing resources, it can allow the terminal device to perform the aforementioned calculations, thereby appropriately reducing the computing load on the server 105. Consequently, the question-answering device can also be located within the terminal devices 101, 102, and 103. In this case, the exemplary system architecture 100 may also exclude the server 105 and the network 104.

[0022] 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.

[0023] Please refer to Figure 2 , Figure 2 A flowchart of a question-and-answer process provided for an embodiment of this disclosure, including process 200.

[0024] Process 200 specifically includes the following steps: Step 201: In response to receiving the question information sent by the user, determine the target user category based on the user's user profile; In embodiments of this disclosure, the entity executing the question-and-answer method (e.g.) Figure 1 The server 105 shown can be interacted with by users via network 104 using terminal devices 101, 102, and 103 as described above. For example, users can provide question information to the executing entity through terminal devices 101, 102, and 103 to seek corresponding answer information.

[0025] In such a situation, if the executing entity receives a problem message from a user, it can respond by determining the user's target user category based on the user's profile.

[0026] Typically, this user profile can be determined based on user configuration information provided by the user in advance (or in this instance). For example, the user can instruct the executing entity to generate this user profile by providing user configuration information such as language preferences, age group, and level of knowledge in technical and knowledge domains.

[0027] In some embodiments, users may also provide the user profile to the executing entity after generating the desired user profile locally on their terminal device.

[0028] Alternatively, users can generate a user profile locally on their terminal device and store it on a third-party device, based on their actual needs. The user can then provide this user profile to the executing entity by instructing the executing entity to send a retrieval request to that third-party device.

[0029] User profiles can be used to characterize user traits, allowing the implementing entity to determine which pre-defined user category a user belongs to based on these traits. For ease of understanding, the user category to which a user belongs can be called the "target user category."

[0030] For example, at least two user categories can be pre-defined based on age groups. Accordingly, the format and content of the presented answer information can be adjusted based on the processing strategies corresponding to the user categories, so that users can read the provided answer information according to their reading habits, while avoiding providing expressions that may not be easily understood by users of the corresponding age group (e.g., avoiding the use of obscure characters, words, etc. that are difficult for users of that age group to understand).

[0031] For example, multiple user categories can be pre-defined based on language style standards, with at least two categories. This allows for the use of language styles that users within a given category might prefer or favor in providing and expressing answer information once the target user category is determined.

[0032] For example, users can be categorized based on whether they belong to a specific field. For instance, users could be categorized based on whether they belong to the finance or technology sectors. Accordingly, if a user belongs to a specific field (or has sufficient knowledge of that field), the answer provided can directly use specialized and well-known terms from that field to construct the final answer information. For users who do not belong to a specific field, the use of specialized and well-known terms can be avoided as much as possible, instead using easier-to-understand words or substitutions for these specialized and well-known terms (or, providing explanations of the specialized and well-known terms) to construct the answer information.

[0033] In some embodiments, the user profile can also be summarized and generated by the executing entity based on historical information (e.g., related question information) from the user's question and inquiry interactions. For example, after receiving a question from a user, the executing entity can, in response, read the related question information. For instance, the executing entity can read question information previously provided by the user in the same round as related question information. Then, the obtained related question information is combined with the question information used in this instance to determine the user profile.

[0034] In some embodiments, the executing entity may filter historical problem information based on semantic relevance and problem content relevance in order to filter out related problem information that is related to the problem information, so as to avoid introducing irrelevant historical problem information.

[0035] Then, as discussed above, the executing entity can combine the problem information and related problem information, and generate a user profile based on the problem information and related problem information. For example, the executing entity can summarize the user's word usage habits, language forms, etc., based on the combination of problem information and related problem information, in order to determine and generate the aforementioned user profile.

[0036] Accordingly, in this step, the executing entity can use the user profile generated based on the question information and related question information to determine the target user category. This allows the executing entity to summarize the user's "user profile" based on the questions the user has provided (especially recent and current rounds) even without the user providing additional configuration information or a user profile, and to determine the target user category accordingly. This not only avoids requiring users to provide information beyond the question information, reducing user costs, but also better protects the user's personal information security and makes the "user profile" more focused on the user's inquiry scenario, improving the relevance of the "features" described in the "user profile" to the question-and-answer scenario.

[0037] Step 202: Query the target knowledge base for target rewritten answer information corresponding to the question information and target user category; In the embodiments of this disclosure, after determining the target user category based on step 201 above, the executing entity can perform a query in the target knowledge base based on the question information and the target user category to determine whether the target rewritten answer information can be found.

[0038] The target knowledge base can be a pre-built library for providing answer information (e.g., a library pre-built offline that can be directly accessed by the executing entity to support online question answering). This target knowledge base maintains preset question information, corresponding standard answer information, and rewritten answer information corresponding to different user categories, obtained by rewriting the standard answer information. Thus, through this pre-maintained offline database, the online question-and-answer service (system) can efficiently and securely provide rewritten standard answer information (i.e., rewritten answer information) corresponding to the question information in a combined online and offline manner.

[0039] For example, pre-defined question information and corresponding standard answer information can be maintained based on publicly available expert knowledge and publicly available datasets.

[0040] In some embodiments, after collecting the above information, the collected data can be organized and cleaned to form an initial knowledge base. Then, this initial database is filtered and updated using a predetermined security policy to remove pre-defined question and standard answer information that does not comply with the security policy. For example, pre-defined question and standard answer information that does not comply with policies and regulations can be removed, resulting in pre-defined question and standard answer information used to construct the target knowledge base. This ensures the information quality and data security of the target knowledge base.

[0041] Then, within this target knowledge base, rewritten answer information is stored and maintained corresponding to the standard answer information. Specifically, the rewritten answer information can correspond to the user categories mentioned above. For example, the rewritten answer information corresponding to a user category can be obtained by rewriting the standard answer information based on the category corpus corresponding to the user category after acquiring the standard answer information.

[0042] Typically, categorized corpora can be pre-built by identifying the characters, words, and sentence structures that users prefer to use within a user category. Using this categorized corpus, by replacing characters and words in the standard answer information and reorganizing sentence structures, rewritten answer information can be obtained that, while having a different textual form than the standard answer information, has the same (or, meets the similarity requirements) semantic meaning within the corresponding user category of the categorized corpus.

[0043] Accordingly, the implementing entity can utilize the rewritten answer information in the target knowledge base to provide users with rewritten answer information that is easier for users to understand and has a better reading experience, including language style and text content.

[0044] It should be understood that the user classifications involved in rewriting answer information are usually consistent with the classification system and criteria used by the executing entity when determining the target user classification. This allows the executing entity to query the rewritten answer information corresponding to the target user classification after determining the target user classification.

[0045] In some scenarios, when rewriting answer information, the same standard answer may not have corresponding rewritten answer information under all user categories. For example, in some specific user categories, the substantive content of the standard answer information may not be suitable or intended for users in certain user categories. In such cases, rewritten answer information corresponding to the standard answer information in those user categories may not be generated. This is to avoid mistakenly providing certain "standard answer information" to users in user categories with different needs.

[0046] In this step, after determining the user's problem information, the executing entity can match it with preset problem information in order to attempt to summarize and correspond it to a specific preset problem information.

[0047] For example, an embedding service can be set up so that after receiving the problem information, the executing entity can convert the problem information into a corresponding embedding vector, and then use the embedding vector matching method to summarize and match the specific preset problem information mentioned above (for example, based on the similarity comparison between vectors, and select the preset problem information with the highest similarity that is greater than or equal to a predetermined similarity threshold).

[0048] If the specific preset question information can be queried and mapped, the executing entity can use the target user category to select the rewritten answer information of the standard answer information corresponding to the specific preset question information under that target user category as the target rewritten answer information. That is, the executing entity can use the rewritten answer information corresponding to the target user category as the target rewritten answer information.

[0049] Next, if the executing entity can find the target rewritten answer information in the target knowledge base, it can respond to this and continue to execute step 203.

[0050] Step 203: Provide the user with the target rewritten answer information.

[0051] In embodiments of this disclosure, if the executing entity can query the target rewritten answer information based on step 202 above, it can provide the target rewritten answer information to the user and the terminal device used by the user based on a predetermined communication link (e.g., network 104).

[0052] In some embodiments, during the execution of this step, the executing entity may first determine the relevance between the target rewritten answer information and the question information. For example, the executing entity may determine the content pointed to by the question information and whether the semantics are asking about the content in the target rewritten answer information based on semantic analysis, thereby determining the relevance between the two.

[0053] Accordingly, if the relevance is greater than or equal to a relevance threshold (for example, this relevance threshold may be pre-determined based on criteria such as whether the two are credibly strongly related), the executing entity may respond by choosing to provide the user with the target rewritten answer information. This ensures the quality of the response.

[0054] In some optional implementations of this embodiment, if the aforementioned relevance is less than the relevance threshold, the executing entity may respond by selecting the target classification corpus corresponding to the target user classification as a reference, calling the large language model to rewrite the target rewritten answer information, and obtaining the rewritten answer information.

[0055] Large Language Model (LLM) is an artificial intelligence model designed to understand and generate human language. Based on its understanding, LLM can perform processing operations to obtain corresponding results. For example, in cases where rewriting target answer information to obtain further rewritten answer information, LLM can use prompts, such as "Referring to the target classification corpus, rewrite the target answer information to obtain further rewritten answer information with a higher relevance to the question information," to instruct LLM to perform the rewriting action and achieve the desired purpose. (For example, the prompt can be a pre-configured template, allowing the executing entity to instruct LLM to perform specific actions after filling in the specific target user classification, question information, and target rewritten answer information.)

[0056] LLMs can be trained on large amounts of text data and perform a wide range of tasks, including text summarization, translation, sentiment analysis, and more. A key characteristic of LLMs is their large scale; they typically include a large number of parameters to help them learn complex patterns in language data. These models are often based on deep learning architectures, such as transformers, which contributes to their superior performance across various NLP tasks.

[0057] Furthermore, for LLM, the "guide word" can be omitted through default configuration. For example, regarding the aforementioned objective of "rewriting the target rewritten answer information based on the target category corpus corresponding to the target user category, and then rewriting the answer information again," LLM, based on its default configuration, naturally understands the need to refer to the provided target category corpus to rewrite the target rewritten answer information according to the goal of improving the relevance to the question information, thus obtaining the rewritten answer information again. Therefore, through default configuration, the generative model can stably and purposefully complete the rewriting task. This allows generative large language models to perform rewriting tasks more efficiently and with higher quality.

[0058] Accordingly, after obtaining the rewritten answer information, the executing entity can choose to use the rewritten answer information as a replacement for the target rewritten answer information and provide it to the user. This avoids reducing the quality of the response due to improper rewriting of content in the target knowledge base.

[0059] It should be understood that, when rewriting answer information, the executing entity can also determine whether to provide it directly to the user or continue rewriting it by comparing the relevance, which will not be repeated here.

[0060] The question-answering method provided in this disclosure, in response to receiving a question from a user, determines the user's target user category based on the user's user profile; queries a target knowledge base for target rewritten answer information corresponding to the question information and the target user category. The target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; in response to finding the target rewritten answer information, the method provides the target rewritten answer information to the user. This improves the readability of the answer information and avoids reading risks for users due to inappropriate wording or other reasons.

[0061] In some embodiments, for question information, it is also possible to pre-classify question types, enabling the implementing entity to avoid answering or responding to certain inappropriate or potentially risky questions. For example, question types that are allowed to be answered and allowed to flow into subsequent processing steps can be pre-classified as "target question types".

[0062] Then, in such a case, during the execution of step 201 above, the executing entity may, as an alternative, after receiving the question information sent by the user, first generate the credibility of the question information belonging to the target question type (for ease of description, the credibility here can be described as the first credibility).

[0063] For example, the implementing entity can determine the first degree of confidence that the problem information can be classified into the target problem type based on the semantic content included in the problem information. For example, the probability that the problem information can be summarized and classified into the target problem type can be used as the "(first) confidence".

[0064] Next, if the first credibility is greater than or equal to the first credibility threshold (for example, the first credibility threshold can be determined based on the criteria that can be considered to reliably classify the problem information as the target problem type), the executing entity can respond by selecting to determine the target user category based on the user's user profile and proceeding with subsequent steps.

[0065] In some embodiments, the aforementioned first confidence level may not be greater than or equal to a first confidence threshold; that is, it may not meet the criteria for reliably classifying the information as the "target problem type," but the first confidence level also may not reach the criteria for reliably determining that the problem information does not belong to the target problem type. For example, the first confidence level may not be less than a second confidence threshold. The second confidence threshold can correspond to the criteria for reliably considering that the problem information cannot be classified as the target problem type. Accordingly, the value of the second confidence threshold is less than the first confidence threshold.

[0066] In such a situation, the implementing entity may choose to try to improve credibility by rewriting the question information, thereby increasing the likelihood that the question information can be answered.

[0067] For easier understanding, please refer to the following: Figure 3 Please refer to this. Figure 3 , Figure 3 A flowchart of another question-and-answer process provided for an embodiment of this disclosure, including process 300.

[0068] Process 300 specifically includes the following steps: Step 301: In response to receiving the question information sent by the user, generate the first confidence level that the question information belongs to the target question type; Specifically, as discussed above, if the implementing entity receives a problem message sent by the user, it can respond by first generating a first degree of confidence that the problem message belongs to the target problem type.

[0069] Next, if the first confidence level is greater than or equal to the first confidence level threshold, the executing entity can choose to continue executing step 302.

[0070] Step 302: Based on the user profile, determine the target user category; Step 303: Query the target rewritten answer information corresponding to the question information and target user category from the target knowledge base; Step 304: Provide the user with the target rewritten answer information.

[0071] The implementation process of steps 302-304 above is as follows: Figure 2 The content discussed in steps 201-203 is similar. For the same parts, please refer to the corresponding parts of the previous embodiment. They will not be repeated here.

[0072] Next, if, as discussed above, in some embodiments the first confidence level is less than a first confidence threshold but greater than or equal to a second confidence threshold—that is, the first confidence level is greater than or equal to the second confidence threshold but less than the first confidence threshold—then the executing entity may choose to rewrite the problem information.

[0073] In such a case, process 300 may also include step 305, in which the subject being executed may choose to execute if the first confidence level is greater than or equal to the second confidence level threshold and less than the first confidence level threshold.

[0074] Step 305: Using the associated problem information as a reference, call the large language model to rewrite the problem information and generate the rewritten problem information; Specifically, the executing entity can obtain the question information, as well as the related question information provided by the user in the historical dialogue and questioning process (for example, the related question information obtained by filtering historical question information based on semantic relevance and question content relevance as discussed above), and instruct and utilize the related question information of the LLM reference question information to rewrite the question information by calling the LLM.

[0075] For example, prompts for such actions can be supplemented with related question information as material and reference, thus clarifying and defining the intended message. Therefore, the implementing entity can use LLM to supplement the question information with related question information, further clarifying the user's intended question through rewriting, thereby enhancing credibility.

[0076] In some embodiments, the LLM here can be the same as the LLM discussed above, so that the executing entity can perform different actions through the same LLM to maintain consistency in the processing.

[0077] Step 306: Generate a second level of credibility for the rewritten problem information to belong to the target problem type; Specifically, after obtaining the rewritten problem information from the LLM rewritten problem information based on step 305 above, the executing entity can similarly generate a second confidence level that the rewritten problem information belongs to the target problem type.

[0078] Next, if the second confidence level is greater than or equal to the first confidence level threshold, the executing entity can similarly continue to execute step 307.

[0079] Step 307: Based on the user profile, determine the target user category; Step 308: Query the target rewrite answer information corresponding to the rewrite question information and target user category from the target knowledge base; Specifically, in step 308, since the rewritten question information is the only one that can be reliably determined and understood as belonging to the target question type compared to the question information, the executing entity can further select the rewritten question information as a substitute for the question information when querying the target rewritten answer information using the target knowledge base.

[0080] Therefore, while ensuring the security of the answer, it is possible to use rewritten question information with richer and more accurate semantic content to perform queries, thus enabling more accurate retrieval of "target rewritten answer information".

[0081] Next, if the "target rewrite answer information" can be found, the executing entity can further choose to jump back to step 304 to provide the target rewrite answer information to the user.

[0082] In some optional implementations of this embodiment, as discussed above, if the first confidence level is less than the second confidence level threshold mentioned above, that is, if it can be reliably determined that the target problem does not belong to the target problem type, the executing entity can provide the user with pre-configured feedback prompts.

[0083] Accordingly, the above process 300 may also include step 309, which may be performed after step 301, if the first confidence level is less than the second confidence level threshold, the subject to be executed may choose to execute.

[0084] Step 309: Provide the user with pre-configured feedback prompts.

[0085] Specifically, this feedback message can be used to indicate that a question cannot be answered. For example, in text form, the feedback message could be, "Since your question is not within the scope of answers, we cannot provide you with a corresponding reply." This avoids answering questions that shouldn't be answered, ensuring the security of the question-and-answer process. Furthermore, it allows for timely synchronization and feedback to the user, improving the user's interactive experience.

[0086] In some embodiments, if the rewritten problem information still fails to meet the first credibility threshold (i.e., greater than or equal to the first credibility threshold), the executing entity may similarly choose to provide the user with pre-configured feedback prompts, which will not be repeated here.

[0087] In some embodiments, for the case of rewriting the answer information as described above, the executing entity may also similarly choose to provide the pre-configured feedback prompt information if the answer information still cannot be rewritten after another round of rewriting, or if the rewritten answer information obtained after the target round of rewriting still cannot be greater than or equal to the relevance threshold. (In such cases, the text form of the feedback prompt information may vary depending on the scenario) to avoid providing answer information with low relevance and causing user confusion.

[0088] In some embodiments, if the executing entity fails to find the target rewritten answer information corresponding to the question information and the target user category in the target knowledge base in step 202, the executing entity may also choose to search for online answer information corresponding to the question information online to enhance its ability to answer user questions and improve the user experience.

[0089] For easier understanding, you can also refer to Figure 4 . Figure 4 A flowchart of another question-answering process provided for embodiments of this disclosure, including process 400.

[0090] Process 400 specifically includes the following steps: Step 401: In response to receiving the question information sent by the user, determine the target user category based on the user's user profile; Step 402: Query the target rewritten answer information corresponding to the question information and target user category from the target knowledge base; The implementation process of steps 401-402 above is as follows: Figure 2 The content discussed in steps 201-202 is similar. For the same parts, please refer to the corresponding parts of the previous embodiment. They will not be repeated here.

[0091] Next, if the executing entity fails to find the target rewritten answer information corresponding to the question information and target user category in the target knowledge base, it can respond to this and continue to select to execute step 403.

[0092] Step 403: Search online for the answer information corresponding to the question; Specifically, if the executing entity cannot find the target rewritten answer information corresponding to the question information and target user category in the target knowledge base, the executing entity can choose to search for online answer information corresponding to the question information through online search. For example, the executing entity can use online resources such as keyword matching to find online answer information corresponding to the question information from public knowledge bases and data sources.

[0093] If the executing entity can find the target online answer information for the question, it can further determine whether the target online answer information meets the requirements of the target user classification. For example, as discussed above, the executing entity can use the target user classification corpus to determine whether the online answer information meets the requirements in terms of word choice, sentence structure, etc.

[0094] Next, if the target online answer information found through online search meets the requirements of the target user's classification, the executing entity can continue to step 404 to provide the online answer information as the final response result to the user, similar to what was discussed above.

[0095] Step 404: Provide online answer information to users.

[0096] In some optional implementations of this embodiment, if the searched target online answer information fails to meet the requirements of the target user classification, the executing entity can also adjust it by rewriting.

[0097] Accordingly, process 400 may also include step 405, which may be executed by the executing entity after step 403 above, if the target online answer information found online fails to meet the target user classification requirements (i.e., in such a case, the executing entity will choose to execute step 405 instead of step 404).

[0098] Step 405: Using the target category corpus corresponding to the target user category as a reference, call the large language model to rewrite the target online answer information to obtain the target rewritten online answer information; Specifically, in such a case, the executing entity can choose to invoke LLM to rewrite the target online answer information with reference to the target category corpus corresponding to the target user category, so as to obtain target rewritten online answer information that meets the requirements of the target user category.

[0099] Similarly, the LLM can be the same LLM that performs the other steps discussed above, so that the same LLM can be used to process question information and answer information in different situations.

[0100] Accordingly, after completing the rewriting and obtaining the target rewritten online answer information, the executing entity can continue to execute step 406 to provide the target rewritten online answer information to the user.

[0101] Step 406: Provide the user with the target rewritten online answer information.

[0102] This allows the implementing entity to adjust the final provided answer information by rewriting it in scenarios where online search provides answer information (for example, the final provided answer information is a rewritten version of the target online answer information) in order to ensure and improve the readability of the answer information and avoid reading risks or interference for users due to inappropriate wording or other reasons.

[0103] In some optional implementations of this embodiment, if online answer information is used, the executing entity can also, after providing (whether it is online answer information or online answer information that is rewritten), refer to the classification corpus corresponding to each user category and call the above-mentioned LLM (or other LLMs) to rewrite the online answer information and generate supplementary rewritten answer information corresponding to the question information.

[0104] Then, the question information is used as the aforementioned preset question information, and the online answer information is used as the standard answer information corresponding to the preset question information. These, along with the supplemented and rewritten answer information, are maintained in the target knowledge base. Accordingly, in this target knowledge base, for the online answer information used as the standard answer information, corresponding supplemented and rewritten answer information for each user category and the corresponding rewritten answer information for that user category can be maintained.

[0105] Therefore, this approach allows the implementing entity to update the target knowledge base in a feedback-based manner, based on question information and corresponding online answer information generated from user interactions, even when the target knowledge base lacks corresponding preset question information and standard answer information. For example, new preset question information, along with the corresponding standard answer information and its rewritten answers under various user categories, can be stored in the target knowledge base. This ensures the target knowledge base can be continuously updated, improving its knowledge quality.

[0106] To enhance understanding, this disclosure also provides a specific implementation scheme based on a particular application scenario. Please refer to the following for details. Figure 5 . Figure 5 This is a flowchart illustrating a problem-solving process implemented in an application scenario according to an embodiment of the present disclosure, including process 500.

[0107] For ease of understanding, an example is given. Figure 1The architecture 100 shown illustrates process 500. For example, in process 500, it can be... Figure 1 The server 105 shown acts as the "executor" to provide answers to questions from the terminal device 101 used by the user (not shown in the figure).

[0108] In process 500, the user can use terminal device 101 to provide problem information 511 to server 105 by executing S501, in order to request server 105 to provide a problem response (i.e., provide answer information corresponding to problem information 511).

[0109] Accordingly, upon receiving the problem information 511, the server 105 can respond by executing S502 to determine the user's target user category 522 based on the user's user profile 521.

[0110] Next, server 105 can continue to execute S503 to query the target rewritten answer information corresponding to question information 510 and target user category 522 from the target knowledge base 530.

[0111] For example, if server 105 can query target rewritten answer information 531 in target knowledge base 530 by executing S503, then in such a case, server 105 can continue to execute S504 to determine the relevance between target rewritten answer information 531 and question information 511.

[0112] For example, the correlation is less than the correlation threshold discussed above.

[0113] In this case, server 105 can continue in step S505, using the target category corpus 523 corresponding to target user category 522 as a reference, and call the large language model 540 to rewrite the target rewritten answer information 531 to obtain the rewritten answer information 541.

[0114] Then, after obtaining the rewritten answer information 541, the server 105 can execute S506 to provide the rewritten answer information 541 as the final "answer information" corresponding to the question information 511 to the terminal device 101 for the user to use and complete the process of answering the user's question.

[0115] Further reference Figure 6 As an implementation of the methods shown in the above figures, this disclosure provides an embodiment of a question-answering apparatus, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0116] like Figure 6As shown, the question-answering device 600 of this embodiment may include: a user classification determination unit 601, a knowledge base query unit 602, and a rewritten answer providing unit 603. The user classification determination unit 601 is configured to determine the user's target user classification based on the user's user profile in response to receiving a question information sent by a user. The knowledge base query unit 602 is configured to query target rewritten answer information corresponding to the question information and the target user classification from the target knowledge base. The target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user classifications obtained by rewriting the standard answer information. The rewritten answer providing unit 603 is configured to provide the target rewritten answer information to the user in response to finding the target rewritten answer information.

[0117] In this embodiment, the specific processing and technical effects of the user classification determination unit 601, knowledge base query unit 602, and answer rewriting and provision unit 603 in the question-answering device 600 can be referred to respectively. Figure 2 The relevant descriptions of steps 201-203 in the corresponding embodiments will not be repeated here.

[0118] In some optional implementations of this embodiment, the user classification determination unit 601 includes: a first context information reading subunit, configured to read associated question information in response to receiving question information sent by a user; a user profile generation subunit, configured to generate a user profile based on the question information and associated question information; and a first user classification determination subunit, configured to determine the target user classification of the user based on the user profile.

[0119] In some optional implementations of this embodiment, the user classification determination unit 601 includes: a first credibility generation subunit, configured to generate a first credibility that the question information belongs to the target question type in response to receiving question information sent by the user; and a second user classification determination subunit, configured to determine the target user classification of the user based on the user profile in response to the first credibility being greater than or equal to the first credibility threshold.

[0120] In some optional implementations of this embodiment, the user classification determination unit 601 may further include: a second context information reading subunit, configured to read associated question information of the question information in response to a first credibility being greater than or equal to a second credibility threshold and less than the first credibility threshold, wherein the value of the second credibility threshold is less than the value of the first credibility threshold; a question rewriting subunit, configured to call a large language model to rewrite the question information and generate rewritten question information with reference to the associated question information of the question information; a second credibility generation subunit, configured to generate a second credibility that the rewritten question information belongs to the target question type; a third user classification determination subunit, configured to determine the target user classification of the user based on the user profile in response to a second credibility of the rewritten question information being greater than or equal to the first credibility threshold; and a knowledge base query unit 602 further configured to query target rewritten answer information corresponding to the rewritten question information and the target user classification from the target knowledge base.

[0121] In some optional implementations of this embodiment, the device 600 further includes: a prompting information feedback unit, configured to provide a pre-configured feedback prompting information to the user in response to the credibility being less than a second credibility threshold, wherein the feedback prompting information is used to indicate that the question information cannot be answered.

[0122] In some optional implementations of this embodiment, the target knowledge base is maintained in the following way: obtaining standard answer information; rewriting the standard answer information based on the classification corpus corresponding to the user classification to obtain rewritten answer information corresponding to the user classification.

[0123] In some optional implementations of this embodiment, the rewritten answer providing unit 603 includes: a relevance determination subunit, configured to determine the relevance between the target rewritten answer information and the question information; and a rewritten answer providing subunit, configured to provide the target rewritten answer information to the user in response to the relevance being greater than or equal to a relevance threshold.

[0124] In some optional implementations of this embodiment, the answer rewriting unit 603 may further include: an answer rewriting subunit, configured to respond to a relevance degree less than a relevance degree threshold, using the target classification corpus corresponding to the target user classification as a reference, call a large language model to rewrite the target rewritten answer information to obtain the rewritten answer information; and a rewritten answer providing subunit, configured to provide the rewritten answer information to the user.

[0125] In some optional implementations of this embodiment, the apparatus 600 further includes: an online search unit configured to, in response to the failure to find target rewritten answer information corresponding to the question information and the target user classification from the target knowledge base, search online for online answer information corresponding to the question information; and an online search answer providing unit configured to, in response to the online search for target online answer information that meets the requirements of the target user classification, provide online answer information to the user.

[0126] In some optional implementations of this embodiment, the device 600 further includes: an online answer rewriting unit, configured to, in response to the searched target online answer information failing to meet the requirements of the target user classification, refer to the target classification corpus corresponding to the target user classification, call a large language model to rewrite the target online answer information, and obtain the target rewritten online answer information; and a rewritten online answer providing unit, configured to provide the target rewritten online answer information to the user.

[0127] In some optional implementations of this embodiment, the device 600 further includes: an input / rewrite answer generation unit, configured to use the classification corpus corresponding to each user category as a reference, call a large language model to rewrite the online answer information, and generate input / rewrite answer information corresponding to the question information; and a knowledge base maintenance unit, configured to maintain the question information as preset question information, the online answer information as standard answer information corresponding to the preset question information, and the input / rewrite answer information together in the target knowledge base.

[0128] This embodiment exists as a device embodiment corresponding to the above method embodiment. The question answering device provided in this embodiment can improve the readability of the answer information and avoid the risk of the answer information being read by the user due to improper wording or other reasons.

[0129] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0130] Figure 7 A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 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.

[0131] like Figure 7As shown, device 700 includes a computing unit 701, which can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) 702 or a computer program loaded into random access memory (RAM) 703 from storage unit 708. The RAM 703 may also store various programs and data required for the operation of device 700. The computing unit 701, ROM 702, and RAM 703 are interconnected via bus 704. Input / output (I / O) interface 705 is also connected to bus 704.

[0132] Multiple components in device 700 are connected to I / O interface 705, including: input unit 706, such as keyboard, mouse, etc.; output unit 707, such as various types of monitors, speakers, etc.; storage unit 708, such as disk, optical disk, etc.; and communication unit 709, such as network card, modem, wireless transceiver, etc. Communication unit 709 allows device 700 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0133] The computing unit 701 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 701 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 701 performs the various methods and processes described above, such as the question-and-answer method. For example, in some embodiments, the question-and-answer method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 708. In some embodiments, part or all of the computer program may be loaded and / or installed on device 700 via ROM 702 and / or communication unit 709. When the computer program is loaded into RAM 703 and executed by the computing unit 701, one or more steps of the question-and-answer method described above may be performed. Alternatively, in other embodiments, the computing unit 701 may be configured to perform the question-and-answer method by any other suitable means (e.g., by means of firmware).

[0134] 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.

[0135] 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.

[0136] 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.

[0137] 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).

[0138] 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.

[0139] Computer systems can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, also known as cloud computing servers or cloud hosts, which are hosting products within the cloud computing service ecosystem to address the management difficulties and weak business scalability inherent in traditional physical hosts and Virtual Private Servers (VPS) services. Servers can also be categorized as distributed system servers or servers incorporating blockchain technology.

[0140] According to the technical solution of this disclosure, in response to receiving a question from a user, a target user category is determined based on the user's user profile; target rewritten answer information corresponding to the question and target user category is queried from a target knowledge base. The target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; in response to finding the target rewritten answer information, the target rewritten answer information is provided to the user. This improves the readability of the answer information and avoids reading risks for users due to inappropriate wording or other reasons.

[0141] 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 provided in this disclosure can be achieved, and this is not limited herein.

[0142] 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. A method for answering a question, comprising: In response to receiving a question from a user, the system determines the user's target user category based on the user's user profile. Query the target rewritten answer information corresponding to the question information and the target user category from the target knowledge base, wherein the target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; In response to the retrieval of the target rewritten answer information, the target rewritten answer information is provided to the user.

2. The method according to claim 1, wherein, The step of responding to receiving a question from a user and determining the user's target user category based on the user's user profile includes: In response to receiving a problem message sent by a user, read the associated problem information of the problem message; A user profile of the user is generated based on the question information and the associated question information; Based on the user profile of the user, the target user category of the user is determined.

3. The method according to claim 1, wherein, The step of responding to receiving a question from a user and determining the user's target user category based on the user's user profile includes: In response to receiving a question message from a user, a first confidence level is generated indicating that the question message belongs to the target question type; In response to the first confidence level being greater than or equal to the first confidence level threshold, the target user category of the user is determined based on the user profile of the user.

4. The method according to claim 3, further comprising: In response to the first confidence level being greater than or equal to the second confidence level threshold, and less than the first confidence level threshold, the associated problem information of the problem information is read, wherein the value of the second confidence level threshold is less than the value of the first confidence level threshold; Using the associated problem information as a reference, the large language model is invoked to rewrite the problem information and generate the rewritten problem information; Generate a second level of confidence that the rewritten problem information belongs to the target problem type; In response to the second credibility of the rewritten question information being greater than or equal to the first credibility threshold, the target user category of the user is determined based on the user's user profile; and The step of querying the target rewritten answer information corresponding to the question information and the target user category from the target knowledge base includes: Retrieve target rewrite answer information from the target knowledge base that corresponds to the rewrite question information and the target user category.

5. The method according to claim 4, further comprising: In response to the first confidence level being less than the second confidence level threshold, a pre-configured feedback prompt is provided to the user, wherein the feedback prompt is used to indicate that the question information cannot be answered.

6. The method according to claim 1, wherein, The target knowledge base is maintained in the following manner: Obtain standard answer information; Based on the classification corpus corresponding to the user classification, the standard answer information is rewritten to obtain the rewritten answer information corresponding to the user classification.

7. The method according to claim 1, wherein, Providing the user with the target rewritten answer information includes: Determine the correlation between the target rewritten answer information and the question information; In response to the relevance being greater than or equal to the relevance threshold, the target rewritten answer information is provided to the user.

8. The method according to claim 7, further comprising: In response to the correlation degree being less than the correlation degree threshold, the target classification corpus corresponding to the target user classification is used as a reference to call the large language model to rewrite the target rewritten answer information to obtain the rewritten answer information; The rewritten answer information is provided to the user.

9. The method according to any one of claims 1-8, further comprising: In response to the failure to find target rewritten answer information corresponding to the question information and the target user category from the target knowledge base, an online search is conducted to find online answer information corresponding to the question information. In response to the online answer information found through online search meeting the requirements of the target user classification, the online answer information is provided to the user.

10. The method of claim 9, further comprising: If the searched target online answer information fails to meet the requirements of the target user classification, the target classification corpus corresponding to the target user classification is used as a reference to rewrite the target online answer information by calling a large language model to obtain the target rewritten online answer information. Provide the user with the target rewritten online answer information.

11. The method of claim 10, further comprising: Using the respective category corpus of each user category as a reference, the large language model is invoked to rewrite the online answer information, generating supplementary and rewritten answer information corresponding to the question information; The question information is used as the preset question information, and the online answer information is used as the standard answer information corresponding to the preset question information. Together with the supplemented and rewritten answer information, they are maintained in the target knowledge base.

12. A device for answering a question, comprising: The user classification determination unit is configured to, in response to receiving a question information sent by a user, determine the target user classification of the user based on the user profile; The knowledge base query unit is configured to query target rewritten answer information corresponding to the question information and the target user category from the target knowledge base, wherein the target knowledge base maintains preset question information, standard answer information corresponding to the preset question information, and rewritten answer information corresponding to different user categories obtained by rewriting the standard answer information; The rewritten answer providing unit is configured to provide the target rewritten answer information to the user in response to a query for the target rewritten answer information.

13. 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 executable by the at least one processor to enable the at least one processor to perform the method of answering the question as described in any one of claims 1-11.

14. A non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform a method for answering a question according to any one of claims 1-11.

15. A computer program product comprising a computer program that, when executed by a processor, implements a method for answering a question according to any one of claims 1-11.

Images