Intelligent question and answer guiding method and device based on reasoning path, computer equipment, readable storage medium and program product

Abstract

The invention relates to an intelligent question and answer guiding method and device based on a reasoning path, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps that a first question input by a user is received and recognized, and a first user intention is obtained; under the condition that the first user intention successfully matches the standard problem and any problem in the reasoning path corresponding to the standard problem, taking the successfully matched problem as a first target problem, and obtaining a next logic chain sub-problem in the successive logic relationship of the first target problem; in response to the first user intention, generating at least two first guide questions, one of which comprises a first target guide question; and when the first target guide question is the last logic chain sub-question in the reasoning path and the user selects the first target guide question, in response to a selection operation of the user, matching and displaying a standard answer. By adopting the method, the accuracy of information recommendation can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an intelligent question-answering guidance method, apparatus, computer equipment, computer-readable storage medium, and computer program product based on a reasoning path. Background Art

[0002] With the development of artificial intelligence, intelligent question answering (QA) systems have been widely used in many fields such as customer service, knowledge management, medical consultation, and e-commerce shopping guides, and automatically answer user questions in a retrieval-based and generative-based manner.

[0003] In traditional technology, the keywords entered by the user are matched with the preset question library to return the most similar answer. Alternatively, search engine technology is used to retrieve relevant documents or paragraphs based on the user's question and extract possible answers from them.

[0004] However, the current AI-based question-answering method does not provide accurate answers to users. Summary of the invention

[0005] Based on this, it is necessary to provide an intelligent question-answering guidance method, device, computer equipment, computer-readable storage medium and computer program product based on reasoning path that can improve the accuracy of information recommendation in response to the above technical problems.

[0006] In a first aspect, the present application provides an intelligent question-answering guidance method based on a reasoning path, the method comprising:

[0007] Receive a first question input by a user, and perform intent recognition according to the first question to obtain a first user intent;

[0008] In the case where the first user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, the successfully matched question is used as the first target question, and the next logical chain question in the sequential logical relationship of the first target question is obtained; wherein the reasoning path includes at least two of the logical chain questions, and each of the logical chain questions has a sequential logical relationship;

[0009] In response to the first user intention, at least two first guiding questions are generated, wherein one of the first guiding questions includes a first target guiding question, and the first target guiding question is generated based on the next logical chain question in the sequential logical relationship of the first target question;

[0010] When the first goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the first goal-guiding question, in response to the user's selection operation, a standard answer is matched and displayed, wherein the standard answer is recommended information corresponding to the standard question and the reasoning path.

[0011] In one embodiment, the method further comprises:

[0012] In the case where the first target guiding question is not the last logical chain sub-question in the reasoning path and the user selects the first target guiding question, in response to the user's selection operation, at least two second guiding questions are generated, wherein one of the second guiding questions includes a second target guiding question, and the second target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target guiding question;

[0013] When the second goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the second goal-guiding question, the standard answer is matched and displayed again in response to the user's selection operation.

[0014] In one embodiment, the method further comprises:

[0015] In a case where the user does not select the first target guiding question, and the user selects other first guiding questions that do not include the first target guiding question or the user inputs a second question, in response to the user's operation, the corresponding answer is matched according to the other first guiding question or the second user intention, and the answer is displayed; the second user intention is determined based on the second question.

[0016] In one embodiment, the method further comprises:

[0017] When the second user intention successfully matches the standard question and any one of the questions in the reasoning path, the successfully matched question is used as the second target question, and the next logical chain question in the sequential logical relationship of the second target question is obtained;

[0018] In response to the second user intention, at least two third guiding questions are generated, wherein one of the third guiding questions includes a third target guiding question, and the third target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the second target question;

[0019] When the third goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the third goal-guiding question, the standard answer is matched and displayed in response to the user's selection operation.

[0020] In one embodiment, the method of generating the reasoning path includes:

[0021] According to the standard question and the standard answer, at least two of the logical chain sub-questions having the sequential logical relationship are generated as an initial reasoning path;

[0022] The initial reasoning path is adjusted according to a preset condition to obtain the reasoning path, wherein the preset condition is set based on business requirements.

[0023] In one embodiment, the updating method of the reasoning path includes:

[0024] Acquiring selection operation data of the user within a preset time period, the selection operation data including other first guidance questions selected by the user, a second user intention, the number of positive evaluations of each of the first guidance questions, and the number of negative evaluations of each of the first guidance questions, the other first guidance questions refer to guidance questions that do not include the first target guidance question; the second user intention is determined based on the second question input by the user;

[0025] The reasoning path is optimized and updated according to the selection operation data to obtain a new reasoning path.

[0026] In a second aspect, the present application further provides an intelligent question-answering guidance device based on an inference path, the device comprising:

[0027] A receiving module, configured to receive a first question input by a user, and perform intent recognition according to the first question to obtain a first user intent;

[0028] A matching acquisition module, configured to, when the first user intention successfully matches a standard question and any question in the reasoning path corresponding to the standard question, use the successfully matched question as the first target question, and acquire the next logical chain question in the sequential logical relationship of the first target question; wherein the reasoning path includes at least two logical chain questions, and each of the logical chain questions has a sequential logical relationship;

[0029] A first response generation module, configured to respond to the first user intention and generate at least two first guiding questions, wherein one of the first guiding questions includes a first target guiding question, and the first target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target question;

[0030] The first response matching module is used to match and display a standard answer in response to the user's selection operation when the first target-guided question is the last logical chain sub-question in the reasoning path and the user selects the first target-guided question, wherein the standard answer is recommended information corresponding to the standard question and the reasoning path.

[0031] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0032] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0033] In a fifth aspect, the present application also provides a computer program product, including a computer program, which implements the steps of the above method when executed by a processor.

[0034] The above-mentioned intelligent question-answering guidance method, device, computer equipment, computer-readable storage medium and computer program product based on reasoning path, by receiving the first question input by the user, first, identify the first user intention corresponding to the first question, and determine whether the first user intention matches the standard question and any question in the reasoning path corresponding to the standard question. If matched, the successfully matched question is used as the first target question, and the corresponding logical chain sub-question and the next logical chain sub-question in the reasoning path where the first target question is located are obtained; secondly, in response to the first user intention, the next logical chain sub-question is used as one of the first guidance questions, that is, the first target guidance question, and is displayed to the user; finally, if the first target guidance question is the last logical chain sub-question in the reasoning path and the user selects the first target guidance question, then in response to the user's selection operation, the standard answer is matched, and the standard answer is the recommended information corresponding to the standard question and the reasoning path, and the recommended information is displayed. That is, between identifying user intention and outputting preset standard answers according to user intention, an inference path is added, and the user's selection operation is assisted. If the first target guiding question generated according to the logical chain sub-questions in the inference path is selected, and other first guiding questions are abandoned, it indicates that the user is interested in the logical chain sub-questions, and according to the user's selection behavior, it is further determined that the questions in the inference path meet the user's needs. If the first target guiding question selected by the user is the last logical chain question in the inference path, the standard answer is recommended to the user. By matching the questions in the inference path according to whether the user's selection matches, that is, combining user behavior and multiple rounds of reasoning to further verify the user's intention, it avoids the situation where the user's intention is inaccurately judged based on the user's input question only once, thereby improving the accuracy of information recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 1 is a flow chart of an intelligent question-answering guidance method based on reasoning paths in one embodiment;

[0037] Figure 2 is a flowchart of an intelligent question-answering guidance method based on reasoning paths in another embodiment;

[0038] Figure 3 is a flowchart of an intelligent question-answering guidance method based on reasoning paths in another embodiment;

[0039] Figure 4 is a flow chart of a method for generating a reasoning path in an embodiment;

[0040] Figure 5 is a flowchart of a method for updating a reasoning path in one embodiment;

[0041] Figure 6 is a structural block diagram of an intelligent question-answering guiding device based on reasoning path in one embodiment;

[0042] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0044] In an exemplary embodiment, Figure 1 As shown, an intelligent question-answering guidance method based on reasoning paths is provided. This embodiment takes the method applied to a terminal as an example, wherein the terminal may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, Internet of Things devices and portable wearable devices, and the Internet of Things devices may be smart speakers, smart TVs, smart air conditioners, smart car-mounted devices, projection devices, etc. Portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The head-mounted devices may be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps S102 to S108. Among them:

[0045] Step S102: receiving a first question input by a user, performing intent recognition based on the first question, and obtaining a first user intent.

[0046] Optionally, the terminal receives a first question input by a user through an interface, and uses natural language processing (Natural Language Processing, NLP) to perform intent recognition on the first question through semantic analysis to obtain a first user intent.

[0047] Furthermore, the terminal performs a word segmentation operation on the first question. For example, if the question input by the user is "What will the weather be like in Beijing tomorrow?", the word segmentations are: ["tomorrow", "Beijing", "of", "weather", "what about", "?"], and each word segment and the contextual relationship of each word segment in the sentence are converted into a vector by means of word embedding and context information; the vector is mapped to the first user intention through a classifier, and key entities in the text, such as names, time and place, are identified.

[0048] Step S104, when the first user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, the successfully matched question is used as the first target question, and the next logical chain question in the sequential logical relationship of the first target question is obtained.

[0049] The reasoning path is the logical steps and decision-making process that the artificial intelligence (AI) model goes through when solving a problem or generating an answer. It describes the thinking chain of the model from input to output. The reasoning path includes at least two logical chain problems, and each logical chain problem has a logical relationship of sequence. The reasoning path includes at least two logical chain problems, and the specific number is not limited and is set according to the actual situation.

[0050] Optionally, the reasoning path is generated by an AI model or set by other information, for example, the reasoning path includes n logical chain questions that have a sequential logical relationship. The terminal first associates the standard question, the standard case, and the reasoning path, that is, forming a logical chain of "standard question-logical chain question 1-logical chain question 2-logical chain question 3-...-logical chain question n-standard answer".

[0051] Optionally, if the first user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, the successfully matched question is used as the first target question, and the next logical chain question in the sequential logical relationship of the first target question is obtained. If the first user intention successfully matches the "standard question" as the first target question, "logical chain question 1" is obtained at the same time; if the first user intention successfully matches the "logical chain question 3" as the first target question, "logical chain question 4" is obtained at the same time.

[0052] Step S106, responding to the first user intention, and generating at least two first guiding questions, wherein there is a first guiding question including a first target guiding question, and the first target guiding question is generated based on the next logical chain question in the sequential logical relationship of the first target question.

[0053] There are at least two first guiding questions, and the specific number is not limited here. It is set according to the actual situation, and can be 2 or 3. The first guiding question can appear in the form of a "Guess what you want to ask" label.

[0054] Optionally, in response to the first user intention, the terminal matches the initial answer corresponding to the first user intention from the database, and generates multiple first guiding questions based on at least one of the user behavior characteristics, the reasoning path relationship and the historical interaction data. The initial answer and the multiple first guiding questions are presented to the user through an interface. Among them, the multiple first guiding questions include a first target guiding question generated based on the next logical chain sub-question in the sequential logical relationship of the first target question. Among them, the database can be MySQL or MongoDB.

[0055] Furthermore, if the number of first guiding questions is set to 3, and the first user intention successfully matches "logical chain question 3" as the first target question, and obtains "logical chain question 4" at the same time, then logical chain question 4 will be used as a first target guiding question, and the other two first guiding questions are non-standard questions and questions outside the reasoning path.

[0056] Step S108, when the first goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the first goal-guiding question, in response to the user's selection operation, the standard answer is matched and displayed.

[0057] Among them, the standard answer is the recommended information corresponding to the standard question and reasoning path.

[0058] Optionally, if there are 4 logical chain questions in the reasoning path, the first guiding question is set to 3. When the first target guiding question is "Logical Chain Question 4" and the user selects "Logical Chain Question 4" from the 3 first guiding questions, in response to the user's selection operation, the standard answer is matched from the database and displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0059] Optionally, if there are 5 logical chain questions in the reasoning path, the first guiding question is set to 4. When the first target guiding question is "Logical Chain Question 5" and the user selects "Logical Chain Question 5" from the 4 first guiding questions, in response to the user's selection operation, the standard answer is matched from the database and displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0060] Optionally, when the first target-guided question is not the last logical chain sub-question in the reasoning path and the user selects the first target-guided question, the next logical chain sub-question of the first target-guided question in the reasoning path is generated as one of the “guess what you want to ask” questions for the user to select.

[0061] Optionally, when the user does not select the first target guiding question, there are two situations: one is that the first target guiding question is the last logical chain question in the reasoning path; the other is that the first target guiding question is not the last logical chain question in the reasoning path. As long as the user does not select the first target guiding question but selects other first guiding questions, the terminal will respond to the user's selection operation, match and display the answers corresponding to other first guiding questions.

[0062] It should be noted that the number of the first guiding questions and the number of logical chain questions in the reasoning path can be greater than or equal to 2, and the specific number is not limited in this solution.

[0063] In the above-mentioned intelligent question-answering guidance method based on reasoning paths, by receiving the first question input by the user, first, the first user intention corresponding to the first question is identified, and it is determined whether the first user intention matches the standard question and any question in the reasoning path corresponding to the standard question. If it matches, the successfully matched question is used as the first target question, and the corresponding logical chain sub-question and the next logical chain sub-question in the reasoning path where the first target question is located are obtained; secondly, in response to the first user intention, the next logical chain sub-question is used as one of the first guidance questions, that is, the first target guidance question, and is displayed to the user; finally, if the first target guidance question is the last logical chain sub-question in the reasoning path and the user selects the first target guidance question, then in response to the user's selection operation, the standard answer is matched, and the standard answer is the recommended information corresponding to the standard question and the reasoning path, and the recommended information is displayed. That is, between identifying user intention and outputting preset standard answers according to user intention, an inference path is added, and the user's selection operation is assisted. If the first target guiding question generated according to the logical chain sub-questions in the inference path is selected, and other first guiding questions are abandoned, it indicates that the user is interested in the logical chain sub-questions, and based on the user's selection behavior, it is further determined that the questions in the inference path meet the user's needs. If the first target guiding question selected by the user is the last logical chain question in the inference path, the standard answer is recommended to the user. By determining whether the user's choice matches the question in the inference path, the user's actual needs are further verified, thereby avoiding the situation where the user's intention is inaccurately judged based on the user's input question only once, thereby improving the accuracy of information recommendation.

[0064] In an exemplary embodiment, Figure 2As shown, in the case where the first target guiding question is not the last logical chain question in the reasoning path and the user selects the first target guiding question, an intelligent question-answering guiding method based on the reasoning path further includes steps S202 to S204. Among them:

[0065] Step S202, when the first target guiding question is not the last logical chain sub-question in the reasoning path and the user selects the first target guiding question, at least two second guiding questions are generated in response to the user's selection operation, wherein there is a second guiding question including the second target guiding question.

[0066] Among them, the second target-guided question is generated based on the next logical chain question in the sequential logical relationship of the first target-guided question.

[0067] Optionally, when the first target guiding question is not the last logical chain sub-question in the reasoning path and the user selects the first target guiding question, for example, if there are 5 logical chain sub-questions in the reasoning path, the first guiding question is set to 4, and when the first target guiding question is "Logical Chain Question 2", and the user selects "Logical Chain Question 2" from the 4 first guiding questions, at least two second guiding questions are generated in response to the user's selection operation, of which one second guiding question includes the second target guiding question, that is, "Logical Chain Question 3" is taken as the second target guiding question, and the other three second guiding questions are non-standard questions and questions outside the reasoning path.

[0068] Step S204, when the second goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the second goal-guiding question, the standard answer is matched and displayed again in response to the user's selection operation.

[0069] Optionally, when the second target guiding question is "Logical Chain Question 5" and the user selects "Logical Chain Question 5", in response to the user's selection operation, the standard answer is matched from the database again, and the standard answer is displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0070] Optionally, when the second target-guided question is not the last logical chain sub-question in the reasoning path and the user selects the second target-guided question, the next logical chain sub-question of the first target-guided question in the reasoning path will continue to be generated as one of the “guess what you want to ask” questions, such as “logical chain question 3” for the user to select.

[0071] Optionally, when the user does not select the second target guiding question, there are two situations: one is that the second target guiding question is the last logical chain question in the reasoning path; the other is that the second target guiding question is not the last logical chain question in the reasoning path. As long as the user does not select the second target guiding question but selects other second guiding questions, the terminal will respond to the user's selection operation, match and display the answers corresponding to other second guiding questions.

[0072] In this embodiment, by setting guiding questions for multiple rounds through the reasoning path, the user can be guided to achieve multiple rounds of interaction in combination with the user's selection behavior, and the user's intention can be further determined, thereby improving the accuracy of the standard answer recommendation.

[0073] In an exemplary embodiment, when the user does not select the first target guiding question, and the user selects other first guiding questions that do not include the first target guiding question or the user inputs the second question, in response to the user's operation, the corresponding answer is matched according to the other first guiding questions or the second user intention, and the answer is displayed.

[0074] The second user intention is determined based on the second question. The second question may be any question input by the user other than the first question.

[0075] Optionally, when the user does not select the first target guiding question, and the user selects other first guiding questions that do not include the first target guiding question, in response to the user's operation, the corresponding answers are matched from the database according to the other first guiding questions, and the answers are displayed through the interface. If the user further inputs questions, it is further determined whether the questions input by the user match the standard questions and the reasoning path.

[0076] Optionally, when the user does not select the first target-guided question and the user inputs the second question, the first question is segmented, and each segmented word and the contextual relationship of each segmented word in the sentence are converted into a vector by means of word embedding and context information; the vector is mapped to the second user intent through a classifier, and key entities in the text are identified, such as names, time and place, etc. In response to the user's input operation, the corresponding answer is matched from the database according to the second user intent, and the answer is displayed through the interface.

[0077] In this embodiment, when the user does not select the first target guiding question, and the user selects other first guiding questions that do not include the first target guiding question or the user inputs the second question, multiple rounds of interaction are implemented based on the other first guiding questions or the user input second question to determine and verify the user's intention, thereby improving the accuracy of the standard answer, that is, the recommended information.

[0078] In an exemplary embodiment, Figure 3 As shown, in the case where the second user intention successfully matches the standard question and any one of the questions in the reasoning path, an intelligent question-answering guidance method based on the reasoning path also includes steps S302 to S306. Among them:

[0079] Step S302, when the second user intention successfully matches the standard question and any question in the reasoning path, the successfully matched question is used as the second target question, and the next logical chain question in the sequential logical relationship of the second target question is obtained.

[0080] Optionally, if the second user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, the successfully matched question is used as the second target question, and the next logical chain question in the sequential logical relationship of the second target question is obtained. If the second user intention successfully matches the "standard question" as the second target question, "logical chain question 1" is obtained at the same time; if the second user intention successfully matches the "logical chain question 3" as the second target question, "logical chain question 4" is obtained at the same time.

[0081] Step S304, responding to the second user intention, and generating at least two third guiding questions, wherein there is a third guiding question including a third target guiding question, and the third target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the second target question.

[0082] Optionally, the terminal matches the initial answer corresponding to the second user intention from the database in response to the second user intention, and generates multiple third guiding questions at the same time. The initial answer and the multiple third guiding questions are presented to the user through an interface. Among the multiple third guiding questions, a third target guiding question is generated based on the next logical chain question in the sequential logical relationship of the third target question.

[0083] Furthermore, if the third guiding question is set to 3, the second user intention successfully matches "logical chain question 3" as the third target question, and obtains "logical chain question 4" at the same time, then logical chain question 4 will be used as a third target guiding question, and the other two third guiding questions are non-standard questions and questions outside the reasoning path.

[0084] Step S306: When the third goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the third goal-guiding question, the standard answer is matched and displayed in response to the user's selection operation.

[0085] Optionally, if there are 4 logical chain questions in the reasoning path, the third guiding question is set to 3. When the third target guiding question is "Logical Chain Question 4" and the user selects "Logical Chain Question 4" from the 3 third guiding questions, in response to the user's selection operation, the standard answer is matched from the database and displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0086] Optionally, if there are 5 logical chain questions in the reasoning path and the third guiding question is set to 4, when the third target guiding question is "Logical Chain Question 5" and the user selects "Logical Chain Question 5" from the 4 third guiding questions, in response to the user's selection operation, the standard answer is matched from the database and displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0087] In this embodiment, even if the user does not choose according to the set reasoning path, by re-entering the form of the second question, it is determined whether the second user intention of the second question matches the standard question and the reasoning path. If it is successfully matched, the user is redirected to the reasoning path. That is, through multiple rounds of dialogue, the user's true intention is determined to improve the accuracy of information recommendation.

[0088] In an exemplary embodiment, Figure 4 As shown, the method of generating the reasoning path includes steps S402 to S404. Among them:

[0089] Step S402: Generate at least two logical chain sub-questions with a sequential logical relationship based on the standard question and the standard answer as an initial reasoning path.

[0090] Optionally, the terminal inputs "standard questions" and "standard answers" through the AI ​​big model, combines expert knowledge and at least one of the preset rules to generate at least two logical chain problems with a chronological logical relationship as the initial reasoning path. If the initial reasoning path passes the review, it can be used as the reasoning path and the reasoning path is published for subsequent application; if the initial reasoning path fails the review, the initial reasoning path needs to be further optimized and adjusted, where the preset rules are to generate corresponding rules for the logical chain problems.

[0091] Optionally, the terminal also constructs a problem-initial reasoning path mapping through a directed acyclic graph (DAG) model to generate an initial reasoning path to ensure the stability and optimization capability of the initial reasoning path. The terminal also uses a finite state machine (FSM) to enable the reasoning path to have a flexible state transition mechanism to adapt to different types of problem guidance and ensure the hierarchy and rationality of the reasoning process.

[0092] Step S404, adjusting the initial reasoning path according to preset conditions to obtain a reasoning path, wherein the preset conditions are set based on business requirements.

[0093] Optionally, Rule-Based Path Optimization is applicable to custom business logic. Preset conditions are set according to the custom business logic. The terminal also adjusts the initial reasoning path according to the preset conditions to obtain the reasoning path, ensuring that users acquire knowledge according to the set path rather than free exploratory questions and answers, so as to improve the accuracy of information recommendation. The terminal adjusts the initial reasoning path according to historical user behavior data to generate the final reasoning path. The terminal also adjusts the initial reasoning path through reinforcement learning optimization strategy to generate the final reasoning path.

[0094] In this embodiment, an initial reasoning path is pre-set through standard questions and standard answers, and a mapping relationship between the initial reasoning path and the standard questions and standard answers is established, so that the initial reasoning path is adjusted, which can improve the accuracy of information recommendation.

[0095] In an exemplary embodiment, Figure 5 As shown, the updating method of the reasoning path includes steps S502 to S504. Among them:

[0096] Step S502: obtaining the user's selection operation data within a preset time period.

[0097] The selection operation data includes other first guiding questions selected by the user, the second user intention, the number of positive evaluations of each first guiding question, and the number of negative evaluations of each first guiding question. Other first guiding questions refer to guiding questions that do not include the first target guiding question. The second user intention is determined based on the second question input by the user. Positive evaluations can be collected in the form of likes, and the number of user likes; correspondingly, negative evaluations can be collected in the form of dislikes, and the number of user dislikes can be collected.

[0098] Optionally, the terminal obtains the user's selection operation data within a preset time period from the database. For example, the terminal obtains the user's selection of the first guidance question, the second user intention input by the user, the number of likes for each first guidance question, and the number of dislikes for each first guidance question within one month.

[0099] Step S504: Optimize and update the reasoning path according to the selection operation data to obtain a new reasoning path.

[0100] Optionally, the terminal also optimizes the path through reinforcement learning (RL), combines user interaction behaviors, such as user click behavior data, calculates the reasoning path score, and continuously adjusts the reasoning path through policy gradient optimization (PolicyGradient) to improve the adaptive ability of reasoning path optimization. The terminal also uses the rule optimization model to adjust the weights of the logical chain sub-questions in the reasoning path to dynamically optimize the path structure and generate a new reasoning path. The terminal can also use statistical analysis methods (Statistical Optimization) to dynamically adjust the reasoning path based on user behavior logs, optimize the guidance effect of recommended questions, and improve the adaptability of the system in complex question-and-answer scenarios. Through user click behavior data feedback, optimize recommended questions, improve the personalization and accuracy of question-and-answer guidance, and enable the system to adaptively optimize the reasoning path during multiple rounds of dialogue.

[0101] Optionally, the terminal adds the user's selected first guiding question to the reasoning path; or regenerates the reasoning path based on the collected second input question; or deletes or adds the first guiding question from the reasoning path based on the number of likes for each first guiding question and the number of thumbs-downs for each first guiding question. It should be noted that the terminal can also use the above methods together to optimize and update the reasoning path so as to obtain a new reasoning path.

[0102] In this embodiment, by continuously optimizing and updating the reasoning path, the logical sub-problems in the reasoning path can be made more realistic, and the accuracy of information recommendation can be achieved.

[0103] In an exemplary embodiment, the relationship between the pre-established standard questions, standard answers and reasoning paths is obtained. If the reasoning path is: logical chain question 1-logical chain question 2-logical chain question 3-logical chain question 4, the number of guiding questions generated each time, i.e., "guess what you want to ask", is 3.

[0104] The terminal receives the first question input by the user through the interface, and uses NLP to identify the intent of the first question through semantic analysis to obtain the first user intention. Assume that if the first user intention successfully matches "Logical Chain Question 2" as the first target question, "Logical Chain Question 3" is obtained at the same time. In response to the first user intention, the terminal matches the initial answer corresponding to the first user intention from the database, and generates three first guiding questions at the same time. The initial answer and the three first guiding questions are displayed to the user through the interface. Among them, the multiple first guiding questions include one based on the first target guiding question, and the first target guiding question is "Logical Chain Question 3". That is, the three first guiding questions are the first target question "Logical Chain Question 3", "Sub-question x" and "Sub-question y" respectively.

[0105] If the user does not select "Logical chain sub-question 3" and the user selects "Sub-question x", in response to the user's operation, the corresponding answer is matched from the database according to "Sub-question x" and the answer is displayed through the interface.

[0106] If the first target guiding question is "Logical chain sub-question 3", and the user selects "Logical chain sub-question 3" from the three first guiding questions, in response to the user's current selection operation, three second guiding questions are generated, among which there is a second guiding question that includes the second target guiding question, that is, "Logical chain sub-question 4" is used as the second target guiding question, and the other two second guiding questions are non-standard questions and questions outside the reasoning path. The generated second guiding questions are: "Logical chain sub-question 4", "Sub-question m" and "Sub-question n". If the user selects "Logical chain sub-question 4", in response to the user's current selection operation, the standard answer is matched from the database again, and the standard answer is displayed through the interface, that is, the recommended information is displayed to the user through the interface. If the user does not select "Logical chain sub-question 4", but selects "Sub-question m", the terminal matches and displays the answers corresponding to other second guiding questions in response to the user's current selection operation.

[0107] If the user neither selects "Logical Chain Question 4" nor other guiding questions, but inputs the second question and extracts the second user intention, and the second user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, if the second user intention successfully matches "Logical Chain Question 3" as the second target question, and obtains "Logical Chain Question 4" at the same time. Then, Logic Chain Question 4 is used as a third target guiding question, and the other two third guiding questions are non-standard questions and questions outside the reasoning path. When the third target guiding question is "Logical Chain Question 4" and the user selects "Logical Chain Question 4" from the three third guiding questions, in response to the user's selection operation, the standard answer is matched from the database and displayed through the interface, that is, the recommended information is displayed to the user through the interface.

[0108] If the user does not select "Logical Chain Question 4" or re-enters a new question, the intent is extracted based on the question again, and the intent is matched with the standard question and any question in the reasoning path corresponding to the standard question. If there is a match, the execution continues according to the set reasoning path. If there is no match, the answer corresponding to the new question is obtained from the database.

[0109] Among them, the method of generating the reasoning path includes: the terminal inputs the "standard question" and "standard answer" through the AI ​​big model to generate at least two logical chain questions with a chronological logical relationship as the initial reasoning path. If the initial reasoning path passes the review, it can be used as the reasoning path and the reasoning path is published for subsequent application; if the initial reasoning path fails the review, the initial reasoning path needs to be further optimized and adjusted: Rule-Based Path Optimization is suitable for custom business logic. Preset conditions are set according to the custom business logic. The terminal also adjusts the initial reasoning path according to the preset conditions to obtain the reasoning path, ensuring that users acquire knowledge according to the set path, rather than free exploratory questions and answers, to improve the accuracy of information recommendations.

[0110] The updating method of the reasoning path includes: the terminal obtains the user's selection operation data within a preset time period from the database, for example, the terminal obtains the user's selection of the first guiding question within one month, the second user intention input by the user, the number of likes for each first guiding question, and the number of thumbs-down for each first guiding question. The terminal adds the user's selection of the first guiding question to the reasoning path; or regenerates the reasoning path based on the collected second input question; or deletes or adds the first guiding question from the reasoning path based on the number of likes for each first guiding question and the number of thumbs-down for each first guiding question. It should be noted that the terminal can also use the aforementioned methods together to optimize and update the reasoning path in order to obtain a new reasoning path.

[0111] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0112] Based on the same inventive concept, the embodiment of the present application also provides an intelligent question-answering guidance device based on reasoning paths for implementing the intelligent question-answering guidance method based on reasoning paths involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the intelligent question-answering guidance device based on reasoning paths provided below can refer to the limitations of the intelligent question-answering guidance method based on reasoning paths above, and will not be repeated here.

[0113] In an exemplary embodiment, Figure 6 As shown, an intelligent question-answering guiding device based on reasoning path is provided, including a receiving module 601, a matching acquisition module 602, a first response generating module 603 and a first response matching module 604, wherein:

[0114] The receiving module 601 is used to receive a first question input by a user, perform intent recognition according to the first question, and obtain a first user intent.

[0115] The matching acquisition module 602 is used to take the successfully matched question as the first target question and obtain the next logical chain question in the sequential logical relationship of the first target question when the first user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question; wherein the reasoning path includes at least two logical chain questions, and each logical chain question has a sequential logical relationship.

[0116] The first response generation module 603 is used to respond to the first user intention and generate at least two first guiding questions, wherein there is a first guiding question including a first target guiding question, and the first target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target question.

[0117] The first response matching module 604 is used to match and display the standard answer in response to the user's selection operation when the first target-guided question is the last logical chain sub-question in the reasoning path and the user selects the first target-guided question, wherein the standard answer is the recommended information corresponding to the standard question and the reasoning path.

[0118] In an exemplary embodiment, an intelligent question-and-answer guidance device based on a reasoning path is provided, including a second response generation module, which is used to respond to the user's selection operation and generate at least two second guidance questions when the first target guidance question is not the last logical chain sub-question in the reasoning path and the user selects the first target guidance question, wherein there is a second guidance question including the second target guidance question, and the second target guidance question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target guidance question; when the second target guidance question is the last logical chain question in the reasoning path and the user selects the second target guidance question, the module again responds to the user's selection operation to match and display the standard answer.

[0119] In an exemplary embodiment, an intelligent question-and-answer guidance device based on a reasoning path is provided, including a second response matching module, which is used to match corresponding answers according to other first guidance questions or second user intentions in response to user operations and display answers when the user does not select the first target guidance question and the user selects other first guidance questions that do not include the first target guidance question or the user inputs the second question; the second user intention is determined based on the second question.

[0120] In an exemplary embodiment, an intelligent question-and-answer guidance device based on a reasoning path is provided, including a third response generation module, which is used to, when a second user intention successfully matches a standard question and any question in the reasoning path, use the successfully matched question as the second target question and obtain the next logical chain sub-question in the chronological logical relationship where the second target question is located; respond to the second user intention, and generate at least two third guidance questions, wherein there is a third guidance question including a third target guidance question, and the third target guidance question is generated based on the next logical chain sub-question in the chronological logical relationship where the second target question is located; when the third target guidance question is the last logical chain sub-question in the reasoning path and the user selects the third target guidance question, match and display the standard answer in response to the user's selection operation.

[0121] In an exemplary embodiment, an intelligent question-and-answer guidance device based on reasoning paths is provided, including a reasoning path generation module, which is used to generate at least two logical chain questions with a chronological logical relationship according to standard questions and standard answers as an initial reasoning path; the initial reasoning path is adjusted according to preset conditions to obtain a reasoning path, wherein the preset conditions are set based on business needs.

[0122] In an exemplary embodiment, an intelligent question-and-answer guidance device based on an inference path is provided, including an inference path updating module, which is used to obtain the user's selection operation data within a preset time period, the selection operation data including other first guidance questions selected by the user, the second user intention, the number of positive evaluations for each first guidance question, and the number of negative evaluations for each first guidance question, other first guidance questions refer to guidance questions that do not include the first target guidance question; the second user intention is determined based on the second question input by the user; the reasoning path is optimized and updated according to the selection operation data to obtain a new reasoning path.

[0123] Each module in the above-mentioned intelligent question-answering guidance device based on reasoning path can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0124] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 7 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store reasoning path data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an intelligent question-answering guidance method based on a reasoning path is implemented.

[0125] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0126] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0127] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0128] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0129] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations. If the recommended information of the standard answer is an advertisement, the user can refuse or can easily refuse the advertising push information.

[0130] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0131] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0132] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An intelligent question-answering guidance method based on reasoning path, characterized in that: The method comprises: Receive a first question input by a user, and perform intent recognition according to the first question to obtain a first user intent; In the case where the first user intention successfully matches the standard question and any question in the reasoning path corresponding to the standard question, the successfully matched question is used as the first target question, and the next logical chain question in the sequential logical relationship of the first target question is obtained; wherein the reasoning path includes at least two of the logical chain questions, and each of the logical chain questions has a sequential logical relationship; In response to the first user intention, at least two first guiding questions are generated, wherein one of the first guiding questions includes a first target guiding question, and the first target guiding question is generated based on the next logical chain question in the sequential logical relationship of the first target question; When the first goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the first goal-guiding question, in response to the user's selection operation, a standard answer is matched and displayed, wherein the standard answer is recommended information corresponding to the standard question and the reasoning path.

2. The method according to claim 1, characterized in that The method further comprises: In the case where the first target guiding question is not the last logical chain sub-question in the reasoning path and the user selects the first target guiding question, in response to the user's selection operation, at least two second guiding questions are generated, wherein one of the second guiding questions includes a second target guiding question, and the second target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target guiding question; When the second goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the second goal-guiding question, the standard answer is matched and displayed again in response to the user's selection operation.

3. The method according to claim 1, characterized in that The method further comprises: In a case where the user does not select the first target guiding question, and the user selects other first guiding questions that do not include the first target guiding question or the user inputs a second question, in response to the user's operation, the corresponding answer is matched according to the other first guiding question or the second user intention, and the answer is displayed; the second user intention is determined based on the second question.

4. The method according to claim 3, characterized in that The method further comprises: When the second user intention successfully matches the standard question and any one of the questions in the reasoning path, the successfully matched question is used as the second target question, and the next logical chain question in the sequential logical relationship of the second target question is obtained; In response to the second user intention, at least two third guiding questions are generated, wherein one of the third guiding questions includes a third target guiding question, and the third target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the second target question; When the third goal-guiding question is the last logical chain sub-question in the reasoning path and the user selects the third goal-guiding question, the standard answer is matched and displayed in response to the user's selection operation.

5. The method according to claim 1, characterized in that The method for generating the reasoning path includes: According to the standard question and the standard answer, at least two of the logical chain sub-questions having the sequential logical relationship are generated as an initial reasoning path; The initial reasoning path is adjusted according to a preset condition to obtain the reasoning path, wherein the preset condition is set based on business requirements.

6. The method according to claim 5, characterized in that The updating method of the reasoning path includes: Acquiring selection operation data of the user within a preset time period, the selection operation data including other first guidance questions selected by the user, a second user intention, the number of positive evaluations of each of the first guidance questions, and the number of negative evaluations of each of the first guidance questions, the other first guidance questions refer to guidance questions that do not include the first target guidance question; the second user intention is determined based on the second question input by the user; The reasoning path is optimized and updated according to the selection operation data to obtain a new reasoning path.

7. An intelligent question-answering guidance device based on reasoning path, characterized in that: The device comprises: A receiving module, configured to receive a first question input by a user, and perform intent recognition according to the first question to obtain a first user intent; A matching acquisition module, configured to, when the first user intention successfully matches a standard question and any question in the reasoning path corresponding to the standard question, use the successfully matched question as the first target question, and acquire the next logical chain question in the sequential logical relationship of the first target question; wherein the reasoning path includes at least two logical chain questions, and each of the logical chain questions has a sequential logical relationship; A first response generation module, configured to respond to the first user intention and generate at least two first guiding questions, wherein one of the first guiding questions includes a first target guiding question, and the first target guiding question is generated based on the next logical chain sub-question in the sequential logical relationship of the first target question; The first response matching module is used to match and display a standard answer in response to the user's selection operation when the first target-guided question is the last logical chain sub-question in the reasoning path and the user selects the first target-guided question, wherein the standard answer is recommended information corresponding to the standard question and the reasoning path.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images