Question identification method and device, storage medium and electronic equipment
By acquiring question information from intelligent customer service calls, using large models for intent analysis and matching, and identifying and updating the question information set, the problem of slow speed and low accuracy of user question identification in intelligent outbound calling scenarios is solved, and the accurate identification of new user questions and the improvement of the response system are achieved.
Patent Information
- Application Number
- CN202511890425.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-15
- Publication Date
- 2026-04-07
AI Technical Summary
In existing technologies, the speed and efficiency of user problem identification in intelligent outbound calling scenarios are slow, and manual inspection methods cannot identify problems that occur infrequently, resulting in intelligent customer service being unable to accurately identify user problems.
By acquiring user questions during conversations with intelligent customer service, a large model is used to perform intent analysis to generate a summary of question intent information. This summary is then matched with the question information set to identify target question information that is not included and to update the question information set.
It has enabled comprehensive collection and precise extraction of user questions, improved the accuracy of identifying new questions, and perfected the intelligent customer service response system.
Smart Images
Figure CN121814897A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of data processing technology, and in particular to a problem identification method, apparatus, storage medium and electronic device. Background Technology
[0002] Intelligent outbound calling is widely used in business scenarios such as after-sales follow-up. During the call, users may raise inquiries or requests that are not within the standard operating procedure (SOP). In order to improve the response system, it is necessary to identify such user questions from the call records to supplement relevant knowledge or optimize the intelligent outbound calling process.
[0003] Currently, for user problem identification in intelligent outbound calling scenarios, existing technologies include the following two identification methods: manually reviewing all problems in historical call records and identifying new user problems; and using language models to filter out and store problems that occur more than a certain number of times.
[0004] However, using manual inspection to identify problems results in slow identification speed and low efficiency. Using model screening methods cannot identify problems that occur infrequently, leading to the omission of some problems. Consequently, the intelligent outbound customer service cannot accurately identify users' problems. Summary of the Invention
[0005] In view of this, this application provides a problem identification method, device, storage medium and electronic device, the main purpose of which is to improve the technical problems of the prior art, which are that the method of manually inspecting and identifying problems leads to slow identification speed and low efficiency, and the method of using model screening cannot identify problems that occur infrequently, resulting in the omission of problems, and thus the intelligent outbound customer service cannot accurately identify the user's problems.
[0006] Firstly, this application provides a problem identification method, including: In response to a user engaging in a conversation with the intelligent customer service, the system obtains the user's question information. The user's question intent is analyzed to generate a summary of the user's question intent. The summary of the question intent is matched with the set of question information used by the intelligent customer service for making a call, and target question information not included in the set of question information is generated. The target question information is stored in the question information set to update the question information set used by the intelligent customer service for making calls.
[0007] Optionally, the step of matching the summary of the question intent with the set of question information used by the intelligent customer service for the call, and generating target question information not included in the set of question information, includes: Determine the semantic similarity between the summary of the questioning intent and the question information in the question information set; In response to the semantic similarity not meeting the similarity condition, it is determined that the question information set does not contain the question information corresponding to the summary information of the question intent; The question information corresponding to the summary information of the questioning intent is determined as the target question information.
[0008] Optionally, intent analysis is performed on the question information to generate a summary of the user's question intent, including: Intent analysis is performed on the question information to obtain the type identifier and question content of the question information, wherein the question content includes the subject information and action information in the question information; Based on the type identifier and the question content, a summary of the user's question intent is generated.
[0009] Optionally, storing the target question information in the question information set to update the question information set used by the intelligent customer service for making calls includes: Based on the type identifier of the question information corresponding to the target question information, the target question information is divided into at least one target question set; The at least one set of target questions is stored in the question information set to update the question information set used by the intelligent customer service for making calls.
[0010] Optionally, the method further includes: Obtain the historical call data of the intelligent customer service; Determine the historical question intent summary information corresponding to the historical call data; The historical question intent summary information is clustered, and based on the clustering results, the historical question intent summary information is divided into at least one set of historical question intent summary information. Based on the at least one set of historical question intent summary information, generate historical question information not included in the question information set.
[0011] Optionally, generating historical question information not included in the question information set based on the at least one historical question intent summary information set includes: Determine the semantic similarity between the at least one historical question intent summary information set and the question information set to obtain at least one historical semantic similarity data; At least one set of historical problem information not included in the problem information set is generated based on the at least one set of historical semantic similarity data. The at least one set of historical problem information is used to store in the problem information set for updating the problem information set.
[0012] Secondly, this application provides a problem identification device, comprising: The acquisition module is configured to acquire the user's question information in response to a call between the user and the intelligent customer service; The generation module is configured to perform intent analysis on the question information and generate a summary of the user's question intent. The generation module is also configured to match the summary information of the question intent with the set of question information used by the intelligent customer service for the call, and generate target question information not included in the set of question information; The storage module is configured to store the target question information in the question information set in order to update the question information set used by the intelligent customer service for making calls.
[0013] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the problem identification method described in the first aspect.
[0014] Fourthly, this application provides an electronic device, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the computer program to implement the problem identification method described in the first aspect.
[0015] By employing the above technical solutions, this application provides a problem identification method, apparatus, storage medium, and electronic device. In response to a user's call with intelligent customer service, this application acquires the user's question information; performs intent analysis on the question information to generate a summary of the user's question intent; matches the summary of question intent with a set of question information used by the intelligent customer service during the call to generate target question information not included in the question information set; and stores the target question information in the question information set to update the question information set used by the intelligent customer service during the call. Compared with existing technologies, this application achieves comprehensive collection of user questions during the call by acquiring the user's question information in response to a user's call with intelligent customer service; achieves accurate extraction of the core meaning of the user's question by generating a summary of the user's question intent through intent analysis of the question information; improves the accuracy of identifying new user questions by generating target question information not included in the question information set by matching the summary of question intent with the set of question information used by the intelligent customer service during the call; and achieves continuous improvement of the intelligent customer service response system by storing the target question information in the question information set to update the question information set. Attached Figure Description
[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0018] Figure 1 A flowchart illustrating a problem identification method provided in an embodiment of this application is shown; Figure 2 A schematic diagram illustrating an example of a prompt analysis template provided in an embodiment of this application is shown; Figure 3 This illustration shows a schematic diagram of an example of a filled prompt analysis template provided in an embodiment of this application; Figure 4 A flowchart illustrating a problem identification method provided in an embodiment of this application is shown; Figure 5 A schematic diagram illustrating a semantic clustering example provided in an embodiment of this application is shown; Figure 6 A schematic diagram illustrating a semantic matching example provided in an embodiment of this application is shown; Figure 7 This illustration shows an example of recognizing new user questions by combining a large model intent in an intelligent outbound calling scenario, as provided by an embodiment of this application. Figure 8 This paper shows a schematic diagram of the structure of a problem identification device provided in an embodiment of this application; Figure 9 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation
[0019] The embodiments of this application will now be described in more detail with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.
[0020] To address the technical problems of slow and inefficient manual problem identification in current technologies, and the inability of model-based screening to identify infrequently occurring issues, resulting in missed problems and hindering accurate problem identification by intelligent outbound customer service, this embodiment provides a problem identification method, such as... Figure 1 As shown, the method includes: Step 101: Respond to the user's call with the intelligent customer service and obtain the user's question information.
[0021] In this application embodiment, the intelligent customer service can be an automated customer service in an intelligent outbound calling scenario, where the intelligent customer service can proactively initiate a call and interact with the user. For example, the intelligent customer service in this application implementation can specifically be an intelligent outbound calling customer service in scenarios such as mobile phone repair follow-up visits and after-sales consultations.
[0022] In this embodiment of the application, the question information can be various inquiries or requests raised by the user during a call with the intelligent customer service. The question information can include inquiries about business-related knowledge and requests related to business processes. The specific form of the question information can be the user's natural language expression, without the need for a specific format.
[0023] For example, in a mobile phone repair follow-up scenario, after the intelligent customer service confirms the repair information with the user, the user's question about whether there are any coupons available for the repair can be considered a typical inquiry, while the user's response, such as describing the fault as a green line in the middle of the screen or confirming that they have checked the SMS regarding the repair, can be considered a response but not an inquiry.
[0024] Step 102: Perform intent analysis on the question information to generate a summary of the user's question intent.
[0025] In this embodiment of the application, intent analysis can be a process of refining, filtering and structuring the user's question information using the prompt technology of a large model. Intent analysis can be used to clarify the core demands of the user's question.
[0026] In the embodiments of this application, the question intent summary information may include information containing type identifier, subject information and action information. The question intent summary information can be used to reflect the question category and core content of the question asked by the user.
[0027] In this embodiment of the application, intent analysis can be performed using the large model prompt technique to construct an analysis template. The template can include the scope of content to be excluded, examples, and uniform format requirements. This embodiment of the application provides an example diagram of an analysis template constructed using the prompt technique. Figure 2 As shown, Figure 2 The scope of content to be excluded can be filled in using the `{normal_sop}` parameter. Specifically, it refers to content that does not need to be identified as new issues in the standard operating procedure (SOP) of the corresponding business scenario. For example, in the mobile phone repair follow-up scenario, situations where users express fault descriptions, request model modifications, or confirm information provided by the robot are not considered new information inquiries or process requests. Examples of information inquiries can be filled in using the `{info_examples}` parameter, and examples of process requests can be filled in using the `{sop_examples}` parameter. For example, a sample prompt analysis template after filling in the prompt in the mobile phone repair follow-up scenario could be as follows: Figure 3 As shown.
[0028] In this embodiment, the specific process of intent analysis may include: substituting the complete call record into the template using the {conversation} parameter; after filtering the question information, the large model generates a summary result in the format of information inquiry-topic-action or process request-topic-action xxx, where the type identifier can be information inquiry or process request, the subject information can be the core object of user concern, and the action information can be the user's specific requirements for the subject. It should be noted that if the user does not raise any information inquiry or process request, the large model can return an empty set [].
[0029] Step 103: Match the summary of the question intent with the set of question information used by the intelligent customer service for the call, and generate target question information that is not included in the question information set.
[0030] In this embodiment of the application, the question information set can be a reference set used by the intelligent customer service to compare user intent during a call. The question information set can include a regular intent set, an intent set corresponding to existing Frequently Asked Questions (FAQs), and a newly discovered question intent set. The regular intent set can store intents that have been confirmed not to be new questions, the intent set corresponding to existing FAQs can store common question intents that already have answer content, and the newly discovered question intent set can store new question intents that have been previously identified and supplemented with answer content.
[0031] In this embodiment of the application, the target question information may be information corresponding to a new user question that is not included in the question information set after matching.
[0032] In this embodiment of the application, the semantic similarity between the summary information of the question intent and each intent in the question information set can be calculated by using a semantic representation model to convert each intent in the summary information of the question intent and the question information set into a computable feature vector, and then using a similarity algorithm to calculate the degree of similarity between the two. The semantic representation model may include, but is not limited to, the bge-m3 semantic representation model, and the similarity algorithm may include, but is not limited to, the cosine similarity algorithm, the Euclidean distance algorithm, etc., and is not limited to a specific model or algorithm.
[0033] For example, if the user's question intent summary information and the intent similarity already included in the question information set meet the similarity condition, it can be determined that it is already included; if the user's question intent summary information and the intent similarity already included in the question information set do not meet the similarity condition, it can be determined that it is not included, and target question information not included in the question information set is generated.
[0034] Step 104: Store the target question information in the question information set to update the question information set used by the intelligent customer service for calls.
[0035] In the embodiments of this application, updating the problem information set can be a process of incorporating newly identified target problem information into the original problem information set, thereby expanding the scope of reference intent for intelligent customer service and improving the response system.
[0036] In this embodiment of the application, the target question information is stored in a question information set so that the question information set used for calls by the intelligent customer service can be updated and stored according to its corresponding type identifier. At the same time, the target question information can also be provided to business personnel for review. After the business personnel supplement the corresponding response content, it can be added to the online FAQ library of the intelligent customer service.
[0037] Compared with existing technologies, this application achieves comprehensive collection of user questions during calls by responding to user conversations with intelligent customer service; it generates a summary of user question intent through intent analysis of the question information, enabling precise extraction of the core meaning of user questions; it generates target question information not included in the question information set by matching the summary of question intent with the question information set used by intelligent customer service during the call, improving the accuracy of identifying new user questions; and it continuously improves the intelligent customer service response system by storing the target question information set to update the question information set.
[0038] As an optional approach, when performing the task of "matching the summary of the question intent with the set of question information used by the intelligent customer service for the call, and generating target question information not included in the question information set," the following methods can be used, but are not limited to them: Figure 4 As shown, it includes: Step 201: Determine the semantic similarity between the information summarizing the question's intent and the question information in the question information set.
[0039] In this embodiment of the application, semantic similarity can be a quantitative indicator used to measure the similarity in core meaning between the summary information of the question intent and each question information in the question information set. The value of semantic similarity can range from 0 to 1. The closer the value of semantic similarity is to 1, the closer the meaning of the summary information of the question intent and the question information in the question information set is. The closer the value is to 0, the greater the difference in meaning between the summary information of the question intent and the question information in the question information set.
[0040] In this embodiment of the application, determining the semantic similarity between the summary information of the question intent and the question information in the question information set can be achieved by first converting the summary information of the question intent and each question information in the question information set into feature vectors through a semantic representation model. The semantic representation model can include, but is not limited to, the bge-m3 semantic representation model, or other technologies with equivalent semantic conversion capabilities. The semantic similarity between the two can be calculated based on the converted feature vectors using a preset similarity algorithm. The similarity algorithm can be selected according to the actual application scenario and is not limited to a specific algorithm.
[0041] Step 202: In response to the fact that the semantic similarity does not meet the similarity condition, determine that the question information set does not contain the question information corresponding to the summary information of the question intent.
[0042] In the embodiments of this application, the similarity condition can be a threshold condition used to determine whether the intent matches. The threshold T in the similarity condition can be adjusted according to the needs of the business scenario and the actual test results. For example, the threshold T can be set to 0.8 or other reasonable values.
[0043] In the embodiments of this application, if the semantic similarity is greater than or equal to the threshold T, it can be said that the question information corresponding to the summary information of the question intent has been included in the question information set; if the semantic similarity is less than the threshold T, it can be said that the question information corresponding to the summary information of the question intent has not been included in the question information set and belongs to a potential new question.
[0044] Step 203: Determine the question information corresponding to the summary information of the questioning intent as the target question information.
[0045] In this embodiment of the application, user questions corresponding to question intent summary information not included in the question information set can be new user questions, and the structured information corresponding to the new questions can be determined as target question information. For example, if the user's question intent summary information is process request - door-to-door pickup time - postponement, and the semantic similarity with each intent in the question information set is 0.6, which is lower than the threshold T=0.8, then the question information corresponding to the question intent summary information can be determined as target question information.
[0046] Optionally, when performing "intent analysis on the question information to generate a summary of the user's question intent", the following methods can be used, but are not limited to: performing intent analysis on the question information to obtain the type identifier and question content of the question information, wherein the question content includes the subject information and action information in the question information; and generating a summary of the user's question intent based on the type identifier and question content.
[0047] In this embodiment, the type identifier can be an identifier used to distinguish the category of the query information. For example, in this embodiment, the type identifier can specifically be information inquiry or process request, where information inquiry can correspond to a user's inquiry about business-related knowledge, and process request can correspond to a user's request for business operation procedures.
[0048] In this embodiment of the application, the question content can be a distillation of the core meaning of the user's question. For example, the question content in this embodiment of the application can specifically include subject information and action information, wherein the subject information can be the core object to which the user's question is directed, such as mail-in repair fees, a backup device, or pick-up time; the action information can be the specific requirements made by the user for the core object, such as inquiring, applying, postponing, or urging.
[0049] In this embodiment of the application, the summary information of the user's question intent can be generated according to the preset format requirements based on the type identifier and the question content; during the intent analysis process, content that does not belong to information consultation or process request can be filtered out according to the preset information exclusion rules.
[0050] Optionally, when performing the action of "storing the target question information in the question information set to update the question information set used for calls by the intelligent customer service", the following method may be used, but is not limited thereto: the method includes: dividing the target question information into at least one target question set based on the type identifier of the question information corresponding to the target question information; storing at least one target question set in the question information set to update the question information set used for calls by the intelligent customer service.
[0051] In the embodiments of this application, based on the type identifier of the question information corresponding to the target question information, dividing the target question information into at least one target question set can be done by dividing all target question information with the type identifier of information consultation into one target question set, and dividing all target question information with the type identifier of process request into another target question set, so as to achieve classified storage.
[0052] In the embodiments of this application, during the process of storing at least one set of target questions in a question information set, the set of target questions for information consultation can be added to the new question intent subset in the question information set, and the set of target questions for process requests can also be added to the corresponding category of the new question intent subset. The target question information can be provided to business personnel for review, and after supplementing the corresponding response content, it can be added to the online FAQ library of the intelligent customer service, so that the intelligent customer service can directly call the response content in subsequent calls. After storage, the updated new question intent subset, the regular intent subset in the question information set, and the intent subset corresponding to the existing response resources can be merged to reconstruct the reference intent set of the intelligent customer service, thereby completing the update of the question information set.
[0053] As an optional approach, embodiments of this application may also employ the following methods, but not limited thereto: the method includes: acquiring historical call data of intelligent customer service; determining historical question intent summary information corresponding to the historical call data; performing clustering processing on the historical question intent summary information, and dividing the historical question intent summary information into at least one set of historical question intent summary information based on the clustering results; and generating historical question information not included in the question information set based on at least one set of historical question intent summary information.
[0054] In this embodiment of the application, historical call data may be historical interaction records generated by the intelligent customer service and the user, and historical call data may contain a large amount of information on questions previously raised by the user.
[0055] In this embodiment of the application, the historical question intent summary information can be a structured summary information obtained after performing intent analysis on the user's question information in historical call data. The format of the historical question intent summary information can be consistent with the question intent summary information. The format of the historical question intent summary information can include type identifier, subject information and action information.
[0056] In the embodiments of this application, clustering can be a process of classifying information summarizing historical questioning intentions with similar semantics.
[0057] In this embodiment, clustering the historical question intent summary information can be performed by first converting the core themes of each historical question intent summary information into feature vectors using a semantic representation model, and then using a density clustering algorithm to cluster the feature vectors. The semantic representation model can include, but is not limited to, the bge-m3 semantic representation model, and the density clustering algorithm can include, but is not limited to, the DBSCAN density clustering algorithm, or other algorithms with equivalent clustering effects. Semantically similar core themes can be integrated into a single cluster. This embodiment also provides a semantic clustering example, such as... Figure 5 As shown, in Figure 5 In this context, the cost of mail-in repair and the shipping cost can be integrated into one cluster, and the repair time, repair cycle, and warranty period can be integrated into another cluster. Each cluster can serve as a set of information summarizing historical question intents, and a representative intent can be selected from each set for subsequent matching with the question information set.
[0058] Optionally, when performing the action of "generating historical question information not included in the question information set based on at least one historical question intent summary information set", the following method may be used, but is not limited thereto: the method includes: determining the semantic similarity between at least one historical question intent summary information set and the question information set to obtain at least one historical semantic similarity data; generating at least one historical question information set not included in the question information set based on at least one historical semantic similarity data, wherein the at least one historical question information set is used to store in the question information set for updating the question information set.
[0059] In this embodiment of the application, the semantic similarity between at least one set of historical question intent summary information and the question information set is determined. Obtaining at least one set of historical semantic similarity data can be achieved by converting it into a feature vector through a semantic representation model and then using a similarity algorithm to calculate the semantic similarity between the representative intent of each set of historical question intent summary information and each intent in the question information set.
[0060] In this embodiment, semantic matching is performed on at least one set of historical question intent summary information and a set of question information based on semantic similarity. This embodiment also provides a semantic matching example, such as... Figure 6 As shown, if the semantic similarity between the representative intent of the historical question intent summary information set and each intent in the question information set does not meet the similarity threshold, then the question information corresponding to the historical question intent summary information set can be identified as historical question information not included in the question information set. Based on the historical question information not included, a historical question information set can be constructed. Then, the historical question information set can be manually verified to mark whether each intent is a new question. If it is not a new question, it can be added to the regular intent set; if it is a new question, it can be added to the new question intent set and provided to business personnel for review. Business personnel can supplement the answer content of the new question, add the new question and answer content to the online FAQ library, and merge the updated sets to complete the update of the question information set.
[0061] As an optional approach, embodiments of this application also provide an example of combining large model intent recognition to identify new user questions in an intelligent outbound calling scenario, such as... Figure 7 As shown, Figure 7 The process includes two parts: an online process and an offline process. The online process involves summarizing the hierarchical intent of real-time call records (i.e., generating question intent summary information by analyzing the question information in this embodiment), and then matching it with the user's regular intent library and the FAQ library (i.e., the question set in this embodiment). If no match is found, manual verification can be performed; if a match is found, it can be ignored. The offline process involves summarizing and clustering the hierarchical intent of historical dialogue data, then manually annotating it in conjunction with the FAQ library, and finally updating it to the user's regular intent library and the new question library.
[0062] Compared with existing technologies, this application's embodiments achieve precise filtering of new user questions by determining the semantic similarity between the summary information of question intent and the question information in the question information set, and determining the target question information in response to the semantic similarity not meeting the similarity condition; by dividing and storing at least one target question set based on the type identifier corresponding to the target question information, the classification, organization, and efficient updating of the question information set are achieved; by acquiring historical call data of intelligent customer service and performing clustering processing on the corresponding historical question intent summary information, historical question information not included in the question information set is generated, thereby realizing the mining of potential new questions in historical calls; and by determining the historical semantic similarity data between the historical question intent summary information set and the question information set and generating the historical question information set not included, the comprehensive supplementation and updating of the question information set is achieved.
[0063] Furthermore, as Figure 1 and Figure 4The specific implementation of the method shown in this embodiment provides a problem identification device, such as... Figure 8 As shown, the device includes: an acquisition module 31, a generation module 32, and a storage module 33.
[0064] Module 31 is configured to retrieve user questions in response to a conversation between the user and the intelligent customer service. The generation module 32 is configured to perform intent analysis on the question information and generate a summary of the user's question intent. The generation module 32 is also configured to match the summary information of the question intent with the question information set used by the intelligent customer service for the call, and generate target question information not included in the question information set; Storage module 33 is configured to store target question information in a question information set for updating the question information set used by the intelligent customer service for making calls.
[0065] In some examples of this embodiment, the generation module 32 is specifically configured to determine the semantic similarity between the summary information of the question intent and the question information in the question information set; in response to the semantic similarity not meeting the similarity condition, determine that the question information set does not contain the question information corresponding to the summary information of the question intent; and determine the question information corresponding to the summary information of the question intent as the target question information.
[0066] In some examples of this embodiment, the generation module 32 is further configured to perform intent analysis on the question information to obtain the type identifier and question content of the question information, wherein the question content includes the subject information and action information in the question information; and generate summary information of the user's question intent based on the type identifier and question content.
[0067] In some examples of this embodiment, the storage module 33 is specifically configured to divide the target question information into at least one target question set based on the type identifier of the question information corresponding to the target question information; and to store at least one target question set in the question information set so as to update the question information set used by the intelligent customer service for making calls.
[0068] In some examples of this embodiment, the storage module 33 is further configured to: acquire historical call data of the intelligent customer service; determine historical question intent summary information corresponding to the historical call data; perform clustering processing on the historical question intent summary information, and divide the historical question intent summary information into at least one set of historical question intent summary information based on the clustering results; and generate historical question information not included in the question information set based on at least one set of historical question intent summary information.
[0069] In some examples of this embodiment, the storage module 33 is further configured to determine the semantic similarity between at least one historical question intent summary information set and the question information set, and obtain at least one historical semantic similarity data; generate at least one historical question information set not included in the question information set based on the at least one historical semantic similarity data, and the at least one historical question information set is used to store in the question information set for updating the question information set.
[0070] It should be noted that other corresponding descriptions of the functional units involved in the problem identification device provided in this embodiment can be found in [reference needed]. Figure 1 and Figure 4 The corresponding descriptions in [the document] will not be repeated here.
[0071] Based on the above, Figure 1 and Figure 4 Accordingly, this embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. Figure 1 and Figure 4 The method shown.
[0072] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause a computer device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.
[0073] like Figure 9 The diagram shown is a hardware structure schematic of an electronic device according to the present invention, comprising: At least one processor 401; and, Memory 402 is communicatively connected to at least one processor 401; wherein, The memory 402 stores instructions that can be executed by at least one processor, which enables the at least one processor to perform the problem identification method as described above.
[0074] Figure 9 Take a processor 401 as an example.
[0075] The electronic device may also include an input device 404 and an output device 404.
[0076] The processor 401, memory 402, input device 403, and output device 404 can be connected via a bus or other means. Figure 9 Taking the example of a connection between China and Israel via a bus.
[0077] Memory 402, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer-executable programs, and modules, such as the program instructions / modules corresponding to the problem identification method in the embodiments of this application, for example, Figure 1 and Figure 4 The method flow is shown. The processor 401 executes various functional applications and communications by running non-volatile software programs, instructions, and modules stored in the memory 402, thereby implementing the problem identification method in the above embodiments.
[0078] Memory 402 may include a program storage area and a data storage area, wherein the program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the problem identification method, etc. Furthermore, memory 402 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, memory 402 may optionally include memory remotely located relative to processor 401, and these remote memories may be connected via a network to the apparatus performing the problem identification method. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
[0079] The input device 403 can receive user clicks and generate signal inputs related to user settings and function control for the problem identification method. The output device 403 may include a display device such as a screen.
[0080] One or more modules are stored in memory 402, and when run by one or more processors 401, the problem identification method in any of the above method embodiments is executed.
[0081] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.
[0082] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.
[0083] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.
[0084] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms, or it can be implemented by hardware. By applying the solution of this embodiment, compared with the existing technology, this application embodiment obtains the user's question information by responding to the user's call with intelligent customer service, realizing the comprehensive collection of user questions during the call; by performing intent analysis on the question information to generate the user's question intent summary information, it realizes the accurate extraction of the core meaning of the user's question; by matching the question intent summary information with the question information set used by intelligent customer service in the call, it generates target question information not included in the question information set, improving the accuracy of identifying new user questions; by storing the target question information in the question information set to update the question information set, it realizes the continuous improvement of the intelligent customer service response system; by determining the question intent summary information and the question information... The system calculates the semantic similarity of question information in the information set, identifies target question information when the semantic similarity does not meet the similarity condition, and achieves accurate screening of new user questions. It also classifies and stores at least one target question set based on the type identifier corresponding to the target question information, enabling efficient updating and organization of the question information set. Furthermore, it acquires historical call data from the intelligent customer service system and clusters the corresponding historical question intent summary information to generate historical question information not included in the question information set, thus enabling the discovery of potential new questions in historical calls. Finally, it determines the historical semantic similarity data between the historical question intent summary information set and the question information set, and generates a set of historical question information not included, achieving comprehensive supplementation and updating of the question information set.
[0085] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.
[0086] The above are merely specific embodiments of this application, enabling those skilled in the art to understand or implement this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to these embodiments, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.
Claims
1. A problem identification method, characterized in that, include: In response to a user engaging in a conversation with the intelligent customer service, the system obtains the user's question information. The user's question intent is analyzed to generate a summary of the user's question intent. The summary of the question intent is matched with the set of question information used by the intelligent customer service for making a call, and target question information not included in the set of question information is generated. The target question information is stored in the question information set to update the question information set used by the intelligent customer service for making calls.
2. The method according to claim 1, characterized in that, The step of matching the summary of the question intent with the set of question information used by the intelligent customer service for the call, and generating target question information not included in the set of question information, includes: Determine the semantic similarity between the summary of the questioning intent and the question information in the question information set; In response to the semantic similarity not meeting the similarity condition, it is determined that the question information set does not contain the question information corresponding to the summary information of the question intent; The question information corresponding to the summary information of the questioning intent is determined as the target question information.
3. The method according to claim 1, characterized in that, The step of performing intent analysis on the question information to generate a summary of the user's question intent includes: Intent analysis is performed on the question information to obtain the type identifier and question content of the question information, wherein the question content includes the subject information and action information in the question information; Based on the type identifier and the question content, a summary of the user's question intent is generated.
4. The method according to claim 1, characterized in that, The step of storing the target question information in the question information set to update the question information set used by the intelligent customer service for making calls includes: Based on the type identifier of the question information corresponding to the target question information, the target question information is divided into at least one target question set; The at least one set of target questions is stored in the question information set to update the question information set used by the intelligent customer service for making calls.
5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: Obtain the historical call data of the intelligent customer service; Determine the historical question intent summary information corresponding to the historical call data; The historical question intent summary information is clustered, and based on the clustering results, the historical question intent summary information is divided into at least one set of historical question intent summary information. Based on the at least one set of historical question intent summary information, generate historical question information not included in the question information set.
6. The method according to claim 5, characterized in that, The step of generating historical question information not included in the question information set based on the at least one historical question intent summary information set includes: Determine the semantic similarity between the at least one historical question intent summary information set and the question information set to obtain at least one historical semantic similarity data; At least one set of historical problem information not included in the problem information set is generated based on the at least one set of historical semantic similarity data. The at least one set of historical problem information is used to store in the problem information set for updating the problem information set.
7. A problem identification device, characterized in that, include: The acquisition module is configured to acquire the user's question information in response to a call between the user and the intelligent customer service; The generation module is configured to perform intent analysis on the question information and generate a summary of the user's question intent. The generation module is also configured to match the summary information of the question intent with the set of question information used by the intelligent customer service for the call, and generate target question information not included in the set of question information; The storage module is configured to store the target question information in the question information set in order to update the question information set used by the intelligent customer service for making calls.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.
9. An electronic device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.
10. A computer program product, the computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.