Question pushing method and device, computer device and storage medium

Abstract

This application relates to a question push method, apparatus, computer device, and storage medium. The method includes: acquiring user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location; establishing a filter rule based on the user tag data, user behavior data, and business scenario characteristics; filtering original questions in a preset first question library according to the filter rule to obtain a first candidate question set; determining target keywords based on the user tag data, historical interaction data, and user behavior data; recalling questions from the first question library based on the target keywords to obtain a second candidate question set; determining a first question sequence based on the historical interaction data, the first candidate question set, and the second candidate question set, and pushing the first question sequence to the user's corresponding display interface. This method can improve the accuracy of question push.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a question push method, apparatus, computer device, and storage medium. Background Technology

[0002] With the rapid development of artificial intelligence technology, intelligent question recommendation technology has emerged. Intelligent question recommendation refers to a service that can understand the user's needs when a chat session is established, and recommend questions that the user wants to ask based on those needs.

[0003] In related technologies, frequently asked questions from the backend customer service system are directly selected and recommended to users. However, different customers often have different questions, and directly pushing frequently asked questions can easily lead to inaccurate question recommendations. Therefore, how to improve the accuracy of question recommendations has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer device, and storage medium for pushing issues that can improve the accuracy of such issues, in order to address the aforementioned technical problems.

[0005] Firstly, this application provides a method for pushing questions. The method includes:

[0006] Acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location;

[0007] Establish baffle rules based on the user tag data, user behavior data, and business scenario characteristics;

[0008] The original questions in the preset first question library are filtered according to the baffle rules to obtain the first candidate question set;

[0009] Target keywords are determined based on the user tag data, the historical interaction data, and the user behavior data;

[0010] Based on the target keywords, the first question database is retrieved to obtain a second candidate question set;

[0011] Based on the historical interaction data, the first candidate question set, and the second candidate question set, a first question sequence is determined, and the first question sequence is pushed to the user's corresponding display interface.

[0012] In one embodiment, the method further includes:

[0013] In response to a first operation triggered on the display interface, the target question selected by the user is determined;

[0014] The target question is classified and matched according to a preset classification question library to obtain the classification matching result;

[0015] The user is verified based on the classification and matching results, or the question answer corresponding to the target question is determined based on the user's user attribute information and a preset standard answer library.

[0016] In one embodiment, determining the first question sequence based on the historical interaction data, the first candidate question set, and the second candidate question set includes:

[0017] The target question set is obtained based on the first candidate question set and the second candidate question set;

[0018] The target problem set is sorted, and multiple first problems are determined from the sorted target problem set;

[0019] Multiple second questions are identified from the second question bank based on historical interaction data;

[0020] The first question sequence is obtained by combining the plurality of first questions and the plurality of second questions.

[0021] In one embodiment, the method further includes:

[0022] In response to a second operation triggered on the display interface, or if no operation is triggered on the display interface within a preset time period, the plurality of first questions are removed from the target question set, and the plurality of second questions are removed from the second question library;

[0023] Multiple third questions are identified from the removed target question set, and multiple fourth questions are identified from the removed second question set;

[0024] The plurality of third questions and the plurality of fourth questions are combined to obtain a second question sequence;

[0025] The second question sequence is pushed to the display interface.

[0026] In one embodiment, before acquiring the user's user tag data, user behavior data, historical interaction data, and business scenario features corresponding to the user's current location, the method further includes:

[0027] Obtain the operation information triggered by the user on the customer service system, and the response information generated by the customer service system based on the operation information;

[0028] The user behavior data is determined based on the operation information and the response information.

[0029] In one embodiment, the method further includes:

[0030] Obtain the user's historical interaction information;

[0031] The historical interaction information is converted into interaction vectors according to a preset pre-trained model;

[0032] A cluster of questions is obtained by mining the interaction vectors; the cluster of questions includes at least one question.

[0033] Each question in the question cluster is matched according to the first question database, and the matched question cluster is added to the first question database.

[0034] In one embodiment, converting the historical interaction information into interaction vectors according to a preset pre-trained model includes:

[0035] The historical interaction information is cleaned and deduplicated to obtain the target interaction information;

[0036] The target interaction information is encoded into the interaction vector based on the pre-trained model.

[0037] Secondly, this application also provides a problem push device. The device includes:

[0038] The data acquisition module is used to acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location;

[0039] The rule establishment module is used to establish baffle rules based on the user tag data, user behavior data, and business scenario characteristics.

[0040] The filtering module is used to filter the original questions in the preset first question library according to the baffle rules to obtain a first candidate question set;

[0041] The target keyword determination module is used to determine target keywords based on the user tag data, the historical interaction data, and the user behavior data.

[0042] The recall module is used to recall the first question database based on the target keyword to obtain a second candidate question set;

[0043] The push module is used to determine a first question sequence based on the historical interaction data, the first candidate question set, and the second candidate question set, and push the first question sequence to the user's corresponding display interface.

[0044] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the aforementioned problem-pushing method.

[0045] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the above-described problem-pushing method.

[0046] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the aforementioned problem-pushing method.

[0047] The aforementioned question push method, device, computer equipment, and storage medium acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location. Then, based on these user tag data, user behavior data, and business scenario characteristics, they establish a filter rule. This ensures that the filter rule is highly correlated with the user tag data, user behavior data, historical interaction data, and business scenario characteristics of the user's current location. This improves the relevance between the subsequently selected first candidate question set and the questions the user wants to inquire about, thereby improving the accuracy of question push. Furthermore, by recalling the first question database based on target keywords to obtain second candidate questions, and based on historical interaction data, the first candidate questions, and the second candidate question set, a first question sequence is determined and then pushed to the user. This facilitates the user's selection of the questions they want to inquire about, further improving the accuracy of question push. Attached Figure Description

[0048] Figure 1 This is an application environment diagram of the problem push method in one embodiment;

[0049] Figure 2 This is a schematic diagram of the first process of a problem push method in one embodiment;

[0050] Figure 3 This is a schematic diagram of the second process of the problem push method in one embodiment;

[0051] Figure 4 This is a flowchart illustrating the steps involved in combining the steps to obtain the first problem sequence in one embodiment.

[0052] Figure 5 This is a schematic diagram of the third process of the problem push method in one embodiment;

[0053] Figure 6 This is a schematic diagram of the fourth process of the problem push method in one embodiment;

[0054] Figure 7 This is a structural block diagram of a problem push device in one embodiment;

[0055] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0056] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0057] The problem push method provided in this application embodiment can be applied to, for example, Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system stores the data that server 104 needs to process. The server acquires user tag data, user behavior data on terminal 102, historical interaction data on terminal 102, and business scenario characteristics corresponding to the user's current location. Based on the user tag data, user behavior data, and business scenario characteristics, it establishes swatch rules, filters the original questions in a preset first question library according to the swatch rules to obtain a first candidate question set, then determines target keywords based on the user tag data, historical interaction data, and user behavior data, and recalls the first question library based on the target keywords to obtain a second candidate question set. Based on the historical interaction data, the first candidate question set, and the second candidate question set, it determines a first question sequence and pushes the first question sequence to the user's corresponding display interface. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, and tablets. Server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0058] In one embodiment, such as Figure 2 As shown, a problem push method is provided, which can be applied to Figure 1 Taking server 104 as an example, the following steps are included:

[0059] Step 202: Obtain user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location.

[0060] User tag data refers to user tag data stored on the server and on the terminal. This user tag data may include, but is not limited to, the user's account status, credit limit information, and identity information. Account status includes normal status, blacklisted user, monitored user, invalid user, low-credit user, and high-credit user. Credit limit information refers to the amount of funds a user can borrow. Identity information refers to information used to uniquely identify a user. This identity information can be represented by the user's ID number, mobile phone number, etc.

[0061] User behavior data refers to data on the actions a user takes on a device. This user behavior data may include, but is not limited to, the operations, actions, and behaviors performed by the user on the device.

[0062] In some embodiments, user behavior data may further include responses generated by the terminal based on actions performed by the user.

[0063] Business scenario characteristics can characterize the business scenario a user is in on their current device. For example, this business scenario may include, but is not limited to, online shopping, borrowing money, repaying loans, or contacting customer service. These business scenario characteristics are not unique to the user.

[0064] Historical interaction data can refer to the user's historical inquiries in the customer service system, as well as the interaction data between the terminal and the user. This historical interaction data may include, but is not limited to, historical interaction text and interaction voice.

[0065] For example, user tag data, user behavior data, and historical interaction data can be stored on the server. The server directly retrieves the user tag data, user behavior data, and historical interaction data. The server then determines the characteristics of the business scenario based on the current location of the user as reported by the terminal.

[0066] Step 204: Establish baffle rules based on user tag data, user behavior data, and business scenario characteristics.

[0067] Here, a squeezing rule refers to a rule used to filter the original question. This squeezing rule consists of user-specific characteristics and characteristics that are not unique to the user.

[0068] For example, the business scenario in which the current user is located is determined based on the characteristics of the business scenario. Then, the operation that the current user is performing is determined based on the user behavior data. The problem encountered by the user is determined based on the user tag data. Finally, baffle rules are generated based on the user's business scenario, the operation that the user is performing, and the problem encountered by the user.

[0069] Step 206: Filter the original questions in the preset first question library according to the baffle rules to obtain the first candidate question set.

[0070] The first question database is pre-configured by the server and contains several original questions.

[0071] For example, the original questions in the first question library are filtered according to the baffle rules established in the aforementioned steps to obtain a first candidate question set that conforms to the baffle rules.

[0072] For example, a mapping relationship between the baffle rules and the first question base can be established, and then the first candidate question set can be determined based on the mapping relationship.

[0073] For example, when the user is an unregistered customer, the first set of candidate questions could be questions related to account registration and related business and function introductions.

[0074] Step 208: Determine target keywords based on user tag data, historical interaction data, and user behavior data.

[0075] For example, user tag data, historical interaction data, and user behavior data can be extracted and processed to obtain target keywords.

[0076] Step 210: Recall the first question database based on the target keywords to obtain the second candidate question set.

[0077] For example, target keywords can be input into the recall model to recall a first question base and obtain a second candidate question set.

[0078] For example, the BM25 algorithm is used to recall the first question database based on the target keyword in order to find the second candidate question set with high text similarity to the target keyword. Then, the second candidate question set is sorted according to the target keyword to obtain the sorted second candidate question set.

[0079] Step 212: Based on historical interaction data, the first candidate question set, and the second candidate question set, determine the first question sequence and push the first question sequence to the user's corresponding display interface.

[0080] The first question sequence can refer to the sequence of questions pushed to the user and displayed on the screen interface corresponding to the user's terminal. This first question sequence may include multiple questions.

[0081] The display interface can refer to the display interface corresponding to the terminal used by the user.

[0082] For example, several questions can be selected from the first candidate question set and the second candidate question set. Then, several questions can be selected from the popular question library based on historical interaction data. The selected questions are combined to obtain the first question sequence, and the first question sequence is pushed to the user's corresponding display interface.

[0083] The question push method in this application embodiment obtains user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location. Then, it establishes a baffle rule based on the user tag data, user behavior data, and business scenario characteristics. This makes the baffle rule highly correlated with the user tag data, user behavior data, historical interaction data, and business scenario characteristics of the user's current location, thereby improving the relevance between the subsequently selected first candidate question set and the question the user wants to consult, thus improving the accuracy of question push. Furthermore, it retrieves second candidate questions from the first question library based on target keywords, and determines a first question sequence based on historical interaction data, the first candidate questions, and the second candidate question set. Then, it pushes the first question sequence to the user, making it easier for the user to select the question they want to consult, further improving the accuracy of question push.

[0084] Please refer to Figure 3 In some embodiments, the problem push method also includes, but is not limited to, the following steps:

[0085] Step 302, in response to the first operation triggered on the display interface, determine the target problem selected by the user.

[0086] The first operation can refer to the user's action of selecting a question on the display interface. For example, the user clicks on the first question sequence on the display interface.

[0087] For example, the target question selected by the user is determined based on the user's click operation on the first question sequence on the display interface that displays the first question sequence.

[0088] Step 304: Classify and match the target question according to the preset classification question library to obtain the classification matching result.

[0089] The category question library refers to a set of questions used for categorizing and matching questions. This category question library is pre-configured and can be pre-set by the administrator on the server.

[0090] The classification matching results may include, but are not limited to, the first classification result used to indicate that the target question belongs to the classification question library and the second classification result used to indicate that the target question does not belong to the classification question library.

[0091] For example, the target question is classified and matched according to the classification question library to determine whether the target question belongs to the questions in the classification question library, and the classification result is obtained.

[0092] Step 306: Verify the user based on the classification matching results, or determine the question answer corresponding to the target question based on the user's user attribute information and a preset standard answer library.

[0093] User attribute information refers to a user's basic attribute information. This user attribute information may include the user's gender, age, etc.

[0094] The standard answer database can refer to the set of answers corresponding to a question. This standard answer database can correspond to a first question database and a second question database, and can include the answers to all questions in both databases.

[0095] For example, users can be verified based on classification matching results, or the question answer corresponding to the target question can be determined based on the user's user attribute information and a preset standard answer library.

[0096] For example, if the classification matching result indicates that the target question belongs to the first category of the classification question library, then the user is verified. If the classification result indicates that the target question does not belong to the second category of the classification question library, then the question answer corresponding to the target question is determined based on the user attribute information and the standard answer library.

[0097] For example, if the user is a 30-year-old male and the target question is "How to register an account", this target question does not belong to the question in the classification question library. The user attribute information includes 30 years old and male. Then, the question answer obtained based on the user attribute information and the representative answer library can be "Hello sir, please follow the instructions below to register an account".

[0098] For example, when the target question is "how to borrow money", it is necessary to perform security verification on the user, including but not limited to sending the user an identity authentication request and sending the user a password verification.

[0099] Please refer to Figure 4 In some embodiments, step 212 includes, but is not limited to, the following steps:

[0100] Step 402: Obtain the target problem set based on the first candidate problem set and the second candidate problem set.

[0101] The target problem set can refer to the set of problems in the first candidate problem set and the second candidate problem set.

[0102] For example, the first candidate problem set and the second candidate problem set can be combined to obtain the target problem set.

[0103] Step 404: Sort the target problem set and determine multiple first problems from the sorted target problem set.

[0104] For example, the questions in the target question set can be sorted according to the target keywords to obtain a sorted target question set, and multiple first questions can be selected from the sorted target question set.

[0105] For example, you can calculate the text similarity or relevance between the target keywords and the questions in the target question set, then sort them from high to low according to the text similarity or relevance to obtain a sorted target question set, and then select several first questions from the sorted target question set. For example, you can select the top three first questions in the sorted set.

[0106] Step 406: Identify multiple second questions from the second question library based on historical interaction data.

[0107] The second question bank is a pre-configured question bank. This second question bank can be a popular question bank or it can be pre-configured on the server by the administrator.

[0108] For example, multiple second questions can be selected from a popular question bank based on historical interaction data.

[0109] For example, based on the questions users ask in historical interaction data, similar or identical questions can be selected from a popular question library to obtain multiple second questions.

[0110] Step 408: Obtain the sequence of first questions by combining multiple first questions and multiple second questions.

[0111] For example, the multiple first questions and multiple second questions obtained from the aforementioned steps are combined to obtain a sequence of first questions.

[0112] The technical solution of this application embodiment improves the accuracy of question push by selecting multiple first questions in a first question library, selecting multiple second questions in a second question library, and combining the first and second questions to obtain a first question sequence. Furthermore, the first question sequence is a sorted sequence, which makes it easier for users to select target questions and improves the user experience.

[0113] In some embodiments, see Figure 5 In some embodiments, the problem push method also includes, but is not limited to, the following steps:

[0114] Step 502: In response to the second operation triggered on the display interface, or if no operation is triggered on the display interface within a preset time period, remove multiple first questions from the target question set and remove multiple second questions from the second question library.

[0115] The second operation can refer to a selection action triggered by the user on the display interface. This second operation could be the user clicking the "None" button on the display interface, or it could be the user clicking the "Next Batch" action on the display interface.

[0116] For example, if a user triggers a second action on the display interface, or if no action is triggered on the display interface within a preset time period, it indicates that the user is not satisfied with the first question sequence being pushed, or that the first question sequence does not contain the question the user wants to ask. In this case, the server removes multiple first questions from the target question set and multiple second questions from the second question library, so that questions can be re-pushed to the user.

[0117] It should be noted that the second question will not be completely deleted from the second question database. Instead, when pushing questions to the same user, the second question will no longer be recommended.

[0118] Step 504: Identify multiple third problems from the removed target problem set and multiple fourth problems from the removed second problem set.

[0119] Step 506: Combine multiple third questions and multiple fourth questions to obtain the second question sequence.

[0120] For example, the multiple third questions and multiple fourth questions determined in the aforementioned steps are combined to obtain a second question sequence.

[0121] Step 508: Push the second question sequence to the display interface.

[0122] For example, a second question sequence is pushed to the display interface of the terminal used by the same user, so that the user can select a question from the second question sequence.

[0123] It should be noted that for the target question selected in the second question sequence, it is also necessary to perform classification matching processing on the target question according to the classification question library to obtain the classification matching result, and then verify the user according to the classification matching result, or determine the question answer of the target question based on the user attribute information and the standard answer library.

[0124] The technical solution of this application embodiment will re-push a second question sequence when the first question sequence does not meet the questions the customer wants to consult, thereby improving the accuracy of question push.

[0125] In some embodiments, the problem push method may include, but is not limited to, the following steps: recording the number of problem pushes; if the number of problem pushes exceeds the push count threshold, then no more problem pushes will be made, and a transfer request will be sent to connect to human customer service.

[0126] The technical solution of this application embodiment stops pushing questions when the number of pushes exceeds the push number threshold, and transfers the user to a human customer service representative, which can improve the user experience.

[0127] In some embodiments, the problem push method further includes, but is not limited to, the following steps: obtaining operation information triggered by the user on the customer service system, and response information generated by the customer service system based on the operation information; determining user behavior data based on the operation information and response information.

[0128] The "operation information" refers to the actions performed by the user on the customer service system. The customer service system can be a system used to provide customer service consultation to users; it can be embedded in the terminal and communicate with the server, or it can be set up on the server with a customer service entry point on the terminal.

[0129] Response information can refer to the replies and responses generated by the customer service system in response to user actions.

[0130] For example, the server obtains the operation triggered by the user on the customer service system, as well as the response information generated by the customer service system for the operation information, and then determines the user's user behavior data based on the operation information and the response information.

[0131] Please see Figure 6 In some embodiments, the problem push method also includes, but is not limited to, the following steps:

[0132] Step 602: Obtain the user's historical interaction information.

[0133] Historical interaction information refers to the user's interaction information with the customer service system on the terminal. This historical interaction information may include the aforementioned historical interaction data and the user's current interaction information with the customer service system.

[0134] For example, the server obtains the current interaction information between the user and the customer service system, extracts the user's historical interaction data, and obtains historical interaction information based on the historical interaction data and the current interaction information.

[0135] Step 604: Convert historical interaction information into interaction vectors according to the preset pre-trained model.

[0136] The pre-trained model can refer to a model used to transform historical interaction information. This pre-trained model can be the RoFormer-Sim model, which integrates retrieval and generation.

[0137] For example, this historical interaction information can be input into the Roformer-sim model for processing to obtain the interaction vector.

[0138] Step 606: Obtain a cluster of questions based on the interaction vector mining; the cluster of questions includes at least one question.

[0139] For example, problem clusters can be obtained by mining interaction vectors using methods such as connected subgraphs and free solidification.

[0140] Step 608: Match each question in the question cluster according to the first question database, and add the matched question cluster to the first question database.

[0141] For example, the text similarity between the original questions in the first question library and each question in the question cluster can be calculated, and questions in the question cluster with a text similarity greater than the similarity threshold can be added to the first question library.

[0142] For example, the questions in the question cluster can be clustered and text generated to obtain a new batch of questions, and then these new questions can be added to the first question library.

[0143] In some embodiments, step 604 includes, but is not limited to, the following steps: cleaning and deduplicating historical interaction information to obtain target interaction information; and encoding the target interaction information into an interaction vector according to a pre-trained model.

[0144] Deduplication and cleaning can refer to operations such as deleting sensitive information, symbols, keywords, etc. from information. Sensitive information includes user identity information, passwords, etc.

[0145] For example, sensitive information, symbols, keywords, etc. in the historical interaction information are first cleaned and deduplicated to obtain the target interaction information, and then the target interaction information is input into the pre-trained model for encoding to obtain the interaction vector.

[0146] The technical solution of this application embodiment cleans and deduplicates historical interaction information, thereby facilitating the protection of users' sensitive information and improving the security of user data.

[0147] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0148] Based on the same inventive concept, this application also provides a problem-pushing device for implementing the problem-pushing method described above. The problem-solving solution provided by this device is similar to the solution described in the above method.

[0149] In one embodiment, such as Figure 7 As shown, a problem push device is provided, including: a data acquisition module 702, a rule establishment module 704, a filtering module 706, a target keyword determination module 708, a recall module 710, and a push module 712, wherein:

[0150] The data acquisition module 702 is used to acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location.

[0151] The rule creation module 704 is used to create baffle rules based on user tag data, user behavior data, and business scenario characteristics.

[0152] The filtering module 706 is used to filter the original questions in the preset first question library according to the baffle rules to obtain the first candidate question set.

[0153] The target keyword determination module 708 is used to determine target keywords based on user tag data, historical interaction data, and user behavior data.

[0154] The recall module 710 is used to recall the first question database based on the target keyword to obtain the second candidate question set.

[0155] The push module 712 is used to determine the first question sequence based on historical interaction data, the first candidate question set, and the second candidate question set, and push the first question sequence to the user's corresponding display interface.

[0156] In some embodiments, the problem push device further includes, but is not limited to:

[0157] The target question determination module is used to determine the target question selected by the user in response to the first operation triggered on the display interface.

[0158] The classification and matching module is used to classify and match target questions according to a preset classification question library to obtain classification and matching results.

[0159] The verification module is used to verify users based on classification matching results, or to determine the corresponding question answer based on user attribute information and a preset standard answer library.

[0160] In some embodiments, the push module includes:

[0161] The target problem set determination unit is used to obtain the target problem set based on the first candidate problem set and the second candidate problem set.

[0162] The first problem determination unit is used to sort the target problem set and determine multiple first problems from the sorted target problem set.

[0163] The second question determination unit is used to determine multiple second questions from the second question library based on historical interaction data.

[0164] A combination unit is used to combine multiple first problems and multiple second problems to obtain a sequence of first problems.

[0165] In some embodiments, the problem push device further includes, but is not limited to:

[0166] The elimination module is used to eliminate multiple first questions from the target question set and multiple second questions from the second question library in response to a second operation triggered on the display interface or if no operation is triggered on the display interface within a preset time period.

[0167] The problem identification module is used to identify multiple third problems from the removed target problem set and multiple fourth problems from the removed second problem set.

[0168] The combination module is used to combine multiple third problems and multiple fourth problems to obtain a sequence of second problems.

[0169] The second push module is used to push the second question sequence to the display interface.

[0170] In some embodiments, the problem push device further includes, but is not limited to:

[0171] The information acquisition module is used to acquire the operation information triggered by the user on the customer service system, as well as the response information generated by the customer service system based on the operation information.

[0172] The user behavior data determination module is used to determine user behavior data based on operation information and response information.

[0173] In some embodiments, the problem push device includes, but is not limited to:

[0174] The interaction information acquisition module is used to acquire the user's historical interaction information.

[0175] The conversion module is used to convert historical interaction information into interaction vectors based on a preset pre-trained model.

[0176] The mining module is used to mine question clusters based on interaction vectors; each question cluster includes at least one question.

[0177] The supplementary module is used to match each question in the question cluster with the first question database and add the matched question cluster to the first question database.

[0178] In some embodiments, the conversion module includes:

[0179] The cleaning and deduplication unit is used to clean and deduplicate historical interaction information to obtain the target interaction information.

[0180] The encoding module is used to encode target interaction information into interaction vectors based on the pre-trained model.

[0181] The modules in the aforementioned problem-pushing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0182] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a question-pushing method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0183] Those skilled in the art will understand that Figure 8The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0184] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the problem push method of the above embodiments.

[0185] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the problem push method of the above embodiment.

[0186] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the problem push method of the above embodiment.

[0187] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile 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. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0188] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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 specification.

[0189] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for pushing questions, characterized in that, The method includes: Acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location; Establish baffle rules based on the user tag data, user behavior data, and business scenario characteristics; Based on the mapping relationship established between the baffle rules and the preset first question library, the original questions in the preset first question library are filtered based on the mapping relationship to obtain a first candidate question set; Target keywords are determined based on the user tag data, the historical interaction data, and the user behavior data; Based on the target keyword, the first question database is retrieved to obtain a second candidate question set with high text similarity to the target keyword; A target question set is obtained based on the first candidate question set and the second candidate question set; the text similarity or relevance between the target keyword and the questions in the target question set is calculated, and the questions in the target question set are sorted from high to low according to the obtained text similarity or relevance to obtain a sorted target question set; multiple first questions are determined from the sorted target question set; multiple second questions are determined from a second question library based on historical interaction data, where the second question library is a popular question library; a first question sequence is obtained by combining the multiple first questions and the multiple second questions, and the first question sequence is pushed to the user's corresponding display interface.

2. The method according to claim 1, characterized in that, The method further includes: In response to a first operation triggered on the display interface, the target question selected by the user is determined; The target question is classified and matched according to a preset classification question library to obtain the classification matching result; The user is verified based on the classification and matching results, or the question answer corresponding to the target question is determined based on the user's user attribute information and a preset standard answer library.

3. The method according to claim 1, characterized in that, The method further includes: In response to a second operation triggered on the display interface, or if no operation is triggered on the display interface within a preset time period, the plurality of first questions are removed from the target question set, and the plurality of second questions are removed from the second question library; Multiple third questions are identified from the removed target question set, and multiple fourth questions are identified from the removed second question set; The plurality of third questions and the plurality of fourth questions are combined to obtain a second question sequence; The second question sequence is pushed to the display interface.

4. The method according to claim 1, characterized in that, Before acquiring the user's user tag data, user behavior data, historical interaction data, and business scenario features corresponding to the user's current location, the method further includes: Obtain the operation information triggered by the user on the customer service system, and the response information generated by the customer service system based on the operation information; The user behavior data is determined based on the operation information and the response information.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: Obtain the user's historical interaction information; The historical interaction information is converted into interaction vectors according to a preset pre-trained model; A cluster of questions is obtained by mining the interaction vectors; the cluster of questions includes at least one question. Each question in the question cluster is matched according to the first question database, and the matched question cluster is added to the first question database.

6. The method according to claim 5, characterized in that, The step of converting the historical interaction information into interaction vectors according to a preset pre-trained model includes: The historical interaction information is cleaned and deduplicated to obtain the target interaction information; The target interaction information is encoded into the interaction vector based on the pre-trained model.

7. A problem push device, characterized in that, The device includes: The data acquisition module is used to acquire user tag data, user behavior data, historical interaction data, and business scenario characteristics corresponding to the user's current location; The rule establishment module is used to establish baffle rules based on the user tag data, user behavior data, and business scenario characteristics. The filtering module is used to establish a mapping relationship between the baffle rules and a preset first question library, and to filter the original questions in the preset first question library based on the mapping relationship to obtain a first candidate question set; The target keyword determination module is used to determine target keywords based on the user tag data, the historical interaction data, and the user behavior data. The recall module is used to recall the first question database based on the target keyword to obtain a second candidate question set with high text similarity to the target keyword; The push module includes: The target problem set determination unit obtains the target problem set based on the first candidate problem set and the second candidate problem set; The first problem determination unit calculates the text similarity or relevance between the target keyword and the questions in the target problem set, sorts the questions in the target problem set from high to low according to the obtained text similarity or relevance, obtains the sorted target problem set, and determines multiple first problems from the sorted target problem set. The second question determination unit is used to determine multiple second questions from a second question database based on historical interaction data. The second question database is a popular question database. A combination unit is configured to combine the plurality of first questions and the plurality of second questions to obtain a sequence of first questions; The push module is used to push the first question sequence to the display interface corresponding to the user.

8. The apparatus according to claim 7, characterized in that, The problem push device also includes: The target question determination module is used to determine the target question selected by the user in response to a first operation triggered on the display interface; The classification and matching module is used to classify and match the target question according to a preset classification question library to obtain the classification and matching results; The verification module is used to verify the user based on the classification matching results, or to determine the question answer corresponding to the target question based on the user's user attribute information and a preset standard answer library.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

10. 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 steps of the method according to any one of claims 1 to 6.

Images