Information processing method, device and apparatus

Abstract

The embodiment of the present application discloses an information processing method, which includes: receiving current input content sent by a client device; performing user behavior feature analysis on the current input content to determine user intent information; if a target event map corresponding to the user intent information is determined, sending the target event map to the client device; wherein the target event map is used to be displayed in a display area corresponding to the client device, and the target event map has an association relationship with a target solution corresponding to the user intent information recommended for execution by the client device, including at least one possible event caused by executing the target solution. The embodiment of the present application also discloses an electronic device and apparatus.

Description

Technical Field

[0001] The present application relates to the field of information processing technology, and in particular to an information processing method, device and apparatus. Background Art

[0002] With the rapid development of computer and Internet technology, people's lives are becoming more and more intelligent. Intelligent applications have had a significant impact on people's lives, learning, entertainment, etc., providing people with convenient and fast services. In order to save labor costs, various service providers provide intelligent reply robots to give corresponding intelligent replies to people's questions.

[0003] However, at present, when people use intelligent dialogue systems, there is a problem that users are not actively implementing the solutions provided by the intelligent reply robots. That is, the credibility of the solutions provided by the intelligent robots is low, resulting in a high probability of transferring to manual service, causing the intelligent reply robots to have a low intelligence level.

[0004] Application Contents

[0005] In order to solve the above-mentioned technical problems, the embodiments of the present application hope to provide an information processing method, equipment and device, which solves the problem that the credibility of the solutions to problems provided by the current intelligent reply robots is low, and realizes a method for actively guiding users to implement the solutions to problems provided by the intelligent reply robots, effectively improving the credibility of the solutions to problems provided by the intelligent reply robots, reducing the probability of transferring to manual services, and improving the intelligence level of the intelligent reply robots.

[0006] The technical solution of this application is implemented as follows:

[0007] In a first aspect, an information processing method is provided, the method comprising:

[0008] Receive current input content sent by the client device;

[0009] Performing user behavior feature analysis on the current input content to determine user intent information;

[0010] If it is determined that a target event map corresponding to the user intention information is obtained, the target event map is sent to the client device; wherein the target event map is used to be displayed in a display area corresponding to the client device, and the target event map is associated with a target solution corresponding to the user intention information and recommended for execution by the client device, including at least one possible event caused by executing the target solution.

[0011] Optionally, before performing user behavior feature analysis on the current input content to determine user intent information, the method further includes:

[0012] Obtain a trained Bayesian network for identifying user intent analysis;

[0013] Correspondingly, the performing user behavior feature analysis on the current input content to determine user intention information includes:

[0014] Determine the first characteristic information included in the current input content; wherein the first characteristic information includes at least one of the following information: entity content keywords included in the current input content, sentence patterns of the current input content, emotional words included in the current input content, negative words included in the current input content, and numbers included in the current input content;

[0015] The user behavior feature analysis and processing is performed on the first feature information through the Bayesian network to obtain the user intention information.

[0016] Optionally, if it is determined that a target transaction graph corresponding to the user intention information is obtained, sending the target transaction graph to the client device includes:

[0017] If it is determined that a target solution corresponding to the user intention information is obtained, sending the target solution to the client device;

[0018] Based on the target solution, determine the target event map;

[0019] Send the target event graph to the client device.

[0020] Optionally, determining the target event graph based on the target solution includes:

[0021] If it is determined that the target solution is consistent with the solution provided by the historical manual service for the user intention information, the target matter graph corresponding to the target solution is determined from a preset graph library.

[0022] Optionally, after sending the target event graph to the client device, the method further includes:

[0023] If the first operation information corresponding to the target affair graph is not received from the client device within a preset time after the target affair graph is sent, or the second operation information indicating that the target solution is not executed is received from the client device, guidance information for guiding the user to execute the target solution is generated;

[0024] The guidance information is sent to the client device.

[0025] Optionally, the method further includes:

[0026] If it is determined that the target event graph corresponding to the user intention information is not obtained, it is determined that the current input content is not understood;

[0027] Determine the number of times the current input content is not understood in the current conversation;

[0028] If the number of times of not understanding is less than a preset number, determining second feature information of the current input content; wherein the second feature information is irrelevant to the semantic information of the current input content;

[0029] Determine target fallback reply content based on the current input content and the second feature information;

[0030] Send the target fallback reply content to the client device.

[0031] Optionally, the determining a target fallback reply content based on the current input content and the second feature information includes:

[0032] Determine the context content of the current input content, and determine the target stage to which the current input content belongs in the target dialogue process;

[0033] Based on the second characteristic information, the target fallback response content corresponding to the target stage is determined from a preset fallback response strategy.

[0034] Optionally, the method further includes:

[0035] If the number of misunderstandings is greater than or equal to the preset number, a manual service agent is determined;

[0036] A communication link is established between the client device and the manual service agent; wherein the manual service agent is used to provide manual service for the client device.

[0037] In a second aspect, an electronic device includes: a processor, a memory, and a communication bus; wherein:

[0038] The memory is used to store executable instructions;

[0039] The communication bus is used to realize the communication connection between the processor and the memory;

[0040] The processor is used to execute the information processing program stored in the memory to implement the steps of any of the above-mentioned information processing methods.

[0041] According to a third aspect, an information processing device is provided, the device comprising: a receiving unit, a processing unit and a first sending unit; wherein:

[0042] The receiving unit is used to receive the current input content sent by the client device;

[0043] The processing unit is used to perform user behavior feature analysis on the current input content to determine user intention information;

[0044] The first sending unit is used to send the target affair map to the client device if it is determined that the target affair map corresponding to the user intention information is obtained; wherein the target affair map is used to be displayed in a display area corresponding to the client device, and the target affair map is associated with a target solution corresponding to the user intention information recommended for the client device to execute, including at least one possible event caused by executing the operation corresponding to the target event.

[0045] The embodiment of the present application provides an information processing method, device and apparatus. After receiving the current input content sent by the client device, the current input content is analyzed and processed by the user behavior characteristics to determine the user intention information. If the target matter map corresponding to the user intention information is determined, the target matter map is sent to the client device. In this way, by sending the target matter map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the credibility of the solution provided by the current intelligent reply robot is low, and realizes a method for actively guiding users to implement the solution provided by the intelligent reply robot, which effectively improves the credibility of the solution provided by the intelligent reply robot, reduces the probability of transferring to manual service, and improves the intelligence of the intelligent reply robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A flowchart of an information processing method provided in an embodiment of the present application;

[0047] Figure 2 A flowchart of another information processing method provided in an embodiment of the present application;

[0048] Figure 3 A flowchart of another information processing method provided in an embodiment of the present application;

[0049] Figure 4 A schematic diagram of a target event map provided in an embodiment of the present application;

[0050] Figure 5 A flowchart of an information processing method provided by another embodiment of the present application;

[0051] Figure 6 A flowchart of another information processing method provided by another embodiment of the present application;

[0052] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0053] Figure 8 A schematic diagram of the structure of an information processing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0055] The embodiment of the present application provides an information processing method, referring to Figure 1 As shown, the method is applied to an electronic device, and the method comprises the following steps:

[0056] Step 101: Receive current input content sent by a client device.

[0057] In the embodiment of the present application, the client device is a device that is linked to the electronic device for communication. The client device can be various types of computer devices, such as desktop computers, notebook computers, or car computers, etc. The client device can also be an intelligent mobile terminal device, such as a smart phone and a tablet computer. The electronic device can be a server that provides problem-solving services for the client device. The electronic device can provide an intelligent reply dialogue system for the client device. When the intelligent reply robot in the intelligent reply dialogue system cannot solve the problem sent by the client device, the corresponding manual customer service can also provide corresponding services for the client device. The current input content sent by the client device can be text content or voice content.

[0058] Step 102: Perform user behavior feature analysis on the current input content to determine user intent information.

[0059] In the embodiment of the present application, the electronic device analyzes the current input content according to the preset user behavior feature analysis strategy to determine the user intention information, that is, analyzes the user behavior features included in the current input content to determine the user intention information according to the determined user behavior features.

[0060] Step 103: If it is determined that the target transaction graph corresponding to the user intention information is obtained, the target transaction graph is sent to the client device.

[0061] Among them, the target event map is used to be displayed in the display area corresponding to the client device. The target event map is associated with the target solution corresponding to the user intention information and recommended to be executed by the client device, including at least one possible event caused by executing the target solution.

[0062] In an embodiment of the present application, the target event graph can be a pre-generated target directed acyclic graph, in which the starting point is the target event corresponding to the target solution, and the corresponding end point is the possible event caused by the target solution. Correspondingly, the association relationship between the target event graph and the target solution is that the target event corresponding to the target solution is included in the target event graph. In some application scenarios, the association relationship between the target event graph and the target solution can also be that the target event graph is determined by the target solution. The target event graph is determined and constructed for the solution when the solution in which the intelligent reply robot replies to the user's intention information is consistent with the solution in which the manual customer service replies to the same user's intention information. Since the solution replied by the intelligent reply robot is consistent with that replied by the manual customer service, the generated target event graph is reliable. In this way, the target event graph is sent to the client device, which effectively guarantees the reliability of the target solution recommended for the user.

[0063] An information processing method provided by an embodiment of the present application performs user behavior feature analysis and processing on the current input content after receiving the current input content sent by the client device to determine the user intention information. If it is determined that the target matter map corresponding to the user intention information is obtained, the target matter map is sent to the client device. In this way, by sending the target matter map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the current intelligent reply robot provides low credibility of the solution to the problem, and realizes a method for actively guiding users to implement the solution to the problem provided by the intelligent reply robot, which effectively improves the credibility of the solution to the problem provided by the intelligent reply robot, reduces the probability of transferring to manual service, and improves the intelligence of the intelligent reply robot.

[0064] Based on the above embodiments, the embodiments of the present application provide an information processing method, referring to Figure 2 As shown, the method is applied to an electronic device, and the method comprises the following steps:

[0065] Step 201: Receive current input content sent by a client device.

[0066] In an embodiment of the present application, the current input content sent by the client device is taken as text content as an example. The user enters the current input text content in the client device. After obtaining the current input content, after the user clicks the send button, the client device sends the current input content to the electronic device.

[0067] Step 202: Obtain a trained Bayesian network for identifying user intent analysis.

[0068] In an embodiment of the present application, an electronic device obtains a trained Bayesian network for identifying user intention analysis to perform user behavior feature analysis on the current input content.

[0069] Among them, the trained Bayesian network can be obtained by analyzing historical sample information of known user intention information.

[0070] Step 203: Determine the first feature information included in the current input content.

[0071] Among them, the first feature information includes at least one of the following information: entity content keywords included in the current input content, sentence pattern of the current input content, emotion words included in the current input content, negative words included in the current input content, numbers included in the current input content.

[0072] In an embodiment of the present application, the first feature information can be one or more of the above-listed information. The entity content keywords included in the current input content can be, for example, keywords related to user intention information in the current input content, which can be obtained by keyword extraction. The sentence pattern of the current input content can be determined according to the characteristics of sentence patterns such as declarative sentences, interrogative sentences, and exclamatory sentences. For example, it can be determined according to the punctuation marks or specific modal particles included in the sentences of the current input content. For example, it can be determined that the current input content is an interrogative sentence by the punctuation mark "?" included in the sentence and / or the modal particle "ma", and it can be determined that the current input content is an exclamatory sentence by the punctuation mark "!" included in the sentence and / or the modal particle "a". The emotion words included in the current input content can be obtained by extracting the words expressing emotions in the current input content. The negative words included in the current input content can be, for example, words expressing negative meanings such as "no", "not", "negation", "No".

[0073] Step 204: Perform user behavior feature analysis processing on the first feature information through the Bayesian network to obtain user intention information.

[0074] In an embodiment of the present application, by inputting the first feature information into the Bayesian network, the user intention information can be output.

[0075] Step 205: If the target事理图谱 corresponding to the user intention information is determined, send the target事理图谱 to the client device.

[0076] Among them, the target事理图谱 is used to be displayed in the display area corresponding to the client device. The target事理图谱 has an association relationship with the target solution corresponding to the user intention information recommended for the client device to execute, and includes at least one possible event that occurs after executing the target solution.

[0077] It should be noted that "事理图谱" in the original text seems to be a specific term in Chinese that may not have a direct and common English equivalent. Here, it is directly retained in the translation. If there is a more accurate English expression for this term, it can be adjusted accordingly.In an embodiment of the present application, after the electronic device determines that the user's intention information is obtained, it determines the target event map corresponding to the user's intention information from a large amount of event map information stored in the electronic device, and then sends the target event map to the client device. After the client device receives the target event map sent by the electronic device, it can directly display the target event map in the display area corresponding to the client device. However, in some application scenarios, the client device can detect whether it includes all possible events caused by executing the target solution included in the target event map. If some of the events are missing, the corresponding possible events can be deleted and displayed.

[0078] Alternatively, in some application scenarios, after the electronic device determines that the user's intention information is obtained, it determines the reference event map corresponding to the user's intention information from the large amount of event map information stored in the electronic device, and then determines whether the client device includes the object corresponding to the possible event included in the reference event map. If the client device does not include the object corresponding to the possible event, the corresponding possible event in the reference event map is deleted to obtain the target event map, and the obtained target event map is sent to the client device. In this way, the possible events that do not exist in the displayed map can effectively reduce the misleading of users, resulting in misunderstandings, and improve the user experience.

[0079] Based on the above embodiments, in other embodiments of the present application, step 205 can be implemented by steps 205a to 205c:

[0080] Step 205a: If it is determined that a target solution corresponding to the user intention information is obtained, the target solution is sent to the client device.

[0081] In an embodiment of the present application, after determining that the user intent information is obtained, it is first determined whether there is a target solution corresponding to the user intent information. If there is a target solution corresponding to the user intent information, the target solution is sent to the client device so that the target solution is displayed in the corresponding display area of ​​the client device. In this way, after the user sees the target solution sent by the client device, the corresponding operation can be performed according to the target solution.

[0082] Step 205b: Determine the target event graph based on the target solution.

[0083] In the embodiment of the present application, after the electronic device determines the target solution, it determines the target matter graph corresponding to the target solution from the solution and matter graph library.

[0084] Step 205c, sending the target transaction graph to the client device.

[0085] In the embodiment of the present application, the target event graph can be sent to the client device after the timing time is a certain time. The certain time is determined according to the time required for the user to execute the target solution to determine the experience value. Alternatively, steps 205b to 205c are executed after the timing time is a certain time, that is, step 205b is executed after a certain time, and then step 205c is sent to the client device immediately after determining that the target instance graph is obtained.

[0086] Based on the foregoing embodiments, in other embodiments of the present application, step 205b can be implemented by the following steps: if it is determined that the target solution is consistent with the solution provided by the historical manual service for the user intent information, determine the target matter graph corresponding to the target solution from the preset graph library.

[0087] In an embodiment of the present application, a large number of solutions provided for different user intention information in historical manual service processes are stored in the storage area of ​​the electronic device. In this way, after determining the target solution, the electronic device determines the solution corresponding to the user intention information determined this time in the historical manual service process, that is, the historical manual solution, and determines whether the two are consistent. If they are consistent, the target matter graph corresponding to the target solution is determined.

[0088] Alternatively, if it is determined that a target solution corresponding to the user intent information is obtained, determine whether the target solution is consistent with the solution provided by historical manual services for the user intent information; if the target solution is consistent with the solution provided by historical manual services for the user intent information, send the target solution to the client device; based on the target solution, determine the target matter graph; and send the target matter graph to the client device.

[0089] Based on the above embodiments, in other embodiments of the present application, refer to Figure 3 As shown, after the electronic device executes step 205, it is also used to execute steps 206 to 207:

[0090] Step 206: If the first operation information corresponding to the target affair graph is not received from the client device within a preset time after sending the target affair graph, or the second operation information sent by the client device indicating that the target solution is not executed is received, guidance information is generated to guide the user to execute the target solution.

[0091] In an embodiment of the present application, the preset time length is an empirical value obtained based on actual experience. If the first operation information corresponding to the target affair graph sent by the client device is not received within the preset time length after the target affair graph is sent, it is determined that the user has not executed the target solution. In this way, when it is determined that the user has not executed the target solution, guidance information is generated. For example, the guidance information may be "Execute the operation corresponding to the target solution, and there is a 90% chance that it can solve the problem you need to solve" and other similar words.

[0092] Step 207: Send the guidance information to the client device.

[0093] In an embodiment of the present application, guidance information is sent to a client device and displayed in the client device to guide the user to perform operations corresponding to the target solution.

[0094] Based on the above embodiments, the present application provides an information processing method, and the implementation process can refer to the following steps, including:

[0095] Step a11: The electronic device obtains historical sample data.

[0096] The historical sample data is each user's historical input information in the historical conversation log, and the user behavior information determined for each historical input information. The user behavior information is the aforementioned user intention information.

[0097] Step a12: The electronic device uses non-semantic information, such as entities, sentence patterns, emotional words, negative words, numbers, keyword parts of speech, and punctuation marks, to segment each historical input information and obtain user behavior features corresponding to the determined user behavior information.

[0098] Among them, the corresponding sentence patterns may include: interrogative sentences, declarative sentences, exclamatory sentences, affirmative sentences, negative sentences and imperative sentences, etc.; keyword parts of speech may include: nouns, verbs, adjectives, numerals and interrogative words, etc.; emotional words may include: positive emotional words, negative emotional words and neutral emotional words, etc.; entities may include: names of time, place and frequency; punctuation marks may include: hello, comma, exclamation mark and period, etc.

[0099] User behavior information may include, for example: feedback on the last service, thanking the smart reply robot, not understanding the answer, describing the problem, the answer is useless, or describing a new problem.

[0100] Step a13: The electronic device obtains a trained Bayesian network based on a large amount of determined user behavior features and user behavior information.

[0101] Among them, the random variables involved in a certain research system are plotted in a directed graph according to whether they are conditionally independent, thus forming a Bayesian network. The Bayesian network is mainly used to describe the conditional dependencies between random variables, with circles representing random variables and arrows representing conditional dependencies. In an embodiment of the present application, the random variables included in the Bayesian network are different user behavior characteristics and different user behavior information, and the direction of the arrows in the conditional dependency relationship is that different user behavior characteristics point to different user behavior information.

[0102] It should be noted that steps a11 to a13 can be implemented as an independent embodiment.

[0103] Step a14: the electronic device receives current input content currently input by the user.

[0104] For example, the current input content currently input by the user is "the client device is stuck".

[0105] Step a15: The electronic device analyzes the current input content through a Bayesian network to determine the user behavior information.

[0106] Among them, when the current input content is analyzed by the Bayesian network, a variety of different user behavior information can be determined in the Bayesian network for the current input content. Take the user behavior information Act1 and the corresponding user behavior features x1, x2, x3, x4 and x5 as an example to illustrate. According to the formula The probability value corresponding to the user behavior information Act1 can be calculated when the user behavior features are x1, x2, x3, x4 and x5. Similarly, the probabilities corresponding to other different user behavior information can be calculated; then, the user behavior information Acti corresponding to the maximum probability value is determined as the user behavior information from the multiple probabilities p calculated.

[0107] Step a16: The electronic device determines a target solution corresponding to the user behavior information.

[0108] Exemplarily, the electronic device determines that the target resolution solution corresponding to the user behavior information requires a restart operation on the client device.

[0109] Step a18: The electronic device determines the corresponding target event graph based on the target solution.

[0110] For example, the target event map can be determined by referring to Figure 4 As shown in the figure, restart is the target solution, and the corresponding application shutdown, event 1, event 2, and event 3 are possible events that may occur due to restart.

[0111] The process of obtaining the event graph in advance can be as follows:

[0112] Step b11, the electronic device obtains the conversation data of the same user behavior intention under which the intelligent reply robot and the manual customer service provide the same solution; wherein, the intelligent reply robot provides k solutions for the same user behavior intention in the reply conversation, which can be specifically recorded as X = [x1, ...x k ], there are m solutions provided by the manual service reply dialogue for the same user behavior intention, which can be specifically recorded as Y = [y1,…y m ].

[0113] Step b12: The electronic device performs vector representation on the data in X and Y: Each row in each vector represents a solution. For example, in the vector corresponding to X, the first row represents the first solution, and the kth row represents the kth solution.

[0114] Step b13: The electronic device calculates the relationship between the intelligent reply robot reply dialogue and the manual service reply dialogue.

[0115] Specifically, the formula can be used: To calculate the relationship between two conversations, corr is used to represent the relationship between the i-th solution replied by the intelligent reply robot and the j-th solution replied by the manual customer service. (x i1 … x in )(y j1 … y jn ) represents the calculation of the vector product between the i-th solution of the intelligent reply robot and the j-th solution of the manual customer service, ||x i || represents the norm of the matrix.

[0116] Step b14: If the calculated corr value is greater than the threshold, the electronic device believes that the solution of the intelligent reply robot is consistent with that of the manual service, and can determine that the solution is a reliable solution to the user's behavioral intention.

[0117] Step b15: The electronic device determines a reliable solution to the user's behavioral intention and possible events that may occur after the reliable solution is executed.

[0118] Step b16: The electronic device establishes a cause-and-effect graph based on the reliable solution and the possible events corresponding to the reliable solution.

[0119] Among them, the event graph is a logical knowledge base that describes the evolution rules and patterns between events. Structurally, the event graph is a directed graph, in which nodes represent events and directed edges represent logical relationships between events such as sequence, cause and effect, condition, and hierarchy.

[0120] Step a19: the electronic device sends the target event graph to the client device so that the client device displays it to the user.

[0121] Step a20: The electronic device receives the operation information for the target event graph sent by the client device, and learns whether the user executes the target solution.

[0122] Step a21: If it is determined that the user is executing the target solution, further guide the user to execute the target solution again to help the user solve the problem.

[0123] Among them, if the user has executed the target solution, the conversation can be ended, or the user can be further asked whether the problem has been successfully solved and other follow-up operations.

[0124] It should be noted that, for the description of the same steps and the same contents in this embodiment as those in other embodiments, reference can be made to the description in other embodiments and will not be repeated here.

[0125] An information processing method provided by an embodiment of the present application performs user behavior feature analysis and processing on the current input content after receiving the current input content sent by the client device to determine the user intention information. If it is determined that the target matter map corresponding to the user intention information is obtained, the target matter map is sent to the client device. In this way, by sending the target matter map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the current intelligent reply robot provides low credibility of the solution to the problem, and realizes a method for actively guiding users to implement the solution to the problem provided by the intelligent reply robot, which effectively improves the credibility of the solution to the problem provided by the intelligent reply robot, reduces the probability of transferring to manual service, and improves the intelligence of the intelligent reply robot.

[0126] Based on the above embodiments, the embodiments of the present application provide an information processing method, referring to Figure 5 As shown, the method is applied to an electronic device, and the method comprises the following steps:

[0127] Step 301: Receive current input content sent by a client device.

[0128] Step 302: Perform user behavior feature analysis on the current input content to determine user intent information.

[0129] In the embodiment of the present application, step 302 can be implemented by steps 202 to 204, which will not be described in detail here. The specific implementation process can refer to the implementation process of steps 202 to 204.

[0130] Step 303: If the target event graph corresponding to the user intention information is not obtained, it is determined that the current input content is not understood.

[0131] In an embodiment of the present application, the user intent information may be empty or not empty, but when the corresponding target instance map does not currently exist in the electronic device, it is determined that the electronic device does not understand the current input content. That is, the user's current behavior cannot be accurately identified. When the user intent information is not empty, if it is counted that the electronic device replies to the user's current input content with "Please rephrase your question" or "Would you please rephrase your question?" and other similar descriptions, it can be determined that the electronic device does not understand the current input content.

[0132] Step 304: Determine the number of times the current input content is not understood in the current dialogue.

[0133] In an embodiment of the present application, the electronic device will record the misunderstandings in the current conversation, so that the number of misunderstandings of the current input content in the current conversation can be counted. In some application scenarios, when counting the number of misunderstandings of the input content that is not understood continuously, it can be counted the number of replies of the electronic device to the content of "Please rephrase your question" or "Would you please rephrase your question?" and other similar descriptions. Among them, the reply object to which the electronic device continuously replies with "Please rephrase your question" or "Would you please rephraseyour question?" and other similar descriptions, that is, the input content of the user, can be the same content or different content.

[0134] Step 305: If the number of times of misunderstanding is less than a preset number, determine the second characteristic information of the current input content.

[0135] The second feature information is irrelevant to the semantic information of the current input content.

[0136] In the embodiment of the present application, the preset number of times may be a value set by the user, which can usually be changed accordingly according to actual conditions and is not a fixed value. The second characteristic information may be the number of sentences, sentence patterns, word count, etc. included in the current input content.

[0137] Step 306: Determine the target catch-all reply content based on the current input content and the second feature information.

[0138] In an embodiment of the present application, the target fallback reply content is determined from a preset strategy based on the current input content and the second feature information.

[0139] Step 307: Send the target fallback reply content to the client device.

[0140] In the embodiment of the present application, the target catch-all reply content is sent to the client device, and after the client device receives the target catch-all reply content, the target catch-all reply content is displayed. In this way, in the embodiment of the present application, different catch-all reply contents are determined based on the specific current input content and the second feature information to reply to the user, instead of simply repeating the same sentence to the user, which effectively improves the user's experience and reduces the probability of the user choosing to switch to manual service.

[0141] In other embodiments of the present application, step 306 may be implemented by steps 306a to 306b:

[0142] Step 306a: Determine the context of the current input content, and determine the target stage to which the current input content belongs in the target dialogue process.

[0143] In the embodiment of the present application, in order to make the bottom-up strategy more targeted, the dialogue can be divided into five stages according to the process of solving the problem:

[0144] The first stage is: the starting stage.

[0145] Among them, the period before entering into the main topic or business issues is considered the starting stage.

[0146] The second stage is: intention selection stage.

[0147] Among them, in the intention selection stage, the user has expressed his or her intention, and the electronic device has identified the user's intention according to its own understanding and introduced several solution options for the user to choose.

[0148] The third stage is the slot filling stage.

[0149] In the slot filling stage, in order to know the user's true intention, it is necessary to ask the user to provide some key information. This process of asking the user to cooperate in providing key information is called slot filling.

[0150] The fourth stage is: pushing answers.

[0151] After determining the user's question, the electronic device helps the user to find the answer to the current question and pushes the found answer in a certain order.

[0152] The fifth stage is: transferring to manual or pushing questionnaires.

[0153] Among them, after the user's question has been pushed, the answer to the query has been pushed to the user, and the user is asked to fill out the questionnaire, or because the user's problem has not been solved, it enters the stage of manual handling.

[0154] It should be noted that the above five stages do not exist in every conversation process. There may be only one or several stages, which can be determined according to the actual application scenario.

[0155] Step 306b: Based on the second characteristic information, determine the target fallback response content corresponding to the target stage from the preset fallback response strategy.

[0156] In the embodiment of the present application, corresponding to the same second characteristic information, the target catch-all reply content set may be different in different target stages.

[0157] Based on the above embodiments, in other embodiments of the present application, refer to Figure 6 As shown, after the electronic device executes step 304, it may also choose to execute steps 308 to 309:

[0158] Step 308: If the number of misunderstandings is greater than or equal to the preset number, determine a manual service agent.

[0159] Step 309: Establish a communication link between the client device and the manual service agent.

[0160] Among them, the manual service seat is used to provide manual service for client devices.

[0161] In an embodiment of the present application, after the input content sent by the client device is not understood for a certain number of times, a manual customer service is assigned to the client device and the client is directly switched to the manual customer service so that the manual customer service provides corresponding manual services.

[0162] Based on the foregoing embodiment, the embodiment of the present application provides a specific implementation process of determining the second characteristic information of the current input content, which can refer to the following steps:

[0163] Step c11: The electronic device determines whether the current input content includes at least two sentences.

[0164] Step c12: If the current input content includes at least two sentences, the electronic device determines that the feature of the current input content is a paragraph.

[0165] When the current input content includes at least two sentences, ie, multiple sentences, the second characteristic information is a paragraph.

[0166] Step c13: If the current input content is not at least two sentences, determine whether the current input content is one sentence.

[0167] Step c14: If the current input content is a sentence, determine the sentence pattern of the current input content.

[0168] The sentence pattern may be a question sentence, an exclamatory sentence or a declarative sentence, etc. When the current input content is a sentence, that is, a single sentence, the second characteristic information is the sentence pattern of the current input content.

[0169] Step c15: If the current input content is not a sentence, determine whether the current input content is a word.

[0170] Step c16: If the current input content is a word, determine the number of words.

[0171] Step c17: If the number of words is one, determine the attributes of the words in the current input content.

[0172] The attribute of the word may be a common word, a question word, an interjection word or a quantifier. When the number of the word is one, the second characteristic information is the attribute of the word.

[0173] Step c18: If the number of words is at least two, determine that the current input content is multiple words.

[0174] When the number of words is at least two, the second characteristic information is multiple words.

[0175] Step c19: If the current input content is not a word, determine that the current input content is a character symbol.

[0176] Step c20: Determine the category of the character symbols included in the current input content.

[0177] Step c21: If the category of the character symbols included in the current input content is a single category, determine the specific category.

[0178] The specific categories are letter categories, punctuation mark categories, and numbers. When the category of the character symbols included in the current input content is a single category, the second characteristic information is a specific category.

[0179] Step c22: If the character symbols included in the current input content are of multiple categories, determine that the current input content is other.

[0180] When the character symbols included in the current input content are of multiple categories, the second characteristic information is others.

[0181] Based on the second characteristic information determined above and the different stages described above, the corresponding preset fallback response strategy can be shown in Table 1 below.

[0182] Among them, the notes corresponding to the strategies of each stage in Table 1 are: Re-brief description, applicable to each stage, used to instruct the user to re-describe the problem in concise and brief language; Pull back, the pull back strategies are generally set differently in each stage, used to prompt the user to repeat the previously required operations, for example, in the slot filling stage, the corresponding catch-all response strategy when pulling back is: Please help fill the slot; Start, used to instruct the user to start over; Switch to manual service, used for currently switched to manual service, available in all stages; Prompt to continue operation, used to prompt the user how to continue the operation, such as how to ask questions or how to fill in the slot filling value, for example, the catch-all response is "Please use the following questioning method to describe your problem", or in some cases, when the user's question content has been preliminarily determined but is not sure, the corresponding catch-all response content can be "Is the problem you want to solve a problem with your mobile phone" and so on.

[0183] Table 1 Preset fallback response strategy

[0184]

[0185] It should be noted that, for the description of the same steps and the same contents in this embodiment as those in other embodiments, reference can be made to the description in other embodiments and will not be repeated here.

[0186] An information processing method provided by an embodiment of the present application performs user behavior feature analysis and processing on the current input content after receiving the current input content sent by the client device, determines the user intention information, and if it is determined that the target thing map corresponding to the user intention information is obtained, sends the target thing map to the client device. In this way, by sending the target thing map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the current intelligent reply robot provides a low credibility of the solution to the problem, and realizes a method for actively guiding users to implement the solution to the problem provided by the intelligent reply robot, effectively improving the credibility of the solution to the problem provided by the intelligent reply robot, reducing the probability of transferring to manual service, and improving the intelligence of the intelligent reply robot. Furthermore, different bottom-up reply strategies are provided for different characteristic information at different stages, which effectively improves the naturalness of the conversation between the intelligent robot and the user, improves the dialogue efficiency between the user and the intelligent reply robot, and improves the user experience.

[0187] Based on the above embodiments, the embodiments of the present application provide an electronic device, which can be applied to Figures 1 to 3 and Figures 5-6 In the information processing method provided in the corresponding embodiment, refer to Figure 7 As shown, the electronic device 4 may include: a processor 41, a memory 42 and a communication bus 43, wherein:

[0188] A memory 42, for storing executable instructions;

[0189] A communication bus 43, used to realize the communication connection between the processor 41 and the memory 42;

[0190] The processor 41 is used to execute the information processing program stored in the memory 42 to implement the following steps:

[0191] Receive current input content sent by the client device;

[0192] Analyze the user behavior characteristics of the current input content to determine the user intention information;

[0193] If it is determined that a target event map corresponding to the user intention information is obtained, the target event map is sent to the client device; wherein the target event map is used to be displayed in a display area corresponding to the client device, and the target event map is associated with a target solution corresponding to the user intention information that is recommended for execution by the client device, including at least one possible event caused by executing the target solution.

[0194] In other embodiments of the present application, before the processor executes the step of performing user behavior feature analysis on the current input content and determining the user intent information, it is also used to execute the following steps:

[0195] Obtain a trained Bayesian network for identifying user intent analysis;

[0196] Correspondingly, the processor executes the step of analyzing the user behavior characteristics of the current input content to determine the user intention information, which can be achieved through the following steps:

[0197] Determine the first characteristic information included in the current input content; wherein the first characteristic information includes at least one of the following information: entity content keywords included in the current input content, sentence patterns of the current input content, emotional words included in the current input content, negative words included in the current input content, and numbers included in the current input content;

[0198] The user behavior characteristics of the first feature information are analyzed and processed through a Bayesian network to obtain user intention information.

[0199] In other embodiments of the present application, if the processor executes the step of determining that the target event graph corresponding to the user intention information is obtained and sending the target event graph to the client device, it can be implemented by the following steps:

[0200] If it is determined that a target solution corresponding to the user intent information is obtained, the target solution is sent to the client device;

[0201] Based on the target solution, determine the target event map;

[0202] Send the target event graph to the client device.

[0203] In other embodiments of the present application, when the processor executes the step based on the target solution and determines the target event graph, it can be implemented by the following steps:

[0204] If it is determined that the target solution is consistent with the solution provided by historical manual services for user intent information, the target matter graph corresponding to the target solution is determined from the preset graph library.

[0205] In other embodiments of the present application, after the processor executes the step of sending the target event graph to the client device, it is further used to execute the following steps:

[0206] If the first operation information corresponding to the target business map is not received from the client device within a preset time after the target business map is sent, or the second operation information indicating that the target solution is not executed is received from the client device, guidance information for guiding the user to execute the target solution is generated;

[0207] Sends guidance information to the client device.

[0208] In other embodiments of the present application, the processor is further configured to perform the following steps:

[0209] If the target event graph corresponding to the user's intention information is not obtained, it is determined that the current input content is not understood;

[0210] Determine the number of times that the current input content is not understood in the current dialogue;

[0211] If the number of times of not understanding is less than the preset number, determining the second characteristic information of the current input content; wherein the second characteristic information is irrelevant to the semantic information of the current input content;

[0212] Based on the current input content and the second feature information, determine the target fallback reply content;

[0213] Send the target fallback reply content to the client device.

[0214] In other embodiments of the present application, when the processor executes the step of determining the target fallback reply content based on the current input content and the second feature information, it can be implemented by the following steps:

[0215] Determine the context of the current input content and determine the target stage to which the current input content belongs in the target dialogue process;

[0216] Based on the second characteristic information, the target fallback response content corresponding to the target stage is determined from the preset fallback response strategy.

[0217] In other embodiments of the present application, the processor may also be configured to perform the following steps:

[0218] If the number of misunderstandings is greater than or equal to the preset number, determine the manual service agent;

[0219] A communication link is established between the client device and a human service agent, wherein the human service agent is used to provide human service for the client device.

[0220] It should be noted that the specific implementation process of the steps executed by the processor in this embodiment can refer to Figures 1 to 3 and Figures 5-6 The implementation process of the information processing method provided in the corresponding embodiment will not be repeated here.

[0221] An electronic device provided in an embodiment of the present application performs user behavior feature analysis and processing on the current input content after receiving the current input content sent by the client device to determine the user intention information. If it is determined that the target matter map corresponding to the user intention information is obtained, the target matter map is sent to the client device. In this way, by sending the target matter map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the current intelligent reply robot provides low credibility of the solution to the problem, and realizes a method for actively guiding the user to implement the solution to the problem provided by the intelligent reply robot, which effectively improves the credibility of the solution to the problem provided by the intelligent reply robot, reduces the probability of transferring to manual service, and improves the intelligence of the intelligent reply robot.

[0222] Based on the above embodiments, the embodiments of the present application provide an information processing device, which can be applied to Figures 1 to 3 and Figures 5-6 In the information processing method provided in the corresponding embodiment, refer to Figure 8 As shown, the information processing device 5 may include: a receiving unit 51, a processing unit 52 and a first sending unit 53, wherein:

[0223] The receiving unit 51 is used to receive the current input content sent by the client device;

[0224] The processing unit 52 is used to analyze the user behavior characteristics of the current input content and determine the user intention information;

[0225] The first sending unit 53 is used to send the target affair map to the client device if it is determined that the target affair map corresponds to the user intention information; wherein the target affair map is used to be displayed in the display area corresponding to the client device, and the target affair map is associated with the target solution corresponding to the user intention information that is recommended to be executed by the client device, including at least one possible event caused by executing the operation corresponding to the target event.

[0226] In other embodiments of the present application, before the processing unit, the information processing device further includes an acquisition unit; wherein:

[0227] An acquisition unit, used for acquiring a trained Bayesian network for identifying user intent analysis;

[0228] Correspondingly, the processing unit includes: a determination module and a processing module; wherein:

[0229] A determination module, used to determine the first characteristic information included in the current input content; wherein the first characteristic information includes at least one of the following information: entity content keywords included in the current input content, sentence patterns of the current input content, emotional words included in the current input content, negative words included in the current input content, and numbers included in the current input content;

[0230] The processing module is used to perform user behavior feature analysis on the first feature information through a Bayesian network to obtain user intention information.

[0231] In other embodiments of the present application, the determination module is specifically used to implement the following steps:

[0232] If it is determined that a target solution corresponding to the user intent information is obtained, the target solution is sent to the client device;

[0233] Based on the target solution, determine the target event map;

[0234] Send the target event graph to the client device.

[0235] In other embodiments of the present application, the determination module is used to implement the step based on the target solution, and when determining the target event map, it can be implemented by the following steps:

[0236] If it is determined that the target solution is consistent with the solution provided by historical manual services for user intent information, the target matter graph corresponding to the target solution is determined from the preset graph library.

[0237] In other embodiments of the present application, after the first sending unit, the information processing device includes: a generating unit and a second sending unit; wherein:

[0238] A generating unit, configured to generate guiding information for guiding a user to execute the target solution if the first operation information corresponding to the target issue map sent by the client device is not received within a preset time period after the target issue map is sent, or if the second operation information sent by the client device indicating that the target solution is not executed is received;

[0239] The second sending unit is used to send the guidance information to the client device.

[0240] In other embodiments of the present application,

[0241] The processing unit is further configured to determine that the current input content is not understood if the target event graph corresponding to the user intention information is not determined;

[0242] The processing unit is further used to determine the number of times that the current input content is not understood in the current dialogue, which is a continuous incomprehension of the input content;

[0243] The processing unit is further configured to determine second characteristic information of the current input content if the number of times of not understanding is less than a preset number; wherein the second characteristic information is irrelevant to the semantic information of the current input content;

[0244] The processing unit is further used to determine the target fallback reply content based on the current input content and the second feature information;

[0245] The first sending unit is further used to send the target fallback reply content to the client device.

[0246] In other embodiments of the present application, when the processor unit is used to execute the step of determining the target fallback reply content based on the current input content and the second feature information, it can be implemented by the following steps:

[0247] Determine the context of the current input content and determine the target stage to which the current input content belongs in the target dialogue process;

[0248] Based on the second characteristic information, the target fallback response content corresponding to the target stage is determined from the preset fallback response strategy.

[0249] In other embodiments of the present application, the information processing device further includes: a communication unit, wherein:

[0250] The processing unit is further configured to determine a manual service agent if the number of times of not understanding is greater than or equal to a preset number of times;

[0251] The communication unit is used to establish a communication link between the client device and the manual service agent; wherein the manual service agent is used to provide manual service for the client device.

[0252] It should be noted that the specific implementation process of the steps performed by the units and modules in this embodiment can be referred to Figures 1 to 3 and Figures 5-6 The implementation process of the information processing method provided in the corresponding embodiment will not be repeated here.

[0253] An information processing device provided by an embodiment of the present application performs user behavior feature analysis and processing on the current input content after receiving the current input content sent by the client device to determine the user intention information. If it is determined that the target matter map corresponding to the user intention information is obtained, the target matter map is sent to the client device. In this way, by sending the target matter map corresponding to the user intention information to the client device, the user is actively guided to execute the recommended target solution, which solves the problem that the current intelligent reply robot provides low credibility of the solution to the problem, and realizes a method for actively guiding users to implement the solution to the problem provided by the intelligent reply robot, which effectively improves the credibility of the solution to the problem provided by the intelligent reply robot, reduces the probability of transferring to manual service, and improves the intelligence of the intelligent reply robot.

[0254] Based on the foregoing embodiments, the embodiments of the present application provide a computer-readable storage medium, referred to as a storage medium, which stores one or more programs to implement the following Figures 1 to 3 and Figures 5-6 The implementation process of the information processing method provided in the corresponding embodiment will not be repeated here.

[0255] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of hardware embodiments, software embodiments, or embodiments in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) that contain computer-usable program code.

[0256] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0257] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0258] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0259] The above description is only a preferred embodiment of the present application and is not intended to limit the protection scope of the present application.

Claims

1. An information processing method, the method comprising: Receive current input content sent by the client device; Performing user behavior feature analysis on the current input content to determine user intent information; If it is determined that a target event map corresponding to the user intention information is obtained, the target event map is sent to the client device; wherein the target event map is used to be displayed in a display area corresponding to the client device, and the target event map has an associated relationship with a target solution corresponding to the user intention information and recommended for execution by the client device, including at least one possible event caused by executing the target solution; If the first operation information corresponding to the target affair graph is not received from the client device within a preset time after the target affair graph is sent, or the second operation information indicating that the target solution is not executed is received from the client device, guidance information for guiding the user to execute the target solution is generated; The guidance information is sent to the client device.

2. The method according to claim 1, before performing user behavior feature analysis on the current input content to determine user intent information, the method further comprises: Obtain a trained Bayesian network for identifying user intent analysis; Correspondingly, the performing user behavior feature analysis on the current input content to determine user intention information includes: Determine the first characteristic information included in the current input content; wherein the first characteristic information includes at least one of the following information: entity content keywords included in the current input content, sentence patterns of the current input content, emotional words included in the current input content, negative words included in the current input content, and numbers included in the current input content; The user behavior feature analysis and processing is performed on the first feature information through the Bayesian network to obtain the user intention information.

3. According to the method of claim 2, if it is determined that the target event graph corresponding to the user intention information is obtained, sending the target event graph to the client device comprises: If it is determined that a target solution corresponding to the user intention information is obtained, sending the target solution to the client device; Based on the target solution, determine the target event map; Send the target event graph to the client device.

4. The method according to claim 3, wherein determining the target event graph based on the target solution comprises: If it is determined that the target solution is consistent with the solution provided by the historical manual service for the user intention information, the target matter graph corresponding to the target solution is determined from a preset graph library.

5. The method according to claim 1, further comprising: If it is determined that the target event graph corresponding to the user intention information is not obtained, it is determined that the current input content is not understood; Determine the number of times the current input content is not understood in the current conversation; If the number of times of not understanding is less than a preset number, determining second feature information of the current input content; wherein the second feature information is irrelevant to the semantic information of the current input content; Determine target fallback reply content based on the current input content and the second feature information; Send the target fallback reply content to the client device.

6. The method according to claim 5, wherein determining the target fallback reply content based on the current input content and the second feature information comprises: Determine the context content of the current input content, and determine the target stage to which the current input content belongs in the target dialogue process; Based on the second characteristic information, the target fallback response content corresponding to the target stage is determined from a preset fallback response strategy.

7. The method according to claim 5, further comprising: If the number of misunderstandings is greater than or equal to the preset number, a manual service agent is determined; A communication link is established between the client device and the manual service agent, wherein the manual service agent is used to provide manual service for the client device.

8. An electronic device, comprising: A processor, a memory, and a communication bus; wherein: The memory is used to store executable instructions; The communication bus is used to realize the communication connection between the processor and the memory; The processor is used to execute the information processing program stored in the memory to implement the steps of the information processing method according to any one of claims 1 to 7.

9. An information processing device, comprising: A receiving unit, a processing unit and a first sending unit; wherein: The receiving unit is used to receive the current input content sent by the client device; The processing unit is used to perform user behavior feature analysis on the current input content to determine user intention information; The first sending unit is used to send the target event map to the client device if it is determined that the target event map corresponding to the user intention information is obtained; wherein the target event map is used to be displayed in a display area corresponding to the client device, and the target event map has an association relationship with the target solution corresponding to the user intention information recommended for the client device to execute, including at least one possible event caused by executing the operation corresponding to the target event; A generating unit, configured to generate guiding information for guiding the user to execute the target solution if the first operation information corresponding to the target issue map sent by the client device is not received within a preset time period after the target issue map is sent, or the second operation information sent by the client device indicating that the target solution is not executed is received; The second sending unit is used to send the guiding information to the client device.

Images