A transaction dialogue breakpoint takeover control method and device

Abstract

The present invention provides a transaction dialogue breakpoint takeover control method and device, which relates to the field of artificial intelligence technology and can be used in the financial field or other technical fields. The method includes: obtaining static transaction information and the intention value of the transaction dialogue counterpart in each time stage during the transaction dialogue process of an intelligent dialogue robot; processing the static information and the intention value in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point; polling and monitoring each transaction breakpoint probability value, and controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur based on the polling and monitoring results. The device executes the above method. The transaction dialogue breakpoint takeover control method and device provided in the embodiment of the present invention can facilitate the completion of transactions as soon as possible and avoid missing favorable transaction opportunities.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a transaction dialogue breakpoint takeover control method and device. Background Art

[0002] When enterprises use intelligent bot technology to conduct multiple rounds of price inquiry dialogues with external institutional traders, various reasons can lead to interruptions in the intelligent price inquiry process. These interruptions typically require the caller to manually transfer the call to a human operator at our bank. This delay, coupled with the time it takes for traders to read and understand the information provided, can hinder efficient transaction completion due to the ever-changing market conditions. Summary of the Invention

[0003] In response to the problems in the prior art, embodiments of the present invention provide a transaction session breakpoint takeover control method and device, which can at least partially solve the problems in the prior art.

[0004] In one aspect, the present invention provides a transaction session breakpoint takeover control method, comprising:

[0005] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0006] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0007] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0008] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0009] The step of obtaining the intention value of the transaction dialogue counterparty at each time stage includes:

[0010] Perform text feature extraction on the transaction dialogue to obtain text features corresponding to each segmentation in the transaction dialogue, perform intent recognition on each text feature, and obtain intent values ​​corresponding to each segmentation;

[0011] Based on the preset transaction dialogue intention recognition model, the intention value corresponding to each segmentation is identified to obtain the intention value of the transaction dialogue counterpart at each time stage;

[0012] Among them, the preset transaction conversation intention recognition model is obtained by training a neural network model based on transaction conversation intention recognition sample data, and can be used to obtain intention values ​​in the communication access time period, demand communication time period and demand finalization time period respectively.

[0013] The step of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0014] A transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is used as the target transaction dialogue time point, and the trader is accessed at the target transaction dialogue time point.

[0015] The step of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0016] If it is determined that the probability values ​​of each transaction breakpoint are less than or equal to the preset transaction breakpoint probability value, and the time period of the current transaction conversation has not changed, then monitoring the change in the intention value in the current polling monitoring cycle;

[0017] If it is determined that the change in the intention value is greater than the preset numerical change, the polling monitoring of each transaction breakpoint probability value will continue until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than the preset transaction breakpoint probability value is monitored, and the transaction dialogue time point at this time will be used as the target transaction dialogue time point, and the trader will be connected at the target transaction dialogue time point.

[0018] The transaction conversation volume is also monitored; accordingly, the transaction conversation breakpoint takeover control method further includes:

[0019] Switching the time phase of the transaction conversation based on information about changes in the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases;

[0020] After completing the time phase switching, continue to poll and monitor each transaction breakpoint probability value until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is detected, and use the transaction dialogue time point at this time as the target transaction dialogue time point, and access the trader at the target transaction dialogue time point.

[0021] The switching of the time phase of the transaction conversation according to the change information of the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases includes:

[0022] If it is determined that the change information of the conversation volume in the communication access time period continues to increase, and the first time period of the continuous increase in the change information is greater than the first preset duration threshold corresponding to the communication access time period, the communication access time period is switched to the required communication time period.

[0023] The switching of the time phase of the transaction conversation according to the change information of the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases includes:

[0024] If it is determined that the change information of the conversation volume in the demand communication time period continues to increase, and the second time period in which the change information continues to increase is greater than the second preset duration threshold corresponding to the demand communication time period, the demand communication time period is switched to the demand finalization time period.

[0025] In one aspect, the present invention provides a transaction session breakpoint takeover control device, comprising:

[0026] An acquisition unit is configured to acquire static transaction information and the intention value of the transaction counterparty in each time period during a transaction dialogue conducted by the intelligent dialogue robot; the time periods include a communication access period, a demand communication period, and a demand finalization period;

[0027] a prediction unit, configured to process the static information and the intention value in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain a transaction breakpoint probability value corresponding to each transaction dialogue time point;

[0028] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0029] The control unit is used to poll and monitor the probability values ​​of each transaction breakpoint, and control the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling and monitoring results.

[0030] On the other hand, an embodiment of the present invention provides an electronic device, comprising: a processor, a memory, and a bus, wherein:

[0031] The processor and the memory communicate with each other via the bus;

[0032] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the following method:

[0033] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0034] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0035] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0036] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0037] An embodiment of the present invention provides a non-transitory computer-readable storage medium, including:

[0038] The non-transitory computer-readable storage medium stores computer instructions, which cause the computer to execute the following method:

[0039] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0040] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0041] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0042] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0043] The transaction dialogue breakpoint takeover control method and device provided by the embodiment of the present invention obtain static information of the transaction and the intention value of the transaction dialogue counterpart in each time stage during the process of the transaction dialogue conducted by the intelligent dialogue robot; the time stages include the communication access time period, the demand communication time period and the demand finalization time period; based on the preset transaction dialogue breakpoint prediction model, the static information and the intention value in each time stage are processed to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; wherein, the preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data; the probability value of each transaction breakpoint is polled and monitored, and the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur is controlled according to the polling and monitoring results, which can help to complete the transaction as soon as possible and avoid missing favorable transaction opportunities. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0045] Figure 1 The figure is a flow chart of a transaction session breakpoint takeover control method provided by one embodiment of the present invention.

[0046] Figure 2 It is a structural diagram of a transaction session breakpoint takeover control device provided by an embodiment of the present invention.

[0047] Figure 3 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0048] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of this application can be combined with each other in any manner.

[0049] Figure 1 FIG. 1 is a flow chart of a transaction session breakpoint takeover control method provided by an embodiment of the present invention. Figure 1 As shown, the transaction session breakpoint takeover control method provided by the embodiment of the present invention includes:

[0050] Step S1: During a transaction dialogue conducted by the intelligent dialogue robot, static transaction information and the intention value of the transaction dialogue counterpart in each time stage are obtained; the time stages include a communication access time period, a demand communication time period, and a demand finalization time period.

[0051] Step S2: Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0052] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data.

[0053] Step S3: polling and monitoring the probability values ​​of each transaction breakpoint, and controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling and monitoring results.

[0054] In step S1, the device acquires static transaction information and the transaction counterpart's intent values ​​during various time periods during a transaction conversation with the intelligent conversational robot. These time periods include a communication access period, a demand communication period, and a demand finalization period. The device can be, for example, a computer device that executes the method. It should be noted that all data used in this embodiment of the present invention is user-authorized.

[0055] Because both parties to a transaction must use real names, static transaction information can be returned through the system interface. This information can include the trader's name, ID, counterparty, product type, access time, and other information. This objective information can be directly obtained and recorded in the trading system, while the backend system can be called to obtain more static information, such as historical transaction gross amounts, historical transaction net amounts, total number of transactions, number of successful manual transactions, manual transaction success rate, number of successful robot transactions, robot transaction success rate, and transaction data for the past three, six, and twelve months.

[0056] Credit limit, credit utilization, external rating, internal rating, etc.;

[0057] Available quota, occupied quota, level of risk capital occupied, etc.;

[0058] Current position, credit risk level, bond rating, etc.

[0059] Dynamic information can include conversational information. The basic process involves using natural language processing (NLPS) technology to identify elements and transactions, generate a text-based intended quotation, and then push it. After confirmation, a formatted intended quotation is pushed. This also supports proactive pricing notifications, order splitting reminders, proactive price inquiries, and retained intent.

[0060] The information sent by our intelligent conversation robot and the information replied by the opponent are captured as conversation information.

[0061] The relevant conversation information is processed by NLP as the input parameter of the model, and the intent value corresponding to each word is obtained through NLP processing.

[0062] According to the progress of transaction communication time, it mainly goes through three stages, namely the communication access period, the demand communication period and the demand finalization period.

[0063] Obtaining the intention value of the transaction dialogue counterpart in each time stage, including:

[0064] Text feature extraction is performed on transaction conversations to obtain text features corresponding to each word in the transaction conversation. Intent recognition is performed on each text feature to obtain the intent value corresponding to each word. A word vector-based feature extraction model is applied. Based on a large text corpus and trained with a model similar to a neural network, each word is mapped into a fixed-dimensional vector. Each vector represents this word, and the meaning and grammatical similarity of the word are judged by the similarity between vectors.

[0065] NLP methods are used for dictionary-based analysis. Different intents have different domain-specific dictionaries. Judgment is based on the degree of match or overlap between the intent and the dictionary. Common dictionaries include SentiWordNet and AFINN. Each word in the intent dictionary is assigned a value ranging from -1.0 to 1.0 based on its degree of significance. This value represents the intent value.

[0066] Based on the preset transaction conversation intention recognition model, the intent values ​​corresponding to each segmentation are identified to obtain the transaction conversation counterparty's intent values ​​for each time period. The intent values ​​corresponding to each segmentation can be input into the preset transaction conversation intention recognition model, and the output of the preset transaction conversation intention recognition model is used as the transaction conversation counterparty's intent values ​​for each time period. First, the intent values ​​corresponding to each segmentation are divided into different time periods, and then the weighted average of the intent values ​​divided into different time periods is taken to obtain the intent values ​​for each time period. The weighted weights are denoted as W1 to W3, corresponding to the three different time periods. Taking the intent value c1 corresponding to segmentation x1 as an example, c1 is divided into the communication access time period, the demand communication time period, and the demand finalization time period. The weighted average of the intent of segmentation x1 is calculated according to the following formula:

[0067]

[0068] For each word segment, the calculation shown in word segmentation x1 is performed to obtain the intention value in each time stage.

[0069] The preset transaction conversation intent recognition model is derived by training a neural network model based on sample transaction conversation intent recognition data and can be used to obtain intent values ​​during the communication access period, the demand communication period, and the demand finalization period. The specific type of neural network model is not specifically limited. It is understood that after the preset transaction conversation intent recognition model is trained, the optimized weighted intent values ​​of each word segment at different time periods can be obtained, thereby accurately obtaining the intent values ​​of the transaction conversation counterparty at each time period.

[0070] Before extracting text features from transaction conversations, text preprocessing can be performed, which can include:

[0071] Word segmentation: Apply a dictionary-based word segmentation method to split the sentence into multiple parts in a unified dictionary table, and match each part with the dictionary one by one. If the word is in the dictionary, the word segmentation is successful, otherwise continue word segmentation until a match is successful.

[0072] Remove stop words: Create a stop word dictionary, including some adverbs, adjectives, and conjunctions. By maintaining a stop word list, unnecessary stop words can be removed.

[0073] In the above step S2, the device processes the static information and the intention value in each time stage based on the preset transaction dialogue breakpoint prediction model to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; the static information and the intention value in each time stage can be input into the preset transaction dialogue breakpoint prediction model, and the output result of the preset transaction dialogue breakpoint prediction model is used as the transaction breakpoint probability value corresponding to each transaction dialogue time point. Each transaction breakpoint probability value can be recorded in time series and can be expressed as {P1, P2…PN}, which corresponds to each transaction dialogue time point t1, t2…tN corresponding to each transaction dialogue time point in chronological order.

[0074] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data.

[0075] The weak classifier of the GBDT (Gradient Boosting Decision Tree) model is iteratively trained to obtain the optimal model. Based on GBDT, the LightGBM model can achieve highly efficient parallel training, with the advantages of faster training speed, lower memory consumption, better accuracy, and support for distributed processing of massive data.

[0076] This application scenario involves a large number of input parameters, with a high number of missing values ​​across different data points. LightGBM is well-suited for model training. Training is performed using historical basic information, conversation information, and result information to form a pre-defined transaction conversation breakpoint prediction model. Therefore, the methods of this embodiment of the present invention can utilize LightGBM as the aforementioned machine learning model.

[0077] In step S3, the device polls and monitors the probability values ​​of each transaction breakpoint. Based on the polling and monitoring results, the device controls the trader who will take over the intelligent conversational robot at the target transaction conversation time point when the transaction breakpoint is predicted to occur. A historical conversation message may be displayed to the trader, indicating that the trader has taken over the conversation. After reading the message, the trader enters the message and clicks "Send," which takes over the transaction conversation process.

[0078] The method of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0079] The trading session time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is set as the target trading session time point, and the trader is connected at the target trading session time point. The preset transaction breakpoint probability value can be set independently based on actual circumstances. It is understood that as the trading session progresses, if the transaction breakpoint probability value of the current trading session time point is found to be greater than the preset transaction breakpoint probability value, the current trading session time point is set as the target trading session time point, and the trader is connected at this time point.

[0080] The method of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0081] If it is determined that the probability values ​​of each transaction breakpoint are less than or equal to the preset transaction breakpoint probability value, and the time period of the current transaction dialogue has not changed, then the change in the intention value in the current polling monitoring cycle is monitored; it can be understood that since the transaction dialogue is ongoing, the number of transaction breakpoint probability values ​​monitored by polling in the current polling monitoring cycle is greater than the number of transaction breakpoint probability values ​​monitored by polling in the previous polling monitoring cycle.

[0082] Whether the time stage of the current transaction dialogue has changed can be determined based on the keywords corresponding to the time stage. If the current transaction dialogue hits at least M keywords corresponding to the required communication time period, and the previous transaction dialogue was in the communication access time period, it means that the time stage of the current transaction dialogue has changed from the communication access time period to the required communication time period.

[0083] If the current transaction conversation matches fewer than M keywords corresponding to the required communication time period, and the previous transaction conversation was in the communication access time period, it means that the current transaction conversation is still in the communication access time period.

[0084] If the change in the intent value is determined to be greater than a preset value, the polling monitoring of each transaction breakpoint probability value continues until a transaction conversation time point corresponding to a transaction breakpoint probability value greater than the preset value is detected. This transaction conversation time point is then used as the target transaction conversation time point, and the trader is connected at the target transaction conversation time point. The preset value can be set based on actual circumstances.

[0085] It can be understood that if the change in the intention value is greater than the preset numerical change, it means that the intention value in each time stage, which is one of the inputs of the preset transaction dialogue breakpoint prediction model, has undergone a large change, which is likely to affect the output result of the preset transaction dialogue breakpoint prediction model. Therefore, the probability value of each transaction breakpoint is polled and monitored in a timely manner to ensure that the transaction breakpoint probability value corresponding to each transaction dialogue time point is accurately obtained.

[0086] If it is determined that the change in the intention value is less than or equal to the preset numerical change, the polling monitoring of the probability values ​​of each transaction breakpoint can be suspended. When the transaction dialogue breakpoint takeover control method is periodically executed and the step of polling monitoring the probability values ​​of each transaction breakpoint is executed, the polling monitoring of the probability values ​​of each transaction breakpoint is started again.

[0087] The conversation volume of the transaction conversation is also monitored; the conversation volume can be represented by the number of conversation sentences.

[0088] Accordingly, the transaction session breakpoint takeover control method further includes:

[0089] Switching the time phase of a transaction conversation based on information about changes in the conversation volume of transaction conversations during a current polling monitoring cycle, the time phase of the transaction conversation, and preset duration thresholds corresponding to the respective time phases; switching the time phase of a transaction conversation based on information about changes in the conversation volume of transaction conversations during a current polling monitoring cycle, the time phase of the transaction conversation, and the respective preset duration thresholds corresponding to the respective time phases, includes:

[0090] If it is determined that the change information of the conversation volume during the communication access time period continues to increase, and the first time period during which the change information continues to increase is greater than the first preset duration threshold corresponding to the communication access time period, the communication access time period is switched to the required communication time period. The first preset duration threshold can be set independently based on actual circumstances. That is, if the conversation volume gradually increases for a long period of time during the communication access time period, it indicates that there may be changes in the trading market, which has led to a large amount of communication between the two parties regarding the trading market changes, resulting in a continuous increase in the change information of the conversation volume. Switching the communication access time period to the required communication time period is conducive to facilitating the completion of the transaction as soon as possible and avoiding missing favorable trading opportunities.

[0091] The switching of the time phase of the transaction dialogue according to the change information of the conversation volume of the transaction dialogue in the current polling monitoring cycle, the time phase of the transaction dialogue, and the preset duration thresholds corresponding to the time phases includes:

[0092] If it is determined that the change information of the conversation volume in the demand communication time period continues to increase, and the second time period during which the change information continues to increase is greater than the second preset time threshold corresponding to the demand communication time period, then the demand communication time period is switched to the demand finalization time period. The second preset time threshold can be independently set according to actual conditions, and the second preset time threshold can be the same as or different from the first preset time threshold. That is, if the conversation volume gradually increases for a long time during the demand communication time period, it indicates that there may be changes in the trading market, which has led to a large amount of communication between the two parties regarding the changes in the trading market, resulting in a continuous increase in the change information of the conversation volume. Switching the demand communication time period to the demand finalization time period is conducive to facilitating the completion of the transaction as soon as possible and avoiding missing favorable trading opportunities.

[0093] After completing the time phase switching, continue to poll and monitor each transaction breakpoint probability value until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is detected, and use the transaction dialogue time point at this time as the target transaction dialogue time point, and access the trader at the target transaction dialogue time point.

[0094] It should be noted that if it is determined that the change information of the conversation volume in the demand finalization time period continues to increase, there is no need to switch the time phase.

[0095] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention obtains static information of the transaction and the intention value of the transaction dialogue counterpart in each time stage during the process of the transaction dialogue conducted by the intelligent dialogue robot; the time stages include the communication access time period, the demand communication time period and the demand finalization time period; based on the preset transaction dialogue breakpoint prediction model, the static information and the intention value in each time stage are processed to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; wherein, the preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data; the probability value of each transaction breakpoint is polled and monitored, and the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur is controlled according to the polling and monitoring results, which can help to complete the transaction as soon as possible and avoid missing the favorable transaction opportunity.

[0096] Furthermore, obtaining the intention value of the transaction dialogue counterpart in each time stage includes:

[0097] Perform text feature extraction on the transaction dialogue to obtain text features corresponding to each word in the transaction dialogue, perform intent recognition on each text feature to obtain intent values ​​corresponding to each word; refer to the above embodiment for description and no further details will be given.

[0098] Based on the preset transaction dialogue intention recognition model, the intention value corresponding to each word segment is identified to obtain the intention value of the transaction dialogue counterpart in each time stage; please refer to the above embodiment for description and no further details will be given.

[0099] The preset transaction conversation intent recognition model is obtained by training a neural network model based on transaction conversation intent recognition sample data and can be used to obtain intent values ​​during the communication access period, demand communication period, and demand finalization period. This description can be referred to the above embodiment and will not be repeated here.

[0100] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention can accurately obtain the intention value of the transaction dialogue counterpart in each time stage.

[0101] Furthermore, the control of accessing and taking over the trader of the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0102] The transaction conversation time point corresponding to the transaction breakpoint probability value greater than the preset transaction breakpoint probability value is used as the target transaction conversation time point, and the trader is connected at the target transaction conversation time point. The above embodiment can be referred to for explanation and will not be repeated here.

[0103] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention can accurately determine the target transaction dialogue time point, thereby facilitating the completion of the transaction as soon as possible and avoiding missing favorable transaction opportunities.

[0104] Furthermore, the control of accessing and taking over the trader of the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes:

[0105] If it is determined that the probability values ​​of each transaction breakpoint are less than or equal to the preset transaction breakpoint probability value, and the time period of the current transaction dialogue has not changed, then the change in the intention value in the current polling monitoring cycle is monitored; please refer to the above embodiment for description and no further details will be given.

[0106] If it is determined that the change in the intent value is greater than the preset value change, the polling monitoring of each transaction breakpoint probability value continues until a transaction conversation time point corresponding to a transaction breakpoint probability value greater than the preset transaction breakpoint probability value is detected. This transaction conversation time point is then used as the target transaction conversation time point, and the trader is connected at the target transaction conversation time point. This description can be made with reference to the above embodiment and will not be repeated here.

[0107] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention can accurately determine the target transaction dialogue time point, thereby facilitating the completion of the transaction as soon as possible and avoiding missing favorable transaction opportunities.

[0108] Furthermore, the transaction conversation volume is also monitored; accordingly, the transaction conversation breakpoint takeover control method further includes:

[0109] The time phase of the transaction conversation is switched based on the change information of the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation and the preset duration threshold corresponding to the time phase; please refer to the above embodiment for description and no further details will be given.

[0110] After the time phase switch is completed, the transaction breakpoint probability values ​​are continuously polled and monitored until a transaction session time point corresponding to a transaction breakpoint probability value greater than the preset transaction breakpoint probability value is detected. This transaction session time point is then used as the target transaction session time point, and the trader is connected at the target transaction session time point. This description can be referred to the above embodiment and will not be repeated here.

[0111] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0112] Furthermore, switching the time phase of the transaction conversation based on the change information of the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases includes:

[0113] If it is determined that the change information of the conversation volume in the communication access time period continues to increase, and the first time period of the continuous increase in the change information is greater than the first preset duration threshold corresponding to the communication access time period, the communication access time period is switched to the required communication time period. This can be explained with reference to the above embodiment and will not be repeated here.

[0114] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0115] Furthermore, switching the time phase of the transaction conversation based on the change information of the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases includes:

[0116] If it is determined that the change information of the conversation volume in the required communication time period continues to increase, and the second time period during which the change information continues to increase is greater than a second preset duration threshold corresponding to the required communication time period, the required communication time period is switched to the required finalization time period. This can be explained with reference to the above embodiment and will not be repeated here.

[0117] The transaction dialogue breakpoint takeover control method provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0118] It should be noted that the transaction dialogue breakpoint takeover control method provided in the embodiment of the present invention can be used in the financial field, and can also be used in any technical field other than the financial field. The embodiment of the present invention does not limit the application field of the transaction dialogue breakpoint takeover control method.

[0119] Figure 2 FIG. 1 is a schematic diagram of a transaction session breakpoint takeover control device according to an embodiment of the present invention. Figure 2 As shown, the transaction session breakpoint takeover control device provided by the embodiment of the present invention includes an acquisition unit 201, a prediction unit 202 and a control unit 203, wherein:

[0120] The acquisition unit 201 is used to obtain the static information of the transaction and the intention value of the transaction dialogue partner in each time stage during the transaction dialogue process of the intelligent dialogue robot; the time stages include the communication access time period, the demand communication time period and the demand finalization time period; the prediction unit 202 is used to process the static information and the intention value in each time stage based on the preset transaction dialogue breakpoint prediction model to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; wherein, the preset transaction dialogue breakpoint prediction model is obtained by training the machine learning model based on the transaction dialogue breakpoint prediction sample data; the control unit 203 is used to poll and monitor the probability value of each transaction breakpoint, and control the trader who accesses and takes over the intelligent dialogue robot at the target transaction dialogue time point when the predicted transaction breakpoint occurs according to the polling monitoring result.

[0121] Specifically, the acquisition unit 201 in the device is used to obtain the static information of the transaction and the intention value of the other party in the transaction dialogue in each time stage during the transaction dialogue of the intelligent dialogue robot; the time stages include the communication access time period, the demand communication time period and the demand finalization time period; the prediction unit 202 is used to process the static information and the intention value in each time stage based on the preset transaction dialogue breakpoint prediction model to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; wherein, the preset transaction dialogue breakpoint prediction model is obtained by training the machine learning model based on the transaction dialogue breakpoint prediction sample data; the control unit 203 is used to poll and monitor the probability value of each transaction breakpoint, and control the trader who accesses and takes over the intelligent dialogue robot at the target transaction dialogue time point when the predicted transaction breakpoint occurs according to the polling monitoring result.

[0122] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention obtains static information of the transaction and the intention value of the transaction dialogue counterpart in each time stage during the process of the transaction dialogue conducted by the intelligent dialogue robot; the time stages include the communication access time period, the demand communication time period and the demand finalization time period; based on the preset transaction dialogue breakpoint prediction model, the static information and the intention value in each time stage are processed to obtain the transaction breakpoint probability value corresponding to each transaction dialogue time point; wherein, the preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data; the probability value of each transaction breakpoint is polled and monitored, and the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur is controlled according to the polling and monitoring results, which can help to complete the transaction as soon as possible and avoid missing the favorable transaction opportunity.

[0123] Furthermore, the acquiring unit 201 is specifically configured to:

[0124] Perform text feature extraction on the transaction dialogue to obtain text features corresponding to each segmentation in the transaction dialogue, perform intent recognition on each text feature, and obtain intent values ​​corresponding to each segmentation;

[0125] Based on the preset transaction dialogue intention recognition model, the intention value corresponding to each segmentation is identified to obtain the intention value of the transaction dialogue counterpart at each time stage;

[0126] Among them, the preset transaction conversation intention recognition model is obtained by training a neural network model based on transaction conversation intention recognition sample data, and can be used to obtain intention values ​​in the communication access time period, demand communication time period and demand finalization time period respectively.

[0127] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention can accurately obtain the intention value of the transaction dialogue counterpart in each time stage.

[0128] Furthermore, the control unit 203 is specifically configured to:

[0129] A transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is used as the target transaction dialogue time point, and the trader is accessed at the target transaction dialogue time point.

[0130] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention can accurately determine the target transaction dialogue time point, thereby facilitating the completion of the transaction as soon as possible and avoiding missing favorable transaction opportunities.

[0131] Furthermore, the control unit 203 is further specifically configured to:

[0132] If it is determined that the probability values ​​of each transaction breakpoint are less than or equal to the preset transaction breakpoint probability value, and the time period of the current transaction conversation has not changed, then monitoring the change in the intention value in the current polling monitoring cycle;

[0133] If it is determined that the change in the intention value is greater than the preset numerical change, the polling monitoring of each transaction breakpoint probability value will continue until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than the preset transaction breakpoint probability value is monitored, and the transaction dialogue time point at this time will be used as the target transaction dialogue time point, and the trader will be connected at the target transaction dialogue time point.

[0134] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention can accurately determine the target transaction dialogue time point, thereby facilitating the completion of the transaction as soon as possible and avoiding missing favorable transaction opportunities.

[0135] Furthermore, the transaction conversation volume is also monitored; accordingly, the transaction conversation breakpoint takeover control device is also used to:

[0136] Switching the time phase of the transaction conversation based on information about changes in the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases;

[0137] After completing the time phase switching, continue to poll and monitor each transaction breakpoint probability value until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is detected, and use the transaction dialogue time point at this time as the target transaction dialogue time point, and access the trader at the target transaction dialogue time point.

[0138] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0139] Furthermore, the transaction session breakpoint takeover control device is further specifically configured to:

[0140] If it is determined that the change information of the conversation volume in the communication access time period continues to increase, and the first time period of the continuous increase in the change information is greater than the first preset duration threshold corresponding to the communication access time period, the communication access time period is switched to the required communication time period.

[0141] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0142] Furthermore, the transaction session breakpoint takeover control device is further specifically configured to:

[0143] If it is determined that the change information of the conversation volume in the demand communication time period continues to increase, and the second time period in which the change information continues to increase is greater than the second preset duration threshold corresponding to the demand communication time period, the demand communication time period is switched to the demand finalization time period.

[0144] The transaction dialogue breakpoint takeover control device provided by the embodiment of the present invention is further conducive to facilitating the completion of transactions as quickly as possible and avoiding missing favorable transaction opportunities.

[0145] The embodiment of the transaction session breakpoint takeover control device provided by the embodiment of the present invention can be specifically used to execute the processing flow of the above-mentioned method embodiments. Its functions are not described in detail here, and reference can be made to the detailed description of the above-mentioned method embodiments.

[0146] Figure 3 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 3As shown, the electronic device includes: a processor 301, a memory 302 and a bus 303;

[0147] The processor 301 and the memory 302 communicate with each other via the bus 303.

[0148] The processor 301 is configured to call the program instructions in the memory 302 to execute the methods provided by the above method embodiments, for example, including:

[0149] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0150] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0151] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0152] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0153] This embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can perform the methods provided in the above-mentioned method embodiments, for example, including:

[0154] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0155] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0156] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0157] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0158] This embodiment provides a computer-readable storage medium storing a computer program. The computer program enables the computer to execute the methods provided in the above method embodiments, for example, including:

[0159] During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart in each time stage; the time stages include the communication access time period, the demand communication time period, and the demand finalization time period;

[0160] Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point;

[0161] The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data;

[0162] Polling and monitoring are performed on the probability values ​​of each transaction breakpoint, and according to the polling and monitoring results, the trader who takes over the intelligent dialogue robot is controlled to access the target transaction dialogue time point when the transaction breakpoint is predicted to occur.

[0163] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0164] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, 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 produce 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 flowcharts and / or block diagrams. 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.

[0165] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work 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 The function specified in one or more boxes.

[0166] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0167] Throughout this specification, reference to terms such as "one embodiment," "a specific embodiment," "some embodiments," "for example," "example," "specific example," or "some examples" means that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0168] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A transaction session breakpoint takeover control method, characterized in that: include: During the transaction dialogue process, the intelligent dialogue robot obtains static transaction information and the intention value of the transaction dialogue counterpart at each time stage; The time periods include a communication access period, a demand communication period, and a demand finalization period; Processing the static information and the intention values ​​in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain transaction breakpoint probability values ​​corresponding to each transaction dialogue time point; The preset transaction conversation breakpoint prediction model is obtained by training a machine learning model based on transaction conversation breakpoint prediction sample data; the preset transaction conversation breakpoint prediction model includes a machine learning model based on a gradient boosting framework; Conduct polling monitoring on the probability values ​​of each transaction breakpoint, and control the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring results; The method of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes: The transaction conversation time point corresponding to the transaction breakpoint probability value greater than the preset transaction breakpoint probability value is used as the target transaction conversation time point, and the trader is connected at the target transaction conversation time point; The method of controlling the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling monitoring result includes: If it is determined that the probability values ​​of each transaction breakpoint are less than or equal to the preset transaction breakpoint probability value, and the time period of the current transaction conversation has not changed, then monitoring the change in the intention value in the current polling monitoring cycle; If it is determined that the change in the intention value is greater than the preset numerical change, the polling monitoring of each transaction breakpoint probability value will continue until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than the preset transaction breakpoint probability value is monitored, and the transaction dialogue time point at this time will be used as the target transaction dialogue time point, and the trader will be connected at the target transaction dialogue time point.

2. The transaction session breakpoint takeover control method according to claim 1, characterized in that: Obtaining the intention value of the transaction dialogue counterpart in each time stage, including: Perform text feature extraction on the transaction dialogue to obtain text features corresponding to each word in the transaction dialogue, perform intent recognition on each text feature, and obtain the intent value corresponding to each word; Based on the preset transaction dialogue intention recognition model, the intention value corresponding to each segmentation is identified to obtain the intention value of the transaction dialogue counterpart at each time stage; Among them, the preset transaction conversation intention recognition model is obtained by training a neural network model based on transaction conversation intention recognition sample data, and can be used to obtain intention values ​​in the communication access time period, demand communication time period and demand finalization time period respectively.

3. The transaction session breakpoint takeover control method according to claim 1 or 2, characterized in that: The transaction dialogue volume is also monitored; accordingly, the transaction dialogue breakpoint takeover control method further includes: Switching the time phase of the transaction conversation based on information about changes in the conversation volume of the transaction conversation in the current polling monitoring cycle, the time phase of the transaction conversation, and the preset duration thresholds corresponding to the time phases; After completing the time phase switching, continue to poll and monitor each transaction breakpoint probability value until a transaction dialogue time point corresponding to a transaction breakpoint probability value greater than a preset transaction breakpoint probability value is detected, and use the transaction dialogue time point at this time as the target transaction dialogue time point, and access the trader at the target transaction dialogue time point.

4. The transaction session breakpoint takeover control method according to claim 3, characterized in that: The switching of the time phase of the transaction dialogue according to the change information of the conversation volume of the transaction dialogue in the current polling monitoring cycle, the time phase of the transaction dialogue, and the preset duration thresholds corresponding to the time phases includes: If it is determined that the change information of the conversation volume in the communication access time period continues to increase, and the first time period of the continuous increase in the change information is greater than the first preset duration threshold corresponding to the communication access time period, the communication access time period is switched to the required communication time period.

5. The transaction session breakpoint takeover control method according to claim 3, characterized in that: The switching of the time phase of the transaction dialogue according to the change information of the conversation volume of the transaction dialogue in the current polling monitoring cycle, the time phase of the transaction dialogue, and the preset duration thresholds corresponding to the time phases includes: If it is determined that the change information of the conversation volume in the demand communication time period continues to increase, and the second time period in which the change information continues to increase is greater than the second preset duration threshold corresponding to the demand communication time period, the demand communication time period is switched to the demand finalization time period.

6. A transaction session breakpoint takeover control device, used to execute the method according to any one of claims 1 to 5, characterized in that: The device comprises: An acquisition unit is configured to acquire static transaction information and the intention value of the transaction counterparty in each time period during a transaction dialogue conducted by the intelligent dialogue robot; the time periods include a communication access period, a demand communication period, and a demand finalization period; a prediction unit, configured to process the static information and the intention value in each time stage based on a preset transaction dialogue breakpoint prediction model to obtain a transaction breakpoint probability value corresponding to each transaction dialogue time point; The preset transaction dialogue breakpoint prediction model is obtained by training a machine learning model based on transaction dialogue breakpoint prediction sample data; The control unit is used to poll and monitor the probability values ​​of each transaction breakpoint, and control the trader who takes over the intelligent dialogue robot at the target transaction dialogue time point when the transaction breakpoint is predicted to occur according to the polling and monitoring results.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instructions are executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images