Order cancelling processing method and device, computer equipment and storage medium

Abstract

The invention provides an order cancelling processing method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a call record of a cancelled order, order remark information and order structured information; translating the call record into a user call text and a driver call text; performing feature extraction on the order remark information and the order structured information to obtain order remark features and order structured features; performing sentence intention recognition on the user call text to obtain a first recognition result; performing sentence intention recognition on the driver call text to obtain a second recognition result, and performing entity recognition on the driver call text to obtain an entity recognition result; judging the first recognition result, the second recognition result and the entity recognition result to obtain a judgment prediction result; and performing comprehensive analysis on the judgment prediction result, the order remark features and the order structural features to obtain a responsibility judgment result of the cancelled order. According to the invention, the processing efficiency and the responsibility judgment accuracy of order cancelling are improved.

Description

Technical Field

[0001] The present invention relates to the field of freight technology, and in particular to a method, device, computer equipment and storage medium for processing order cancellation. Background Art

[0002] With the development of Internet technology, orders can be distributed to drivers through online platforms, such as freight platforms. However, on freight platforms, a large number of orders are canceled every day, which will affect the driver's rejection rate. The platform will analyze the reasons for the cancellation of these orders to determine the responsible party. Some of these canceled orders are the responsibility of the user, and some are the responsibility of the driver.

[0003] Currently, when judging the liability for order cancellation, the platform usually judges the liability based on the cancellation reason filled in by the canceling party and some preset rules. However, the overall accuracy of the current liability judgment rules for handling cancelled orders is low, and errors in the judgment results will lead to a large number of drivers filing complaints, which need to be manually reviewed, resulting in increased operating costs. Summary of the invention

[0004] The main purpose of the present invention is to provide a method, device, computer equipment and storage medium for processing canceled orders, aiming to solve the technical problem in the prior art that the overall accuracy of the judgment processing of canceled orders using judgment rules is low.

[0005] To achieve the above object, the present invention provides a method for processing order cancellation, the method comprising: Obtain order data of canceled orders; wherein the order data includes the call recording between the user and the driver, order note information and order structured information; Translating the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text; Performing feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; Perform sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; Perform sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and perform entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result; Based on a preset sequence model, the first recognition result, the second recognition result, and the entity recognition result are discriminated and processed to obtain corresponding discriminant prediction results; A comprehensive analysis is performed on the discrimination prediction result, the order remark feature and the order structured feature to obtain a judgment result corresponding to the canceled order.

[0006] Optionally, the performing feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features includes: Call the preset text feature extractor; Extracting text note features from the order note information based on the text feature extractor to obtain corresponding text note features; Using the text note feature as the order note feature; Structural features are extracted from the order structured information to obtain the order structured features.

[0007] Optionally, the comprehensive analysis of the discrimination prediction result, the order remark feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Calling a preset first comprehensive discriminator; Inputting the discrimination prediction result, the order note feature and the order structured feature into the first comprehensive discriminator; The first comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature and the order structured feature to generate a judgment result corresponding to the canceled order.

[0008] Optionally, the comprehensive analysis of the discrimination prediction result, the order remark feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Invoke the preset judgment rules; Based on the judgment rule, the call text is judged and processed to obtain a corresponding initial judgment result; Calling a preset second comprehensive discriminator; Inputting the discrimination prediction result, the order remark feature, the order structured feature and the initial discrimination result into the second comprehensive discriminator; The second comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature, the order structured feature and the initial judgment result to generate a judgment result corresponding to the canceled order.

[0009] Optionally, before performing sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result, the method further includes: Obtaining a first historical call text corresponding to the historical canceled order from a preset order system; Annotating sentences related to the user in the first historical call text to obtain corresponding first annotated data; Preprocessing the first labeled data to obtain corresponding first sample data; Get the preset initial model; Training the initial model based on the first sample data to obtain a specified model that meets the preset construction requirements; The designated model is used as the first sentence recognition model.

[0010] Optionally, before performing discriminative processing on the first recognition result, the second recognition result, and the entity recognition result based on a preset sequence model to obtain corresponding discriminative prediction results, the method further includes: Obtaining a second historical call text corresponding to the historical canceled order from a preset order system; Based on the first sentence recognition model, the second sentence recognition model and the character recognition model, respectively, the second historical call text is recognized and processed to obtain corresponding recognition data; Annotating the second historical call text to obtain corresponding second annotated data; Constructing corresponding second sample data based on the identification data and the second annotation data; Based on the second sample data, a preset recurrent neural network model is trained to obtain a specified recurrent neural network model that meets preset performance requirements; The specified recurrent neural network model is used as the sequence model.

[0011] Optionally, after comprehensively analyzing the discrimination prediction result, the order remark feature and the order structured feature to obtain a judgment result corresponding to the canceled order, the method further includes: Determine whether the responsible party of the judgment result is the driver; If so, generating a corresponding target resend order based on the order data; Obtaining order requirements of the target resend order, and screening out matching designated drivers based on the order requirements; The target resend order is pushed to the designated driver.

[0012] In addition, to achieve the above-mentioned purpose, the present invention further provides a device for processing order cancellation, the device for processing order cancellation comprising: The first acquisition module is used to acquire the order data of the canceled order; wherein the order data includes the call recording between the user and the driver, the order note information and the order structured information; A translation module, used to translate the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text; An extraction module, used to perform feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; A first recognition module, configured to perform sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; A second recognition module is used to perform sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and to perform entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result; A processing module, used to perform discrimination processing on the first recognition result, the second recognition result and the entity recognition result based on a preset sequence model to obtain corresponding discrimination prediction results; The analysis module is used to comprehensively analyze the discrimination prediction results, the order note features and the order structured features to obtain the judgment results corresponding to the canceled orders.

[0013] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution: The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods for canceling an order proposed in the embodiments of the present application are implemented.

[0014] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the method for canceling an order proposed in any one of the embodiments of the present application are implemented.

[0015] Compared with the prior art, the embodiments of the present application have the following beneficial effects: The present invention provides a method, device, computer equipment and storage medium for processing order cancellation, the method comprising: BRIEF DESCRIPTION OF THE DRAWINGS In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 is an exemplary system architecture diagram to which the present application may be applied; Figure 2 is a flowchart of a method for processing order cancellation provided by an embodiment of the present invention; Figure 3 is a structural schematic diagram of an embodiment of a device for processing order cancellation according to the present application; Figure 4 This is a basic structural block diagram of the computer device in this embodiment. DETAILED DESCRIPTION

[0017] The method for processing order cancellation provided by the embodiment of the present invention is applied to the processing device for canceling order. Unless otherwise defined, all technical and scientific terms used in this document have the same meaning as those commonly understood by technicians in the technical field of this application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" in the specification and claims of this application and the above-mentioned drawings and any variations thereof are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of this application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0018] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0019] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0020] like Figure 1As shown, the system architecture 100 may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0021] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social online platform software, etc.

[0022] Terminal devices 101, 102, 103 can be various electronic devices with display screens and supporting web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 (Moving Picture Experts Group Audio Layer IV), laptop computers, desktop computers, etc.

[0023] The server 105 may be a server that provides various services, such as a background server that provides support for web pages displayed on the terminal devices 101 , 102 , and 103 .

[0024] It should be noted that the method for processing order cancellation provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the processing device for processing order cancellation is generally set in the server / terminal device.

[0025] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0026] With the development of Internet technology, orders can be distributed to drivers through online platforms, such as freight platforms. However, on freight platforms, a large number of orders are canceled every day, which will affect the driver's rejection rate. The platform will analyze the reasons for the cancellation of these orders to determine the responsible party. Some of these canceled orders are the responsibility of the user, and some are the responsibility of the driver.

[0027] Currently, when judging the liability for order cancellation, the platform usually judges the liability based on the cancellation reason filled in by the canceling party and some preset rules. However, the overall accuracy of the current liability judgment rules for handling cancelled orders is low, and errors in the judgment results will lead to a large number of drivers filing complaints, which need to be manually reviewed, resulting in increased operating costs.

[0028] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for processing order cancellation proposed in the present application. The embodiment of the present application can acquire and process relevant data based on artificial intelligence technology.

[0029] The method for processing order cancellation provided by the embodiment of the present invention comprises the following steps: S210, obtaining order data of the canceled order; wherein the order data includes a call recording between the user and the driver, order note information, and order structured information.

[0030] In this step, the execution subject of the present invention may be a freight platform, or an order processing system, referred to as a system. The system of the present invention can be applied to freight companies to conduct judgment and processing of canceled orders. Among them, the canceled orders may refer to freight cancellation orders to be judged and processed. The order note information may include the note information filled in or selected by the user when placing an order. The order structured information may include at least personnel information, cargo information, vehicle information and service information.

[0031] S220, translating the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text.

[0032] In this step, the call recording may be converted into text using a translation tool with a translation function to generate corresponding call text. The user call text refers to the text corresponding to the voice uttered by the user in the call recording, and the driver call text refers to the text corresponding to the voice uttered by the driver in the call recording.

[0033] S230, performing feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features.

[0034] In this step, the feature extraction processing is performed on the order note information and the order structured information respectively to obtain the corresponding order note features and order structured features. The present invention will provide further detailed descriptions of this in subsequent specific embodiments and will not be elaborated on here.

[0035] S240, performing sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result.

[0036] In this step, the user conversation text can be input into the first sentence recognition model to perform sentence intention recognition processing on the user conversation text through the first sentence recognition model, and output a corresponding first recognition result. The specific construction process of the first sentence recognition model will be further described in detail in the subsequent specific embodiments of the present invention, and will not be elaborated on here.

[0037] S250, performing sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and performing entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result.

[0038] In this step, the driver's conversation text can be input into the second sentence recognition model to perform sentence intention recognition on the driver's conversation text, and output a corresponding second recognition result. The driver's conversation text can be input into the character recognition model to perform entity recognition on the driver's conversation text, and output a corresponding entity recognition result. In addition, the specific construction process of the second sentence recognition model and the character recognition model can refer to the construction process of the first sentence recognition model.

[0039] S260, performing discrimination processing on the first recognition result, the second recognition result and the entity recognition result based on a preset sequence model to obtain corresponding discrimination prediction results.

[0040] In this step, the first recognition result, the second recognition result and the entity recognition result can be input into the sequence model to discriminate the first recognition result, the second recognition result and the entity recognition result through the sequence model, and output the corresponding discriminant prediction result. The specific construction process of the sequence model will be further described in detail in the subsequent specific embodiments of the present invention, and will not be elaborated here.

[0041] S270, comprehensively analyzing the discrimination prediction result, the order remark feature and the order structured feature to obtain a judgment result corresponding to the canceled order.

[0042] In this step, the above-mentioned discrimination prediction results, the order remark features and the order structured features are comprehensively analyzed to obtain the specific implementation process of the judgment result corresponding to the canceled order. The present invention will provide further detailed descriptions of this in subsequent specific embodiments and will not be elaborated in detail here.

[0043] In an embodiment of the present invention, firstly, order data of a canceled order is obtained; wherein, the order data includes a call recording between a user and a driver, order note information and order structured information; then, the call recording is translated into a corresponding call text; wherein, the call text data includes a user call text and a driver call text; then, feature extraction processing is performed on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; subsequently, sentence intention recognition processing is performed on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; and sentence intention recognition processing is performed on the driver call text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and entity recognition is performed on the driver call text based on a preset character recognition model to obtain a corresponding entity recognition result; further, the first recognition result, the second recognition result and the entity recognition result are discriminated based on a preset sequence model to obtain a corresponding discriminated prediction result; finally, the discriminated prediction result, the order note feature and the order structured feature are comprehensively analyzed to obtain a judgment result corresponding to the canceled order. The present invention performs accountability processing on canceled orders by combining the discrimination prediction results generated based on the sequence model with the order note features and the order structured features extracted from the order note information and the order structured information respectively. This can achieve rapid and accurate generation of accountability results corresponding to canceled orders, thereby improving the processing efficiency and accountability accuracy of canceled orders, which is beneficial to reducing the amount of driver complaints, and further reducing the operating costs for handling driver complaints.

[0044] Optionally, the performing feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features includes: Call the preset text feature extractor.

[0045] In this step, the text feature extractor is a tool with a text extraction function, such as a keyword extractor or a regular expression extractor.

[0046] Based on the text feature extractor, text note features are extracted from the order note information to obtain corresponding text note features.

[0047] In this step, the text feature extractor may be used to extract text features from the order note information to obtain the text note features.

[0048] The text note feature is used as the order note feature.

[0049] Structural features are extracted from the order structured information to obtain the order structured features.

[0050] In this step, the above-mentioned order structured features may include one or more of the following information: personnel information, cargo information, vehicle information, service information, number of people following the vehicle, whether there is any transportation demand, etc.

[0051] In this embodiment, a preset text feature extractor is called; then, based on the text feature extractor, text note feature extraction is performed on the order note information to obtain corresponding text note features; then, the text note features are used as the order note features; and subsequently, structured feature extraction is performed on the order structured information to obtain the order structured features. The present invention can achieve fast and accurate acquisition of required order note features by performing text note feature extraction on order note information based on the use of a text feature extractor, and can achieve accurate acquisition of required order structured features by performing structured feature extraction on order structured information, thereby improving the feature extraction efficiency of order note information and order structured information.

[0052] Optionally, the comprehensive analysis of the discrimination prediction result, the order remark feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Call the preset first comprehensive discriminator.

[0053] In this step, the historical canceled orders can be first obtained from the order system (such as the order system of the freight company), and then the historical canceled orders are translated to obtain the historical call text, and the historical discrimination prediction results obtained after the historical call text is recognized and processed by the above-mentioned first sentence recognition model, the second sentence recognition model, the character recognition model and the sequence model, and the historical order remark features and the historical order structured features corresponding to the historical call text are obtained, and then the judgment label is annotated for each historical call text according to the historical discrimination prediction results, the historical order remark features and the historical order structured features, and then the historical call text of each historical canceled order and its annotated judgment label are used as a sample to obtain the corresponding sample data. The construction process of the above-mentioned first comprehensive discriminator includes: using the above-mentioned sample data annotated with the judgment label to train the deep learning model, so as to obtain a model with the function of performing judgment analysis on the canceled order, and use it as the above-mentioned first comprehensive discriminator. The first comprehensive discriminator can predict the matching judgment label, or the judgment result, of the canceled order based on the judgment prediction result corresponding to the canceled order, the order remark feature, and the order structured feature.

[0054] The discrimination prediction result, the order note feature and the order structured feature are input into the first comprehensive discriminator.

[0055] In this step, the above-mentioned discrimination prediction results, order note features and order structured features can be used as model inputs and input into the above-mentioned first comprehensive discriminator.

[0056] The first comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature and the order structured feature to generate a judgment result corresponding to the canceled order.

[0057] In this step, the input discrimination prediction results, order note features and order structured features are judged by using the above-mentioned first comprehensive discriminator, and the cancellation reason label and responsible party identification output by the first comprehensive discriminator are obtained as the judgment result. Among them, the cancellation reason label may include but is not limited to "vehicle and cargo mismatch-cargo overweight", "vehicle and cargo mismatch-cargo overlength, width and height", "vehicle and cargo mismatch-car model mismatch", "user temporarily does not use the car", "driver did not read the note requirements" and other labels. The responsible party identification is used to represent the driver or user.

[0058] In this embodiment, the preset first comprehensive discriminator is called; then the discrimination prediction result, the order note feature and the order structured feature are input into the first comprehensive discriminator; subsequently, the discrimination prediction result, the order note feature and the order structured feature are discriminated and processed by the first comprehensive discriminator to generate a judgment result corresponding to the canceled order. The present invention inputs the discrimination prediction result, the order note feature and the order structured feature into the first comprehensive discriminator, and then based on the use of the first comprehensive discriminator, the discrimination prediction result generated based on the sequence model and the order note feature and the order structured feature extracted from the order note information and the order structured information are combined for judgment, so that the judgment result corresponding to the canceled order can be generated quickly and accurately, which improves the accuracy of the judgment of the canceled order, is conducive to reducing the number of driver complaints, improving the processing time of canceled orders, and thus reducing the labor cost for handling driver complaints.

[0059] Optionally, the comprehensive analysis of the discrimination prediction result, the order remark feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Invoke the preset judgment rules; In this step, the above-mentioned judgment rules are rules pre-built according to actual business needs for judging the responsibility of canceled orders. Exemplary content of the judgment rules may include: "If the user clearly forgives the driver, the driver is not judged to be responsible for canceling the order", "If the order type is an emergency order and the driver mistakenly grabs the order and causes a delay, the driver is judged to be responsible", etc.

[0060] The call text is judged and processed based on the judgment rule to obtain a corresponding initial judgment result.

[0061] In this step, the call text may be judged according to the content of the judgment rule, thereby obtaining a corresponding initial judgment result.

[0062] Call the preset second comprehensive discriminator.

[0063] In this step, the historical cancellation orders can be first obtained from the order system (such as the order system of the freight company), and then the historical cancellation orders are translated to obtain the historical call text, and the historical discrimination prediction results obtained after the historical call text is recognized and processed by the above-mentioned first sentence recognition model, the second sentence recognition model, the character recognition model and the sequence model, and the historical judgment results obtained by judging the historical call text by the above-mentioned judgment rules, and the historical order remark features and historical order structured features corresponding to the historical call text are obtained, and then the judgment labels are annotated for each historical call text according to the historical discrimination prediction results, the historical judgment results, the historical order remark features and the historical order structured features, and then the historical call text of each historical cancellation order and its annotated judgment label are used as a sample to obtain the corresponding historical sample data. The construction process of the above-mentioned second comprehensive discriminator includes: using the above-mentioned historical sample data annotated with the judgment label to train the deep learning model to obtain a model with the function of performing judgment analysis on the cancellation order, and serving as the above-mentioned second comprehensive discriminator. Furthermore, the second comprehensive discriminator can predict the matching judgment label, or judgment result, of the canceled order based on the judgment prediction result corresponding to the canceled order, the preliminary judgment result obtained based on the judgment rule, the order note feature and the order structured feature.

[0064] The discrimination prediction result, the order remark feature, the order structured feature and the initial discrimination result are input into the second comprehensive discriminator.

[0065] In this step, the above-mentioned discrimination prediction results, order note features, order structured features and initial judgment results can be used as model inputs and input into the above-mentioned second comprehensive discriminator.

[0066] The second comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature, the order structured feature and the initial judgment result to generate a judgment result corresponding to the canceled order.

[0067] In this step, the input discrimination prediction results, order note features and order structured features are judged by using the above-mentioned second comprehensive discriminator, and the cancellation reason label and responsible party identification output by the second comprehensive discriminator are obtained as the judgment result. Among them, the cancellation reason label may include but is not limited to "vehicle and cargo mismatch-cargo overweight", "vehicle and cargo mismatch-cargo overlength, width and height", "vehicle and cargo mismatch-car model mismatch", "user temporarily does not use the car", "driver did not read the note requirements" and other labels. The responsible party identification is used to represent the driver or user.

[0068] In this embodiment, a preset judgment rule is called; then the call text is judged based on the judgment rule to obtain a corresponding initial judgment result; then a preset second comprehensive discriminator is called; subsequently, the judgment prediction result, the order note feature, the order structured feature and the initial judgment result are input into the second comprehensive discriminator; finally, the judgment prediction result, the order note feature, the order structured feature and the initial judgment result are judged by the second comprehensive discriminator to generate a judgment result corresponding to the canceled order. The present invention performs discrimination processing on the call text based on the use of judgment rules to obtain the corresponding initial judgment result, and then inputs the discrimination prediction result, order note feature, order structured feature and the initial judgment result into a preset second comprehensive discriminator. Then, based on the use of the second comprehensive discriminator, the discrimination prediction result generated based on the sequence model, the order note feature and the order structured feature extracted from the order note information and the order structured information respectively, and the initial judgment result generated based on the judgment rule are combined to perform judgment. It can achieve fast and accurate generation of judgment results corresponding to canceled orders, improve the accuracy of judgment on canceled orders, help reduce the number of driver complaints, improve the processing time of canceled orders, and thus reduce the labor cost for handling driver complaints.

[0069] Optionally, before performing sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result, the method further includes: A first historical call text corresponding to the historical canceled order is obtained from a preset order system.

[0070] In this step, the order system may refer to the order system of the freight company. The historical cancellation orders may be obtained from the order system, and then the historical cancellation orders may be translated to obtain the corresponding first historical call text.

[0071] The sentences related to the user in the first historical call text are annotated to obtain corresponding first annotated data.

[0072] In this step, the sentences related to the user in the first historical call text may be annotated at the sentence level, and the labels include but are not limited to "blame the driver", "forgive the driver", "request the driver to cancel the order", etc. These labels reflect the user's feedback or attitude towards the driver's behavior.

[0073] The first labeled data is preprocessed to obtain corresponding first sample data.

[0074] In this step, the first labeled data may be cleaned to remove irrelevant characters, punctuation marks, and other preprocessing, and the preprocessed data obtained may be used as the first sample data.

[0075] Get the preset initial model.

[0076] In this step, the above initial model can specifically adopt a deep learning model based on Ernie-Medium.

[0077] The initial model is trained based on the first sample data to obtain a specified model that meets preset construction requirements.

[0078] In this step, the above-mentioned initial model can be trained by using the above-mentioned first sample data to obtain a first sentence-level deep model for identifying the user's sentence intention, that is, a first sentence recognition model. Specifically, the construction process of the first sentence recognition model includes: converting the annotated first sample data into a format acceptable to the model, such as converting the text and the corresponding labels into sequence data. Then divide the converted data into a training set and a validation set. Then use the training set to train the model and set appropriate hyperparameters (such as learning rate, batch size, etc.). Subsequently, the validation set is used to evaluate the model, and the model parameters are adjusted to improve performance until a specified model that meets the construction requirements is obtained, and it is used as the final first sentence recognition model.

[0079] The designated model is used as the first sentence recognition model.

[0080] In this embodiment, a first historical call text corresponding to a historical canceled order is obtained from a preset order system; sentences related to the user in the first historical call text are annotated to obtain corresponding first annotated data; the first annotated data is preprocessed to obtain corresponding first sample data; a preset initial model is subsequently obtained; and the initial model is trained based on the first sample data to obtain a designated model that meets preset construction requirements; finally, the designated model is used as the first sentence recognition model, thereby realizing efficient and accurate completion of the construction process of the first sentence recognition model, improving the construction efficiency of the first sentence recognition model, and ensuring the model effect of the obtained first sentence recognition model.

[0081] In some optional implementations, the training process of the second sentence recognition model can refer to the training process of the first sentence recognition model, and no further details are given here. It is only necessary that the training process of the first sentence recognition model includes the process of annotating the sentences related to the user in the first historical call text to obtain the corresponding first annotated data, which is correspondingly modified to annotate the sentences related to the driver in the first historical call text to obtain the corresponding third annotated data. Specifically, by annotating the sentences related to the driver in the first historical call text, the labels include but are not limited to "did not see the instant order reservation form clearly", "someone else grabbed it", "accidentally touched", "did not see the requirements clearly", etc. These labels reflect the behavior or status that the driver may exhibit in the conversation.

[0082] The training process of the above-mentioned character recognition model can refer to the training process of the above-mentioned first sentence recognition model, and will not be elaborated here. It is only necessary that the training process of the first sentence recognition model includes the processing of annotating the sentences related to the user in the first historical call text to obtain the corresponding first annotation data, which is correspondingly modified to identify the key information related to the goods in the first historical call text, such as the weight of the goods, the length of the goods, the vehicle type, the type of service, etc. And these key information are annotated at the character level, that is, a unique label is assigned to each entity, thereby completing the corresponding annotation processing. In addition, the Bert+CRF model can be trained by using the first historical call text after entity annotation to obtain the required character recognition model.

[0083] Optionally, before performing discriminative processing on the first recognition result, the second recognition result, and the entity recognition result based on a preset sequence model to obtain corresponding discriminative prediction results, the method further includes: A second historical call text corresponding to the historical canceled order is obtained from a preset order system.

[0084] In this step, the order system may refer to the order system of the freight company. The historical cancellation orders may be obtained from the order system, and then the historical cancellation orders may be translated to obtain the corresponding second historical call text.

[0085] The second historical call text is recognized and processed based on the first sentence recognition model, the second sentence recognition model and the character recognition model to obtain corresponding recognition data.

[0086] In this step, the second historical call text can be recognized and processed by using the previously trained first sentence recognition model, second sentence recognition model and character recognition model, so as to obtain corresponding user intention labels, driver intention labels and entity labels, and the obtained user intention labels, driver intention labels and entity labels are integrated to obtain the above-mentioned recognition data. The recognition data reflects the behavior or attitude of the driver and the user in the conversation, and includes key information related to the goods in the conversation.

[0087] The second historical call text is annotated to obtain corresponding second annotated data.

[0088] In this step, the second historical call text is annotated to determine whether the driver's mistaken grabbing caused the order to be cancelled. This is a binary classification task with labels of "yes" or "no". In order to ensure the accuracy and consistency of the annotated data, the annotation quality can be improved by annotating by multiple people and taking the average value or using other quality control methods.

[0089] Corresponding second sample data is constructed based on the recognition data and the second labeled data.

[0090] In this step, the above-mentioned recognition data and the second annotation data can be combined into initial sample data, and the initial sample data can be cleaned and normalized, such as removing irrelevant characters, unifying the label format, etc., to obtain corresponding second sample data.

[0091] The preset recurrent neural network model is trained based on the second sample data to obtain a specified recurrent neural network model that meets the preset performance requirements.

[0092] In this step, the above-mentioned recurrent neural network model can specifically adopt an RNN model (such as standard RNN, LSTM, GRU, etc.). By dividing the above-mentioned second sample data into training set data and validation set data. The RNN model is trained by using the training set data. During the training process, the model parameters are updated by the back propagation algorithm and the gradient descent method. Monitor the training process of the model, such as the change in loss value, the accuracy on the validation set, etc. Adjust the hyperparameters of the model (such as learning rate, batch size, etc.) as needed to optimize the model performance, so as to obtain a trained RNN model. Then use the validation set data to evaluate the trained RNN model. Calculate the accuracy, recall rate, F1 score and other indicators of the model on the validation set. Adjust the model architecture or hyperparameters according to the evaluation results to improve the performance until a specified recurrent neural network model that meets the performance requirements is obtained and used as the final sequence model.

[0093] The specified recurrent neural network model is used as the sequence model.

[0094] In this step, the trained RNN model can be used to identify the intent of the new call recording text and determine whether the driver's mistaken grabbing leads to the cancellation of the order. In addition, the generated sequence model can be stored, such as in a database, disk, blockchain, etc., for subsequent calls.

[0095] In this embodiment, a second historical call text corresponding to a historical canceled order is obtained from a preset order system; then the second historical call text is recognized and processed based on the first sentence recognition model, the second sentence recognition model and the character recognition model to obtain corresponding recognition data; then the second historical call text is annotated to obtain corresponding second annotated data; subsequently, corresponding second sample data is constructed based on the recognition data and the second annotated data; and based on the second sample data, a preset recurrent neural network model is trained and processed to obtain a specified recurrent neural network model that meets preset performance requirements; finally, the specified recurrent neural network model is used as the sequence model, thereby realizing efficient and accurate completion of the sequence model construction process, improving the sequence model construction efficiency, and ensuring the model effect of the obtained sequence model.

[0096] Optionally, after comprehensively analyzing the discrimination prediction result, the order remark feature and the order structured feature to obtain a judgment result corresponding to the canceled order, the method further includes: Determine whether the responsible party of the judgment result is the driver.

[0097] In this step, the responsible party for the above-mentioned judgment result may include the driver or the user.

[0098] If so, generate a corresponding target resend order based on the order data.

[0099] In this step, if it is detected that the responsible party of the above judgment result is the driver, a reissue order is automatically generated based on the order data of the above canceled order, that is, the above target reissue order.

[0100] The order requirements of the target resend order are obtained, and matching designated drivers are screened out based on the order requirements.

[0101] In this step, the order requirements of the target reissued order refer to the order acceptance conditions corresponding to the target reissued order. The designated driver matching the order requirements of the target reissued order refers to other drivers who meet the order acceptance requirements except the above-mentioned driver.

[0102] The target resend order is pushed to the designated driver.

[0103] In this step, the order push address of the designated driver can be obtained, and then the target resend order can be pushed to the designated driver according to the order push address.

[0104] In this embodiment, it is determined whether the responsible party of the judgment result is the driver; if so, a corresponding target reissue order is generated based on the order data; then the order requirements of the target reissue order are obtained, and a matching designated driver is screened out based on the order requirements; and the target reissue order is subsequently pushed to the designated driver. When the present invention detects that the responsible party of the judgment result is the driver, it will automatically and intelligently generate a corresponding target reissue order based on the order data, obtain the order requirements based on the obtained target reissue order to screen out a matching designated driver, and then push the target reissue order to the designated driver, so as to automatically and timely complete the retry processing of the canceled order, which is beneficial to improve the success rate of order processing and thus improve the user experience.

[0105] In some optional implementations, the user information obtained is subject to the user's consent and complies with relevant laws and policies.

[0106] In addition, compared with the existing technology, the present invention makes more in-depth and detailed use of the recorded conversations between drivers and users, and specifically optimizes the scenario of drivers mistaking orders, which accounts for a large proportion of cancelled orders, and can further improve the accuracy of the judgment of responsibility in this scenario. At the same time, it can provide a more detailed judgment process, reasons for judgment and conversation details, which is convenient for operators to analyze and study.

[0107] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a device 300 for processing order cancellation. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0108] An embodiment of the present invention provides a device 300 for processing order cancellation, and the device 300 for processing order cancellation includes: The first acquisition module 310 is used to acquire order data of the canceled order; wherein the order data includes the call recording between the user and the driver, order note information and order structured information; The translation module 320 is used to translate the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text; An extraction module 330 is used to perform feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; A first recognition module 340, configured to perform sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; The second recognition module 350 is used to perform sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and to perform entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result; A processing module 360, configured to perform discrimination processing on the first recognition result, the second recognition result and the entity recognition result based on a preset sequence model to obtain corresponding discrimination prediction results; The analysis module 370 is used to comprehensively analyze the discrimination prediction results, the order note features and the order structured features to obtain the judgment results corresponding to the canceled orders.

[0109] Optionally, the extraction module 330 includes: The first calling submodule is used to call a preset text feature extractor; A first extraction submodule, configured to extract text note features from the order note information based on the text feature extractor to obtain corresponding text note features; A determination submodule, configured to use the text note feature as the order note feature; The second extraction submodule is used to extract structured features from the order structured information to obtain the order structured features.

[0110] Optionally, the analysis module 370 includes: A second calling submodule, used for calling a preset first comprehensive discriminator; A first input submodule, used for inputting the discrimination prediction result, the order remark feature and the order structured feature into the first comprehensive discriminator; The first generating submodule is used to perform discrimination processing on the discrimination prediction result, the order remark feature and the order structured feature through the first comprehensive discriminator to generate a judgment result corresponding to the canceled order.

[0111] Optionally, the analysis module 370 includes: The third calling submodule is used to call the preset judgment rules; A determination submodule, used for performing determination processing on the call text based on the determination rule to obtain a corresponding initial determination result; A fourth calling submodule, used for calling a preset second comprehensive discriminator; A second input submodule, used for inputting the discrimination prediction result, the order remark feature, the order structured feature and the initial discrimination result into the second comprehensive discriminator; The second generating submodule is used to perform discrimination processing on the discrimination prediction result, the order remark feature, the order structured feature and the initial judgment result through the second comprehensive discriminator to generate a judgment result corresponding to the canceled order.

[0112] Optionally, the order cancellation processing device 300 further includes: A second acquisition module is used to acquire a first historical call text corresponding to a historical canceled order from a preset order system; A first annotation module, configured to annotate sentences related to the user in the first historical call text to obtain corresponding first annotation data; A preprocessing module, used for preprocessing the first labeled data to obtain corresponding first sample data; The third acquisition module is used to acquire a preset initial model; A first training module, used for training the initial model based on the first sample data to obtain a specified model that meets the preset construction requirements; The first determination module is used to use the specified model as the first sentence recognition model.

[0113] Optionally, the order cancellation processing device 300 further includes: A fourth acquisition module, used to acquire a second historical call text corresponding to the historical canceled order from a preset order system; A first generating module, configured to perform recognition processing on the second historical call text based on the first sentence recognition model, the second sentence recognition model, and the character recognition model, respectively, to obtain corresponding recognition data; A second annotation module, used for annotating the second historical call text to obtain corresponding second annotation data; A construction module, configured to construct corresponding second sample data based on the identification data and the second labeled data; A second training module is used to train a preset recurrent neural network model based on the second sample data to obtain a specified recurrent neural network model that meets preset performance requirements; The second determination module is used to use the specified recurrent neural network model as the sequence model.

[0114] Optionally, the order cancellation processing device 300 further includes: A judgment module, used to judge whether the responsible party of the judgment result is the driver; A second generating module, configured to generate a corresponding target resending order based on the order data if yes; A screening module, used for obtaining the order requirements of the target resend order and screening out matching designated drivers based on the order requirements; A push module is used to push the target resend order to the designated driver.

[0115] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment. The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected through a system bus. It should be noted that the figure only shows a computer device 4 with components 41-43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0116] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0117] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as the program code of the processing method for canceling an order, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0118] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the program code stored in the memory 41 or process data, such as running the program code of the method for processing order cancellation.

[0119] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0120] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores the application crash processing program, and the application crash processing program can be executed by at least one processor to enable the at least one processor to perform the steps of the order cancellation processing method as described above.

[0121] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware online platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0122] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0123] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for processing order cancellation, characterized in that: include: Obtain order data of canceled orders; wherein the order data includes the call recording between the user and the driver, order note information and order structured information; Translating the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text; Performing feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; Perform sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; Perform sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and perform entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result; Based on a preset sequence model, the first recognition result, the second recognition result, and the entity recognition result are discriminated and processed to obtain corresponding discriminant prediction results; A comprehensive analysis is performed on the discrimination prediction result, the order remark feature and the order structured feature to obtain a judgment result corresponding to the canceled order.

2. The method according to claim 1, characterized in that The feature extraction process is performed on the order note information and the order structure information respectively to obtain corresponding order note features and order structure features, including: Call the preset text feature extractor; Extracting text note features from the order note information based on the text feature extractor to obtain corresponding text note features; Using the text note feature as the order note feature; Structural features are extracted from the order structured information to obtain the order structured features.

3. The method according to claim 1, characterized in that The comprehensive analysis of the discrimination prediction result, the order note feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Calling a preset first comprehensive discriminator; Inputting the discrimination prediction result, the order note feature and the order structured feature into the first comprehensive discriminator; The first comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature and the order structured feature to generate a judgment result corresponding to the canceled order.

4. The method according to claim 1, characterized in that: The comprehensive analysis of the discrimination prediction result, the order note feature and the order structured feature to obtain the judgment result corresponding to the canceled order includes: Invoke the preset judgment rules; Based on the judgment rule, the call text is judged and processed to obtain a corresponding initial judgment result; Calling a preset second comprehensive discriminator; Inputting the discrimination prediction result, the order remark feature, the order structured feature and the initial discrimination result into the second comprehensive discriminator; The second comprehensive discriminator performs discriminative processing on the discriminative prediction result, the order remark feature, the order structured feature and the initial judgment result to generate a judgment result corresponding to the canceled order.

5. The method according to claim 1, characterized in that Before performing sentence intention recognition processing on the user call text based on the preset first sentence recognition model to obtain the corresponding first recognition result, the method also includes: Obtaining a first historical call text corresponding to the historical canceled order from a preset order system; Annotating sentences related to the user in the first historical call text to obtain corresponding first annotated data; Preprocessing the first labeled data to obtain corresponding first sample data; Get the preset initial model; Training the initial model based on the first sample data to obtain a specified model that meets the preset construction requirements; The designated model is used as the first sentence recognition model.

6. The method according to claim 1, characterized in that Before performing discriminative processing on the first recognition result, the second recognition result, and the entity recognition result based on the preset sequence model to obtain corresponding discriminative prediction results, the method further includes: Obtaining a second historical call text corresponding to the historical canceled order from a preset order system; Based on the first sentence recognition model, the second sentence recognition model and the character recognition model, respectively, the second historical call text is recognized and processed to obtain corresponding recognition data; Annotating the second historical call text to obtain corresponding second annotated data; Constructing corresponding second sample data based on the identification data and the second annotation data; Based on the second sample data, a preset recurrent neural network model is trained to obtain a specified recurrent neural network model that meets preset performance requirements; The specified recurrent neural network model is used as the sequence model.

7. The method according to claim 1, characterized in that After comprehensively analyzing the discrimination prediction result, the order remark feature, and the order structured feature to obtain a judgment result corresponding to the canceled order, the method further includes: Determine whether the responsible party of the judgment result is the driver; If so, generating a corresponding target resend order based on the order data; Obtaining order requirements of the target resend order, and screening out matching designated drivers based on the order requirements; The target resend order is pushed to the designated driver.

8. A device for processing order cancellation, characterized in that: include: The first acquisition module is used to acquire the order data of the canceled order; wherein the order data includes the call recording between the user and the driver, the order note information and the order structured information; A translation module, used to translate the call recording into corresponding call text; wherein the call text data includes the user call text and the driver call text; An extraction module, used to perform feature extraction processing on the order note information and the order structured information respectively to obtain corresponding order note features and order structured features; A first recognition module, configured to perform sentence intention recognition processing on the user call text based on a preset first sentence recognition model to obtain a corresponding first recognition result; A second recognition module is used to perform sentence intention recognition processing on the driver's conversation text based on a preset second sentence recognition model to obtain a corresponding second recognition result, and to perform entity recognition on the driver's conversation text based on a preset character recognition model to obtain a corresponding entity recognition result; A processing module, used to perform discrimination processing on the first recognition result, the second recognition result and the entity recognition result based on a preset sequence model to obtain corresponding discrimination prediction results; The analysis module is used to comprehensively analyze the discrimination prediction results, the order note features and the order structured features to obtain the judgment results corresponding to the canceled orders.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the method for canceling an order as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for processing order cancellation as described in any one of claims 1 to 7 are implemented.

Images