Intelligent processing method, device and storage medium for work order data

Abstract

The present invention discloses an intelligent processing method, device and storage medium for work order data, which are used to improve the effectiveness and efficiency of processing work order data. The method may include: receiving work order data to be processed; performing pre-judgment processing on the work order data to be processed through a trained work order processing model to obtain a processing result; the processing result includes information characterizing whether the work order data to be processed is valid data; transferring the work order data to be processed according to the processing result through the trained work order processing model to obtain a transfer result; matching the transfer result through the trained work order processing model to obtain processing result information corresponding to the work order data to be processed.

Description

Technical Field

[0001] The present invention relates to the field of communication operation technology, and in particular to a method, device and storage medium for intelligent processing of work order data. Background Art

[0002] Currently, in the telecommunications industry, when users encounter communication problems using their terminal devices, they typically call the operator's consultation phone number. As the number of users using these devices increases, telecommunications operators are likely to receive a large number of consultation calls with diverse questions.

[0003] At present, the following process is generally used to handle received consultation calls:

[0004] 1. Work order judgment requires professionals to manually predict whether the work order data corresponding to the consultation call is valid.

[0005] 2. Work order transfer: The dispatcher needs to transfer the work order data to the corresponding technical personnel for processing based on the prediction results of professionals.

[0006] 3. Work order processing: technicians query the fault location manual to locate the problem in the work order data, then query the business system and respond to the user's inquiry results.

[0007] It can be seen that the three processes mentioned above, namely work order judgment, work order transfer and work order processing, all use manual methods to judge and process work order data. The efficiency of processing work order data is low. There is also the problem of work order prediction errors due to inaccurate manual knowledge points, and the effectiveness of processing work order data is low.

[0008] In summary, it is necessary to provide a solution that can improve the effectiveness and efficiency of processing work order data. Summary of the Invention

[0009] Based on this, the purpose of the present invention is to provide a method, device and storage medium for intelligent processing of work order data, so as to improve the effectiveness and efficiency of processing work order data.

[0010] In a first aspect, the present invention provides a method for intelligently processing work order data, comprising:

[0011] Receive work order data to be processed;

[0012] Pre-judgment processing is performed on the work order data to be processed using the trained work order processing model to obtain a processing result; the processing result includes information indicating whether the work order data to be processed is valid data;

[0013] Transferring the to-be-processed work order data according to the processing result using the trained work order processing model to obtain a transfer result;

[0014] The transfer result is matched with the trained work order processing model to obtain processing result information corresponding to the work order data to be processed.

[0015] In one possible design, if the processing result includes information that the work order data to be processed is valid data, then the trained work order processing model is used to transfer the work order data to be processed according to the processing result to obtain a transfer result, including:

[0016] Performing a first-level transfer process on the to-be-processed work order data using the trained work order processing model to obtain a first-level transfer result;

[0017] The trained work order processing model is used to perform secondary transfer processing on the primary transfer result and the work order data to be processed to obtain a secondary transfer result.

[0018] In one possible design, the trained work order processing model is used to perform a first-level transfer process on the work order data to be processed, and a first-level transfer result is obtained, including:

[0019] Determining a first matching score between the to-be-processed work order data and at least one first business category to obtain at least one first matching score;

[0020] If a target matching score that satisfies the first screening condition exists among the at least one first matching score, determining a first business category corresponding to the target matching score;

[0021] Determining a second matching score between the to-be-processed work order data and each second business category in the first business category;

[0022] A first-level transfer result is obtained based on the second matching score obtained.

[0023] In one possible design, obtaining a first-level transfer result based on the obtained second matching score includes:

[0024] If the target matching score that satisfies the second screening condition does not exist in the obtained second matching scores, transferring the case to manual processing is taken as the first transfer result;

[0025] If there is a target matching score that satisfies the second screening condition among the obtained second matching scores, the pending confirmation process is taken as the first transfer result.

[0026] In one possible design, performing secondary transfer processing on the primary transfer result and the to-be-processed work order data using the trained work order processing model to obtain a secondary transfer result includes:

[0027] If there is a target matching score that satisfies the second screening condition among the obtained second matching scores, determining whether the to-be-processed work order data includes a custom keyword;

[0028] If the work order data to be processed includes a custom keyword, the work order data is transferred to intelligent processing as the second transfer result;

[0029] If the work order data to be processed does not include the custom keyword, manual processing will be taken as the second transfer result.

[0030] In one possible design, matching processing is performed on the secondary transfer result using the trained work order processing model to obtain processing result information corresponding to the work order data to be processed, including:

[0031] determining a target second business category based on a second matching score that satisfies a second screening condition;

[0032] The preset reply information associated with the target second business category is used as the processing result information corresponding to the work order data to be processed.

[0033] In one possible design, the trained work order processing model is trained in the following manner, including:

[0034] Determine a plurality of work order data, and process the plurality of work order data using a synthetic minority class oversampling algorithm to obtain a plurality of invalid work order data to be trained;

[0035] Inputting the plurality of invalid work order data and valid work order data to be trained into a preset work order processing model for training to obtain a first training result, and adjusting the sub-model for performing pre-judgment processing in the preset work order processing model based on the first training result to obtain a sub-model for performing pre-judgment processing in the trained work order processing model;

[0036] Inputting a plurality of valid work order data marked with business types into the preset work order processing model for training to obtain a second training result, and adjusting a sub-model for performing a dispatch process in the preset work order processing model based on the second training result to obtain a sub-model for performing a dispatch process in the trained work order processing model;

[0037] A trained work order processing model is obtained based on the sub-model that performs pre-judgment processing in the trained work order processing model and the sub-model that performs transfer processing in the trained work order processing model.

[0038] In a second aspect, the present invention further provides an intelligent processing device for work order data, comprising:

[0039] A receiving unit, used for receiving work order data to be processed;

[0040] A prejudgment unit, configured to perform prejudgment processing on the work order data to be processed using a trained work order processing model to obtain a processing result; the processing result includes information indicating whether the work order data to be processed is valid data;

[0041] a transfer unit, configured to transfer the work order data to be processed according to the processing result using the trained work order processing model to obtain a transfer result;

[0042] The reply unit is used to match the transfer result through the trained work order processing model to obtain the processing result information corresponding to the work order data to be processed.

[0043] In one possible design, if the processing result includes information that the to-be-processed work order data is valid data, the dispatching unit is configured to:

[0044] Performing a first-level transfer process on the to-be-processed work order data using the trained work order processing model to obtain a first-level transfer result;

[0045] The trained work order processing model is used to perform secondary transfer processing on the primary transfer result and the work order data to be processed to obtain a secondary transfer result.

[0046] In one possible design, the dispatching unit is specifically configured to:

[0047] Determining a first matching score between the to-be-processed work order data and at least one first business category to obtain at least one first matching score;

[0048] If a target matching score that satisfies the first screening condition exists among the at least one first matching score, determining a first business category corresponding to the target matching score;

[0049] Determining a second matching score between the to-be-processed work order data and each second business category in the first business category;

[0050] A first-level transfer result is obtained based on the second matching score obtained.

[0051] In one possible design, the dispatching unit is specifically configured to:

[0052] If the target matching score that satisfies the second screening condition does not exist in the obtained second matching scores, transferring the case to manual processing is taken as the first transfer result;

[0053] If there is a target matching score that satisfies the second screening condition among the obtained second matching scores, the pending confirmation process is taken as the first transfer result.

[0054] In one possible design, the dispatching unit is specifically configured to:

[0055] If there is a target matching score that satisfies the second screening condition among the obtained second matching scores, determining whether the to-be-processed work order data includes a custom keyword;

[0056] If the work order data to be processed includes a custom keyword, the work order data is transferred to intelligent processing as the second transfer result;

[0057] If the work order data to be processed does not include the custom keyword, manual processing will be taken as the second transfer result.

[0058] In one possible design, the reply unit is specifically configured to:

[0059] determining a target second business category based on a second matching score that satisfies a second screening condition;

[0060] The preset reply information associated with the target second business category is used as the processing result information corresponding to the work order data to be processed.

[0061] In one possible design, the apparatus further includes a training unit, wherein the training unit is configured to:

[0062] Determine a plurality of work order data, and process the plurality of work order data using a synthetic minority class oversampling algorithm to obtain a plurality of invalid work order data to be trained;

[0063] Inputting the plurality of invalid work order data and valid work order data to be trained into a preset work order processing model for training to obtain a first training result, and adjusting the sub-model for performing pre-judgment processing in the preset work order processing model based on the first training result to obtain a sub-model for performing pre-judgment processing in the trained work order processing model;

[0064] Inputting a plurality of valid work order data marked with business types into the preset work order processing model for training to obtain a second training result, and adjusting a sub-model for performing a dispatch process in the preset work order processing model based on the second training result to obtain a sub-model for performing a dispatch process in the trained work order processing model;

[0065] A trained work order processing model is obtained based on the sub-model that performs pre-judgment processing in the trained work order processing model and the sub-model that performs transfer processing in the trained work order processing model.

[0066] In a third aspect, the present invention further provides a computer device comprising: at least one memory and at least one processor;

[0067] The at least one memory is used to store one or more programs;

[0068] When the one or more programs are executed by the at least one processor, the method involved in any possible design of the first aspect described above is implemented.

[0069] In a fourth aspect, the present invention further provides a computer-readable storage medium storing at least one program; when the at least one program is executed by a processor, the method involved in any possible design in the first aspect above is implemented.

[0070] The beneficial effects of the present invention are as follows:

[0071] Compared with the existing technology, the present invention receives work order data to be processed; pre-judges the work order data to be processed through a trained work order processing model to obtain a processing result; the processing result includes information characterizing whether the work order data to be processed is valid data; transfers the work order data to be processed according to the processing result through the trained work order processing model to obtain a transfer result; matches the transfer result through the trained work order processing model to obtain processing result information corresponding to the work order data to be processed, so that machine learning can be used to simulate and replace manual data analysis and method construction, and try to eliminate the negative impact caused by misjudgment and incomplete analysis due to the analysis ability limitations and subjectivity of analysts and transfer personnel, thereby improving the accuracy of analysis and transfer to a certain extent, and improving the effectiveness and efficiency of processing work order data.

[0072] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 A flowchart of a method for training a work order processing model provided by the present invention;

[0074] Figure 2 A schematic diagram of an algorithm used in a training work order processing model provided by the present invention;

[0075] Figure 3 A schematic diagram of a page for an algorithm for selecting a training model provided by the present invention;

[0076] Figure 4 A flowchart of an intelligent processing method for work order data provided by the present invention;

[0077] Figure 5A schematic diagram of an intelligent processing device for work order data provided by the present invention;

[0078] Figure 6 This is a structural diagram of a computer device provided by the present invention. DETAILED DESCRIPTION

[0079] The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of embodiments consistent with some aspects of the present invention.

[0080] The terms used in this invention are for the purpose of describing specific embodiments only and are not intended to limit the invention. The singular forms "a," "the," and "the" used in this invention are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in this invention refers to and includes any or all possible combinations of one or more of the associated listed items.

[0081] Unless otherwise specified, ordinal numbers such as "first" and "second" mentioned in the present invention are used to distinguish multiple objects and are not used to limit the order, timing, priority or importance of multiple objects.

[0082] As previously mentioned, existing technologies suffer from low effectiveness and efficiency in processing work order data. In light of this, the present invention provides an intelligent work order data processing method that uses machine learning to determine the effectiveness of pending work order data and intelligently assigns processing. This approach can improve the effectiveness and efficiency of work order data processing.

[0083] An embodiment of the present invention provides a method for intelligently processing work order data. This method can be executed by a computer device. For example, the computer device can be a terminal device such as a laptop, personal computer, smartphone, personal digital assistant (PDA), or personal computer, or a server such as a personal computer, a large or medium-sized computer, or a computer cluster, although this embodiment of the present invention does not limit this.

[0084] The following combination Figure 4 The method flow chart shown in the figure illustrates the intelligent processing method of work order data in the embodiment of the present invention. Figure 4 The steps shown can be executed by the aforementioned computer device.

[0085] In an embodiment of the present invention, new work order data can be processed by a trained work order processing model. Before introducing the intelligent processing method for work order data provided by the present invention, the following first introduces the process of obtaining a trained preset work order processing model in an embodiment of the present invention. Figure 1 As shown, the process may include the following steps:

[0086] S101: Determine a plurality of work order data, and process the plurality of work order data using a synthetic minority class oversampling algorithm to obtain a plurality of invalid work order data to be trained.

[0087] In an embodiment of the present invention, historical work order data can be obtained from the original work order system or manually imported into the original work order system and stored in an Elasticsearch distributed database. The original work order system can be the system used by the operator that receives the user's consultation call, the system can be used by a subsidiary, or the system used by the head office, without limitation in this embodiment of the present invention.

[0088] In an embodiment of the present invention, data preprocessing can be performed on work order data in a distributed database. For example, data preprocessing can include Jieba word segmentation processing and stop word processing. Jieba word segmentation processing can segment sentences in a long text into phrases and, based on a custom operator dictionary, prevent some phrases from being segmented. Stop word processing can remove "invalid words" from the words that have undergone Jieba word segmentation processing by using a custom operator stop word library.

[0089] In embodiments of the present invention, feature processing can also be performed on the work order data in the distributed database. For example, feature processing can include part-of-speech analysis, entity recognition, and feature encoding. Part-of-speech analysis can be understood as extracting nouns from historical work order data, entity recognition can be understood as identifying entities in historical work order data, and feature encoding can be understood as converting text in historical work order data into vectors.

[0090] In some embodiments, the present invention can provide a system that allows users to select a coding algorithm, and then determine the target feature coding algorithm based on the selected algorithm. For example, during the feature coding process, the present invention uses three models for encoding: TF-IDF, Word2vec, and pre-trained BERT, which are currently the mainstream feature coding models.

[0091] In this embodiment of the present invention, considering the relatively small amount of invalid work order data available, the synthetic minority oversampling (SMOTE) algorithm is used to obtain invalid work order data. For example, multiple work order data sets can be obtained from a distributed database and processed using the SMOTE algorithm to obtain multiple invalid work order data sets for training.

[0092] In some embodiments, processing the plurality of work order data using the SMOTE algorithm to obtain a plurality of invalid work order data to be trained may include the following steps:

[0093] Step a: Classify the small amount of work order data into a set M. The small amount of work order data may be part of the work order data obtained from the distributed database.

[0094] Step b: Randomly select a data A from the small sample set M.

[0095] Step c: Calculate the Euclidean distance between A and other samples in the small sample set M, and select the k points with the shortest distance.

[0096] Step d: Randomly select a sample point B from these k points.

[0097] Step e: Take a random data on the line connecting sample point A and sample point B as the latest amplified data C.

[0098] Step f: Repeat step be. The number of iterations is the amount of data that needs to be amplified.

[0099] In this embodiment of the present invention, in step f, the amount of data to be amplified is the amount of invalid work order data. As an example, the ratio of the amount of invalid work order data to the amount of valid work order data can be 1:1. For example, if the amount of valid work order data is 300, the amount of data to be amplified can be 300. In this way, a plurality of invalid work order data can be obtained for training.

[0100] S102. Input multiple invalid work order data and valid work order data to be trained into a preset work order processing model for training to obtain a first training result, and adjust the sub-model that performs pre-judgment processing in the preset work order processing model based on the first training result to obtain a sub-model that performs pre-judgment processing in the trained work order processing model.

[0101] In an embodiment of the present invention, the sub-model that performs the predictive processing can be a binary classification model, that is, after a work order data is input into the trained work order processing model to perform the predictive processing sub-model, the processing result obtained includes information characterizing that the work order data is valid data, or the processing result obtained includes information characterizing that the work order data is invalid data.

[0102] S103. Input multiple valid work order data marked with business types into the preset work order processing model for training to obtain a second training result, and adjust the sub-model that performs transfer processing in the preset work order processing model based on the second training result to obtain the sub-model that performs transfer processing in the trained work order processing model.

[0103] In an embodiment of the present invention, the sub-model that performs the transfer processing is a multi-classification model. Taking into account the data imbalance in the work order transfer, and the different categories of the work order transfer processing scenario mainly have different meanings of the work order description, it is possible to consider generating highly similar texts based on EDA data amplification (synonym replacement, random insertion, random exchange) or SimBert. Among them, Simbert is based on the BERT model and uses the UniLM idea to design a task that integrates retrieval and generation. It is trained using a large number of similar sentence pairs, mainly using BERT and UniLM, a set of supervised models implemented by comparative learning technology, and is a good tool for generating similar work orders.

[0104] Since the two aforementioned methods for handling data imbalance hardly change the meaning of the original work order description and can accurately determine the category of the work order data, they are suitable for this work order transfer processing scenario.

[0105] In order to improve the accuracy of work order transfer, an embodiment of the present invention trains 1 large classification model + n small classification models. Among them, the computer device first trains a large classification model based on the "major business category", and then trains n small models based on the "minor business category" to perform the sub-model of work order transfer. In other words, the sub-model that performs work order transfer includes a large classification model that performs the classification of the "major business category" and n models that perform the classification of the "minor business category". Among them, n is equal to the number of "major business categories" in the data to be trained. It should be noted that, for the sake of convenience of description, in the following text, the "major business category" can also be referred to as the first business category, and the "minor business category" can be referred to as the second business category.

[0106] In this embodiment of the present invention, valid work order data is labeled with two types: "First Service Category" and "Second Service Category," and each "First Service Category" typically includes multiple "Second Service Categories." For example, the major service categories include billing, finance, and broadband. When the first service category is billing, it includes second service categories such as data charges, voice charges, and data and telecommunications charges. When the first service category is broadband, it includes second service categories such as broadband basic information, group broadband, and broadband dedicated line.

[0107] When executing work order transfer in the embodiment of the present invention, the "major business category" is first predicted by the primary model, and then the "minor business category" is predicted based on the secondary model corresponding to the predicted "major business category".

[0108] In an embodiment of the present invention, in the sub-model training for executing work order transfer, the present invention selects the most typical Logistic Regression algorithm in machine learning, two integrated learning models Random Forest and XGBoost algorithms, the deep learning model Bert for fine-tuning and the Bert+LSTM algorithm.

[0109] See Figure 2 , Figure 2 The figure shows three feature encoding algorithms: term frequency–inverse document frequency (TF-IDF), word to vector (Word2vec), and bidirectional encoder representations from transformers (BERT), which uses a multi-head attention mechanism. The figure also shows the algorithms used in classification models: logistic regression, random forest, extreme gradient boosting (XGBoost), long short-term memory (LSTM), and multi-class classification (softmax). For example, if TF-IDF is used for feature encoding, the classification model can use logistic regression and XGBoost.

[0110] like Figure 2 As shown in the figure, in the feature encoding process, the present invention uses three algorithms: TF-IDF, Word2vec, and pre-trained Bert for encoding; in the classification model, the present invention uses the most typical Logistic Regression algorithm in machine learning, two integrated learning models Random Forest and XGBoost algorithms, the deep learning model Bert for fine-tuning, and the Bert+LSTM algorithm.

[0111] S104. Obtain a trained work order processing model according to the sub-model that performs the pre-judgment processing in the trained work order processing model and the sub-model that performs the transfer processing in the trained work order processing model.

[0112] In this embodiment of the present invention, two large models and n small models are trained. One large model is a sub-model used for pre-judgment processing. The execution logic of this model is classic binary classification, which can easily obtain the model's evaluation index f1. Furthermore, one large model and n small models are sub-models used for dispatch processing. Adjustments to this sub-model primarily focus on the evaluation index f2 of the n small models. Therefore, training for the preset work order processing model primarily focuses on the two evaluation indicators f1 and f2. When using a multi-algorithm, multi-parameter model, multiple experiments can be established, and the model with the best evaluation index can be selected for release and application.

[0113] For example, Figure 3 The algorithm training page shown includes the experiment name, a list of the experiment's completion status, information about whether to publish the selected function, model training information, and model evaluation information. It should be noted that in actual implementation, the algorithm training page may also include other information, which is not limited in this embodiment of the present invention.

[0114] In the specific implementation process, the person training the model can Figure 3 Select the model training algorithm on the page shown, and determine the final released work order processing model, which is the trained work order processing model mentioned above.

[0115] In an embodiment of the present invention, after obtaining the trained work order processing model, the trained work order processing model can be connected to a system in a work order automatic flow scenario.

[0116] In some embodiments, if the system in other work order automatic flow scenarios (for example, an object sales scenario that requires a consultation call) needs to access the trained work order processing model, the work order data in the aforementioned step S101 can be the historical work order data corresponding to the system in that scenario, and the aforementioned steps S102-S104 are executed, so that the trained work order processing model corresponding to the system in different scenarios can be obtained.

[0117] Furthermore, in an embodiment of the present invention, after obtaining the trained work order processing model, the work orders to be processed can be transferred for processing according to the model.

[0118] The following will be combined Figure 1-4 , the intelligent processing method of work order data provided by the present invention is introduced in detail.

[0119] like Figure 4 As shown, the intelligent processing method of work order data provided by the present invention may include the following steps:

[0120] S201: Receive work order data to be processed.

[0121] In an embodiment of the present invention, the computer device may receive work order data to be processed, wherein the work order data to be processed may be generated based on user voice information or operation information.

[0122] S202 . Predictively process the work order data to be processed using the trained work order processing model to obtain a processing result. The processing result includes information indicating whether the work order data to be processed is valid data.

[0123] In an embodiment of the present invention, the computer device uses the sub-model for executing pre-judgment processing in the trained work order processing model generated by the training in the aforementioned steps S101 to S104 to perform pre-judgment processing on the work order data to be processed and obtain a processing result.

[0124] S203 , using the trained work order processing model to transfer the work order data to be processed according to the processing result, and obtaining a transfer result.

[0125] In some embodiments, if the processing result includes information that the work order data to be processed is invalid data, the transfer processing is determined to be a return modification processing, and according to the set matching set, the processing result information matching the return modification processing is determined to be a re-modification of the work order content.

[0126] In some embodiments, if the processing result includes information that the work order data to be processed is valid data, the following steps may be performed to determine the transfer result:

[0127] Step A: Perform a first-level transfer process on the work order data to be processed using the trained work order processing model to obtain a first-level transfer result;

[0128] In an embodiment of the present invention, a computer device can determine a first match score between pending work order data and at least one first business category using a trained work order processing model to obtain at least one first match score; if a target match score exists in the at least one first match score that satisfies a first screening condition, the computer device can determine the first business category corresponding to the target match score; determine a second match score between the pending work order data and each second business category within the first business category; and obtain a first-level dispatch result based on the obtained second match scores. The first screening condition is exceeding a preset first threshold, such as 80 points. The preset first threshold can be updated based on actual implementation, and this is not limited in the embodiments of the present invention.

[0129] In some embodiments, if the obtained second matching scores do not contain a target matching score that satisfies the second screening condition, manual processing is selected as the first assignment result. The second screening condition may be, for example, exceeding a preset second threshold, such as 70 points. The preset second threshold may be updated based on actual implementation and is not limited in the embodiments of the present invention.

[0130] In some embodiments, if there is a target matching score that satisfies the second screening condition among the obtained second matching scores, the pending confirmation process is taken as the first transfer result.

[0131] Step B: Perform secondary transfer processing on the primary transfer results and the pending work order data through the trained work order processing model to obtain the secondary transfer results.

[0132] In some embodiments, if the first transfer result is transfer to manual processing, it is possible to analyze whether the work order data to be processed includes specific language vocabulary. If it is determined that the work order data to be processed includes specific language vocabulary, the specific language vocabulary will be transferred to manual processing as a secondary transfer result, wherein the specific language vocabulary can be the language vocabulary of a certain region.

[0133] In some embodiments, if the first transfer result is pending confirmation, it is determined whether the work order data to be processed includes custom keywords; if the work order data to be processed includes custom keywords, it is transferred to intelligent processing as the second transfer result; if the work order data to be processed does not include custom keywords, it is transferred to manual processing as the second transfer result.

[0134] It can be seen that in the embodiment of the present invention, the computer device predicts that the work order data to be processed belongs to the "first business category" through a large classification model in the trained work order processing model, and then selects a specified small classification model according to the predicted "first business category" type to predict the "second business category" type and the probability value of the predicted category, which is the aforementioned matching score.

[0135] In some embodiments, it is determined whether the probability value of the "second business category" to which the work order data to be processed is predicted to belong meets the probability threshold, for example: whether the probability value of the prediction of "broadband basic information" is greater than 50%. If the probability value of the "second business category" to which the work order data to be processed is predicted to belong meets the probability threshold, it is determined whether it contains keywords, otherwise the work order data to be processed is manually transferred and replied. Furthermore, it can be determined whether the work order data to be processed contains keywords customized by the operator. If the work order data to be processed contains keywords customized by the operator, it is automatically transferred, i.e., intelligently processed, otherwise it is manually transferred and replied.

[0136] S204: Match the transfer results using the trained work order processing model to obtain processing result information corresponding to the work order data to be processed.

[0137] In an embodiment of the present invention, a computer device may determine a target second business category based on a second matching score that satisfies a second screening condition, and use a preset response message associated with the target second business category as processing result information corresponding to the pending work order data. The preset response message automatically queries the work order user's business information, provides a work order processing suggestion, and uses the suggested processing suggestion as processing result information, which is then fed back to the user who called the consultation hotline.

[0138] As an example, in order to help those skilled in the art understand the possible forms of expression of the above-mentioned preset reply information, it will be explained below through examples.

[0139] For example, taking the target second service category as abnormal call charges, the pre-set response message associated with abnormal call charges could be: Dear user A (e.g., the user's mobile phone number), your call charge details for month X (e.g., July) are B (e.g., monthly charge 90 yuan, package fee 29 yuan, data charge 45 yuan, and value-added service fee 16 yuan). In the cost details for month X, service C (e.g., data charge) increased by E (e.g., 30 yuan) compared to the previous month's D (e.g., 15 yuan). In this pre-set response message, A, X, B, C, D, and E are variable information and can be obtained by querying the service system. In other words, the pre-set response message associated with abnormal call charges can contain both fixed and variable information, with the variable information being obtained by querying the service system. Fixed information can also be understood as template information.

[0140] It should be noted that the above-mentioned preset reply information can be set according to actual needs, and the present invention does not limit this.

[0141] Based on the same inventive concept, an embodiment of the present invention provides an intelligent processing device for work order data. Figure 5 As shown, the intelligent processing device 300 for work order data includes a receiving unit 301, a pre-judgment unit 302, a transfer unit 303 and a reply unit 304.

[0142] Receiving unit 301, for receiving work order data to be processed;

[0143] The pre-judgment unit 302 is used to pre-judgment the work order data to be processed using the trained work order processing model to obtain a processing result; the processing result includes information indicating whether the work order data to be processed is valid data;

[0144] The transfer unit 303 is used to transfer the work order data to be processed according to the processing result using the trained work order processing model to obtain a transfer result;

[0145] The reply unit 304 is used to match the transfer result with the trained work order processing model to obtain processing result information corresponding to the work order data to be processed.

[0146] In one possible design, if the processing result includes information that the work order data to be processed is valid data, the transfer unit 303 is configured to:

[0147] The trained work order processing model performs first-level transfer processing on the pending work order data to obtain the first-level transfer result;

[0148] The trained work order processing model is used to perform secondary transfer processing on the primary transfer results and pending work order data to obtain the secondary transfer results.

[0149] In one possible design, the dispatching unit 303 is specifically configured to:

[0150] Determining a first matching score between the to-be-processed work order data and at least one first business category to obtain at least one first matching score;

[0151] If there is a target matching score that satisfies the first screening condition among the at least one first matching score, determining a first business category corresponding to the target matching score;

[0152] determining a second matching score between the to-be-processed work order data and each second business category in the first business category;

[0153] A first-level transfer result is obtained based on the second matching score obtained.

[0154] In one possible design, the dispatching unit 303 is specifically configured to:

[0155] If the target matching score that satisfies the second screening condition does not exist in the obtained second matching scores, manual processing will be used as the first assignment result;

[0156] If there is a target matching score that satisfies the second screening condition among the obtained second matching scores, the pending confirmation process is taken as the first transfer result.

[0157] In one possible design, the dispatching unit 303 is specifically configured to:

[0158] If the obtained second matching scores include a target matching score that satisfies the second screening condition, determining whether the work order data to be processed includes the custom keyword;

[0159] If the work order data to be processed includes custom keywords, it will be transferred to intelligent processing as the second transfer result;

[0160] If the work order data to be processed does not include custom keywords, it will be transferred to manual processing as the second transfer result.

[0161] In one possible design, the reply unit 304 is specifically configured to:

[0162] determining a target second business category based on a second matching score that satisfies a second screening condition;

[0163] The preset reply information associated with the target second business category is used as the processing result information corresponding to the work order data to be processed.

[0164] In a possible design, the intelligent processing device for work order data further includes a training unit ( Figure 5 (not shown), the training unit is used to:

[0165] Determine a plurality of work order data, and process the plurality of work order data using a synthetic minority class oversampling algorithm to obtain a plurality of invalid work order data to be trained;

[0166] Inputting a plurality of invalid work order data and valid work order data to be trained into a preset work order processing model for training to obtain a first training result, and adjusting a sub-model for performing a pre-judgment process in the preset work order processing model based on the first training result to obtain a sub-model for performing a pre-judgment process in the trained work order processing model;

[0167] Inputting multiple valid work order data marked with business types into a preset work order processing model for training to obtain a second training result, and adjusting the sub-model for performing the dispatch processing in the preset work order processing model based on the second training result to obtain the sub-model for performing the dispatch processing in the trained work order processing model;

[0168] The trained work order processing model is obtained according to the sub-model that performs the prediction processing in the trained work order processing model and the sub-model that performs the dispatch processing in the trained work order processing model.

[0169] It should be noted that the division of units in the embodiments of the present invention is illustrative and represents only one logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional units in various embodiments of the present invention may be integrated into a single processor, exist physically as separate units, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in either hardware or software functional units.

[0170] The intelligent processing device 300 of the work order data in the embodiment of the present invention is the same as the above Figure 4 The intelligent processing method for work order data shown is an invention based on the same concept. Through the above detailed description of the intelligent processing method for work order data, those skilled in the art can clearly understand the implementation process of the intelligent processing device 300 for work order data in this embodiment, so for the sake of conciseness of the specification, it will not be repeated here.

[0171] Based on the same inventive concept, an embodiment of the present invention further provides a computer device. Figure 6 As shown, the computer device 400 may include: at least one memory 401 and at least one processor 402.

[0172] The at least one memory 401 is used to store one or more programs.

[0173] When one or more programs are executed by at least one processor 402, the above Figure 4 The intelligent processing method of work order data is shown.

[0174] The computer device 400 may also optionally include a communication interface, which is used to communicate with external devices and exchange data.

[0175] It should be noted that the memory 401 may include a high-speed RAM memory, and may also include a non-volatile memory (nonvolatile memory), such as at least one disk memory.

[0176] In a specific implementation, if the memory 401, processor 402, and communication interface are integrated on a single chip, the memory 401, processor 402, and communication interface can communicate with each other via an internal interface. If the memory 401, processor 402, and communication interface are implemented independently, the memory 401, processor 402, and communication interface can be connected to each other via a bus and communicate with each other.

[0177] Based on the same inventive concept, an embodiment of the present invention further provides a computer-readable storage medium, which can store at least one program. When the at least one program is executed by a processor, the above Figure 4 The intelligent processing method of work order data is shown.

[0178] It should be understood that a computer-readable storage medium is any data storage device that can store data or programs that can be subsequently read by a computer system. Examples of computer-readable storage media include read-only memory, random access memory, CD-ROMs, HDDs, DVDs, magnetic tapes, and optical data storage devices.

[0179] The computer readable storage medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

[0180] The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wire, optical cable, radio frequency (RF), etc., or any suitable combination thereof.

[0181] The above-described embodiments merely illustrate several implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. An intelligent processing method for work order data, characterized in that: include: Receive work order data to be processed; Predictively process the work order data to be processed using the trained work order processing model to obtain a processing result; The processing result includes information indicating whether the work order data to be processed is valid data; Transferring the to-be-processed work order data according to the processing result using the trained work order processing model to obtain a transfer result; Matching the transfer result with the trained work order processing model to obtain processing result information corresponding to the work order data to be processed; If the processing result includes information that the work order data to be processed is valid data, the trained work order processing model is used to transfer the work order data to be processed according to the processing result to obtain a transfer result, including: Determine a first matching score between the to-be-processed work order data and at least one first business category to obtain at least one first matching score; if a target matching score that satisfies a first filtering condition exists among the at least one first matching score, determine the first business category corresponding to the target matching score; determine a second matching score between the to-be-processed work order data and each second business category in the first business category; if no target matching score that satisfies a second filtering condition exists among the obtained second matching scores, transfer to manual processing is set as a first transfer result; if a target matching score that satisfies the second filtering condition exists among the obtained second matching scores, set pending confirmation processing as the first transfer result; If the second matching scores obtained include a target matching score that satisfies the second screening condition, it is determined whether the work order data to be processed includes custom keywords; if the work order data to be processed includes custom keywords, the work order data is transferred to intelligent processing as the second transfer result; if the work order data to be processed does not include custom keywords, the work order data is transferred to manual processing as the second transfer result.

2. The method according to claim 1, wherein Matching the transfer result with the trained work order processing model to obtain processing result information corresponding to the work order data to be processed, including: determining a target second business category based on a second matching score that satisfies a second screening condition; The preset reply information associated with the target second business category is used as the processing result information corresponding to the work order data to be processed.

3. The method according to claim 1 or 2, wherein: The trained work order processing model is obtained by training in the following manner, including: Determine a plurality of work order data, and process the plurality of work order data using a synthetic minority class oversampling algorithm to obtain a plurality of invalid work order data to be trained; Inputting the plurality of invalid work order data and valid work order data to be trained into a preset work order processing model for training to obtain a first training result, and adjusting the sub-model for performing pre-judgment processing in the preset work order processing model based on the first training result to obtain a sub-model for performing pre-judgment processing in the trained work order processing model; Inputting a plurality of valid work order data marked with business types into the preset work order processing model for training to obtain a second training result, and adjusting a sub-model for performing a dispatch process in the preset work order processing model based on the second training result to obtain a sub-model for performing a dispatch process in the trained work order processing model; A trained work order processing model is obtained based on the sub-model that performs pre-judgment processing in the trained work order processing model and the sub-model that performs transfer processing in the trained work order processing model.

4. An intelligent processing device for work order data, characterized in that: include: The receiving unit is used to receive the work order data to be processed; The pre-judgment unit is used to pre-judgment the work order data to be processed using the trained work order processing model to obtain a processing result; The processing result includes information indicating whether the work order data to be processed is valid data; The transfer unit is used to transfer the work order data to be processed according to the processing result using the trained work order processing model to obtain a transfer result; The reply unit is used to match the transfer result with the trained work order processing model to obtain processing result information corresponding to the work order data to be processed; If the processing result includes information that the work order data to be processed is valid data, the transfer unit, when performing transfer processing on the work order data to be processed according to the processing result using the trained work order processing model to obtain the transfer result, is specifically configured to: Determine a first matching score between the to-be-processed work order data and at least one first business category to obtain at least one first matching score; if a target matching score that satisfies a first filtering condition exists among the at least one first matching score, determine the first business category corresponding to the target matching score; determine a second matching score between the to-be-processed work order data and each second business category in the first business category; if no target matching score that satisfies a second filtering condition exists among the obtained second matching scores, transfer to manual processing is set as a first transfer result; if a target matching score that satisfies the second filtering condition exists among the obtained second matching scores, set pending confirmation processing as the first transfer result; If the second matching scores obtained include a target matching score that satisfies the second screening condition, it is determined whether the work order data to be processed includes custom keywords; if the work order data to be processed includes custom keywords, the work order data is transferred to intelligent processing as the second transfer result; if the work order data to be processed does not include custom keywords, the work order data is transferred to manual processing as the second transfer result.

5. A computer device, characterized in that: include: at least one memory and at least one processor; The at least one memory is used to store one or more programs; When the one or more programs are executed by the at least one processor, the method according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one program; when the at least one program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images