CN111611353A

CN111611353A - Screening method and device, electronic equipment and computer readable storage medium

Info

Publication number: CN111611353A
Application number: CN201910139175.3A
Authority: CN
Inventors: 冯浩; 吴康康; 徐江; 王鹏; 李奘
Original assignee: Beijing Didi Infinity Technology and Development Co Ltd
Current assignee: Beijing Didi Infinity Technology and Development Co Ltd
Priority date: 2019-02-25
Filing date: 2019-02-25
Publication date: 2020-09-01
Anticipated expiration: 2039-02-25
Also published as: CN111611353B

Abstract

The application provides a screening method, a screening device, electronic equipment and a computer-readable storage medium, wherein the method comprises the following steps: acquiring work order data in an online service process, and performing word segmentation processing on the work order data; performing target processing on each alternative keyword in the work order data to obtain at least one new work order data; classifying the work order data and at least one new work order data by using the target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not; determining a target keyword in the alternative keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safety type work order. The embodiment of the application can screen the work order data in a machine learning mode to determine whether the work order data is a safe work order or not, so that the technical problems of low screening efficiency and poor accuracy when the emergency safe work order is screened in the prior art are solved.

Description

Screening method and device, electronic equipment and computer readable storage medium

Technical Field

The present application relates to the field of data processing technologies, and in particular, to a screening method, an apparatus, an electronic device, and a computer-readable storage medium.

Background

An online service mode exists in a customer service system of a network car appointment platform, the online service mode mainly solves the problems of users in a text information mode, for example, the users can reflect various service problems in an online service mode, and customer service personnel solve and process the problems reflected by the users online.

In order to ensure the service quality of the network car booking, the network car booking platform generally screens abnormal work orders from a large number of work orders based on the safety problem fed back to customer service by a user. The current screening method for the abnormal work orders of the network appointment platform is mainly a manual method. With the continuous increase of the number of service work orders, the efficiency and the accuracy of the manual screening mode are low, and the actual requirements cannot be met.

Disclosure of Invention

In view of this, an object of the embodiments of the present application is to provide a screening method, an apparatus, an electronic device, and a computer-readable storage medium, which can screen work order data in a machine learning manner to determine whether the work order data to be processed is a safe work order, so as to alleviate technical problems of low screening efficiency and poor accuracy when screening an emergency safe work order by using the prior art.

According to one aspect of the present application, an electronic device is provided that may include one or more storage media and one or more processors in communication with the storage media. One or more storage media store machine-readable instructions executable by a processor. When the electronic device is operated, the processor communicates with the storage medium through the bus, and the processor executes the machine readable instructions to perform one or more of the following operations:

acquiring work order data in an online service process, and performing word segmentation processing on the work order data; each work order data comprises session data between a session service provider and a target object, and the target object comprises an order service provider and/or an order service requester; performing target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the worksheet data; classifying the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not; determining target keywords in the alternative keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

In a preferred embodiment of the present application, the classifying the work order data and the at least one new work order data using the target classification model includes: if the number of the alternative keywords exceeds the preset number, processing the at least one new worksheet data to obtain at least one data to be processed, wherein each data to be processed comprises a plurality of new worksheets and one worksheet data; and classifying the new work order data in each data to be batched and the work order data by using a target classification model to obtain a classification processing result.

In a preferred embodiment of the present application, processing the at least one new work order data to obtain at least one data to be batched includes: acquiring a candidate keyword set; the candidate keyword set comprises a plurality of keywords for representing that the work order data is not a safe work order, and each keyword in the candidate keyword set comprises a weight value which is used for representing the importance degree of the keyword; and processing the at least one new work order data based on the candidate keyword set to obtain the at least one data to be batched.

In a preferred embodiment of the present application, processing the at least one new work order data based on the candidate keyword set to obtain at least one data to be batched includes: determining a first target alternative keyword in the alternative keywords, wherein the first target alternative keyword is a keyword contained in the candidate keyword set; sorting the first target alternative keywords according to the weight values of the first target alternative keywords to obtain a first sorting result; determining a target sorting result based on the first sorting result and a second sorting result, wherein the first sorting result is located before the second sorting result, the second sorting result is obtained after sorting second target alternative keywords, and the second target alternative keywords are other keywords except the first target alternative keywords in the alternative keywords; and according to the sorting sequence in the target sorting result, classifying the new work order data corresponding to each keyword in the target sorting result to obtain the at least one piece of data to be batched.

In a preferred embodiment of the present application, determining the candidate keyword set by the following method specifically includes: acquiring historical work order data, wherein the historical work order data is session data between a session service provider and a target object, and the target object comprises an order service provider and a service requester; determining a set of candidate keywords based on the historical work order data.

In a preferred embodiment of the present application, determining the candidate keyword set based on the historical work order data comprises: performing word segmentation processing on each historical worksheet data to obtain word segmentation results, wherein the word segmentation results comprise a plurality of words; determining target word segmentation in the word segmentation results, wherein the target word segmentation is the word segmentation with the occurrence frequency higher than a preset threshold value in the word segmentation results of the historical work order data; converting the target participles into target data in a one-hot form, wherein the target data comprises a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data; determining the set of candidate keywords using the target data.

In a preferred embodiment of the present application, determining the candidate keyword set using the target data includes: determining training samples and test samples based on the target data; training an initial classification prediction model by using the training samples to obtain a target classification prediction model; processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation; and determining the candidate keyword set based on the weight value of each target participle.

In a preferred embodiment of the present application, performing a word segmentation process on each piece of historical work order data to obtain a word segmentation result includes: determining a preset word segmentation template; and performing word segmentation processing on each historical worksheet data through a word segmentation device according to the preset word segmentation template to obtain word segmentation results.

In a preferred embodiment of the present application, the method further comprises: constructing a training sample according to the word segmentation result of each piece of historical work order data and the label information of each piece of historical work order data, wherein the label information is used for representing whether the historical work order data is a safe work order; and training an initial model of the target classification model by using the training sample to obtain the target classification model.

In a preferred embodiment of the present application, the classifying the work order data and the at least one new work order data by using the target classification model further includes: if the number of the alternative keywords does not exceed the preset number, taking the at least one new work order data and the work order data as data to be processed in batches; and classifying the new work order data in the data to be batched and the work order data by using a target classification model to obtain a classification processing result.

In a preferred embodiment of the present application, the performing target processing on each candidate keyword in the work order data includes: and replacing the alternative keywords Ai in the work order data with preset data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, I is the number of the alternative keywords, and the preset data is useless words.

In a preferred embodiment of the present application, the performing target processing on each candidate keyword in the work order data includes: and deleting the alternative keywords Ai in the work order data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, and I is the number of the alternative keywords.

In a preferred embodiment of the present application, the classification processing result includes a plurality of sub-processing results, and the plurality of sub-processing results include a first sub-processing result and a second sub-processing result, where the classification processing result of the work order data is the first sub-processing result, and each new work order data corresponds to one second sub-processing result.

In a preferred embodiment of the present application, determining a target keyword among the candidate keywords based on the classification processing result includes: calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values; determining the target keyword based on the plurality of variance values.

In a preferred embodiment of the present application, determining the target keyword based on the plurality of variance values includes: determining a first target variance value among the plurality of variance values, wherein the first target variance value is the first N largest variance values among the plurality of variance values, and N is a positive integer greater than zero; determining new work order data corresponding to the first target change value; and determining the target keyword according to the third target alternative keyword after the replacement or deletion in the corresponding new work order data.

In a preferred embodiment of the present application, determining the target keyword according to a third target candidate keyword after replacement or deletion in the corresponding new work order data includes: if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords comprise keywords which are recorded in the candidate keyword set, determining M keywords with the largest weight values in the fourth target alternative keywords; the candidate keyword set comprises a plurality of keywords for representing whether the work order data is a safe work order, each keyword in the candidate keyword set comprises a weight value, the weight values are used for representing the importance degrees of the keywords, and the fourth target alternative keyword is a keyword except the third target alternative keyword in the plurality of change values; and determining the keywords with the largest weight values in the first M determined keywords and the keywords contained in the candidate keyword set in the third target candidate keywords as the target keywords.

In a preferred embodiment of the present application, taking the alternative keyword after replacement or deletion in the corresponding new work order data as the target keyword further includes: and if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords do not comprise the keywords which are recorded in the candidate keyword set, determining the third target alternative keywords which are replaced or deleted in the corresponding new work order data as the target keywords.

According to another aspect of the present application, there is also provided a screening apparatus comprising: the system comprises an acquisition unit, a word segmentation unit and a word segmentation unit, wherein the acquisition unit is used for acquiring work order data in an online service process and performing word segmentation on the work order data; each work order data comprises session data between a session service provider and a target object, and the target object comprises an order service provider and/or an order service requester; the keyword processing unit is used for carrying out target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the worksheet data; the classification unit is used for classifying the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not; a determining unit, configured to determine a target keyword among the candidate keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

In a preferred embodiment of the present application, the classification unit includes: the processing module is used for processing the at least one new worksheet data to obtain at least one data to be processed in batches if the number of the alternative keywords exceeds a preset number, wherein each data to be processed comprises a plurality of new worksheets and one worksheet; and the first classification module is used for classifying the new work order data in each data to be batched and the work order data by using a target classification model to obtain the classification processing result.

In a preferred embodiment of the present application, the processing module is configured to: acquiring a candidate keyword set; the candidate keyword set comprises a plurality of keywords for representing that the work order data is not a safe work order, and each keyword in the candidate keyword set comprises a weight value which is used for representing the importance degree of the keyword; and processing the at least one new work order data based on the candidate keyword set to obtain the at least one data to be batched.

In a preferred embodiment of the present application, the processing module is further configured to: determining a first target alternative keyword in the alternative keywords, wherein the first target alternative keyword is a keyword contained in the candidate keyword set; sorting the first target alternative keywords according to the weight values of the first target alternative keywords to obtain a first sorting result; determining a target sorting result based on the first sorting result and a second sorting result, wherein the first sorting result is located before the second sorting result, the second sorting result is obtained after sorting second target alternative keywords, and the second target alternative keywords are other keywords except the first target alternative keywords in the alternative keywords; and according to the sorting sequence in the target sorting result, classifying the new work order data corresponding to each keyword in the target sorting result to obtain the at least one piece of data to be batched.

In a preferred embodiment of the present application, the apparatus further determines the candidate keyword set by the following method, which specifically includes: acquiring historical work order data, wherein the historical work order data is session data between a session service provider and a target object, and the target object comprises an order service provider and a service requester; determining a set of candidate keywords based on the historical work order data.

In a preferred embodiment of the present application, the apparatus is further configured to: performing word segmentation processing on each historical worksheet data to obtain word segmentation results, wherein the word segmentation results comprise a plurality of words; determining target word segmentation in the word segmentation results, wherein the target word segmentation is the word segmentation with the occurrence frequency higher than a preset threshold value in the word segmentation results of the historical work order data; converting the target participles into target data in a one-hot form, wherein the target data comprises a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data; determining the set of candidate keywords using the target data.

In a preferred embodiment of the present application, the apparatus is further configured to: determining training samples and test samples based on the target data; training an initial classification prediction model by using the training samples to obtain a target classification prediction model; processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation; and determining the candidate keyword set based on the weight value of each target participle.

In a preferred embodiment of the present application, the apparatus is further configured to: determining a preset word segmentation template; and performing word segmentation processing on each historical worksheet data through a word segmentation device according to the preset word segmentation template to obtain word segmentation results.

In a preferred embodiment of the present application, the apparatus is further configured to: constructing a training sample according to the word segmentation result of each piece of historical work order data and the label information of each piece of historical work order data, wherein the label information is used for representing whether the historical work order data is a safe work order; and training an initial model of the target classification model by using the training sample to obtain the target classification model.

In a preferred embodiment of the present application, the classifying unit further includes: the first determining module is used for taking the at least one new work order data and the work order data as data to be batched if the number of the alternative keywords does not exceed the preset number; and the second classification module is used for classifying the new work order data in the data to be batched and the work order data by using a target classification model to obtain a classification processing result.

In a preferred embodiment of the present application, the keyword processing unit is configured to: and replacing the alternative keywords Ai in the work order data with preset data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, I is the number of the alternative keywords, and the preset data is useless words.

In a preferred embodiment of the present application, the keyword processing unit is further configured to: and deleting the alternative keywords Ai in the work order data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, and I is the number of the alternative keywords.

In a preferred embodiment of the present application, the determining unit includes: the calculation module is used for calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values; a second determination module to determine the target keyword based on the plurality of variance values.

In a preferred embodiment of the present application, the determining module is configured to: determining a first target variance value among the plurality of variance values, wherein the first target variance value is the first N largest variance values among the plurality of variance values, and N is a positive integer greater than zero; determining new work order data corresponding to the first target change value; and determining the target keyword according to the third target alternative keyword after the replacement or deletion in the corresponding new work order data.

In a preferred embodiment of the present application, the determining module is further configured to: if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords comprise keywords which are recorded in the candidate keyword set, determining M keywords with the largest weight values in the fourth target alternative keywords; the candidate keyword set comprises a plurality of keywords for representing whether the work order data is a safe work order, each keyword in the candidate keyword set comprises a weight value, the weight values are used for representing the importance degrees of the keywords, and the fourth target alternative keyword is a keyword except the third target alternative keyword in the plurality of change values; and determining the keywords with the largest weight values in the first M determined keywords and the keywords contained in the candidate keyword set in the third target candidate keywords as the target keywords.

In a preferred embodiment of the present application, the determining module is further configured to: and if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords do not comprise the keywords which are recorded in the candidate keyword set, determining the third target alternative keywords which are replaced or deleted in the corresponding new work order data as the target keywords.

According to another aspect of the present application, there is also provided an electronic device including: the fir tree screening method comprises a processor, a storage medium and a bus, wherein the storage medium stores machine-readable instructions executable by the processor, when an electronic device runs, the processor and the storage medium are communicated through the bus, and the processor executes the machine-readable instructions to execute the steps of the screening method of the fir tree.

According to another aspect of the present application, there is also provided a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the screening method described above.

The method comprises the steps of firstly obtaining work order data in an online service process, carrying out word segmentation processing on the work order data, and then carrying out replacement or deletion processing on each alternative keyword in the work order data in sequence to obtain at least one new work order data; then, classifying the work order data and at least one new work order data by using the target classification model to obtain a classification result; and finally, determining the target keywords in the alternative keywords based on the classification processing result. According to the method and the device, the work order data can be screened in a machine learning mode to determine whether the work order data is a safe work order, so that the technical problems that the screening efficiency is low and the accuracy is poor when an emergency safe work order is screened in the prior art are solved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and for those skilled in the art, other related drawings can be obtained from the drawings without inventive effort.

Fig. 1 shows a schematic structural diagram of an electronic device provided in an embodiment of the present application;

FIG. 2 is a flow chart illustrating a screening method provided by an embodiment of the present application;

fig. 3 is a flowchart illustrating a first alternative screening method provided in an embodiment of the present application;

FIG. 4 is a flow chart illustrating a second alternative screening method provided by embodiments of the present application;

FIG. 5 is a flow chart of a third alternative screening method provided in embodiments of the present application;

fig. 6 shows a flowchart of a fourth alternative screening method provided in the embodiments of the present application;

fig. 7 is a flowchart illustrating a fifth alternative screening method provided in the embodiments of the present application;

FIG. 8 is a schematic flow chart of another screening method provided in the embodiments of the present application;

fig. 9 shows a schematic view of a screening apparatus provided in an embodiment of the present application.

Detailed Description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it should be understood that the drawings in the present application are for illustrative and descriptive purposes only and are not used to limit the scope of protection of the present application. Additionally, it should be understood that the schematic drawings are not necessarily drawn to scale. The flowcharts used in this application illustrate operations implemented according to some embodiments of the present application. It should be understood that the operations of the flow diagrams may be performed out of order, and steps without logical context may be performed in reverse order or simultaneously. One skilled in the art, under the guidance of this application, may add one or more other operations to, or remove one or more operations from, the flowchart.

In addition, the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. The components of the embodiments of the present application, generally described and illustrated in the figures herein, can be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, presented in the accompanying drawings, is not intended to limit the scope of the claimed application, but is merely representative of selected embodiments of the application. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that in the embodiments of the present application, the term "comprising" is used to indicate the presence of the features stated hereinafter, but does not exclude the addition of further features.

The terms "service" and "order" are used interchangeably herein to refer to a service request initiated by a passenger, a service requester, a driver, a service provider, a supplier, or the like, or any combination thereof. Accepting the "service" or "order" may be a passenger, a service requester, a driver, a service provider, a supplier, or the like, or any combination thereof. The service may be charged or free.

Example 1:

fig. 1 illustrates a schematic diagram of exemplary hardware and software components of an electronic device 100 that may implement the screening methods provided herein, according to some embodiments of the present application.

The electronic device 100 may be a general purpose computer or a special purpose computer, both of which may be used to implement the screening methods of the present application. Although only a single computer is shown, for convenience, the functions described herein may be implemented in a distributed fashion across multiple similar platforms to balance processing loads.

For example, electronic device 100 may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and a storage medium 140 of a different form.

The processor 102 may be a Central Processing Unit (CPU) or other form of Processing Unit having data Processing capabilities and/or instruction execution capabilities, and may control other components in the electronic device 100 to perform desired functions.

The storage medium 104 may include one or more computer program products that may include various forms of computer-readable storage media, such as a disk, ROM, or RAM, or any combination thereof. Illustratively, the computer platform may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The method of the present application may be implemented in accordance with these program instructions. The electronic device 100 also includes an Input/Output (I/O) interface 150 between the computer and other Input/Output devices (e.g., keyboard, display screen).

The storage medium 140 stores machine-readable instructions executable by the processor 120, and when the electronic device is operated, the processor 120 communicates with the storage medium 140 through the bus, and the processor executes the machine-readable instructions to perform the steps of the screening method described below. In addition, the storage medium may also be referred to as a computer-readable storage medium, on which a computer program is stored, which computer program, when being executed by a processor, performs the steps of the screening method described below.

In particular, when the electronic device is operating, the processor 120 executes machine-readable instructions by communicating between the processor 120 and the storage medium 140 via a bus.

An obtaining unit in the processor 120 is configured to obtain work order data in an online service process, and perform word segmentation on the work order data; each of the work order data includes session data between a session service provider and a target object, and the target object includes an order service provider and/or an order service requester.

Then, the keyword processing unit in the processor 120 performs target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the work order data.

Next, the classification unit in the processor 120 performs classification processing on the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; and the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not.

Finally, a determining unit in the processor 120 determines a target keyword among the alternative keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

For ease of illustration, only one processor is depicted in electronic device 100. However, it should be noted that the electronic device 100 in the present application may also comprise a plurality of processors, and thus the steps performed by one processor described in the present application may also be performed by a plurality of processors in combination or individually. For example, if the processor of the electronic device 100 executes steps a and B, it should be understood that steps a and B may also be executed by two different processors together or separately in one processor. For example, a first processor performs step a and a second processor performs step B, or the first processor and the second processor perform steps a and B together.

Example 2:

see figure 2 for a flow chart of one screening method.

The screening method shown in fig. 2 is described by taking an application at a server side as an example, and the method includes the following steps:

step S202, acquiring work order data in an online service process, and performing word segmentation processing on the work order data; each of the work order data includes session data between a session service provider and a target object, and the target object includes an order service provider and/or an order service requester.

During the service order execution process, the service requester can perform session communication with the session service provider through a telephone service mode or an online service mode. The session communication content can be used as the work order data of the service order.

In the process of executing the service order, the order service provider can also communicate with the session service provider in a telephone service mode or an online service mode. The session communication content can be used as the work order data of the service order.

Step S204, performing target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the work order data.

In this embodiment, the data included in the work order data is session data (or session information) between the session service provider and the target object.

The alternative keywords are obtained after word segmentation processing is performed on the work order data, and the alternative keywords may include keywords capable of determining whether the work order data is a safety work order.

In this embodiment, the candidate keywords in the work order data may be replaced with useless words, so as to obtain new work order data; or, the alternative keywords in the work order data can be deleted to obtain new work order data.

One purpose of performing targeted processing on the candidate keywords is to determine the probability that the work order data is determined to be an insecure type work order in the event that the work order data does not contain the candidate keywords.

Step S206, classifying the work order data and the at least one new work order data by using a target classification model to obtain a classification result; and the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not.

In this embodiment, after the new work order data and the work order data are obtained, the new work order data and the work order data may be classified by using the target classification model, so as to obtain a classification result.

And the classification processing result corresponding to the new work order data is a classification processing result obtained after the alternative keywords are removed, and the classification processing result can determine the influence degree of the alternative keywords on the classification processing result.

Step S208, determining target keywords in the alternative keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

In this embodiment, after obtaining the new work order data and the classification processing result of the work order data, a target keyword capable of representing the work order data as an unsafe work order may be determined in the candidate keywords by combining the classification processing result.

As can be seen from the above description, in the present embodiment, during the execution of the service order, the session data between the session service provider (i.e., customer service) and the target object can be recorded in real time. And using the session data as work order data for the service order. After the work order data is obtained, the work order data may be stored in a data platform. The session data may be voice call information or text communication information. The target object includes an order service provider (i.e., a driver of the network appointment platform) and/or a service requester (passenger).

After the execution instruction of the above step S202 to step S208 is acquired, the work order data is acquired from the data platform as the work order data to be processed. And then, performing word segmentation processing on the worksheet data to obtain a word segmentation result, wherein the word segmentation result comprises at least one word segmentation. The at least one word segmentation is a candidate keyword in the work order data.

Thereafter, each candidate keyword in the work order data may be targeted.

In this embodiment, each candidate keyword in the work order data may be subjected to target processing in the following two ways.

In a first manner, as shown in fig. 3, in step S204, the target processing for each candidate keyword in the work order data includes the following steps:

step S301, replacing alternative keywords Ai in the work order data with preset data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, I is the number of the alternative keywords, and the preset data are useless words.

In this embodiment, the unnecessary vocabulary is a vocabulary that has no influence on whether or not the work order data is a safe work order. Optionally, in this embodiment, the garbage vocabulary may be selected as "OOV", and in addition, other garbage vocabularies may be selected, which is not specifically limited in this embodiment.

For example, the work order data includes I candidate keywords. Then the alternative keyword a1 in the work order data can be replaced by "OOV", and new work order data B1 is obtained after replacement; then, replacing the alternative keyword A2 in the work order data with OOV to obtain new work order data B2; by analogy, each of the alternative keywords from a1 to AI is replaced by "OOV" in the manner described above, and new worksheet data B1 to BI are obtained after the replacement.

It should be noted that, if the candidate keyword Ai in the work order data occurs multiple times, each candidate keyword Ai in the work order data needs to be replaced with a useless vocabulary (e.g., OOV).

As can be seen from the above description, in this embodiment, replacement processing is sequentially performed on each candidate keyword in the work order data, and after the candidate keyword is replaced, the influence of the candidate keyword on the classification processing result of the work order data can be determined, so as to determine whether the candidate keyword is the target keyword. The method for mining the target keywords by eliminating the alternative keywords to evaluate the change of the work order data classification processing result greatly improves the mining efficiency of the keywords, can also ensure that the keywords really play a key role in the safety work orders and reflect the essence of the safety work orders.

In a second way, as shown in fig. 4, in step S204, the target processing for each candidate keyword in the work order data includes the following steps:

step S401, deleting the alternative keywords Ai in the work order data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, and I is the number of the alternative keywords.

In this embodiment, the work order data includes I candidate keywords. Then the alternative keyword a1 in the work order data can be deleted, and new work order data B1 is obtained after deletion; then, deleting the alternative keyword A2 in the work order data to obtain new work order data B2; and by analogy, deleting each alternative keyword from A1 to AI in the manner described above, and obtaining new worksheet data B1 to BI after deletion.

It should be noted that, if the candidate keyword Ai in the work order data appears multiple times, each candidate keyword Ai in the work order data needs to be deleted.

As can be seen from the above description, in this embodiment, deletion processing is performed on each candidate keyword in the work order data in sequence, and after the candidate keyword is deleted, the influence of the candidate keyword on the classification processing result of the work order data can be determined, so as to determine whether the candidate keyword is a target keyword. The method for mining the target keywords by eliminating the alternative keywords to evaluate the change of the work order data classification processing result greatly improves the mining efficiency of the keywords, can also ensure that the keywords really play a key role in the safety work orders and reflect the essence of the safety work orders.

It should be further noted that, in this embodiment, target processing may be performed on each candidate keyword in the work order data in combination with the above-mentioned first and second manners, for example, performing replacement processing on part of the candidate keywords, performing deletion processing on part of the candidate keywords, and the like.

In the invention, after at least one new work order data is obtained according to the above-described manner, the work order data and the at least one new work order data can be classified by using the target classification model, so as to obtain a classification result.

In an alternative embodiment, as shown in fig. 5, in step S206, the classifying the work order data and the at least one new work order data by using the target classification model further includes:

step S501, if the number of the alternative keywords does not exceed the preset number, the at least one new work order data and the work order data are used as data to be batched;

step S502, a target classification model is used for classifying new work order data in the data to be batched and the work order data to obtain a classification processing result.

In order to improve the evaluation speed and efficiency, the invention adopts a batch processing mode to process one piece of dialogue data at a time, wherein the first line of record in the batch is complete data (namely, work order data), and the data of other lines are data after the alternative keywords are removed or replaced (namely, at least one piece of new work order data).

Alternatively, the size of the batch can be selected to be 64, and 63 candidate keywords can be evaluated at most at one time. If the stop words are removed from the work order data and the number of the alternative keywords is less than 63, at least one new work order data and the work order data can be used as data to be batched, and then the new work order data and the work order data in the data to be batched are classified by using the target classification model to obtain a classification result.

In an alternative embodiment, as shown in fig. 6, in step S206, the classifying the work order data and the at least one new work order data by using the target classification model includes the following steps:

step S601, if the number of the alternative keywords exceeds the preset number, processing the at least one new worksheet data to obtain at least one data to be processed, wherein each data to be processed comprises a plurality of new worksheets and one worksheet data;

step S602, a target classification model is used for classifying new work order data in each data to be batched and the work order data to obtain a classification processing result.

Alternatively, the size of the batch can be selected to be 64, and 63 candidate keywords can be evaluated at most at one time. In this embodiment, if the number of candidate keywords excluding the stop word is greater than 63 in the work order data, at least one new work order data may be batch processed. For example, at least one new work order data is batch processed to obtain at least one data to be batch processed. For example, each pending data may include one work order data and 63 new work order data.

In this embodiment, when step S601 is executed, at least one new work order data may be processed by combining the candidate keyword set to obtain at least one data to be processed, where the specific process is as follows:

firstly, acquiring a candidate keyword set; the candidate keyword set comprises a plurality of keywords for representing that the work order data is not a safe work order, and each keyword in the candidate keyword set comprises a weight value which is used for representing the importance degree of the keyword;

and then, processing the at least one new work order data based on the candidate keyword set to obtain the at least one data to be processed in batch.

The candidate keyword set is a pre-constructed word set, the candidate keyword set comprises a large number of keywords, and the keywords are used for representing the work order data to be the unsafe work orders. And in the candidate keyword set, each keyword corresponds to a weight value, and the weight value is used for representing the importance degree of the keyword.

In this embodiment, the determining the candidate keyword set may include:

acquiring historical work order data, wherein the historical work order data is session data between a session service provider and a target object, and the target object comprises an order service provider and a service requester. Specifically, in this embodiment, the marked work order data, i.e., the historical work order data, may be pulled from the data platform. The historical work order data comprises a safe work order and an unsafe work order, and the label of the safe work order and the unsafe work order is a second classification. For example, the label of the secure work order is "1", and the label of the non-secure work order is "0". After the historical work order data is obtained, a candidate keyword set can be constructed based on the historical work order data.

When a candidate keyword set is constructed based on historical work order data, word segmentation processing may be performed on each historical work order data to obtain a word segmentation result, where the word segmentation result includes a plurality of words. In this embodiment, a "jieba participle device" may be used to perform the word segmentation processing, so as to obtain the word segmentation result.

When performing word segmentation processing on each historical work order data, a preset word segmentation template can be determined firstly; and then, performing word segmentation processing on each historical work order data through a word segmentation device according to a preset word segmentation template to obtain word segmentation results.

The preset word segmentation template is a new word added under a certain network car booking scene, such as new words of 'service score', 'windward driving', 'customer service of XX network car booking platform' and the like, and the keyword can be acquired more accurately by setting the new word through the preset word segmentation template.

After the word segmentation result is obtained, determining a target word segmentation in the word segmentation result, wherein the target word segmentation is a word segmentation with the occurrence frequency higher than a preset threshold value in the word segmentation results of each historical work order data. In this embodiment, after the word segmentation result is obtained, the occurrence frequency of each word segmentation in the word segmentation result can be counted, and then the first N words with the highest occurrence frequency are selected as the target words. After the word segmentation result is obtained, the occurrence frequency of each word segmentation in the word segmentation result can be counted, and then the word segmentation with the occurrence frequency higher than a preset threshold value is used as a target word segmentation.

After the target participles are obtained, converting the target participles into target data in a one-hot form, wherein the target data comprise a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data; and finally, determining the candidate keyword set by using the target data.

For example, the set of vectors can be represented as: [1,0,1, …,0], wherein each value in the set of vectors indicates whether its corresponding target participle is present in the historical work order data, e.g., "1" indicates present and "0" indicates absent.

Assuming that the number of target participles is 10, taking the first historical work order data C1 as an example, it is assumed that the vector group corresponding to the first historical work order data C1 is represented as: 1,0,1,1,0,1,0,1,0, wherein a first vector value "1" in the vector group indicates that a first target participle occurs in the first historical work order data C1, a second vector value "0" in the vector group indicates that a second target participle does not occur in the first historical work order data C1, and so on, and other vector values in the vector group indicate that the first historical work order data C1 does not occur (or occurs) in its corresponding target participle, which will not be described herein one by one.

It should be noted that, in this embodiment, the number of the target segmented words is not necessarily 10, alternatively, the number of the target segmented words may be selected to be 5 ten thousand, and may be specifically set according to actual needs, which is not specifically limited in this embodiment.

In this embodiment, after the target data in the one-hot form is obtained, the candidate keyword set may be determined by using the target data in the one-hot form.

Specifically, when determining the candidate keyword set using the target data, first, a training sample and a test sample are determined based on the target data. In the present embodiment, the target data in the form of one-hot may be divided into a training sample and a test sample.

And then, training the initial classification prediction model by using the training samples to obtain a target classification prediction model. And then, processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation.

In this embodiment, the target classification prediction model may be selected as a Logistic Regression model (LR). In addition, other models may be selected, and this embodiment is not particularly limited.

And finally, determining the candidate keyword set based on the weight value of each target participle. The weighted value obtained through prediction of the target classification prediction model can be used as the importance degree of the target word segmentation.

In this embodiment, after the candidate keyword set is obtained in the manner described above, at least one new work order data may be processed based on the candidate keyword set, so as to obtain at least one data to be processed in batch.

In this embodiment, if the number of the candidate keywords is large, the keywords included in the candidate keyword set in the candidate keywords may be preferentially processed. Since the keywords contained in the candidate keyword set are more likely to be the target keywords. Therefore, by adopting the processing mode, the generation effect of the target keywords can be further improved, the accuracy of the safe work order recall is improved, the human resources are saved, and the working efficiency is improved.

In an optional implementation manner, as shown in fig. 7, the processing the at least one new worksheet data based on the candidate keyword set to obtain at least one pending data includes the following steps:

step S701, determining a first target alternative keyword in the alternative keywords, wherein the first target alternative keyword is a keyword contained in the candidate keyword set.

In this embodiment, the candidate keywords may be compared with the keywords in the candidate keyword set, so as to select the keywords included in the candidate keyword set from the candidate keywords, which are referred to as first target candidate keywords.

Step S702, the first target alternative keywords are ranked according to the weight values of the first target alternative keywords, and a first ranking result is obtained.

In this embodiment, as can be seen from the above description, all the keywords in the candidate keyword set include corresponding weight values. At this time, the weight value of the keyword in the candidate keyword set that is the same as the first target candidate keyword may be used as the weight value of the first target candidate keyword.

Then, the first target alternative keywords can be ranked according to the weight values of the first target alternative keywords, and a first ranking result is obtained. For example, ordering from high to low, or ordering from low to high.

Step S703, determining a target ranking result based on the first ranking result and the second ranking result, where the first ranking result is located before the second ranking result, the second ranking result is obtained after ranking a second target alternative keyword, and the second target alternative keyword is another keyword in the alternative keywords except the first target alternative keyword.

After the first ranking result is obtained, other candidate keywords (namely, the second target candidate keywords) except the first target candidate keyword in the candidate keywords are randomly ranked, and then a second ranking result is obtained.

Then, the first sorting result and the second sorting result may be concatenated to obtain a target sorting result. Preferably, the first sorting result may be located before the second sorting result, and besides, the first sorting result may be located after the second sorting result, which is not specifically limited in this embodiment.

Step S704, according to the sorting order in the target sorting result, performing classification processing on the new work order data corresponding to each keyword in the target sorting result to obtain the at least one data to be batched.

For example, 63 new work order data are allowed to be included in one batch data. Then the keywords are grouped in the order of the rank in the target ranking result, e.g., the 1 st through 63 rd keywords are grouped together, the 64 th through 126 th keywords are grouped together, and so on. And then, taking the new work order data corresponding to the 1 st to 63 th keywords as the sum work order data as the data to be batched. And then, taking the new work order data corresponding to the 64 th to 126 th keywords as the sum work order data as another data to be batched. By analogy, the processing mode is adopted for the new worksheet data corresponding to other keywords, and the processing mode is not introduced one by one here.

As can be seen from the above description, in this embodiment, a batch processing mode is introduced to process the work order data and the new work order data, so as to determine a mode of the target keyword, and improve the acquisition efficiency of the target keyword.

In this embodiment, after obtaining at least one piece of data to be batch processed in the manner described above, the new work order data and the work order data in each piece of data to be batch processed may be classified by using the target classification model, so as to obtain a classification result.

Optionally, in this embodiment, the selected target classification model may be an han (hierarchical assignment network) model. Before classifying new work order data and work order data in data to be batch processed by using the target classification model, training an initial model of the target classification model is needed, and a target classification model is obtained after training, wherein the training process is described as follows:

constructing a training sample according to the word segmentation result of each piece of historical work order data and the label information of each piece of historical work order data, wherein the label information is used for representing whether the historical work order data is a safe work order; and then, training an initial model of the target classification model by using the training sample to obtain the target classification model.

As can be seen from the above description, in the present embodiment, when a candidate keyword set is constructed, historical work order data is acquired from a data platform. After the word segmentation processing is performed on the historical work order data, an initial model of a target classification model can be trained by using word segmentation results and label information to obtain the target classification model.

Preferably, in this embodiment, the target classification model is selected as HAN model, which is a top-down vector-based text representation model as a dialogue classification training model, and has good performance in the classification task.

After the work order data and the at least one new work order data are classified in the manner described above to obtain the classification result, the target keyword may be determined among the candidate keywords based on the classification result.

In this embodiment, the classification processing result includes a plurality of sub-processing results, and at this time, the plurality of sub-processing results may be divided into a first sub-processing result and a second sub-processing result, where the classification processing result of the work order data is the first sub-processing result, and each new work order data corresponds to one second sub-processing result.

That is, the result of the classification processing of the work order data by the target classification model is a first sub-processing result; and the result of the classification processing of each new work order data by the target classification model is a second sub-processing result.

It should be noted that, because each piece of to-be-batch processing data includes the work order data, when the target classification model performs classification processing on the work order data in each piece of to-be-batch processing data, a first sub-processing result is obtained. And the obtained first sub-processing results may be the same or different, but the difference between the first sub-processing results is within the preset requirement range.

Based on this, in step S208, determining the target keyword among the candidate keywords based on the classification processing result includes the following steps:

step S2081, calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values;

step S2082, the target keyword is determined based on the plurality of variation values.

In this embodiment, after obtaining the classification processing result of each piece of data to be batch processed, the variation value between each second sub-processing result and the first sub-processing result in the classification processing result may be calculated to obtain a plurality of variation values.

After obtaining a plurality of variation values, a first target variation value may be determined among the plurality of variation values, where the first target variation value is the first N largest variation values among the plurality of variation values, and N is a positive integer greater than zero. And then determining new work order data corresponding to the first target change value. And determining a replaced or deleted alternative keyword (marked as a third target alternative keyword) in the new work order data corresponding to the first target change value. And finally, determining a target keyword according to the third target alternative keyword after the replacement or deletion in the corresponding new work order data.

According to the method and the device, the work order data can be screened in a machine learning mode to determine whether the work order data is a safe work order, so that the technical problems that the screening efficiency is low and the accuracy is poor when an emergency safe work order is screened in the prior art are solved.

Optionally, in this embodiment, determining the target keyword according to the third target candidate keyword after replacement or deletion in the corresponding new work order data includes:

firstly, if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords comprise keywords which are recorded in the candidate keyword set, determining the first M keywords with the maximum weight values in the fourth target alternative keywords; the candidate keyword set comprises a plurality of keywords used for representing whether the work order data is a safe work order, each keyword in the candidate keyword set comprises a weight value, the weight values are used for representing the importance degree of the keywords, and the fourth target alternative keyword is a keyword except the third target alternative keyword in the plurality of change values.

Specifically, after the third target candidate keywords are obtained, if the third target candidate keywords include keywords that are not recorded in the candidate keyword set, these keywords need to be replaced with keywords in the candidate keyword set.

At this time, in this embodiment, new work order data corresponding to a plurality of change values may be determined, and then candidate keywords after replacement or deletion in the new work order data may be determined. Then, determining that, in the candidate keywords, except for the third target candidate keyword, other keywords in the candidate keyword set are included, and then determining M keywords with the largest weight values among the keywords, where M is the number of keywords that are not recorded in the candidate keyword set and are included in the third target candidate keyword. And finally, determining the keywords with the maximum weight values in the first M determined keywords and the keywords contained in the candidate keyword set in the third target candidate keywords as target keywords.

In this embodiment, if the third target candidate keyword includes M keywords that are not recorded in the candidate keyword set, and the fourth target candidate keyword does not include a keyword that is recorded in the candidate keyword set, the third target candidate keyword after being replaced or deleted in the corresponding new work order data is determined as the target keyword.

That is, if the keywords corresponding to the plurality of variation values do not include the keyword recorded in the candidate keyword set, at this time, the third target candidate keyword may be determined as the target keyword.

Example 3:

the screening method will be described with reference to FIG. 8.

As shown in fig. 8, in this embodiment, first, an initial model of the target classification model needs to be trained, and a candidate keyword set is constructed. As shown in fig. 8, the process is described as follows:

first, the marked work order data (i.e., the historical work order data described above) is pulled from the database of the data platform, and the work orders include the safe type work order and the non-safe type work order, and the label of the work order is the second category. Then, these worksheet data are subjected to word segmentation processing. In the embodiment, a jieba word segmentation device is adopted to segment words of the work order data to obtain word segmentation results. In the process of word segmentation, new words under a certain network car booking scene can be added as word lists, such as service score, windward driving, certain customer service and the like, so that keywords can be acquired more accurately.

Thereafter, a candidate set of keywords is generated. Specifically, the occurrence frequency of each segmented word may be counted in the segmentation result, and then, 5 ten thousand words with the highest occurrence frequency are selected from the segmentation result according to the occurrence frequency, and the words substantially cover 99% of the words. After the target participles are obtained, converting the target participles into target data in a one-hot form, wherein the target data comprise a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data; and finally, determining the candidate keyword set by using the target data.

Then, a Logistic Regression (LR) model is adopted for training, and a target classification prediction model is obtained. And then, processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation, wherein the weight value can be regarded as the importance degree of the word segmentation.

Thereafter, an initial model of the target classification model is trained. When the initial model is trained, the initial model is trained by adopting the word segmentation result and the label information of each work order data. The selected target classification model may be an han (hierarchical assignment network) model.

After the candidate keyword set is constructed and the target classification model is obtained through training, the steps described in any one of the above embodiments 2 may be executed by using the target classification model, which is not described in detail herein.

It should be noted that, in this embodiment, in order to improve the evaluation speed and improve the efficiency, the present invention processes one piece of dialogue data at a time by using a batch mode, where the first row of records in the batch is complete data, the data in other rows are data after potential keywords are removed or replaced, the size of the batch is 64, 63 potential keywords can be evaluated at most at one time, if the dialogue records are too long, the first 63 words included in the candidate keyword library in the current record need to be selected as the yellow marks 1 in fig. 8, and if less than 63 stop words are removed from the dialogue, the potential keyword library does not need to be used for corresponding to the yellow marks 2.

Example 4:

fig. 9 is a block diagram illustrating a screening apparatus of some embodiments of the present application, which implements functions corresponding to the steps performed by the above-described method. The apparatus may be understood as the server or the processor of the server, or may be understood as a component that is independent of the server or the processor and implements the functions of the present application under the control of the server, as shown in the figure, the screening apparatus may include an obtaining unit 910, a keyword processing unit 920, a classifying unit 930, and a determining unit 940.

An obtaining unit 910, configured to obtain work order data in an online service process, and perform word segmentation processing on the work order data; each work order data comprises session data between a session service provider and a target object, and the target object comprises an order service provider and/or an order service requester;

a keyword processing unit 920, configured to perform target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the worksheet data;

a classifying unit 930, configured to perform classification processing on the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not;

a determining unit 940, configured to determine a target keyword among the candidate keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

Optionally, the classification unit includes: the processing module is used for processing the at least one new worksheet data to obtain at least one data to be processed in batches if the number of the alternative keywords exceeds a preset number, wherein each data to be processed comprises a plurality of new worksheets and one worksheet; and the first classification module is used for classifying the new work order data in each data to be batched and the work order data by using a target classification model to obtain the classification processing result.

Optionally, the processing module is configured to: acquiring a candidate keyword set; the candidate keyword set comprises a plurality of keywords for representing that the work order data is not a safe work order, and each keyword in the candidate keyword set comprises a weight value which is used for representing the importance degree of the keyword; and processing the at least one new work order data based on the candidate keyword set to obtain the at least one data to be batched.

Optionally, the processing module is further configured to: determining a first target alternative keyword in the alternative keywords, wherein the first target alternative keyword is a keyword contained in the candidate keyword set; sorting the first target alternative keywords according to the weight values of the first target alternative keywords to obtain a first sorting result; determining a target sorting result based on the first sorting result and a second sorting result, wherein the first sorting result is located before the second sorting result, the second sorting result is obtained after sorting second target alternative keywords, and the second target alternative keywords are other keywords except the first target alternative keywords in the alternative keywords; and according to the sorting sequence in the target sorting result, classifying the new work order data corresponding to each keyword in the target sorting result to obtain the at least one piece of data to be batched.

Optionally, the apparatus further determines the candidate keyword set by the following method, specifically including: acquiring historical work order data, wherein the historical work order data is session data between a session service provider and a target object, and the target object comprises an order service provider and a service requester; determining a set of candidate keywords based on the historical work order data.

Optionally, the apparatus is further configured to: performing word segmentation processing on each historical worksheet data to obtain word segmentation results, wherein the word segmentation results comprise a plurality of words; determining target word segmentation in the word segmentation results, wherein the target word segmentation is the word segmentation with the occurrence frequency higher than a preset threshold value in the word segmentation results of the historical work order data; converting the target participles into target data in a one-hot form, wherein the target data comprises a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data; determining the set of candidate keywords using the target data.

Optionally, the apparatus is further configured to: determining training samples and test samples based on the target data; training an initial classification prediction model by using the training samples to obtain a target classification prediction model; processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation; and determining the candidate keyword set based on the weight value of each target participle.

Optionally, the apparatus is further configured to: determining a preset word segmentation template; and performing word segmentation processing on each historical worksheet data through a word segmentation device according to the preset word segmentation template to obtain word segmentation results.

Optionally, the apparatus is further configured to: constructing a training sample according to the word segmentation result of each piece of historical work order data and the label information of each piece of historical work order data, wherein the label information is used for representing whether the historical work order data is a safe work order; and training an initial model of the target classification model by using the training sample to obtain the target classification model.

Optionally, the classification unit further includes: the first determining module is used for taking the at least one new work order data and the work order data as data to be batched if the number of the alternative keywords does not exceed the preset number; and the second classification module is used for classifying the new work order data in the data to be batched and the work order data by using a target classification model to obtain a classification processing result.

Optionally, the keyword processing unit is configured to: and replacing the alternative keywords Ai in the work order data with preset data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, I is the number of the alternative keywords, and the preset data is useless words.

Optionally, the keyword processing unit is further configured to: and deleting the alternative keywords Ai in the work order data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, and I is the number of the alternative keywords.

Optionally, the classification processing result includes a plurality of sub-processing results, and the plurality of sub-processing results include a first sub-processing result and a second sub-processing result, where the classification processing result of the work order data is the first sub-processing result, and each new work order data corresponds to one second sub-processing result.

Optionally, the determining unit includes: the calculation module is used for calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values; a second determination module to determine the target keyword based on the plurality of variance values.

Optionally, the determining module is configured to: determining a first target variance value among the plurality of variance values, wherein the first target variance value is the first N largest variance values among the plurality of variance values, and N is a positive integer greater than zero; determining new work order data corresponding to the first target change value; and determining the target keyword according to the third target alternative keyword after the replacement or deletion in the corresponding new work order data.

Optionally, the determining module is further configured to: if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords comprise keywords which are recorded in the candidate keyword set, determining M keywords with the largest weight values in the fourth target alternative keywords; the candidate keyword set comprises a plurality of keywords for representing whether the work order data is a safe work order, each keyword in the candidate keyword set comprises a weight value, the weight values are used for representing the importance degrees of the keywords, and the fourth target alternative keyword is a keyword except the third target alternative keyword in the plurality of change values; and determining the keywords with the largest weight values in the first M determined keywords and the keywords contained in the candidate keyword set in the third target candidate keywords as the target keywords.

Optionally, the determining module is further configured to: and if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords do not comprise the keywords which are recorded in the candidate keyword set, determining the third target alternative keywords which are replaced or deleted in the corresponding new work order data as the target keywords.

The modules may be connected or in communication with each other via a wired or wireless connection. The wired connection may include a metal cable, an optical cable, a hybrid cable, etc., or any combination thereof. The wireless connection may comprise a connection over a LAN, WAN, bluetooth, ZigBee, NFC, or the like, or any combination thereof. Two or more modules may be combined into a single module, and any one module may be divided into two or more units.

The invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the screening method of any of the above-described method embodiments.

It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to corresponding processes in the method embodiments, and are not described in detail in this application. In the several embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical division, and there may be other divisions in actual implementation, and for example, a plurality of modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or modules through some communication interfaces, and may be in an electrical, mechanical or other form.

The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A screening method, comprising:

acquiring work order data in an online service process, and performing word segmentation processing on the work order data; each work order data comprises session data between a session service provider and a target object, and the target object comprises an order service provider and/or an order service requester;

performing target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the worksheet data;

classifying the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not;

determining target keywords in the alternative keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

2. The method of claim 1, wherein classifying the work order data and the at least one new work order data using a target classification model comprises:

if the number of the alternative keywords exceeds the preset number, processing the at least one new worksheet data to obtain at least one data to be processed, wherein each data to be processed comprises a plurality of new worksheets and one worksheet data;

and classifying the new work order data in each data to be batched and the work order data by using a target classification model to obtain a classification processing result.

3. The method of claim 2, wherein processing the at least one new work order data to obtain at least one pending data comprises:

acquiring a candidate keyword set; the candidate keyword set comprises a plurality of keywords for representing that the work order data is not a safe work order, and each keyword in the candidate keyword set comprises a weight value which is used for representing the importance degree of the keyword;

and processing the at least one new work order data based on the candidate keyword set to obtain the at least one data to be batched.

4. The method of claim 3, wherein processing the at least one new worksheet data based on the set of candidate keywords to obtain at least one pending data comprises:

determining a first target alternative keyword in the alternative keywords, wherein the first target alternative keyword is a keyword contained in the candidate keyword set;

sorting the first target alternative keywords according to the weight values of the first target alternative keywords to obtain a first sorting result;

determining a target sorting result based on the first sorting result and a second sorting result, wherein the first sorting result is located before the second sorting result, the second sorting result is obtained after sorting second target alternative keywords, and the second target alternative keywords are other keywords except the first target alternative keywords in the alternative keywords;

and according to the sorting sequence in the target sorting result, classifying the new work order data corresponding to each keyword in the target sorting result to obtain the at least one piece of data to be batched.

5. The method according to claim 3, wherein determining the set of candidate keywords by:

acquiring historical work order data, wherein the historical work order data is session data between a session service provider and a target object, and the target object comprises an order service provider and a service requester;

determining a set of candidate keywords based on the historical work order data.

6. The method of claim 5, wherein determining a set of candidate keywords based on the historical work order data comprises:

performing word segmentation processing on each historical worksheet data to obtain word segmentation results, wherein the word segmentation results comprise a plurality of words;

determining target word segmentation in the word segmentation results, wherein the target word segmentation is the word segmentation with the occurrence frequency higher than a preset threshold value in the word segmentation results of the historical work order data;

converting the target participles into target data in a one-hot form, wherein the target data comprises a plurality of vector groups, and a vector value in each vector group represents whether the target participles appear in each historical worksheet data;

determining the set of candidate keywords using the target data.

7. The method of claim 6, wherein determining the set of candidate keywords using the target data comprises:

determining training samples and test samples based on the target data;

training an initial classification prediction model by using the training samples to obtain a target classification prediction model;

processing the test sample by using the target classification prediction model to obtain the weight value of each target word segmentation;

and determining the candidate keyword set based on the weight value of each target participle.

8. The method of claim 6, wherein performing a word segmentation process on each of the historical work order data to obtain a word segmentation result comprises:

determining a preset word segmentation template;

and performing word segmentation processing on each historical worksheet data through a word segmentation device according to the preset word segmentation template to obtain word segmentation results.

9. The method of claim 6, further comprising:

constructing a training sample according to the word segmentation result of each piece of historical work order data and the label information of each piece of historical work order data, wherein the label information is used for representing whether the historical work order data is a safe work order;

and training an initial model of the target classification model by using the training sample to obtain the target classification model.

10. The method of claim 2, wherein classifying the work order data and the at least one new work order data using a target classification model further comprises:

if the number of the alternative keywords does not exceed the preset number, taking the at least one new work order data and the work order data as data to be processed in batches;

and classifying the new work order data in the data to be batched and the work order data by using a target classification model to obtain a classification processing result.

11. The method of claim 1, wherein targeting each candidate keyword in the work order data comprises:

and replacing the alternative keywords Ai in the work order data with preset data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, I is the number of the alternative keywords, and the preset data is useless words.

12. The method of claim 1, wherein targeting each candidate keyword in the work order data comprises:

and deleting the alternative keywords Ai in the work order data to obtain new work order data corresponding to the alternative keywords Ai, wherein I is 1 to I in sequence, and I is the number of the alternative keywords.

13. The method of claim 1, wherein the classification result comprises a plurality of sub-processing results, the plurality of sub-processing results comprises a first sub-processing result and a second sub-processing result, wherein the classification result of the work order data is the first sub-processing result, and each new work order data corresponds to one second sub-processing result.

14. The method of claim 13, wherein determining a target keyword among the alternative keywords based on the classification processing result comprises:

calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values;

determining the target keyword based on the plurality of variance values.

15. The method of claim 14, wherein determining the target keyword based on the plurality of variance values comprises:

determining a first target variance value among the plurality of variance values, wherein the first target variance value is the first N largest variance values among the plurality of variance values, and N is a positive integer greater than zero;

determining new work order data corresponding to the first target change value;

and determining the target keyword according to the third target alternative keyword after the replacement or deletion in the corresponding new work order data.

16. The method of claim 15, wherein determining the target keyword according to a third target alternative keyword after replacement or deletion in the corresponding new work order data comprises:

if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords comprise keywords which are recorded in the candidate keyword set, determining M keywords with the largest weight values in the fourth target alternative keywords; the candidate keyword set comprises a plurality of keywords for representing whether the work order data is a safe work order, each keyword in the candidate keyword set comprises a weight value, the weight values are used for representing the importance degrees of the keywords, and the fourth target alternative keyword is a keyword except the third target alternative keyword in the plurality of change values;

and determining the keywords with the largest weight values in the first M determined keywords and the keywords contained in the candidate keyword set in the third target candidate keywords as the target keywords.

17. The method of claim 16, wherein using the alternative keyword after replacement or deletion in the corresponding new work order data as the target keyword further comprises:

and if the third target alternative keywords comprise M keywords which are not recorded in the candidate keyword set, and the fourth target alternative keywords do not comprise the keywords which are recorded in the candidate keyword set, determining the third target alternative keywords which are replaced or deleted in the corresponding new work order data as the target keywords.

18. A screening apparatus, comprising:

the system comprises an acquisition unit, a word segmentation unit and a word segmentation unit, wherein the acquisition unit is used for acquiring work order data in an online service process and performing word segmentation on the work order data; each work order data comprises session data between a session service provider and a target object, and the target object comprises an order service provider and/or an order service requester;

the keyword processing unit is used for carrying out target processing on each alternative keyword in the work order data to obtain at least one new work order data; the target processing is deleting or replacing alternative keywords with useless words, and the alternative keywords are words obtained after word segmentation processing is carried out on the worksheet data;

the classification unit is used for classifying the work order data and the at least one new work order data by using a target classification model to obtain a classification processing result; the classification processing result is used for representing whether the work order data and the new work order data are safe work orders or not;

a determining unit, configured to determine a target keyword among the candidate keywords based on the classification processing result; the target keywords are used for representing that the work order data is not a safe work order.

19. The apparatus of claim 18, wherein the classification unit comprises:

the processing module is used for processing the at least one new worksheet data to obtain at least one data to be processed in batches if the number of the alternative keywords exceeds a preset number, wherein each data to be processed comprises a plurality of new worksheets and one worksheet;

and the first classification module is used for classifying the new work order data in each data to be batched and the work order data by using a target classification model to obtain the classification processing result.

20. The apparatus of claim 19, wherein the processing module is configured to:

21. The apparatus of claim 20, wherein the processing module is further configured to:

22. The apparatus of claim 20, wherein the apparatus further determines the set of candidate keywords by:

23. The apparatus of claim 22, wherein the apparatus is further configured to:

determining the set of candidate keywords using the target data.

24. The apparatus of claim 23, wherein the apparatus is further configured to:

determining training samples and test samples based on the target data;

25. The apparatus of claim 23, wherein the apparatus is further configured to:

determining a preset word segmentation template;

26. The apparatus of claim 23, wherein the apparatus is further configured to:

27. The apparatus of claim 19, wherein the classification unit further comprises:

the first determining module is used for taking the at least one new work order data and the work order data as data to be batched if the number of the alternative keywords does not exceed the preset number;

and the second classification module is used for classifying the new work order data in the data to be batched and the work order data by using a target classification model to obtain a classification processing result.

28. The apparatus of claim 18, wherein the keyword processing unit is configured to:

29. The apparatus of claim 18, wherein the keyword processing unit is further configured to:

30. The apparatus of claim 18, wherein the classification result comprises a plurality of sub-processing results, the plurality of sub-processing results comprising a first sub-processing result and a second sub-processing result, wherein the classification result for the work order data is the first sub-processing result, and each new work order data corresponds to a second sub-processing result.

31. The apparatus of claim 30, wherein the determining unit comprises:

the calculation module is used for calculating a variation value between each second sub-processing result and the first sub-processing result to obtain a plurality of variation values;

a second determination module to determine the target keyword based on the plurality of variance values.

32. The apparatus of claim 31, wherein the determining module is configured to:

determining new work order data corresponding to the first target change value;

33. The apparatus of claim 32, wherein the determining module is further configured to:

34. The apparatus of claim 33, wherein the determining module is further configured to:

35. An electronic device, comprising: a processor, a storage medium and a bus, the storage medium storing machine-readable instructions executable by the processor, the processor and the storage medium communicating via the bus when the electronic device is running, the processor executing the machine-readable instructions to perform the steps of the screening method according to any one of claims 1 to 17 when executed.

36. A computer-readable storage medium, having stored thereon a computer program for performing, when being executed by a processor, the steps of the screening method according to any one of claims 1 to 17.