Training method of production problem classification model, production problem classification method and device

Abstract

The present disclosure provides a training method for a production problem classification model, which can be applied to the financial field and the field of artificial intelligence technology. The production problem classification model includes a keyword extraction layer and a classification layer. The training method for the production problem classification model includes: obtaining a training sample data set, the training sample data set includes a known problem description text; inputting the known problem description text into the keyword extraction layer of the production problem classification model to be trained, and outputting a keyword set; and using the keyword set to train the classification layer of the production problem classification model to be trained to obtain a trained production problem classification model. The present disclosure also provides a production problem classification method, a training device for a production problem classification model, a production problem classification device, equipment, storage medium and program product.

Description

Technical Field

[0001] The present disclosure relates to the fields of finance and artificial intelligence technology, and more specifically to a training method for a production problem classification model, a production problem classification method, a device, equipment, a medium and a program product. Background Art

[0002] In related technologies, when major banks and enterprises classify production problems, they usually classify the production problems in the problem description text based on manual experience.

[0003] In the process of implementing the concept of the present disclosure, the inventors discovered that the related art of using manual experience to classify production problems in problem description texts has a technical problem of insufficient accuracy. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a training method for a production problem classification model, a production problem classification method, an apparatus, a device, a medium and a program product.

[0005] According to a first aspect of the present disclosure, a method for training a production problem classification model is provided, wherein the production problem classification model includes a keyword extraction layer and a classification layer, and the method includes:

[0006] Obtain a training sample data set, wherein the training sample data set includes a known problem description text;

[0007] Input the above-mentioned known problem description text into the keyword extraction layer of the production problem classification model to be trained, and output a keyword set; and

[0008] The above keyword set is used to train the classification layer of the above production problem classification model to be trained to obtain a trained production problem classification model.

[0009] According to an embodiment of the present disclosure, the above-mentioned known problem description text is input into the keyword extraction layer of the production problem classification model to be trained, and the output keyword set includes:

[0010] The known issue description text is input into the keyword extraction layer so that the keyword extraction layer performs the following operations:

[0011] Preprocess the above-mentioned known problem description text to generate a set of candidate keywords;

[0012] Generate a candidate keyword graph based on the candidate keyword set, wherein the nodes in the candidate keyword graph are composed of keywords in the candidate keyword set, and the nodes that meet the preset conditions are connected by edges;

[0013] Using the TextRANK formula, iteratively calculate the score values ​​of the nodes in the candidate keyword graph until the score value of any of the nodes is less than a first preset threshold;

[0014] Taking the candidate keywords corresponding to the nodes whose score values ​​are greater than the second preset threshold as keywords;

[0015] The keyword set is generated according to the keywords.

[0016] According to an embodiment of the present disclosure, the preset condition includes: the candidate keywords corresponding to the nodes at both ends of the edge co-occur in a word window of a preset length.

[0017] According to an embodiment of the present disclosure, the above-mentioned edges include directed edges, and each of the above-mentioned directed edges has a corresponding weight value.

[0018] According to an embodiment of the present disclosure, the above-mentioned preprocessing of the known problem description text to generate a candidate keyword set includes:

[0019] Perform word segmentation on the above-mentioned known problem description text to obtain a preprocessing keyword set;

[0020] The pre-processed keyword set is screened to delete interference words in the pre-processed keyword set to generate the candidate keyword set.

[0021] According to an embodiment of the present disclosure, the classification layer of the production problem classification model to be trained using the keyword set includes:

[0022] The above keyword set is matched with a pre-configured standard keyword set, and a keyword matrix is ​​generated according to the matching result, so as to use the above keyword matrix to train the classification layer of the above production problem classification model to be trained.

[0023] A second aspect of the present disclosure provides a production problem classification method, comprising:

[0024] Obtaining description text of the problem to be classified; and

[0025] The above-mentioned description text of the problem to be classified is input into the production problem classification model, and the classification result is output, wherein the above-mentioned production problem classification model is trained by the training method of the production problem classification model provided by the embodiment of the present disclosure.

[0026] A third aspect of the present disclosure provides a training device for a production problem classification model, comprising:

[0027] A first acquisition module is used to acquire a training sample data set, wherein the training sample data set includes a known problem description text;

[0028] A first training module is used to input the known problem description text into the keyword extraction layer of the production problem classification model to be trained, so as to train the keyword extraction layer of the production problem classification model to be trained and output a keyword set; and

[0029] The second training module is used to train the classification layer of the production problem classification model to be trained using the keyword set to obtain a trained production problem classification model, wherein the trained production problem classification model includes a keyword extraction layer and a classification layer.

[0030] A fourth aspect of the present disclosure provides a production problem classification device, comprising:

[0031] A second acquisition module is used to acquire a description text of the problem to be classified; and

[0032] A classification module is used to input the above-mentioned description text of the problem to be classified into a production problem classification model and output a classification result, wherein the above-mentioned production problem classification model is trained by the training method of the production problem classification model provided by the embodiment of the present disclosure.

[0033] The fifth aspect of the present disclosure provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned production problem classification model training method and production problem classification method.

[0034] The sixth aspect of the present disclosure also provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned production problem classification model training method and production problem classification method.

[0035] The seventh aspect of the present disclosure also provides a computer program product, including a computer program, which, when executed by a processor, implements the above-mentioned production problem classification model training method and production problem classification method. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0037] Figure 1 A diagram schematically illustrates an application scenario of a production problem classification model training method, a production problem classification method, a production problem classification model training device, and a production problem classification device according to an embodiment of the present disclosure;

[0038] Figure 2A flowchart of a method for training a production problem classification model according to an embodiment of the present disclosure is schematically shown;

[0039] Figure 3 Schematically shows a flowchart of inputting a known problem description text into a keyword extraction layer of a production problem classification model to be trained and outputting a keyword set according to an embodiment of the present disclosure;

[0040] Figure 4 A schematic diagram of a candidate keyword graph generated by an embodiment of the present disclosure is schematically shown;

[0041] Figure 5 The schematic diagram of the word window of the embodiment of the present disclosure is schematically shown;

[0042] Figure 6 A flowchart of preprocessing a known problem description text to generate a candidate keyword set according to an embodiment of the present disclosure is schematically shown;

[0043] Figure 7 A schematic diagram of a classification layer according to an embodiment of the present disclosure is schematically shown;

[0044] Figure 8 A flowchart of a production problem classification method according to an embodiment of the present disclosure is schematically shown;

[0045] Fig. 9 The structural block diagram of the training device of the production problem classification model according to the embodiment of the present disclosure is schematically shown;

[0046] Fig.10 A structural block diagram of a production problem classification device according to an embodiment of the present disclosure is schematically shown; and

[0047] Fig.11 A block diagram of an electronic device suitable for implementing a training method for a production problem classification model and a production problem classification method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0048] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0049] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0050] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0051] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0052] At present, major banks still basically use traditional production problem solving methods to solve production problems. That is, first, the branch sends the production problems to the software development center through email and other channels, and then the corresponding internal personnel manually summarize the production problems of each branch, and then analyze each production problem and locate the type of production problem through the problem description text provided by the branch, and finally distribute the corresponding type of production problems to the corresponding person in charge for analysis and resolution. This traditional production problem classification process has the following defects:

[0053] 1) The analysis and location of production problems requires a high level of experience, especially for newcomers. Once there are changes and adjustments in personnel, it will affect the analysis and location of production problems, thus extending the time to solve production problems.

[0054] 2) The time spent on locating and analyzing different production problems varies. Based on current experience, some production problems only take about ten minutes to solve, but some special and complex production problems may take several hours or even a whole day. Such production problems take a long time to solve and there is currently no better solution.

[0055] 3) When faced with the occasional phenomenon of many production problems occurring at the same time, the workload of employees will increase exponentially, which not only brings tremendous pressure to employees, but also reduces the efficiency of handling production problems and seriously delays the closed-loop time of production problems.

[0056] Therefore, it is required to study a method to efficiently complete the analysis and location of production problems within an effective time, while reducing the workload and pressure of employees. Currently, there is no effective and reliable solution to this problem.

[0057] In order to solve the technical problems existing in the related art, the present disclosure provides a training method for a production problem classification model, the production problem classification model includes a keyword extraction layer and a classification layer, and the training method for the production problem classification model includes: obtaining a training sample data set, the training sample data set includes a known problem description text; inputting the known problem description text into the keyword extraction layer of the production problem classification model to be trained, and outputting a keyword set; and using the keyword set to train the classification layer of the production problem classification model to be trained to obtain a trained production problem classification model. The present disclosure also provides a production problem classification method, a training device for a production problem classification model, a production problem classification device, an apparatus, a storage medium, and a program product.

[0058] It should be noted that the method and device provided in the embodiments of the present disclosure can be used in the financial field, and can also be used in any field other than the financial field. The application field of the method and device provided in the embodiments of the present disclosure is not limited.

[0059] Figure 1 The application scenario diagram of the production problem classification model training method, the production problem classification method, the production problem classification model training device, and the production problem classification device according to the embodiments of the present disclosure is schematically shown.

[0060] like Figure 1 As shown, the application scenario 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0061] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only examples).

[0062] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0063] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0064] It should be noted that the training method of the production problem classification model and the production problem classification method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the training device of the production problem classification model and the production problem classification device provided in the embodiment of the present disclosure can generally be set in the server 105. The training method of the production problem classification model and the production problem classification method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the training device of the production problem classification model and the production problem classification device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0065] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0066] The following will be based on Figure 1 The scene described by Figure 2 to Figure 6 The training method of the production problem classification model of the disclosed embodiment is described in detail.

[0067] Figure 2 A flowchart of a method for training a production problem classification model according to an embodiment of the present disclosure is schematically shown.

[0068] According to an embodiment of the present disclosure, the production question classification model may include a keyword extraction layer and a classification layer.

[0069] like Figure 2 As shown, the training method of the production problem classification model of this embodiment includes operations S201 to S203.

[0070] In operation S201, a training sample data set is obtained, where the training sample data set includes a known problem description text.

[0071] According to an embodiment of the present disclosure, the problem description text may be text description information used to describe a production problem.

[0072] According to an embodiment of the present disclosure, when the description information of the production problem is not in a text format, a suitable format conversion method may be first used to convert the description information of the production problem into a text format.

[0073] According to an embodiment of the present disclosure, a known problem description text may include a problem description text in which the type of the described production problem is known. For example, the known problem description text may include a problem description text with label information, and the label information may characterize the type of the production problem described in the problem description text.

[0074] In operation S202, the known problem description text is input into the keyword extraction layer of the production problem classification model to be trained, and a keyword set is output.

[0075] According to an embodiment of the present disclosure, a keyword set may include one or more keywords, and the keywords in the keyword set may be keywords extracted from a known problem description text by a keyword extraction layer.

[0076] According to an embodiment of the present disclosure, the keyword extraction layer may include, for example, a keyword extraction layer constructed based on a TextRank algorithm.

[0077] In operation S203, the classification layer of the production problem classification model to be trained is trained using the keyword set to obtain a trained production problem classification model.

[0078] According to an embodiment of the present disclosure, the classification layer may include, for example, a classification layer constructed based on RNN (Recurrent Neural Network).

[0079] According to the embodiments of the present disclosure, the number of output neurons of the classification layer can match the number of production problem types. For example, for ATM production problems, the production problem types include five categories: native, front-end, middleware, upstream application, and manufacturer, so the number of output neurons of the classification layer can be five. It can be understood by those skilled in the art that the above examples are only used to help those skilled in the art understand the embodiments of the present disclosure, and are not any improper limitations on the embodiments of the present disclosure.

[0080] The embodiment of the present disclosure obtains a trained production problem classification model by first using a known problem description text to train the keyword extraction layer of the production problem classification model to be trained, and then using the keyword set output by the keyword extraction layer of the production problem classification model to be trained to train the classification layer of the production problem classification model to be trained. When the trained production problem classification model is used to classify the production problems described in the problem description text, the technical problem of insufficient accuracy in classifying production problems in the problem description text using manual experience in the related technology can be at least partially solved, thereby achieving the technical effect of improving the accuracy of production problem classification.

[0081] Figure 3 A flowchart schematically illustrates inputting a known problem description text into a keyword extraction layer of a production problem classification model to be trained and outputting a keyword set according to an embodiment of the present disclosure.

[0082] like Figure 3 As shown, in this embodiment, the known problem description text is input into the keyword extraction layer of the production problem classification model to be trained, and outputting the keyword set includes inputting the known problem description text into the keyword extraction layer, so that the keyword extraction layer performs the following operations S301 to S305.

[0083] In operation S301, the known problem description text is preprocessed to generate a candidate keyword set.

[0084] In operation S302, a candidate keyword graph is generated according to the candidate keyword set, wherein nodes in the candidate keyword graph are composed of keywords in the candidate keyword set, and nodes meeting preset conditions are connected by edges.

[0085] Figure 4 The schematic diagram of the candidate keyword graph generated by the embodiment of the present disclosure is shown schematically. Figure 4 In , Ks={·} can represent a candidate keyword set, 401 can represent a candidate keyword graph generated according to the candidate keyword set, 1 to 5 in 401 can be nodes in the keyword graph, respectively representing a large amount, extraction, positioning, application, and time in the candidate keyword set; in 401, node 1 and node 3 are connected by an edge, indicating that the candidate keywords represented by node 1 and node 3 respectively meet the preset conditions.

[0086] In operation S303, the score values ​​of the nodes in the candidate keyword graph are iteratively calculated using the TextRANK formula until the score value of any node is less than a first preset threshold.

[0087] According to an embodiment of the present disclosure, the edges in the candidate keyword graph include directed edges, and each directed edge has a corresponding weight value.

[0088] According to an embodiment of the present disclosure, the TextRANK formula may include the following equation (1).

[0089]

[0090] Among them, V i represents the target node in the candidate keyword graph, WS(V i ) indicates V i The rating value, d represents the damping coefficient, In(V i ) indicates pointing to V i The node set, Out(V i ) indicates that from V i The set of nodes pointed to by the edge, WS(V j ) indicates pointing to V i Node V j The rating value, W ji Indicates V i To V j The weight of V i To V k The weight of .

[0091] According to an embodiment of the present disclosure, in formula (1), d may be taken as 0.85.

[0092] According to an embodiment of the present disclosure, the first preset threshold value may be, for example, 0.0001.

[0093] According to the embodiments of the present disclosure, the values ​​of the first preset threshold and the damping coefficient can be flexibly adjusted by those skilled in the art according to actual application conditions, and the embodiments of the present disclosure do not specifically limit the values ​​of the first preset threshold and the damping coefficient.

[0094] According to an embodiment of the present disclosure, a node can be randomly selected as a target node in the candidate keyword graph each time, and the score value of the target node can be calculated using the above formula (1). When the score value of the target node is less than a first preset threshold, a node is randomly selected as the target node from nodes other than the nodes that have completed the iterative calculation, until all nodes in the candidate keyword graph are traversed.

[0095] According to an embodiment of the present disclosure, when the score value of any node in the candidate keyword graph is less than the first preset threshold, a trained keyword extraction layer can be obtained.

[0096] In operation S304, candidate keywords corresponding to nodes having a score greater than a second preset threshold are used as keywords.

[0097] According to an embodiment of the present disclosure, for example, Figure 4In the candidate keyword graph shown, after iterative calculation, the score values ​​of nodes 1, 4 and 5 are greater than the second preset threshold, and the candidate keywords corresponding to nodes 1, 4 and 5, respectively, a large amount, application and time can be used as keywords. Among them, the value of the second preset threshold can be flexibly set by those skilled in the art according to actual applications, and the embodiment of the present disclosure does not specifically limit the value of the second preset threshold.

[0098] In operation S305, a keyword set is generated according to the keywords.

[0099] According to an embodiment of the present disclosure, by iterative calculation of the candidate keyword graph, the keyword extraction layer can extract keywords that can more accurately describe the production problem from the problem description text, thereby improving the accuracy of using the classification layer to classify the production problem described in the problem description text based on keywords.

[0100] According to an embodiment of the present disclosure, in operation S302, the preset condition may include that candidate keywords corresponding to nodes at both ends of the edge respectively co-occur in a word window of a preset length.

[0101] Figure 5 The word window diagram of the embodiment of the present disclosure is schematically shown.

[0102] According to an embodiment of the present disclosure, for example, there are two candidate keywords: lighten, technology. The preset length of the word window may be 6. Figure 5 In the figure, 501 can represent a word window of preset length. Suppose in the problem description text, there is a sentence "effectively reducing the work pressure of scientific and technological personnel and improving the intelligence and efficiency of the entire process". In the above sentence, the candidate keywords "relief" and "technology" appear together in the word window 501, that is, the candidate keywords "relief" and "technology" meet the preset conditions. In the candidate keyword graph, these two candidate keywords can be connected by an edge.

[0103] According to an embodiment of the present disclosure, by connecting keywords that co-occur in a word window of a preset length with edges, the correlation between nodes in the candidate word graph can be improved, thereby achieving the technical effect of improving the training efficiency of the keyword extraction layer.

[0104] Figure 6 The flowchart of preprocessing a known problem description text and generating a candidate keyword set according to an embodiment of the present disclosure is schematically shown.

[0105] like Figure 6 As shown, in this embodiment, preprocessing the known problem description text to generate a candidate keyword set includes operations S601 to S602.

[0106] In operation S601, word segmentation is performed on the known problem description text to obtain a preprocessing keyword set.

[0107] According to the embodiments of the present disclosure, the known problem description text can be first split into sentences to obtain N sentences, for example, the text between two periods can be split into one sentence. Then each sentence can be split with a comma as a separator, that is, if there is a comma in the middle of a sentence, the sentence is split into two short sentences with the comma as a separator, and the two short sentences are used as preprocessing keywords respectively.

[0108] In operation S602, the pre-processed keyword set is screened to delete interference words in the pre-processed keyword set to generate a candidate keyword set.

[0109] According to an embodiment of the present disclosure, the preprocessed keywords in the preprocessed keyword set can be segmented and POS tagged, and stop words can be filtered out to retain only words with specified POS, such as nouns, verbs, and adjectives, to generate a candidate keyword set.

[0110] According to an embodiment of the present disclosure, stop words may include stop words in a preconfigured stop word list, for example, may include the, is, at, which, on, ah, good, well, etc., which have no actual meaning in a sentence.

[0111] According to an embodiment of the present disclosure, by preprocessing the known problem description text and screening out candidate keywords, the accuracy of the keyword extraction layer in extracting keywords can be improved, and data contamination of the keyword extraction layer by interference words in the known problem description text can be avoided.

[0112] According to an embodiment of the present disclosure, training the classification layer of the production problem classification model to be trained using a keyword set includes the following operations:

[0113] The keyword set is matched with a pre-configured standard keyword set, and a keyword matrix is ​​generated according to the matching result, so as to train the classification layer of the production problem classification model to be trained by using the keyword matrix.

[0114] According to the embodiments of the present disclosure, based on the classified production problems, several keywords can be extracted for each production problem, and then all the extracted keywords are sorted and screened, and the keywords are sorted according to the number of times the keywords appear, and multiple keywords with a number of appearances greater than a preset threshold are selected as standard keywords to form a standard keyword set.

[0115] According to an embodiment of the present disclosure, matching a keyword set with a preconfigured standard keyword set may include matching each keyword in the keyword set with the standard keyword set respectively. If a keyword in the keyword set appears in the standard keyword set, the position of this keyword in the keyword set is recorded as 1, otherwise it is recorded as 0, thereby generating a keyword matrix.

[0116] According to an embodiment of the present disclosure, for example, the standard keyword set Kg = {a, b, c, d, e}, the keyword set K = {r, t, a, e, b, c}, each keyword in the keyword set is traversed and matched with the standard keywords in the standard keyword set, for example, keyword r, after traversing and comparing keyword r with the standard keywords, the standard keyword set does not contain keyword r, so the position corresponding to r in the keyword set can be recorded as 0; for example, keyword a, after traversing and comparing keyword a with the standard keywords, the standard keyword set contains keyword a, so the position corresponding to a in the keyword set can be recorded as 1. After traversing each keyword in the keyword set, a keyword matrix Kj = {0, 0, 1, 1, 1, 1} can be generated, and then the keyword matrix Kj is used to train the classification layer of the production problem classification model to be trained.

[0117] Figure 7 A schematic diagram of a classification layer according to an embodiment of the present disclosure is schematically shown.

[0118] According to the embodiments of the present disclosure, Figure 7 As shown, the classification layer may include an input layer, a hidden layer, and an output layer. The input parameters of the input layer may include a keyword matrix K j , the activation function of the input layer can include σ(z), the output values ​​of the hidden layer and the output layer are a1 and a2 respectively. Assuming that the lLth layer has m neurons and the Lth layer has n neurons, the linear coefficients w of the Lth layer form an n×m matrix W L The bias b of the Lth layer forms an n×1 vector b L , the output a1 of the L-1th layer forms an m×1 vector a L -1, the linear output z before the activation of the Lth layer forms an n×1 vector z L , the output a2 of the Lth layer forms an n×1 vector a L . Using matrix method, the output of layer L is: a l =σ(z l )=σ(W l a l-1 +b l ).

[0119] According to an embodiment of the present disclosure, after the output neuron of the output layer generates an output value, the output neuron with the largest output value can be selected from multiple output neurons according to a greedy algorithm, and the production problem type corresponding to this neuron is used as the output result.

[0120] According to an embodiment of the present disclosure, Figure 7 The classification layer shown is only an illustrative classification layer, and the number of neurons in the input layer, hidden layer and output layer of the classification layer can be flexibly adjusted by those skilled in the art according to actual needs.

[0121] According to an embodiment of the present disclosure, when the output result of the classification layer does not match the label value of the known problem description text, the gradient descent algorithm can be used to reversely adjust the network parameters of the classification layer and the keyword extraction layer until the output result of the classification layer matches the label value of the known problem description text.

[0122] Figure 8 The flowchart of the production problem classification method according to the embodiment of the present disclosure is schematically shown.

[0123] like Figure 8 As shown, the production problem classification method of this embodiment includes operations S801 to S802.

[0124] In operation S801, a description text of a problem to be classified is obtained.

[0125] In operation S802, the description text of the problem to be classified is input into the production problem classification model, and the classification result is output, wherein the production problem classification model is trained by the training method of the production problem classification model provided by the embodiment of the present disclosure.

[0126] According to the embodiments of the present disclosure, by utilizing the trained production problem classification model to classify the problem description text to be classified, it is possible to quickly extract keywords from the problem description text and then output the type and location of the production problem. This saves the cost of manual operation with a shorter time and higher accuracy, avoids human errors, and realizes the intelligent improvement of the steps for solving production problems.

[0127] According to an embodiment of the present disclosure, the problem description text may be sent to corresponding developers based on the classification result output by the production problem classification model.

[0128] Based on the above-mentioned training method of the production problem classification model, the present disclosure also provides a training device for the production problem classification model. Fig. 9 The device is described in detail.

[0129] Fig. 9 The structural block diagram of a training device for a production problem classification model according to an embodiment of the present disclosure is schematically shown.

[0130] like Fig. 9 As shown, the training device 900 for the production problem classification model of this embodiment includes a first acquisition module 901 , a first training module 902 , and a second training module 903 .

[0131] The first acquisition module 910 is used to acquire a training sample data set, and the training sample data set includes a known problem description text. In one embodiment, the first acquisition module 901 can be used to perform the operation S201 described above, which will not be repeated here.

[0132] The first training module 902 is used to input the known problem description text into the keyword extraction layer of the production problem classification model to be trained, so as to train the keyword extraction layer of the production problem classification model to be trained and output a keyword set. In one embodiment, the first training module 902 can be used to perform the operation S202 described above, which will not be repeated here.

[0133] The second training module 903 is used to train the classification layer of the production problem classification model to be trained using the keyword set to obtain a trained production problem classification model, wherein the trained production problem classification model includes a keyword extraction layer and a classification layer. In one embodiment, the second training module 903 can be used to perform the operation S203 described above, which will not be repeated here.

[0134] According to an embodiment of the present disclosure, the first training module 820 includes a first generating unit, a second generating unit, a calculating unit, a determining unit, and a generating unit.

[0135] The first generating unit is used to pre-process the known problem description text to generate a candidate keyword set.

[0136] The second generating unit is used to generate a candidate keyword graph according to the candidate keyword set, wherein the nodes in the candidate keyword graph are composed of keywords in the candidate keyword set, and the nodes meeting the preset conditions are connected by edges.

[0137] The calculation unit is used to iteratively calculate the score values ​​of the nodes in the candidate keyword graph using the TextRANK formula until the score value of any node is less than a first preset threshold.

[0138] The determination unit is configured to use the candidate keyword corresponding to the node having a score greater than a second preset threshold as a keyword.

[0139] The generating unit is used to generate a keyword set according to the keywords.

[0140] According to an embodiment of the present disclosure, the preset condition of the edge in the candidate keyword graph generated by the second generating unit may include that the candidate keywords corresponding to the nodes at both ends of the edge co-occur in a word window of a preset length.

[0141] According to an embodiment of the present disclosure, the edges in the candidate keyword graph generated by the second generating unit include directed edges, and each directed edge has a corresponding weight value.

[0142] According to an embodiment of the present disclosure, the first generating unit includes a first processing subunit and a second processing subunit.

[0143] The first processing subunit is used to perform word segmentation processing on the known problem description text to obtain a preprocessing keyword set.

[0144] The second processing subunit is used to perform screening processing on the pre-processed keyword set to delete interference words in the pre-processed keyword set and generate a candidate keyword set.

[0145] According to an embodiment of the present disclosure, the second training module 903 includes a matching unit.

[0146] The matching unit is used to match the keyword set with the pre-configured standard keyword set and generate a keyword matrix according to the matching result so as to use the keyword matrix to train the classification layer of the production problem classification model to be trained.

[0147] Fig.10 The structural block diagram of the production problem classification device according to an embodiment of the present disclosure is schematically shown.

[0148] like Fig.10 As shown, the production problem classification device 1000 of this embodiment includes a second acquisition module 1001 and a classification module 1002 .

[0149] The second acquisition module 1001 is used to obtain the description text of the problem to be classified.

[0150] The classification module 1002 is used to input the description text of the problem to be classified into the production problem classification model and output the classification result, wherein the production problem classification model is trained by the training method of the production problem classification model provided by the embodiment of the present disclosure.

[0151] According to an embodiment of the present disclosure, any multiple modules of the first acquisition module 901, the first training module 902, the second training module 903, the second acquisition module 1001 and the classification module 1002 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the first acquisition module 901, the first training module 902, the second training module 903, the second acquisition module 1001 and the classification module 1002 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware and firmware or by a suitable combination of any of them. Alternatively, at least one of the first acquisition module 901, the first training module 902, the second training module 903, the second acquisition module 1001 and the classification module 1002 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0152] Fig.11 A block diagram of an electronic device suitable for implementing a training method for a production problem classification model and a production problem classification method according to an embodiment of the present disclosure is schematically shown.

[0153] like Fig.11 As shown, the electronic device 1100 according to an embodiment of the present disclosure includes a processor 1101, which can perform various appropriate actions and processes according to a program stored in a random access memory (ROM) 1102 or a program loaded from a storage part 1108 into a random access memory (RAM) 1103. The processor 1101 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1101 may also include an onboard memory for caching purposes. The processor 1101 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0154] In RAM 1103, various programs and data required for the operation of electronic device 1100 are stored. Processor 1101, ROM 1102 and RAM 1103 are connected to each other through bus 1104. Processor 1101 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 1102 and / or RAM 1103. It should be noted that the program can also be stored in one or more memories other than ROM 1102 and RAM 1103. Processor 1101 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0155] According to an embodiment of the present disclosure, the electronic device 1100 may further include an input / output (I / O) interface 1105, which is also connected to the bus 1104. The electronic device 1100 may further include one or more of the following components connected to the I / O interface 1105: an input portion 1106 including a keyboard, a mouse, etc.; an output portion 1107 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 1108 including a hard disk, etc.; and a communication portion 1109 including a network interface card such as a LAN card, a modem, etc. The communication portion 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the I / O interface 1105 as needed. A removable medium 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1110 as needed, so that a computer program read therefrom is installed into the storage portion 1108 as needed.

[0156] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0157] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 1102 and / or RAM 1103 described above and / or one or more memories other than ROM 1102 and RAM 1103.

[0158] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the item recommendation method provided by the embodiment of the present disclosure.

[0159] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 1101. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0160] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 1109, and / or installed from a removable medium 1111. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0161] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1109, and / or installed from the removable medium 1111. When the computer program is executed by the processor 1101, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0162] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0163] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0164] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0165] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A training method for a production problem classification model, wherein: The production problem classification model includes a keyword extraction layer and a classification layer, and the method includes: Acquire a training sample data set, wherein the training sample data set includes a known problem description text; Inputting the known problem description text into the keyword extraction layer of the production problem classification model to be trained, and outputting a keyword set; and Using the keyword set to train the classification layer of the production problem classification model to be trained, to obtain a trained production problem classification model; Wherein, the step of training the classification layer of the production problem classification model to be trained by using the keyword set includes: Matching the keyword set with a pre-configured standard keyword set, and generating a keyword matrix according to the matching result, so as to train the classification layer of the production problem classification model to be trained by using the keyword matrix; The step of matching the keyword set with a pre-configured standard keyword set and generating a keyword matrix according to the matching result includes: Each keyword in the keyword set is matched with the standard keyword set respectively. If a keyword in the keyword set appears in the standard keyword set, the position of the keyword in the keyword set is recorded as 1, otherwise it is recorded as 0, thereby generating the keyword matrix.

2. The method according to claim 1, wherein: The known problem description text is input into the keyword extraction layer of the production problem classification model to be trained, and the output keyword set includes: The known problem description text is input into the keyword extraction layer so that the keyword extraction layer performs the following operations: Preprocessing the known problem description text to generate a candidate keyword set; Generate a candidate keyword graph according to the candidate keyword set, wherein the nodes in the candidate keyword graph are composed of keywords in the candidate keyword set, and the nodes meeting the preset conditions are connected by edges; Using the TextRANK formula, iteratively calculate the score values ​​of the nodes in the candidate keyword graph until the score value of any of the nodes is less than a first preset threshold; Taking the candidate keywords corresponding to the nodes whose score values ​​are greater than the second preset threshold as keywords; The keyword set is generated according to the keywords.

3. The method according to claim 2, wherein: The preset conditions include: The candidate keywords corresponding to the nodes at both ends of the edge co-occur in a word window of a preset length.

4. The method according to claim 2 or 3, wherein: The edges include directed edges, and each of the directed edges has a corresponding weight value.

5. The method according to claim 2, wherein: The preprocessing of the known problem description text to generate a candidate keyword set includes: Performing word segmentation processing on the known problem description text to obtain a preprocessing keyword set; The pre-processed keyword set is screened to delete interference words in the pre-processed keyword set to generate the candidate keyword set.

6. A production problem classification method comprising: Get the description text of the problem to be classified; as well as The description text of the problem to be classified is input into a production problem classification model, and a classification result is output, wherein the production problem classification model is trained by the training method of the production problem classification model described in any one of claims 1 to 5.

7. A training device for a production problem classification model, comprising: A first acquisition module is used to acquire a training sample data set, where the training sample data set includes a known problem description text; A first training module is used to input the known problem description text into the keyword extraction layer of the production problem classification model to be trained, so as to train the keyword extraction layer of the production problem classification model to be trained and output a keyword set; as well as A second training module is used to train the classification layer of the production problem classification model to be trained by using the keyword set to obtain a trained production problem classification model, wherein the trained production problem classification model includes a keyword extraction layer and a classification layer; Wherein, the second training module includes: A matching unit, used for matching the keyword set with a pre-configured standard keyword set, and generating a keyword matrix according to the matching result, so as to train the classification layer of the production problem classification model to be trained by using the keyword matrix; The step of matching the keyword set with a pre-configured standard keyword set and generating a keyword matrix according to the matching result includes: Each keyword in the keyword set is matched with the standard keyword set respectively. If a keyword in the keyword set appears in the standard keyword set, the position of the keyword in the keyword set is recorded as 1, otherwise it is recorded as 0, thereby generating the keyword matrix.

8. A production problem classification device, comprising: The second acquisition module is used to obtain the description text of the problem to be classified; as well as A classification module is used to input the description text of the problem to be classified into a production problem classification model and output a classification result, wherein the production problem classification model is trained by the training method of the production problem classification model described in any one of claims 1 to 5.

9. An electronic device, comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors execute the method according to any one of claims 1 to 6.

10. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 6.

11. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images