Full-automatic work order processing information generating and pushing method based on fault knowledge base

Abstract

The invention relates to the technical field of information processing, in particular to a full-automatic work order processing information generating and pushing method based on a fault knowledge base. The method comprises the following steps: acquiring an initial decision tree; identifying a core fault work order in a plurality of fault work orders corresponding to the target fault type; pruning the initial decision-making tree based on the core fault work order to obtain a target decision-making tree, the number of nodes of the target decision-making tree being smaller than the number of nodes of the initial decision-making tree; and processing a plurality of fault work orders corresponding to the target fault type according to the target decision tree to obtain work order processing information of each fault work order. According to the technical scheme of the invention, a lighter target decision tree is obtained through decision tree pruning, and batch processing is carried out on the plurality of fault work orders corresponding to the target fault type, so that the overall time consumption of the plurality of fault work orders corresponding to the target fault type during processing is reduced, and the overall response efficiency of a large batch of fault work orders is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of information processing, and in particular to a full-automatic work order processing information generation and pushing method based on a fault knowledge base. BACKGROUND

[0002] In order to improve the efficiency of work order processing, it is particularly important to generate and push full-automatic work order processing information based on a fault knowledge base. Automatic pushing not only ensures that work order processing information can be timely and accurately delivered to relevant personnel, helping them to quickly respond and effectively solve problems. At the same time, automated processes can significantly reduce operating costs and reduce dependence on human resources, thereby saving enterprise manpower investment. In addition, automated work order processing improves processing accuracy and response speed, not only improving customer satisfaction, but also enhancing the service capabilities and market competitiveness of enterprises.

[0003] In the prior art, although the decision tree of the full-automatic work order processing information generation and pushing itself has good interpretability, when the structure of the tree becomes too complex, understanding and interpretation become more difficult. As the depth and breadth of the tree increase, the basis for judging each decision node may become more detailed and multi-level, increasing the difficulty of interpretation and the time-consuming of work order disposal. Especially in the process of processing multiple work orders in batches, it will lead to a very low overall response efficiency of a large number of fault work orders. SUMMARY

[0004] The purpose of the present application is to provide a full-automatic work order processing information generation and pushing method based on a fault knowledge base, which solves the technical problem of low overall response efficiency of a large number of fault work orders in the prior art.

[0005] In a first aspect, an embodiment of the present application provides a full-automatic work order processing information generation and pushing method based on a fault knowledge base, the method comprising: obtaining an initial decision tree, the initial decision tree being constructed based on a fault knowledge base, the end layer nodes of the initial decision tree being used for disposal of fault work orders, the non-end nodes in the initial decision tree being used to describe fault problems corresponding to fault work orders, the non-end nodes being other nodes in the decision tree except the end layer nodes; identifying core fault work orders in a plurality of fault work orders corresponding to a target fault type; pruning the initial decision tree based on the core fault work orders to obtain a target decision tree, the number of nodes of the target decision tree being less than the number of nodes of the initial decision tree; processing a plurality of fault work orders corresponding to the target fault type according to the target decision tree to obtain work order disposal information of each fault work order.

[0006] In an embodiment, the pruning the initial decision tree based on the core fault ticket to obtain a target decision tree comprises: performing association analysis on each non-terminal node in the initial decision tree and the core fault ticket to determine a retention index of each non-terminal node; pruning the initial decision tree according to a preset retention threshold and the retention index of each non-terminal node to obtain a pruned decision tree; performing similarity analysis on each non-terminal node in the pruned decision tree to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree; performing node fusion on each non-terminal node in the pruned decision tree according to the plurality of node similarity indexes of each non-terminal node in the pruned decision tree to obtain the target decision tree.

[0007] In an embodiment, the performing association analysis on each non-terminal node in the initial decision tree and the core fault ticket to determine a retention index of each non-terminal node comprises: obtaining a ticket association index and a node association index of each non-terminal node in the initial decision tree, the ticket association index indicating a correlation degree between the corresponding non-terminal node and the core fault ticket, and the node association index indicating a correlation degree between the corresponding non-terminal node and its adjacent non-terminal node; determining the retention index of each non-terminal node according to the ticket association index, the node association index and a node depth of each non-terminal node in the initial decision tree.

[0008] In an embodiment, the determining the retention index of each non-terminal node according to the ticket association index, the node association index and a node depth of each non-terminal node in the initial decision tree comprises: calculating a product of the ticket association index and the node association index of each non-terminal node in the initial decision tree to obtain a node effective index of each non-terminal node in the initial decision tree; calculating a ratio of the node effective index and the node depth of each non-terminal node in the initial decision tree to obtain the retention index of each non-terminal node.

[0009] In an embodiment, the obtaining the ticket association index of each non-terminal node in the initial decision tree comprises: analyzing a semantic similarity degree between each phrase of the core fault ticket and each non-terminal node in the initial decision tree to obtain a plurality of semantic similarity values corresponding to each non-terminal node in the initial decision tree; determining a maximum semantic similarity value as a semantic target value corresponding to each non-terminal node in the initial decision tree from the plurality of semantic similarity values corresponding to each non-terminal node in the initial decision tree; According to the information entropy of each non-terminal node in the initial decision tree and the semantic target value corresponding to the non-terminal node, a ticket association index of each non-terminal node in the initial decision tree is determined.

[0010] In one embodiment, the semantic similarity value is determined based on the number of historical fault tickets associated with the corresponding non-terminal node, the number of historical fault tickets associated with the corresponding phrase, the fault semantic similarity degree between the historical fault tickets associated with the corresponding non-terminal node and the historical fault tickets associated with the corresponding phrase, and the treatment semantic similarity degree between the historical fault tickets associated with the corresponding non-terminal node and the historical fault tickets associated with the corresponding phrase.

[0011] In one embodiment, the step of obtaining the node association index of each non-terminal node in the initial decision tree comprises: According to the position of each non-terminal node in the initial decision tree in the core fault ticket and the position of the adjacent non-terminal node of the non-terminal node in the core fault ticket, a node association index of each non-terminal node in the initial decision tree is determined.

[0012] In one embodiment, the similarity analysis on each non-terminal node in the pruned decision tree to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree comprises: obtaining a source semantic similarity index of the first non-terminal node and the second non-terminal node at each depth, wherein the first non-terminal node and the second non-terminal node are any two different non-terminal nodes in the pruned decision tree, and the source semantic similarity index represents the semantic similarity degree between the two upper nodes of the first non-terminal node and the second non-terminal node at the corresponding depth; determining a node similarity index of the first non-terminal node to the second non-terminal node according to the source semantic similarity index of the first non-terminal node and the second non-terminal node at each depth, the number of upper nodes of the first non-terminal node, the number of upper nodes of the second non-terminal node, and the semantic similarity value between the first non-terminal node and the second non-terminal node.

[0013] In one embodiment, the node similarity index is negatively correlated with the number of upper nodes of the corresponding non-terminal node, the node similarity index is positively correlated with the corresponding source semantic similarity index, and the node similarity index is positively correlated with the corresponding semantic similarity value.

[0014] In one embodiment, the step of identifying the core fault ticket in the plurality of fault tickets corresponding to the target fault type comprises: In the plurality of fault tickets corresponding to the target fault type, feature extraction is performed on each fault ticket to obtain a plurality of feature vectors; In the plurality of fault tickets corresponding to the target fault type, a vector similarity of the feature vectors between each fault ticket and other fault tickets is calculated to obtain a plurality of vector similarity values corresponding to each fault ticket. summing the plurality of vector similarity values corresponding to each fault work order to obtain a vector similarity total value corresponding to each fault work order; In the plurality of fault work orders corresponding to the target fault type, the fault work order corresponding to the highest vector similarity total value is identified as the core fault work order.

[0015] In a second aspect, another embodiment of the present application further provides an automatic work order processing information generation and push system based on a fault knowledge base, comprising: An initial decision tree acquisition module is configured to acquire an initial decision tree, wherein the initial decision tree is constructed based on a fault knowledge base, the end layer nodes of the initial decision tree are used for the disposal of fault work orders, and the non-end nodes in the initial decision tree are used to describe the fault problems corresponding to the fault work orders, wherein the non-end nodes are other nodes in the decision tree except the end layer nodes. A core identification module is configured to identify a core fault work order in the plurality of fault work orders corresponding to the target fault type. A decision tree pruning module is configured to prune the initial decision tree based on the core fault work order to obtain a target decision tree, wherein the number of nodes of the target decision tree is less than that of the initial decision tree. A work order disposal module is configured to process the plurality of fault work orders corresponding to the target fault type based on the target decision tree to obtain work order disposal information of each fault work order.

[0016] In a third aspect, another embodiment of the present application further provides an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program is executed by the processor to implement the steps of the method of the first aspect.

[0017] In a fourth aspect, another embodiment of the present application further provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the method of the first aspect.

[0018] The present application has the following advantages: The present application identifies a core fault work order in the plurality of fault work orders corresponding to the target fault type, prunes an initial decision tree based on a fault knowledge base using the core fault work order to obtain a target decision tree with fewer nodes, and processes the plurality of fault work orders corresponding to the target fault type in batches using the more lightweight target decision tree, thereby reducing the overall time consumption of the plurality of fault work orders corresponding to the target fault type in disposal and improving the overall response efficiency of a large number of fault work orders. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings required to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0020] Figure 1 is a flowchart of a full-automatic work order processing information generation and push method based on a fault knowledge base provided by an embodiment of the present application. Figure 2 is a structural diagram of a full-automatic work order processing information generation and push system based on a fault knowledge base provided by an embodiment of the present application. Figure 3 is a schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0021] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the full-automatic work order processing information generation and push method based on a fault knowledge base according to the present application, its specific implementation, structure, features and effects are described in detail as follows by combining with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0023] The specific scheme of the full-automatic work order processing information generation and push method based on a fault knowledge base provided by the present application is specifically described below in combination with the drawings.

[0024] The present application proposes a full-automatic work order processing information generation and push method based on a fault knowledge base. Please refer to Figure 1 which shows a flowchart of a full-automatic work order processing information generation and push method based on a fault knowledge base provided by one embodiment of the present application. The method comprises: Step S1, obtaining an initial decision tree.

[0025] The initial decision tree is constructed based on a fault knowledge base. The end layer node of the initial decision tree is used for disposal of a fault work order. The non-end node in the initial decision tree is used to describe a fault problem corresponding to the fault work order. The non-end node is a node other than the end layer node in the decision tree.

[0026] The above fault knowledge base can be understood as a knowledge base composed of relevant information of historical fault work orders that have been solved. The fault indication library records information such as work order generation time of the historical fault work order, fault problem description text of the historical fault work order, and fault solution measure of the historical fault work order.

[0027] In some embodiments, the obtaining process of the initial decision tree is as follows: The fault problem description text of each historical fault work order in the fault knowledge base is subjected to text cleaning respectively to remove stop words, punctuation marks, and other irrelevant noise data included in the text. Subsequently, the TF-IDF (Term Frequency-Inverse Document Frequency) method is used to vectorize the cleaned fault problem description text to obtain the TF-IDF feature vector of each fault problem description text. Then, PCA is used to reduce the dimension of the TF-IDF feature vectors of multiple fault problem description texts, and the TF-IDF feature vectors of multiple fault problem description texts after dimension reduction are input into the decision tree model together with the fault solution measures corresponding to each fault problem description text for training to obtain the initial decision tree.

[0028] In actual application, a new initial decision tree can be generated every other month (using the historical fault work orders solved in the previous month for training) to ensure the timeliness and decision accuracy of the initial decision tree.

[0029] Step S2, identifying a core fault work order from the multiple fault work orders corresponding to the target fault type.

[0030] The above target fault type can be understood as any one of all fault types involved in a business scenario. The above business scenario can be a software application scenario. All fault types involved in the business scenario can include logical fault type, computing fault type, data fault type, interface fault type, resource leakage fault type, etc.

[0031] It should be understood that the above initial decision tree is a complex decision tree that can be used to solve fault work orders of all fault types involved in a business scenario (in the case that the fault knowledge base records fault work orders and their solutions corresponding to all fault types).

[0032] The multiple fault work orders corresponding to the target fault type can be understood as a number of fault work orders corresponding to the target fault type received within a set processing period (such as ten minutes).

[0033] In some embodiments, the identifying of the core fault work order from the multiple fault work orders corresponding to the target fault type comprises: In the multiple fault work orders corresponding to the target fault type, feature extraction is performed on each fault work order to obtain a plurality of feature vectors. In the plurality of fault work orders corresponding to the target fault type, vector similarities of feature vectors between each fault work order and other fault work orders are calculated to obtain a plurality of vector similarity values corresponding to each fault work order. The plurality of vector similarity values corresponding to each fault work order are summed to obtain a vector similarity total value corresponding to each fault work order. In the plurality of fault work orders corresponding to the target fault type, the fault work order corresponding to the highest vector similarity total value is identified as the core fault work order.

[0034] Wherein, the feature extraction of the fault work order should be understood as the semantic feature (such as through the Word Embedding algorithm) extraction of the fault problem description text of the fault work order. The vector similarity values of the feature vectors between different fault work orders can be the Euclidean distance / cosine similarity / Pearson correlation coefficient, etc. between the feature vectors of different fault work orders.

[0035] It should be noted that if the total number of the plurality of fault work orders corresponding to the target fault type is set as sum, the total number of the plurality of vector similarity values corresponding to each fault work order is sum-1.

[0036] Based on the above settings, through vector extraction and vector similarity calculation and summation, the core degree of each fault work order in the plurality of fault work orders corresponding to the target fault type can be accurately evaluated (the higher the vector similarity total value corresponding to the fault work order, the higher the core degree), and then the fault work order with the highest core degree is selected as the core fault work order representing the plurality of fault work orders corresponding to the target fault type, so as to ensure that the target decision tree obtained by pruning can effectively cope with the fault processing requirements of the plurality of fault work orders corresponding to the target fault type, while reducing the number of fault work orders relied on by subsequent pruning operation, thereby reducing the resource consumption and time consumption of the pruning process.

[0037] Step S3, pruning the initial decision tree based on the core fault work order to obtain a target decision tree.

[0038] Wherein, the number of nodes of the target decision tree is less than the number of nodes of the initial decision tree.

[0039] In some embodiments, the pruning of the initial decision tree based on the core fault work order to obtain a target decision tree comprises: associating each non-terminal node in the initial decision tree with the core fault work order to determine a retention index of each non-terminal node; pruning the initial decision tree according to a preset retention threshold and the retention index of each non-terminal node to obtain a pruned decision tree; performing similarity analysis on each non-terminal node in the pruned decision tree to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree; performing node fusion on each non-terminal node in the pruned decision tree according to the plurality of node similarity indexes of each non-terminal node in the pruned decision tree to obtain the target decision tree.

[0040] In the above arrangement, in addition to pruning the initial decision tree based on the relevance between each non-terminal node in the initial decision tree and the core fault work order, the pruned decision tree (i.e., the pruned decision tree) is further subjected to node fusion processing to further reduce the number of nodes in the decision tree, thereby improving the overall efficiency of the decision tree in handling a plurality of fault work orders corresponding to the target fault type.

[0041] The retention index is used to represent the importance of the corresponding non-terminal node. The higher the retention index, the higher the importance of the corresponding non-terminal node in handling a plurality of fault work orders corresponding to the target fault type, and the more important it is to retain.

[0042] The retention index of each non-terminal node in the pruned decision tree is greater than or equal to the retention threshold (which can be set to 0.75 based on experience).

[0043] It should be noted that during the pruning process, if the retention index of a non-terminal node is less than the retention threshold, not only will the non-terminal node be pruned, but also the nodes that expand downward from the non-terminal node will be pruned.

[0044] The number of nodes in the target decision tree is less than the number of nodes in the pruned decision tree.

[0045] During the node fusion process, when the node similarity index (with a value of 0-1, the larger the value, the more similar the two non-terminal nodes) between two non-terminal nodes is greater than a set similarity threshold (such as 0.85 based on experience), one of the two non-terminal nodes can be randomly selected to be retained, and the other non-terminal node can be deleted to complete the node fusion operation of the two non-terminal nodes. In this process, the superior and inferior nodes of the deleted non-terminal node are inherited by the retained non-terminal node.

[0046] The association analysis of each non-terminal node in the initial decision tree with the core fault work order to determine the retention index of each non-terminal node includes: obtaining a work order association index and a node association index of each non-terminal node in the initial decision tree, the work order association index indicating the degree of correlation between the corresponding non-terminal node and the core fault work order, and the node association index indicating the degree of correlation between the corresponding non-terminal node and its adjacent non-terminal node; According to the ticket association index of each non-terminal node in the initial decision tree, the node association index and the node depth, a reservation index of each non-terminal node is determined.

[0047] In the above setting, the correlation between the non-terminal node and the core fault ticket is analyzed in order to evaluate the content correlation between the non-terminal node and the core fault ticket, that is, to directly determine from the content dimension whether the corresponding non-terminal node can be used for the fault handling of the several fault tickets represented by the core fault ticket. The correlation between the non-terminal node and its adjacent non-terminal node is analyzed in order to evaluate the logical coherence of the region where the non-terminal node is located, that is, by judging the correlation between the corresponding non-terminal node and its adjacent non-terminal node (i.e. the last non-terminal node in the decision tree), it is inferred whether the corresponding non-terminal node and its adjacent non-terminal node are continuous in logic, and then it is determined whether the corresponding non-terminal node can be used for the fault handling of the several fault tickets represented by the core fault ticket. As for the analysis of the node depth of the non-terminal node, it is to adapt to the situation that the deeper the node depth, the more specific and localized the corresponding fault information content is, so as to evaluate whether the non-terminal node can be effectively used for the fault handling of the several fault tickets represented by the core fault ticket from the specific degree of the fault information content corresponding to the non-terminal node.

[0048] It should be noted that since the calculation of the reservation index involves each non-terminal node in the initial decision tree, and the subsequent pruning process will prune all non-terminal nodes that do not have reservation value, therefore, the judgment of the logical coherence of the region where the non-terminal node is located only needs to analyze the non-terminal node and its adjacent non-terminal node, and does not need to perform correlation analysis on all non-terminal nodes on the logical chain where the non-terminal node is located. This can guarantee the accuracy of pruning while avoiding the problem of repeated calculation, reducing the resource consumption and time consumption when judging the logical coherence of the non-terminal node.

[0049] Exemplarily, the reservation index of the A th non-terminal node in the initial decision tree can be represented as: wherein, represents the ticket association index of the A th non-terminal node, represents the node association index of the A th non-terminal node, represents the node depth of the A th non-terminal node, represents a normalization function.

[0050] In some embodiments, the step of obtaining the ticket association index of each non-terminal node in the initial decision tree comprises: analyzing the semantic similarity degree between each phrase of the core fault ticket and each non-terminal node in the initial decision tree to obtain a plurality of semantic similarity values corresponding to each non-terminal node in the initial decision tree; ​In the initial decision tree, among the multiple semantic similarity values ​​corresponding to each non-terminal node, the largest semantic similarity value is determined as the semantic target value corresponding to each non-terminal node in the initial decision tree; Based on the information entropy of each non-terminal node in the initial decision tree and its corresponding semantic target value, the work order association index of each non-terminal node in the initial decision tree is determined.

[0051] In the above setup, semantic similarity is evaluated between non-terminal nodes and each phrase in the core fault work order. The work order association index of the non-terminal node is determined based on the maximum semantic similarity value and the information entropy of the non-terminal node, thus determining the degree of matching between the non-terminal node and the core fault work order at the semantic content level. A higher information entropy indicates a higher probability that the corresponding non-terminal node contains important information, and a stronger ability to support decision-making in the core fault work order, resulting in a higher degree of matching between the non-terminal node and the core fault work order at the semantic content level. Correspondingly, a higher maximum semantic similarity value indicates a higher probability that the non-terminal node matches a certain phrase in the core fault work order, and a higher degree of matching between it and the core fault work order at the semantic content level.

[0052] The information entropy of the aforementioned non-terminal node can be understood as the normalized value of the frequency of occurrence of the non-terminal node in the fault knowledge base.

[0053] For example, the retention index of the A-th non-terminal node in the initial decision tree It can be represented as: in, This represents the information entropy of the A-th non-terminal node. This indicates the A-th non-terminal node and the core fault work order. The semantic similarity value between phrases, where N is the total number of phrases included in the core fault work order, and max represents the maximum value function.

[0054] Furthermore, the semantic similarity value is determined based on the number of historical fault work orders associated with the corresponding non-terminal node, the number of historical fault work orders associated with the corresponding phrase, the degree of fault semantic similarity between the historical fault work orders associated with the corresponding non-terminal node and the historical fault work orders associated with the corresponding phrase, and the degree of handling semantic similarity between the historical fault work orders associated with the corresponding non-terminal node and the historical fault work orders associated with the corresponding phrase.

[0055] The association between historical fault work orders and non-terminal nodes should be understood as follows: the fault problem description text of the historical fault work order includes the node content of the corresponding non-terminal node. Similarly, the association between historical fault work orders and phrases should be understood as follows: the fault problem description text of the historical fault work order includes the corresponding phrases.

[0056] The fault semantic similarity between the historical fault work order associated with the non-terminal node and the historical fault work order associated with the word group can be quantified as: the semantic similarity value between the fault problem description texts of one historical fault work order associated with the non-terminal node and one historical fault work order associated with the word group (e.g., after text vectorization processing by a sentence encoder such as a Sentence-BERT model, the cosine similarity between the obtained text vectors is calculated as the semantic similarity value).

[0057] The treatment semantic similarity between the historical fault work order associated with the non-terminal node and the historical fault work order associated with the word group can be quantified as a binary value (1 or 0), if the fault resolution measures between one historical fault work order associated with the non-terminal node and one historical fault work order associated with the word group are the same, the value is 1, if they are different, the value is 0.

[0058] Exemplarily, the semantic similarity value between the A-th non-terminal node in the initial decision tree and the Q-th word group in the core fault work order can be represented as: Wherein, exp represents the exponential function with the natural constant e as the base, represents the fault semantic similarity between the Q-th historical fault work order associated with the A-th non-terminal node and the R-th historical fault work order associated with the Q-th word group, represents the treatment semantic similarity between the Q-th historical fault work order associated with the A-th non-terminal node and the R-th historical fault work order associated with the Q-th word group, represents the number of historical fault work orders associated with the A-th non-terminal node, represents the number of historical fault work orders associated with the Q-th word group.

[0059] In some embodiments, the step of obtaining the node association index of each non-terminal node in the initial decision tree comprises: According to the position of each non-terminal node in the initial decision tree in the core fault work order and the position of its adjacent non-terminal node in the core fault work order, the node association index of each non-terminal node in the initial decision tree is determined.

[0060] In the above setting, by analyzing the position difference between each non-terminal node and its adjacent non-terminal node in the core fault work order, the logical continuity between each non-terminal node and its adjacent non-terminal node is determined by using the content distribution characteristics of logically continuous different content appearing adjacent in the fault problem description text of the core fault work order, which can make the determined node association index more accurate and reliable. ​​​​​

[0061] It is to be noted that the non-terminal node adjacent to the non-terminal node is understood as the last non-terminal node of the non-terminal node in the decision tree (both are directly connected).

[0062] Exemplarily, the node association index of the A th non-terminal node in the initial decision tree may be expressed as: wherein exp represents the exponential function with the natural constant e as the base, represents the position (i.e. the occurrence order) of the A th non-terminal node in the fault problem description text of the core fault work order, represents the position of the adjacent non-terminal node of the A th non-terminal node in the fault problem description text of the core fault work order.

[0063] It is to be noted that when the non-terminal node does not appear in the fault problem description text of the core fault work order, the position of the non-terminal node in the fault problem description text of the core fault work order is considered to be infinite.

[0064] In some embodiments, the similarity analysis of each non-terminal node in the pruned decision tree is performed to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree, comprising: obtaining a source semantic similarity index of a first non-terminal node and a second non-terminal node at each depth, wherein the first non-terminal node and the second non-terminal node are any two different non-terminal nodes in the pruned decision tree, and the source semantic similarity index is used to represent the semantic similarity degree between the two upper nodes of the first non-terminal node and the second non-terminal node at the corresponding depth; determining a node similarity index of the first non-terminal node to the second non-terminal node according to the source semantic similarity index of the first non-terminal node and the second non-terminal node at each depth, the number of upper nodes of the first non-terminal node, the number of upper nodes of the second non-terminal node, and the semantic similarity value between the first non-terminal node and the second non-terminal node.

[0065] wherein the node similarity index is negatively correlated with the number of upper nodes of the corresponding non-terminal node, the node similarity index is positively correlated with the corresponding source semantic similarity index, and the node similarity index is positively correlated with the corresponding semantic similarity value.

[0066] The source semantic similarity index can be understood as a semantic similarity value between the upper node of the first non-terminal node at a certain depth and the upper node of the second non-terminal node at a certain depth. For example, the cosine similarity between the two vectors obtained by respectively performing vectorization processing on the upper node of the first non-terminal node at the kp depth and the upper node of the second non-terminal node at the kp depth through the Sentence-BERT model can be used as the source semantic similarity index of the first non-terminal node and the second non-terminal node at the kp depth.

[0067] It should be noted that the depth corresponding to the source semantic similarity index is not only the upper depth of the depth of the first non-terminal node, but also the upper depth of the depth of the second non-terminal node.

[0068] The semantic similarity value between the first non-terminal node and the second non-terminal node can refer to the acquisition process of the plurality of semantic similarity values corresponding to each non-terminal node in the aforementioned initial decision tree. To avoid repetition, details are not repeated.

[0069] Exemplarily, if the first non-terminal node and the second non-terminal node are the Bth non-terminal node and the Cth non-terminal node in the pruned decision tree, respectively, the node similarity index of the first non-terminal node to the second non-terminal node can be represented as: wherein, represents a minimum value function, represents the number of upper nodes of the Bth non-terminal node in the pruned decision tree, represents the number of upper nodes of the Cth non-terminal node in the pruned decision tree, represents the semantic similarity between the Bth non-terminal node and the Cth non-terminal node in the pruned decision tree, represents the semantic similarity between the upper node (indicated by ) of the Bth non-terminal node at the jth depth and the upper node (indicated by ) of the Cth non-terminal node at the jth depth in the pruned decision tree.

[0070] Step S4: processing a plurality of fault work orders corresponding to the target fault type according to the target decision tree to obtain work order handling information of each fault work order.

[0071] ​Specifically, the text features of the processed fault work order (i.e., text cleaning, vector conversion using the TF-IDF algorithm, and vector dimension reduction using the PCA algorithm) will be used as the input of the target decision tree, and the result output by the target decision tree after reasoning will be used as the work order handling information of the corresponding fault work order, which can include the recommended processing scheme of the fault work order, the contact information of the relevant technical support personnel, and the estimated processing time, etc.

[0072] Subsequently, the work order handling information can be timely copied to the relevant responsible person or technical team through email, SMS or workflow notification, etc., to ensure that the corresponding fault work order can be processed in the shortest time.

[0073] In summary, the application identifies the core fault work order in the multiple fault work orders corresponding to the target fault type, prunes the initial decision tree based on the fault knowledge base using the core fault work order to obtain a target decision tree with fewer nodes, and uses the more lightweight target decision tree to process the multiple fault work orders corresponding to the target fault type in batches, thereby reducing the overall time consumption of the multiple fault work orders corresponding to the target fault type when handling, and improving the overall response efficiency of a large number of fault work orders.

[0074] The application provides a full-automatic work order processing information generation and pushing system based on a fault knowledge base, which is used for processing the multiple fault work orders corresponding to the target fault type. Figure 2 The system includes: A decision tree acquisition module 201 is configured to acquire an initial decision tree, wherein the initial decision tree is constructed based on a fault knowledge base, the end layer node of the initial decision tree is used for handling the fault work order, the non-end node in the initial decision tree is used for describing the fault problem corresponding to the fault work order, and the non-end node is a node other than the end layer node in the decision tree. A core identification module 202 is configured to identify a core fault work order in the multiple fault work orders corresponding to the target fault type. A decision tree pruning module 203 is configured to prune the initial decision tree based on the core fault work order to obtain a target decision tree, wherein the number of nodes of the target decision tree is less than that of the initial decision tree. A work order handling module 204 is configured to handle the multiple fault work orders corresponding to the target fault type according to the target decision tree to obtain the work order handling information of each fault work order.

[0075] It should be noted that the system provided in the above embodiment is only used as an example for the division of the above functional modules, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the automatic work order processing information generation and pushing system based on the fault knowledge base and the automatic work order processing information generation and pushing method based on the fault knowledge base provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0076] The embodiment of the present application also provides an electronic device. Referring to Figure 3 , the electronic device can include a processor 301, a memory 302, and a program 3021 stored in the memory 302 and executable on the processor 301.

[0077] The program 3021 is executed by the processor 301 to implement Figure 1 any step in the corresponding method embodiment and achieve the same beneficial effects, which will not be repeated here.

[0078] Those skilled in the art can understand that all or part of the steps of the above-mentioned embodiment method can be completed by program instructions related to hardware, and the program can be stored in a readable medium.

[0079] The embodiment of the present application also provides a readable storage medium, and the readable storage medium stores a computer program, and the computer program is executed by a processor to implement Figure 1 any step in the corresponding method embodiment, and can achieve the same technical effects, to avoid repetition, which will not be repeated here.

[0080] The computer readable storage medium of the embodiment of the present application can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium may, for example, be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples (non-exhaustive list) of the computer readable storage medium include: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus.

[0081] Computer readable signal media can include a propagated data signal with computer readable program code embodied therein. For example, a propagated signal can be an electromagnetic signal, an optical signal, and / or any suitable combination thereof. Computer readable program code embodied on a computer readable medium can direct a computer to function in a particular manner, such as implementing an embodiment of the present application.

[0082] Program code embodied on a computer readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire line, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0083] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). These network connections are

[0084] The embodiment of the present application also provides a computer program product, when the computer program product runs on a computer, causes the computer to execute the above related steps, so as to realize the automatic work order processing information generation and pushing method based on the fault knowledge base provided by the above embodiment.

[0085] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0086] Each of the embodiments in the specification is described in a progressive manner, and the same and similar parts between the embodiments can be referred to each other. Each embodiment focuses on the difference from other embodiments.

Claims

1. A method for generating and pushing information based on a full-automatic work order processing knowledge base, characterized in that, The method comprises: obtaining an initial decision tree, the initial decision tree being constructed based on a fault knowledge base, a terminal node of the initial decision tree being used for handling a fault work order, and a non-terminal node in the initial decision tree being used for describing a fault problem corresponding to the fault work order, the non-terminal node being a node other than the terminal node in the decision tree; identifying a core fault work order from a plurality of fault work orders corresponding to a target fault type; pruning the initial decision tree based on the core fault work order to obtain a target decision tree, the target decision tree having a smaller number of nodes than the initial decision tree; processing the plurality of fault work orders corresponding to the target fault type according to the target decision tree to obtain work order handling information of each of the fault work orders.

2. The fully automated trouble knowledge base based incident processing information generation push method according to claim 1, characterized in that, The pruning of the initial decision tree based on the core fault work order to obtain the target decision tree comprises: performing correlation analysis on each non-terminal node in the initial decision tree and the core fault work order to determine a retention index of each non-terminal node; pruning the initial decision tree based on a preset retention threshold and the retention index of each non-terminal node to obtain a pruned decision tree; performing similarity analysis on each non-terminal node in the pruned decision tree to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree; performing node fusion on each non-terminal node in the pruned decision tree based on the plurality of node similarity indexes of each non-terminal node in the pruned decision tree to obtain the target decision tree. 3.The method of claim 2, wherein, The correlation analysis on each non-terminal node in the initial decision tree and the core fault work order to determine a retention index of each non-terminal node comprises: obtaining a work order correlation index and a node correlation index of each non-terminal node in the initial decision tree, the work order correlation index indicating a correlation degree between the corresponding non-terminal node and the core fault work order, and the node correlation index indicating a correlation degree between the corresponding non-terminal node and an adjacent non-terminal node thereof; determining the retention index of each non-terminal node based on the work order correlation index, the node correlation index, and a node depth of each non-terminal node in the initial decision tree.

4. The fully automated trouble ticket processing information generation push method based on a failure knowledge base according to claim 3, characterized by, The determination of the retention index of each non-terminal node based on the work order correlation index, the node correlation index, and the node depth of each non-terminal node in the initial decision tree comprises: calculating a product of the work order correlation index and the node correlation index of each non-terminal node in the initial decision tree to obtain a node effective index of each non-terminal node in the initial decision tree; calculating a ratio of the node effective index and the node depth of each non-terminal node in the initial decision tree to obtain the retention index of each non-terminal node. 5.The method of claim 3, wherein, The obtaining of the work order correlation index of each non-terminal node in the initial decision tree comprises: analyzing a semantic similarity degree between each phrase of the core fault work order and each non-terminal node in the initial decision tree to obtain a plurality of semantic similarity values corresponding to each non-terminal node in the initial decision tree; determining a maximum semantic similarity value as a semantic target value corresponding to each non-terminal node in the initial decision tree from the plurality of semantic similarity values corresponding to each non-terminal node in the initial decision tree; According to the information entropy of each non-terminal node in the initial decision tree and the semantic target value corresponding to the non-terminal node, a ticket association index of each non-terminal node in the initial decision tree is determined.

6. The fully automated trouble ticket processing information generation push method based on a failure knowledge base according to claim 5, characterized by, The semantic similarity value is determined based on the number of historical fault tickets associated with the corresponding non-terminal node, the number of historical fault tickets associated with the corresponding word group, the fault semantic similarity degree between the historical fault tickets associated with the corresponding non-terminal node and the historical fault tickets associated with the corresponding word group, and the treatment semantic similarity degree between the historical fault tickets associated with the corresponding non-terminal node and the historical fault tickets associated with the corresponding word group.

7. The fully automated trouble ticket processing information generation push method based on a failure knowledge base according to claim 3, characterized by, The step of obtaining the node association index of each non-terminal node in the initial decision tree comprises: According to the position of each non-terminal node in the initial decision tree in the core fault ticket and the position of the adjacent non-terminal node of the non-terminal node in the core fault ticket, a node association index of each non-terminal node in the initial decision tree is determined.

8. The fully automated trouble ticket processing information generation push method based on a failure knowledge base according to claim 2, characterized by, The similarity analysis of each non-terminal node in the pruned decision tree is performed to determine a plurality of node similarity indexes of each non-terminal node in the pruned decision tree, comprising: obtaining a source semantic similarity index of a first non-terminal node and a second non-terminal node at each depth, wherein the first non-terminal node and the second non-terminal node are any two different non-terminal nodes in the pruned decision tree, and the source semantic similarity index is used to represent the semantic similarity degree between the first non-terminal node and the second non-terminal node at the corresponding depth between two upper nodes; determining a node similarity index of the first non-terminal node to the second non-terminal node according to the source semantic similarity index of the first non-terminal node and the second non-terminal node at each depth, the number of upper nodes of the first non-terminal node, the number of upper nodes of the second non-terminal node, and the semantic similarity value between the first non-terminal node and the second non-terminal node. 9.The method of claim 8, wherein, The node similarity index is negatively correlated with the number of upper nodes of the corresponding non-terminal node, the node similarity index is positively correlated with the corresponding source semantic similarity index, and the node similarity index is positively correlated with the corresponding semantic similarity value. 10.The method of claim 1, wherein, The core fault ticket is identified in the plurality of fault tickets corresponding to the target fault type, comprising: In the plurality of fault tickets corresponding to the target fault type, feature extraction is performed on each fault ticket to obtain a plurality of feature vectors. In the plurality of fault tickets corresponding to the target fault type, the vector similarity of the feature vectors between each fault ticket and other fault tickets is calculated to obtain a plurality of vector similarity values corresponding to each fault ticket. The plurality of vector similarity values corresponding to each fault ticket are summed to obtain a vector similarity total value corresponding to each fault ticket. In the plurality of fault tickets corresponding to the target fault type, the fault ticket corresponding to the highest vector similarity total value is identified as the core fault ticket.

Images