Work order processing method and apparatus

Abstract

The application provides a work order processing method and device, the method comprising: obtaining work order data; extracting fault information from the work order data to obtain the fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; and calling the corresponding processing strategies and executing according to the matched work order instances. The application extracts fault information from the obtained work order data to clearly define the core content of the fault, remove irrelevant information interference, improve the accuracy and efficiency of subsequent processing, matches the extracted fault information with work order instances in the fault feature database to quickly locate the verified processing strategy corresponding to the matched work order instances, improves the accuracy of work order processing, reduces the possibility of incorrect processing, reduces the time and cost of starting from scratch to analyze and solve problems, and improves the speed and success rate of fault processing.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a work order processing method and apparatus. Background Technology

[0002] As business expands, the number of work orders generated in customer service has increased dramatically. Traditional manual work order processing is not only time-consuming but also prone to errors, failing to meet the demands for efficient and accurate service. Many of these work orders involve repetitive or similar issues. Traditional work order processing typically relies on manual operation, resulting in slow processing speeds, high error rates, and heavy workloads, making it difficult to meet the company's growing work order processing needs. Therefore, developing an intelligent work order system that can automatically identify and handle common problems is particularly important.

[0003] While there have been some initial attempts at automation in existing technologies, most rely on preset rules and have limited ability to handle complex and ever-changing actual work order content. They lack self-optimization and learning capabilities, and are slow, error-prone, and have a large workload. They cannot automatically analyze and process customer-submitted work orders, and the same problems require repeated manual processing, which is difficult to meet the growing work order processing needs of enterprises. Summary of the Invention

[0004] This invention provides a work order processing method and apparatus to address the limitations of existing technologies in processing complex and ever-changing actual work order content, thereby improving work order processing efficiency, reducing manual intervention, and enhancing customer service quality.

[0005] This invention provides a work order processing method, comprising: acquiring work order data; extracting fault information from the work order data to obtain fault information; matching the fault information with work order instances in a pre-established fault feature database to obtain matching work order instances; wherein, the fault feature database is established based on work order instances and corresponding processing strategies; and invoking and executing the corresponding processing strategy according to the matching work order instance.

[0006] According to a work order processing method provided by the present invention, fault information is extracted from work order data to obtain fault information, including: extracting features from the work order data as image data to obtain feature vectors; extracting fault information from the feature vectors using natural language processing technology to obtain fault information; extracting fault information from the work order data as text data using natural language processing technology to obtain fault information; and converting the fault information into vectors using word embedding technology.

[0007] According to a work order processing method provided by the present invention, natural language processing technology is used to extract fault information from work order data in feature vector or text format to obtain fault information. The method includes: segmenting the feature vector or text format into words to obtain multiple words; performing part-of-speech tagging on each word to obtain part-of-speech tags; performing entity recognition on the word segments based on the part-of-speech tags to obtain entity recognition results; and extracting fault information from each word segment based on the entity recognition results to obtain fault information.

[0008] According to a work order processing method provided by the present invention, after extracting fault information from work order data to obtain fault information, the method includes: using a pre-established sentiment dictionary to identify the context of each word segment and determine the contextual sentiment tendency of each word segment; wherein, the contextual sentiment tendency includes the positive sentiment tendency corresponding to positive words and the negative sentiment tendency corresponding to negative words, the sentiment dictionary is constructed based on pre-collected positive and negative words, positive words are used to characterize the urgency and importance of the fault, and negative words are used to characterize the fault and the problems existing in the fault; according to the contextual sentiment tendency of each word, combined with a preset weight allocation rule, a sentiment score is determined; wherein, the preset weight allocation rule is used to limit the allocation weight corresponding to the contextual sentiment tendency; according to the sentiment score, the corresponding work order data is prioritized.

[0009] Matching fault information with work order instances in the previously established fault feature database includes: selecting the work order data with the highest priority based on the priority sorting result; and matching the fault information corresponding to the selected work order data with the work order instances in the previously established fault feature database.

[0010] According to a work order processing method provided by the present invention, fault information is matched with work order instances in a pre-established fault feature database to obtain a matched work order instance, including: using Jaccard similarity to determine the similarity between fault information and work order instances in the pre-established fault feature database; determining whether the similarity exceeds a preset threshold, and based on the similarity exceeding the preset threshold, using the corresponding work order instance as a matched work order instance.

[0011] According to a work order processing method provided by the present invention, the preset threshold is obtained by optimizing the corresponding preset threshold based on feedback information obtained from the processing strategy corresponding to the previously acquired work order data; the preset threshold is the preset threshold used to determine the matching work order instance that matches the previously acquired work order data.

[0012] According to a work order processing method provided by the present invention, after calling and executing the corresponding processing strategy based on a matched work order instance, the method includes: obtaining feedback information based on the execution of the corresponding processing strategy; and optimizing the corresponding processing strategy in the fault feature database based on the feedback information.

[0013] The application further provides a work order processing device, comprising: a data acquisition module, which acquires work order data; an information extraction module, which extracts fault information from the work order data to obtain the fault information; a work order matching module, which matches the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; and a work order processing module, which calls the corresponding processing strategies and executes them according to the matched work order instances.

[0014] The application further provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the work order processing method according to any of the above when executing the computer program.

[0015] The application further provides a non-transitory computer-readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement the work order processing method according to any of the above.

[0016] The application further provides a computer program product, comprising a computer program, wherein the computer program is executed by a processor to implement the work order processing method according to any of the above.

[0017] The application provides a work order processing method and device, which extracts fault information from acquired work order data to determine the core content of the fault, removes irrelevant information interference, improves the accuracy and efficiency of subsequent processing, matches the extracted fault information with work order instances in a fault feature database to quickly locate the verified processing strategy corresponding to the matched work order instance, improves the accuracy of work order processing, reduces the possibility of error processing, reduces the time and cost of starting from scratch to analyze and solve problems, and improves the speed and success rate of fault processing. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 is one of the flowcharts of the work order processing method provided by the present application;

[0020] Figure 2 is another flowchart of the work order processing method provided by the present application;

[0021] Figure 3 is a structural schematic diagram of the work order processing device provided by the present application;

[0022] Figure 4 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0023] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0024] Figure 1 A flowchart of a work order processing method provided by the present application is shown, and the method comprises the following steps:

[0025] S11, obtaining work order data;

[0026] S12, extracting fault information from the work order data to obtain the fault information;

[0027] S13, matching the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies;

[0028] S14, calling the corresponding processing strategy and executing according to the matched work order instances.

[0029] It should be noted that the step number "S1N" in the present specification does not represent the sequence of the work order processing method, and the following will be specifically combined Figure 2 to describe the work order processing method of the present application.

[0030] Step S11, obtaining work order data.

[0031] It should be noted that the work order data can be data submitted by a user based on a preset platform, including customer questions, processing processes, solution results and other information, which is determined based on the actual received work order data submitted by the user. In addition, the preset platform can be a designated platform or other user platforms, which is not limited further here.

[0032] In an optional embodiment, after obtaining the work order data, the method comprises: pre-processing the work order data, wherein the pre-processing comprises at least one of data cleaning and text conversion. It should be noted that by data cleaning, special characters, punctuation marks and other elements in the work order description that may interfere with subsequent processing can be removed, the interference of noise is reduced, and subsequent text analysis and understanding is more accurate and efficient; by text conversion, standardized text representation is used to reduce format interference and facilitate quick extraction and understanding of the main content of the work order.

[0033] In step S12, the fault information is extracted from the work order data.

[0034] In one possible implementation, the work order data is in a picture format, and the fault information is extracted from the work order data, including: extracting features from the work order data to obtain a feature vector; and using a natural language processing (NLP) technique to extract the fault information from the feature vector to obtain the fault information.

[0035] Further, the feature extraction from the work order data includes: inputting the work order data into a feature extraction model to obtain a feature vector output by the feature extraction model; and wherein the feature extraction model is trained according to historical work order data and feature labels corresponding to the historical work order data.

[0036] It should be noted that the feature extraction model can use a convolutional neural network (CNN) model or other models that can be used to extract feature vectors, which are not limited here.

[0037] In another possible implementation, the work order data is in a text format, and the fault information is extracted from the work order data, further including: using an NLP technique to extract the fault information from the work order data; and using a word embedding technique to convert the fault information into a vector.

[0038] It should be noted that the word embedding technique is used to convert the fault information into a vector, including: inputting the fault information into a word embedding model to obtain a vector form of the fault information output by the word embedding model; and wherein the word embedding model is trained according to historical fault information extracted based on historical work order data and feature labels corresponding to the historical fault information.

[0039] It should be noted that the fault information used for training is extracted from the historical work order data using the NLP technique, which facilitates the word embedding model to learn the context relationship of the words. By using the historical fault information and the feature labels corresponding to the historical fault information, and combining the previously configured model parameters, the word embedding model is trained to continuously adjust the vector representation of the fault information, minimize the loss function, and thus better capture the semantic association and context information between the words, providing more valuable vector representation for subsequent text analysis and understanding.

[0040] In an optional embodiment, natural language processing technology is used to extract fault information from the feature vector or text format work order data to obtain the fault information, including: performing word segmentation on the feature vector or text format work order data to obtain a plurality of segmented words; performing part-of-speech tagging on each segmented word to obtain part-of-speech tags; performing entity recognition on the segmented words based on the part-of-speech tags to obtain entity recognition results; and performing fault information extraction on each segmented word based on the entity recognition results to obtain the fault information.

[0041] Further, the word segmentation can use a preset word segmentation tool, such as Jieba word segmentation, without further limitation. For example, the obtained work order data is “server hard disk fault light turns on, please contact me during working hours if on-site handling is required”, after data cleaning, word segmentation is used to obtain segmented words including “server”, “hard disk”, “fault”, “light”, “turns on”, “request”, “as soon as possible”, “handle”, “if required”, “on-site”, “handle”, “please”, “in”, “working hours”, “and”, “me”, and “contact”.

[0042] Correspondingly, the part-of-speech of each segmented word is tagged to determine the part-of-speech of each segmented word, such as noun, verb, adjective, etc., to help better understand the role of the word in the sentence, provide important clues for subsequent entity recognition, and improve the accuracy and efficiency of entity recognition.

[0043] In addition, based on the part-of-speech tags, entity recognition is performed on the segmented words to obtain entity recognition results, including: inputting the segmented words and the part-of-speech tags corresponding to each segmented word into an entity recognition model to obtain the entity recognition results output by the entity recognition model; wherein the entity recognition model is trained according to segmented training data and entity tags corresponding to the segmented training data, and the segmented training data includes training segmented words and part-of-speech corresponding to the training segmented words.

[0044] It should be noted that the entity recognition model can use a deep learning model such as BiLSTM-CRF, which can be selected and trained according to actual design requirements. A large amount of segmented training data with labeled entity tags is used to train the model to enable the model to learn entity features and automatically identify entities in the work order data.

[0045] In addition, assuming that the segmented words include “server”, “hard disk”, “fault”, “light”, “turns on”, “request”, “as soon as possible”, “handle”, “if required”, “on-site”, “handle”, “please”, “in”, “working hours”, “and”, “me”, and “contact”, correspondingly, the entity recognition results include device entities (server, hard disk, etc.), fault type entities (fault light turns on, etc.), action entities (handle, on-site, etc.), and time entities (working hours, etc.).

[0046] In an optional embodiment, after the fault information extraction on the work order data, the fault information is obtained, comprising: using the previously established sentiment dictionary to identify the context of each word, and determining the context sentiment tendency of each word; wherein the context sentiment tendency includes a positive sentiment tendency corresponding to positive words and a negative sentiment tendency corresponding to negative words, the sentiment dictionary is constructed based on the previously collected positive words and negative words, the positive words are used to represent the urgency and importance of the fault, and the negative words are used to represent the fault and the problems existing in the fault; according to the context sentiment tendency of each word, the preset weight distribution rule is combined to determine the sentiment score; wherein the preset weight distribution rule is used to limit the distribution weight corresponding to the context sentiment tendency; according to the sentiment score, the priority of the corresponding work order data is sorted.

[0047] It should be noted that the positive words and the negative words can be set according to the actual work order data involved and the actual design requirements, such as the work order data "server hard disk fault light is on, please contact me during working hours if you need on-site processing", wherein "as soon as possible", "require" and the like can be regarded as positive words expressing urgency and importance, and "fault", "problem" and the like can be regarded as negative words, which are not further limited here. The priority is determined through sentiment analysis and rapid response to determine the urgency of the work order, so as to more reasonably allocate resources and arrange work, and ensure that the work order with high urgency can be processed in priority, thereby improving the efficiency of fault processing and service quality.

[0048] In addition, using the previously established sentiment dictionary to identify the context of each word and determine the context sentiment tendency of each word, comprising: inputting each word into a language model to identify the context of each word using a preset context window, and combining the previously established sentiment dictionary to determine the context sentiment tendency; wherein the language model is trained based on the word segmentation training data and the sentiment tendency label corresponding to the word segmentation training data.

[0049] It should be noted that the contextual sentiment tendency is determined according to the context containing each segmented word, for example, if there is a positive word in the upper and lower segmented words of any segmented word, the positive sentiment tendency of the current segmented word may be enhanced, and if there is a negative word in the context, the negative sentiment tendency of the current segmented word may be enhanced. In addition, the language model can be selected based on actual design requirements, and can be a pre-trained transformer model, such as a bidirectional encoder representation from transformers (BERT) model, a generative pretrained transformer (GPT) model, etc., or a traditional language model, such as an n-gram model, which is not limited further here.

[0050] In an optional embodiment, according to the contextual sentiment tendency of each segmented word, the sentiment score is determined in combination with a preset weight distribution rule, including: based on the preset weight distribution rule, assigning an initial weight to each segmented word; adjusting the initial weight assigned to each segmented word according to the contextual sentiment tendency of each segmented word, and using the adjusted weight to determine the sentiment score of each segmented word in combination with the sentiment tendency; and determining the sentiment score of the corresponding work order data according to the sentiment scores of all segmented words of the same work order data.

[0051] It should be noted that the design of the preset distribution rule can be set according to actual design requirements or prior experience, for example, the negative weight of “serious fault” can be relatively high, and the positive weight of “rapid solution” can also be high; or the preset distribution rule is used to limit the weight distribution according to the statistical result of the sentiment tendency of each segmented word, for example, if the statistical result of the positive sentiment tendency of each segmented word is higher than that of the negative sentiment tendency, it indicates that the positive sentiment tendency has a greater influence, and the corresponding weight can be appropriately increased, which is not limited further here.

[0052] In addition, it is worth noting that when adjusting the initial weight assigned to each segmented word according to the contextual sentiment tendency of each segmented word, the adjustment of the weight can be based on the frequency of the appearance of the segmented word in the context of the sentiment tendency, for example, if the segmented word appears more frequently in the positive context, the positive weight can be further increased, and if the segmented word appears more frequently in the negative context, the negative weight can be further increased. The adjustment of the weight can be configured according to actual design or prior experience, which is not limited further here.

[0053] In step S13, the fault information is matched with the work order instance in the previously established fault feature database to obtain a matched work order instance; wherein the fault feature database is established based on the work order instance and the corresponding processing strategy.

[0054] In an optional embodiment, matching the fault information with the work order instance in the previously established fault feature database comprises: selecting the work order data with the highest priority based on the priority ranking result; and matching the fault information corresponding to the selected work order data with the work order instance in the previously established fault feature database. It should be noted that the priority ranking result is used to determine the urgency of the work order, and the work order with high priority is processed first, so as to reduce the impact of the fault on the business and improve customer satisfaction.

[0055] Specifically, matching the fault information with the work order instance in the previously established fault feature database to obtain a matched work order instance comprises: determining the similarity between the fault information and the work order instance in the previously established fault feature database by using the Jaccard similarity; and determining whether the similarity exceeds a preset threshold, and based on the similarity exceeding the preset threshold, taking the corresponding work order instance as the matched work order instance.

[0056] It should be noted that by using the Jaccard similarity, the similarity between sets is measured to compare the fault information with a large number of work order instances in the fault feature database in a short time, quickly match the instance most similar to the current fault information, and thus more accurately reflect the essential features of the fault and better distinguish different degrees of similarity; by determining whether the similarity exceeds the preset threshold, the matched work order instance is quickly determined, and thus the processing scheme thereof is directly referred to, the work order processing strategy is quickly located, and the accuracy and reliability of work order processing are improved.

[0057] Further, with reference to Figure 2 , the preset threshold is obtained by optimizing the corresponding previously preset threshold based on the feedback information obtained by executing the processing strategy of the previously acquired work order data.

[0058] It should be noted that by using the feedback information after executing the processing strategy, the advantages and disadvantages of the processing strategy are comprehensively evaluated, the preset threshold is continuously optimized, the preset threshold is more suitable for specific fault scenarios and processing requirements, and the accuracy of work order instance matching is improved. In addition, the feedback information includes correction suggestions and customer satisfaction, etc.

[0059] In an optional embodiment, with reference to Figure 2After determining the similarity between the fault information and the work order instance in the previously established fault feature database by using the Jaccard similarity, the method comprises: determining that the similarity does not exceed a preset threshold, and processing the work order data by using a preset processing mode. It should be noted that the preset processing mode can be manual processing or further intelligent analysis processing. The intelligent analysis processing can be further analysis of the work order fault type to determine the corresponding processing mode according to the work order fault type. The specific configuration can be determined according to the actual design requirements, and is not limited here.

[0060] Further, after processing the work order data by using the preset processing mode, the method comprises: storing the fault description of the corresponding work order data and the preset processing mode into the fault feature database to optimize the fault feature database.

[0061] Step S14, according to the matched work order instance, the corresponding processing strategy is called and executed.

[0062] It should be noted that the matched work order instance usually represents a problem that has been successfully solved, and its processing strategy has been verified in practice. Therefore, the processing strategy is quickly located according to the matched work order instance to improve the accuracy and reliability of fault processing.

[0063] In an optional embodiment, after the corresponding processing strategy is called and executed according to the matched work order instance, the method comprises: obtaining feedback information based on the execution of the corresponding processing strategy; and optimizing the corresponding processing strategy in the fault feature database according to the feedback information, perfecting the fault feature database, and improving the accuracy and efficiency of future fault processing.

[0064] In summary, the embodiments of the present application extract fault information from the obtained work order data to clearly define the core content of the fault, remove irrelevant information interference, improve the accuracy and efficiency of subsequent processing, and match the extracted fault information with the work order instance in the fault feature database to quickly locate the verified processing strategy corresponding to the matched work order instance, improve the accuracy of work order processing, reduce the possibility of error processing, reduce the time and cost of starting from scratch to analyze and solve problems, and improve the speed and success rate of fault processing.

[0065] The work order processing device provided by the present application is described below. The work order processing device described below can be referred to in conjunction with the work order processing method described above.

[0066] Figure 3 A structural schematic diagram of a work order processing device is shown. The device comprises:

[0067] The data acquisition module 31 acquires work order data.

[0068] The information extraction module 32 extracts fault information from the work order data to obtain fault information.

[0069] The work order matching module 33 matches the fault information with the work order instances in the previously established fault feature database to obtain a matched work order instance; wherein, the fault feature database is established based on the work order instances and corresponding processing strategies;

[0070] The work order processing module 34 calls the corresponding processing strategy and executes according to the matched work order instance.

[0071] It should be noted that the work order data can be data submitted by users based on a preset platform, including customer questions, processing processes, solution results and other information, which is determined based on the actual received work order data submitted by users. In addition, the preset platform can be a designated platform or other user platforms, which is not limited here.

[0072] In an optional embodiment, the device further comprises a preprocessing module that pre-processes the work order data after obtaining the work order data, the preprocessing including at least one of data cleaning and text conversion. It should be added that through data cleaning, special characters, punctuation marks and other elements that may interfere with subsequent processing in the work order description can be removed, reducing the interference of noise, making the subsequent text analysis and understanding more accurate and efficient; through text conversion, standardized text representation is used to reduce format interference and facilitate quick extraction and understanding of the main content of the work order.

[0073] In a possible implementation, the work order data is in a picture format, and the information extraction module 32 comprises: a feature extraction unit that extracts features from the work order data to obtain a feature vector; and an information extraction unit that extracts fault information from the feature vector using natural language processing (NLP) technology to obtain the fault information.

[0074] Further, the feature extraction unit is configured to: input the work order data into a feature extraction model to obtain a feature vector output by the feature extraction model; wherein, the feature extraction model is trained according to historical work order data and feature labels corresponding to the historical work order data.

[0075] It should be added that the feature extraction model can use a convolutional neural network (CNN) model or other models that can be used to extract feature vectors, which is not limited here.

[0076] In another possible implementation, the work order data is in a text format, and the information extraction module 32 further includes: an information extraction unit configured to extract fault information from the work order data by using a natural language processing (NLP) technique; and a word embedding unit configured to convert the fault information into a vector by using a word embedding technique.

[0077] It should be noted that the word embedding unit is configured to: input the fault information into a word embedding model to obtain fault information in a vector form output by the word embedding model; and the word embedding model is trained according to historical fault information and feature labels corresponding to the historical fault information.

[0078] It should be noted that the fault information used for training is extracted from historical work order data by using an NLP technique, so as to facilitate the word embedding model to learn the context relationship of words, and the word embedding model is trained by using the historical fault information and the feature labels corresponding to the historical fault information, in combination with the model parameters configured in advance, to continuously adjust the vector representation of the fault information, so as to minimize a loss function, thereby facilitating the word embedding model to better capture the semantic association and context information between words, and providing more valuable vector representation for subsequent text analysis and understanding.

[0079] In an optional embodiment, the feature extraction unit and / or the word embedding unit include: a word segmentation subunit configured to perform word segmentation on the feature vector or the work order data in the text format to obtain a plurality of segmented words; a part-of-speech tagging subunit configured to perform part-of-speech tagging on the segmented words to obtain part-of-speech tags; an entity recognition subunit configured to perform entity recognition on the segmented words based on the part-of-speech tags to obtain entity recognition results; and an information extraction subunit configured to extract fault information from the segmented words based on the entity recognition results to obtain the fault information.

[0080] Further, the word segmentation can use a preset word segmentation tool, such as the Jieba word segmentation tool, without further limitation. Correspondingly, the part-of-speech of each segmented word is tagged to determine the part-of-speech of each segmented word, such as a noun, a verb, and an adjective, to help better understand the role of the word in a sentence and provide an important clue for subsequent entity recognition, thereby improving the accuracy and efficiency of entity recognition.

[0081] In addition, the entity recognition subunit is configured to: input the segmented words and the part-of-speech tags corresponding to the segmented words into an entity recognition model to obtain the entity recognition results output by the entity recognition model; and the entity recognition model is trained according to segmented word training data and entity labels corresponding to the segmented word training data, and the segmented word training data includes training segmented words and part-of-speech tags corresponding to the training segmented words.

[0082] In an optional embodiment, the device further includes: an emotion recognition module, which, after extracting fault information from the work order data and obtaining the fault information, uses a pre-established emotion dictionary to identify the context of each word segment and determine the contextual emotion tendency of each word segment; wherein, the contextual emotion tendency includes the positive emotion tendency corresponding to positive words and the negative emotion tendency corresponding to negative words, and the emotion dictionary is constructed based on the pre-collected positive and negative words, with positive words used to characterize the urgency and importance of the fault, and negative words used to characterize the fault and the problems existing in the fault; an emotion scoring module, which determines the emotion score according to the contextual emotion tendency of each word and in combination with a preset weight allocation rule; wherein, the preset weight allocation rule is used to limit the allocation weight corresponding to the contextual emotion tendency; and a sorting module, which prioritizes the corresponding work order data according to the emotion score.

[0083] It should be noted that positive and negative words can be set according to the actual work order data and actual design requirements, and no further restrictions are made here.

[0084] In addition, the sentiment recognition module includes a sentiment analysis unit that inputs each word segment into the language model to identify the context of each word segment using a preset context window, and determines the sentiment tendency by combining it with a pre-established sentiment dictionary; wherein, the language model is trained based on the word segmentation training data and the sentiment tendency labels corresponding to the word segmentation training data.

[0085] In one optional embodiment, the sentiment scoring module includes: a weight allocation unit, which assigns initial weights to each word segment based on a preset weight allocation rule; a weight adjustment unit, which adjusts the initial weights assigned to each word segment according to the contextual sentiment tendency of each word segment; a sentiment scoring unit, which uses the adjusted weights and combines them with the sentiment tendency to determine the sentiment score of each word segment; and a score synthesis unit, which determines the sentiment score of the corresponding work order data based on the sentiment scores of all words in the same work order data.

[0086] It should be noted that the design of the preset allocation rules can be set according to actual design needs or prior experience. For example, the negative weight of "serious fault" can be relatively high, and the positive weight of "quickly resolve" can also be relatively high. Alternatively, the preset allocation rules can be used to limit the allocation of weights based on the statistical results of the sentiment tendency of each word segment. For example, if the statistical results of the positive sentiment tendency in each word segment are higher than the statistical results of the negative sentiment tendency, it indicates that the positive sentiment tendency has a greater influence, and the corresponding weight can be appropriately increased. No further restrictions are made here.

[0087] In addition, it should be noted that, when the weight of each segmented word is adjusted based on the preset allocation rule according to the identified context, the adjustment of the weight can be based on the frequency of the segmented word appearing in the context of the sentiment tendency, for example, if the segmented word appears more frequently in a positive context, the positive weight can be further increased, and if the segmented word appears more frequently in a negative context, the negative weight can be further increased. The adjustment of the weight can be configured according to actual design or prior experience, which is not limited further herein.

[0088] In an optional embodiment, the work order matching module 33 is configured to: based on the priority sorting result, select work order data with the highest priority; and match the fault information corresponding to the selected work order data with the work order instances in the previously established fault feature database. It should be noted that the priority sorting result is used to determine the urgency of the work order, and the work order with high priority is processed preferentially, thereby reducing the impact of the fault on the business and improving customer satisfaction.

[0089] Specifically, the work order matching module 33 includes: a similarity determination unit configured to determine the similarity between the fault information and the work order instances in the previously established fault feature database by using the Jaccard similarity; and a threshold matching unit configured to determine whether the similarity exceeds a preset threshold, and based on the similarity exceeding the preset threshold, determine the corresponding work order instance as a matching work order instance.

[0090] Further, the preset threshold is obtained by optimizing the corresponding previously preset threshold based on the feedback information obtained by executing the processing strategy corresponding to the previously acquired work order data.

[0091] In an optional embodiment, the apparatus further includes: an other processing module configured to, after determining that the similarity does not exceed the preset threshold by using the Jaccard similarity to determine the similarity between the fault information and the work order instances in the previously established fault feature database, process the work order data by using a preset processing manner.

[0092] Further, the apparatus further includes: a storage module configured to, after processing the work order data by using the preset processing manner, store the fault description corresponding to the work order data and the preset processing manner into the fault feature database, so as to optimize the fault feature database.

[0093] In an optional embodiment, the apparatus further includes: an information acquisition module configured to, after invoking the corresponding processing strategy and executing the corresponding processing strategy according to the matching work order instance, acquire feedback information based on the execution of the corresponding processing strategy; and an optimization module configured to optimize the corresponding processing strategy in the fault feature database according to the feedback information, perfect the fault feature database, and improve the accuracy and efficiency of future fault processing.

[0094] To sum up, the embodiment of the present application extracts fault information from the work order data obtained by the data acquisition module through the information extraction module, so as to clearly determine the core content of the fault, remove irrelevant information interference, improve the accuracy and efficiency of subsequent processing, and match the extracted fault information with the work order instances in the fault feature database through the work order matching module, so as to quickly locate the verified processing strategy corresponding to the matched work order instance, improve the accuracy of work order processing of the work order processing module, reduce the possibility of error processing, reduce the time and cost of starting from scratch to analyze and solve problems, and improve the speed and success rate of fault processing.

[0095] Figure 4 An example of a schematic diagram of the physical structure of an electronic device is shown in Figure 4 As shown, the electronic device can include a processor (processor) 410, a communication interface (communications interface) 420, a memory (memory) 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call the logical instructions in the memory 430 to execute the work order processing method, which includes: obtaining work order data; extracting fault information from the work order data to obtain fault information; matching the fault information with the work order instances in the previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on the work order instances and the corresponding processing strategies; and calling the corresponding processing strategy and executing according to the matched work order instances.

[0096] In addition, the logical instructions in the memory 430 described above can be implemented in the form of a software functional unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0097] In another aspect, the present application also provides a computer program product comprising a computer program, the computer program being stored in a non-transitory computer-readable storage medium, and the computer program being executable by a processor to enable a computer to perform the work order processing method provided by any of the above methods, the method comprising: obtaining work order data; extracting fault information from the work order data to obtain the fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; and invoking and executing the corresponding processing strategy according to the matched work order instances.

[0098] In another aspect, the present application also provides a computer program product comprising a computer program, the computer program being stored in a non-transitory computer-readable storage medium, and the computer program being executable by a processor to enable a computer to perform the work order processing method provided by any of the above methods, the method comprising: obtaining work order data; extracting fault information from the work order data to obtain the fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matched work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; and invoking and executing the corresponding processing strategy according to the matched work order instances.

[0099] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment scheme according to actual needs. Those skilled in the art can understand and implement it without creative labor.

[0100] From the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software plus a necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0101] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; and although the present application has been described in detail with reference to the foregoing embodiments, it should be appreciated by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features thereof can be replaced equivalently; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A work order processing method, characterized by, The method comprises the following steps: obtaining work order data; extracting fault information from the work order data to obtain fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matching work order instances; wherein the fault feature database is established based on work order instances and corresponding processing strategies; calling the corresponding processing strategy and executing according to the matching work order instances; extracting fault information from the work order data, comprising: performing feature extraction on the work order data based on the fact that the work order data is in picture format to obtain a feature vector; using natural language processing technology, extracting fault information from the feature vector to obtain fault information; based on the fact that the work order data is in text format, using natural language processing technology to extract fault information from the work order data; using word embedding technology to convert the fault information into a vector; using natural language processing technology to extract fault information from the feature vector or the work order data in text format, comprising: performing word segmentation on the feature vector or the work order data in text format to obtain a plurality of segmented words; performing part-of-speech tagging on each segmented word to obtain a part-of-speech tag; based on the part-of-speech tag, performing entity recognition on the segmented word to obtain an entity recognition result; based on the entity recognition result, extracting fault information from each segmented word to obtain fault information; matching the fault information with work order instances in a previously established fault feature database to obtain matching work order instances, comprising: using Jaccard similarity to determine the similarity between the fault information and the work order instances in the previously established fault feature database; determining whether the similarity exceeds a preset threshold, and based on the fact that the similarity exceeds the preset threshold, regarding the corresponding work order instance as the matching work order instance.

2. The work order processing method of claim 1, wherein, After extracting fault information from the work order data, comprising: using a previously established sentiment dictionary to identify the context of each segmented word and determine the contextual sentiment tendency of each segmented word; wherein the contextual sentiment tendency includes a positive sentiment tendency corresponding to positive words and a negative sentiment tendency corresponding to negative words, the sentiment dictionary is constructed based on previously collected positive words and negative words, the positive words are used to represent the urgency and importance of the fault, and the negative words are used to represent the fault and the problems existing in the fault; determining a sentiment score according to the contextual sentiment tendency of each segmented word and combining a preset weight distribution rule; wherein the preset weight distribution rule is used to limit the distribution weight corresponding to the contextual sentiment tendency; according to the sentiment score, prioritizing the corresponding work order data; matching the fault information with work order instances in a previously established fault feature database to determine the similarity between the fault information and the work order instances in the previously established fault feature database, comprising: based on the prioritization result, selecting the work order data with the highest priority; matching the fault information corresponding to the selected work order data with the work order instances in the previously established fault feature database.

3. The work order processing method of claim 1, wherein, The preset threshold is obtained by optimizing a corresponding previous preset threshold based on feedback information obtained by executing a corresponding processing strategy on previously acquired work order data. The previous preset threshold is a preset threshold used to determine a matching work order instance matching the previously acquired work order data.

4. The work order processing method of claim 1, wherein, After the matching work order instance is obtained, a corresponding processing strategy is invoked and executed, including: Obtaining feedback information based on the execution of the corresponding processing strategy; Optimizing the corresponding processing strategy in the fault feature database according to the feedback information.

5. A work order processing apparatus characterized by comprising: Including: A data acquisition module acquires work order data; An information extraction module extracts fault information from the work order data to obtain fault information; A work order matching module matches the fault information with work order instances in a previously established fault feature database to obtain a matching work order instance; wherein the fault feature database is established based on work order instances and corresponding processing strategies; A work order processing module invokes and executes a corresponding processing strategy according to the matching work order instance; Based on the work order data being in a picture format, the information extraction module includes: A feature extraction unit extracts features from the work order data to obtain a feature vector; An information extraction unit uses natural language processing technology to extract fault information from the feature vector to obtain fault information; Based on the work order data being in a text format, the information extraction module further includes: An information extraction unit uses natural language processing technology to extract fault information from the work order data to obtain fault information; A word embedding unit uses word embedding technology to convert the fault information into a vector; The information extraction unit includes: A word segmentation subunit segments the feature vector or the work order data in a text format to obtain a plurality of segmented words; A part-of-speech tagging subunit tags the parts of speech of each segmented word to obtain a part-of-speech tag; An entity recognition subunit recognizes entities based on the part-of-speech tag to obtain an entity recognition result; An information extraction subunit extracts fault information from each segmented word based on the entity recognition result to obtain fault information; The work order matching module includes: A similarity determination unit determines the similarity between the fault information and work order instances in a previously established fault feature database using Jaccard similarity; A threshold matching unit determines whether the similarity exceeds a preset threshold, and based on the similarity exceeding the preset threshold, a corresponding work order instance is taken as a matching work order instance.

6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the work order processing method of any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the work order processing method of any one of claims 1 to 4.

Images