Service message processing method and device, storage medium and program product

Abstract

The invention discloses a service message processing method and device, a storage medium and a program product, and relates to the technical field of data processing, and the service message processing method comprises the steps: obtaining original service message data; performing text vectorization processing on the original service message data to obtain vectorized service message data; and based on a pre-trained large model, performing label prediction on the vectorized service message data to obtain the tagged service message data, the pre-trained large model is obtained by training an annotation data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model and a stack reverse tracing algorithm. According to the scheme, accurate identification of the service information in the service message data can be realized, and the service message processing efficiency is effectively improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a service message processing method, device, storage medium and program product. Background Art

[0002] Identifying specific business scenarios from massive business message data so that enterprises can perform operation and maintenance management of business services is conducive to discovering potential service quality issues in business scenarios. Existing business message processing methods usually use keyword extraction technology or word segmentation algorithms to screen out keywords or phrases related to specific business scenarios. However, there are still some limitations: First, it still relies on manual labeling of business scenarios and business links for business messages, which is not only inefficient but also prone to errors. Second, traditional word segmentation algorithms often find it difficult to accurately capture the complex relationships and contextual dependencies between words, and are prone to word segmentation errors or omissions, affecting the accuracy of subsequent business message labeling processing.

[0003] Therefore, the existing service message processing method has the problem of low accuracy and efficiency in service message labeling processing.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a business message processing method, device, storage medium and program product, aiming to solve the technical problems of low accuracy and efficiency of business message labeling processing in existing business message processing methods.

[0006] To achieve the above object, the present application proposes a service message processing method, the method comprising:

[0007] Obtain original business message data;

[0008] Performing text vectorization processing on the original business message data to obtain vectorized business message data;

[0009] Based on a pre-trained large model, label prediction is performed on the vectorized business message data to obtain labeled business message data, wherein the pre-trained large model is trained based on an annotated data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

[0010] In one embodiment, before the step of performing label prediction on the vectorized service message data based on the pre-trained large model to obtain the labeled service message data, the following steps are included:

[0011] Based on historical business message data, an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm are used to generate a labeled data set;

[0012] Dividing the labeled data set into a training set, a validation set, and a test set;

[0013] According to the training set, the service message label prediction capability of the large model is trained to obtain a pre-trained large model; and / or

[0014] Based on the validation set and the test set, the model performance of the pre-trained large model is evaluated, and the model is optimized according to the model performance evaluation result to obtain the pre-optimized large model.

[0015] In one embodiment, the step of generating a labeled data set based on historical service message data using an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm includes:

[0016] Based on an unsupervised word segmentation algorithm, the historical business message data is subjected to word segmentation ambiguity processing to obtain first business message data;

[0017] Based on the pre-trained word embedding model, perform text vectorization processing on the first business message data to obtain second business message data;

[0018] Based on a stack reverse tracing algorithm and a preset message data vector set, the second service message data is labeled to obtain a labeled data set.

[0019] In one embodiment, the step of performing word segmentation ambiguity processing on the historical service message data based on an unsupervised word segmentation algorithm to obtain the first service message data includes:

[0020] Based on a preset dictionary word segmentation strategy, the historical service message data is subjected to preliminary word segmentation processing to obtain a plurality of maximum intersection ambiguous fields;

[0021] Based on an unsupervised word segmentation algorithm, the mutual information and boundary entropy of the maximum intersection ambiguous field are calculated to obtain the mutual information value and boundary entropy value of the maximum intersection ambiguous field;

[0022] Based on a preset word segmentation ambiguity processing strategy, word segmentation processing is performed on the maximum intersection ambiguous field according to the mutual information value and the boundary entropy value to obtain first business message data.

[0023] In one embodiment, the step of performing text vectorization processing on the first service message data based on the pre-trained word embedding model to obtain the second service message data includes:

[0024] Based on the pre-trained word embedding model, load the preset word vector set;

[0025] Performing data preprocessing and text segmentation processing on the first business message data to obtain segmented first business message data;

[0026] The segmented first business message data is subjected to text vectorization processing by using a pre-trained word embedding model and a preset word vector set to obtain second business message data.

[0027] In one embodiment, the historical service message data has a service request identifier, and the step of labeling the second service message data based on a stack reverse tracing algorithm and a preset message data vector set to obtain a labeled data set includes:

[0028] According to the business request identifier, call the business full link information to construct a full link directed acyclic graph DAG;

[0029] Based on a depth-first search algorithm, a key message node is located in the full-link DAG, and business information of the historical business message data at the key message node is obtained;

[0030] According to a preset message data vector set, initial label prediction is performed on the second service message data to obtain an initial predicted label;

[0031] The second business message data is labeled according to the business information and the initial prediction label to obtain a labeled data set.

[0032] In one embodiment, the step of locating a key message node in the full-link DAG based on a depth-first search algorithm and acquiring the service information of the historical service message data at the key message node includes:

[0033] If the current message node is abnormal, a reverse search is performed from the current message node in the full-link DAG by calling reverse tracing of the link according to the depth-first search algorithm and the service request identifier until a key message node is found;

[0034] Obtain the corresponding message parsing result of the historical business message data at the key message node to obtain the business information.

[0035] In addition, to achieve the above-mentioned purpose, the present application also proposes a business message processing device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the business message processing method described above.

[0036] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the service message processing method described above are implemented.

[0037] In addition, to achieve the above-mentioned purpose, the present application also proposes a computer program product, which includes a computer program, and when the computer program is executed by a processor, the steps of the service message processing method described above are implemented.

[0038] One or more technical solutions proposed in this application have at least the following technical effects:

[0039] The business message processing method, device, storage medium and program product proposed in the embodiments of the present application are specifically obtained by obtaining original business message data; performing text vectorization processing on the original business message data to obtain vectorized business message data; based on a pre-trained large model, performing label prediction on the vectorized business message data to obtain labeled business message data, wherein the pre-trained large model is trained based on a labeled data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model and a stack reverse tracing algorithm.

[0040] The present application obtains original business message data, performs text vectorization processing on the original business message data first, then inputs the business message data after text vectorization processing into a pre-trained large model, and uses the pre-trained large model to perform label prediction of relevant business information, thereby realizing automatic labeling processing of relevant business information of the business message data, and effectively improving the accuracy and efficiency of business message labeling processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0043] Figure 1 A flowchart of the first embodiment of the method for processing business messages of this application is provided;

[0044] Figure 2 A flowchart of the second embodiment of the method for processing business messages of this application is provided;

[0045] Figure 3 A flowchart of the third embodiment of the method for processing business messages of this application is provided;

[0046] Figure 4 A flowchart of the fourth embodiment of the method for processing business messages of this application is provided;

[0047] Figure 5 This is a logic flow chart of a method for coloring a service message according to a fourth embodiment of the present application;

[0048] Figure 6 A schematic diagram of the device structure of the hardware operating environment involved in the service message processing method in the embodiment of the present application.

[0049] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0050] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0051] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0052] The main solution of the embodiment of the present application is: obtaining original business message data; performing text vectorization processing on the original business message data to obtain vectorized business message data; based on a pre-trained large model, performing label prediction on the vectorized business message data to obtain labeled business message data, wherein the pre-trained large model is trained based on a labeled data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

[0053] Technical terms involved in this application:

[0054] Big model: A big model is a model with powerful prediction and decision-making capabilities that is trained using massive amounts of data through advanced algorithms and technologies. Big models usually have hundreds of millions of parameters, can capture complex patterns in data, capture complex language structures and semantic information, and in many cases can reach or even exceed human performance. Big models can be widely used in many fields, such as natural language processing, image recognition, financial risk control, intelligent recommendation, etc.

[0055] Word embedding: Word embedding is a technique that converts text data into low-dimensional dense vector representations. These vectors can capture the semantic information in the text, so that the similarities and differences between texts can be measured by the distance between vectors. Word embedding is widely used in various natural language processing tasks, such as information retrieval, text classification, clustering, etc. Through word embedding, the model can better understand and process text data and improve the performance of the task.

[0056] Backtracking algorithm: The backtracking algorithm is a search algorithm that is both systematic and jumpy. It searches the solution space tree from the root node according to the depth-first strategy in the solution space tree that contains all the solutions to the problem. When the algorithm searches to any node in the solution space tree, it always first determines whether the node definitely does not contain the solution to the problem. If it definitely does not contain it, it skips the systematic search of the subtree with the node as the root and backtracks to its ancestor node layer by layer. Otherwise, it enters the subtree and continues to search according to the depth-first strategy.

[0057] Stack: A stack is a data structure that follows the last-in-first-out principle. It only allows insertion and deletion operations at one end (called the top of the stack) of an ordered linear data set. Stacks are widely used in computer science, such as function calls, recursive implementations, expression evaluation, etc.

[0058] Depth-first search algorithm: The depth-first search algorithm is an algorithm used to traverse or search a tree or graph. The depth-first search algorithm starts from a vertex of the graph and visits the deepest possible node along the edge of the graph until the node has no unvisited adjacent nodes in the graph, then backtracks to the previous node and continues to search for other unvisited adjacent nodes until all nodes in the graph have been visited. There are many ways to implement the depth-first search algorithm, including recursion and explicit stacks (for iterative implementation). This algorithm is often used for path search, connectivity testing, topological sorting and other problems.

[0059] As the basic unit of data transmission, business messages usually contain rich business information. Identifying specific business scenarios from massive business message data allows enterprises to operate and manage business services, which helps to discover potential service quality issues in business scenarios.

[0060] Since existing business message processing methods usually use keyword extraction technology or word segmentation algorithms to screen out keywords or phrases related to specific business scenarios, but different business links often follow specific message protocols, the current business message processing methods still need to rely on manual labeling of business scenarios and business links. In addition, traditional word segmentation algorithms often find it difficult to accurately capture the complex relationships and contextual dependencies between words, especially when processing business messages containing a large number of professional terms and phrases, which are prone to word segmentation errors or omissions, affecting the accuracy of subsequent labeling.

[0061] Therefore, the existing service message processing method has the problem of low accuracy and efficiency in service message labeling processing.

[0062] The present application provides a solution by obtaining original business message data, first performing text vectorization processing on the original business message data, and then inputting the business message data after text vectorization processing into a pre-trained large model, and using the pre-trained large model to perform label prediction of relevant business information, thereby realizing automatic labeling processing of relevant business information of the business message data, and effectively improving the accuracy and efficiency of business message labeling processing.

[0063] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or a business message processing system capable of realizing the above functions, etc. The following takes the business message processing system as an example to illustrate this embodiment and the following embodiments.

[0064] Based on this, the present application embodiment provides a method for processing a service message, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the service message processing method of the present application.

[0065] In this embodiment, the service message processing method includes steps S110 to S130:

[0066] Step S110, obtaining original service message data;

[0067] It is understandable that since the technical solution proposed in this application is actually applied to the labeling processing of relevant business information of business messages, step S110 needs to be performed. First, the business message processing system needs to obtain the original business message data and use it as the business message to be labeled, so as to use the large model to understand and identify the business information contained in the original business message data and annotate the original business message with relevant business information. Among them, the original business message data refers to the message data generated by the business system in response to the user's business request, which usually includes business scenario information, corresponding business links, response results of business requests, business status information, user information and other contents.

[0068] Step S120, performing text vectorization processing on the original service message data to obtain vectorized service message data;

[0069] Specifically, based on the original business message data obtained above, the text in the original business message data can be segmented and converted into corresponding word vectors through the bag-of-words model, word embedding model, sentence vector model based on deep learning, etc., thereby forming a word vector set as the vectorized business message data.

[0070] Step S130, based on the pre-trained big model, perform label prediction on the vectorized business message data to obtain labeled business message data, wherein the pre-trained big model is trained based on an annotated data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

[0071] It should be noted that the pre-trained big model is obtained by relevant personnel based on actual business message processing needs. In order to make the big model output more accurate label information of the business message, the big model is pre-trained. In this embodiment, the big model can be one or more combinations of large models such as Tongyi Qianwen (Qwen), Meta Large Language Model (Large Language Model MetaAI), and Wenxin Yiyan.

[0072] It should be noted that the pre-trained word embedding model is a word embedding model that is pre-trained in the corpus of the relevant industry by relevant personnel based on the actual word segmentation requirements of the business message. In this embodiment, the word embedding model can be a Word2Vec model.

[0073] Unsupervised word segmentation algorithms are word segmentation algorithms that do not require manual annotation of data. They are algorithms that divide continuous character strings into individual words based on statistical or machine learning methods. Unsupervised word segmentation algorithms use features such as word frequency, contextual information, and co-occurrence relationships between characters to automatically identify word boundaries, thereby achieving word segmentation of character strings.

[0074] In this embodiment, the stack reverse tracing algorithm refers to an algorithm constructed based on the link positioning requirements of the actual business message, utilizing the 'last in, first out' data structure characteristics of the stack and combining the idea of ​​the backtracking algorithm to locate the link request path of the business message in order to obtain the message parsing result of the business message.

[0075] Specifically, the vectorized business message data obtained above is input into the preset training big model, and the big model is used to predict and infer the relevant business information of the vectorized business message data, and label information that can be used for labeling is given, and the business message data is labeled using the label information, so as to obtain the labeled business message data. Among them, the label information includes the business scenario information, business link information, business status information, etc. associated with the current business message data.

[0076] Furthermore, in order to use the big model to predict the labels of relevant business information of the business message data, it is necessary to train and optimize the big model in advance. Therefore, before the above step S130, steps A01 to A04 may also be included:

[0077] Step A01, generating a labeled data set based on historical business message data using an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm;

[0078] Specifically, the business message processing system first needs to collect the business message data generated historically in a certain industry, and then use an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm to perform text vectorization and automatic labeling on the historical business message data, thereby generating a data set with label information, namely, a labeled data set.

[0079] Step A02, dividing the labeled data set into a training set, a validation set and a test set;

[0080] Step A03, training the service message label prediction capability of the large model according to the training set to obtain a pre-trained large model; and / or

[0081] Step A04, performing model performance evaluation on the pre-trained large model according to the validation set and the test set, and performing model optimization according to the model performance evaluation result to obtain a pre-optimized large model.

[0082] Then, in order to train and optimize the large model, the business message processing system needs to divide the labeled data set into training set, validation set and test set.

[0083] The business message processing system trains the business message label prediction capability of the large model based on the divided training set, so that the large model learns the patterns and features of the business message data through the training set, and fine-tunes the parameters of the large model by setting the preset recognition accuracy and recognition time indicators, such as the weight parameters, bias parameters, and layer normalization parameters in the neural network model, so as to obtain a large model that meets the preset recognition accuracy and recognition time, that is, a pre-trained large model. It should be understood that the preset recognition accuracy and recognition time can be data set by relevant personnel based on actual needs or experience, and are not specifically limited here.

[0084] Furthermore, in order to further ensure the business message labeling processing capability of the pre-trained large model, the business message processing system can use the validation set to evaluate the generalization capability of the large model on unseen data, thereby adjusting the hyperparameters of the large model, such as learning rate, batch size, etc., to avoid overfitting of the large model, so that the large model can better adapt to new data and ensure the recognition accuracy and generalization capability of the large model on the test set. The business message processing system uses the test set to ultimately evaluate the performance and generalization capability of the large model, and confirms that the performance of the large model on the test set can reflect its performance in real application scenarios.

[0085] It should be understood that when using the validation set and the test set to optimize and adjust the large model, the performance and generalization ability of the large model can be evaluated through the loss function or the preset recognition accuracy and recognition time indicators, and the label prediction ability of the business message data of the large model can be continuously optimized by continuously reducing the loss function or reaching the preset indicators, so as to obtain the pre-optimized large model, and use the pre-optimized large model to label the business message data.

[0086] It is understandable that when large models are applied to business message data labeling processing tasks in different fields, the validation set can be used to compare the performance of different combinations of models on unseen data to help select the best large model.

[0087] The present embodiment provides a business message processing method, which includes obtaining original business message data; performing text vectorization processing on the original business message data to obtain vectorized business message data; and performing label prediction on the vectorized business message data based on a pre-trained large model to obtain labeled business message data, wherein the pre-trained large model is trained based on a labeled data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

[0088] Through the above-mentioned scheme, the present application obtains the original business message data, and first performs text vectorization processing on the original business message data, and then inputs the business message data after text vectorization processing into a pre-trained large model, and uses the pre-trained large model to perform label prediction of relevant business information, thereby realizing automatic labeling processing of relevant business information of the business message data, and effectively improving the accuracy and efficiency of business message labeling processing.

[0089] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above-mentioned embodiment 1 can refer to the above introduction, and will not be repeated later. Figure 2 , the step A01 further includes steps S210 to S230:

[0090] Step S210, performing word segmentation ambiguity processing on the historical service message data based on an unsupervised word segmentation algorithm to obtain first service message data;

[0091] Specifically, first, the business message processing system needs to automatically identify data features and semantic information such as the frequency of words, contextual information, and co-occurrence relationships between characters in historical business message data based on an unsupervised word segmentation algorithm, so as to automatically identify the boundaries of words, handle the ambiguity caused by different word segmentation methods, and further divide continuous character strings into independent words, so as to implement word segmentation ambiguity processing on historical business message data and obtain business message data after word segmentation ambiguity processing, i.e., the first business message data.

[0092] It should be understood that when using an unsupervised word segmentation algorithm to handle word segmentation ambiguity in historical business message data, in order to further improve the word segmentation accuracy, the historical business message data can be subjected to data preprocessing operations, such as data cleaning and data normalization, and removing special characters, punctuation marks, Hypertext Markup Language (HTML) tags, etc. in the business message text.

[0093] Step S220: performing text vectorization processing on the first business message data based on the pre-trained word embedding model to obtain second business message data;

[0094] It is understandable that business message data is often in text format or other non-numerical formats, while large models usually recognize numerical format data. Therefore, it is necessary to convert the labeled data set into numerical data in order to optimize the large model to use the labeled data set for training and optimization.

[0095] Specifically, the business message processing system can segment the text in the first business message data obtained after word segmentation and convert it into corresponding word vectors through a pre-trained word embedding model, thereby forming a word vector set, and use the vectorized first business message data as the second business message data.

[0096] In one practicable manner, step 220 includes steps B01 to B03:

[0097] Step B01, based on the pre-trained word embedding model, load a preset word vector set;

[0098] Step B02, performing data preprocessing and text segmentation processing on the first service message data to obtain segmented first service message data;

[0099] It is understandable that after pre-training the word embedding model, a word vector set corresponding to the corpus of the relevant industry can be obtained, that is, a preset word vector set, which can be used to achieve more accurate recognition of specific industry terms. In addition, the preset word vector can also be a related word vector set of an open source, trained word embedding model, such as the open source Word2Vec model.

[0100] Specifically, the business message processing system first needs to load a preset word vector set from a pre-trained word embedding model, because the word embedding model is usually stored in a binary format and can be loaded using a corresponding library (such as the gensim library).

[0101] Then, before performing text vectorization processing on the first business message data, it is necessary to preprocess it into data that meets the processing standards of the word embedding model. Specifically, the business message processing system needs to perform data preprocessing operations on the first business message data, including removing stop words, special characters, irrelevant characters, etc. in the business message data, and then perform text segmentation processing on the first business message data, and segment the text in the first business message data into words or sub-words, thereby obtaining the segmented first business message data.

[0102] Step B03, performing text vectorization processing on the segmented first business message data through a pre-trained word embedding model and a preset word vector set to obtain second business message data.

[0103] Finally, through the pre-trained word embedding model, text features are extracted and semantic information is recognized for the words or sub-words in the segmented first business message data, so as to find word vectors that are semantically similar or close to the words or sub-words in the segmented first business message data, and these word vectors are used to form the vectorized business message data, i.e., the second business message data.

[0104] In this embodiment, the pre-trained word embedding model can be used to realize vectorized processing of business message data, which facilitates the subsequent large model to use the vectorized business message data to predict the relevant business information of the business message, thereby improving the accuracy and efficiency of business message labeling processing.

[0105] Step S230: label the second service message data based on the stack reverse tracing algorithm and the preset message data vector set to obtain a labeled data set.

[0106] Specifically, the business message processing system performs similarity matching with the second business message data through a preset message data vector set, and when the similarity reaches a preset threshold, a preliminary predicted label can be obtained. The preset message data vector set is a message data vector set pre-constructed by relevant personnel based on known labeled message data; the preset threshold refers to the vector similarity threshold information set by relevant personnel based on actual needs or experience.

[0107] The stack reverse tracing algorithm is used simultaneously to locate the link request path of the business message according to the historical business message data corresponding to the second business message data, and to find the corresponding relevant business information according to the link request path of the business message, including but not limited to business scenario information, business link information, and business status information. The business message processing system uses these relevant business information and the initially predicted labels to further supplement the label information of the second business message data, thereby obtaining an automatically labeled data set.

[0108] The present embodiment provides a method for processing business messages, which performs word segmentation ambiguity processing on the historical business message data based on an unsupervised word segmentation algorithm to obtain first business message data; performs text vectorization processing on the first business message data based on a pre-trained word embedding model to obtain second business message data; performs label annotation on the second business message data based on a stack reverse tracing algorithm and a preset message data vector set to obtain an annotation data set; divides the annotation data set into a training set, a validation set, and a test set; trains the business message label prediction capability of a large model based on the training set to obtain a pre-trained large model; and / or performs model performance evaluation on the pre-trained large model based on the validation set and the test set, and performs model optimization based on the model performance evaluation result to obtain a pre-optimized large model; obtains original business message data; performs text vectorization processing on the original business message data to obtain vectorized business message data; performs label prediction on the vectorized business message data based on the pre-trained large model to obtain labeled business message data.

[0109] This embodiment uses the above scheme to construct a labeled data set through an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm, and uses the labeled data set to train the large model, thereby obtaining a pre-trained large model, so that the pre-trained large model can be used to perform automatic labeling processing on business message data in the subsequent process, thereby solving the problem of low efficiency in the prior art of relying on manual business message labeling and the problem of low word segmentation accuracy in traditional word segmentation algorithms, and effectively improving the accuracy and efficiency of business message labeling processing.

[0110] Based on the second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the above-mentioned second embodiment can refer to the above introduction, and will not be repeated in the following. Figure 3 The step S210 may further include steps S310 to S330:

[0111] Step S310, based on a preset dictionary word segmentation strategy, performing preliminary word segmentation processing on the historical service message data to obtain a plurality of maximum intersection ambiguous fields;

[0112] Step S320, calculating the mutual information and boundary entropy of the maximum intersection ambiguous field based on an unsupervised word segmentation algorithm to obtain the mutual information value and boundary entropy value of the maximum intersection ambiguous field;

[0113] In this embodiment, the unsupervised word segmentation algorithm is a word segmentation ambiguity processing algorithm based on multi-word mutual information and left and right boundary entropy as word segmentation measurement indicators. The unsupervised word segmentation algorithm includes a preset dictionary word segmentation strategy and a preset word segmentation ambiguity processing strategy.

[0114] It should be noted that the preset dictionary word segmentation strategy is an initial word segmentation strategy based on a dictionary that is pre-set by relevant personnel based on actual word segmentation needs, and is used to find the maximum intersection ambiguous field in historical business message data, thereby performing an initial word segmentation strategy, which may be based on a maximum matching method or a bidirectional maximum matching method, etc. In this embodiment, the preset dictionary word segmentation strategy is a combination strategy based on a forward maximum matching method and a one-word fallback mechanism.

[0115] Specifically, first, the business message processing system needs to adopt a dictionary-based forward maximum matching method. First, according to the longest character string length in the dictionary, the character string of the message text is matched in the forward order, for example, from left to right, in the business message data. When the match is successful, the character string is segmented. When the match fails, a one-word backoff mechanism is adopted, and the length of the character string that failed to match is reduced by one bit, and rematch is performed until all the character strings of the message text are segmented. It should be understood that in order to identify as many maximum intersection ambiguous fields as possible, when using the dictionary-based forward maximum matching method, when a character string is successfully segmented, it is possible to back off one bit from the left end of the successfully matched character string and continue to use the dictionary-based forward maximum matching method to find the character string.

[0116] When there are overlapping intersections in the segmented character string, the intersection fields are combined to obtain a number of maximum intersection ambiguous fields.

[0117] It should be understood that the dictionary mentioned above may be a terminology dictionary of a professional field of a specific industry pre-set by relevant personnel.

[0118] Among them, the maximum intersection ambiguous field refers to the longest intersection segmentation field that cannot be further decomposed in the text. The definition of the maximum intersection ambiguous field is as follows: suppose the string S = c1 c2...cn (S is not a word; c1, c2,..., cn are Chinese characters), the string Smax = ci...cj is a substring of S (1≤i≤j≤n), and Smax is an intersection ambiguous segmentation field. If there is no larger intersection ambiguous segmentation field containing Smax in S, then Smax is called the maximum intersection ambiguous field of S. For example, in the text "He has worked hard to promote Mandarin", "Mandarin", "phone fees", "exhausted", "dedicated", and "heart and effort" may all be words and overlap with each other, so "Mandarin fees" and "Mandarin spends all his heart and effort" both belong to the intersection field, but "Mandarin spends all his heart and effort" is not included in any intersection field, so "Mandarin spends all his heart and effort" is the maximum intersection ambiguous field in this text.

[0119] Then, determine that there may be a segmentation method corresponding string in the maximum intersection ambiguous field, and according to the calculation formula of multi-word mutual information and left and right boundary entropy, calculate the multi-word mutual information and left and right boundary entropy of the string corresponding to the above-mentioned possible segmentation method, and obtain the mutual information value and left and right boundary entropy value corresponding to different segmentation methods of several maximum intersection ambiguous fields. It should be understood that in step S320, the mutual information and boundary entropy of the maximum intersection ambiguous field, that is, the mutual information value and left and right boundary entropy value corresponding to different segmentation methods of the maximum intersection ambiguous field. Wherein, the multi-word mutual information calculation formula is as follows:

[0120]

[0121] Among them, x 1…n Refers to a string of length n.

[0122] In the above formula, the present application expands the mutual information calculation method for multi-character strings. The traditional mutual information calculation method is only applicable to the combination evaluation of two Chinese characters. In the present application, the calculation of multi-character mutual information requires first calculating the mutual information of each pair of adjacent characters in the multi-character word, and then taking the average value of the total mutual information. The above multi-character mutual information calculation method can more accurately reflect the possibility of multi-character words in the corpus, thereby improving the accuracy and reliability of word segmentation.

[0123] Among them, the left and right boundary entropy calculation formula is as follows:

[0124]

[0125] In the above formula, when x is x i-1 When x is x i…j The set of left adjacent words is called the conditional entropy of the left word (referred to as left entropy, denoted by hl(x i…j )), when x is xj+1 When x is x i…j The set of right adjacent words is called the conditional entropy of the right word (referred to as right entropy, denoted as hr(x i…j )). In the above formula, lb refers to the logarithm with base 2, p(x|x i…j ) is the conditional probability.

[0126] In the above formula, this application introduces boundary entropy to measure the possibility of a string becoming a word in the corpus. Considering the diversity of the left adjacent word set and the right adjacent word set of the string, by calculating the left entropy and right entropy of each string in the corpus and taking their average as the criterion, the potential probability of the string becoming a word can be evaluated. The larger the left entropy and right entropy of a string, the more likely it is to be a word.

[0127] It should be understood that the corpus mentioned above is a corpus in a professional field of a specific industry pre-set by relevant personnel.

[0128] Step S330: Based on a preset word segmentation ambiguity processing strategy, word segmentation processing is performed on the maximum intersection ambiguous field according to the mutual information value and the boundary entropy value to obtain first service message data.

[0129] It should be noted that, in this embodiment, the preset word segmentation ambiguity processing strategy refers to a combination strategy of the word segmentation methods of string frequency segmentation, mutual information value and left and right boundary entropy value.

[0130] Specifically, first determine the comprehensive index of the mutual information value and the left and right boundary entropy values ​​corresponding to different segmentation methods of the maximum intersection ambiguous field, where the comprehensive index is a weighted calculation of the mutual information value and the left and right boundary entropy values ​​of the string according to a certain weight ratio. The weight ratio is set by relevant personnel based on actual word segmentation needs or experience.

[0131] Then, determine whether the comprehensive index corresponding to different segmentation methods of the maximum intersection ambiguous field exceeds the preset index threshold. When the comprehensive index exceeds the preset index threshold, use the word segmentation method of mutual information value and left and right boundary entropy value. Specifically, find the segmentation method with the highest comprehensive index from the comprehensive index corresponding to different segmentation methods of the maximum intersection ambiguous field as the final text segmentation method.

[0132] It should be noted that the preset indicator threshold is set by relevant personnel based on actual conditions or expert experience, or based on empirical values ​​verified by multiple large models, and this application does not make specific limitations.

[0133] When the comprehensive index does not exceed the preset index threshold, the string frequency segmentation method is used. Specifically, first determine the words belonging to the dictionary in the string corresponding to the different segmentation methods of the maximum intersection ambiguous field, and then count the sum of the frequencies of each word appearing in the corpus under different segmentation methods, that is, the string frequency of the string under different segmentation methods, and finally decide how to segment the words based on the string frequency of the string corresponding to the different segmentation methods of the maximum intersection ambiguous field. When the string frequency of a segmentation method is significantly higher than other possible segmentation methods, the method with a higher string frequency is adopted as the final segmentation result.

[0134] For example, suppose there is a maximum intersection ambiguous field abc, where a, b, c are all non-empty strings, and ab and bc are words in the dictionary. Then abc has two segmentation methods: ab / c and a / bc. If the frequency of ab in the corpus is greater than the frequency of bc, the segmentation result is ab / c. It should be understood that the frequency of a single character is not included in the statistics.

[0135] This embodiment provides a business message processing method, which performs preliminary word segmentation processing on historical business message data through a preset dictionary word segmentation strategy to find the maximum intersection ambiguous field; then combines mutual information, boundary entropy, and the preset word segmentation ambiguity processing strategy to perform word segmentation and word meaning processing on the maximum intersection ambiguous field, thereby realizing word segmentation processing of business message data, effectively improving word segmentation accuracy, and laying the foundation for subsequent automatic label processing of business message data.

[0136] Based on the second embodiment and / or the third embodiment of the present application, in the fourth embodiment of the present application, the same or similar contents as those of the above-mentioned second embodiment and / or the third embodiment can refer to the above introduction, and will not be repeated later. Figure 4 , the historical service message data has a service request identifier, step S230 includes steps S410 to S440:

[0137] Step S410, calling the full-link information of the service according to the service request identifier, and constructing a full-link directed acyclic graph DAG;

[0138] Step S420: Based on a depth-first search algorithm, locate the key message node in the full-link DAG, and obtain the service information of the historical service message data at the key message node;

[0139] It should be noted that the service request identifier refers to a specific request ID used to identify a service request in the service system, and the request ID is globally unique in the entire link path of the service system. In this embodiment, the service request identifier refers to the traceId in the distributed link tracing technology, and the traceID is a unique identifier used to track requests in a distributed system.

[0140] Specifically, the business message processing system first needs to construct a full-link directed acyclic graph DAG of the business system in real time based on the business request identifier by calling the global link data of the business system.

[0141] Then, according to the business request identifier, the depth-first search algorithm is used to realize reverse tracing of the request path in the full-link DAG, determine the key message nodes in the request path of the business request, and obtain the message data corresponding to the historical business message data at the key message nodes, and parse the message data to obtain the business information corresponding to the historical business message data at the key message nodes.

[0142] In one practicable manner, step 420 includes steps C01 to C02:

[0143] Step C01: if the current message node is abnormal, according to the depth-first search algorithm and the service request identifier, by invoking the reverse tracing of the link, a reverse search is performed from the current message node in the full-link DAG until a key message node is found;

[0144] Step C02, obtaining the corresponding message parsing result of the historical service message data at the key message node to obtain the service information.

[0145] Specifically, if the parsing result of the historical service message data in the current message node is empty, that is, the current message node is abnormal, the service message processing system needs to use a depth-first search algorithm to call reverse tracking of the link based on the service request identifier.

[0146] First, mark the current message node as visited. Then recursively visit each unvisited adjacent message node associated with the current message node in the full-link DAG to confirm whether there is a critical message node. Recursively visit the adjacent message nodes to ensure that all possible link paths are fully traversed.

[0147] When all adjacent message nodes have been visited and there is still no critical message node, it is necessary to trace back to the previous layer of message nodes based on the full-link DAG and visit the previous layer of message nodes. When there is no critical message node in the previous layer of message nodes, continue to visit and confirm its adjacent message nodes, and continue to cycle until the critical message node is found. Among them, the critical message node mentioned above refers to the existence of the parsing result of the message data corresponding to the historical business message data at the critical message node, and the parsing result contains scene information. It should be understood that the critical message node corresponding to each historical business message data is different.

[0148] After the key message node is searched through the above method, the corresponding message of the historical business message data at the key message node is determined, and its message parsing result is obtained. Then, according to the corresponding relevant business information in the message parsing result, the historical business message data is business-colored, that is, the relevant business information obtained by the above parsing is used as a label and annotated to the historical business message data, thereby realizing the business message labeling processing of the historical business message data.

[0149] In this implementation, the reverse tracking of the link is realized through the depth-first search method and the full-link DAG. By calling the reverse tracking of the link, it is ensured that any message node on the request link path of a certain historical business message data normally obtains the message parsing, thereby obtaining the relevant business information of the historical business message data, and realizing the automatic synchronization and annotation of the business information.

[0150] For example, refer to Figure 5 , Figure 5 The present invention is a logic flow chart of an exemplary method for coloring business messages.

[0151] exist Figure 5 In the example, the business links involved in the business request include business link A, business link B, business link C, and business link D. The traceIds of different business links on the call chain are consistent. Among them, the spanId on the call chain is used to identify the position of a remote procedure call in a distributed request. After the business request reaches each business link, the business link will generate a spanId for the request. The parentSpanId is used to identify the position of the upstream remote procedure call in the distributed request. When the business request reaches each business link, the spanId of the upstream business link transmitted from the upstream along with the business request will be recorded as the parentSpanId.

[0152] exist Figure 5 In the process, when business link D, that is, the current message node, cannot parse the message, the depth-first search method and the full-link DAG are used to realize reverse tracing of the call link until it is traced back to business link A, and the business information in its message parsing result is used to color the message data corresponding to the downstream business link.

[0153] Step S430, performing initial label prediction on the second service message data according to a preset message data vector set to obtain an initial predicted label;

[0154] Step S440: annotate the second service message data according to the service information and the initial prediction label to obtain an annotated data set.

[0155] Specifically, the business message processing system performs similarity matching with the second business message data through a preset message data vector set, and when the similarity reaches a preset threshold, a preliminary predicted label, i.e., an initial predicted label, can be obtained. The preset message data vector set is a message data vector set pre-constructed by relevant personnel based on known labeled message data; the preset threshold refers to the vector similarity threshold information set by relevant personnel based on actual needs or experience.

[0156] Then, the business message processing system uses the aforementioned business information, combined with the initial predicted label, to further supplement the label information for the second business message data, thereby obtaining an automatically labeled data set.

[0157] This embodiment provides a method for processing business messages, which realizes reverse tracking of links through a depth-first search method and a full-link DAG. By calling the reverse tracking of the link, it is ensured that after a message is parsed normally for a certain historical business message data at any message node on its request link path, the relevant business information of the historical business message data is obtained, and the automatic synchronization and labeling of the business information can be realized, which effectively solves the problem of being unable to obtain business information when the current message node is abnormal, and lays the foundation for the automatic label processing of subsequent business message data.

[0158] The present application provides a business message processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the business message processing method in the above-mentioned implementation example.

[0159] Reference below Figure 6 , which shows a schematic diagram of the structure of a service message processing device suitable for implementing the embodiment of the present application. The service message processing device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The service message processing device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0160] like Figure 6As shown, the service message processing device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the service message processing device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the service message processing device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a service message processing device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided alternatively.

[0161] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0162] The service message processing device provided by the present application adopts the service message processing method in the above embodiment, which can solve the technical problem of low accuracy and efficiency of service message labeling processing in the existing service message processing method. Compared with the prior art, the beneficial effects of the service message processing device provided by the present application are the same as the beneficial effects of the service message processing method provided by the above embodiment, and the other technical features in the service message processing device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0163] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0164] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0165] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, wherein the computer-readable program instructions are used to execute the service message processing method in the above-mentioned embodiment.

[0166] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), 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 embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0167] The above-mentioned computer-readable storage medium may be included in the service message processing device; or may exist independently without being assembled into the service message processing device.

[0168] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the business message processing device, the business message processing device: obtains original business message data; performs text vectorization processing on the original business message data to obtain vectorized business message data; based on a pre-trained large model, performs label prediction on the vectorized business message data to obtain labeled business message data, wherein the pre-trained large model is trained based on an annotated data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

[0169] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may 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 may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0170] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0171] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0172] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned service message processing method, and can solve the technical problems of low accuracy and efficiency of service message labeling processing in the existing service message processing method. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the service message processing method provided by the above-mentioned embodiment, and will not be repeated here.

[0173] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the above-mentioned service message processing method when executed by a processor.

[0174] The computer program product provided by the present application can solve the technical problem of low accuracy and efficiency of the existing service message processing method for service message labeling. Compared with the prior art, the beneficial effects of the computer program product provided by the embodiment of the present application are the same as the beneficial effects of the service message processing method provided by the above embodiment, which will not be repeated here.

[0175] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent processing scope of the present application.

Claims

1. A method for processing a service message, characterized in that: The method comprises: Obtain original business message data; Performing text vectorization processing on the original business message data to obtain vectorized business message data; Based on a pre-trained large model, label prediction is performed on the vectorized business message data to obtain labeled business message data, wherein the pre-trained large model is trained based on an annotated data set generated by an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm.

2. The method according to claim 1, characterized in that Before the step of performing label prediction on the vectorized service message data based on the pre-trained large model to obtain the labeled service message data, the method includes: Based on historical business message data, an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm are used to generate a labeled data set; Dividing the labeled data set into a training set, a validation set, and a test set; According to the training set, the service message label prediction capability of the large model is trained to obtain a pre-trained large model; and / or Based on the validation set and the test set, the model performance of the pre-trained large model is evaluated, and the model is optimized according to the model performance evaluation result to obtain the pre-optimized large model.

3. The method according to claim 2, characterized in that The step of generating a labeled data set based on historical business message data using an unsupervised word segmentation algorithm, a pre-trained word embedding model, and a stack reverse tracing algorithm includes: Based on an unsupervised word segmentation algorithm, the historical business message data is subjected to word segmentation ambiguity processing to obtain first business message data; Based on the pre-trained word embedding model, perform text vectorization processing on the first business message data to obtain second business message data; Based on a stack reverse tracing algorithm and a preset message data vector set, the second service message data is labeled to obtain a labeled data set.

4. The method according to claim 3, characterized in that The step of performing word segmentation ambiguity processing on the historical service message data based on an unsupervised word segmentation algorithm to obtain the first service message data includes: Based on a preset dictionary word segmentation strategy, the historical service message data is subjected to preliminary word segmentation processing to obtain a plurality of maximum intersection ambiguous fields; Based on an unsupervised word segmentation algorithm, the mutual information and boundary entropy of the maximum intersection ambiguous field are calculated to obtain the mutual information value and boundary entropy value of the maximum intersection ambiguous field; Based on a preset word segmentation ambiguity processing strategy, word segmentation processing is performed on the maximum intersection ambiguous field according to the mutual information value and the boundary entropy value to obtain first business message data.

5. The method according to claim 3, characterized in that The step of performing text vectorization processing on the first service message data based on the pre-trained word embedding model to obtain second service message data includes: Based on the pre-trained word embedding model, load the preset word vector set; Performing data preprocessing and text segmentation processing on the first business message data to obtain segmented first business message data; The segmented first business message data is subjected to text vectorization processing by using a pre-trained word embedding model and a preset word vector set to obtain second business message data.

6. The method according to claim 3, characterized in that The historical service message data has a service request identifier, and the step of labeling the second service message data based on the stack reverse tracing algorithm and the preset message data vector set to obtain a labeled data set includes: According to the business request identifier, call the business full link information to construct a full link directed acyclic graph DAG; Based on a depth-first search algorithm, a key message node is located in the full-link DAG, and business information of the historical business message data at the key message node is obtained; According to a preset message data vector set, initial label prediction is performed on the second service message data to obtain an initial predicted label; The second business message data is labeled according to the business information and the initial prediction label to obtain a labeled data set.

7. The method according to claim 6, characterized in that The step of locating the key message node in the full-link DAG based on the depth-first search algorithm and obtaining the business information of the historical business message data at the key message node includes: If the current message node is abnormal, a reverse search is performed from the current message node in the full-link DAG by calling reverse tracing of the link according to the depth-first search algorithm and the service request identifier until a key message node is found; Obtain the corresponding message parsing result of the historical business message data at the key message node to obtain the business information.

8. A service message processing device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the service message processing method according to any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the service message processing method according to any one of claims 1 to 7 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the service message processing method according to any one of claims 1 to 7 are implemented.

Images