Business processing method and device

Abstract

The present invention discloses a business processing method and device, which relate to the field of natural language processing. A specific implementation of the method includes: in response to a selection operation on a target user, obtaining a business item name set corresponding to the identifier of the target user; calculating the similarity between each business item name and other business item names in the business item name set, and screening similar business item names whose similarity exceeds a preset similarity threshold; extracting overlapping corpora of each business item name and similar business item names, splitting and deduplicating the overlapping corpora to obtain target corpora, and using the target corpora as user tags for each business item name; counting the user tags of all business item names, so as to perform corresponding business processing operations based on the obtained user tags. This implementation only needs to obtain records of users handling specific business items, and the complexity of the algorithm used is relatively low, and there is no threshold for technical implementation, thereby forming a unique label processing system.

Description

Technical Field

[0001] The present invention relates to the field of natural language processing in the field of artificial intelligence, and in particular to a business processing method and device. Background Art

[0002] In recent years, big data development has entered a new phase. From traditional industries to internet companies, from physical goods to financial services, big data-based services are proliferating. However, most companies face challenges with limited data volumes, single data sources, and insufficient data dimensionality. Furthermore, with the increasing emphasis on data security and compliance, data flow is becoming increasingly difficult. Furthermore, data related to a company's core value or government-related data is difficult to open up and share.

[0003] To unlock the value of massive amounts of data, mining user behavior data has become a key area. This data comes from user clicks and application records in various applications, as well as from transaction processing systems. Currently, mining this data remains limited to the user behavior sequence dimension, lacking the underlying content and attributes of transactions. Consequently, it's impossible to construct labels representing user attributes. Summary of the Invention

[0004] In view of this, the embodiments of the present invention provide a business processing method and apparatus, which can at least solve the problem that the existing technology does not mine the content and attributes contained in the matter, and thus cannot process and form labels representing user attributes.

[0005] To achieve the above object, according to one aspect of an embodiment of the present invention, a service processing method is provided, including:

[0006] In response to a selection operation on a target user, obtaining a set of business item names corresponding to an identifier of the target user;

[0007] Calculating the similarity between each business item name and other business item names in the business item name set, and screening similar business item names whose similarity exceeds a preset similarity threshold;

[0008] Extracting overlapping corpora of each business item name and similar business item names, splitting and removing duplicates from the overlapping corpora to obtain target corpora, and using the target corpora as user tags for each business item name;

[0009] Count the user tags of all business item names and perform corresponding business processing operations based on the obtained user tags.

[0010] Optionally, after obtaining the business item name set corresponding to the target user identifier, the method further includes:

[0011] Remove unusual information from each business item name by presetting an unusual information table; and

[0012] De-duplicate the business item names after removing non-routine information.

[0013] Optionally, also include:

[0014] Perform word segmentation on each business item name and remove stop words from the segmentation based on a preset stop word library. Stop words are words that are irrelevant to user attributes.

[0015] For the remaining words, the word frequencies of the words at the preset positions are counted, and a preset number of words ranked at the top are selected in descending order of word frequencies to construct a business verb library;

[0016] After extracting the overlapping corpus of each business item name and the similar business item name, the method further includes: removing the business verbs from the overlapping expectations based on the business verb library to obtain updated overlapping expectations.

[0017] Optionally, before performing corresponding service processing operations based on the obtained user tag, the method further includes:

[0018] Based on the business verb library, screening a first business item name including a preset business verb, and determining a first user tag for the first business item name;

[0019] Filtering all business item names corresponding to the first user tag from the set of business item names corresponding to the target user to determine the second business item name with the latest processing time;

[0020] In response to the second business item name containing the preset business verb, the first user tag is determined to be a non-taggable tag.

[0021] Optionally, calculating the similarity between each business item name and other business item names in the business item name set includes:

[0022] Calculate the Levenshtein distance between each business item name and other business item names;

[0023] Determine the character string length of each business item name and the character string length of the other business item names respectively, and add them up to obtain a sum of the character string lengths;

[0024] The difference between the sum of the string lengths and the Levenshtein distance is calculated, and the ratio of the difference to the sum of the string lengths is used as the similarity between each business item name and the other business item names.

[0025] Optionally, extracting overlapping corpus of each business item name and similar business item names includes:

[0026] determining a first character that ranks first in each of the business item names, and in response to retrieving the first character in the similar business item names, continuing to determine a second character that ranks second in each of the business item names;

[0027] In response to the second character being located after the first character in the similar business item name, continuing to determine the third character ranked third in each of the business item names, repeating the above matching operation until a match fails, and arranging the matched characters in order to obtain a first overlapping corpus;

[0028] Determine first remaining information after removing the first overlapping corpus from each business item name, and second remaining information after removing the first overlapping corpus from the similar business item name, and determine a fourth character ranked first in the first remaining information;

[0029] In response to retrieving the fourth character in the second remaining information, continue to determine the fifth character ranked second in the first remaining information, repeat the above matching operation until the match fails, and arrange the matched characters in order to obtain a second overlapping corpus;

[0030] Continue to determine the remaining information after removing the second overlapping corpus from the first remaining information and the second remaining information, and repeat the above matching operation until all characters in the name of each business item are matched, thereby obtaining multiple repeated corpora;

[0031] The longest overlapping corpus among the multiple repeated corpora is used as the target overlapping corpus for each business item name and the similar business item name.

[0032] Optionally, the taking the longest overlapping corpus among the multiple repeated corpora as the target overlapping corpus for each business item name and the similar business item name includes:

[0033] If there are multiple overlapping corpora with the longest length, multiple overlapping corpora with the longest length are retained.

[0034] Optionally, splitting and removing duplicates from the overlapping corpora to obtain target corpora includes:

[0035] Determine the overlapping part and the non-overlapping part of each overlapping corpus and each other overlapping corpus, and in response to the overlapping part being not empty, use the corpus of the overlapping part and the non-overlapping part as the first corpus to be screened;

[0036] Determine the overlapping part and the non-overlapping part of each first corpus to be screened and each other first corpus to be screened, and in response to the overlapping part not being empty, continue to use the corpus of the overlapping part and the non-overlapping part as the second corpus to be screened;

[0037] The overlapping part and the non-overlapping part of each second corpus to be screened and each other second corpus to be screened are determined, and in response to the overlapping part being empty, the corpus of the non-overlapping part is used as the target corpus.

[0038] To achieve the above object, according to another aspect of an embodiment of the present invention, a service processing device is provided, including:

[0039] An acquisition module, configured to acquire, in response to a selection operation on a target user, a set of business item names corresponding to an identifier of the target user;

[0040] A calculation module, configured to calculate the similarity between each business item name and other business item names in the business item name set, and filter similar business item names whose similarity exceeds a preset similarity threshold;

[0041] an extraction module, configured to extract overlapping corpora of each business item name and similar business item names, split and remove duplicates from the overlapping corpora to obtain target corpora, and use the target corpora as user tags for each business item name;

[0042] The processing module is used to count the user tags of all business item names and perform corresponding business processing operations based on the obtained user tags.

[0043] Optionally, the acquisition module is further configured to:

[0044] Remove unusual information from each business item name by presetting an unusual information table; and

[0045] De-duplicate the business item names after removing non-routine information.

[0046] Optionally, the acquisition module is further configured to:

[0047] Perform word segmentation on each business item name and remove stop words from the segmentation based on a preset stop word library. Stop words are words that are irrelevant to user attributes.

[0048] For the remaining words, the word frequencies of the words at the preset positions are counted, and a preset number of words ranked at the top are selected in descending order of word frequencies to construct a business verb library;

[0049] The extraction module is further used to: based on the business verb library, remove the business verbs from the overlapping expectations to obtain updated overlapping expectations.

[0050] Optionally, the processing module is further configured to:

[0051] Based on the business verb library, screening a first business item name including a preset business verb, and determining a first user tag for the first business item name;

[0052] Filtering all business item names corresponding to the first user tag from the set of business item names corresponding to the target user to determine the second business item name with the latest processing time;

[0053] In response to the second business item name containing the preset business verb, the first user tag is determined to be a non-taggable tag.

[0054] Optionally, the computing module is configured to:

[0055] Calculate the Levenshtein distance between each business item name and other business item names;

[0056] Determine the character string length of each business item name and the character string length of the other business item names respectively, and add them up to obtain a sum of the character string lengths;

[0057] The difference between the sum of the string lengths and the Levenshtein distance is calculated, and the ratio of the difference to the sum of the string lengths is used as the similarity between each business item name and the other business item names.

[0058] Optionally, the extraction module is used to:

[0059] determining a first character that ranks first in each of the business item names, and in response to retrieving the first character in the similar business item names, continuing to determine a second character that ranks second in each of the business item names;

[0060] In response to the second character being located after the first character in the similar business item name, continuing to determine the third character ranked third in each of the business item names, repeating the above matching operation until a match fails, and arranging the matched characters in order to obtain a first overlapping corpus;

[0061] Determine first remaining information after removing the first overlapping corpus from each business item name, and second remaining information after removing the first overlapping corpus from the similar business item name, and determine a fourth character ranked first in the first remaining information;

[0062] In response to retrieving the fourth character in the second remaining information, continue to determine the fifth character ranked second in the first remaining information, repeat the above matching operation until the match fails, and arrange the matched characters in order to obtain a second overlapping corpus;

[0063] Continue to determine the remaining information after removing the second overlapping corpus from the first remaining information and the second remaining information, and repeat the above matching operation until all characters in the name of each business item are matched, thereby obtaining multiple repeated corpora;

[0064] The longest overlapping corpus among the multiple repeated corpora is used as the target overlapping corpus for each business item name and the similar business item name.

[0065] Optionally, the extraction module is used to:

[0066] If there are multiple overlapping corpora with the longest length, multiple overlapping corpora with the longest length are retained.

[0067] Optionally, the extraction module is used to:

[0068] Determine the overlapping part and the non-overlapping part of each overlapping corpus and each other overlapping corpus, and in response to the overlapping part being not empty, use the corpus of the overlapping part and the non-overlapping part as the first corpus to be screened;

[0069] Determine the overlapping part and the non-overlapping part of each first corpus to be screened and each other first corpus to be screened, and in response to the overlapping part not being empty, continue to use the corpus of the overlapping part and the non-overlapping part as the second corpus to be screened;

[0070] The overlapping part and the non-overlapping part of each second corpus to be screened and each other second corpus to be screened are determined, and in response to the overlapping part being empty, the corpus of the non-overlapping part is used as the target corpus.

[0071] To achieve the above objective, according to another aspect of an embodiment of the present invention, a service processing electronic device is provided.

[0072] The electronic device of an embodiment of the present invention includes: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement any of the business processing methods described above.

[0073] To achieve the above object, according to another aspect of an embodiment of the present invention, a computer-readable medium is provided, on which a computer program is stored. When the program is executed by a processor, any of the above-mentioned business processing methods is implemented.

[0074] To achieve the above object, according to another aspect of an embodiment of the present invention, a computer program product is provided. A computer program product according to an embodiment of the present invention includes a computer program, which, when executed by a processor, implements the service processing method provided by an embodiment of the present invention.

[0075] According to the solution provided by the present invention, one embodiment of the above invention has the following advantages or beneficial effects: the whole does not rely on massive training data, and there is no need to obtain specific data from relevant departments. It only needs to obtain records of users handling specific business matters. Therefore, it can be widely used in different business matter names. The complexity of the algorithm used as a whole is relatively low, and there is no need for the participants to have overly professional knowledge. There is no threshold for technical implementation, thereby forming a unique label processing system. The whole can be regarded as an original application of natural language processing technology to extract user labels in business matter names.

[0076] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] The accompanying drawings are provided for a better understanding of the present invention and are not intended to limit the present invention.

[0078] Figure 1 This is a schematic diagram of the main flow of a service processing method according to an embodiment of the present invention;

[0079] Figure 2 is a flowchart of an optional service processing method according to an embodiment of the present invention;

[0080] Figure 3 is a flowchart of another optional service processing method according to an embodiment of the present invention;

[0081] Figure 4 is a flowchart of another optional service processing method according to an embodiment of the present invention;

[0082] Figure 5 1 is a schematic diagram of main modules of a service processing device according to an embodiment of the present invention;

[0083] Figure 6 is an exemplary system architecture diagram in which embodiments of the present invention may be applied;

[0084] Figure 7 It is a schematic diagram of the structure of a computer system of a mobile device or server suitable for implementing the embodiments of the present invention. DETAILED DESCRIPTION

[0085] The following description of exemplary embodiments of the present invention is made in conjunction with the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding. These details should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0086] It should be noted that the embodiments and features of the embodiments of the present invention can be combined with each other without conflict. The acquisition, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations.

[0087] Natural language processing technology is primarily used to identify text topics, identify the objects they target, and construct knowledge graphs. It's not used to infer user attributes or label users based on transaction data. Natural language processing can also be used in the field of intelligent annotation, but it primarily focuses on training and labeling text. This requires manual annotation of correct and incorrect results and repeated training iterations, without considering the attributes of the text content itself.

[0088] This solution can be applied to areas where users need to be labeled with attribute labels, such as the government service field. Government data is generally stored in various government departments. For example, identity card information comes from government department 1, marital status and household registration information comes from government department 2, educational information comes from government department 3, motor vehicle driver's license is obtained in government department 4, business registration information comes from government department 5, etc. There is less data that can be shared between each other, and it is difficult to open it to the outside world considering user privacy. Therefore, it is difficult to label users in various fields.

[0089] Insights into user attributes require interfacing with every government department, and obtaining relevant data can be difficult in situations where data confidentiality is high. Therefore, this solution can be seen as an innovative application of natural language processing technology in user profiling scenarios within the government service sector. User labels processed using government data provide more comprehensive user attribute information and eliminate data silos. For example, user label processing scenarios are generally divided into user attributes and user behavior.

[0090] 1. User attributes include gender, age, education level, marital status, household registration type, driver's license, and business registration information. The data sources for these tags are disparate and highly private. Furthermore, the number of tags processed is directly proportional to the labor cost, requiring data cleansing and the rationalization of tag processing logic.

[0091] 2. The processing logic of user behavior is generally "time period + behavior + statistical indicators", such as "the number of times a user has obtained a marriage certificate in the past month" and "the number of types of government affairs handled by a user in the past six months." Such tag-based scenarios require high customization and business-driven customized development. Therefore, the application of user behavior data cannot be effectively and maximized.

[0092] This solution can solve two problems: one is the insufficient mining of user attribute information covered by user affairs during the formation of user tags; the other is the low processing efficiency, few usage scenarios and obvious business-driven attributes of user behavior tags.

[0093] See also Figure 1 , which shows a main flow chart of a service processing method provided by an embodiment of the present invention, including the following steps:

[0094] S101: In response to a selection operation on a target user, obtaining a business item name set corresponding to an identifier of the target user;

[0095] S102: Calculating the similarity between each business item name and other business item names in the business item name set, and screening similar business item names whose similarity exceeds a preset similarity threshold;

[0096] S103: extracting overlapping corpora of each business item name and similar business item names, splitting and removing duplicates from the overlapping corpora to obtain target corpora, and using the target corpora as user tags for each business item name;

[0097] S104: Counting the user tags of all business item names, and performing corresponding business processing operations based on the obtained user tags.

[0098] In the above implementation, for step S101, this solution is primarily used in the field of government services, such as customized content display scenarios on government portals (PC, App, Mini Program, etc.), policy recommendations and precise matching of policy matters, and "policy search for people" scenarios. Government affairs processing data is used, and similar data is not available in other fields. Furthermore, the naming of government affairs follows a unified pattern, which also ensures the accuracy and representativeness of user tags extracted using natural language processing technology.

[0099] After manually or automatically selecting a target user, a user profile operation is triggered. The process begins by obtaining the government service item handling log corresponding to the user and then extracting the business item name from the log. Business item names often contain punctuation and garbled characters, which can affect subsequent processing. Therefore, data cleansing is first performed on the business item names to remove any unconventional information from them using a pre-set table of unconventional information, including symbols and geographic information.

[0100] For example, "Handling the procedures for project quality supervision (Science and Technology Service Platform Section # Workshop and Underground Garage, # Workshop and Corridor and Underground Garage)" becomes "Handling the procedures for project quality supervision (Science and Technology Service Platform Section Workshop and Underground Garage Workshop and Corridor and Underground Garage)" after data cleaning, "Living allowance for disabled people in need is paid?" becomes "Living allowance for disabled people in need is paid" after data cleaning, "Change in labor dispatch business license (registered capital, change in labor dispatch business license (registered capital)" becomes "Change in registered capital of labor dispatch business license" after data cleaning, and "Beijing Gas Registration" does not contain non-routine information, so no data cleaning is required.

[0101] It's important to note that geographic information in unconventional information tables typically refers to administrative divisions, such as underground garages and corridors, and must be excluded using a pre-set vocabulary. However, this information is low-frequency and will not be extracted as user tags in subsequent operations. Therefore, even if it is not removed, it will not affect the final labeling results.

[0102] Furthermore, it can be further expanded according to the actual operation scenario. For example, in order to distinguish the subject of the label object, the handling of matters can be divided into natural person matters and legal person matters according to the type of handling entity. Natural person labels and legal person labels are two sets of label systems. Natural person labels are extracted from the natural person related matters handling log, and legal person labels are extracted from the legal person matters handling log.

[0103] After the above processing is completed, some business item names may be duplicated. For example, the aforementioned "Labor Dispatch Business License Change (Registered Capital)" will both become "Labor Dispatch Business License Change Registered Capital" after data cleansing. To reduce the subsequent calculation workload, you can perform a deduplication operation and retain only one "Labor Dispatch Business License Change Registered Capital."

[0104] For step S102, the similarity of each two business item names is calculated using the Levenshtein distance calculation formula. The Levenshtein distance is also known as the name edit distance, which refers to the minimum number of edits required to convert one string into another between two strings. The editing behavior includes replacing characters, inserting characters, and deleting characters. The edit distance is calculated using a dynamic programming algorithm. Among them, the edit distance is a quantitative measure of the degree of difference between two strings (such as English words). The measurement method is to see how many times it takes to convert one string into another. The edit distance can be used in natural language processing. For example, spell checking can determine which one (or which ones) is the more likely word based on the edit distance between a misspelled word and other correct words.

[0105] Dynamic programming algorithms are often used to solve problems with some optimality property. These problems may have many feasible solutions. Each solution corresponds to a value, and the goal is to find the solution with the optimal value. Similar to the divide-and-conquer algorithm, the basic idea of ​​dynamic programming is to decompose the problem into several subproblems, first solving the subproblems, and then deriving the solution to the original problem from the solutions to these subproblems. However, unlike the divide-and-conquer algorithm, the subproblems suitable for dynamic programming are often not independent of each other. Using the divide-and-conquer algorithm to solve such problems can result in an excessive number of subproblems, and some subproblems may be repeatedly calculated. Saving the answers to solved subproblems and retrieving them when needed can avoid extensive recalculation and save time. A table can be used to record the answers to all solved subproblems. Regardless of whether a subproblem is used again, as long as it has been calculated, its result is entered into the table. This is the basic idea of ​​dynamic programming.

[0106] The Levenshtein distance calculation formula is as follows:

[0107]

[0108] For strings X and Y, |X| represents the length of string X, and |Y| represents the length of string Y. Lev(i,j) represents the Levenshtein distance between the first i characters of string X and the first j characters of string Y, and Lev(|X|,|Y|) represents the Levenshtein distance between the two strings X and Y. X|i-1|≠Y|j-1| Represents an exponential function, which is 1 if the (i-1)th item in string X is not equal to the (j-1)th item in string Y, and 0 otherwise.

[0109] Specifically, X[0] is the first character of string X, and X[1] is the second character of the string. The Levenshtein distance calculation formula must be deduced starting from i = 0 and j = 0. To calculate Lev(1,1), the formula shows that we must first calculate the values ​​of Lev(0,1) + 1, Lev(1,0) + 1, and Lev(0,0) + the indicator function, and then take the minimum value of these three. Substituting Lev(0,1) into the formula, for i = 0, Lev(0,1) = j = 1. Similarly, Lev(1,0) = i = 1 and Lev(0,0) = 0.

[0110] To calculate Lev(7,9), you need to calculate Lev(7,8), Lev(6,9), and Lev(6,8). However, to calculate these three, you also need to calculate Lev(7,7), Lev(6,7), Lev(5,9), Lev(5,8), Lev(6,7), and Lev(5,7). And so on, ultimately coming down to calculating Lev(0,0), Lev(0,1), and Lev(1,0). Therefore, each step requires large amounts of data and computation, and calculating the distance between two strings increases exponentially.

[0111] Let's use a specific example to calculate the edit distance between the strings "The weather is great today" and "I was in a bad mood yesterday." With |X| and |Y| being 6 and 6, respectively, we need to calculate the edit distance of the first six characters of the two strings, which is Lev(6,6). Substituting this into the Levenshtein distance formula, since i = j = 6, the third case in the formula is equal to Lev(5,6)+1, Lev(6,5)+1, and Lev(5,5)+the value of the indicator function. In this example, the fifth character of string X is "true" and the fifth character of string Y is "not," making them unequal. The indicator function satisfies the condition and evaluates to 1, so the third row in the otherwise case evaluates to Lev(5,5)+1. To calculate Lev(5,6), Lev(6,5), and Lev(5,5), we need to repeat the formula by substituting Lev(i,j). The key to calculating each step is the indicator function value; the similarities and differences between characters are crucial in determining the edit distance.

[0112] After calculating the Levenshtein distance between two business item names, calculate the Levenshtein ratio between them. The calculation formula is as follows:

[0113] r=(sum-Lev(|X|,|Y|)) / sum

[0114] Here, sum refers to the sum of the lengths of string X and string Y. For example, to convert "The weather is great today" to "I had a bad temper yesterday," we first replace "zuo" and "pi" and then delete "bu." The Levenshtein distance is 3. The string length of "The weather is great today" is 6, and the string length of "I had a bad temper yesterday" is 7. The Levenshtein ratio is ((6+7)-3) / (6+7) = 76.92%, which represents the similarity between strings X and Y.

[0115] This solution uses the Levenshtein ratio as its similarity algorithm. When calculating edit distance, it can use simple matching, partial matching, or matching that ignores order. Regardless of which approach is used, this solution produces relatively accurate labeling results. Therefore, the computing resource requirements can be adjusted based on actual conditions, allowing for simple matching or complex matching with increased computing resources. This offers advantages such as high flexibility and guaranteed practicality.

[0116] For step S103, through the above steps, the similarity between each business item name and the names of all other items can be calculated, and only similar business item names with a similarity of more than 70% (i.e., a preset similarity threshold, which is actually adjustable) are retained, and the overlapping corpus of each business item name and each similar business item name is extracted. Assume that the current business item name is "Subsidy Payment for Economically Disadvantaged Elderly People" and its similarity with "Subsidy Payment for Economically Disadvantaged Elderly People in Cities" is 92%, and the overlapping corpus is "Subsidy Payment for Economically Disadvantaged Elderly People."

[0117] For the same user, there may be multiple business item names. The same process needs to be repeated for the next business item name to extract the user attribute label for that name. Assuming there are 100 business item names, the logic for processing only the first business item name is shown here. The similarity between the first item and all other items is calculated. Then, items with a similarity greater than 70% are retained and entered into the corpus extraction process. This process needs to be repeated for the next 99 business item names.

[0118] For step S104, according to the above steps, the user label of each business item name can be determined. The confirmed label is stored in the user's portrait result table, and an interface service is provided for various downstream applications to call. For example, it is used for user feature analysis to analyze the attributes of users on a certain platform, and then optimize the platform's services based on these attributes. For example, if most users have a "rural" label, then the service process provided by the platform should be as simple as possible to facilitate user understanding. If some users have a "two-child" label, then the platform should provide customized guidance and reinforcement in terms of fertility services, or provide some new subsidy policies for such users.

[0119] Or, if a government department releases a new policy, to ensure personalized service, policies related to "rural" will be pushed to users with the "rural" tag. Policies encouraging childbirth or providing subsidies for second children will only be communicated to users with the "two-child" tag. Policies can be displayed on government portals or government service apps based on user tags, or they can be proactively sent text messages to notify users of their policy targets, and so on.

[0120] The method provided in the above embodiment does not need to aggregate data from various relevant departments. It only needs to obtain records of users handling business matters to extract rich and valuable tags. The entire solution does not rely on the complexity of the algorithm, nor does it require the participation of personnel with professional business knowledge and business capabilities. The overall investment cost is low, and the consumption of computer resources is also low, so it is highly reusable and scalable.

[0121] See also Figure 2 , shows a schematic flow chart of an optional service processing method according to an embodiment of the present invention, including the following steps:

[0122] S201: extracting overlapping corpus of each business item name and similar business item names;

[0123] S202: Determine the overlapping part and the non-overlapping part of each overlapping corpus and each other overlapping corpus, and in response to the overlapping part being not empty, use the corpus of the overlapping part and the non-overlapping part as the first corpus to be screened;

[0124] S203: Determine the overlapping and non-overlapping parts of each first corpus to be screened and each other first corpus to be screened, and in response to the overlapping parts not being empty, continue to use the corpus of the overlapping and non-overlapping parts as the second corpus to be screened;

[0125] S204: Determine the overlapping parts and non-overlapping parts of each second corpus to be screened and each other second corpus to be screened, and in response to the overlapping parts being empty, use the corpus of the non-overlapping parts as the target corpus.

[0126] In the above implementation, for step S201, this implementation is used to describe how to generate overlapping corpus. Directly extracting duplicate corpus will cause information interference, and only words with very low information entropy such as "insurance", "qualification", and "company" will be extracted. Only matters with similar names can cover the most complete attribute information, such as "medical insurance", "teacher qualification", "joint-stock company", etc. This solution is based on screening out similar business matter names.

[0127] There is a unified naming rule for the names of business items in the field of government affairs services. Usually, they start or end with a verb, such as "Apply for XXXX", "Collect XXXX", "Register XXXX", "Change XXXX", "Approve XXXX". Based on this naming rule and business attributes, the names of business items can be segmented (usually using jieba segmentation), and the word frequencies of the first two words at the beginning and the last two words at the end are calculated, sorted in descending order of word frequency, and the top 100 high-frequency words are selected. After manual screening, a special verb library for government affairs (i.e., a business verb library) is determined. Since the probability of change in the names of government affairs items is low and they are the same in each region, high-frequency model training is not required, and it has the advantages of reusability and generalizability.

[0128] In addition, a stop word library can be created, such as words unrelated to user attributes like "of", "or", "and", etc. Stop words are used for high-frequency word screening. For example, "of" is a high-frequency word, but it cannot be extracted as a user label because this word itself has no meaning.

[0129] See Table 1 below:

[0130] Table 1 Extracting the overlapping corpus of the current business item name and similar business item names

[0131]

[0132] The processing logic of the first business item name among all the obtained business item names shown in Table 1. There may be multiple other business item names with a similarity greater than 70% to the first business item name. The overlapping parts are extracted by using the first business item name and these similar business item names respectively. So there will be many extracted corpora, and they may be repeated. Therefore, there may also be the same corpora among other business item names. The overlapping corpora are de-duplicated. For example, the overlapping corpora in Table 1 after de-duplication result in five corpora, namely "Subsidy payment for economically disadvantaged elderly people", "Subsidy for economically disadvantaged elderly people", "Economically disadvantaged elderly people", "Economically disadvantaged", and "Subsidy payment".

[0133] Using the aforementioned special verb library for government affairs, the special verbs for government affairs in the de-duplicated overlapping corpora are removed. Assume the special verb for government affairs is "payment". For the aforementioned "Subsidy payment for economically disadvantaged elderly people", "Subsidy for economically disadvantaged elderly people", "Economically disadvantaged elderly people", "Economically disadvantaged", and "Subsidy payment", the "payment" in "Subsidy payment for economically disadvantaged elderly people" needs to be removed to get "Subsidy for economically disadvantaged elderly people", and the "payment" in "Subsidy payment" needs to be removed to get "Subsidy". After de-duplication again, the actual obtained corpora are these four: "Economically disadvantaged elderly people", "Subsidy for economically disadvantaged elderly people", "Economically disadvantaged", and "Subsidy".

[0134] For steps S202 to S204, the secondary corpus is extracted next. Each overlapping corpus is matched with other overlapping corpora to be divided into overlapping parts and non-overlapping parts. The corpus that does not overlap with other overlapping corpora is retained, and this process is repeated until all overlapping corpora no longer have overlapping parts. For example, the overlapping part of "elderly people with financial difficulties (called a)" and "subsidy for elderly people with financial difficulties (called b)" is "elderly people with financial difficulties", the overlapping part with "financial difficulties (called c)" is "financial difficulties", and there is no overlapping part with "subsidy (called d). See Table 2 for the overlapping and non-overlapping parts of each overlapping corpus and each other overlapping corpus:

[0135] Table 2

[0136]

[0137] All overlapping and non-overlapping parts are sorted together to obtain the first corpus to be screened, including "elderly people with financial difficulties (referred to as e)", "economic difficulties (referred to as f)", "subsidies (referred to as g)", "elderly people (referred to as h)", and "elderly subsidies (referred to as k)". Due to the presence of overlapping parts, further screening is required. For example, e and f overlap in the part "economic difficulties", but do not overlap with g, and overlap with h and k in the part "elderly". f does not overlap with g, h, or k. g does not overlap with h, but overlaps with k in the part "subsidies". h and k overlap in the part "elderly".

[0138] Therefore, we continue sorting the overlapping and non-overlapping parts together to obtain the second set of corpora to be screened, including "economic difficulties," "elderly people," and "subsidies." Since no two corpora overlap, we use "economic difficulties," "elderly people," and "subsidies" as the user attribute labels corresponding to the current business item name, "Subsidies for Elderly People in Economic Difficulties."

[0139] It should be noted that the repeated corpus extraction process described above is only for the business item name currently being processed. For the same user, there may be multiple business item names. For the next business item name, the same operation needs to be repeated to extract the user label of the next business item name.

[0140] The method provided in the above embodiment adopts the method of calculating the similarity between business item names and extracting repeated corpus to ensure the business relevance of repeated corpus, solves the information interference problem caused by directly extracting repeated corpus, and further ensures the accuracy of user tags.

[0141] See also Figure 3, shows a schematic flow chart of another optional service processing method according to an embodiment of the present invention, including the following steps:

[0142] S301: determining a first character that ranks first in each business item name, and in response to retrieving the first character in similar business item names, continuing to determine a second character that ranks second in each business item name;

[0143] S302: In response to the second character being located after the first character in the similar business item name, continuing to determine the third character ranked third in each of the business item names, repeating the above matching operation until a match fails, and arranging the matched characters in order to obtain a first overlapping corpus;

[0144] S303: Determine first remaining information after removing the first overlapping corpus from each business item name, and second remaining information after removing the first overlapping corpus from the similar business item name, and determine the fourth character ranked first in the first remaining information;

[0145] S304: In response to retrieving the fourth character in the second remaining information, continue to determine the fifth character ranked second in the first remaining information, repeat the above matching operation until the match fails, and arrange the matched characters in order to obtain a second overlapping corpus;

[0146] S305: Continue to determine the remaining information after removing the second overlapping corpus from the first remaining information and the second remaining information, and repeat the above matching operation until all characters in the name of each business item are matched, thereby obtaining multiple repeated corpora;

[0147] S306: Using the longest overlapping corpus among the multiple repeated corpora as the target overlapping corpus for each business item name and the similar business item name.

[0148] In the above implementation, for steps S301 to S306, there may be multiple overlapping corpora. For example, "Subsidy for Elderly People with Financial Difficulties (referred to as A)" and "Living Subsidy for Elderly People with Financial Difficulties (referred to as B)" are examples. Because of the word "living" in "Living Subsidy for Elderly People with Financial Difficulties", the overlapping parts are separated. Here is the loop matching logic:

[0149] 1) First, process the first character in A, which is "经". It is found that this character also exists in B. Then continue to loop to the second character in A, which is "济". If the character adjacent to "经" in B is also "济", then continue the matching. And so on. When A matches to "补", the character after "人" in B is "生", which is different, and the matching process ends. The first overlapping corpus obtained is "经济困难老年人".

[0150] 2) The characters in A have not been completely matched. Continue to process "补". It is found that "补" also exists in B. So continue to process the next character after "补" in A, which is "贴". It is found that the character adjacent to "补" in B is also "贴", then continue to process the next character in A, which is "给". Since there are no more characters after "贴" in B, the matching process ends. The overlapping part is "补贴".

[0151] 3) The characters in A have not been completely matched. Continue to process "给". Since this character does not exist in B, this is the matching result.

[0152] 4) The characters in A have not been completely matched. Continue to process "付". Since this character does not exist in B, this is the matching result.

[0153] 5) All the characters in A have been completely matched, and the extraction of the overlapping corpus ends.

[0154] Through the above loop regular matching operation, it can be known that the completely overlapping and continuous corpora of "经济困难老年人补贴给付" and "经济困难老年人生活补贴)" are two: "经济困难老年人" and "补贴". Extract the one with the longest string length, "经济困难老年人", as the target overlapping corpus of the two. Among them, regular matching describes a pattern of string matching, which can be used to check whether a string contains a certain substring, replace the matching substring, or extract a substring that meets a certain condition from a string, etc.

[0155] However, if there are multiple overlapping parts with the same length, for example, the overlapping corpora of "经济困难老年人补贴给付" and "经济困难群体补贴给付" are two: "重合预料" and "补贴给付", both with a length of 4, then all are retained.

[0156] [[ID=...]]This solution depends on the high overlap of the smallest business attribute corpora in the business item names. If you want to extract the complete repeated corpus, it cannot be completed by relying on the word segmentation algorithm. For example, "经济困难" and "最低生活保障", under the word segmentation algorithm, they are usually segmented into "经济", "困难" and "最低", "生活", "保障". Such fragmented words cannot refine clear user attribute information. And through similarity calculation and loop regular matching, the method of extracting the longest repeated corpus in the business item names can retain the smallest user attribute unit.

[0157] The method provided in the above embodiment extracts the longest repeated corpus in the business item name through similarity calculation and cyclic regular matching, which can retain the smallest user attribute unit and ensure the clarity, accuracy and effectiveness of describing user attributes.

[0158] See also Figure 4 , shows a flow chart of another optional service processing method according to an embodiment of the present invention, including the following steps:

[0159] S401: Filtering a first business item name including a preset business verb based on a business verb library, and determining a first user tag for the first business item name;

[0160] S402: Filtering all business item names corresponding to the first user tag from the set of business item names corresponding to the target user to determine the second business item name with the latest processing time;

[0161] S403: In response to the second business item name containing the preset business verb, determining that the first user tag is a non-taggable tag.

[0162] In the above implementation, after obtaining user tags for each business item name in steps S401 to S403, some tags cannot be directly used to create a user profile because the verbs in the business item name play a decisive role in determining whether the tag is suitable. Furthermore, tag processing is generally performed on a t+1 basis, meaning it is calculated once a day, so computational efficiency is not a problem.

[0163] For example, based on the user attribute labels obtained from the above operations, the user behavior is as follows:

[0164] Table 3

[0165] Business item name Processing time Handler User Attributes Tags Individual business registration 2020 / 1 / 1 Zhang San Individual business owners Registration of change in organizational form of individual industrial and commercial households 2020 / 3 / 12 Zhang San Individual business owners Change of address of individual business owners 2021 / 1 / 1 Zhang San Individual business owners Individual business registration and cancellation 2022 / 1 / 22 Zhang San Individual business owners

[0166] If the labeling is done before January 22, 2022, then the user attribute label "individual business owner" is accurate. However, if the labeling is done after January 22, 2022, it is inaccurate because the user has cancelled their individual business owner status. Therefore, it is necessary to use the government-specific verb library constructed above and combine it with user behavior to determine whether the label should be applied. For example, see Table 4:

[0167] Table 4

[0168]

[0169] First, search the government verb database to determine whether each business item name contains verbs that imply termination or doubt, such as "termination," "cancellation," "preliminary review," "inquiry," "verification," and "review." If so, the label status is set to "unprintable." If the business item name "Registration and Cancellation of Individual Business Owners" contains the preset business verb "cancellation," its label status is changed to "unprintable."

[0170] Determine the label corresponding to the business item name "Individual Business Owner Registration and Cancellation" with a label status of "Unprintable." In this case, it is "Individual Business Owner." Then, filter out all business item names containing the "Individual Business Owner" label from the business item names corresponding to the handler. Here, assume that these are the four business item names in Table 4. These four business item names are all labeled "Individual Business Owner." However, after the individual business owner cancels their business, this label no longer applies to the user because, after canceling their business, they are no longer an individual business owner.

[0171] For these four business items, sort them in chronological order, determine the name of the target business item with the latest time, and judge whether the target business item name is registration, change or cancellation. If it contains a preset business verb such as "cancellation", it is determined that the label "individual business household" cannot be given to the user, that is, these four records cannot be labeled "individual business household", so the label "individual business household" cannot be stored as a user attribute in the user portrait. However, if there is no processing of the 2022-01-22 item, it can be labeled, or if the 2021-01-01 record is cancellation, the 2022-01-22 label can be labeled, then the user as a whole can still be labeled as an individual business household, so it is necessary to pull out the basis and processing time of these labels.

[0172] The method provided in the above embodiment summarizes the verbs related to government affairs based on the naming characteristics of government affairs names through word frequency analysis and business experience accumulation, so as to establish a verb library dedicated to government affairs, which is used to filter out invalid or expired labels in the label confirmation link, ensuring that the labels finally given to users are accurate and timely.

[0173] See also Figure 5 , which shows a schematic diagram of the main modules of a service processing device 500 provided by an embodiment of the present invention, including:

[0174] 1. An acquisition module 501 is configured to acquire a set of business item names corresponding to an identifier of a target user in response to a selection operation on the target user;

[0175] And by presetting the unusual information table, remove the unusual information in each business item name; and perform deduplication processing on the business item names after removing the unusual information;

[0176] Each business item name is segmented, and stop words are removed from the segmented words based on a preset stop word library; stop words are words that are irrelevant to user attributes; for the remaining words, the frequency of the words in the preset position is counted, and a preset number of words with the highest frequency are selected in descending order to construct a business verb library;

[0177] 2. A calculation module 502 is configured to calculate the similarity between each business item name and other business item names in the business item name set, screen similar business item names whose similarity exceeds a preset similarity threshold, and calculate the Levenshtein distance between each business item name and the other business item names; separately determine the string length of each business item name and the string length of the other business item names to accumulate and obtain a sum of the string lengths; calculate the difference between the sum of the string lengths and the Levenshtein distance, and use the ratio of the difference to the sum of the string lengths as the similarity between each business item name and the other business item names;

[0178] 3. An extraction module 503 is configured to extract overlapping corpora of each business item name and similar business item names, split and remove duplicates from the overlapping corpora to obtain target corpora, and use the target corpora as user tags for each business item name;

[0179] And based on the business verb library, the business verbs are removed from the overlapping expectations to obtain updated overlapping expectations.

[0180] Specifically: determine the first character that ranks first in each business item name, in response to retrieving the first character in the similar business item name, continue to determine the second character that ranks second in each business item name; in response to the second character that is located after the first character in the similar business item name and is adjacent to the second character, continue to determine the third character that ranks third in each business item name, repeat the above matching operation until the match fails, and arrange the matched characters in order to obtain a first overlapping corpus; respectively determine the first remaining information of each business item name after removing the first overlapping corpus, and the second remaining information of the similar business item name after removing the first overlapping corpus, Determine the fourth character that ranks first in the first remaining information; in response to retrieving the fourth character in the second remaining information, continue to determine the fifth character that ranks second in the first remaining information, and repeat the above matching operation until the matching fails, and arrange the matched characters in order to obtain a second overlapping corpus; continue to determine the remaining information after removing the second overlapping corpus from the first remaining information and the second remaining information respectively, and repeat the above matching operation until the characters in each business item name are matched, to obtain multiple repeated corpora; use the longest overlapping corpus among the multiple repeated corpora as the target overlapping corpus for each business item name and the similar business item name.

[0181] And if there are multiple overlapping corpora with the longest length, multiple overlapping corpora with the longest length are retained.

[0182] And determine the overlapping part and non-overlapping part of each overlapping corpus and each other overlapping corpus, and in response to the overlapping part being not empty, use the corpus of the overlapping part and the non-overlapping part as the first corpus to be screened; determine the overlapping part and non-overlapping part of each first corpus to be screened and each other first corpus to be screened, and in response to the overlapping part being not empty, continue to use the corpus of the overlapping part and the non-overlapping part as the second corpus to be screened; determine the overlapping part and non-overlapping part of each second corpus to be screened and each other second corpus to be screened, and in response to the overlapping part being empty, use the corpus of the non-overlapping part as the target corpus.

[0183] 4. Processing module 504 is used to count the user tags of all business item names and perform corresponding business processing operations based on the obtained user tags.

[0184] And based on the business verb library, the first business item name including the preset business verb is filtered to determine the first user tag of the first business item name; from the business item name set corresponding to the target user, all business item names corresponding to the first user tag are filtered out to determine the second business item name with the latest processing time; in response to the second business item name containing the preset business verb, the first user tag is determined to be an unmarkable tag.

[0185] In addition, the specific implementation content of the device in the embodiment of the present invention has been described in detail in the above method, so the repeated content will not be described again here.

[0186] Figure 6 An exemplary system architecture 600 to which embodiments of the present invention may be applied is shown, including terminal devices 601 , 602 , 603 , a network 604 and a server 605 (only an example).

[0187] Terminal devices 601, 602, and 603 can be various electronic devices with display screens and support web browsing, and are installed with various communication client applications. Users can use terminal devices 601, 602, and 603 to interact with server 605 through network 604 to receive or send messages, etc.

[0188] The network 604 is used to provide a medium for communication links between the terminal devices 601, 602, 603 and the server 605. The network 604 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0189] The server 605 may be a server that provides various services. It should be noted that the method provided in the embodiment of the present invention is generally executed by the server 605 , and accordingly, the device is generally set in the server 605 .

[0190] It should be understood that Figure 6 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0191] Reference below Figure 7 , which shows a schematic structural diagram of a computer system 700 of a terminal device suitable for implementing an embodiment of the present invention. Figure 7 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0192] like Figure 7As shown, the computer system 700 includes a central processing unit (CPU) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage unit 708 into a random access memory (RAM) 703. Various programs and data required for the operation of the system 700 are also stored in the RAM 703. The CPU 701, ROM 702, and RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0193] The following components are connected to the I / O interface 705: an input section 706 including a keyboard, a mouse, and the like; an output section 707 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 708 including a hard disk; and a communication section 709 including a network interface card such as a LAN card or a modem. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I / O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 710 as needed, so that computer programs read therefrom can be installed into the storage section 708 as needed.

[0194] In particular, according to the embodiments disclosed in the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising 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 via the communication section 709, and / or installed from a removable medium 711. When the computer program is executed by the central processing unit (CPU) 701, the above-mentioned functions defined in the system of the present invention are executed.

[0195] It should be noted that the computer-readable medium described in the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media can include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wireline, optical fiber cable, RF, or any suitable combination thereof.

[0196] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the above-mentioned module, program segment, or a part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0197] The modules described in the embodiments of the present invention may be implemented in software or hardware. The modules described may also be provided in a processor. For example, they may be described as comprising an acquisition module, a calculation module, an extraction module, and a processing module. The names of these modules do not, in some cases, limit the modules themselves. For example, a calculation module may also be described as a "similarity calculation module."

[0198] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiments, or may exist independently and not incorporated into the device. The computer-readable medium carries one or more programs, and when executed by the device, the device executes any of the above-described service processing methods.

[0199] The computer program product of the present invention includes a computer program, and when the computer program is executed by a processor, the service processing method in the embodiment of the present invention is implemented.

[0200] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A business processing method, characterized in that: include: In response to a selection operation on a target user, obtaining a set of business item names corresponding to an identifier of the target user; Calculating the similarity between each business item name and other business item names in the business item name set, and screening similar business item names whose similarity exceeds a preset similarity threshold; Extracting overlapping corpus of each business item name and the similar business item name, including: determining the first character ranked first in each business item name, in response to retrieving the first character in the similar business item name, continuing to determine the second character ranked second in each business item name; in response to the second character being located after the first character in the similar business item name and adjacent to the second character, continuing to determine the third character ranked third in each business item name, repeating the above matching operation until the matching fails, and arranging the matched characters in order to obtain the first overlapping corpus; respectively determining the first remaining information of each business item name after removing the first overlapping corpus, and the similar business item name after removing the first remaining information. the first overlapping corpus is determined according to the second remaining information after the first overlapping corpus is obtained; in response to the fourth character being retrieved in the second remaining information, the fifth character being ranked second in the first remaining information is further determined, and the above matching operation is repeated until the matching fails, and the matched characters are arranged in order to obtain a second overlapping corpus; the remaining information after the second overlapping corpus is removed from the first remaining information and the second remaining information is further determined, and the above matching operation is repeated until the characters in each business item name are matched, thereby obtaining multiple repeated corpora; the longest overlapping corpus among the multiple repeated corpora is used as the target overlapping corpus for each business item name and the similar business item name; The overlapping corpus is split and de-duplicated to obtain a target corpus, including: determining the overlapping part and the non-overlapping part of each overlapping corpus and each other overlapping corpus, and in response to the overlapping part being not empty, taking the corpus of the overlapping part and the non-overlapping part as the first corpus to be screened; determining the overlapping part and the non-overlapping part of each first corpus to be screened and each other first corpus to be screened, and in response to the overlapping part being not empty, continuing to take the corpus of the overlapping part and the non-overlapping part as the second corpus to be screened; determining the overlapping part and the non-overlapping part of each second corpus to be screened and each other second corpus to be screened, and in response to the overlapping part being empty, taking the corpus of the non-overlapping part as the target corpus; Using the target corpus as a user label for each business item name; Count the user tags of all business item names and perform corresponding business processing operations based on the obtained user tags.

2. The method according to claim 1, characterized in that After obtaining the business item name set corresponding to the identifier of the target user, the method further includes: Remove unusual information from each business item name by presetting an unusual information table; and De-duplicate the business item names after removing non-routine information.

3. The method according to claim 1 or 2, characterized in that Also includes: Perform word segmentation on each business item name and remove stop words from the segmentation based on a preset stop word library. Stop words are words that are irrelevant to user attributes. For the remaining words, the word frequencies of the words at the preset positions are counted, and a preset number of words ranked at the top are selected in descending order of word frequencies to construct a business verb library; After extracting the overlapping corpus of each business item name and the similar business item name, the method further includes: removing the business verbs from the overlapping expectations based on the business verb library to obtain updated overlapping expectations.

4. The method according to claim 3, characterized in that Before performing corresponding business processing operations based on the obtained user tag, the method further includes: Based on the business verb library, screening a first business item name including a preset business verb, and determining a first user tag for the first business item name; Filtering all business item names corresponding to the first user tag from the set of business item names corresponding to the target user to determine the second business item name with the latest processing time; In response to the second business item name containing the preset business verb, the first user tag is determined to be a non-taggable tag.

5. The method according to claim 1, characterized in that Calculating the similarity between each business item name and other business item names in the business item name set includes: Calculate the Levenshtein distance between each business item name and other business item names; Determine the character string length of each business item name and the character string length of the other business item names respectively, and add them up to obtain a sum of the character string lengths; The difference between the sum of the string lengths and the Levenshtein distance is calculated, and the ratio of the difference to the sum of the string lengths is used as the similarity between each business item name and the other business item names.

6. The method according to claim 1, characterized in that The method of using the longest overlapping corpus among the multiple repeated corpora as the target overlapping corpus for each business item name and the similar business item name includes: If there are multiple overlapping corpora with the longest length, multiple overlapping corpora with the longest length are retained.

7. A business processing device, characterized in that: include: An acquisition module, configured to acquire, in response to a selection operation on a target user, a set of business item names corresponding to an identifier of the target user; A calculation module, configured to calculate the similarity between each business item name and other business item names in the business item name set, and filter similar business item names whose similarity exceeds a preset similarity threshold; An extraction module is used to extract overlapping corpora of each business item name and the similar business item names, including: determining a first character ranked first in each business item name, in response to retrieving the first character in the similar business item name, continuing to determine a second character ranked second in each business item name; in response to the second character being located after the first character in the similar business item name and adjacent to the first character, continuing to determine a third character ranked third in each business item name, repeating the above matching operation until the match fails, and arranging the matched characters in order to obtain a first overlapping corpus; respectively determining a first remaining information after removing the first overlapping corpus from each business item name and a second remaining information after removing the first overlapping corpus from the similar business item name, determining a fourth character ranked first in the first remaining information; in response to retrieving the fourth character in the second remaining information, continuing to determine a fifth character ranked second in the first remaining information, and repeating the above matching operation until the match fails, and arranging the matched characters in order to obtain a second overlapping corpus; Continue to determine the remaining information after removing the second overlapping corpus from the first remaining information and the second remaining information, and repeat the above matching operation until the characters in each business item name are matched, thereby obtaining multiple repeated corpora; use the longest overlapping corpus among the multiple repeated corpora as the target overlapping corpus for each business item name and the similar business item name; perform deduplication processing on the overlapping corpora to obtain the target corpus, including: determining the overlapping part and non-overlapping part of each overlapping corpus and each other overlapping corpus, and responding to the overlapping corpora. If the overlapping part is not empty, the corpus of the overlapping part and the non-overlapping part is used as the first corpus to be screened; the overlapping part and the non-overlapping part of each first corpus to be screened and each other first corpus to be screened are determined, and in response to the overlapping part being not empty, the corpus of the overlapping part and the non-overlapping part is continued to be used as the second corpus to be screened; the overlapping part and the non-overlapping part of each second corpus to be screened and each other second corpus to be screened are determined, and in response to the overlapping part being empty, the corpus of the non-overlapping part is used as the target corpus; the target corpus is used as the user label of each business item name; The processing module is used to count the user tags of all business item names and perform corresponding business processing operations based on the obtained user tags.

8. The device according to claim 7, characterized in that The acquisition module is further used to: Perform word segmentation on each business item name and remove stop words from the segmentation based on a preset stop word library. Stop words are words that are irrelevant to user attributes. For the remaining words, the word frequencies of the words at the preset positions are counted, and a preset number of words ranked at the top are selected in descending order of word frequencies to construct a business verb library; The extraction module is further used to: based on the business verb library, remove the business verbs from the overlapping expectations to obtain updated overlapping expectations.

9. The device according to claim 8, characterized in that The processing module is further configured to: Based on the business verb library, screening a first business item name including a preset business verb, and determining a first user tag for the first business item name; Filtering all business item names corresponding to the first user tag from the set of business item names corresponding to the target user to determine the second business item name with the latest processing time; In response to the second business item name containing the preset business verb, the first user tag is determined to be a non-taggable tag.

10. The device according to claim 7, characterized in that The computing module is configured to: Calculate the Levenshtein distance between each business item name and other business item names; Determine the character string length of each business item name and the character string length of the other business item names respectively, and add them up to obtain a sum of the character string lengths; The difference between the sum of the string lengths and the Levenshtein distance is calculated, and the ratio of the difference to the sum of the string lengths is used as the similarity between each business item name and the other business item names.

11. An electronic device, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 6.

12. A computer-readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

13. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images