CN117333191A

CN117333191A - Complaint event association method and device and complaint event association system

Info

Publication number: CN117333191A
Application number: CN202311266639.XA
Authority: CN
Inventors: 李亚梦; 梁世栋; 季元; 廖大春; 傅鹏佳
Original assignee: Postal Savings Bank of China Ltd
Current assignee: Postal Savings Bank of China Ltd
Priority date: 2023-09-27
Filing date: 2023-09-27
Publication date: 2024-01-02

Abstract

The application provides a complaint event association method, a complaint event association device and a complaint event association system. The method comprises the following steps: acquiring worksheet data; constructing a correlation diagram according to the complaint questions and the investigation results; and under the condition that the target complaint problem is received, searching the association complaint problem and the association investigation result in the association diagram. Through the constructed association diagram, the association relation between the complaint questions can be determined, and the association relation between the complaint questions and the investigation result can also be determined, so that when new complaint questions are fed back, the associated complaint questions and investigation result can be excavated according to the new complaint questions, further analysis can be assisted by complaint treatment personnel, and the investigation efficiency of the complaint questions is improved.

Description

Complaint event association method and device and complaint event association system

Technical Field

The present application relates to the technical field of complaint management, and in particular, to a method and apparatus for associating a complaint event, a computer readable storage medium, and a system for associating a complaint event.

Background

Complaints are an important channel for reflecting consumer demands, and are an important channel for guaranteeing consumer benefits and improving customer satisfaction, and management, analysis and solution of complaint problems are well done. The complaint scene has the characteristics of diversity of consumer appeal, non-standard description mode, non-uniform solution, diversity of clients, diversity of business and the like, and is important for analysis and treatment of complaint problems.

In the current complaint problem analysis and treatment process, insufficient finding is carried out on the complaint problem, so that the investigation efficiency of the complaint problem is lower.

Disclosure of Invention

The main objective of the present application is to provide a method, an apparatus, a computer readable storage medium and a system for associating complaint events, so as to at least solve the problem of low efficiency of complaint problem investigation in the prior art.

To achieve the above object, according to one aspect of the present application, there is provided a method for associating complaint events, including: acquiring work order data, wherein the work order data at least comprises a complaint problem and a survey result, the complaint problem is a problem of a user on a complaint event, and the survey result is a generation reason of the user on the complaint event; constructing an association graph according to the complaint questions and the investigation results, wherein nodes in the association graph are the complaint questions and the investigation results, and edges in the association graph are association relations between any two complaint questions or between any one complaint question and any one investigation result; under the condition that a target complaint problem is received, searching an associated complaint problem and an associated investigation result in the associated graph, and displaying the associated complaint problem and the associated investigation result, wherein the associated complaint problem is the complaint problem connected with the target complaint problem, and the associated investigation result is the investigation result connected with the target complaint problem.

Optionally, constructing a correlation graph according to the complaint problem and the investigation result includes: calculating a first association strength between any two complaint problems according to a first formula, wherein the first formula is as follows:

W _i,j representing the first correlation strength, p, between complaint problem i and complaint problem j _i,j Indicating whether complaint i and complaint j occur in the work order p at the same time, N indicating the work order set, N indicating the number of work orders, q _i,j Representing whether complaint problem i and complaint problem j occur simultaneously in complaints of the same user q, M representing a set of complaint users, M representing the number of complaint users, a ₁ Representing a first weight, a ₂ Representing a second weight; connecting a complaint problem i and a complaint problem j according to the first association strength; calculating a second association strength between any one of the complaint problems and any one of the investigation results according to a second formula, wherein the second formula is as follows:

X _i,j representing the second correlation strength between complaint problem i and survey result j, b _i,j Whether the complaint problem i and the investigation result j occur in the work order q at the same time or not is indicated, K indicates a complaint work order set, and K indicates the number of complaint work orders; and connecting the complaint problem i with the investigation result j according to the second association strength.

Optionally, after acquiring the worksheet data, the method further comprises: word segmentation is carried out on the worksheet data to obtain first worksheet data; performing stop word removal processing on the first work order data to obtain second work order data; determining whether the second work order data comprises a filtered vocabulary, wherein the filtered vocabulary is a vocabulary in a pre-configured dictionary to be filtered; and deleting the filtered vocabulary from the second work order data under the condition that the second work order data comprises the filtered vocabulary, so as to obtain target work order data.

Optionally, after acquiring the worksheet data, the method further comprises: calculating the TF-IDF value of each word in the work order data according to a TF-IDF algorithm; and selecting a first preset number of words from the target sentences according to the sequence of the TF-IDF values from large to small to obtain keywords, wherein the keywords are used for constructing the association graph.

Optionally, the worksheet data further includes user information, and after constructing a correlation map according to the complaint questions and the investigation result, the method further includes: constructing an auxiliary investigation model, wherein the auxiliary investigation model is trained by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises historical user information acquired in a historical time period, historical complaint problems corresponding to the historical user information and historical investigation results corresponding to the historical user information; and inputting the current user information into the auxiliary investigation model to obtain the current complaint problem and the current investigation result corresponding to the current user information.

Optionally, before building the auxiliary survey model, the method further comprises: at the history complaint problem v ₁ And the history survey result v ₂ Extracting the historical complaint problem v under the condition of edges ₁ First history information of all relevant history worksheets and extracting the history investigation result v ₂ Obtaining the history user information by the second history information of all the related history worksheets; at the history complaint problem v ₁ And the history survey result v ₂ Extracting the history complaint problem v without edges ₁ And obtaining the historical user information according to the first historical information of any related historical worksheet, wherein the historical user information is used for model training.

Optionally, after constructing a correlation graph according to the complaint problem and the investigation result, the method further includes: clustering the complaint problems to obtain at least one cluster; and selecting a second preset number of complaint problems from the cluster according to the order of the centrality from high to low to obtain a hot spot complaint problem, wherein the centrality is the distance between the average cluster of the data points and the cluster center.

According to another aspect of the present application, there is provided an apparatus for associating complaint events, comprising: the system comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring worksheet data, the worksheet data at least comprise complaint questions and investigation results, the complaint questions are questions of a user for a complaint event, and the investigation results are the generation reasons of the user for the complaint event; the first construction unit is used for constructing a correlation graph according to the complaint questions and the investigation results, wherein nodes in the correlation graph are the complaint questions and the investigation results, edges in the correlation graph are the correlation between any two complaint questions, or edges in the correlation graph are the correlation between any one complaint question and any one investigation result; the first processing unit is used for searching the associated complaint questions and the associated investigation results in the associated graph under the condition that the target complaint questions are received, and displaying the associated complaint questions and the associated investigation results, wherein the associated complaint questions are the complaint questions connected with the target complaint questions, and the associated investigation results are the investigation results connected with the target complaint questions.

According to still another aspect of the present application, there is provided a computer readable storage medium, where the computer readable storage medium includes a stored program, and when the program runs, the program controls a device in which the computer readable storage medium is located to execute any one of the association methods of complaint events.

According to yet another aspect of the present application, there is provided a complaint event correlation system comprising: one or more processors, memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including an associated method for performing any one of the complaint events.

By applying the technical scheme, the association relation between the complaint questions can be determined through the constructed association diagram, and the association relation between the complaint questions and the investigation result can also be determined, so that when new complaint questions are fed back, the associated complaint questions and investigation result can be mined according to the new complaint questions, deeper analysis can be assisted by complaint treatment personnel, and the investigation efficiency of the complaint questions is improved.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application. In the drawings:

FIG. 1 shows a block diagram of a hardware architecture of a mobile terminal that performs a method of association of complaint events provided in an embodiment of the present application;

FIG. 2 illustrates a flow diagram of a method of association of complaint events provided in accordance with an embodiment of the present application;

FIG. 3 shows a schematic diagram of the overall solution of the present application;

FIG. 4 shows a schematic diagram of a relationship diagram constructed in the present application;

FIG. 5 shows a schematic flow chart of work order data preprocessing of the present application;

FIG. 6 shows a flow chart of investigation of the cause of complaints in some scenarios;

FIG. 7 shows a flow chart of the auxiliary survey model online prediction of the present application;

FIG. 8 shows a schematic diagram of the association graph and auxiliary survey model combination of the present application;

fig. 9 shows a block diagram of a device for associating complaint events according to an embodiment of the present application.

Wherein the above figures include the following reference numerals:

102. a processor; 104. a memory; 106. a transmission device; 108. and an input/output device.

Detailed Description

It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.

In order to make the present application solution better understood by those skilled in the art, the following description will be made in detail and with reference to the accompanying drawings in the embodiments of the present application, it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.

It should be noted that the terms "first," "second," and the like in the description and claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe the embodiments of the present application described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

As described in the background art, the efficiency of complaint problem investigation in the prior art is low, and in order to solve the above problems, embodiments of the present application provide a method, an apparatus, a computer-readable storage medium, and a system for associating complaint events.

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

The method embodiments provided in the embodiments of the present application may be performed in a mobile terminal, a computer terminal or similar computing device. Taking the operation on the mobile terminal as an example, fig. 1 is a block diagram of a hardware structure of the mobile terminal of a method for associating complaint events according to an embodiment of the present invention. As shown in fig. 1, a mobile terminal may include one or more (only one is shown in fig. 1) processors 102 (the processor 102 may include, but is not limited to, a microprocessor MCU or a processing device such as a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 for communication functions and an input-output device 108. It will be appreciated by those skilled in the art that the structure shown in fig. 1 is merely illustrative and not limiting of the structure of the mobile terminal described above. For example, the mobile terminal may also include more or fewer components than shown in fig. 1, or have a different configuration than shown in fig. 1.

The memory 104 may be used to store a computer program, for example, a software program of application software and a module, such as a computer program corresponding to a display method of device information in an embodiment of the present invention, and the processor 102 executes the computer program stored in the memory 104 to perform various functional applications and data processing, that is, to implement the above-described method. Memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory remotely located relative to the processor 102, which may be connected to the mobile terminal via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof. The transmission device 106 is used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, simply referred to as NIC) that can connect to other network devices through a base station to communicate with the internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module, which is configured to communicate with the internet wirelessly.

In this embodiment, a method of associating complaints with events running on a mobile terminal, a computer terminal, or similar computing device is provided, and it is to be noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer-executable instructions, and that although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from that shown or described herein.

FIG. 2 is a flow chart of a method of associating complaint events according to an embodiment of the present application. As shown in fig. 2, the method comprises the steps of:

step S201, acquiring worksheet data, wherein the worksheet data at least comprise complaint questions and investigation results, the complaint questions are questions of a user for a complaint event, and the investigation results are generation reasons of the user for the complaint event;

specifically, as shown in fig. 3, in the data preparation process, the first step is the process of data acquisition, and complaint work order data in a period of time can be acquired, and the fields are shown in table 1.

TABLE 1

Sequence number	Fields
		1	Work order id
2	Sex (sex)
		3	Age of
4	Occupation of
		5	Residence ground
6	Name of name
		7	Complaint content
8	Investigation result
		…………	…………

Step S202, constructing a correlation graph according to the complaint questions and the investigation results, wherein nodes in the correlation graph are the complaint questions and the investigation results, edges in the correlation graph are the correlation between any two of the complaint questions, or edges in the correlation graph are the correlation between any one of the complaint questions and any one of the investigation results;

in some approaches, it is difficult to find relationships between different complaint problems, resulting in incomplete analysis of the complaint problems. In the case of customer appeal, there may be an association between different appeal, such as customer desire to exempt interest, possibly because the account is suspended from non-counter transaction presentation resulting in untimely repayment, and there is a causal relationship between "exempt interest" and "suspended non-counter transaction".

In some approaches, complaint problems are mostly obtained by clustering or classification. Classification approaches are not good at discovering new problems beyond those known and require more manual labeling effort. Although a part of new problems can be found by the clustering method, whether the relation exists between different problems can not be judged. In the related technology of the association relationship, the association relationship between similar descriptions is mostly constructed according to word senses, for example, the association relationship between the similar descriptions can be correctly expressed, such as 'complaints' and 'complaints' are similar, when the literal meanings of complaint problems are not similar, but when the internal association exists, such as 'reduced interest' and 'suspended non-counter exchange', the association relationship cannot be expressed.

Specifically, in the scheme, an association graph of the complaint event can be constructed, association relations among the complaint problems can be analyzed, association relations among the complaint problems of users and investigation results can be found, the association problems of a certain complaint problem can be inquired and mined, and further analysis is carried out by complaint treatment personnel.

Step S203, when a target complaint problem is received, searching for a related complaint problem and a related investigation result in the related graph, and displaying the related complaint problem and the related investigation result, wherein the related complaint problem is the complaint problem connected to the target complaint problem, and the related investigation result is the investigation result connected to the target complaint problem.

Specifically, in the scheme, association analysis of the complaint problems can be realized, association relation among the complaint problems can be analyzed, the association problem of a certain complaint problem can be queried, and investigation results of association corresponding to the complaint problem can be queried to assist complaint processors in deeper analysis.

Through the embodiment, through the constructed association diagram, the association relation between the complaint questions can be determined, and the association relation between the complaint questions and the investigation result can also be determined, so that when new complaint questions are fed back, the associated complaint questions and investigation result can be excavated according to the new complaint questions, deeper analysis can be assisted by complaint treatment personnel, and further the investigation efficiency of the complaint questions is improved.

Specifically, as shown in fig. 3, the whole flow of the scheme takes work order information as input, and extracts complaint questions and investigation results after preprocessing, and constructs a complaint event diagram by taking the complaint questions and the investigation results as nodes.

In the specific implementation process, the association diagram is constructed according to the complaint problem and the investigation result, and the specific implementation process can be realized by the following steps: calculating a first association strength between any two complaints according to a first formula, wherein the first formula is as follows:

X _i,j Representing the second correlation strength between complaint i and survey j, b _i,j Whether the complaint problem i and the investigation result j occur in the work order q at the same time or not is indicated, K indicates a complaint work order set, and K indicates the number of complaint work orders; and connecting the complaint question i with the investigation result j according to the second association strength. The above formulas are exemplary only, and variations of any formulas fall within the scope of the present application.

In the scheme, the first association strength between any two complaint questions can be determined through the first formula, the complaint questions can be connected through the first association strength, the second association strength between any one complaint question and any one investigation result can be determined through the second formula, and the complaint questions and the investigation result can be connected through the second association migration, so that the association diagram can be obtained by connecting according to edges and nodes, and therefore the association diagram can be constructed accurately and efficiently through the scheme of the embodiment.

Specifically, as shown in fig. 3, a complaint event graph (association graph) shows the relationship between complaint questions, between complaint questions and investigation results in the form of a graph, with the complaint questions and investigation results as nodes, and the relationship between complaint questions and the complaint-investigation relationship as edges. The method comprises the following steps:

(1) Node computation

Nodes include two classes: complaint questions and findings. Obtained by preprocessing and keyword extraction based on the complaint content and the investigation result in the work order respectively.

(2) Edge computation

Edges are divided into two types, edges between complaint questions and investigation results.

The strength of the side association relationship between the complaint questions is determined by the work orders and clients commonly associated with the two 'complaint questions'.

P in the first formula _i,j Take the value of {0,1}, q _i,j The value is {0,1}.

The magnitude of the side association relationship between the complaint question and the investigation result is determined by the work order number which is commonly associated with the complaint question and the investigation result.

B in the second formula _i,j The value is {0,1}.

Specifically, the constructed association diagram is shown in fig. 4.

(3) Associated problem mining

Based on the complaint event diagram constructed above, the related problem of a certain problem can be conveniently mined through inquiry.

The scheme of the application defines a complaint event association diagram and a calculating method thereof. The complaint problems and the investigation results are used as nodes, the strength of the edges is defined through the complaint behaviors of the users, and compared with the similar strength relationship of literal meaning constructed based on word vector similarity, the characteristics of the complaint behaviors of the users can be expressed. And (3) carrying out associated problem searching based on the graph, and inquiring the associated problem of a certain complaint problem to assist complaint treatment personnel in carrying out deeper analysis.

In order to understand and analyze the work order data more efficiently, after acquiring the work order data, the above method further includes the steps of: word segmentation processing is carried out on the work order data to obtain first work order data; performing stop word removal processing on the first work order data to obtain second work order data; determining whether the second work order data comprises a filtering vocabulary, wherein the filtering vocabulary is a vocabulary in a pre-configured dictionary to be filtered; and deleting the filtering vocabulary from the second work order data to obtain target work order data when the second work order data comprises the filtering vocabulary.

In the scheme, the work order text can be split into individual words through word segmentation, key information and topics are extracted from the words, the content and the problem of the work order can be better understood, key words can be more accurately identified, the word segmentation is carried out on the work order text and then the work order text can be expressed as a word set, so that the dimension of data is reduced, the subsequent data analysis and modeling are facilitated, the processing efficiency and accuracy are improved, words which are common but have no practical meaning can be removed through word segmentation and word filtering removal, such as prepositions, conjunctions and the like, interference can be reduced, more valuable key words are extracted, the work order content can be better understood and analyzed, word segmentation processing is carried out on the work order data, word filtering removal and word filtering removal can be improved, meaningful key words are extracted, the dimension of data is reduced, and irrelevant information is removed.

Specifically, as shown in fig. 3 for data preprocessing, preprocessing such as word segmentation, word deactivation, part-of-speech tagging and the like is performed on "complaint content", and the data is processed into individual words, while the word segmentation can be adjusted and specific words can be reserved by using a user dictionary. If "i am contacting customer service repayment in advance, and as a result, you say that you cannot handle the present month, and do not relieve the interest of the next month, you are required to solve" the result after the treatment is "contact/customer service/repayment in advance/result/present month/unable to handle/next month/handle/multi-receipts/interest/unable to reduce/require/solve".

Word segmentation: a sentence is divided into a list of words.

User dictionary: the user-defined dictionary can ensure that word segmentation in the dictionary is correct, and is not filtered out because of part of speech. If the "advance repayment" before configuration is divided into "advance/repayment", the words can be correctly divided after configuration.

Stop words: "one's" "one kind of" don't care "and the like are nonsensical or don't care words.

Part-of-speech filtering: only required parts of speech, such as only verbs and nouns, can be reserved according to business requirements and part of speech tagging results.

Specifically, as shown in fig. 5, the flow of the data preprocessing includes the following steps:

Step (1), acquiring a content text of the job ticket complaints;

step (2), word segmentation is carried out, and words in a user dictionary are not segmented;

step (3), marking parts of speech;

step (4), defining a word segmentation list A, and initializing to be empty;

the step (5) is as follows:

step a, whether the word is a user dictionary, if yes, adding the word into a word segmentation post list A, ending the judgment, starting the judgment of the next word, and if no, turning to b, and carrying out subsequent judgment;

b, judging whether the word is a stop word or not, if yes, judging the next word, and if no, judging the next word;

step c, judging whether the part of speech is a word to be filtered or not, and turning to step 4) to judge the next word, and adding the word into the word segmentation list A if not.

In order to identify and extract the key information in the text, after acquiring the worksheet data, the method further comprises the following steps: calculating the TF-IDF value of each word in the work order data according to a TF-IDF algorithm; and selecting a first preset number of words from the target sentence according to the sequence of the TF-IDF values from large to small to obtain keywords, wherein the keywords are used for constructing the association graph.

In the scheme, words with higher importance in the text can be obtained by calculating the TF-IDF values of the words, and the words can reflect the theme or the characteristics of the text. The keywords obtained through screening can accurately reflect the theme or the characteristics of the text, the extraction effect of the key information is further improved, the words with smaller influence on the text theme can be filtered by excluding common words, the noise interference is further reduced, the understanding and the processing of the text content can be facilitated, and the effect of automatic processing is further improved.

TF-IDF (terminal-Inverse Document Frequency) is a method for assessing the importance or criticality of words to text. It combines the frequency of words in Text (TF) with the importance of words in the whole document set (IDF) so that keywords that frequently occur in text but are rare in the whole document set can be accurately found.

Specifically, for keyword extraction, the TF-IDF algorithm may be used for keyword extraction on a pre-processing basis. For example, "i am contacted customer service to pay in advance, and as a result, you say that you cannot handle the present month, and how much interest is paid in the next month cannot be exempted, you are required to solve the problem that" the result of drawing on the basis of 4.1.1 processing is "pay in advance interest exemption". For the list A corresponding to each sentence (obtained by preprocessing and calculating the previous section), the keyword extraction steps are as follows:

step one, firstly, calculating TF values;

secondly, calculating an IDF value;

thirdly, calculating TF-IDF values of each word;

thirdly, selecting three words with highest TF-IDF in the sentence, namely the keywords of the sentence.

The first preset number may be 3, 5, 10, or other numbers.

The keyword extraction may use other keyword extraction methods instead of TF-IDF, such as TextRank, LDA, deep learning-based methods, etc.

Specifically, the scheme mainly realizes the analysis of major complaint problems and assists in complaint investigation. The discovery of major complaints is based on a large number of user complaints, and the integration analysis is performed to extract important problems and analyze the related problems of a certain problem.

In some schemes, there is a problem that the investigation is time-consuming, and after receiving a complaint, a series of investigation and comparison are required to confirm the actual cause of the complaint problem of the user. As shown in fig. 6, the conventional complaint work order processing flow is: the agent receives the complaints, records the problems, surveys and verifies the problems or reports the leader, the operator solves the processing problems, and the agent returns to visit the user.

In some embodiments, the worksheet data further includes user information, and after constructing the association graph according to the complaint problem and the investigation result, the method further includes the steps of: constructing an auxiliary investigation model, wherein the auxiliary investigation model is trained by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises historical user information acquired in a historical time period, historical complaint problems corresponding to the historical user information and historical investigation results corresponding to the historical user information; and inputting the current user information into the auxiliary investigation model to obtain the current complaint problem and the current investigation result corresponding to the current user information.

In the scheme, the automatic auxiliary problem investigation can be realized, a plurality of possible investigation results and complaint problems can be automatically provided based on the information of the user and the complaint content after the complaint of the user is received, and the investigation difficulty of the problems is reduced.

Specifically, the scheme of the application can design an auxiliary investigation model, takes basic information of a user, complaint problems and investigation processes as input to complete model training, and can automatically provide a plurality of possible investigation results based on the information of the client and complaint contents after receiving the complaints in a prediction stage, so that the problem investigation difficulty is reduced.

Specifically, the auxiliary investigation is to automatically give several possible investigation results based on basic information and description of the user after the complaint of the user occurs, such as a job ticket recording client description "account number is not changed", and the investigation results are "account is paused for non-counter transaction", i.e. "account is frozen and paused for non-counter transaction".

The scheme designs a graph neural network model, and can automatically provide a plurality of possible investigation results after receiving complaints, so that the problem investigation difficulty is reduced. The scheme of the application is an algorithm model independent of specific tasks, can be suitable for problem discovery and analysis of complaint scenes of multiple consumers, can quickly discover problems, and improves service quality.

In some embodiments, the method further comprises the steps of, prior to constructing the auxiliary survey model: the above-mentioned history complaint problem v ₁ And the above history investigation result v ₂ Extracting the history complaint problem v under the condition of edges ₁ First history information of all relevant history worksheets and extracting and searching result v of the history ₂ Obtaining the history user information by the second history information of all the related history worksheets; the above-mentioned history complaint problem v ₁ And the above history investigation result v ₂ Extracting the history complaint problem v without edges ₁ And obtaining the historical user information according to the first historical information of any relevant historical work order, wherein the historical user information is used for model training.

In the scheme, the historical user information for training the model can be processed on the basis of the association diagram, whether an edge exists between two nodes in the association diagram or not can be determined, and the obtained historical user information can be determined, so that an auxiliary investigation model obtained through training can be ensured to be more accurate, and complaint problems and investigation results corresponding to the user information can be determined more accurately.

Specifically, as shown in fig. 3, on the basis of a complaint event map, an auxiliary investigation model can be trained, a map neural network model is built on the basis of the complaint event map, embedded representations of two types of nodes of a complaint problem and a investigation result on the map neural network model and user information are used as inputs, whether edges exist between the two nodes is used as output (1 when edges exist), and the auxiliary investigation model is trained; after model training is finished, when online prediction is finished, the worksheet is preprocessed, user information and complaint problem information mapped on the event map are extracted, all 'investigation result' nodes which are related to 'complaint problems' mapped by the worksheet on the event map are extracted, the nodes are input into an auxiliary investigation model, and the first M candidate investigation results can be obtained according to score sorting.

Specifically, as shown in fig. 7, a graph neural network model may be constructed based on a complaint event graph, embedded representations of two types of nodes of a "complaint problem" and a "investigation result" on the graph neural network model and embedded representations of user information are used as inputs, the user information includes gender, age, occupation, residence … … and the like of the user, the embedded manner is obtained by using a one-hot encoding manner, whether edges exist between the two nodes is used as output (1 when edges exist), an auxiliary investigation model is trained, after all candidate investigation results are obtained, the auxiliary investigation model may be scored, and the first M candidate investigation results are selected as recommendation reasons.

Specifically, the specific steps for combining the association graph and the auxiliary investigation model are as follows:

step I, using word vectors as initialization vectors of nodes,

step ii, for k=1, 2, … …, K: (K is generally between 2 and 4),

for each node:

the neighboring nodes of the aggregation node v,

normalization:

step III, complaint problem v ₁ And investigation result v ₂ When there is a side between the two types of nodes, the user information h in all worksheets related to the two types of nodes is taken as a positive sample, and y=1 (the output of the association diagram of fig. 8) _userv1,v2 And (3) calculating:

Step IV, complaint problem v ₁ And investigation result v ₂ When there is no edge between, y=0 (the correlation diagram output of fig. 8) is taken as a negative sample, and v is taken as ₁ User information h of any work order related to node _userv1 And (3) calculating:

the L training model is minimized according to the following loss function, and each parameter is obtained.

y'＝softmax(w ₂ *z+b ₂ )；

L＝-∑ylog(y')+(1-y)log(y')；

After model training is finished, when online prediction is finished, the worksheet is preprocessed, user information and complaint problem information mapped on the event map are extracted, all 'investigation result' nodes which are related to 'complaint problems' mapped by the worksheet on the event map are extracted, the nodes are input into an auxiliary investigation model, and the first M candidate investigation results can be obtained according to score sorting.

After the model has been trained, the model is set up,that is, the ebedding code of each node is used for other tasks to makeIs used.

In some schemes, it is difficult to find out important complaints, thereby affecting the overall control of the complaints. In the face of a large number of complaints each day, it is often difficult to quickly acquire the hot spot problem concerned by the user from the complaint problem, and after a period of fermentation of one kind of complaint problem is often needed, corresponding solving measures can be formulated to reduce the complaint of the same kind of problem, which causes the reduction of user experience and service quality.

In a specific implementation process, after constructing the association graph according to the complaint problem and the investigation result, the method further comprises the following steps: clustering the complaint problems to obtain at least one cluster; and selecting a second preset number of complaint problems from the cluster according to the order of the centrality from high to low to obtain a hot spot complaint problem, wherein the centrality is the distance between the average cluster of the data points and the cluster center.

In the scheme, the complaint problems can be subjected to cluster analysis, different clusters can represent the complaints of users of different groups, and the closer the cluster is to a central point, the more times of complaints are, so that the hot spot complaint problems can be selected based on the complaint times, and further, the overall situation can be effectively controlled according to the hot spot complaint problems.

Specifically, the problem of finding the 'hot spot complaint problem' can be automatically realized based on the complaint work order, and the complaint of different customer groups can be represented as far as possible.

Specifically, the design diagram segmentation algorithm can be adopted to cluster complaint problems, the complaint problems are clustered into a plurality of types, and the design rule finally outputs a hot complaint problem list, so that the work order key problem list is automatically obtained, and nodes with weak association relations are more easily divided into different clusters because the association is formed based on the number of customer complaints, so that different clusters can represent the complaints of customers of different groups.

Specifically, as shown in fig. 3, on the basis of a complaint event map, hot spot complaint problems are acquired, on the basis of the event map, complaint problem nodes are clustered, after the complaint problems are clustered into a plurality of categories, K nodes with highest centrality in each category are selected as key problems.

Specifically, node clustering can be performed on the association graph, and because the association strength between nodes on the graph is the number of times that the clients appear in the same work order at the same time, the nodes in the same sub-graph are expected to represent the appeal of the same batch of clients, so that when the graph segmentation algorithm is designed, the lower the association strength between different sub-graphs is required to be, the better the higher the association strength between all the nodes in the sub-graph is.

Defining a subgraph contact strength calculation method:

a _vi,vj for edge strength between nodes vi and vj, when two nodes belong to the same community (cluster), δ (C) _vi ,C _vj ) 1, otherwise 0.

The method comprises the following specific steps:

step (1), the number N of the clustered categories is set in advance, wherein N can be set to be slightly larger according to experience;

step (2), initializing N communities under the assumption that each node is 1 community, and initializing N communities by N nodes;

step (3), traversing each node one by one: finding joins enables overall link strength

The method comprises the steps of (1) lifting the largest community;

step (4), N clusters are obtained until the segmentation condition of the whole graph is not changed;

and (5) selecting M nodes with highest centrality in each category as key problems, and finally obtaining N.times.M key problems.

According to the scheme, a graph clustering mode is designed, the problem of hot spot complaints can be obtained, the complaint problems of users are gathered into a plurality of classes, the problem of hot spot complaints is determined by designing a selection mechanism on the basis of clustering, and because edges on the graph are calculated for customer complaint work orders, clustering based on the graph can represent the complaints of different customers to a certain extent.

In addition, the data used in the application adopts the existing data as much as possible, and manual labeling is avoided. According to the scheme, on the basis of solving the defect of the existing classification and clustering method on the complaint problem discovery, manual labeling is avoided, and the complaint problem investigation efficiency is improved.

In the above scheme of the application, an automatic method is designed, which can excavate hot spot complaint events and assist in complaint investigation, and has the following functions: firstly, the hot spot problem of the recent complaints can be automatically acquired, and the complaints of different customer groups can be represented as far as possible; secondly, the association relation between complaint problems can be analyzed, the association problem of a certain complaint problem can be queried, and the complaint processing personnel can be assisted to perform deeper analysis; thirdly, after receiving complaints, the suspicious automatically provides a plurality of possible investigation results, so that the problem investigation difficulty is reduced.

The scheme of the application provides a method for analyzing complaint problems and assisting investigation. The hot problem in the current complaint can be automatically found and the associated problem mining of a certain problem is supported; after receiving the complaints, the suspicious automatically provides a plurality of possible investigation results, thereby reducing the problem investigation difficulty. The main advantages include the following:

(1) The complaint event association diagram can be automatically constructed and updated based on the worksheet, and the worksheet complaint problems and investigation results are organized in the form of a diagram;

(2) Supporting analysis of association relation among complaint problems and inquiry of association problems of a certain problem, and assisting complaint treatment personnel in further analysis;

(3) The hot spot complaint content can be automatically mined, and the result covers more customer groups;

(4) The design diagram neural network model realizes auxiliary investigation, can effectively utilize personal information of users, complaint behaviors of users and various information of complaint behaviors of similar users, and solves the time-consuming difficulty of investigation of complaint problems;

(5) A great deal of labeling work is not needed to be manually conducted to train the hot spot classification model and the like.

In order to enable those skilled in the art to more clearly understand the technical solutions of the present application, the implementation process of the method for associating complaint events of the present application will be described in detail below with reference to specific embodiments.

The embodiment relates to a specific complaint event association method, and the implementation scene of the embodiment is that in telephone customer service complaints, the method of the scheme is utilized to automatically discover and analyze problems and accelerate the solution of the problems, and the method comprises the following steps:

step S1: acquiring a complaint work sheet of a history period of time, wherein the complaint work sheet comprises fields such as work sheet id, gender, age, occupation, residence, complaint content, investigation results and the like;

step S2: preprocessing complaint content and investigation results, extracting keywords, and extracting complaint problems and investigation results at the positions;

step S3: taking the complaint problems extracted in the second step and the investigation results as node construction diagrams;

step S4: training an auxiliary investigation model based on the constructed graph in the step three;

step S5: combining the graph in the third step and a graph segmentation algorithm to complete graph clustering and obtain a hot spot complaint problem list;

step S6: when the method is applied online, a work order is newly obtained, complaint questions corresponding to the work order, user information and candidate investigation results (neighbor nodes of the complaint questions corresponding to the work order are used as candidate investigation results) are obtained, the candidate investigation results are input into the trained model in the fourth step one by one, and the candidate investigation results are ordered according to the scores.

The embodiment of the application also provides a device for associating the complaint event, and the device for associating the complaint event can be used for executing the method for associating the complaint event provided by the embodiment of the application. The device is used for realizing the above embodiments and preferred embodiments, and is not described in detail. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.

The following describes a device for associating complaint events provided in the embodiments of the present application.

Fig. 9 is a block diagram of a device for correlating complaint events according to an embodiment of the present application. As shown in fig. 9, the apparatus includes:

an obtaining unit 100, configured to obtain a worksheet data, where the worksheet data at least includes a complaint problem and a survey result, where the complaint problem is a problem of a user for a complaint event, and the survey result is a cause of the user for the complaint event;

a first construction unit 200, configured to construct a correlation graph according to the complaint problem and the investigation result, where a node in the correlation graph is the correlation between the complaint problem and the investigation result, and an edge in the correlation graph is a correlation between any two of the complaint problems, or an edge in the correlation graph is a correlation between any one of the complaint problems and any one of the investigation result;

And a first processing unit 300 configured to, when receiving a target complaint problem, search for a correlated complaint problem and a correlated investigation result in the correlation graph, and display the correlated complaint problem and the correlated investigation result, where the correlated complaint problem is the complaint problem connected to the target complaint problem, and the correlated investigation result is the investigation result connected to the target complaint problem.

In a specific implementation process, the first construction unit comprises a first calculation module, a first connection module, a second calculation module and a second connection module, wherein the first calculation module is used for calculating first association strength between any two complaints according to a first formula, and the first formula is as follows:

W _i,j Representing the first correlation strength, p, between complaint problem i and complaint problem j _i,j Indicating whether complaint i and complaint j occur in the work order p at the same time, N indicating the work order set, N indicating the number of work orders, q _i,j Representing whether complaint problem i and complaint problem j occur simultaneously in complaints of the same user q, M representing a set of complaint users, M representing the number of complaint users, a ₁ Representing a first weight, a ₂ Representing a second weight; the first connection module is used for connecting the complaint problem i and the complaint problem j according to the first association strength; the second calculation module is configured to calculate a second association strength between any one of the complaint problems and any one of the investigation results according to a second formula, where the second formula is:

X _i,j representing the second correlation strength between complaint i and survey j, b _i,j Whether the complaint problem i and the investigation result j occur in the work order q at the same time or not is indicated, K indicates a complaint work order set, and K indicates the number of complaint work orders; the second connection module is used for connecting the complaint question i and the investigation result j according to the second association strength. The above formulas are exemplary only, and variations of any formulas fall within the scope of the present application.

In order to understand and analyze the work order data more efficiently, the device further comprises a second processing unit, a third processing unit, a determining unit and a fourth processing unit, wherein the second processing unit is used for performing word segmentation processing on the work order data after acquiring the work order data to obtain first work order data; the third processing unit is used for performing stop word removal processing on the first work order data to obtain second work order data; the determining unit is used for determining whether the second worksheet data comprise filtering vocabulary or not, wherein the filtering vocabulary is a vocabulary in a pre-configured dictionary to be filtered; and the fourth processing unit is used for deleting the filtering vocabulary from the second work order data to obtain target work order data when the second work order data comprises the filtering vocabulary.

In order to identify and extract key information in the text, the device further comprises a calculating unit and a fifth processing unit, wherein the calculating unit is used for calculating the TF-IDF value of each word in the work order data according to a TF-IDF algorithm after the work order data are acquired; and the fifth processing unit is used for selecting a first preset number of words from the target sentences according to the sequence of the TF-IDF values from large to small to obtain keywords, wherein the keywords are used for constructing the association graph.

In some embodiments, the worksheet data further includes user information, and the apparatus further includes a second construction unit and a sixth processing unit, where the second construction unit is configured to construct an auxiliary survey model after constructing a correlation graph according to the complaint problem and the survey result, where the auxiliary survey model is trained using a plurality of sets of training data, and each set of training data in the plurality of sets of training data includes historical user information acquired in a historical time period, a historical complaint problem corresponding to the historical user information, and a historical survey result corresponding to the historical user information; the sixth processing unit is used for inputting the current user information into the auxiliary investigation model to obtain the current complaint problem and the current investigation result corresponding to the current user information.

In some embodiments, the apparatus further comprises a seventh processing unit and an eighth processing unit, the seventh processing unit configured to, prior to constructing the auxiliary survey model, generate the historical complaint problem v ₁ And the above history investigation result v ₂ With edges therebetweenExtracting the history complaint problem v ₁ First history information of all relevant history worksheets and extracting and searching result v of the history ₂ Obtaining the history user information by the second history information of all the related history worksheets; an eighth processing unit for solving the above-mentioned historical complaint problem v ₁ And the above history investigation result v ₂ Extracting the history complaint problem v without edges ₁ And obtaining the historical user information according to the first historical information of any relevant historical work order, wherein the historical user information is used for model training.

In a specific implementation process, the device further comprises a clustering unit and a ninth processing unit, wherein the clustering unit is used for clustering the complaint problems to obtain at least one cluster after constructing a correlation diagram according to the complaint problems and the investigation results; and the ninth processing unit is used for selecting a second preset number of complaint problems from the cluster according to the order of the centrality from high to low to obtain a hot spot complaint problem, wherein the centrality is the distance between the average cluster of the data points and the cluster center.

The association device of the complaint event comprises a processor and a memory, wherein the acquisition unit, the first construction unit, the first processing unit and the like are all stored in the memory as program units, and the processor executes the program units stored in the memory to realize corresponding functions. The modules are all located in the same processor; alternatively, the above modules may be located in different processors in any combination.

The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel can be provided with one or more than one kernel, and the problem that the efficiency of complaint problem investigation is lower in the prior art is solved by adjusting kernel parameters.

The memory may include volatile memory, random Access Memory (RAM), and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip.

The embodiment of the invention provides a computer readable storage medium, which comprises a stored program, wherein the program is used for controlling equipment where the computer readable storage medium is located to execute a method for associating the complaint event.

The embodiment of the invention provides a processor which is used for running a program, wherein the program runs to execute the association method of the complaint event.

The present application also provides a complaint event correlation system, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include a correlation method for executing any one of the complaint events.

The embodiment of the invention provides equipment, which comprises a processor, a memory and a program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the associated method steps of complaint events. The device herein may be a server, PC, PAD, cell phone, etc.

The present application also provides a computer program product adapted to perform a program of associated method steps initialized with at least complaint events when executed on a data processing device.

It will be appreciated by those skilled in the art that the modules or steps of the invention described above may be implemented in a general purpose computing device, they may be concentrated on a single computing device, or distributed across a network of computing devices, they may be implemented in program code executable by computing devices, so that they may be stored in a storage device for execution by computing devices, and in some cases, the steps shown or described may be performed in a different order than that shown or described herein, or they may be separately fabricated into individual integrated circuit modules, or multiple modules or steps of them may be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.

It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises an element.

From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

1) According to the association method of the complaint event, the association relation between the complaint problems can be determined through the constructed association diagram, and the association relation between the complaint problems and the investigation result can also be determined, so that when new complaint problems are fed back, the associated complaint problems and investigation result can be excavated according to the new complaint problems, deeper analysis can be assisted by complaint treatment personnel, and the investigation efficiency of the complaint problems is improved.

2) According to the association device of the complaint event, through the constructed association diagram, the association relation between the complaint problems can be determined, and the association relation between the complaint problems and the investigation result can also be determined, so that when new complaint problems are fed back, the associated complaint problems and investigation result can be excavated according to the new complaint problems, deeper analysis can be assisted by complaint treatment personnel, and the investigation efficiency of the complaint problems is improved.

The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method of correlating complaint events, comprising:

acquiring work order data, wherein the work order data at least comprises a complaint problem and a survey result, the complaint problem is a problem of a user on a complaint event, and the survey result is a generation reason of the user on the complaint event;

constructing an association graph according to the complaint questions and the investigation results, wherein nodes in the association graph are the complaint questions and the investigation results, and edges in the association graph are association relations between any two complaint questions or between any one complaint question and any one investigation result;

under the condition that a target complaint problem is received, searching an associated complaint problem and an associated investigation result in the associated graph, and displaying the associated complaint problem and the associated investigation result, wherein the associated complaint problem is the complaint problem connected with the target complaint problem, and the associated investigation result is the investigation result connected with the target complaint problem.

2. The method of claim 1, wherein constructing a correlation graph from the complaint questions and the survey results comprises:

Calculating a first association strength between any two complaint problems according to a first formula, wherein the first formula is as follows:

W _i,j representing the first correlation strength, p, between complaint problem i and complaint problem j _i,j Indicating whether complaint i and complaint j occur in the work order p at the same time, N indicating the work order set, N indicating the number of work orders, q _i,j Indicating whether complaint i and complaint j occur simultaneously in the same user qIn complaints, M represents a set of complaint users, M represents the number of complaint users, a ₁ Representing a first weight, a ₂ Representing a second weight;

connecting a complaint problem i and a complaint problem j according to the first association strength;

calculating a second association strength between any one of the complaint problems and any one of the investigation results according to a second formula, wherein the second formula is as follows:

X _i,j representing the second correlation strength between complaint problem i and survey result j, b _i,j Whether the complaint problem i and the investigation result j occur in the work order q at the same time or not is indicated, K indicates a complaint work order set, and K indicates the number of complaint work orders;

and connecting the complaint problem i with the investigation result j according to the second association strength.

3. The method of claim 1, wherein after acquiring the work order data, the method further comprises:

Word segmentation is carried out on the worksheet data to obtain first worksheet data;

performing stop word removal processing on the first work order data to obtain second work order data;

determining whether the second work order data comprises a filtered vocabulary, wherein the filtered vocabulary is a vocabulary in a pre-configured dictionary to be filtered;

and deleting the filtered vocabulary from the second work order data under the condition that the second work order data comprises the filtered vocabulary, so as to obtain target work order data.

4. The method of claim 1, wherein after acquiring the work order data, the method further comprises:

calculating the TF-IDF value of each word in the work order data according to a TF-IDF algorithm;

and selecting a first preset number of words from the target sentences according to the sequence of the TF-IDF values from large to small to obtain keywords, wherein the keywords are used for constructing the association graph.

5. The method of claim 1, wherein the work order data further includes user information, and after constructing a correlation graph from the complaint questions and the survey results, the method further comprises:

constructing an auxiliary investigation model, wherein the auxiliary investigation model is trained by using a plurality of sets of training data, and each set of training data in the plurality of sets of training data comprises historical user information acquired in a historical time period, historical complaint problems corresponding to the historical user information and historical investigation results corresponding to the historical user information;

And inputting the current user information into the auxiliary investigation model to obtain the current complaint problem and the current investigation result corresponding to the current user information.

6. The method of claim 5, wherein prior to constructing the auxiliary survey model, the method further comprises:

at the history complaint problem v ₁ And the history survey result v ₂ Extracting the historical complaint problem v under the condition of edges ₁ First history information of all relevant history worksheets and extracting the history investigation result v ₂ Obtaining the history user information by the second history information of all the related history worksheets;

at the history complaint problem v ₁ And the history survey result v ₂ Extracting the history complaint problem v without edges ₁ And obtaining the historical user information according to the first historical information of any related historical worksheet, wherein the historical user information is used for model training.

7. The method of claim 1, wherein after constructing a correlation graph from the complaint questions and the survey results, the method further comprises:

clustering the complaint problems to obtain at least one cluster;

And selecting a second preset number of complaint problems from the cluster according to the order of the centrality from high to low to obtain a hot spot complaint problem, wherein the centrality is the distance between the average cluster of the data points and the cluster center.

8. A device for correlating complaint events, comprising:

the system comprises an acquisition unit, a control unit and a control unit, wherein the acquisition unit is used for acquiring worksheet data, the worksheet data at least comprise complaint questions and investigation results, the complaint questions are questions of a user for a complaint event, and the investigation results are the generation reasons of the user for the complaint event;

the first construction unit is used for constructing a correlation graph according to the complaint questions and the investigation results, wherein nodes in the correlation graph are the complaint questions and the investigation results, edges in the correlation graph are the correlation between any two complaint questions, or edges in the correlation graph are the correlation between any one complaint question and any one investigation result;

the first processing unit is used for searching the associated complaint questions and the associated investigation results in the associated graph under the condition that the target complaint questions are received, and displaying the associated complaint questions and the associated investigation results, wherein the associated complaint questions are the complaint questions connected with the target complaint questions, and the associated investigation results are the investigation results connected with the target complaint questions.

9. A computer-readable storage medium, characterized in that the computer-readable storage medium comprises a stored program, wherein the program, when run, controls a device in which the computer-readable storage medium is located to perform the method of correlating complaint events according to any one of claims 1 to 7.

10. A complaint event correlation system, comprising: one or more processors, memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising an association method for performing the complaint event of any one of claims 1-7.