Harbor departure training recommendation method and device based on grouping model, storage medium and equipment

Abstract

The invention provides a departure training recommendation method and device based on a grouping model, and the method comprises the steps: a departure user behavior data collection step: collecting various types of data when a user uses a departure system, and carrying out the analysis and storage of a plurality of dimensions; a departure user grouping step: according to the collected data, performing grouping by using a clustering algorithm to obtain different user group type function lists; a training content recommendation step: real-time learning progress analysis is carried out, and a personalized training recommendation list is recommended for the user; and a departure front-end display step: displaying the training function recommendation list to the user, constructing test data based on the service object characteristics, and customizing a personalized training scheme for the user. According to the method, the clustering algorithm is adopted, the operation behavior data of the users are deeply mined and analyzed, the requirements of each user can be more accurately recognized, and personalized recommendation is achieved; and an advanced stream processing technology and a real-time analysis framework are utilized to respond to a user request in real time.

Description

Technical Field

[0001] The present application belongs to the field of data processing technology, and in particular, relates to a departure training recommendation method and apparatus based on a clustering model, a storage medium and a device. Background Art

[0002] With the economic development of today's society, the rapid development of the civil aviation industry, and the increasing living standards of the people, airplanes have become the preferred mode of transportation for people to work and travel. Airport staff need to handle a large number of passenger service businesses every day, and departure business also has high requirements for the timeliness of staff handling. In the aviation departure system, staff need to quickly and accurately master the use of the system. However, traditional departure system training methods are mostly manuals and training documents, which lack business pertinence, and the quality of training documents compiled by various airports is uneven. What's worse, there are no training documents, and they basically rely on old employees to lead the practical operation step by step. This leads to high training costs and slow results, and no systematic continuous learning tools are provided for new or uncommon businesses. Summary of the invention

[0003] In view of the shortcomings of the prior art, the present application proposes a departure training recommendation method and device based on a clustering model, a storage medium and a device. The technical problem to be solved by the present invention is achieved through the following technical solutions.

[0004] The first aspect of the present application provides a departure system training method based on a grouping model, comprising the following steps:

[0005] Departure user behavior data collection step: used to collect various data when users use the departure system, analyze and store it through multiple dimensions;

[0006] Steps for grouping outbound users: Based on the collected data, clustering algorithms are used to group users based on their service business similarities, and different user group types, service object characteristics, and high / low frequency function lists are obtained;

[0007] Training content recommendation steps: Conduct real-time learning progress analysis and recommend a personalized training recommendation list to users based on personalized training recommendation rules;

[0008] Departure front-end display steps: Display the recommended list of training functions to users, build test data based on the characteristics of service objects, and user training process data to customize personalized training plans for users.

[0009] In combination with the first aspect, the above-mentioned departure user behavior data collection step further includes: the data includes function operation sequence, dwell time, number of clicks, user role and business attributes, and the multiple dimensions include user role, business attributes, function operation frequency, function usage / dwell time and function operation sequence: function operation sequence information of users at each counter, the sequence information operation time, operation function ID and operation counter number.

[0010] In combination with the first aspect, the above-mentioned step of grouping outbound users further includes the following steps:

[0011] Data preprocessing step: clean the collected behavioral data;

[0012] Feature extraction step: extracting the user's feature vector from the preprocessed data;

[0013] Clustering algorithm processing: All points are assigned based on the K-means algorithm, and the center point of the cluster is recalculated based on all the points in the cluster and iterated until the change of the cluster center point is small or the specified number of iterations is reached.

[0014] Combined with the first aspect, the processing steps of the above clustering algorithm are as follows: first, estimate the value range of the cluster number, and according to the business scenario, airport and role dimensions of the behavior data, split the business as finely as possible and then classify it. The number n obtained by classification is the K-means algorithm cluster number, and the algorithm is used to select the optimal cluster number; second, calculate the optimal cluster number, and select the optimal cluster number by iterating the intra-class squared error and WCSS of all calculated cluster numbers k in the number range. The WCSS formula is:

[0015] Among them C i represents the i-th cluster, which contains all the points belonging to this cluster.

[0016] x j It is a cluster C i A data point in

[0017] μ i It is a cluster C i The center of mass, C i The mean of all points,

[0018] ||x j -μ i || represents the data point x j To the center of mass μ i The Euclidean distance of

[0019] The values ​​of parameters i, n and k are natural numbers, and the value range of k is [n, 2n].

[0020] In combination with the first aspect, the above clustering algorithm processing further includes: the output clustering results include customer group names, customer group series characteristics and behavioral characteristics, and the clustering results are evaluated, and the advantages and disadvantages of the clustering results are evaluated by the elbow method to ensure the quality of clustering.

[0021] In combination with the first aspect, the above training content recommendation step further includes the following steps:

[0022] Step of building an exclusive training library: based on the customer group type selected by the user, obtaining the function list as the user's exclusive training library;

[0023] Business operation accuracy assessment steps: Based on the user's production business operation behavior data and business scenario path rule definition, identify the correctness and accuracy of business operations;

[0024] Real-time learning progress analysis step: obtaining the user's real-time learning progress through the acquired user training progress data, as well as the correctness and accuracy of the business operations;

[0025] Personalized training recommendation rule step: recommending a personalized training recommendation list to the user based on the user-specific training library and the user's real-time learning progress;

[0026] Training recommendation list step: outputting a training recommendation list, wherein the training recommendation list includes user ID, customer group type, customer group service object characteristics and a training recommendation function list.

[0027] In combination with the first aspect, the above departure front-end display step further includes the following steps:

[0028] DGU I training function recommendation step: presenting the training function recommendation list to the user, and constructing test data based on the service object characteristics, so as to facilitate the user to learn the departure system in a targeted manner;

[0029] DGU I functional training steps: Collect training progress data to facilitate real-time analysis of learning progress.

[0030] The second aspect of the present application provides a departure system training device based on a grouping model, comprising:

[0031] Departure user behavior data collection module: used to collect various data when users use the departure system, analyze and store them through multiple dimensions;

[0032] Departure user grouping module: Based on the collected data, the clustering algorithm is used to group users based on their service business similarities, and obtain different user group types and service object characteristics, as well as high / low frequency function lists;

[0033] Training content recommendation module: conducts real-time learning progress analysis and recommends personalized training recommendation lists to users based on personalized training recommendation rules;

[0034] Departure front-end display module: displays a recommended list of training functions to users, builds test data based on service object characteristics, and user training process data to customize personalized training plans for users.

[0035] A third aspect of the present application provides a computer-readable storage medium storing one or more programs, which, when executed, can implement the above-mentioned departure system training method based on the grouping model.

[0036] The fourth aspect of the present application provides a device, including a processor, a communication interface, a computer-readable storage medium and a communication bus; wherein the processor, the communication interface, and the computer-readable storage medium communicate with each other via the communication bus; and the above-mentioned processor is used to execute a program stored in the computer-readable storage medium.

[0037] Compared with the prior art, this application has the following advantages:

[0038] Personalized recommendations: Using clustering algorithms, we deeply mine and analyze users’ operational behavior data, and automatically divide users into segments with similar behavior patterns, so that the recommendation system can more accurately identify the needs of each user and achieve personalized recommendations.

[0039] Systematic training: Adopting a multi-dimensional and refined strategy, dynamically constructing a training content recommendation algorithm based on key technical indicators such as business operation frequency, quantitative evaluation of function learning complexity, existing knowledge and skill foundation (learned), actual application scenario coverage (used), and actual application effect (accuracy). This algorithm aims to guide users to follow the path of capability advancement, starting from high-frequency basic functions and gradually transitioning to low-frequency complex functions, while encouraging users to extend from core service areas to related and expanded business areas. Through technology-driven personalized learning path planning, the systematic integration of learning content and the optimization of difficulty gradients are achieved, ensuring that users can master the required skills in an efficient and orderly manner, and accelerating the specialization and standardization of the skill system.

[0040] Real-time: It integrates high-performance data collection and processing modules, and can capture and integrate user behavior data (such as operation logs, interaction feedback, etc.) and training progress data in real time. Using advanced stream processing technology and real-time analysis framework, it can perform complex data processing and analysis tasks as soon as data is generated, achieving a response speed of seconds or even milliseconds.

[0041] This application uses a clustering algorithm to deeply mine and analyze user operation behavior data, which can more accurately identify the needs of each user and achieve personalized recommendations; and integrates a high-performance data collection and processing module, using advanced stream processing technology and real-time analysis framework to respond to user requests in real time.

[0042] Other features and advantages of the present application will be described in the following description, and partly become apparent from the description, or be understood by practicing the present application. The purpose and other advantages of the present application can be realized and obtained by the structures indicated in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0044] Figure 1 The structural schematic diagram of the departure training recommendation device based on the clustering model is shown;

[0045] Figure 2 A schematic flow chart of a departure training recommendation device based on a clustering model is shown;

[0046] Figure 3 The DGUI system training function interface is shown;

[0047] Figure 4 It is a schematic diagram of the internal structure of a device according to an embodiment of the present application. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0049] The English abbreviation DGUI used in this application refers to the China Aviation Information Service's open general departure front-end software, and its full English name is: ETERM DEPARTURE.

[0050] like Figure 1As shown, a departure system training device based on a grouping model of the present application includes four modules: a departure user behavior data collection module, a departure user grouping module, a training content recommendation module, and a departure front-end display module.

[0051] 1. Departure user behavior data collection module:

[0052] Used to collect various data when users use the departure system, including operation records, stay time, number of clicks, user roles, business attributes, etc.

[0053] 2. Departure user grouping module:

[0054] Based on the collected data, clustering algorithms are used to group users based on their service business similarities, and different user group types, service object characteristics, and high / low frequency function lists are obtained.

[0055] 3. Training content recommendation module:

[0056] Build an exclusive training library based on the high / low frequency of group business, difficulty of function learning, and definition of uncommon functions, and conduct real-time learning progress analysis based on training progress, on-site business operation accuracy, etc. Through personalized training recommendation rules, support the recommendation of suitable training recommendation lists for employees with different learning progress.

[0057] 4. Departure front-end display module:

[0058] A recommended list of training functions is displayed to users, and test data is constructed based on the characteristics of service objects to facilitate users to learn the departure system in a targeted manner. At the same time, user training process data is collected to continuously optimize the personalized training analysis mechanism.

[0059] The departure training recommendation device based on the clustering model of this application is as follows Figure 2 As shown, it is specifically divided into a departure user data collection module, a departure user grouping module, a training content recommendation module and an departure front-end display module.

[0060] 1. Departure user data collection module:

[0061] The departure system client is used to transmit back the user's function operation sequence, dwell time, number of clicks, user role, business attributes and other data, and parse and store them in the following dimensions.

[0062] User role: the user's identity or role (such as user ID, level, size of the airport, etc.).

[0063] Business attributes: corresponding to the attributes of the service object in the departure system, such as whether the flight is China Southern Airlines, China Eastern Airlines, a large airline, or a small airline, whether the passenger is in a wheelchair, VIP, checked in but no luggage added, and other attribute information.

[0064] Function operation frequency: the number of times users use each function.

[0065] Feature Usage / Dwell Time: The average time users spend on each feature.

[0066] Function operation sequence: Function operation sequence information of the user at each counter, such as operation time, operation function ID, operation counter number, etc.

[0067] 2. Departure user grouping module:

[0068] Based on the collected data, clustering algorithms are used to group users based on their service business similarities to obtain different user groups.

[0069] Data preprocessing: Clean the collected behavioral data, such as cleaning up some missing, duplicate and dirty data, to facilitate subsequent analysis.

[0070] Feature extraction: Extract the user’s feature vector from the preprocessed data.

[0071] Clustering algorithm processing: Since all types of data are discrete data, the K-Means clustering algorithm is selected.

[0072] The main idea of ​​the K-means algorithm is: given the K value and K initial cluster center points, each point (that is, data record) is assigned to the cluster represented by the cluster center point closest to it. After all points are assigned, the center point of a cluster is recalculated based on all the points in the cluster (taking the average value), and then the steps of assigning points and updating cluster center points are iterated until the change of the cluster center point is very small, or the specified number of iterations is reached.

[0073] The clustering algorithm processing steps of this application are as follows:

[0074] (1) Estimation of the range of cluster number

[0075] In the process of grouping departing users, business experts first need to manually split the business as finely as possible according to the business scenarios, airports, roles and other dimensions of the behavior data and then classify them. The number n obtained by classification is the K-means algorithm cluster number k, which is in the range of [n, 2n]. The algorithm is used to select the optimal cluster number. For example, if business experts manually analyze that there are a total of three counters: high-end counters, special passenger counters, and special baggage handling counters, then the algorithm cluster number k is in the range of [3, 6].

[0076] (2) Calculate the optimal number of clusters

[0077] The optimal number of clusters is selected by iteratively taking the sum of squared errors (WCSS) of all the calculated cluster numbers k in the number interval.

[0078] For each different k value, use the WCSS formula to calculate the corresponding WCSS value:

[0079]

[0080] Among them C i Represents the i-th cluster, which contains all the points belonging to this cluster.

[0081] x j It is a cluster C i A data point in .

[0082] μ i It is a cluster C i The center of mass, C i The mean of all points.

[0083] ||x j -μ i || represents the data point x j To the center of mass μ i The Euclidean distance of .

[0084] As the k value increases, the WCSS value will continue to decrease, and usually there will be an inflection point, and the k at this inflection point is the optimal number of clusters.

[0085] Model output content: Output group function list, including group type name, service object characteristics, and high / low frequency function list.

[0086] 3. Training content recommendation module:

[0087] Based on the high / low frequency of group business, difficulty of function learning, definition of uncommon functions, and personalized training analysis rules, it supports recommending suitable training recommendation lists for employees with different learning progress.

[0088] The specific implementation is as follows:

[0089] Build an exclusive training library: Based on the customer group type selected by the user, obtain the customer group's high- and low-frequency function lists as the user's exclusive training library, and build difficulty labels for the functions by combining the previous analysis of historical training effect data. So far, the user training library has four types of training function lists: customer group high-frequency difficult functions, customer group high-frequency easy functions, customer group low-frequency difficult functions, and customer group low-frequency easy functions. The difficulty evaluation dimension is shown in Table 1.

[0090] Rule Name Learning time First-time exam pass rate Difficulty Rule 1 Within 30 minutes Less than 60% Disaster Rule 2 Within 60 minutes Less than 80% Disaster Rule 3 Within 2 hours Less than 90% Disaster Rule 4 other other easy

[0091] Table 1 Difficulty assessment dimensions

[0092] Business operation accuracy assessment: Based on user production business operation behavior data and business scenario path rule definitions, identify the correctness and accuracy of business operations.

[0093] Real-time learning progress analysis: After obtaining the user's DGUI training progress data and the above-calculated production business operation accuracy, the proficiency of each functional operation is calculated according to the following rules to obtain the user's real-time learning progress. The functional mastery proficiency evaluation rules are shown in Table 2.

[0094]

[0095] Table 2 Functional proficiency assessment rules

[0096] As shown in Table 3, personalized training recommendation rules: Based on the above user-exclusive training library and real-time learning progress, the personalized training recommendation rules configured in the early stage are used to output the user training recommendation list content.

[0097]

[0098] Table 3 Personalized training recommendation rules

[0099] Training recommendation list: Based on personalized training recommendation rules, output training recommendation list, including user ID, customer type, customer service object characteristics, and training recommendation function list.

[0100] 4. Off-post front-end display module:

[0101] DGU I training function recommendation: Display the recommended list of training functions to users, and build test data based on the characteristics of service objects to facilitate users to learn the departure system in a targeted manner.

[0102] DGU I functional training: collects training progress data, such as user ID, training customer group type, learned function type and specific functions, to facilitate real-time analysis of learning progress.

[0103] The content of this application will be explained below with reference to specific embodiments.

[0104] 1.2.1. Departure user data collection module

[0105] Use the departure system client to transmit back data such as user roles, business attributes, number of operations, dwell time, number of clicks, etc., and parse and store them in the following dimensions.

[0106] For example: User role: check-in clerk / check-in supervisor / boarding, etc.; User company size: tens of millions of level airports / millions of level airports; Counter attributes: check-in / baggage check-in / high-end / employee; Business attributes: check-in / boarding; Airport region: South China / Southwest China, etc.; User account creation time: 2013.3.1; User login time: 2013.5.1 08:30; User system exit time: 2013.5.1 20:28.

[0107] Business attributes: Flight status - open for check-in, Passenger check-in status - checked in, Passenger type - general / senior / infant / high-end, Baggage status - with / without free baggage allowance, Ticket status - unused / checked in / refunded, Itinerary type - domestic one-way / domestic connecting / international one-way / mixed connecting, Seat status - reserved / not reserved, etc.

[0108] Operation behavior: operation business ID, operation content, operation method, operation time, operation result, error content, etc.

[0109] 2. Departure user grouping module

[0110] Based on the collected data, clustering algorithms are used to group users based on their service business similarities to obtain different user groups.

[0111] Data preprocessing: If the returned data of the same key operation is performed multiple times due to the user's fast hand speed, the data of one operation can be retained for analysis. Or if the returned data is missing due to program or other reasons and does not meet the analysis conditions, it can be directly cleaned up.

[0112] Feature extraction: Extract the following dimensional feature vectors of the user's basic information, behavior, etc. from the preprocessed data. Basic information features: Employee attributes: new employee / old employee; Employee activity: active employee / general employee / inactive employee; Employee airport scale: large scale / small scale; Employee business scope: many businesses / few businesses; Employee business attributes: check-in / boarding, etc.

[0113] Behavioral characteristics: operation preferences: mouse / shortcut keys; operation ID, operation content, operation method, etc. of each business attribute.

[0114] Clustering algorithm processing: Since all types of data are discrete data, the K-Means clustering algorithm is selected. After the above-mentioned large number of historical features are extracted and trained, the clustering results are output in the following format: 1) Customer group name: large-scale airport ordinary check-in staff handle international check-in business customer group; 2) Customer group series characteristics: employee attributes-new employees; employee airport scale-large-scale; employee role-ordinary employees; employee business-check-in; service object check-in status-not checked in; itinerary status-international one-way; seat selection status-no seat selection; luggage status-free baggage allowance; 3) Behavioral characteristics: high-frequency operation scenarios and business ID-international check-in (document verification 22220 / seat selection check-in 33330) / advance seat selection (seat selection 55550) / international check-in and consignment (document verification 222 20 / seat selection check-in 33330 / baggage check-in 666660) / additional baggage purchase (add excess baggage 77770 / price check 99990); international check-in operation path - extract passenger information 34353@document verification 22220@seat selection check-in 33330, advance seat selection operation path - extract passenger information 34353@seat selection 55550, international check-in operation path - extract passenger information 34353@document verification 22220@seat selection check-in 33330@baggage check-in 666660, additional baggage purchase operation path - extract passenger information 34353@add excess baggage 77770@price check 99990. Low-frequency operation scenarios and business IDs - view passenger details (view details 12222) / add member (add member 13333) / add special service (add unaccompanied 14444)

[0115] Clustering result evaluation: The elbow method is used to evaluate the quality of clustering results to ensure the quality of clustering. For example, by observing the curve, find the inflection point where the SSE value drops sharply as the number of clusters increases and then tends to stabilize. The number of clusters corresponding to this inflection point is considered to be the optimal number of clusters.

[0116] Model output content: Output group function list, including group type name, service object characteristics, high / low frequency function list. The example is the same as above.

[0117] 3. Training content recommendation module:

[0118] Based on the high / low frequency of group business, difficulty of function learning, definition of uncommon functions, and personalized training analysis rules, it supports recommending suitable training recommendation lists for employees with different learning progress.

[0119] The following describes the specific application scenarios of this application:

[0120] Build an exclusive training library: Based on the user's self-selected customer type, obtain a list of high- and low-frequency functions for the customer group as the user's exclusive training library, and build difficulty labels for the functions based on the previous analysis of historical training effect data. So far, the user training library has four types of training function lists: high-frequency and difficult functions for customer groups, high-frequency and easy functions for customer groups, low-frequency and difficult functions for customer groups, and low-frequency and easy functions for customer groups.

[0121] For example, Zhang San has bound the customer group type in the DGU I training system to the customer group of large-scale airport general check-in staff handling international check-in services, and obtained the function list. The content of the exclusive training library is as follows:

[0122] User ID: 001; Name: Zhang San; Customer group: ordinary check-in staff at large airports handling international check-in services; High-frequency and difficult functions for this customer group: document verification 22220 / adding overweight baggage 77770 / price check 99990; High-frequency and easy functions for this customer group: seat selection and check-in 33330 / baggage check-in 666660; Low-frequency and difficult functions for this customer group: adding members 13333 / adding a person without a companion 14444; Low-frequency and easy functions for this customer group: viewing details 12222.

[0123] Business operation accuracy assessment: Based on user production business operation behavior data and business scenario path rule definitions, identify the correctness and accuracy of business operations.

[0124] For example, Zhang San has no production business operation data.

[0125] Real-time learning progress analysis: Obtain the user's DGUI training progress data and the user's production and operation function data, compare the two, and obtain the user's real-time learning progress.

[0126] For example, if Zhang San is a new employee who has never received any training, the learning progress he obtains is: the proficiency function points are empty.

[0127] Personalized training recommendation rules: Based on the above-mentioned user-exclusive training library and real-time learning progress, the user training recommendation list content is output through the personalized training recommendation rules configured in advance.

[0128] For example, Zhang San recommends a training function type that is a high-frequency and easy-to-use function for customers.

[0129] Training recommendation list: Based on personalized training recommendation rules, output training recommendation list, including user ID, customer type, customer service object characteristics, and training recommendation function list.

[0130] Such as 1) User ID: 001; 2) Customer group type: ordinary check-in staff at large airports handling international check-in services; 3) Customer group service object characteristics Check-in status - not checked in; Itinerary status - international one-way; Seat selection status - not selected; Baggage status - free baggage allowance; 4) Training recommended function list: Seat selection check-in 33330 / Baggage check-in 666660.

[0131] 4. Training content recommendation module:

[0132] DGU I training function recommendation: Display the training function recommendation list to users and build test data based on the service object characteristics to facilitate users to learn the departure system in a targeted manner. Figure 3 shown.

[0133] DGU I functional training: collects training progress data, such as user ID, training customer group type, learned function type and specific functions, to facilitate real-time analysis of learning progress.

[0134] For example, after Zhang San has finished his training, he has learned the following: seat selection and check-in 33330 / baggage check-in 666660. When he undergoes the training again for real-time learning progress analysis, the content he has learned will be different.

[0135] The advantages of this application are mainly reflected in the following aspects:

[0136] Personalized recommendations: Using clustering algorithms, we deeply mine and analyze users’ operational behavior data, and automatically divide users into segments with similar behavior patterns, so that the recommendation system can more accurately identify the needs of each user and achieve personalized recommendations.

[0137] Systematic training: Adopting a multi-dimensional and refined strategy, dynamically constructing a training content recommendation algorithm based on key technical indicators such as business operation frequency, quantitative evaluation of function learning complexity, existing knowledge and skill foundation (learned), actual application scenario coverage (used), and actual application effect (accuracy). This algorithm aims to guide users to follow the path of capability advancement, starting from high-frequency basic functions and gradually transitioning to low-frequency complex functions, while encouraging users to extend from core service areas to related and expanded business areas. Through technology-driven personalized learning path planning, the systematic integration of learning content and the optimization of difficulty gradients are achieved, ensuring that users can master the required skills in an efficient and orderly manner, and accelerating the specialization and standardization of the skill system.

[0138] Real-time: It integrates high-performance data collection and processing modules, and can capture and integrate user behavior data (such as operation logs, interaction feedback, etc.) and training progress data in real time. Using advanced stream processing technology and real-time analysis framework, it can perform complex data processing and analysis tasks as soon as data is generated, achieving a response speed of seconds or even milliseconds.

[0139] In summary, the advantages of this application lie in its personalized recommendations, systematic training recommendation rules, real-time performance, etc. These innovations make the device highly practical and have broad application prospects in the field of aviation information technology.

[0140] Based on the same inventive concept, the present application also provides a computer-readable storage medium storing one or more programs. When the one or more programs are executed, the aforementioned electricity market transaction decision-making method with the solar thermal power station as the main body can be implemented.

[0141] like Figure 4 As shown, an embodiment of the present application also provides a device, which includes a processor, an internal memory and a network interface connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external terminal via a network connection.

[0142] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the electronic device to which the solution of the present invention is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements. The computer-readable storage medium may be included in the device / apparatus described in the above embodiments; or it may exist independently without being assembled into the device / apparatus.

[0143] Although the present application has been described in detail with reference to the aforementioned embodiments, a person of ordinary skill in the art should understand that it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent replacements for some of the technical features therein; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A departure system training method based on a clustering model, characterized in that: The following steps are involved: Departure user behavior data collection step: used to collect various data when users use the departure system, analyze and store it through multiple dimensions; Steps for grouping outbound users: Based on the collected data, clustering algorithms are used to group users based on their service business similarities, and different user group types, service object characteristics, and high / low frequency function lists are obtained; Training content recommendation steps: Conduct real-time learning progress analysis and recommend a personalized training recommendation list to users based on personalized training recommendation rules; Departure front-end display steps: Display the recommended list of training functions to users, build test data based on the characteristics of service objects, and user training process data to customize personalized training plans for users.

2. The method according to claim 1, characterized in that The step of collecting departure user behavior data further includes: the data includes function operation sequence, dwell time, number of clicks, user role and business attributes, the multiple dimensions include user role, business attributes, function operation frequency, function usage / dwell time and function operation sequence: function operation sequence information of users at each counter, the sequence information operation time, operation function ID and operation counter number.

3. The method according to claim 2, characterized in that The step of grouping outbound users further comprises the following steps: Data preprocessing step: clean the collected behavioral data; Feature extraction step: extracting the user's feature vector from the preprocessed data; Clustering algorithm processing: All points are assigned based on the K-means algorithm, and the center point of the cluster is recalculated based on all the points in the cluster and iterated until the change of the cluster center point is small or the specified number of iterations is reached.

4. The method according to claim 3, characterized in that The processing steps of the clustering algorithm are as follows: first, estimate the range of cluster number values, and according to the business scenarios, airports and role dimensions of the behavior data, split the business as finely as possible and then classify it. The number n obtained by classification is the K-means algorithm cluster number, and the algorithm is used to select the optimal cluster number; second, calculate the optimal cluster number, and select the optimal cluster number by iterating the intra-class squared errors and WCSS of all calculated cluster numbers k within the number range. The WCSS formula is: Among them C i represents the i-th cluster, which contains all the points belonging to this cluster. x j It is a cluster C i A data point in μ i It is a cluster C i The center of mass, C i The mean of all points, ||x j -μ i || represents the data point x j To the center of mass μ i The Euclidean distance of The values ​​of parameters i, n and k are natural numbers, and the value range of k is [n, 2n].

5. The method according to claim 4, characterized in that The clustering algorithm processing further includes: outputting clustering results including customer group names, customer group series characteristics and behavioral characteristics, and evaluating the clustering results, evaluating the pros and cons of the clustering results by the elbow method to ensure the quality of clustering.

6. The method according to claim 5, characterized in that The training content recommendation step further includes the following steps: Step of building an exclusive training library: based on the customer group type selected by the user, obtaining the function list as the user's exclusive training library; Business operation accuracy assessment steps: Based on the user's production business operation behavior data and business scenario path rule definition, identify the correctness and accuracy of business operations; Real-time learning progress analysis step: obtaining the user's real-time learning progress through the acquired user training progress data, as well as the correctness and accuracy of the business operations; Personalized training recommendation rule step: recommending a personalized training recommendation list to the user based on the user-specific training library and the user's real-time learning progress; Training recommendation list step: outputting a training recommendation list, wherein the training recommendation list includes user ID, customer group type, customer group service object characteristics and a training recommendation function list.

7. The method according to claim 6, characterized in that The departure front-end display step further comprises the following steps: DGUI training function recommendation step: presenting the training function recommendation list to the user, and constructing test data based on the service object characteristics, so as to facilitate the user to learn the departure system in a targeted manner; DGUI function training steps: Collect training progress data to facilitate real-time analysis of learning progress.

8. A departure system training device based on a clustering model, characterized in that: include: Departure user behavior data collection module: used to collect various data when users use the departure system, analyze and store them through multiple dimensions; Departure user grouping module: Based on the collected data, the clustering algorithm is used to group users based on their service business similarities, and obtain different user group types and service object characteristics, as well as high / low frequency function lists; Training content recommendation module: conducts real-time learning progress analysis and recommends personalized training recommendation lists to users based on personalized training recommendation rules; Departure front-end display module: displays a recommended list of training functions to users, builds test data based on service object characteristics, and user training process data to customize personalized training plans for users.

9. A computer-readable storage medium storing one or more programs, characterized in that: When the one or more programs are executed, the departure system training method based on the grouping model described in any one of claims 1 to 7 can be implemented.

10. A device comprising a processor, a communication interface, the computer-readable storage medium of claim 9, and a communication bus; wherein: The processor, the communication interface, and the computer-readable storage medium communicate with each other via a communication bus; It is characterized in that The processor is configured to execute a program stored in a computer-readable storage medium.

Images