A method and system for intelligently generating personalized overseas study consulting reports

By constructing the approximate completion number hit probability judgment of the form pointing to time in the decentralized scenario, and selecting the appropriate form as the training data set, the problem of inconsistent data diversity and review standards during the study abroad consultation process is solved, and the study abroad consultation is improved Report accuracy and recall.

CN118484601BActive Publication Date: 2024-11-22HUANYU THINK TANK INT EDUCATION CONSULTING CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410676789.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-29
Publication Date
2024-11-22
Estimated Expiration
2044-05-29

AI Technical Summary

Technical Problem

During the study abroad consultation process, due to the diverse data sources and inconsistent audit standards, the results of deviations in deep learning model training, and the generated study abroad consultation report is inaccurate, especially in the absence and error of sensitive data in decentralized scenarios, resulting in recommendations. Reduced accuracy.

Method used

By constructing a form to determine the hit probability of the approximate completion number in time in the decentralized scenario, selecting the appropriate form as the training data set to reduce data contradictions and errors, and improving model training accuracy and recall rate.

Benefits of technology

It effectively reduces contradictions and errors in the training dataset, improves the accuracy and recall rate of deep learning models, and ensures that the generated study abroad consultation report is more accurate and reliable.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118484601B_ABST
    Figure CN118484601B_ABST
Patent Text Reader

Abstract

The present invention belongs to the technical field of natural language processing, and provides a personalized study abroad consulting report intelligent generation method and system, by creating a form on each client in a decentralized scenario, the form includes multiple form items, each form item corresponds to the characteristic information in the user text data one by one; the form guide transmitted from each client is selected as a training data set; the neural network model is trained by the training data set to obtain a pre-trained model; the user text data to be recommended is input into the pre-trained model to obtain the success rate of the study abroad information and push the study abroad information. It can reduce the probability of conflicts, contradictions and errors in the data, and can timely clear all forms that may have errors in the training data set, thereby adjusting the accuracy of the pre-trained model during training and ensuring the recall rate of the pre-trained model in subsequent steps.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of natural language processing, and in particular relates to a method and system for intelligently generating a personalized study abroad consulting report. Background Art

[0002] In the current overseas study consultation process, generally, intelligent customer service is used to generate personalized overseas study consultation reports for each customer. However, since overseas study involves a variety of overseas universities and course information, it is difficult for traditional overseas study consultation agencies to standardize the consultation work based on personal experience alone. Each school uses different quantitative and qualitative measurement standards when reviewing applications, and gives different weights to similar indicators when considering. Moreover, most schools do not even publish specific admission consideration standards. Therefore, the existing technology generally generates overseas study consultation reports based on the feature data provided by the customer by training a large model. With the training of deep learning models with a large amount of data, the use of deep learning models has become more and more mature.

[0003] However, when similar training data is obtained from different sources, there are sometimes a large number of contradictions or errors, which will cause problems and deviations in the trained deep learning model, resulting in many problems in the study abroad consulting report generated by the final deep learning model, thus affecting the accuracy of the recommendation; especially in the currently more common decentralized scenario of transmitting data between multiple clients. Although this scenario ensures that sensitive data stays locally and does not need to be exposed or transmitted, due to the lack of sensitive data when training the large model through this scenario, there are a large number of contradictions or errors, resulting in the generated study abroad consulting report producing inaccurate or vague answers when dealing with complex problems or requiring a deep understanding of the context. Summary of the invention

[0004] The purpose of the present invention is to propose a method and system for intelligently generating personalized study abroad consulting reports to solve one or more technical problems existing in the prior art and at least provide a beneficial choice or create conditions.

[0005] In order to achieve the above object, according to one aspect of the present invention, a method for intelligently generating a personalized study abroad consulting report is provided, and the method for intelligently generating a personalized study abroad consulting report comprises the following steps:

[0006] S100, create a form in each client;

[0007] S200, selecting the form guide transmitted from each client as a training data set;

[0008] S300, training the neural network model using the training data set to obtain a pre-trained model;

[0009] S500: Input the user text data to be recommended into the pre-trained model to obtain the success rate of the study abroad information and push the study abroad information.

[0010] Furthermore, the method further includes S600, using the user text data to be recommended and the corresponding success rate and information of studying abroad as the text of the study abroad consultation report.

[0011] Among them, study abroad information includes: the study abroad institutions and study abroad majors to be applied for.

[0012] The form includes a plurality of form items, and each form item corresponds one-to-one to feature information in user text data.

[0013] Further, in step S100, the user sends user text data to the server through a client in a decentralized scenario, wherein the client includes a mobile device or a desktop computer.

[0014] Among them, the decentralized scenario is multiple clients and at least one server, each client is connected to the server, and the neural network model is trained in real time through the data transmitted by each (any one or more) client as a training data set, or instructions are sent through any client for query, and the instructions at least include the user text data to be recommended.

[0015] The characteristic information in the user text data includes: user characteristic information in text format; the user characteristic information includes any one or more of the user name, academic credit points, intended region, education level, language scores, graduation school, graduation major, competition award experience, internship experience and graduation time. It can also include: overseas study school and overseas study major information. In a decentralized scenario, some of the characteristic information in the user text data will be blocked or encrypted or tampered due to privacy protection, so there will be a large number of contradictions or tampering problems.

[0016] Further, in S300, the neural network model is a SA-NET neural network model.

[0017] In a decentralized scenario where multiple clients transmit data, when a large number of clients transmit similar training data to the server in a similar or simultaneous time period, there are a large number of contradictions or tampering problems due to the lack of sensitive data, which will cause problems and deviations in the trained deep learning model. To avoid the following problems, the present application provides the following guided selection method to construct a training data set by judging the hit probability of the approximate number of completed items, and to perform guided selection to avoid these problems from interfering with the subsequent training data of the deep learning model and improve the recall rate of the model;

[0018] Furthermore, in S200, the method of selecting the form guides transmitted from each client as the training data set includes:

[0019] The time when the current client creates the form is recorded as the table creation time, and the time when all form items are filled in or the time when the form is transmitted to the server is recorded as the completion time; the period between the table creation time and the completion time is recorded as the form pointing time of the form; (the form pointing time refers to the time period when the current client's form is completed or submitted. In this time period, in a decentralized scenario, a large number of clients may transmit similar training data to the server in a similar or simultaneous time period).

[0020] During the form pointing time, the number of completed form items in all clients that are identical to any one or more form items in the form of the current client is the approximate number of completed items CompN;

[0021] Construct empty sets ASET and BSET respectively, record each moment when CompN>0 in the form pointing time as the conflict moment; let the mean of the difference between the durations of each conflict moment be MEANGAP; if the current conflict moment is CAT; then in the form pointing time, record the next conflict moment of CAT as CBT, and the next conflict moment of CBT as CCT in the time sequence;

[0022] For each approximate number of completed items, if the CCT moment is within the time period from the CAT moment to the CBT+MEANGAP moment within the form pointing time, the approximate number of completed items CAN at the moment CAT within the form pointing time, the approximate number of completed items CBN at the moment CBT, and the approximate number of completed items CCN at the moment CCT within the form pointing time are sequentially added to the set ASET; otherwise, CAN, CBN, and CCN are sequentially added to the set BSET (the set ASET contains the approximate number of completed items with a high hit probability within the form pointing time, and the set BSET contains the approximate number of completed items with a general hit probability within the form pointing time);

[0023] ASET(i) represents the number of approximately completed items in the i-th set ASET, where i is the sequence number;

[0024] Guide selection for each form, including the following steps:

[0025] For each ASET(i), if the condition is satisfied: BSetMean≥ASET(i)≥BSetSho, or, if the condition is satisfied: ASET(i)<BSetSho+JustIndi×BSETBig, and if the corresponding form is not in the training data set, then add the form to the training data set;

[0026] For each ASET(i), if the condition is satisfied: BSetMean<ASET(i)<BSETBig, or, if the condition is satisfied: ASET(i)≥BSetSho+JustIndi×BSETBig, and if the corresponding form already exists in the training data set, then delete the form from the training data set;

[0027] Where, JustIndi = exp(ASETSho÷ASetMean) / exp(ASetMean÷ASETBig);

[0028] Among them, exp represents the natural index, JustIndi is the form hit index; BSetMean is the mean of the number of approximately completed items in the set BSET; BSetSho represents the minimum number of approximately completed items in the set BSET; BSETBig represents the maximum number of approximately completed items in the set BSET; ASetMean is the mean of the number of approximately completed items in the set ASET; ASETSho represents the minimum number of approximately completed items in the set ASET; ASETBig represents the maximum number of approximately completed items in the set ASET.

[0029] Among them, the present application adds form items with a low hit probability of the approximate number of completed items (hit probability, that is, in a decentralized scenario where multiple clients transmit data, the possibility that the client transmits similar training data to the server at a similar or simultaneous time period at the same time and the data has errors or contradictions) to the training data set to reduce the problem of contradictions or errors in the forms in the training data set, so as to improve the accuracy of the pre-trained model obtained by training the neural network model through the training data set in the subsequent steps;

[0030] The form hit index JustIndi shows the limit degree ratio of the number of approximate completed items tolerating contradictions. When calculated through BSetSho+JustIndi×BSETBig, a reference value can be obtained after the predictive correction of the limit degree ratio by the minimum value of the approximate number of completed items in the correction set BSET. The comparison between this reference value and ASET(i) can accurately reflect the prediction of the form under the maximum limit case where the hit probability of the approximate number of completed items is low, and determine whether the corresponding form can be used as a training set.

[0031] Preferably, further, in S200, selecting the form guide transmitted from each client as a training data set also includes: performing data monitoring on the training data set, and the specific steps include: when the total number of approximate completed items in the set ASET that are greater than BSETBig is greater than the total number of approximate completed items in the set ASET that are less than BSetSho, if the sum of the approximate completed items in the set BSET that are greater than ASETBig is greater than the sum of the approximate completed items in BSET that are less than ASETSho, then all forms in the training data set are deleted.

[0032] Among them, selecting the form guide transmitted from each client as the training data set can accurately filter out all forms with low probability of hitting in the form presentation of all clients in the decentralized scenario, greatly reducing the probability of data conflicts, contradictions, and errors. By monitoring the training data set, when the training data set has an extremely high hit rate, all forms that may have errors in the training data set in the decentralized scenario can be cleared in time, thereby adjusting the accuracy of the model during training and ensuring the recall rate of the pre-trained model in subsequent steps.

[0033] Among them, the user text data to be recommended includes: user characteristic information in text format; the user characteristic information includes user name, academic credit grade point average, intended region, education level, language score, graduation school, graduation major, competition award experience, internship experience and graduation time, among which the required items are: user name, academic credit grade point average, intended region, education level, language score, graduation school, graduation major.

[0034] The present invention also provides a personalized study abroad consulting report intelligent generation system, the system comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor executing the computer program to run in the following system units:

[0035] A form creation unit, used to create a form on each client, wherein the form includes a plurality of form items, each of which corresponds one-to-one to feature information in the user text data;

[0036] A guide selection unit, used for selecting form guides transmitted from various clients as training data sets;

[0037] A model training unit, used to train a neural network model using a training data set to obtain a pre-trained model;

[0038] The model application unit is used to input the user text data to be recommended into the pre-trained model to obtain the success rate of studying abroad information and push the studying abroad information.

[0039] The beneficial effects of the present invention are as follows: the present invention provides a personalized study abroad consulting report intelligent generation method and system, which selects the form guide transmitted from each client as a training data set, and can accurately screen out all forms with low probability of hitting in the form presentation of all clients, greatly reducing the probability of data conflicts, contradictions, and errors, and by monitoring the training data set, when the training data set has an extremely high hit rate, all forms that may have errors in the training data set can be cleared in time, thereby adjusting the accuracy of the model during training and ensuring the recall rate of the pre-trained model in subsequent steps. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The above and other features of the present invention will become more obvious through a detailed description of the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings of the present invention represent the same or similar approximate completion items. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these accompanying drawings without creative work. In the accompanying drawings:

[0041] Figure 1 Shown is a flow chart of a method for intelligently generating a personalized study abroad consulting report;

[0042] Figure 2 Shown is a structural diagram of an intelligent generation system for personalized study abroad consulting reports. DETAILED DESCRIPTION

[0043] The following will be combined with the embodiments and drawings to clearly and completely describe the concept, specific structure and technical effects of the present invention, so as to fully understand the purpose, scheme and effect of the present invention. It should be noted that the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0044] like Figure 1 The figure shows a flowchart of a method for intelligently generating a personalized study abroad consulting report. Figure 1 A personalized study abroad consulting report intelligent generation method according to an embodiment of the present invention is described below. The method comprises the following steps:

[0045] S100, creating a form on each client, the form including a plurality of form items, each form item corresponding one-to-one to feature information in user text data;

[0046] S200, selecting the form guide transmitted from each client as a training data set;

[0047] S300, training the neural network model using the training data set to obtain a pre-trained model;

[0048] S500: Input the user text data to be recommended into the pre-trained model to obtain the success rate of the study abroad information and push the study abroad information.

[0049] Furthermore, the method further includes S600, using the user text data to be recommended and the corresponding success rate and information of studying abroad as the text of the study abroad consultation report.

[0050] Among them, study abroad information includes: the study abroad institutions and study abroad majors to be applied for.

[0051] Further, in step S100, the user sends user text data to the server via a client, wherein the client includes a mobile device or a desktop computer.

[0052] The characteristic information in the user text data includes: user characteristic information in text format; the user characteristic information includes any one or more of the following: user name, academic credit points, intended region, education level, language scores, graduation school, graduation major, competition award experience, internship experience and graduation time. It may also include: overseas study school and overseas study major information.

[0053] Further, in S300, the neural network model is a SA-NET neural network model.

[0054] Furthermore, in S200, the method of selecting the form guides transmitted from each client as the training data set includes:

[0055] The time when the current client creates the form is recorded as the table creation time, and the time when all form items in the form are filled in or the time when the form is transmitted to the server is recorded as the completion time; the period from the table creation time to the completion time is recorded as the form pointing time of the form;

[0056] During the form pointing time, the number of completed form items in all clients that are identical to any one or more form items in the form of the current client is the approximate number of completed items CompN;

[0057] Construct empty sets ASET and BSET respectively, record each moment when CompN>0 in the form pointing time as the conflict moment; let the mean of the difference between the durations of each conflict moment be MEANGAP; if the current conflict moment is CAT; then in the form pointing time, record the next conflict moment of CAT as CBT, and the next conflict moment of CBT as CCT in the time sequence;

[0058] For each approximate number of completed items, if the CCT moment is within the time period from the CAT moment to the CBT+MEANGAP moment within the form pointing time, the approximate number of completed items CAN at the moment CAT within the form pointing time, the approximate number of completed items CBN at the moment CBT, and the approximate number of completed items CCN at the moment CCT within the form pointing time are sequentially added to the set ASET; otherwise, CAN, CBN, and CCN are sequentially added to the set BSET (the set ASET contains the approximate number of completed items with a high hit probability within the form pointing time, and the set BSET contains the approximate number of completed items with a general hit probability within the form pointing time);

[0059] ASET(i) represents the number of approximately completed items in the i-th set ASET, where i is the sequence number;

[0060] Guide selection for each form, including the following steps:

[0061] For each ASET(i), if the condition is satisfied: BSetMean≥ASET(i)≥BSetSho, or, if the condition is satisfied: ASET(i)<BSetSho+JustIndi×BSETBig, and if the corresponding form is not in the training data set, then add the form to the training data set;

[0062] For each ASET(i), if the condition is satisfied: BSetMean<ASET(i)<BSETBig, or, if the condition is satisfied: ASET(i)≥BSetSho+JustIndi×BSETBig, and if the corresponding form already exists in the training data set, then delete the form from the training data set;

[0063] Where, JustIndi = exp(ASETSho÷ASetMean) / exp(ASetMean÷ASETBig);

[0064] Among them, exp represents the natural index, JustIndi is the form hit index; BSetMean is the mean of the number of approximately completed items in the set BSET; BSetSho represents the minimum number of approximately completed items in the set BSET; BSETBig represents the maximum number of approximately completed items in the set BSET; ASetMean is the mean of the number of approximately completed items in the set ASET; ASETSho represents the minimum number of approximately completed items in the set ASET; ASETBig represents the maximum number of approximately completed items in the set ASET.

[0065] Preferably, further, in S200, selecting the form guide transmitted from each client as a training data set also includes: performing data monitoring on the training data set, and the specific steps include: when the total number of approximate completed items in the set ASET that are greater than BSETBig is greater than the total number of approximate completed items in the set ASET that are less than BSetSho, if the sum of the approximate completed items in the set BSET that are greater than ASETBig is greater than the sum of the approximate completed items in BSET that are less than ASETSho, then all forms in the training data set are deleted.

[0066] Among them, the user text data to be recommended includes: user characteristic information in text format; the user characteristic information includes user name, academic credit points, intended region, education level, language scores, graduation school, graduation major, competition award experience, internship experience and graduation time.

[0067] The embodiment of the present invention provides a personalized study abroad consulting report intelligent generation system, such as Figure 2 The figure shows a structure diagram of a personalized study abroad consulting report intelligent generation system of the present invention. A personalized study abroad consulting report intelligent generation system of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned personalized study abroad consulting report intelligent generation system embodiment are implemented.

[0068] The system comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to run in the following units of the system:

[0069] A form creation unit, used to create a form on each client, wherein the form includes a plurality of form items, each of which corresponds one-to-one to feature information in the user text data;

[0070] A guide selection unit, used for selecting form guides transmitted from various clients as training data sets;

[0071] A model training unit, used to train a neural network model using a training data set to obtain a pre-trained model;

[0072] The model application unit is used to input the user text data to be recommended into the pre-trained model to obtain the success rate of studying abroad information and push the studying abroad information.

[0073] The personalized study abroad consulting report intelligent generation system can be run in computing devices such as desktop computers, notebooks, PDAs, and cloud servers. The system that can run the personalized study abroad consulting report intelligent generation system may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the example is only an example of a personalized study abroad consulting report intelligent generation system, and does not constitute a limitation on a personalized study abroad consulting report intelligent generation system. It can include more or fewer components than the example, or combine certain components, or different components. For example, the personalized study abroad consulting report intelligent generation system may also include input and output devices, network access devices, buses, etc.

[0074] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the operation system of the personalized study abroad consulting report intelligent generation system, and uses various interfaces and lines to connect the various parts of the entire operation system of the personalized study abroad consulting report intelligent generation system.

[0075] The memory can be used to store the computer program and / or module, and the processor implements the various functions of the personalized study abroad consulting report intelligent generation system by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card, etc.

[0076] (SmArray1tMediaCArray1d, SMC), Secure Digital (SecureDigital, SD) card, flash memory card (FlashCArray1d), at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0077] Although the description of the present invention has been quite detailed and has been described in particular with respect to several described embodiments, it is not intended to be limited to any of these details or embodiments or any particular embodiment, so as to effectively cover the intended scope of the present invention. In addition, the present invention is described above with the embodiments foreseeable by the inventors, and its purpose is to provide a useful description, and those non-substantial changes to the present invention that are not currently foreseen may still represent equivalent changes of the present invention.

Claims

1. A method for intelligently generating a personalized study abroad consulting report, characterized in that: The method comprises the following steps: S100, create a form in each client; S200, selecting the form guide transmitted from each client as a training data set; S300, training the neural network model using the training data set to obtain a pre-trained model; S500, inputting the user text data to be recommended into the pre-trained model to obtain the success rate of studying abroad information and push the studying abroad information; In S200, the method of selecting the form guides transmitted from each client as a training data set includes: The time when the current client creates the form is recorded as the table creation time, and the time when all form items in the form are filled in or the time when the form is transmitted to the server is recorded as the completion time; the period from the table creation time to the completion time is recorded as the form pointing time of the form; within the form pointing time, the number of form items in all clients that are the same as any one or more form items in the form of the current client and have been filled in is the approximate number of completed items CompN; Construct empty sets ASET and BSET respectively, record each moment when CompN>0 in the form pointing time as the conflict moment; let the mean of the difference between the durations of each conflict moment be MEANGAP; if the current conflict moment is CAT; then in the form pointing time, record the next conflict moment of CAT as CBT, and the next conflict moment of CBT as CCT in the time sequence; For each approximate number of completed items, if the CCT moment is within the time period from the CAT moment to the CBT+MEANGAP moment within the form pointing time, the approximate number of completed items CAN at the moment CAT within the form pointing time, the approximate number of completed items CBN at the moment CBT, and the approximate number of completed items CCN at the moment CCT are sequentially added to the set ASET; otherwise, CAN, CBN, and CCN are sequentially added to the set BSET; ASET(i) represents the number of approximately completed items in the i-th set ASET, where i is the sequence number; Guide selection for each form; The method for guiding selection of each form specifically includes the following steps: For each ASET(i), if the condition is satisfied: BSetMean≥ASET(i)≥BSetSho, and if the corresponding form is not in the training data set, then add the form to the training data set; For each ASET(i), if the condition is satisfied: BSetMean<ASET(i)<BSETBig, and if the corresponding form already exists in the training data set, then delete the form from the training data set; Among them, BSetMean is the mean of the number of approximately completed items in the set BSET; BSetSho represents the minimum number of approximately completed items in the set BSET; BSETBig represents the maximum number of approximately completed items in the set BSET.

2. According to claim 1, a personalized study abroad consulting report intelligent generation method is characterized in that: The method further includes S600, wherein the user text data to be recommended and the corresponding success rate and information of studying abroad are used as the text of the studying abroad consultation report.

3. The method for intelligently generating a personalized overseas study consulting report according to claim 1, characterized in that: The characteristic information in the user text data includes: user characteristic information in text format; the user characteristic information includes any one or more of the user name, academic credit points, intended region, education level, language scores, graduation school, graduation major, competition award experience, internship experience and graduation time.

4. The method for intelligently generating a personalized overseas study consulting report according to claim 1, characterized in that: In S200, selecting the form guide transmitted from each client as a training data set also includes: performing data monitoring on the training data set, and the specific steps include: when the total number of approximate completed items in the set ASET that is greater than BSETBig is greater than the total number of approximate completed items in the set ASET that is less than BSetSho, if the sum of the approximate completed items in the set BSET that is greater than ASETBig is greater than the sum of the approximate completed items in BSET that is less than ASETSho, then all forms in the training data set are deleted; wherein ASETSho represents the minimum value of the approximate completed items in the set ASET; and ASETBig represents the maximum value of the approximate completed items in the set ASET.

5. A personalized study abroad consulting report intelligent generation system, characterized by: The personalized study abroad consulting report intelligent generation system includes: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the personalized study abroad consulting report intelligent generation method described in any one of claims 1 to 4 are implemented.