Method and system for multi-dimensionally evaluating and recommending managers based on machine learning

Through machine learning technology, combined with transaction behavior and sentiment analysis, a comprehensive performance score is generated, which solves the shortcomings of traditional manager evaluation methods, provides accurate personalized recommendations, and improves the effectiveness of investment decisions and user satisfaction.

CN119005786APending Publication Date: 2024-11-22CHINA CITIC BANK CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411026763.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Traditional manager evaluation and recommendation methods rely on historical performance data, ignore trading behavior patterns and market sentiment tendencies, resulting in incomplete and accurate evaluation results, and it is difficult to provide personalized recommendations in line with users' specific investment preferences and risk tolerance.

Method used

A multi-dimensional evaluation method based on machine learning is adopted to analyze transaction behavior through clustering algorithms, combine sentiment analysis models, generate comprehensive performance scores, and provide accurate manager recommendations based on user preferences. The specific steps include: collecting transaction data and public remarks, using the K-means clustering algorithm and BERT sentiment analysis model, generating transaction behavior labels and emotional tendency analysis results, calculating comprehensive performance scores, and finally generating a recommendation list based on user preferences.

Benefits of technology

It improves the comprehensiveness and accuracy of manager evaluation and recommendations, provides accurate and personalized recommendations, improves user satisfaction and effectiveness of investment decisions, and maintains the timeliness of recommendation results through dynamic updates.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005786A_ABST
    Figure CN119005786A_ABST
Patent Text Reader

Abstract

The invention relates to a method and system for evaluating and recommending managers in multiple dimensions based on machine learning, and the method comprises the steps: collecting transaction data of managers, classifying transaction behaviors of the managers through employing a clustering algorithm, and obtaining transaction behavior labels; collecting public speech of a manager, and performing emotional tendency analysis on the public speech of the manager by using an emotional analysis model; generating a comprehensive performance score for each manager based on the transaction behavior tag, the emotional tendency analysis result and a third evaluation factor; according to preference information selected by a user, managers conforming to user preferences are recommended, and recommendation list information is generated, behavior analysis and emotion analysis are introduced, a multi-angle and deep evaluation model is created, comprehensive performance scoring is utilized, the user preferences are deeply analyzed, and highly-customized manager recommendation is provided. The accurate matching not only improves the satisfaction degree of the user, but also remarkably improves the effectiveness of investment decision.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and system for multi-dimensional evaluation and recommendation of managers based on machine learning. Background Art

[0002] With the rapid development of the global financial market, its complexity and diversity continue to increase. When faced with numerous quantitative products and managers, it is more difficult for investors to choose suitable investment targets. In this context, the evaluation and recommendation methods of managers are of great significance in guiding investors' decision-making.

[0003] Traditional managers’ evaluation and recommendation mainly rely on historical performance data, such as yield, Sharpe ratio and other indicators. However, this method ignores the manager’s trading behavior patterns and market sentiment tendencies, resulting in evaluation results that may not be comprehensive and accurate. In addition, there are major deficiencies in existing personalized recommendations, making it difficult to provide accurate recommendations based on users’ specific investment preferences and risk tolerance. Summary of the invention

[0004] To address the deficiencies of the prior art, the present invention has created a multi-angle, in-depth evaluation model by introducing behavioral analysis and sentiment analysis, and provided highly customized manager recommendations by using comprehensive performance scores and in-depth analysis of user preferences. This precise matching not only improves user satisfaction, but also significantly improves the effectiveness of investment decisions. To achieve the above objectives, the technical solutions adopted by the present invention include:

[0005] A method for multi-dimensional evaluation and recommendation of managers based on machine learning, characterized by comprising:

[0006] S1. Collect the manager's transaction data, use the clustering algorithm to classify the manager's transaction behavior, and obtain the transaction behavior label;

[0007] Specifically, the manager's trading data is collected from financial databases and public information sources, including buying and selling timing, trading frequency and trading strategies.

[0008] Data structure: Each transaction record includes manager ID, transaction time, transaction type (buy / sell), transaction amount, transaction subject, etc.

[0009] Use K-means clustering algorithm to classify managers' trading behaviors.

[0010] Code example:

[0011] from sklearn.cluster import KMeans

[0012] import numpy as np

[0013] #Assume that data is a matrix containing the manager's trading behavior data

[0014] data = np.array([...])

[0015] #Set the number of clusters

[0016] num_clusters=5

[0017] kmeans=KMeans(n_clusters=num_clusters)

[0018] #Train model

[0019] kmeans.fit(data)

[0020] #Get the clustering label of each manager

[0021] labels = kmeans.labels_

[0022] For example, a transaction behavior label can be defined as:

[0023] Tag 1: "High-frequency traders"

[0024] Features: The manager conducts a large number of short-term transactions, buys and sells frequently, and conducts multiple transactions every day.

[0025] Behavior description: Prefers to achieve quick gains in a short period of time and is usually very sensitive to market fluctuations.

[0026] Example: A manager makes more than 100 trades in a month, and the average holding period for each trade does not exceed one day.

[0027] With trading behavior labels, investors can more clearly identify and understand managers’ trading behaviors and strategies. Transparency greatly enhances investors’ confidence in investment decisions and enables them to make more informed choices based on detailed information.

[0028] S2. Collect the public statements of the administrator and use the sentiment analysis model to analyze the sentiment tendency of the public statements of the administrator;

[0029] Specifically, the BERT (Bidirectional Encoder Representations from Transformers) model is used for sentiment analysis. The results of sentiment and public opinion analysis can reflect the performance and decision-making tendencies of managers in market fluctuations, help evaluate their risk management capabilities, optimize investment portfolios based on managers' emotional tendencies, and avoid over-concentration on managers with consistent emotions.

[0030] S3. Generate a comprehensive performance score for each manager based on the trading behavior labels, sentiment analysis results and the third evaluation factor;

[0031] Specifically, the random forest algorithm is used to calculate the manager's comprehensive performance score. The comprehensive performance score includes indicators in multiple dimensions, including rate of return, maximum drawdown, risk management capability, management scale, etc. The random forest regression model is trained and set up using a large amount of crawled data. In addition, the training data of the manager's data crawling is updated regularly to obtain timely comprehensive performance scores, sentiment tendency types, and transaction label types.

[0032] S4. Based on the preference information and comprehensive performance score selected by the user, recommend managers that meet the user's preferences and generate recommendation list information.

[0033] Specifically, based on the survey data submitted by the user through the front end, such as the expected return level, tolerable risk level, tolerable investment time, annual income, family size, age, occupation and other information, the user's preferred manager type (i.e., trading behavior label and emotional tendency type preference) is determined, such as preference for high-frequency traders and emotional neutrals, and a recommendation list is generated based on comprehensive matching of comprehensive performance scores, including managers that meet the user's preferences. The recommendation list contains key information about the manager, such as name, quantitative products managed, trading behavior patterns, emotional tendencies and comprehensive performance scores. The recommendation module also includes

[0034] Exemplarily, S1 collects transaction data of managers, classifies transaction behaviors of managers using a clustering algorithm, and obtains transaction behavior labels, including associating and storing the transaction behavior labels with the manager's ID.

[0035] Specifically, the transaction behavior tags of the clustering results are associated with the manager's ID and stored. One manager ID can be associated with multiple transaction behavior tags. The transaction behavior tags can better understand the manager's investment style and strategy, so as to make more informed investment decisions.

[0036] Exemplarily, S2 also includes using crawler technology to collect public statements of managers from portal websites and company data sources, performing preliminary cleaning, preprocessing and feature extraction on the obtained text data, and storing the processed data in a database; the sentiment analysis model in S2 is a sentiment analysis performed using the BERT model.

[0037] Exemplarily, the S3 generates a comprehensive performance score for each manager based on trading behavior labels, sentiment tendency analysis results and a third evaluation factor, including using a random forest algorithm to train a comprehensive performance scoring model, and using the comprehensive performance scoring model to calculate the manager's comprehensive performance score; the third evaluation factor includes but is not limited to rate of return, maximum drawdown, risk management capability, and management scale.

[0038] Exemplarily, the random forest algorithm is used to train the comprehensive performance scoring model, and the comprehensive performance scoring model is used to calculate the manager's comprehensive performance score. It also includes using a Kafka message queue to regularly update the collected transaction data and the public statements, and retrain the comprehensive performance scoring model based on the newly collected text data.

[0039] Exemplarily, S4 recommends managers who meet the user's preferences based on the preference information submitted by the user, generates recommendation list information, and also provides an interactive interface to help the user provide feedback on the evaluation of the recommended manager.

[0040] The present invention also provides a system for multi-dimensional evaluation and recommendation of managers based on machine learning, the system comprising:

[0041] The data collection module is used to collect the manager's transaction data and public statements;

[0042] Behavior analysis module, used to obtain transaction behavior labels;

[0043] Sentiment analysis module, used to analyze managers’ emotional tendencies;

[0044] A scoring module, which is used to generate a comprehensive performance score for managers;

[0045] The recommendation module is used to generate recommendation list information based on preference information and comprehensive performance scores.

[0046] The present invention also provides a computer-readable storage medium, characterized in that a computer program is stored on the storage medium, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

[0047] The present invention also provides an electronic device, characterized in that it includes a processor and a memory;

[0048] The memory is used to store computer programs;

[0049] The processor is configured to execute the method according to any one of claims 1 to 6 by calling a stored computer program.

[0050] The present invention also relates to a computer program product, comprising a computer program and / or instructions, characterized in that the computer program and / or instructions implement the steps of the above method when executed by a processor.

[0051] The beneficial effects of the present invention are:

[0052] Comprehensive data analysis: By integrating multiple data sources, the comprehensiveness and accuracy of managers' evaluations and recommendations are improved.

[0053] Multi-dimensional evaluation system: Introduce behavioral analysis and sentiment analysis to establish a multi-dimensional manager evaluation system to help investors understand the manager's capabilities and strategies more comprehensively.

[0054] Accurate personalized recommendations: Utilize comprehensive performance scores and user preference analysis to provide accurate personalized recommendations, improve user satisfaction and the effectiveness of investment decisions.

[0055] Dynamic update and optimization: The system can be updated and optimized in real time to reflect market changes and user feedback, and maintain the timeliness and accuracy of recommendation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of the method for multi-dimensional evaluation and manager recommendation based on machine learning of the present invention.

[0057] Figure 2 This is a schematic diagram of the execution logic of the method for multi-dimensional evaluation and recommendation of managers based on machine learning of the present invention.

[0058] Figure 3 This is a schematic diagram of the structure of the system for multi-dimensional evaluation and recommendation of managers based on machine learning in the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0060] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0061] Embodiment 1

[0062] The first aspect of the present invention relates to a multi-dimensional evaluation and manager recommendation process based on machine learning. Figure 1 and Figure 2 The method shown comprises:

[0063] S1. Collect the manager's transaction data, use the clustering algorithm to classify the manager's transaction behavior, and obtain the transaction behavior label;

[0064] Specifically, the manager's trading data is collected from financial databases and public information sources, including buying and selling timing, trading frequency and trading strategies.

[0065] Data structure: Each transaction record includes manager ID, transaction time, transaction type (buy / sell), transaction amount, transaction subject, etc.

[0066] Use K-means clustering algorithm to classify managers' trading behaviors.

[0067] Code example:

[0068] from sklearn.cluster import KMeans

[0069] import numpy as np

[0070] #Assume that data is a matrix containing the manager's trading behavior data

[0071] data = np.array([...])

[0072] #Set the number of clusters

[0073] num_clusters=5

[0074] kmeans=KMeans(n_clusters=num_clusters)

[0075] #Train model

[0076] kmeans.fit(data)

[0077] #Get the clustering label of each manager

[0078] labels = kmeans.labels_

[0079] For example, a transaction behavior label can be defined as:

[0080] Tag 1: "High-frequency traders"

[0081] Features: The manager conducts a large number of short-term transactions, buys and sells frequently, and conducts multiple transactions every day.

[0082] Behavior description: Prefers to achieve quick gains in a short period of time and is usually very sensitive to market fluctuations.

[0083] Example: A manager makes more than 100 trades in a month, and the average holding period for each trade does not exceed one day.

[0084] Tag 2: "Long-term investors"

[0085] Features: The manager's trading frequency is low and the holding period is long, usually holding assets for months or even years.

[0086] Behavior description: Focus on long-term investment returns and tend to choose stocks or quantitative products with good fundamentals for long-term holding.

[0087] Example: A manager has only made 10 trades in the past year, and the average holding period for each trade is more than half a year.

[0088] Tag 3: "Market Follower"

[0089] Features: The manager's trading behavior is highly correlated with market trends, and he tends to buy when the market rises and sell when the market falls.

[0090] Behavior description: Trading behavior is relatively passive, and investment decisions are mainly based on market trends.

[0091] Example: A manager increases his holdings when the market rises and reduces his holdings when the market falls, and has a high correlation with market trends.

[0092] Tag 4: "Contrarian Investor"

[0093] Features: The manager's trading behavior is opposite to the market trend, tending to buy when the market falls and sell when the market rises.

[0094] Behavior description: The trading behavior is relatively aggressive, attempting to obtain excess returns through reverse operations.

[0095] Example: A manager increases his holdings when the market is generally pessimistic and reduces his holdings when the market is generally optimistic, adopting an investment strategy that goes against the market trend.

[0096] Tag 5: "Event-driven investors"

[0097] Features: The manager's trading behavior is driven by specific events, such as company financial report releases, policy changes, mergers and acquisitions, and restructuring.

[0098] Behavior description: The transaction frequency is not fixed, and large-scale transactions are mainly carried out when major events occur.

[0099] Example: A manager buys a large amount of stock when the company releases major positive news and sells it quickly when negative news appears. The trading behavior is closely related to specific events.

[0100] S2. Collect the public statements of the administrator and use the sentiment analysis model to analyze the sentiment tendency of the public statements of the administrator;

[0101] Use the BERT (Bidirectional Encoder Representations from Transformers) model for sentiment analysis. Code example:

[0102] #Load the pre-trained BERT sentiment analysis model

[0103] sentiment_analysis=pipeline("sentiment-analysis")

[0104] #Assume that texts is the Chinese text data list of the administrator

[0105] texts=[

[0106] "The recent market declines are just temporary setbacks. I believe the fundamentals of the economy are strong and we will see a rebound soon."

[0107] "While there are some risks in the current geopolitical situation, we remain cautiously optimistic about the growth potential of the technology sector."

[0108] "Our portfolio performed below expectations last quarter due to unforeseen market volatility. We are re-evaluating our strategy to mitigate these risks."

[0109] "Our new investments in renewable energy projects have been warmly welcomed by investors and the public. We are committed to achieving sustainable growth."

[0110] "Just had a very good meeting with our stakeholders. The prospects are exciting as we plan to launch new financial products that meet the changing market needs." ]

[0112] #Analyze the sentiment of each text

[0113] results=sentiment_analysis(texts)

[0114] # Output sentiment analysis results

[0115] for i, result in enumerate(results):

[0116] print(f"text:{texts[i]}")

[0117] print(f"Sentiment: {result['label']}, Rating: {result['score']: .2f}\n")

[0118] Output example

[0119] Text: The recent market decline is only a temporary setback. I believe the fundamentals of the economy are strong and we will see a rebound soon.

[0120] Sentiment: Positive, Rating: 0.95

[0121] Text: While there are some risks in the current geopolitical situation, we remain cautiously optimistic about the growth potential of the technology industry.

[0122] Sentiment: Neutral, Rating: 0.65

[0123] Text: Our portfolio performed below expectations last quarter due to unforeseen market volatility. We are re-evaluating our strategy to mitigate these risks.

[0124] Sentiment: Negative, Rating: 0.85

[0125] Text: Our new investments in renewable energy projects have been warmly welcomed by investors and the public. We are committed to achieving sustainable growth.

[0126] Sentiment: Positive, Rating: 0.92

[0127] Text: Just had a very good meeting with our stakeholders. The prospects are exciting as we plan to launch new financial products that meet the changing market needs.

[0128] Sentiment: Positive, Rating: 0.88

[0129] S3. Generate a comprehensive performance score for each manager based on the trading behavior labels, sentiment analysis results and the third evaluation factor;

[0130] The random forest algorithm is used to calculate the manager's comprehensive performance score, which includes indicators in multiple dimensions, such as yield, maximum drawdown, risk management ability, management scale, etc. Specific examples are as follows:

[0131] from sklearn.ensemble import RandomForestRegressor

[0132] import numpy as np

[0133] import pandas as pd

[0134] # Sample data, a lot of data is actually needed for training

[0135] data={

[0136] 'Yield':[0.15,0.10,0.20,0.12,0.08],#Unit: percentage

[0137] 'Maximum Drawdown':[-0.05,-0.03,-0.08,-0.04,-0.02],#Unit: Percent

[0138] 'Risk management ability':[3,2,1,3,2],#High: 3,Medium: 2,Low: 1

[0139] 'Management scale':[10,8,12,15,6],#Unit: 100 million yuan

[0140] 'Trading behavior pattern':[2,1,4,2,3],#High frequency traders = 1, long term investors = 2, contrarian investors = 4, market followers = 3

[0141] 'Emotional tendency': [1,0,-1,1,0]# Positive = 1, Neutral = 0, Negative = -1

[0142] }

[0143] #Overall performance score

[0144] performance=[85,75,65,90,70]# Input the score based on the manager's performance and expert opinions

[0145] #Create DataFrame

[0146] df = pd.DataFrame(data)

[0147] X = df.values

[0148] y = np.array(performance)

[0149] #Set up comprehensive performance scoring model

[0150] model=RandomForestRegressor(n_estimators=100,max_depth=10,random_state=42)

[0151] #Train model

[0152] model.fit(X,y)

[0153] # Example prediction data

[0154] new_data = np.array([[0.14,-0.04,3,11,2,1]])#Corresponding to a manager's rate of return, maximum drawdown, risk management ability, management scale, trading behavior pattern, and emotional tendency

[0155] Predicted_performance = model.predict(new_data) #Automatically calculate the score based on the trained model

[0156] Output example

[0157] Overall performance score: 82.50

[0158] S4. Based on the preference information and comprehensive performance score selected by the user, recommend managers that meet the user's preferences and generate recommendation list information.

[0159] Embodiment 2

[0160] Step 1: Crawl data to analyze manager behavior information and collect trading data of all public fund managers from multiple financial databases and public information sources, including buying and selling timing, trading frequency, trading strategies, and public statements.

[0161] For example, the administrator Li Ming:

[0162] Basic information: Manages 3 funds with a total management scale of 1.5 billion and has high risk management capabilities.

[0163] Transaction data: The transaction frequency was high in the past year, with about 30 transactions per quarter.

[0164] Market Comments: 1 recently published market comment

[0165] Return during tenure: 15%

[0166] Maximum drawdown during tenure: -5%

[0167] Administrator Wang Gang:

[0168] Basic information: Manages 2 funds with a total management scale of 1 billion and has medium risk management capabilities.

[0169] Transaction data: The transaction frequency in the past year was moderate, with about 10 transactions per quarter.

[0170] Market Comments: 2 recently published market comments

[0171] Return during tenure: 36%

[0172] Maximum drawdown during tenure: -15%

[0173] Administrator Liu Yi:

[0174] Basic information: Manages 4 funds with a total management scale of 2 billion and has low risk management capabilities.

[0175] Transaction data: The transaction frequency was low in the past year, with approximately 5 transactions per quarter.

[0176] Market Comments: 1 recently published market comment

[0177] Return during tenure: 23%

[0178] Maximum drawdown during tenure: -10%

[0179] Data preprocessing: removing duplicate records, filling missing values, feature extraction:

[0180] Manager Li Ming: Trading frequency (high)

[0181] Manager Wang Gang: Trading frequency (medium)

[0182] Manager Liu Yi: Trading frequency (low)

[0183] Data storage, the cleaned data is stored in the database, forming the following table:

[0184]

[0185] Step 2: Manager’s emotional tendency analysis:

[0186] Li Ming: The recent market decline is just a temporary setback. I believe the fundamentals of the economy are strong and we will see a rebound soon.

[0187] Wang Gang Although there are some risks in the current geopolitical situation, we remain cautiously optimistic about the growth potential of the technology industry.

[0188] Liu Yi: Due to unforeseen market fluctuations, our portfolio performed below expectations last quarter. We are re-evaluating our strategy to mitigate these risks.

[0189] Behavior analysis uses cluster analysis to classify the trading behaviors of all managers and train a model, for example:

[0190] Manager Li Ming: Long-term investor

[0191] Manager Wang Gang: Market follower

[0192] Manager Liu Yi: High-frequency trader

[0193] Emotional public opinion analysis: Use the sentiment analysis model to analyze the market comments of all managers and train the model, for example, to obtain:

[0194] Manager Li Ming: Positive emotions (score 0.9)

[0195] Manager Wang Gang: Neutral sentiment (score 0.5)

[0196] Manager Liu Yi: Negative emotions (score 0.2)

[0197] Step 3: Comprehensive scoring: Combine the historical performance, trading behavior patterns and sentiment analysis results of typical managers (such as well-known managers with excellent, average and poor performance), and train a comprehensive performance scoring model through the random forest algorithm to calculate the newly input manager data and calculate the comprehensive performance score.

[0198] Enter the performance data, behavior analysis labels, sentiment analysis results and other data of managers Li Ming, Wang Gang and Liu Yi to calculate the scores:

[0199] Manager Li Ming

[0200] Overall performance score: 72

[0201] Administrator Wang Gang

[0202] Overall performance score: 56

[0203] Administrator Liu Yi

[0204] Overall performance score: 23

[0205] Step 4: Personalized recommendation, analyzing user preferences. Assume that the survey results entered by the user through the front end are:

[0206] Expected return level: Stable

[0207] Tolerable risk level: medium to low risk

[0208] Affordable investment period: 5-10 years

[0209] Annual income: 800,000

[0210] Family size: 5

[0211] Age: 40

[0212] Identify managers who prefer long-term investments and stable sentiment (neutral or positive sentiment).

[0213] Based on the analysis and matching of user preference information and comprehensive performance scores, eligible managers are screened out and ranked for recommendation:

[0214] Manager Li Ming: Overall score 72, positive sentiment, recommended

[0215] Manager Wang Gang: Overall score 56, neutral sentiment, recommended

[0216] Manager Liu Yi: Overall score 23, negative sentiment, not recommended

[0217] Recommended results

[0218] Manager Li Ming: Overall score 72, positive sentiment, long-term investor

[0219] Manager Wang Gang: Overall score 56, neutral sentiment, market follower

[0220] Embodiment 3

[0221] Another aspect of the present invention also relates to a system for evaluating and recommending managers based on multi-dimensional machine learning, the structure of which is as follows: Figure 3 As shown, including:

[0222] The data collection module is used to collect the manager's transaction data and public statements.

[0223] Behavior analysis module, used to obtain transaction behavior labels.

[0224] Sentiment analysis module, used to analyze managers’ emotional tendencies;

[0225] A scoring module, which is used to generate a comprehensive performance score for managers;

[0226] The recommendation module is used to generate recommendation list information based on preference information and comprehensive performance scores.

[0227] The data collection module also includes:

[0228] Data standardization module: performs pre-processing operations such as cleaning, word segmentation, and removal of stop words on the collected raw data, and converts it into a unified format for subsequent processing and analysis.

[0229] The behavior analysis module also includes:

[0230] Behavioral feature extraction module, used to extract key features from transaction data, such as transaction frequency, position holding time, etc.;

[0231] The transaction behavior classification module is used to classify the transaction behavior data using a clustering algorithm and determine the transaction behavior label;

[0232] Label storage module: associates the clustering results with the manager ID and stores them in the database.

[0233] The sentiment analysis module also includes:

[0234] The sentiment tendency analysis module is used to perform sentiment analysis on the pre-processed text using pre-trained models such as BERT to determine the sentiment tendency type;

[0235] The sentiment feature quantification module is used to quantify the sentiment analysis results into numerical values ​​for subsequent comprehensive performance scoring.

[0236] The scoring module also includes:

[0237] Performance indicator definition module, used to define the third evaluation factor, including rate of return, maximum drawdown, risk management capability, management scale, etc.;

[0238] The comprehensive performance scoring model training module is used to train the comprehensive performance scoring model using algorithms such as random forest.

[0239] The recommendation module also includes:

[0240] User preference analysis module, used to analyze the preference information submitted by users and determine the transaction behavior labels and emotional tendency types of managers preferred by users;

[0241] The matching module is used to generate recommendation list information using a matching algorithm based on the user's preference information and comprehensive performance score;

[0242] The screening and generation module is used for users to screen and sort the recommendation list information according to different indicators (such as comprehensive performance score, transaction behavior pattern, and emotional tendency type);

[0243] The user feedback evaluation module is used to provide an interactive interface to collect user feedback on recommendation results.

[0244] By using this system, the above-mentioned operation processing method can be executed and the corresponding technical effects can be achieved.

[0245] Embodiment 4

[0246] An embodiment of the present invention also provides a computer-readable storage medium capable of implementing all the steps in the method in the above embodiment, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, all the steps in the method in the above embodiment are implemented.

[0247] Embodiment 5

[0248] An embodiment of the present invention also provides an electronic device for executing the above method. As an implementation device of the method, the electronic device has at least a processor and a memory, and in particular, the memory stores data and related computer programs required for executing the method, such as storage, etc., and the processor calls the data and programs in the memory to execute all the steps of the implementation method and obtains the corresponding technical effect.

[0249] Preferably, the electronic device may include a bus architecture, the bus may include any number of interconnected buses and bridges, and the bus will include various circuits linked together by one or more processors and memories. The bus may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are all well known in the art and are therefore not further described herein. The bus interface provides an interface between the bus and the receiver and transmitter. The receiver and transmitter may be the same component, i.e., a transceiver, providing a unit for communicating with various other systems over a transmission medium. The processor is responsible for managing the bus and general processing, while the memory may be used to store data used by the processor when performing operations.

[0250] In addition, the electronic device may further include components such as a communication module, an input unit, an audio processor, a display, and a power supply. The processor (or controller, operation control) used may include a microprocessor or other processor device and / or a logic device, which receives input and controls the operation of various components of the electronic device; the memory may be one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory, or other suitable devices, which may store the above-mentioned related data information, and may also store a program for executing related information, and the processor may execute the program stored in the memory to realize information storage or processing, etc.; the input unit is used to provide input to the processor, for example, it may be a key or a touch input device; the power supply is used to provide power to the electronic device; the display is used to display display objects such as images and text, for example, it may be an LCD display. The communication module is a transmitter / receiver that sends and receives signals via an antenna. The communication module (transmitter / receiver) is coupled to the processor to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal. Based on different communication technologies, multiple communication modules may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless local area network module, etc. The communication module (transmitter / receiver) is also coupled to a speaker and a microphone via an audio processor to provide an audio output via the speaker and receive an audio input from the microphone, thereby realizing a common telecommunication function. The audio processor may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor is also coupled to a central processor, so that recording can be performed on the machine through the microphone, and the sound stored on the machine can be played through the speaker.

[0251] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, storage media, electronic devices or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0252] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A system that specifies the functions of a box or multiple boxes.

[0253] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a product including an instruction system, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0254] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps of the functions specified in a box or multiple boxes. Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they know the basic creative concept. Therefore, the attached claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0255] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for evaluating and recommending managers in multiple dimensions based on machine learning, characterized in that: The following steps are involved: S1. Collect the manager's transaction data, use the clustering algorithm to classify the manager's transaction behavior, and obtain the transaction behavior label; S2. Collect the public statements of the administrator and use the sentiment analysis model to analyze the sentiment tendency of the public statements of the administrator; S3. Generate a comprehensive performance score for each manager based on the trading behavior labels, sentiment analysis results and the third evaluation factor; S4. Based on the preference information and comprehensive performance score selected by the user, recommend managers that meet the user's preferences and generate recommendation list information.

2. The method according to claim 1, characterized in that S1 collects the manager's transaction data, uses a clustering algorithm to classify the manager's transaction behavior, and obtains transaction behavior labels. This includes associating and storing the transaction behavior label with the manager's ID.

3. The method according to claim 2, characterized in that S2 also includes using crawler technology to collect public statements of managers from portal websites and company data sources, performing preliminary cleaning, preprocessing and feature extraction on the obtained text data, and storing the processed data in a database; The sentiment analysis model in S2 is a sentiment analysis performed using the BERT model.

4. The method according to claim 3, characterized in that The S3 generates a comprehensive performance score for each manager based on the transaction behavior labels, the sentiment tendency analysis results and the third evaluation factor, including using a random forest algorithm to train a comprehensive performance score model and using the comprehensive performance score model to calculate the manager's comprehensive performance score; The third evaluation factors include but are not limited to rate of return, maximum drawdown, risk management capability, and management scale.

5. The method according to claim 4, characterized in that The method of using a random forest algorithm to train a comprehensive performance scoring model and using the comprehensive performance scoring model to calculate the comprehensive performance score of the manager also includes using a Kafka message queue to regularly update the collected transaction data and public statements, and retraining the comprehensive performance scoring model based on the newly collected text data.

6. The method according to any one of claims 1 to 5, characterized in that: The S4 recommends managers that meet the user's preferences based on the preference information and comprehensive performance scores submitted by the user, generates recommendation list information, and also provides an interactive interface to help users provide feedback on their evaluation of the recommended managers.

7. A system for evaluating and recommending managers in multiple dimensions based on machine learning, characterized in that: The system includes: The data collection module is used to collect the manager's transaction data and public statements; Behavior analysis module, used to obtain transaction behavior labels; Sentiment analysis module, used to analyze managers’ emotional tendencies; A scoring module, which is used to generate a comprehensive performance score for managers; The recommendation module is used to generate recommendation list information based on preference information and comprehensive performance scores.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store computer programs; The processor is configured to execute the method according to any one of claims 1 to 6 by calling a stored computer program.

10. A computer program product, characterized in that The method comprises a computer program and / or an instruction, wherein the computer program and / or the instruction implements the steps of the method according to any one of claims 1 to 6 when the computer program and / or the instruction is executed by a processor.