Investment behavior analysis system based on social network aggregation

Abstract

The present invention relates to the technical field of investment behavior analysis, and discloses an investment behavior analysis system based on social network aggregation, including: a communication information collection module, which is used to obtain the communication information and investment information of each investor on the investment communication platform in the past T months; a communication information processing module, which is used to perform semantic analysis on the communication information, and determine the investment project and recommendation value corresponding to the communication information. Before recommending financial products to each investor, the present invention quantifies the influence value and the influenced value of each investor, and fully considers the influence of the communication information of investors on the investment communication platform on each investor. Compared with the general method of analyzing the investment behavior of investors, the present invention fully considers the influence of the communication between investors on the subsequent investment of investors, so the analysis of the investment behavior of investors by the present invention is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of investment behavior analysis, and more specifically, to an investment behavior analysis system based on social network aggregation and distribution. Background Art

[0002] Investment exchange platforms are a common investment channel for most investors. During the operation of investment exchange platforms, in order to increase trading volume and improve investors' user experience, they often recommend the most likely financial products to investors, so it is necessary to analyze investors' investment behavior.

[0003] However, a common analysis method is to analyze based on information such as investors' historical investment behavior and browsing history, without fully considering the impact of other investors on the target investors, resulting in inaccurate analysis of the target investors' investment behavior. Summary of the invention

[0004] The present invention provides an investment behavior analysis system based on social network aggregation and distribution, which solves the technical problem that the results of analyzing the investment behavior of target investors in the above-mentioned background technology are not accurate enough.

[0005] The present invention provides an investment behavior analysis system based on social network aggregation and distribution, comprising:

[0006] A communication information collection module is used to obtain the communication information and investment information of each investor on the investment communication platform in the past T months;

[0007] A communication information processing module, which is used to perform semantic analysis on the communication information and determine the investment projects and recommended values ​​corresponding to the communication information;

[0008] An investment behavior analysis module, which is used to analyze the influence value and the influenced value of each investor based on the communication information and investment information of each investor on the investment communication platform in the past T months;

[0009] The investment behavior prediction module is used to predict the probability of the target investor investing in each financial product based on the influence value and influenced value of each investor in the past T months;

[0010] T is a custom parameter.

[0011] Furthermore, the method for obtaining the communication information of each investor on the investment communication platform in the past T months is as follows:

[0012] Step S101, obtaining a message sent by an investor;

[0013] Step S102, obtaining messages browsed by investors on the investment exchange platform, presetting a standard browsing time, and determining whether the time for investors to browse messages exceeds the standard browsing time, where the standard browsing time is a custom parameter;

[0014] If yes, keep it;

[0015] If not, delete it.

[0016] Furthermore, the specific steps of semantic analysis of communication information are as follows:

[0017] Step S201, removing irrelevant characters, punctuation marks and stop words in the communication information;

[0018] Step S202, constructing an investment character library, extracting stems according to the content in the investment character library, and deleting communication information irrelevant to investment;

[0019] The investment character library includes the names, numbers and abbreviations of financial products and common investment terms;

[0020] Step S203: Use the attention mechanism to analyze the communication information to obtain the semantics of the communication information.

[0021] Furthermore, the steps for analyzing the impact value of each investor are as follows:

[0022] Step S301, obtaining the reading volume of the communication information released by each investor;

[0023] Step S302, presetting a first reading volume, and determining whether the reading volume of the communication information released by each investor in the past T months exceeds the first reading volume;

[0024] If yes, then mark as first investor;

[0025] If not, then record as a second investor;

[0026] Step S303, randomly selecting a first investor, and determining whether a second investor who has read the communication information published by the first investor has made an investment behavior consistent with the communication information;

[0027] If yes, then it will be included in the affected queue;

[0028] Step S304, calculating the influence value of the investor, the calculation formula is as follows:

[0029] ;

[0030] In the formula, represents the influence value of the i-th investor; represents the total number of readings of the communication information released by the i-th investor in the t-th month; It represents the sum of the number of investors who took consistent actions after reading the investor information; and Respectively express about The first and second weight parameters of ; and These are custom parameters.

[0031] Furthermore, the specific steps for analyzing the impact value of each investor are as follows:

[0032] Step S311, obtaining the number of times each investor has made investment actions following the communication information released by the first investor in the past T months;

[0033] Step S312, obtaining the number of times each investor participated in the communication in the past T months;

[0034] Step S313, calculating the impact value of the investor, the calculation formula is as follows:

[0035] ;

[0036] In the formula, represents the impact value of the i-th investor; It indicates the number of times the i-th investor made investment behaviors following the communication information released by the first investor in the past T months; represents the number of times the i-th investor participated in the communication in the past T months; and Respectively express about The first and second weight parameters of ; and These are custom parameters.

[0037] Furthermore, the steps to predict the probability of the target investor investing in each financial product are as follows:

[0038] Step S401, constructing investment graph network data;

[0039] The investment graph network data includes a first node and edges between nodes;

[0040] The first node is generated based on the mapping between the investor’s influence value and the influenced value;

[0041] The existence of edges between nodes indicates: an investor makes consistent investment behavior after viewing the communication information released by another investor, an investor views the communication information released by another investor with the same meaning more than n times, or the number of communications between the two investors exceeds m times and they maintain consistent views;

[0042] Step S402, input the investment graph network data into the investment analysis network model, and output the probability of the target investor investing in each financial product.

[0043] Further, the investment analysis network model includes a first hidden layer and a classifier;

[0044] The first hidden layer inputs the investment graph network data and outputs the first updated feature;

[0045] The classifier inputs the first updated feature and outputs the probability that the target investor invests in each financial product.

[0046] Furthermore, the first hidden layer of the investment analysis network model is constructed based on the graph neural network model.

[0047] Furthermore, the calculation formula of the first hidden layer of the investment analysis network model is as follows:

[0048] ;

[0049] In the formula, Indicated in Layer Time Node The eigenvector of Representation and Nodes The set of nodes with which there are edges; Representation and Nodes The node with the edge is Characteristics of the layer; Indicated in Layer Time Node The feature vector of .

[0050] Furthermore, the training samples used to train the investment analysis network model include:

[0051] Randomly select an investor who has invested for 2T months, select T months as the training period, and obtain the investment behavior and communication information of the investor during the training period as sample data of a training sample;

[0052] After the training period, the financial products invested by the investor are the sample labels of the training samples;

[0053] Repeat the above steps until the number reaches K;

[0054] K is a custom parameter.

[0055] The beneficial effect of the present invention is that before recommending financial products to each investor, the influence value and the influenced value of each investor are quantified, and the influence of the communication information of investors on the investment communication platform on each investor is fully considered. Compared with the general method of analyzing the investment behavior of investors, the influence of the communication between investors on the subsequent investment of investors is fully considered, so the present invention is more accurate in analyzing the investment behavior of investors. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is the overall flow chart of the present invention;

[0057] Figure 2 is a flowchart of obtaining the communication information of each investor on the investment communication platform in the past T months of the present invention;

[0058] Figure 3 It is a flowchart of the semantic analysis of communication information of the present invention;

[0059] Figure 4 is a flowchart of analyzing the impact value of each investor of the present invention;

[0060] Figure 5 It is a flowchart of analyzing the impact value of each investor of the present invention;

[0061] Figure 6 It is a flowchart of the present invention for predicting the probability of target investors investing in each financial product.

[0062] In the figure: 10, communication information acquisition module; 20, communication information processing module; 30, investment behavior analysis module; 40, investment behavior prediction module. DETAILED DESCRIPTION

[0063] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0064] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in one or more embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] like Figure 1-Figure 6 As shown, in one embodiment of the present invention, a common method for analyzing an investor's investment behavior is to analyze based on information such as the user's assets and historical investment behavior. However, this method does not take into account the mutual influence between investors. Therefore, an investment behavior analysis system based on social network distribution is proposed, including a communication information collection module 10, a communication information processing module 20, an investment behavior analysis module 30 and an investment behavior prediction module 40.

[0066] First, the communication information collection module 10 is used to obtain the communication information and investment information of each investor on the investment communication platform in the past T months;

[0067] The method for obtaining the investment information of each investor on the investment exchange platform in the past T months is as follows:

[0068] When an investor conducts investment behavior on the investment exchange platform, the investment exchange platform records the investor's investment behavior, and then the exchange information collection module 10 collects and obtains it;

[0069] The method for obtaining the communication information of each investor on the investment communication platform in the past T months is as follows:

[0070] Step S101, obtaining a message sent by an investor;

[0071] Step S102, obtaining messages browsed by investors on the investment exchange platform, presetting a standard browsing time, and determining whether the time for investors to browse messages exceeds the standard browsing time, the standard browsing time being a custom parameter, preferably, the standard browsing time being 30 seconds;

[0072] If yes, keep it;

[0073] If not, delete it.

[0074] Then, semantic analysis is performed on the communication information of investors. However, in the process of semantic analysis, the communication between investors adopts industry terms or abbreviations, which may cause incomprehensibility. Therefore, the communication information processing module 20 is used to perform semantic analysis on the communication information and determine the investment project and recommendation value corresponding to the communication information;

[0075] The specific steps for semantic analysis of communication information are as follows:

[0076] Step S201, removing irrelevant characters, punctuation marks and stop words in the communication information;

[0077] Step S202, constructing an investment character library, extracting stems according to the content in the investment character library, and deleting communication information irrelevant to investment;

[0078] The investment character library includes the names, numbers and abbreviations of financial products and common investment terms;

[0079] Step S203: Use the attention mechanism to analyze the communication information to obtain the semantics of the communication information.

[0080] It is worth noting that the investment character library is constructed based on the communication information between investors, and the names, numbers and abbreviations of financial products are included in the investment character library in real time, and commonly used investment terms are selected and included in the investment character library, such as: buy, sell, etc.;

[0081] In addition, as the operation time of the investment communication platform increases, the possibility of popular words appearing also greatly increases, so popular words can be included in the investment character library through manual marking. However, manual marking may have a certain delay, so when a character that has not appeared in a certain investment character library appears multiple times, it will be included in the character library, and it will be determined whether the character appears during the next manual marking. If it does not appear, the character will be deleted from the investment character library; if it appears, it will be retained.

[0082] The investment behavior analysis module 30 is used to analyze the influence value and the influenced value of each investor based on the communication information and investment information of each investor on the investment communication platform in the past T months;

[0083] The steps to analyze the impact value of each investor are as follows:

[0084] Step S301, obtaining the reading volume of the communication information released by each investor;

[0085] Step S302, presetting a first reading volume, and determining whether the reading volume of the communication information released by each investor in the past T months exceeds the first reading volume;

[0086] If yes, then mark as first investor;

[0087] If not, then record as a second investor;

[0088] Step S303, randomly selecting a first investor, and determining whether a second investor who has read the communication information published by the first investor has made an investment behavior consistent with the communication information;

[0089] If yes, then it will be included in the affected queue;

[0090] Step S304, calculating the influence value of the investor, the calculation formula is as follows:

[0091] ;

[0092] In the formula, represents the influence value of the i-th investor; represents the total number of readings of the communication information released by the i-th investor in the t-th month; It represents the sum of the number of investors who took consistent actions after reading the investor information; and Respectively express about The first and second weight parameters of ; and These are custom parameters.

[0093] The specific steps to analyze the impact value of each investor are as follows:

[0094] Step S311, obtaining the number of times each investor has made investment actions following the communication information released by the first investor in the past T months;

[0095] Step S312, obtaining the number of times each investor participated in the communication in the past T months;

[0096] Step S313, calculating the impact value of the investor, the calculation formula is as follows:

[0097] ;

[0098] In the formula, represents the impact value of the i-th investor; It indicates the number of times the i-th investor made investment behaviors following the communication information released by the first investor in the past T months; represents the number of times the i-th investor participated in the communication in the past T months; and Respectively express about The first and second weight parameters of ; and These are custom parameters.

[0099] The investment behavior prediction module 40 is used to predict the probability of the target investor investing in each financial product based on the influence value and the influenced value of each investor in the past T months;

[0100] T is a custom parameter. Preferably, T is 24.

[0101] The steps to predict the probability of a target investor investing in each financial product are as follows:

[0102] Step S401, constructing investment graph network data;

[0103] The investment graph network data includes a first node and edges between nodes;

[0104] The first node is generated based on the mapping between the investor’s influence value and the influenced value;

[0105] The existence of edges between nodes indicates: an investor makes consistent investment behavior after viewing the communication information released by another investor, an investor views the communication information released by another investor with the same meaning more than n times, or the number of communications between the two investors exceeds m times and they maintain consistent views;

[0106] Step S402, input the investment graph network data into the investment analysis network model, and output the probability of the target investor investing in each financial product.

[0107] The investment analysis network model includes a first hidden layer and a classifier;

[0108] The first hidden layer inputs the investment graph network data and outputs the first updated feature;

[0109] The classifier inputs the first updated feature and outputs the probability that the target investor invests in each financial product.

[0110] The first hidden layer of the investment analysis network model is built based on the graph neural network model.

[0111] The calculation formula of the first hidden layer of the investment analysis network model is as follows:

[0112] ;

[0113] In the formula, Indicated in Layer Time Node The eigenvector of Representation and Nodes The set of nodes with which there are edges; Representation and Nodes The node with the edge is Characteristics of the layer; Indicated in Layer Time Node The eigenvector of Function is used to aggregate all ; Functions are used for aggregation and .

[0114] The training samples used to train the investment analysis network model include:

[0115] Randomly select an investor who has invested for 2T months, select T months as the training period, and obtain the investment behavior and communication information of the investor during the training period as sample data of a training sample;

[0116] After the training period, the financial products invested by the investor are the sample labels of the training samples;

[0117] Repeat the above steps until the number reaches K, where K is a custom parameter. Preferably, K is 3000.

[0118] After obtaining the probability of the target investor investing in each financial product, the corresponding financial products can be recommended to the target investor based on the above information to improve the investor's user experience and the transaction volume of the investment exchange platform.

[0119] The above describes an embodiment of the present embodiment, but the present embodiment is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present embodiment, ordinary technicians in this field can also make many forms, all of which are within the protection of the present embodiment.

Claims

1. An investment behavior analysis system based on social network aggregation and distribution, characterized by: include: A communication information collection module (10) is used to obtain the communication information and investment information of each investor on the investment communication platform in the past T months; A communication information processing module (20), which is used to perform semantic analysis on the communication information and determine the investment project and recommendation value corresponding to the communication information; An investment behavior analysis module (30), which is used to analyze the influence value and the influenced value of each investor based on the communication information and investment information of each investor on the investment communication platform in the past T months; The steps for analyzing the impact value of each investor are as follows: Step S301, obtaining the reading volume of the communication information released by each investor; Step S302, presetting a first reading volume, and determining whether the reading volume of the communication information released by each investor in the past T months exceeds the first reading volume; If yes, then mark as first investor; If not, then record as a second investor; Step S303, randomly selecting a first investor, and determining whether a second investor who has read the communication information published by the first investor has made an investment behavior consistent with the communication information; If yes, then it will be included in the affected queue; Step S304, calculating the influence value of the investor, the calculation formula is as follows: ； In the formula, represents the influence value of the i-th investor; represents the total number of readings of the communication information released by the i-th investor in the t-th month; It represents the sum of the number of investors who took consistent actions after reading the investor information; and Respectively express about The first and second weight parameters of ; and All are custom parameters; The specific steps to analyze the impact value of each investor are as follows: Step S311, obtaining the number of times each investor has made investment actions following the communication information released by the first investor in the past T months; Step S312, obtaining the number of times each investor participated in the communication in the past T months; Step S313, calculating the impact value of the investor, the calculation formula is as follows: ； In the formula, represents the impact value of the i-th investor; It indicates the number of times the i-th investor made investment behaviors following the communication information released by the first investor in the past T months; represents the number of times the i-th investor participated in the communication in the past T months; and Respectively express about The first and second weight parameters of ; and All are custom parameters; An investment behavior prediction module (40) is used to predict the probability of a target investor investing in each financial product based on the influence value and the influence value of each investor in the past T months; The steps to predict the probability of a target investor investing in each financial product are as follows: Step S401, constructing investment graph network data; The investment graph network data includes a first node and edges between nodes; The first node is generated based on the mapping between the investor’s influence value and the influenced value; The existence of edges between nodes indicates: an investor makes consistent investment behavior after viewing the communication information released by another investor, an investor views the communication information released by another investor with the same meaning more than n times, or the number of communications between the two investors exceeds m times and they maintain consistent views; Step S402, inputting the investment graph network data into the investment analysis network model, and outputting the probability of the target investor investing in each financial product; T is a custom parameter.

2. The investment behavior analysis system based on social network aggregation and distribution according to claim 1 is characterized in that: The method for obtaining the communication information of each investor on the investment communication platform in the past T months is as follows: Step S101, obtaining a message sent by an investor; Step S102, obtaining messages browsed by investors on the investment exchange platform, presetting a standard browsing time, and determining whether the time for investors to browse messages exceeds the standard browsing time, where the standard browsing time is a custom parameter; If yes, keep it; If not, delete it.

3. The investment behavior analysis system based on social network aggregation and distribution according to claim 1 is characterized in that: The specific steps for semantic analysis of communication information are as follows: Step S201, removing irrelevant characters, punctuation marks and stop words in the communication information; Step S202, constructing an investment character library, extracting stems according to the content in the investment character library, and deleting communication information irrelevant to investment; The investment character library includes the names, numbers and abbreviations of financial products and common investment terms; Step S203: Use the attention mechanism to analyze the communication information to obtain the semantics of the communication information.

4. The investment behavior analysis system based on social network aggregation and distribution according to claim 1 is characterized in that: The investment analysis network model includes a first hidden layer and a classifier; The first hidden layer inputs the investment graph network data and outputs the first updated feature; The classifier inputs the first updated feature and outputs the probability that the target investor invests in each financial product.

5. The investment behavior analysis system based on social network aggregation and distribution according to claim 1 is characterized in that: The first hidden layer of the investment analysis network model is built based on the graph neural network model.

6. The investment behavior analysis system based on social network aggregation and distribution according to claim 1 is characterized in that: The training samples used to train the investment analysis network model include: Randomly select an investor who has invested for 2T months, select T months as the training period, and obtain the investment behavior and communication information of the investor during the training period as sample data of a training sample; After the training period, the financial products invested by the investor are the sample labels of the training samples; Repeat the above steps until the number reaches K; K is a custom parameter.

Images