CN115357729A - Method and device for constructing securities relation map and electronic equipment - Google Patents

Abstract

The embodiment of the invention provides a method and a device for constructing a security relationship map and electronic equipment, wherein the method comprises the following steps: aiming at a plurality of securities of a security relation map to be constructed, acquiring security data of each security; for each security, determining factor data under a plurality of specified categories of the security from the security data of the security, and determining a security vector under each specified category of the security based on the factor data under each specified category of the security; for each appointed category, determining appointed relation data among the securities under the appointed category based on the securities vector under the appointed category of the securities; and constructing a security relation map taking each appointed category as a map dimension respectively based on appointed relation data among the securities under each appointed category by taking each security as a map vertex. According to the scheme, the richness and the connectivity of the security relationship map can be improved, so that the security relationship map can be better constructed.

Description

Method, device and electronic equipment for constructing a securities relationship graph

technical field

The present invention relates to the technical field of knowledge graphs, in particular to a method, device and electronic equipment for constructing a securities relationship graph.

Background technique

At present, by constructing a securities relationship map, it is possible to assist in the analysis of the price influence mode between securities, the risk transmission process, and the setting of constraints on position portfolios in investment.

In related technologies, the securities relationship graph constructed based on the industry to which the securities belong cannot represent the correlation of stocks in different industries, and the number of securities involved in the securities relationship graph constructed by connecting securities mentioned in news or public opinion at the same time is too small, and The securities relationship graph constructed based on the relationship between the upstream and downstream supply chains of the companies where the securities are located can only represent the relationship between the securities of the companies in a supply chain, and so on.

It can be seen that the richness and connectivity of the securities relationship graph constructed by related technologies are not ideal, resulting in the construction of the securities relationship graph is not conducive to providing reference for investors' investment decisions and risk prevention. Therefore, how to improve the richness and connectivity of the securities relationship graph, so as to better construct the securities relationship graph is an urgent problem to be solved.

Contents of the invention

The purpose of the embodiments of the present invention is to provide a method, device, and electronic device for constructing a securities relationship graph, so as to improve the richness and connectivity of the securities relationship graph, so as to better construct the securities relationship graph. The specific technical scheme is as follows:

In the first aspect, an embodiment of the present invention provides a method for constructing a securities relationship graph, the method comprising:

Obtaining the securities data of each of the securities for which the securities relationship graph is to be constructed; wherein the securities data includes the financial statement of the issuer of the securities, the technical index data of the securities and/or all Intraday transaction data for the securities mentioned;

for each of said securities, determining factor data for a plurality of specified categories of the security from the security data for the security, and determining each specified category of the security based on the factor data for each specified category of the security The securities vector under ; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the security description data under the specified category;

For each specified category, based on the securities vectors of each security under the specified category, determine the specified relationship data between the securities under the specified category; wherein, the specified relationship data is used to represent the association relationship data;

Taking each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a security relationship graph with each specified category as a graph dimension.

Optionally, said factor data comprises at least one factor value;

The determining of the security vectors under each specified category of the security based on the factor data under each specified category of the security includes:

For each specified category, each factor value contained in the factor data of the security under the specified category is used as a vector dimension to generate a security vector of the security under the specified category.

Optionally, for each specified category, based on the securities vectors of each security under the specified category, determining the specified relationship data between the securities under the specified category includes:

For each specified category, calculate the vector similarity of the securities vectors under the specified category for every two securities, and obtain the specified relationship data between every two securities under the specified category.

Optionally, using each security as a graph vertex, and based on the specified relationship data between securities under each specified category, constructing a securities relationship graph with each specified category as a graph dimension, including:

For each specified category, based on the vector similarity of the securities vectors under the specified category for each two securities, determine the related securities corresponding to each security under the specified category; wherein, the related securities corresponding to each security are: Securities whose vector similarity of the security is greater than a predetermined threshold, or a specified number of securities corresponding to the security that are sorted according to the vector similarity;

Using each security as a graph vertex, and using the connecting line corresponding to each specified category as a graph edge of the specified category, generate a securities relationship graph with each specified category as a graph dimension;

Wherein, the connection line corresponding to each specified category is the connection line between the graph vertex of each security under the specified category and the graph vertex of the corresponding related securities.

Optionally, the method for determining the specified quantity corresponding to each security includes:

For each specified category, when the number of related securities corresponding to each security under the specified category is 1 to M, the specified category is used as the M information entropy of the securities relationship graph of a graph dimension; where M is the the total number of securities;

Select the minimum value point with the smallest value from the obtained information entropy as the target information entropy;

Determine the quantity of associated securities corresponding to the target information entropy as the specified quantity corresponding to each security under the specified category.

Optionally, the acquisition of securities data of each of the securities for the plurality of securities to be constructed in the securities relationship graph includes:

Whenever a predetermined construction cycle is entered, the securities data of each of the securities in the current construction cycle is obtained for multiple securities of the securities relationship graph to be constructed; wherein, the duration of the construction cycle is based on the securities data Determined by the length of the update cycle.

Optionally, the method also includes:

Using the community mining algorithm and the specified relationship data of each security under each graph dimension in the securities relationship graph, the securities are divided into different clusters from different graph dimensions.

In a second aspect, an embodiment of the present invention provides a device for constructing a securities relationship graph, the device comprising:

The obtaining module is used to obtain the securities data of each of the securities for the plurality of securities to be constructed in the securities relationship graph; wherein, the securities data includes the financial statements of the issuer of the securities and the technical indicators of the securities data and intraday transaction data for said securities;

The vector determination module is configured to, for each of the securities, determine the factor data under multiple specified categories of the security from the security data of the security, and determine the factor data under each specified category of the security based on the factor data of the security. Securities vectors under each specified category of securities; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the securities description data under the specified category;

The relationship determination module is used to determine, for each specified category, specified relationship data between securities of the specified category based on the securities vectors of each security under the specified category; wherein, the specified relationship data is Data used to characterize the relationship;

The construction module is used to use each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a securities relationship graph with each specified category as a graph dimension.

In a third aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete communication with each other through the communication bus;

memory for storing computer programs;

The processor is used to implement the steps of the above method for constructing the securities relationship graph when executing the program stored in the memory.

In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned method for constructing a securities relationship graph are implemented .

Beneficial effects of the embodiments of the present invention:

In the method for constructing a securities relationship graph provided in an embodiment of the present invention, the securities data of each security is obtained for multiple securities to be constructed in the securities relationship graph; The factor data under multiple specified categories of the security, and based on the factor data under each specified category of the security, determine the security vector under each specified category of the security; wherein, each specified category is used for securities description A category, the factor data under each specified category is the securities description data under the specified category; for each specified category, based on the securities vectors under the specified category of each security, determine each security under the specified category Among them, the specified relationship data is the data used to represent the relationship; each security is used as the graph vertex, and based on the specified relationship data between each security under each specified category, each A securities relationship graph that specifies categories as a graph dimension. In this scheme, for multiple securities, the factor data under multiple specified categories of each security is determined according to the acquired securities data, and the securities relationship map is constructed from multiple dimensions, which improves the richness of the securities relationship map. At the same time, the factor data is used to The data is vectorized for each security to obtain the security vector, and based on the obtained security vector, the specified relationship data between the securities in each category is determined, so that the specified relationship data can be used to construct the security relationship map, and improve Connectivity between securities. Therefore, through this scheme, the securities relationship graph can be better constructed.

Of course, implementing any product or method of the present invention does not necessarily need to achieve all the above-mentioned advantages at the same time.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments of the present invention, and those skilled in the art can also obtain other embodiments according to these drawings.

Fig. 1 is a flowchart of a method for constructing a securities relationship graph provided by an embodiment of the present invention;

FIG. 2 is another flowchart of a method for constructing a securities relationship graph provided by an embodiment of the present invention;

FIG. 3 is another flowchart of a method for constructing a securities relationship graph provided by an embodiment of the present invention;

FIG. 4 is a schematic structural diagram of a program module for realizing the securities relationship graph method provided by the embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a device for constructing a securities relationship graph provided by an embodiment of the present invention;

FIG. 6 is a schematic structural diagram of an electronic device provided by an embodiment of the present invention.

Detailed ways

The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art based on the present application belong to the protection scope of the present invention.

Graphs can essentially reveal the relationship between vertices, such as protein interaction graphs, chemical molecular formula graphs, etc., but currently there are few graphs that can reflect the relationship between different securities, such as stocks, funds, bonds, etc. , if it can be constructed to reflect the relationship between different securities and displayed through visualization technology, it can be used to assist investors in investment decision-making and risk prevention.

There are currently three main construction methods for securities relationship graphs. First, based on the securities relationship graph constructed by the industry to which the securities belong, this method connects securities in the same industry, and securities in different industries are not connected by edges; second, based on news or public opinion data Constructed securities relationship graph, this method connects securities mentioned in news or public opinion at the same time; third, based on the securities relationship graph constructed based on the upstream and downstream supply chain relationship of the company where the securities are located, this method connects securities with supply chain relationships . However, there are obvious defects in the above three methods.

First, the securities relationship graph constructed according to the industry to which the securities belong cannot reflect the relationship between stocks in different industries, and the update frequency of the industry in which the securities are located is extremely low. Once the industry is determined, unless the company's main business changes, the industry will basically not change , so the constructed securities relationship graph is relatively fixed and cannot dynamically reflect the relationship between different securities; secondly, the securities relationship graph constructed based on news or public opinion data, the securities in it only appear in the news or public opinion, and the securities involved are all The number of hot securities is small. In addition, crawler technology is required to crawl news data from different sources, which has potential legal compliance issues, and natural language understanding technology is required to extract securities names or securities codes, which has a high technical threshold; third, based on The securities relationship map constructed by the upstream and downstream supply chain relationship of the company where the securities are located. Since it is difficult to obtain the supply chain relationship of each listed company, and each supply chain relationship is a tree diagram, in the optimal case, the constructed securities relationship map It is only a tree diagram, the number of edges and the number of vertices are of the same order of magnitude, and not all companies in the supply chain are listed companies, not all issue securities, and securities under different supply chains cannot form a tree For example, medical industry securities and animal husbandry securities, so the connectivity of the constructed securities relationship graph is low. In addition, the supply chain of the company is relatively stable and the frequency of change is low. Therefore, the securities constructed based on the upstream and downstream supply chain relationship of the company where the securities are The relationship graph cannot dynamically reflect the relationship between securities.

It can be seen that the connectivity of the securities relationship graph constructed by related technologies is low, which is not conducive to the analysis of the community structure of securities, and the dimension of the securities relationship graph constructed by related technologies is 1, and related technologies cannot construct multi-dimensional more meaningful Graph of securities relationships. Therefore, the securities relationship graph constructed by using related technologies is not conducive to providing reference for investors' investment decisions and risk prevention.

In order to solve the above problems, improve the richness and connectivity of the securities relationship graph, so as to better construct the securities relationship graph, the embodiment of the present invention provides a construction method, device and electronic equipment of the securities relationship graph, which can be applied to In an electronic device, the electronic device may be a computer, a server, or the like. The method includes:

Obtain the securities data of each of the securities for which the securities relationship graph is to be constructed; wherein, the securities data includes the financial statements of the issuer of the securities, the technical index data of the securities, and the intraday transaction data of the securities;

For each security, from the security data of the security, determine the factor data under a plurality of specified categories of the security, and determine the factor data under each specified category of the security based on the factor data under each specified category of the security Securities vector; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the security description data under the specified category;

For each specified category, based on the securities vectors of each security under the specified category, determine the specified relationship data between the securities under the specified category; wherein the specified relationship data is data used to represent the relationship;

Taking each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a security relationship graph with each specified category as a graph dimension.

In this scheme, for multiple securities, the factor data under multiple specified categories of each security is determined according to the acquired securities data, and the securities relationship map is constructed from multiple dimensions, which improves the richness of the securities relationship map. At the same time, the factor data is used to The data is vectorized for each security to obtain the security vector, and based on the obtained security vector, the specified relationship data between the securities in each category is determined, so that the specified relationship data can be used to construct the security relationship map, and improve Connectivity between securities. Therefore, through this scheme, the securities relationship graph can be better constructed.

A method for constructing a securities relationship map provided by an embodiment of the present invention is introduced below in conjunction with the accompanying drawings. As shown in FIG. 1, the method may include the following steps:

S101. Obtain the securities data of each of the securities for which the securities relationship graph is to be constructed; wherein, the securities data includes the financial statement of the issuer of the securities, the technical index data of the securities and/or the intraday transaction data of the securities;

Wherein, the securities may be stocks, funds, bonds, and the like. The multiple securities to be constructed in the securities relationship graph can be any multiple securities in the market. The issuer of the securities is generally a listed company. The financial statements of the listed company are publicly released. The technical index data refers to the data collection of all securities prices calculated by mathematical formulas. The intraday transaction data is the data that represents the intraday transaction. Refers to the trading method of holding positions for a short time without leaving overnight positions. The technical index data and intraday transaction data are all public data on the Internet; therefore, the financial statements of the securities issuer, the technical index data of securities, and the intraday transaction data of securities are all It can be obtained from the Internet, that is to say, the securities data of each security can be directly obtained from the Internet for multiple securities whose securities relationship graph is to be constructed, and the obtained securities data can be at least one of them.

S102, for each security, from the security data of the security, determine the factor data under multiple specified categories of the security, and determine each specified category of the security based on the factor data under each specified category of the security The securities vector under ; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the security description data under the specified category;

The above specified categories can be preset categories based on experience and needs, and at the same time set the factor data to be obtained under each specified category. The factor data can be the data specified in the securities data, or the data specified in the securities data. Data obtained after certain calculations. When it is necessary to establish a securities relationship graph, for each specific category, the specified data can be obtained from the securities data as the factor data under the specified category, or the specified data can be obtained from the securities data first, and the specified data will be processed The data obtained after specific calculations are used as factor data under the specified category.

Taking the security data as the technical indicator data of the security as an example, the specified category can be set to the following categories:

Momentum category, reversal category, technology composite category, factor data under momentum category can include: long-term momentum, June momentum, Williams index, etc.; factor data under reversal category can include: 1-month stock price reversal, 3 Monthly stock price reversal, 6-month stock price reversal, etc.; technical composite factor data can include: price-volume divergence, price linear regression coefficient, trading volume variance, length of highest price, distance from highest point, relative strength of stock price, etc.

For each security, the factor data under each specified category of the security may be vectorized to obtain a security vector under each specified category of the security.

In one implementation, the aforementioned factor data includes at least one factor value;

For example, when the securities data is the financial statement of the issuer of the securities, the specified category can be set to the following categories:

a. Scale category, the factor values included in the factor data under this category can be: total assets, working capital, financial expenses, free cash flow of equity, depreciation and amortization, total operating income, net working capital and operating costs, etc. at least one;

b. Profit category, the factor values included in the factor data under this category can be: return on equity before tax, return on net assets, return on total assets, net profit rate, gross profit rate, total profit rate, operating profit rate and operating profit at least one of proportion etc.;

c. Capital structure category, the factor data included in this category can include: asset-liability ratio, fixed asset ratio, equity-liability ratio, current asset ratio, total debt ratio, shareholder equity ratio, operating capital ratio, and retained earnings at least one of proportion etc.;

d. Valuation category, the factor values included in the factor data under this category can be: price-earnings ratio, price-to-book ratio, price-to-current ratio, price-to-sales ratio, book-to-market value ratio, net assets per share, non-price-earnings ratio and sales ratio, etc. at least one of

e. Quality category. The factor values included in the factor data under this category can be: total asset cash recovery rate, receivable-payable ratio, net profit fluctuation, interest coverage ratio, sales period expense ratio, retained earnings ratio, and net profit volatility and at least one of the cash-to-sales ratio;

f. Growth category, the factor values included in the factor data under this category can be: quarterly year-on-year growth rate of return on net assets, quarterly year-on-year growth rate of net profit, operating profit growth rate, main operating income growth rate, total asset growth rate, At least one of the growth rate of total profit, growth rate of shareholders' equity and growth rate of net assets per share;

g. Operation category, the factor values included in the factor data under this category can be: accounts receivable turnover ratio, fixed asset turnover ratio, current asset turnover ratio, inventory turnover ratio, accounts payable turnover ratio, shareholder equity turnover ratio and At least one of the total asset turnover ratio, etc.;

h. Shareholder-related category, the factor data included in this category can include factor values: proportion of executives’ shareholding, proportion of employee stock ownership plan, change in institutional shareholding, change rate of number of shareholders, proportion of tradable shares, institutional shareholding At least one of the proportion of the total share capital, the proportion of the largest shareholder and the number of shares held by the management;

i. Debt solvency category, the factor data included in this category can include: current ratio, quick ratio, cash ratio, cash flow ratio, long-term debt ratio, cash flow and debt ratio, super quick ratio and long-term At least one of debt to operating income ratio etc.

When the security data is the technical indicator data of the security, the specified category can be set to the following categories:

a. Momentum category, the factor values contained in the factor data under this category can be: long-term momentum, June momentum, William index, seasonality, stochastic index, homeopathic index (CCI, Commodity Channel Index), momentum line (MOM, Momentum ) and at least one of moving average (MACD, Moving Average Convergence/Divergence);

b. Inversion category, the factor data contained in this category can contain factor values: 1-month stock price reversal, 3-month stock price reversal, 6-month stock price reversal, 30-day rise and fall, 90-day rise and fall At least one of the price increase and the 180-day increase and decrease;

c. Technical composite category, the factor data contained in this category can include factor values: price-volume divergence, capacity ratio (a technical indicator to measure buying and selling power), price linear regression coefficient, volume variance, maximum price length, maximum At least one of point distance and stock price relative strength.

When the securities data is the intraday trading data of securities, the specified category can be set as high-frequency category, and the factor values contained in the factor data under this category can be: high-frequency fluctuation, high-frequency idiosyncratic fluctuation, high-frequency specificity, High Frequency Systematic Volatility, High Frequency Return Variance, High Frequency Return Skewness, and High Frequency Return Peak.

The factor values mentioned above can be obtained from the securities data, or obtained by performing specific operations on the data obtained from the securities data. The specific operations can be operations in the prior art, which do not belong to the invention point of the present invention. I won't go into details here. The above specified categories and factor data are only exemplary introductions, and the embodiment of the present invention does not limit the specified categories and factor data.

The aforementioned determination of the security vectors under each specified category of the security based on the factor data under each specified category of the security may include:

For each specified category, each factor value contained in the factor data of the security under the specified category is used as a vector dimension to generate a security vector of the security under the specified category.

Taking the specified category as the scale category as an example, assuming that according to the financial statements of the issuer of a security, the factor values contained in the factor data under the scale category are: total assets = A, operating capital = B, financial expenses = C... ..., then the security vector of the security under the scale category can be generated as [A, B, C...]. Alternatively, it is also possible to normalize the factor values contained in the factor data under the specified category, and then generate the stock vector of the stock according to the normalized value of each factor.

In this implementation, for each specified category, each factor value contained in the factor data of the security under the specified category is used as a vector dimension to generate the security vector of the security under the specified category. It is easier to reflect the correlation between various securities.

S103, for each specified category, based on the securities vectors of each security under the specified category, determine the specified relationship data between the securities under the specified category; wherein the specified relationship data is used to represent the relationship data;

The above-mentioned specified relationship data may be the vector similarity of the securities vectors of each two securities. Specified relationship data between securities, which may include:

For each specified category, calculate the vector similarity of the securities vectors under the specified category for every two securities, and obtain the specified relationship data between every two securities under the specified category.

Wherein, the manner of calculating the vector similarity may be calculating the normalized cosine similarity or the Pearson correlation coefficient.

Exemplarily, when the specified category is the scale category, under the specified category, the security vector of security 1 is [A1, B1, C1...], and the security vector of security 2 is [A2, B2, C2...], then The vector similarity of the security vectors of the security 1 and the security 2 can be calculated as specified relationship data between the security 1 and the security 2.

In this implementation, for each specified category, by calculating the vector similarity of the securities vectors of each two securities under the specified category, the specified relationship data between each two securities under the specified category can be obtained. Further, it can be Build a securities relationship graph based on specified relationship data.

S104, taking each security as a map vertex, and constructing a securities relationship graph with each specified category as a graph dimension, based on specified relationship data between securities under each specified category.

It can be understood that the graph is composed of graph vertices and graph edges composed of two connected graph vertices. In this embodiment, each security can be used as the graph vertex, and for each two securities, according to the The specified relationship data between, determine whether the graph vertices of the two securities are connected.

In one implementation, the above-mentioned securities are used as graph vertices, and based on the specified relationship data between securities under each specified category, a securities relationship graph with each specified category as a graph dimension is constructed, which can be Include steps A1-A2:

Step A1, for each specified category, based on the vector similarity of the securities vectors under the specified category of every two securities, determine the related securities corresponding to each security under the specified category; wherein, the related securities corresponding to each security It is: the securities whose vector similarity with the security is greater than a predetermined threshold, or a specified number of securities corresponding to the security that are sorted according to the vector similarity;

In this implementation, a predetermined threshold can be set first, and the predetermined threshold can be set according to experience and needs, and then for each security under each specified category, it is determined that the vector similarity with the security vector of the security is greater than the predetermined threshold Securities, as related securities;

Alternatively, the designated quantity can also be preset, and the designated quantity can also be set according to experience and needs, and then, for each security under each specified category, determine the vector similarity with the security vector of the security with the largest or minimum vector similarity. A specified number of securities, as affiliated securities. In this implementation, the specified number of securities with the highest vector similarity with the security vector of the security is determined, as related securities, which can relatively better reflect the relationship between the various securities.

In the specific implementation process, for each specified category, it is also possible to first generate the specified category as an original securities relationship graph of a graph dimension, and each vertex in the original securities relationship graph is fully connected, that is, each security is connected in pairs , at the same time, the vector similarity between the securities vectors of each two securities is used as the weight of the graph edges of the two securities, and then for each graph vertex, the specified number of graph edges with the largest or smallest weight is reserved to obtain the A securities relationship graph that specifies categories as a graph dimension.

Step A2, using each security as a graph vertex, and using the connecting line corresponding to each specified category as a graph edge of the specified category, to generate a securities relationship graph with each specified category as a graph dimension; wherein, each specified The connection line corresponding to the category is the connection line between the graph vertex of each security under the specified category and the graph vertex of the corresponding related securities.

In this implementation, for each security under each specified category, the graph vertex of the security is connected with the graph vertex of the corresponding related securities in the securities relationship graph, so as to obtain the graph vertex of the security and the corresponding For the specified category, each connecting line of the specified category is used as a graph edge, so as to obtain a securities relationship graph with the specified category as a graph dimension. Finally, a multi-dimensional securities relationship graph can be obtained in which each specified category is used as a graph dimension.

A multi-dimensional securities relationship graph can be a collection of securities relationship graphs in each graph dimension, that is, a securities relationship graph in each graph dimension can be used as an independent graph, or it can also be a graph containing multiple types of graph edges. Graph, that is to say, in a multi-dimensional securities relationship graph, there may be multiple graph edges between every two securities, and each graph edge is used to represent a different specified category.

After the securities relationship graph is generated, the securities relationship graph can be stored in the graph database to realize the storage of the securities graph. The terminal device of the client can also obtain the securities graph from the graph database and store the The securities relationship graph is visualized.

In this embodiment, for multiple securities, according to the acquired securities data, the factor data under multiple specified categories of each security is determined, and the securities relationship map is constructed from multiple dimensions, which improves the richness of the securities relationship map. The factor data vectorizes each security to obtain a security vector, and based on the obtained security vector, determines the specified relationship data between each security under each category, so as to use the determined specified relationship data to construct a securities relationship graph. Improved connectivity between securities. Therefore, through this scheme, the securities relationship graph can be better constructed.

Optionally, in another embodiment of the present invention, as shown in FIG. 2 , the method of determining the designated quantity corresponding to each of the above-mentioned securities may include steps S201-step S203:

S201, for each specified category, calculate the M information entropies of the specified category as a securities relationship map of a map dimension when the number of related securities corresponding to each security under the specified category is 1 to M; wherein, M is the total number of each security;

That is, assuming that the number of related securities corresponding to each security under the specified category is 1, 2, 3...M, calculate the information entropy of the specified category as a graph dimension of the securities relationship graph in each case, where information Entropy can be used to measure the uncertainty of the securities relationship graph of the specified category as a graph dimension, that is, the reliability of the securities relationship graph of the specified category as a graph dimension, for example, the associated securities corresponding to each security under the specified category The number of securities is n, and when the number of related securities corresponding to each security under the specified category is n, the information entropy of the specified category as a graph dimension of the securities relationship map can be calculated.

Assuming that the securities are respectively securities 1, securities 2, securities 3 ... securities M, the information entropy can be calculated according to the following formula:

为证券m的关联证券数量为n时，该n只关联证券的证券向量的向量相似度之和，该n只关联证券可以为与证券m的证券向量的向量相似度最大的n只证券，m的取值范围为1到M。Among them, H(n) is the information entropy of the securities relationship map with the specified category as a graph dimension when the number of related securities corresponding to each security under the specified category is n, and the value range of n is 1 to M; M is the total number of securities;

When the number of related securities of security m is n, the sum of the vector similarities of the securities vectors of the n related securities, the n related securities can be the n securities with the largest vector similarity with the securities vector of security m, m The value range is from 1 to M.

S202. Select the minimum value point with the smallest value from the obtained information entropy as the target information entropy;

It can be understood that selecting the smallest minimum value point can ensure that the uncertainty of the securities relationship graph of the specified category as a graph dimension is as small as possible under the condition of excluding meaningless minimum value points.

S203. Determine the quantity of associated securities corresponding to the target information entropy as the specified quantity corresponding to each security under the specified category.

It can be understood that, for each specified category, the specified quantity corresponding to each security under the specified category can be determined by calculating information entropy, so that the uncertainty of the finally generated multi-dimensional security relationship map is lower.

In this embodiment, for each specified category, when the number of related securities corresponding to each security under the specified category is 1 to M respectively, the specified category is used as M information entropies of a securities relationship graph of a map dimension; Select the minimum value point with the smallest value from the obtained information entropy as the target information entropy; determine the number of related securities corresponding to the target information entropy as the specified quantity corresponding to each security under the specified category. This scheme determines the specified quantity by calculating the information entropy, which can make the constructed securities relationship map more reliable.

Optionally, in another embodiment of the present invention, the acquisition of the securities data of each of the securities for the multiple securities to be constructed of the securities relationship graph may include:

Whenever a predetermined construction cycle is entered, the securities data of each security in the current construction cycle is obtained for multiple securities of the securities relationship graph to be constructed; wherein, the duration of the construction cycle is determined based on the duration of the update cycle of the securities data .

In this embodiment, the securities relationship graph can be periodically updated, that is, periodically for multiple securities to be constructed in the securities relationship graph, obtain the securities data of each security in the current construction cycle, and execute the subsequent construction of the securities relationship Mapping process. The above-mentioned construction cycle can be consistent with the update cycle of securities data, such as one quarter, one month, one day, one hour, etc., which are all possible. Different designated categories can also have different construction periods. For example, the factor data under the designated category comes from the financial statements of the issuer of the securities, and the construction period can be one quarter; the factor data under the designated category comes from the technical indicators of the securities data, the construction period can be one day; the factor data under the specified category comes from the intraday transaction data of securities, and the construction period can be one hour.

In this embodiment, for multiple securities, according to the acquired securities data, the factor data under multiple specified categories of each security is determined, and the securities relationship map is constructed from multiple dimensions, which improves the richness of the securities relationship map. The factor data vectorizes each security to obtain a security vector, and builds a security relationship graph based on the obtained security vector, which improves the connectivity between various securities. Therefore, through this scheme, the securities relationship graph can be better constructed. Further, by obtaining the securities data of each security in the current construction cycle for multiple securities to be constructed whenever a predetermined construction period is entered, the securities relationship diagram can be dynamically updated, thereby dynamically reflecting Relationship between different securities.

Optionally, in another embodiment of the present invention, the method further includes:

Using the community mining algorithm and the specified relationship data of each security under each graph dimension in the securities relationship graph, each security is divided into different clusters from different graph dimensions.

Among them, community mining refers to dividing entities into different types, that is, different clusters, so that there are as many graph edges in the same cluster as possible, and as few graph edges as possible between different clusters. The community mining algorithm may adopt UEOC (community mining algorithm based on integrated network overlap) algorithm, FNCA (a fast community mining algorithm) algorithm, etc., which are not specifically limited in the embodiment of the present invention.

In the actual application process, the community mining algorithm can be used to process the securities relationship graph under each graph dimension. Since the community mining algorithm generally uses the weight of each graph edge, when the specified relationship data is every two graphs in each graph dimension When the vector similarity of the securities vectors of only two securities is used, the vector similarity of the securities vectors of each two securities can be used as the weight of the two securities as the graph edge of the graph vertex, and further, each graph under each graph dimension The weight of the graph edge is brought into the calculation process of the above-mentioned community mining algorithm, and then the calculation results of dividing each security under the graph dimension into different clusters can be obtained.

In this embodiment, by using the community mining algorithm and the specified relational data of each stock under each map dimension in the stock relation map, each stock is divided into different clusters from different map dimensions, which can be more Intuitively reflect the relationship between different securities, so as to better provide reference for investors' investment decisions and risk prevention.

For the convenience of understanding, the method for constructing the securities relationship graph provided by the embodiment of the present invention will be further introduced below in conjunction with the accompanying drawings.

As shown in Figure 3, there are N types of specified categories, specified category 1, specified category 2 ... specified category N, for specified category 1, the multi-factor vector data of specified category 1 can be constructed first, that is, each security under specified category 1 The stock vectors of each stock; calculate the vector similarity between the stock vectors of each two stocks according to the stock vectors of each stock; use the vector similarity between the stock vectors of each two stocks as the weight of the graph edge of the two stocks, In this way, a fully connected original securities relationship graph is constructed; the number n of graph edges is determined according to the minimum value of information entropy, and the n graph edges with the largest weight are reserved for each security to obtain a securities relationship graph of the specified category 1, that is, the specified category 1 A securities relationship graph as a graph dimension.

The construction method of the securities relationship graph of other specified categories is the same. After the securities relationship graphs of each specified category are constructed, the securities relationship graphs of all specified categories are superimposed to generate a multi-dimensional securities relationship graph. After that, graph storage and visual display can be carried out, that is, the securities relationship graph is stored in the graph database, and the securities relationship graph is displayed on the client; at the same time, community mining and analysis among securities can be carried out, that is, using community mining For the specified relationship data of each security under each map dimension, the algorithm divides each stock into different clusters from different map dimensions.

In an implementation manner, the computer program implementing the method provided by the embodiment of the present invention can be divided into multiple functional modules as shown in FIG. 4:

The securities multi-factor data calculation module is used to obtain the securities data of each security for multiple securities to be constructed in the securities relationship graph, and obtain and calculate the factor data under multiple specified categories of each security from the securities data, Generate security vectors for each security under each specified category;

The securities similarity calculation module is used to calculate the vector similarity of the securities vectors of each two securities under each specified category;

The securities full connection graph building module is used to construct and store a fully connected securities relationship graph between all securities under each specified category, wherein the weight of the graph edge is the vector similarity of the securities vectors of the two securities on the graph edge ;

A module for establishing the number of securities edges is used to calculate the specified number of graph edges of each security under each specified category, and reserve a specified number of graph edges for each security under each specified category;

The securities multi-dimensional relationship graph overlay and generation module is used to combine the securities relationship graph of each specified category into a multi-dimensional securities relationship graph;

The securities relationship map storage and visualization module is used to store the constructed securities relationship map in the graph database, and use the advantages of the graph database display to visualize the multi-dimensional relationship between different securities;

The securities relationship map community mining and analysis module is used to use the community mining algorithm and the specified relationship data of each stock under each map dimension in the stock relation map to divide each stock into different map dimensions from different map dimensions Among the clusters, the securities relationship within each cluster and between different clusters is analyzed.

In this embodiment, the factor data under multiple specified categories of each security is determined according to the acquired securities data, and the securities relationship map is constructed from multiple dimensions, which improves the richness of the securities relationship map. Stocks are vectorized to obtain stock vectors, and a stock relationship graph is constructed based on the obtained stock vectors, which improves the connectivity between each stock. Therefore, through this scheme, the securities relationship graph can be better constructed.

The embodiment of the present invention also provides a device for constructing a securities relationship graph, as shown in Figure 5, the device includes:

An acquisition module 510, configured to acquire securities data of each of the securities for which the securities relationship graph is to be constructed; wherein, the securities data includes the financial statement of the issuer of the securities, the technical information of the securities Indicator data and/or intraday trading data for said securities;

The vector determination module 520 is configured to, for each of the securities, determine the factor data under multiple specified categories of the security from the security data of the security, and based on the factor data under each specified category of the security, determine The securities vector under each specified category of the security; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the securities description data under the specified category;

The relationship determination module 530 is configured to, for each specified category, determine the specified relationship data between securities under the specified category based on the securities vectors of each security under the specified category; wherein, the specified relationship data It is the data used to characterize the relationship;

The construction module 540 is configured to use each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a security relationship graph with each specified category as a graph dimension.

Optionally, said factor data comprises at least one factor value;

The vector determination module is specifically used for:

For each specified category, each factor value contained in the factor data of the security under the specified category is used as a vector dimension to generate a security vector of the security under the specified category.

Optionally, the relationship determination module is specifically configured to:

For each specified category, calculate the vector similarity of the securities vectors under the specified category for every two securities, and obtain the specified relationship data between every two securities under the specified category.

Optionally, the building blocks include:

The related securities determination sub-module is used to determine the related securities corresponding to each security under the specified category based on the vector similarity of the securities vectors under the specified category of every two securities for each specified category; wherein, each The associated securities corresponding to the securities are: securities whose vector similarity with the securities is greater than a predetermined threshold, or a specified number of securities corresponding to the securities that are sorted according to the similarity of the vectors;

Generate a sub-module, which is used to use each security as a graph vertex, and use the connecting line corresponding to each specified category as a graph edge of the specified category to generate a securities relationship graph with each specified category as a graph dimension; wherein, The connection line corresponding to each specified category is the connection line between the graph vertex of each security under the specified category and the graph vertex of the corresponding related securities.

Optionally, the device also includes:

A calculation module, for each specified category, when the number of related securities corresponding to each security under the specified category is 1 to M respectively, the specified category is used as M information entropies of a securities relationship map of a map dimension; Among them, M is the total number of securities;

The selection module is used to select the minimum value point with the smallest value from the obtained information entropy as the target information entropy;

A quantity determining module, configured to determine the quantity of associated securities corresponding to the target information entropy as the specified quantity corresponding to each security under the specified category.

Optionally, the acquisition module is specifically used for:

Whenever a predetermined construction cycle is entered, the securities data of each of the securities in the current construction cycle is obtained for multiple securities of the securities relationship graph to be constructed; wherein, the duration of the construction cycle is based on the securities data Determined by the length of the update cycle.

Optionally, the device also includes:

The division module is used to divide each security into different clusters from different graph dimensions by using the community mining algorithm and the specified relationship data of each security under each graph dimension in the securities relationship graph.

The embodiment of the present invention also provides an electronic device, as shown in FIG. complete the mutual communication,

Memory 603, used to store computer programs;

The processor 601 is configured to implement the above method for constructing the securities relationship graph when executing the program stored in the memory 603 .

The communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus or the like. The communication bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface is used for communication between the electronic device and other devices.

The memory may include a random access memory (Random Access Memory, RAM), and may also include a non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the aforementioned processor.

Above-mentioned processor can be general-purpose processor, comprises central processing unit (Central Processing Unit, CPU), network processor (Network Processor, NP) etc.; It can also be Digital Signal Processor (Digital Signal Processor, DSP), ASIC (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

In yet another embodiment provided by the present invention, a computer-readable storage medium is also provided, and a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the construction of the above-mentioned securities relationship graph is realized. method steps.

In yet another embodiment provided by the present invention, there is also provided a computer program product containing instructions, which, when run on a computer, causes the computer to execute the method for constructing a securities relationship graph in the above embodiment.

In the above embodiments, all or part of them may be implemented by software, hardware, firmware or any combination thereof. When implemented using software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, all or part of the processes or functions according to the embodiments of the present invention will be generated. The computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from a website, computer, server or data center Transmission to another website site, computer, server, or data center by wired (eg, coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center integrated with one or more available media. The available medium may be a magnetic medium (for example, a floppy disk, a hard disk, or a magnetic tape), an optical medium (for example, DVD), or a semiconductor medium (for example, a Solid State Disk (SSD)).

It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is a relationship between these entities or operations. any such actual relationship or order exists between them. Furthermore, the term "comprises", "comprises" or any other variation thereof is intended to cover a non-exclusive inclusion such that a process, method, article, or apparatus comprising a set of elements includes not only those elements, but also includes elements not expressly listed. other elements of or also include elements inherent in such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising said element.

Each embodiment in this specification is described in a related manner, the same and similar parts of each embodiment can be referred to each other, and each embodiment focuses on the differences from other embodiments.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present invention are included in the protection scope of the present invention.

Claims (10)

1. A method for building a securities relational graph, characterized in that the method comprises: Obtaining the securities data of each of the securities for which the securities relationship graph is to be constructed; wherein the securities data includes the financial statement of the issuer of the securities, the technical index data of the securities and/or all Intraday transaction data for the securities mentioned; for each of said securities, determining factor data for a plurality of specified categories of the security from the security data for the security, and determining each specified category of the security based on the factor data for each specified category of the security The securities vector under ; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the security description data under the specified category; For each specified category, based on the securities vectors of each security under the specified category, determine the specified relationship data between the securities under the specified category; wherein, the specified relationship data is used to represent the association relationship data; Taking each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a security relationship graph with each specified category as a graph dimension. 2. The method according to claim 1, wherein said factor data comprises at least one factor value; The determining of the security vectors under each specified category of the security based on the factor data under each specified category of the security includes: For each specified category, each factor value contained in the factor data of the security under the specified category is used as a vector dimension to generate a security vector of the security under the specified category. 3. The method according to claim 1 or 2, characterized in that, for each specified category, based on the securities vectors of each security under the specified category, determine the Specified relational data for , including: For each specified category, calculate the vector similarity of the securities vectors under the specified category for every two securities, and obtain the specified relationship data between every two securities under the specified category. 4. The method according to claim 3, characterized in that, using each of the securities as graph vertices, based on the specified relationship data between securities under each specified category, constructing A graph-dimensional securities relationship graph, including: For each specified category, based on the vector similarity of the securities vectors under the specified category for each two securities, determine the related securities corresponding to each security under the specified category; wherein, the related securities corresponding to each security are: Securities whose vector similarity of the security is greater than a predetermined threshold, or a specified number of securities corresponding to the security that are sorted according to the vector similarity; Using each security as a graph vertex, and using the connecting line corresponding to each specified category as a graph edge of the specified category, generate a securities relationship graph with each specified category as a graph dimension; Wherein, the connection line corresponding to each specified category is the connection line between the graph vertex of each security under the specified category and the graph vertex of the corresponding related securities. 5. The method according to claim 4, wherein the method for determining the designated quantity corresponding to each security includes: For each specified category, when the number of related securities corresponding to each security under the specified category is 1 to M, the specified category is used as the M information entropy of the securities relationship graph of a graph dimension; where M is the the total number of securities; Select the minimum value point with the smallest value from the obtained information entropy as the target information entropy; Determine the quantity of associated securities corresponding to the target information entropy as the specified quantity corresponding to each security under the specified category. 6. The method according to claim 1, characterized in that, for the plurality of securities to be constructed in the securities relationship graph, acquiring the securities data of each said securities comprises: Whenever a predetermined construction cycle is entered, the securities data of each of the securities in the current construction cycle is obtained for multiple securities of the securities relationship graph to be constructed; wherein, the duration of the construction cycle is based on the securities data Determined by the length of the update cycle. 7. The method according to claim 1, further comprising: Using the community mining algorithm and the specified relationship data of each security under each graph dimension in the securities relationship graph, the securities are divided into different clusters from different graph dimensions. 8. A device for building a securities relationship graph, characterized in that the device comprises: The obtaining module is used to obtain the securities data of each of the securities for the plurality of securities to be constructed in the securities relationship graph; wherein, the securities data includes the financial statements of the issuer of the securities and the technical indicators of the securities data and intraday transaction data for said securities; The vector determination module is configured to, for each of the securities, determine the factor data under multiple specified categories of the security from the security data of the security, and determine the factor data under each specified category of the security based on the factor data of the security. Securities vectors under each specified category of securities; wherein, each specified category is a category used for securities description, and the factor data under each specified category is the securities description data under the specified category; The relationship determination module is used to determine, for each specified category, specified relationship data between securities of the specified category based on the securities vectors of each security under the specified category; wherein, the specified relationship data is Data used to characterize the relationship; The construction module is used to use each security as a graph vertex, and based on the specified relationship data between securities under each specified category, construct a securities relationship graph with each specified category as a graph dimension. 9. An electronic device, characterized in that it comprises a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory complete mutual communication through the communication bus; memory for storing computer programs; When the processor is used to execute the program stored in the memory, it realizes the method steps described in any one of claims 1-7. 10. A computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, the method steps of any one of claims 1-7 are implemented.

Images