Invocation information recommendation method, electronic equipment and readable storage medium

Abstract

The embodiment of the invention discloses an investment attraction information recommendation method, electronic equipment and a readable storage medium. The method comprises the steps that investment attraction party data and investor party data are collected; constructing an investment attracting party and investor knowledge graph, an investment attracting party attribute graph and an investor attribute graph; an investment attraction information recommendation model is constructed, the investment attraction information recommendation model comprises an input layer, a graph convolution layer and a prediction layer, and the input layer fuses internal interaction information and cross interaction information of an investment attraction party attribute graph and an investor attribute graph to obtain embedded representations of an investment attraction party and an investor; embedding representations of the investment attracting party and the investor are input into a graph convolution layer, the graph convolution layer obtains final representations of the investment attracting party and the investor through multi-layer aggregation operation, the final representations are input into a prediction layer, and the prediction layer obtains correlation scores of the investment attracting party and the investor through a three-layer multi-layer perceptron; and selecting a preset number of investors with the highest score to recommend to the investment attracting party.

Description

Technical Field

[0001] The present invention relates to the technical field of content recommendation, and in particular to a method and system for recommending investment information, an electronic device and a readable storage medium. Background Art

[0002] Investment promotion is a series of activities carried out by the government, economic development zones, commercial organizations, etc. to attract foreign capital, technology, talents and management experience and promote local economic development. Investment promotion is usually a need for regional economic development, especially in the context of economic globalization. In order to gain an advantage in competition, various places will attract foreign and domestic capital, which can promote local economic development, increase employment opportunities, and increase tax revenue, thereby enhancing regional economic vitality. In addition, investors may bring new technologies, new concepts and management experience, which will help upgrade and transform industries in the investment promotion region and improve industrial competitiveness.

[0003] However, due to the information asymmetry between the business invitation party and the investor, it is difficult for both parties to accurately assess each other's needs and capabilities. It may be difficult to find an investor that matches the industry characteristics and development strategy of the business invitation party's region. In addition, when selecting investors, the business invitation party needs to accurately assess the matching degree between the investor and the local industry and the investment risks, including financial risks, operational risks and market risks. Therefore, for the business invitation party, finding a suitable investor is still a big challenge.

[0004] Therefore, designing an accurate and efficient investment information recommendation method is an urgent problem to be solved by those skilled in the art. Summary of the invention

[0005] In view of this, the embodiments of the present invention provide a method and system for recommending investment information, an electronic device and a readable storage medium, which recommend suitable investors to the investment party, improve the recommendation efficiency, and accurately match the investment party's needs.

[0006] In order to achieve the purpose of the invention, the present invention adopts the following technical solutions:

[0007] In a first aspect, a method for recommending investment information is provided, characterized by comprising:

[0008] Collecting the data of the business invitation party and the investor party, and pre-processing the data of the business invitation party and the investor party;

[0009] Build the knowledge graph of the merchant and investor, as well as the merchant attribute graph and investor attribute graph based on the preprocessed data;

[0010] Constructing an investment promotion information recommendation model, the investment promotion information recommendation model includes an input layer, a graph convolution layer, and a prediction layer, the input layer fuses the intrinsic interaction information and the cross-interaction information of the investment promotion party attribute graph and the investor attribute graph to obtain embedded representations of the investment promotion party and the investor, the embedded representations of the investment promotion party and the investor are input to the graph convolution layer, the graph convolution layer obtains the final representations of the investment promotion party and the investor through multi-layer aggregation operations, the final representation is input to the prediction layer, and the prediction layer obtains the correlation score of the investment promotion party and the investor through a three-layer multi-layer perceptron;

[0011] A predetermined number of investors with the highest scores are selected and recommended to the investment invitation party.

[0012] Optionally, the collecting of the merchant data and the investor data includes:

[0013] The investment promotion data include local industrial development planning, industrial structure, industrial policy, financial subsidies, land use policy, industrial support policy, park infrastructure, information on settled enterprises, and industrial positioning of the park;

[0014] The investor data includes investment intention, site selection preference, enterprise financial information, profitability, market size, enterprise litigation information, enterprise business scope, enterprise qualifications, and historical investment information;

[0015] Optionally, preprocessing the merchant data and investor data includes:

[0016] Data cleaning, data format conversion, data normalization, and manual labeling; the data cleaning includes removing duplicate data, processing missing values, correcting erroneous data, and deleting outliers; the data format conversion includes converting the cleaned data into a unified format; the data normalization includes scaling the format-converted data to the [0,1] interval; the manual labeling includes manually labeling the normalized data with categories.

[0017] Optionally, a knowledge graph of the merchant and investor, as well as a merchant attribute graph and an investor attribute graph are constructed based on the preprocessed data, including:

[0018] In the merchant's attribute graph and the investor's attribute graph, the merchant's attributes and the investor's attributes are respectively used as nodes of the graph, and the two nodes are connected by an edge;

[0019] use Represents the embedding vector of node i in the merchant attribute graph, using Represents the embedding vector of node i in the investor attribute graph.

[0020] Optionally, the input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including:

[0021] Get the internal interaction information between nodes, expressed as:

[0022] z ij =MLP(u i ,u j )

[0023] Among them, u i and u j are respectively the embedding vectors of node i and node j of the merchant attribute graph or the investor attribute graph, MLP() is a multi-layer perceptron function, z ij is the internal interaction information between node i and node j;

[0024] Get the aggregated internal interaction information of the node, expressed as:

[0025]

[0026] Among them, z i is the aggregated internal interaction information of node i in the merchant attribute graph or the investor attribute graph, ij is the internal interaction information between node i and node j, N i is the neighbor set of node i.

[0027] Optionally, the input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including:

[0028] Get the cross-interaction information between nodes, expressed as:

[0029]

[0030] Among them, u i and are respectively the embedding vector of node i of the merchant attribute graph and the embedding vector of node j of the investor attribute graph, or the embedding vector of node i of the investor attribute graph and the embedding vector of node j of the merchant attribute graph, s ij is the cross-interaction information between node i and node j, and e is the element-by-element multiplication;

[0031] Get the aggregated cross-interaction information of the nodes, expressed as:

[0032]

[0033] Among them, si and are respectively the aggregated cross-interaction information of node i of the merchant attribute graph and the node set of the investor attribute graph, or the aggregated cross-interaction information of node i of the investor attribute graph and the node set of the merchant attribute graph, s ij is the cross interaction information between node i and node j.

[0034] Optionally, the input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including:

[0035] Get the fusion information of the node's internal interaction information and cross-interaction information, expressed as:

[0036] u′ i =GRU(u i ,z i ,s i )

[0037] Among them, u i ′ is the fusion information of the internal interaction information and cross-interaction information of node i in the merchant attribute graph or the investor attribute graph, u i is the embedding vector of node i, s i is the aggregated cross-interaction information of node i, z i is the aggregated internal interaction information of node i, GRU() is the gated recurrent unit function;

[0038] Get the embedded representation of the merchant and investor, expressed as:

[0039]

[0040] Among them, u i ′ is the fusion information of the internal interaction information and the cross-interaction information of the node i of the investment attraction attribute map or the investor attribute map, and V is the node set of the investment attraction attribute map or the investor attribute map.

[0041] Optionally, the graph convolution layer obtains the final representation of the merchant and the investor through a multi-layer aggregation operation, including:

[0042] The graph convolution layer has a multi-layer structure. In the knowledge graph of the investment party and the investor, each layer of the graph convolution layer performs weighted summation of the neighboring nodes of the target node to obtain an embedded representation of each layer, which is expressed as:

[0043]

[0044] in, and are the embedding representations of investors at the kth and k+1th layers respectively. and are the embedding representations of the merchant agents at the kth and k+1th layers, respectively. u is the neighbor node set of the target investment attraction, N i It is the set of neighbor nodes of the target investor.

[0045] Optionally, the graph convolution layer obtains the final representation of the merchant and the investor through a multi-layer aggregation operation, including:

[0046] The embedding representations of different layers are merged to get the final representation, which is expressed as:

[0047]

[0048] in, and are the embedding representations of the merchant and investor at the kth layer, α k is the weight of the kth layer, K is the number of layers of the graph convolutional layer, e u and e i They are the final statements of the business invitation party and the investor respectively.

[0049] Optionally, the prediction layer obtains the correlation scores between the merchant and the investor through a three-layer multilayer perceptron, including:

[0050] The correlation scores between the merchant and the investor are expressed as:

[0051] r=φ3(φ2(φ1(P T e u ,Q T e i )))

[0052] Among them, φ3, φ2 and φ1 are the mapping functions of the third, second and first layers of multilayer perceptrons respectively, P T and Q T are the potential factor matrices of the merchant and investor, respectively, u and e i are the final representations of the merchant and investor respectively, and r is the correlation score between the merchant and investor.

[0053] Optionally, selecting a predetermined number of investors with the highest scores and recommending them to the investment invitation party comprises:

[0054] Sort the investors by scores from high to low, and select a preset number of investors ranked at the top to recommend to the investment promotion party.

[0055] Optionally, a loss function is used to optimize the investment information recommendation model, and the loss function is a BPR loss function.

[0056] In a second aspect, a system for recommending investment information is provided, characterized in that it includes:

[0057] A collection module, used to collect the data of the business invitation party and the investor party, and pre-process the data of the business invitation party and the investor party;

[0058] A recommendation module, which constructs a knowledge graph of the investment attraction party and the investor, as well as an investment attraction party attribute graph and an investor attribute graph according to the preprocessed data; the recommendation module includes an investment attraction information recommendation model, which includes an input layer, a graph convolution layer, and a prediction layer. The input layer fuses the intrinsic interaction information and cross-interaction information of the investment attraction party attribute graph and the investor attribute graph to obtain an embedded representation of the investment attraction party and the investor, which is input into the graph convolution layer. The graph convolution layer obtains the final representation of the investment attraction party and the investor through multi-layer aggregation operations, and the final representation is input into the prediction layer. The prediction layer obtains the correlation score of the investment attraction party and the investor through a three-layer multi-layer perceptron.

[0059] The output module is used to select a predetermined number of investors with the highest scores and recommend them to the investment invitation party.

[0060] In a third aspect, the present invention provides an electronic device, characterized in that the electronic device comprises:

[0061] at least one processor; and

[0062] a memory communicatively connected to the at least one processor; wherein,

[0063] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the investment promotion information recommendation method described in the first aspect.

[0064] In a fourth aspect, the present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the investment information recommendation method described in the first aspect when executed.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The investment promotion information recommendation method, electronic device and readable storage medium provided by the present invention, aiming at the problem of investment promotion information recommendation, fuse internal interaction information and cross-interaction information in the input layer, which can better capture the complex relationship between entities, provide richer data support for the recommendation algorithm, and improve the performance of the recommendation system; the graph convolution layer can capture the high-order connectivity between the investment promotion party and the investor through multi-layer aggregation of the neighbor nodes of the target node, thereby improving the recommendation quality; the prediction layer uses a three-layer multi-layer perceptron to calculate the correlation score, which better captures the data characteristics of the interaction between the investment promotion party and the investor, and improves the accuracy of the correlation score.

[0067] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The above and other purposes, features and advantages of the present invention will become more apparent by describing the embodiments of the present invention in more detail in conjunction with the accompanying drawings. The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0069] Figure 1 It is a flow chart of a method for recommending investment information provided by an exemplary embodiment of the present invention.

[0070] Figure 2 It is a structural diagram of an investment information recommendation system provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0071] Below, the exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments of the present invention, and it should be understood that the present invention is not limited to the exemplary embodiments described here.

[0072] It should be noted that the relative arrangement of components and steps, the numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.

[0073] Those skilled in the art can understand that the terms "first" and "second" in the embodiments of the present invention are only used to distinguish different steps, devices or modules, etc., and neither represent any specific technical meaning nor indicate the necessary logical order between them.

[0074] It should also be understood that, in the embodiments of the present invention, “plurality” may refer to two or more than two, and “at least one” may refer to one, two or more than two.

[0075] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more, unless explicitly limited or otherwise indicated in the context.

[0076] Example 1

[0077] This embodiment provides a method for recommending investment information, which can be performed by an investment information recommendation device. The method is suitable for analyzing the investment party data and the investor party data using an investment information recommendation system, thereby recommending investors to the investment party. The investment information recommendation device can be implemented in the form of hardware and / or software, and the investment information recommendation device can be configured in an electronic device with investment information recommendation method capabilities.

[0078] like Figure 1 As shown, the method provided in this embodiment includes the following steps:

[0079] S110, collecting the merchant data and the investor data, and pre-processing the merchant data and the investor data;

[0080] Among them, investor data can be obtained by searching the target company's information on the Internet, including official websites, social media, industry forums, etc., or by using industry databases or professional data service platforms to provide detailed company information, including financial status, business scope, management information, etc., and industry statistics and company registration information can also be obtained through relevant government departments.

[0081] S120, constructing a knowledge graph of the merchant and investor, as well as a merchant attribute graph and an investor attribute graph according to the preprocessed data;

[0082] S130, constructing an investment promotion information recommendation model, the investment promotion information recommendation model comprising an input layer, a graph convolution layer, and a prediction layer, the input layer fuses the intrinsic interaction information and the cross-interaction information of the investment promotion party attribute graph and the investor attribute graph to obtain embedded representations of the investment promotion party and the investor, the embedded representations of the investment promotion party and the investor are input to the graph convolution layer, the graph convolution layer obtains the final representations of the investment promotion party and the investor through multi-layer aggregation operations, the final representations are input to the prediction layer, the prediction layer obtains the correlation scores of the investment promotion party and the investor through three-layer multi-layer perceptron;

[0083] S140: Select a predetermined number of investors with the highest scores and recommend them to the investment invitation party.

[0084] In step S110, the investment promotion data includes local industrial development planning, industrial structure, industrial policy, financial subsidies, land use policy, industrial support policy, park infrastructure, information on settled enterprises, and industrial positioning of the park.

[0085] The investor data includes investment intentions, site preferences, corporate financial information, profitability, market size, corporate litigation information, business scope, corporate qualifications, and historical investment information.

[0086] In step S110, the collected raw data may contain errors, missing values ​​or outliers. Through data cleaning and preprocessing, these errors can be corrected, missing values ​​can be filled, and outliers can be identified and processed, thereby improving the overall quality of the data. At the same time, data preprocessing helps to ensure that the data used by the recommendation system is accurate and reliable, which can improve the relevance of the recommendation and the satisfaction of the investment promotion party.

[0087] Preprocessing includes data cleaning, data format conversion, data normalization, and manual labeling; the data cleaning includes removing duplicate data, processing missing values, correcting erroneous data, and deleting outliers; the data format conversion includes converting the cleaned data into a unified format; the data normalization includes scaling the format-converted data to the interval [0,1]; the manual labeling includes manually labeling the normalized data with categories.

[0088] In step S120, a knowledge graph of the business attracting parties and investors is constructed, with the business attracting parties and investors as entity nodes in the knowledge graph, and edges are created to represent the relationship between the entities.

[0089] In the business invitation attribute graph and the investor attribute graph, the attributes of the business invitation and the attributes of the investor are respectively used as nodes of the graph, and the two nodes are connected by an edge.

[0090] use Represents the embedding vector of node i in the merchant attribute graph, using Represents the embedding vector of node i in the investor attribute graph.

[0091] In step S130, the investment information recommendation model includes an input layer, a graph convolution layer, and a prediction layer. The input layer can better capture the complex relationship between entities by distinguishing the internal interactions and cross-interactions between the investment party attributes and the investor party attributes. This distinction helps the model learn the embedded representation of the node more accurately.

[0092] Internal interactions refer to the relationship between the internal attributes of the same entity. In the merchant attribute map, internal interactions refer to the relationship between the various attributes of the merchant itself, such as the relationship between the geographical location and the preferential policies it provides. Therefore, internal interactions help to deeply understand the characteristics of a single entity.

[0093] In this embodiment, the internal interaction information between nodes is obtained, which is expressed as:

[0094] z ij=MLP(u i ,u j )

[0095] Among them, u i and u j are respectively the embedding vectors of node i and node j of the merchant attribute graph or the investor attribute graph, MLP() is a multi-layer perceptron function, z ij is the internal interaction information between node i and node j;

[0096] Get the aggregated internal interaction information of the node, expressed as:

[0097]

[0098] Among them, z i is the aggregated internal interaction information of node i in the merchant attribute graph or the investor attribute graph, ij is the internal interaction information between node i and node j, N i is the neighbor set of node i.

[0099] In addition, cross-interaction refers to the attribute relationship between different entities. In this embodiment, cross-interaction may refer to the relationship between the industrial advantages of the business attracting party and the investment preferences of the investor, or the relationship between the financial strength of the investor and the project funding needs of the business attracting party. Cross-interaction helps to discover the potential connections and matching between different entities.

[0100] Get the cross-interaction information between nodes, expressed as:

[0101]

[0102] Among them, u i and are respectively the embedding vector of node i of the merchant attribute graph and the embedding vector of node j of the investor attribute graph, or the embedding vector of node i of the investor attribute graph and the embedding vector of node j of the merchant attribute graph, s ij is the cross-interaction information between node i and node j, and e is the element-by-element multiplication;

[0103] Get the aggregated cross-interaction information of the nodes, expressed as:

[0104]

[0105] Among them, s i and are respectively the aggregated cross-interaction information of node i of the merchant attribute graph and the node set of the investor attribute graph, or the aggregated cross-interaction information of node i of the investor attribute graph and the node set of the merchant attribute graph, s ij is the cross interaction information between node i and node j.

[0106] At the same time, in this embodiment, the input layer aggregates the attribute vector representations of the merchant and the investor as the input of the graph convolution layer. This aggregation method can integrate more feature information, thereby providing richer data support for the recommendation algorithm.

[0107] Get the fusion information of the node's internal interaction information and cross-interaction information, expressed as:

[0108] u′ i =GRU(u i ,z i ,s i )

[0109] Among them, u i ′ is the fusion information of the internal interaction information and cross-interaction information of node i in the merchant attribute graph or the investor attribute graph, u i is the embedding vector of node i, s i is the aggregated cross-interaction information of node i, z i is the aggregated internal interaction information of node i, GRU() is the gated recurrent unit function;

[0110] Get the embedded representation of the merchant and investor, expressed as:

[0111]

[0112] Among them, u i ′ is the fusion information of the internal interaction information and the cross-interaction information of the node i of the investment attraction attribute map or the investor attribute map, and V is the node set of the investment attraction attribute map or the investor attribute map.

[0113] In step S130, the graph convolution layer updates the node embedding by aggregating the information of neighbor nodes through multi-layer aggregation of the target node's neighbor nodes, which helps to generate richer and more accurate representations of the investment and investment parties. The graph convolution layer only aggregates the information of the first-order neighbors, which can better control the signal strength and reduce the impact of noise, thereby improving the robustness of the recommendation system. At the same time, the graph convolution layer avoids additional activation functions and linear transformation layers, which can obtain a more comprehensive embedding without adding additional adjustable parameters.

[0114] The graph convolution layer has a multi-layer structure. In the knowledge graph of the investment party and the investor, each layer of the graph convolution layer performs weighted summation of the neighboring nodes of the target node to obtain an embedded representation of each layer, which is expressed as:

[0115]

[0116] in, and are the embedding representations of investors at the kth and k+1th layers respectively. and are the embedding representations of the merchant agents at the kth and k+1th layers, respectively. u is the neighbor node set of the target investment attraction, N i It is the set of neighbor nodes of the target investor.

[0117] The embedding representations of different layers are merged to get the final representation, which is expressed as:

[0118]

[0119] in, and are the embedding representations of the merchant and investor at the kth layer, α k is the weight of the kth layer, K is the number of layers of the graph convolutional layer, e u and e i They are the final statements of the business invitation party and the investor respectively.

[0120] In step S130, the prediction layer uses a three-layer multilayer perceptron (MLP). The MLP network is used as an interaction function between the investment agent and the investor. It predicts the investment agent's interest in the investor by learning the complex relationship between the investment agent and the investor's embedding vectors, thereby improving the overall performance of the recommendation system. The three-layer setting can learn the hidden vectors of the investment agent and the investor well without making the model too complicated.

[0121] The correlation scores between the merchant and the investor are expressed as:

[0122] r=φ3(φ2(φ1(P T e u ,Q T e i )))

[0123] Among them, φ3, φ2 and φ1 are the mapping functions of the third, second and first layers of multilayer perceptrons respectively, P T and Q T are the potential factor matrices of the merchant and investor, respectively, u and e iare the final representations of the merchant and investor respectively, and r is the correlation score between the merchant and investor.

[0124] In step S130, the investment information recommendation model is optimized using a loss function, where the loss function is a BPR loss function.

[0125] The BPR loss function is a loss function used in recommender systems and information retrieval, based on the Bayesian Personalized Ranking (BPR) framework. BPR is a recommendation algorithm based on a probabilistic model that aims to optimize users' predictions of item rankings.

[0126] The advantage of the BPR loss function is that it directly optimizes the quality of item ranking rather than just predicting the rating of a single item. This enables the model to better capture the user's overall preference for items, thereby improving the accuracy and satisfaction of the recommendation system.

[0127] In step S140, for an investment invitation, investors are sorted from high to low according to the scores, and a preset number of investors ranked top are selected and recommended to the investment invitation.

[0128] Example 2

[0129] This embodiment provides a system for recommending investment information. Figure 2 As shown, the system includes:

[0130] The collection module 100 is used to collect the merchant data and the investor data, and pre-process the merchant data and the investor data;

[0131] The recommendation module 200 constructs a knowledge graph of the investment attraction party and the investor, as well as an investment attraction party attribute graph and an investor attribute graph according to the preprocessed data; the recommendation module includes an investment attraction information recommendation model, and the investment attraction information recommendation model includes an input layer 210, a graph convolution layer 220, and a prediction layer 230. The input layer 210 fuses the intrinsic interaction information and the cross-interaction information of the investment attraction party attribute graph and the investor attribute graph to obtain an embedded representation of the investment attraction party and the investor, and the embedded representation of the investment attraction party and the investor is input to the graph convolution layer 220. The graph convolution layer 220 obtains the final representation of the investment attraction party and the investor through multi-layer aggregation operations, and the final representation is input to the prediction layer 230. The prediction layer 230 obtains the correlation score between the investment attraction party and the investor through a three-layer multi-layer perceptron.

[0132] The output module 300 is used to select a predetermined number of investors with the highest scores and recommend them to the investment invitation party.

[0133] Example 3

[0134] The present invention provides an electronic device, characterized in that the electronic device comprises:

[0135] at least one processor; and

[0136] a memory communicatively connected to the at least one processor; wherein,

[0137] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the investment promotion information recommendation method described in the first aspect.

[0138] Example 4

[0139] The present invention provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the investment information recommendation method described in the first aspect when executed.

[0140] The basic principles of the present disclosure are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present disclosure are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present disclosure. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, and are not limitations. The above details do not limit the present disclosure to the necessity of adopting the above specific details to be implemented.

[0141] Each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0142] The block diagrams of the devices, apparatuses, equipment, and systems involved in this disclosure are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, referring to "including but not limited to," and can be used interchangeably therewith. The words "or" and "and" used herein refer to the words "and / or," and can be used interchangeably therewith, unless the context clearly indicates otherwise. The word "such as" used herein refers to the phrase "such as but not limited to," and can be used interchangeably therewith.

[0143] The method and apparatus of the present disclosure may be implemented in many ways. For example, the method and apparatus of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present disclosure may also be implemented as a program recorded in a recording medium, which includes machine-readable instructions for implementing the method according to the present disclosure. Therefore, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

[0144] It should also be noted that in the apparatus, equipment and method of the present disclosure, each component or each step can be decomposed and / or recombined. These decompositions and / or recombinations should be regarded as equivalent schemes of the present disclosure. The above description of the disclosed aspects is provided to enable any technician in the field to make or use the present disclosure. Various modifications to these aspects are very obvious to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the present disclosure. Therefore, the present disclosure is not intended to be limited to the aspects shown here, but to the widest scope consistent with the principles and novel features disclosed herein.

[0145] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present disclosure to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A method for recommending investment information, characterized in that: include: Collecting the data of the business invitation party and the investor party, and pre-processing the data of the business invitation party and the investor party; Build the knowledge graph of the merchant and investor, as well as the merchant attribute graph and investor attribute graph based on the preprocessed data; Constructing an investment promotion information recommendation model, the investment promotion information recommendation model includes an input layer, a graph convolution layer, and a prediction layer, the input layer fuses the intrinsic interaction information and the cross-interaction information of the investment promotion party attribute graph and the investor attribute graph to obtain embedded representations of the investment promotion party and the investor, the embedded representations of the investment promotion party and the investor are input to the graph convolution layer, the graph convolution layer obtains the final representations of the investment promotion party and the investor through multi-layer aggregation operations, the final representation is input to the prediction layer, and the prediction layer obtains the correlation score of the investment promotion party and the investor through a three-layer multi-layer perceptron; A predetermined number of investors with the highest scores are selected and recommended to the investment invitation party.

2. The method according to claim 1, characterized in that The collection of merchant data and investor data includes: The investment promotion data include local industrial development planning, industrial structure, industrial policy, financial subsidies, land use policy, industrial support policy, park infrastructure, information on settled enterprises, and industrial positioning of the park; The investor data includes investment intentions, site preferences, corporate financial information, profitability, market size, corporate litigation information, business scope, corporate qualifications, and historical investment information.

3. The method according to claim 1, characterized in that Preprocessing the merchant data and investor data includes: Data cleaning, data format conversion, data normalization, and manual labeling; the data cleaning includes removing duplicate data, processing missing values, correcting erroneous data, and deleting outliers; the data format conversion includes converting the cleaned data into a unified format; the data normalization includes scaling the format-converted data to the [0,1] interval; the manual labeling includes manually labeling the normalized data with categories.

4. The method according to claim 1, characterized in that: Based on the preprocessed data, the knowledge graph of the merchant and investor is constructed, as well as the merchant attribute graph and the investor attribute graph, including: In the merchant's attribute graph and the investor's attribute graph, the merchant's attributes and the investor's attributes are respectively used as nodes of the graph, and the two nodes are connected by an edge; use Represents the embedding vector of node i in the merchant attribute graph, using Represents the embedding vector of node i in the investor attribute graph.

5. The method according to claim 1, characterized in that The input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including: Get the internal interaction information between nodes, expressed as: With ij =MLP(u i ,at j ) Among them, u i and u j are respectively the embedding vectors of node i and node j of the merchant attribute graph or the investor attribute graph, MLP() is a multi-layer perceptron function, z ij is the internal interaction information between node i and node j; Get the aggregated internal interaction information of the node, expressed as: Among them, z i is the aggregated internal interaction information of node i in the merchant attribute graph or the investor attribute graph, ij is the internal interaction information between node i and node j, N i is the neighbor set of node i.

6. The method according to claim 5, characterized in that The input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including: Get the cross-interaction information between nodes, expressed as: Among them, u i and are respectively the embedding vector of node i of the merchant attribute graph and the embedding vector of node j of the investor attribute graph, or the embedding vector of node i of the investor attribute graph and the embedding vector of node j of the merchant attribute graph, s ij is the cross-interaction information between node i and node j, and e is the element-by-element multiplication; Get the aggregated cross-interaction information of the nodes, expressed as: Among them, s i and are respectively the aggregated cross-interaction information of node i of the merchant attribute graph and the node set of the investor attribute graph, or the aggregated cross-interaction information of node i of the investor attribute graph and the node set of the merchant attribute graph, s ij is the cross interaction information between node i and node j.

7. The method according to claim 6, characterized in that The input layer fuses the intrinsic interaction information and cross-interaction information of the merchant attribute graph and the investor attribute graph to obtain embedded representations of the merchant and the investor, including: Get the fusion information of the node's internal interaction information and cross-interaction information, expressed as: in' i =GRU(in i ,with i ,with i ) Among them, u i ′ is the fusion information of the internal interaction information and cross-interaction information of node i in the merchant attribute graph or the investor attribute graph, u i is the embedding vector of node i, s i is the aggregated cross-interaction information of node i, z i is the aggregated internal interaction information of node i, GRU() is the gated recurrent unit function; Get the embedded representation of the merchant and investor, expressed as: Among them, u′ i is the fusion information of the internal interaction information and the cross-interaction information of the node i of the investment attraction attribute map or the investor attribute map, and V is the node set of the investment attraction attribute map or the investor attribute map.

8. The method according to claim 1, characterized in that The graph convolution layer obtains the final representation of the merchant and investor through multi-layer aggregation operations, including: The graph convolution layer has a multi-layer structure. In the knowledge graph of the investment party and the investor, each layer of the graph convolution layer performs weighted summation of the neighboring nodes of the target node to obtain an embedded representation of each layer, which is expressed as: in, and are the embedding representations of investors at the kth and k+1th layers respectively. and are the embedding representations of the merchant agents at the kth and k+1th layers, respectively. u is the neighbor node set of the target investment attraction, N i It is the set of neighbor nodes of the target investor.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the investment information recommendation method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the investment information recommendation method according to any one of claims 1 to 8 when executed.

Images