A Method, System, Device and Medium for Data Retrieval and Recommendation of Scientific and Technological Innovation Projects

Abstract

The present invention discloses a method, system, device and medium for data retrieval and recommendation of scientific and technological innovation projects, specifically relating to the technical field of data association. The technical key points are as follows: constructing a knowledge graph by taking each entity in the scientific and technological innovation project data as a graph node; determining the mapped entity of the retrieved entity in the knowledge graph, calculating the association strength between the mapped entity and the related entities, constructing a neighborhood representation of the mapped entity based on the association strength, and expanding the features of the mapped entity according to the neighborhood representation; propagating the historical retrieval interaction entities in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entities at each layer of the knowledge graph, calculating the correlation between the historical retrieval interaction entities and the neighbor entities at each layer, and aggregating all the calculated correlations to obtain the retrieval interest features; performing weighted calculation on the first prediction result and the second prediction result, and outputting the project data retrieval and recommendation result based on the weighted prediction value.

Description

Technical Field

[0001] The present invention relates to the technical field of data association, and particularly to a method, system, device and medium for data retrieval and recommendation of scientific and technological innovation projects. Background Art

[0002] The scientific and technological innovation management system is an enterprise-level centralized management system for scientific research projects, including project management, high-tech enterprise management, scientific and technological achievement management, achievement transformation, knowledge base, etc. By means of a unified platform, unified planning and unified construction, the data channels are opened up, information islands are eliminated, vertical penetration of data and horizontal integration of functions are realized, which also supports the subsequent goals of building and managing a data center and an enterprise service bus.

[0003] In the existing scientific and technological innovation management system, when an enterprise needs to form some certification materials or project reports and needs to retrieve or query data of the same theme or the same type in multiple sections, since the information retrieval and data recommendation of each section are relatively independent and there is a large amount of data in each section, the traditional method is to separately retrieve relevant data in each section and then fill in the form. Therefore, it is difficult to match data of the same theme under different sections, it is difficult to generate a complete data report or certification material, and data support cannot be provided efficiently and accurately.

[0004] Therefore, the present invention aims to provide a method, system, device and medium for data retrieval and recommendation of scientific and technological innovation projects to solve the above-mentioned related problems. Summary of the Invention

[0005] The technical problem to be solved by the present invention is that it is difficult to match data of the same theme under different sections, it is difficult to generate a complete data report or certification material, and data support cannot be provided efficiently and accurately. The purpose is to provide a method, system, device and medium for data retrieval and recommendation of scientific and technological innovation projects. By constructing a knowledge graph for scientific and technological innovation project data, it is convenient to integrate relevant data in each section and facilitate the subsequent call of data recommendation; at the same time, the mapping entity corresponding to the retrieval entity is determined in the knowledge graph, the association strength between the mapping entity and the entities with relationships in the knowledge graph is calculated, and then the neighborhood representation of the mapping entity is constructed based on the association strength, and the features of the mapping entity are expanded according to the neighborhood representation, which can increase the feature representation of the entity during the data retrieval process, enrich the semantic information of the recommended data, avoid recommendation deviation and data recommendation missing caused by insufficient information, and solve the technical problem that it is difficult to match data of the same theme under different sections, it is difficult to generate a complete data report or certification material, and data support cannot be provided efficiently and accurately.

[0006] The present invention is realized by the following technical solutions:

[0007] A data retrieval and recommendation method for scientific and technological innovation projects, the method comprising:

[0008] Obtain scientific and technological innovation project data, and construct a knowledge graph with each entity in the scientific and technological innovation project data as a graph node;

[0009] Extract the retrieval entity in the retrieval information, determine the corresponding mapped entity of the retrieval entity in the knowledge graph, calculate the association strength between the mapped entity and the entities having relationships in the knowledge graph, construct a neighborhood representation of the mapped entity based on the association strength, and perform feature augmentation on the mapped entity according to the neighborhood representation, and input the mapped entity after feature augmentation into a pre-constructed first prediction model to calculate a first prediction result;

[0010] Propagate the historical retrieval interaction entities in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entities in each layer of the knowledge graph, calculate the correlation between the historical retrieval interaction entities and the neighbor entities in each layer, and aggregate all the calculated correlations to obtain a retrieval interest feature, and input the retrieval interest feature into a pre-constructed second prediction model to calculate a second prediction result;

[0011] Calculate the relative entropy between the first prediction result and the second prediction result to obtain a prediction result deviation value, and perform weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output a final project data retrieval and recommendation result according to the fused prediction value.

[0012] Further, each entity in the knowledge graph includes an index entity and a project entity.

[0013] Further, extracting the retrieval entity in the retrieval information, determining the corresponding mapped entity of the retrieval entity in the knowledge graph, calculating the association strength between the mapped entity and the entities having relationships in the knowledge graph, constructing a neighborhood representation of the mapped entity based on the association strength, and performing feature augmentation on the mapped entity according to the neighborhood representation, specifically:

[0014] Extract the retrieval entity in the retrieval information, and determine the corresponding mapped entity of the retrieval entity in the knowledge graph;

[0015] Calculate a first association strength between the mapped entity and the project entities having relationships, and calculate a second association strength between the mapped entity and the index entities having relationships, and construct a neighborhood representation of the mapped entity based on the first association strength and the second association strength;

[0016] Use the SUM aggregation method to aggregate the neighborhood representation and the mapped entity to obtain the mapped entity after feature augmentation.

[0017] Further, calculate the second association strength between the mapped entity and the index entity with a relationship, specifically as follows:

[0018] In the knowledge graph, filter out the index entities related to the mapped entity, and find all the relationship paths from the mapped entity to the index entity;

[0019] Use the pre-built knowledge representation model to calculate the semantic similarity of all relationship paths, specifically as follows: , where represents the number of hops of the relationship path; represents the weight of the th hop; represents the semantic similarity of the th hop; represents the relationship path; represents the mapped entity, and the pre-built knowledge representation model is obtained by training the TransE model;

[0020] Use the average pooling algorithm to aggregate the semantic similarities of all relationship paths to obtain the second association strength, where the path aggregation weight of each relationship path is dynamically adjusted through the attention mechanism, specifically as follows: , where represents the second association strength; represents the path aggregation weight.

[0021] Further, propagate the historical retrieval interaction entity in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity in each layer of the knowledge graph, calculate the correlation between the historical retrieval interaction entity and the neighbor entities in each layer, and aggregate all the calculated correlations to obtain the retrieval interest feature, specifically as follows:

[0022] Use the historical retrieval interaction entity as the starting point and propagate it in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity in each layer of the knowledge graph;

[0023] Calculate the correlation between the historical retrieval interaction entity and the neighbor entities in each layer, and multiply the correlation by the neighbor entities in the previous layer to calculate the entity response in each layer;

[0024] Perform hierarchical neighborhood summation on the entity responses in each layer to obtain the retrieval interest feature.

[0025] Further, after performing weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output the final project data retrieval recommendation result according to the fused prediction value, specifically as follows: , where represents the fused predicted value; represents the weight value, ; respectively represent the first prediction result and the second prediction result; represents the prediction result deviation value.

[0026] The present invention also provides a data retrieval and recommendation system for scientific and technological innovation projects. This system is used in any one of the above-mentioned data retrieval and recommendation methods for scientific and technological innovation projects. The system includes:

[0027] A knowledge graph construction module, which is used to obtain scientific and technological innovation project data and construct a knowledge graph with each entity in the scientific and technological innovation project data as graph nodes;

[0028] A first prediction module, which is used to extract the retrieval entity in the retrieval information, determine the mapped entity corresponding to the retrieval entity in the knowledge graph, calculate the association strength between the mapped entity and the entities related in the knowledge graph, construct a neighborhood representation of the mapped entity based on the association strength, and expand the features of the mapped entity according to the neighborhood representation, and input the mapped entity with expanded features into a pre-constructed first prediction model to calculate the first prediction result;

[0029] A second prediction module, which is used to propagate the historical retrieval interaction entities in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entities in each layer of the knowledge graph, calculate the correlation between the historical retrieval interaction entities and the neighbor entities in each layer, and aggregate all the calculated correlations to obtain the retrieval interest feature, and input the retrieval interest feature into a pre-constructed second prediction model to calculate the second prediction result;

[0030] A retrieval and recommendation module, which is used to calculate the relative entropy between the first prediction result and the second prediction result to obtain the prediction result deviation value, and perform weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output the final project data retrieval and recommendation result according to the fused predicted value.

[0031] The present invention also provides a computer device, including a system memory and a processor. The system memory stores a computer program, and when the processor executes the computer program, it implements the steps of the method described in any one of the above.

[0032] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method described in any one of the above.

[0033] The present invention also provides a computer program product including instructions, which, when run on a computer device cluster, cause the computer device cluster to execute the method described in any one of the above.

[0034] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0035] In the present invention, by constructing a knowledge graph for the scientific and technological innovation project data, it is convenient to integrate the relevant data of each section, which is convenient for the subsequent invocation of data recommendation; at the same time, in the knowledge graph, the mapping entity corresponding to the retrieved entity is determined, the association strength between the mapping entity and the related entities in the knowledge graph is calculated, and then the neighborhood representation of the mapping entity is constructed based on the association strength, and the features of the mapping entity are expanded according to the neighborhood representation, which can increase the feature representation of the entity during the data retrieval process, enrich the semantic information of the recommended data, avoid the recommendation deviation and the lack of data recommendation caused by insufficient information, and solve the technical problem that it is difficult to match the data of the same theme under different sections, it is difficult to generate a complete data report or certification materials, and it is impossible to provide data support efficiently and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0037] Figure 1 is a schematic flowchart of a method for retrieving and recommending scientific and technological innovation project data in this embodiment;

[0038] Figure 2 is a schematic diagram of the system modules of a scientific and technological innovation project data retrieval and recommendation system in this embodiment;

[0039] Figure 3 is a schematic diagram of the structure of a computer device in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] The following describes exemplary embodiments of the present disclosure, including various details of the embodiments of the present disclosure to facilitate understanding. They should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described here without departing from the scope of the present disclosure. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted below.

[0041] In this disclosure, unless otherwise specified, the use of terms such as "first" and "second" to describe various elements does not intend to limit the positional relationship, temporal relationship, or importance relationship of these elements. Such terms are only used to distinguish one element from another. In some examples, the first element and the second element may refer to the same instance of the element, and in certain cases, based on the context description, they may also refer to different instances.

[0042] The terms used in the description of various examples in this disclosure are only for the purpose of describing specific examples and are not intended to be restrictive. Unless the context clearly indicates otherwise, if the number of elements is not specifically limited, the element can be one or more. In addition, the term "and / or" used in this disclosure covers any one of the listed items and all possible combinations.

[0043] Embodiment 1

[0044] See Figure 1 As shown, this embodiment provides a method for retrieving and recommending data of scientific and technological innovation projects. The method includes:

[0045] S1: Obtain the data of scientific and technological innovation projects, and construct a knowledge graph with each entity in the data of scientific and technological innovation projects as graph nodes;

[0046] It should be noted that in this embodiment, the data of scientific and technological innovation projects are stored in the database of the scientific and technological innovation management system, and the data of scientific and technological innovation projects are extracted from the database. Among them, the data of scientific and technological innovation projects include R & D activity data, technology product data, and intellectual property data. Among them, the R & D activity data includes R & D activity number information, R & D activity name information, technology field information, technology source information, technology purpose, etc.; the technology product data includes product number information, technology source information, key technology information, technology neighborhood information, etc.; the intellectual property data includes intellectual property name information, intellectual property category information, application year information, intellectual property number information, authorized patent number information, authorized date information, etc.; in other embodiments, other information data may also be included, which will not be elaborated here;

[0047] At the same time, each entity in the knowledge graph includes a project entity and an index entity. The index entity is composed of each data information index in the data of scientific and technological innovation projects. Therefore, the entities in the knowledge graph correspond one-to-one with the data of scientific and technological innovation projects; at the same time, the knowledge graph is composed of a head entity, a tail entity, and a relationship, and can be described by a triple <head entity, relationship, tail entity>.

[0048] S2: Extract the retrieval entities in the retrieval information, determine the corresponding mapped entities in the knowledge graph, calculate the association strength between the mapped entities and the entities related in the knowledge graph, construct the neighborhood representation of the mapped entities based on the association strength, and expand the features of the mapped entities according to the neighborhood representation. Then input the mapped entities with expanded features into the pre-constructed first prediction model to calculate the first prediction result;

[0049] Specifically, in this embodiment, first extract the retrieval entities in the retrieval information and determine the corresponding mapped entities in the knowledge graph;

[0050] It should be noted that in this embodiment, the retrieval information can be obtained by the user inputting a retrieval formula or automatically obtained by the system according to the form generation requirements. This technical means is a conventional technical means and will not be elaborated here. At the same time, the extracted retrieval entities can be 1, 2, 3 or more, depending on the actual retrieval information;

[0051] Meanwhile, it should be noted that in this embodiment, the mapped entities refer to the entities found in the knowledge graph corresponding to the retrieval entities. For example, the retrieval entities are: "transportation", "rotating device", etc. The entities found in the knowledge graph based on these retrieval entities are the mapped entities.

[0052] Calculate the first association strength between the mapped entities and the related item entities, and calculate the second association strength between the mapped entities and the related index entities, and construct the neighborhood representation of the mapped entities based on the first association strength and the second association strength;

[0053] It should be noted that in this embodiment, the item entities or index entities related to the mapped entities refer to that the mapped entities are used as a head entity or a tail entity in a triple, and the other head entity or tail entity of the same triple is the item entity or index entity related to the mapped entity;

[0054] Calculate the first association strength between the mapped entities and the related item entities: It is necessary to calculate the number of associations between the mapped entities and the item entities under the relationship by using the inner product method, and then normalize all the association numbers related to the mapped entity to calculate the first association strength between the mapped entities and the related item entities , ,where represents the number of associations between the mapped entity and the item entity , represents the entity set, represents the mapped entity, represents the entity set connected by the mapped entity;

[0055] Calculate the second association strength between the mapped entity and the metric entity with a relationship: Filter the metric entities related to the mapped entity in the knowledge graph and find all the relationship paths from the mapped entity to the metric entity; Use the pre-constructed knowledge representation model to calculate the semantic similarity of all relationship paths. Specifically: , where represents the hop count of the relationship path; represents the weight of the th hop; represents the semantic similarity of the th hop; represents the metric entity; represents the relationship path; represents the mapped entity; Aggregate the semantic similarities of all relationship paths using the average pooling algorithm to obtain the second association strength. Among them, the path aggregation weight of each relationship path is dynamically adjusted through the attention mechanism. Specifically: , where represents the second association strength; represents the path aggregation weight;

[0056] It should be noted that in this embodiment, the pre-constructed knowledge representation model is obtained by training the TransE model.

[0057] Construct the neighborhood representation of the mapped entity based on the first association strength and the second association strength: Multiply the first association strength, the second association strength, and the corresponding in the entity set to obtain the vector of the triple, and sum the vectors of the triples with the same head entity as the outer neighborhood representation , .

[0058] Use the SUM aggregation method to aggregate the neighborhood representation and the mapped entity to obtain the mapped entity with feature expansion , , where represents the non-linear function; represent the weight and bias respectively.

[0059] Input the mapped entity with feature expansion into the pre-constructed first prediction model to calculate the first prediction result. It should be noted that in this embodiment, the first prediction model is the KGCN model, which is a knowledge graph convolutional neural network model and is used to predict the recommended data based on the mapped entity with feature expansion in this retrieval and recommendation method; It improves the accuracy and diversity of recommendations by combining the structural information of the knowledge graph and the user's personalized preferences; This technical solution is a conventional technical means and will not be elaborated here too much.

[0060] S3: Propagate the historical retrieval interaction entity in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity in each layer of the knowledge graph. Calculate the correlation between the historical retrieval interaction entity and the neighbor entities in each layer, and aggregate all the calculated correlations to obtain the retrieval interest feature. Input the retrieval interest feature into the pre-constructed second prediction model to calculate the second prediction result;

[0061] Specifically, in this embodiment, the historical retrieval interaction entity is used as the starting point and propagated in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity in each layer of the knowledge graph;

[0062] Calculate the correlation between the historical retrieval interaction entity and the neighbor entities in each layer , , where represents the transpose of the vector representation of the historical retrieval interaction entity, represents the th relation vector, represents the th head entity vector. Multiply the correlation with the neighbor entities in the previous layer to calculate the entity response in each layer; perform hierarchical neighborhood summation on the entity responses in each layer to obtain the retrieval interest feature;

[0063] Input the retrieval interest feature into the pre-constructed second prediction model to calculate the second prediction result. It should be noted that, in this embodiment, the second prediction model adopts the RippleNet model, which is a recommendation model based on the knowledge graph. It expands the user's interest feature through the user's historical retrieval feature. This technical solution is a conventional technical solution in the art and will not be elaborated here.

[0064] S4: Calculate the relative entropy between the first prediction result and the second prediction result to obtain the prediction result deviation value, and perform weighted calculation on the first prediction result and the second prediction result. Integrate the prediction result deviation value with the weighted calculation result, and output the final project data retrieval recommendation result according to the integrated prediction value.

[0065] Specifically, in this embodiment, by constructing a knowledge graph from the scientific and technological innovation project data, it is convenient to integrate the relevant data in each section and facilitate the subsequent invocation of data recommendation. At the same time, the mapping entity corresponding to the retrieved entity is determined in the knowledge graph, the association strength between the mapping entity and the related entities in the knowledge graph is calculated, and then the neighborhood representation of the mapping entity is constructed based on the association strength, and the features of the mapping entity are expanded according to the neighborhood representation, which can increase the feature representation of the entity during the data retrieval process, enrich the semantic information of the recommended data, avoid the recommendation deviation and the lack of data recommendation caused by insufficient information, and solve the technical problems that it is difficult to match the data with the same theme under different sections, it is difficult to generate a complete data report or certification materials, and it is impossible to provide data support efficiently and accurately.

[0066] As a possible implementation manner, the method further includes:

[0067] Calculating the relative entropy between the first prediction result and the second prediction result to obtain a prediction result deviation value;

[0068] After performing weighted calculation on the first prediction result and the second prediction result, the prediction result deviation value is fused with the weighted calculation result, and the final project data retrieval recommendation result is output according to the fused prediction value, specifically: , where, represents the fused prediction value; represents the weight value, ; respectively represent the first prediction result and the second prediction result; represents the prediction result deviation value.

[0069] Specifically, in this embodiment, the prediction result deviation value obtained by calculating the relative entropy between the first prediction result and the second prediction result is used to balance the deviation between the two prediction results, so that the data recommendation result is more in line with the retrieval information.

[0070] Embodiment 2

[0071] Refer to Figure 2 As shown, the present invention further provides a scientific and technological innovation project data retrieval and recommendation system, which is used in any one of the above-mentioned scientific and technological innovation project data retrieval and recommendation methods. The system includes:

[0072] A knowledge graph construction module 100, configured to obtain scientific and technological innovation project data and construct a knowledge graph with each entity in the scientific and technological innovation project data as graph nodes;

[0073] The first prediction module 200 is configured to extract the retrieved entity in the retrieved information, determine the mapped entity corresponding to the retrieved entity in the knowledge graph, calculate the association strength between the mapped entity and the entities related to it in the knowledge graph, construct a neighborhood representation of the mapped entity based on the association strength, and perform feature augmentation on the mapped entity according to the neighborhood representation. Then, the mapped entity with augmented features is input into a pre-constructed first prediction model to calculate the first prediction result.

[0074] The second prediction module 300 is configured to propagate the historical retrieval interaction entity in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity in each layer of the knowledge graph, calculate the correlation between the historical retrieval interaction entity and the neighbor entities in each layer, and aggregate all the calculated correlations to obtain the retrieval interest feature. Then, the retrieval interest feature is input into a pre-constructed second prediction model to calculate the second prediction result.

[0075] The retrieval recommendation module 400 is configured to calculate the relative entropy between the first prediction result and the second prediction result to obtain the prediction result deviation value, perform weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output the final item data retrieval recommendation result according to the fused prediction value.

[0076] It should be noted that the modules in the system of Embodiment 2 correspond to the steps in the method of Embodiment 1. The steps in the method of Embodiment 1 have been elaborated in detail in Embodiment 1, and the content of the modules in the system will not be elaborated in detail in Embodiment 2 here.

[0077] Embodiment 3

[0078] See Figure 3 As shown, this embodiment also provides a computer device, including a system memory 1005 and a processor 1001. The system memory 1005 stores a computer program, and when the processor 1001 executes the computer program, it implements the steps of the method in any one of the above.

[0079] It should be noted that the processor 1001 is configured to execute the steps in the above method embodiments according to the instructions in the program code. Alternatively, when the processor 1001 executes the computer program, it implements the functions of each module / unit in the above system / device embodiments.

[0080] Specifically, in this embodiment, the computer program can be divided into one or more modules / units. One or more modules / units are stored in the system memory 1005 and executed by the processor 1001 to complete this application. One or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0081] The terminal device can be a computing device such as a desktop computer, a notebook, a handheld computer, and a cloud server. The terminal device may include, but is not limited to, a processor 1001 and a system memory 1005. Those skilled in the art can understand that this does not constitute a limitation on the terminal device, and it may include more or fewer components than those shown in the figure, or combine certain components, or different components. For example, the terminal device may further include an input / output device 1003, a network access device 1002, a bus 1006, etc.

[0082] The processor 1001 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may also be any conventional processor, etc.

[0083] The system memory 1005 may be an internal storage unit of the terminal device, such as the hard disk or memory of the terminal device. The system memory 1005 may also be a storage device 1004 of the terminal device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal device. Further, the system memory 1005 may also include both the internal storage unit of the terminal device and the storage device 1004. The system memory 1005 is used to store computer programs and other programs and data required by the terminal device. The system memory 1005 may also be used to temporarily store data that has been output or will be output.

[0084] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described systems, systems, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0085] Embodiment 4

[0086] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the above methods are implemented.

[0087] Among them, a computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk. Random Access Memory (RAM), Read-Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), a register, a hard disk, an optical fiber, a portable compact disc read-only memory (Compact Disc Read-Only Memory, CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above, or any other form of computer-readable storage medium in the art of numerical values.

[0088] An exemplary storage medium is coupled to a processor, enabling the processor to read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an Application Specific Integrated Circuit (ASIC). In an embodiment of the present invention, the computer-readable storage medium can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, apparatus, or device.

[0089] Embodiment 5

[0090] This embodiment also provides a computer program product containing instructions. When the instructions are run by a computer device cluster, the computer device cluster is caused to execute the method described in Embodiment 1.

[0091] The specific embodiments described above further elaborate on the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above is only the specific embodiment of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for retrieving and recommending scientific and technological innovation project data, characterized in that: include: Obtain scientific and technological innovation project data, and use each entity in the scientific and technological innovation project data as a graph node to construct a knowledge graph; Extract the search entity in the search information, determine the mapping entity corresponding to the search entity in the knowledge graph, calculate the association strength between the mapping entity and the entity with relationship in the knowledge graph, construct a neighborhood representation of the mapping entity based on the association strength, and perform feature expansion on the mapping entity according to the neighborhood representation, input the mapping entity after feature expansion into the pre-constructed first prediction model, and calculate and obtain the first prediction result; The historical retrieval interaction entity is propagated in the knowledge graph according to the entity connection relationship to obtain the neighbor entity of the historical retrieval interaction entity in each layer of the knowledge graph, the correlation between the historical retrieval interaction entity and the neighbor entity of each layer is calculated, and all the calculated correlations are aggregated to obtain the retrieval interest feature, and the retrieval interest feature is input into the pre-built second prediction model to calculate the second prediction result; Calculate the relative entropy between the first prediction result and the second prediction result to obtain the prediction result deviation value, perform weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output the final project data retrieval recommendation result according to the fused prediction value; Among them, the first prediction model is the KGCN model, and the second prediction model adopts the RippleNet model; the first prediction result refers to the data retrieval prediction result of the project data retrieval prediction based on the mapping entity after feature expansion using the KGCN model; the second prediction result refers to the data retrieval prediction result of the project data retrieval prediction based on the retrieval interest feature using the RippleNet model.

2. A method for retrieving and recommending scientific and technological innovation project data according to claim 1, characterized in that: The entities in the knowledge graph include indicator entities and project entities.

3. A method for retrieving and recommending scientific and technological innovation project data according to claim 2, characterized in that: Extract the search entity in the search information, determine the mapping entity corresponding to the search entity in the knowledge graph, calculate the association strength between the mapping entity and the entity with relationship in the knowledge graph, build a neighborhood representation of the mapping entity based on the association strength, and expand the features of the mapping entity according to the neighborhood representation, specifically: Extract the search entity in the search information and determine the mapping entity corresponding to the search entity in the knowledge graph; Calculating a first association strength between a mapping entity and a project entity having a relationship, and calculating a second association strength between the mapping entity and an indicator entity having a relationship, and constructing a neighborhood representation of the mapping entity based on the first association strength and the second association strength; The neighborhood representation and the mapping entity are aggregated using the SUM aggregation method to obtain the mapping entity after feature expansion.

4. A method for retrieving and recommending scientific and technological innovation project data according to claim 3, characterized in that: The second association strength between the mapping entity and the indicator entity having a relationship is calculated, specifically: Filter indicator entities that have relationships with mapping entities in the knowledge graph, and find all relationship paths from mapping entities to indicator entities; The pre-built knowledge representation model is used to calculate the semantic similarity of all relationship paths, specifically: Where k represents the number of hops in the relationship path; β i represents the weight of the i-th hop; Sim() represents the semantic similarity of the i-th hop; e i-1 Represents the head entity; r i Represents the relationship path; e i Representation mapping entity, the pre-built knowledge representation model is obtained by training the TransE model; The average pooling algorithm is used to aggregate the semantic similarities of all relationship paths to obtain the second association strength, where the path aggregation weight of each relationship path is dynamically adjusted through the attention mechanism, specifically: g = ∑ p∈P α p A(p), where g represents the second association strength; α p Indicates the path aggregation weight.

5. The method for retrieving and recommending scientific and technological innovation project data according to claim 1, characterized in that: The historical retrieval interaction entities are propagated in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entities in each layer of the knowledge graph, and the correlation between the historical retrieval interaction entities and the neighbor entities in each layer is calculated. All the calculated correlations are aggregated to obtain the retrieval interest features, which are as follows: Taking the historical retrieval interaction entity as the starting point, propagate it in the knowledge graph according to the entity connection relationship to obtain the neighbor entities of the historical retrieval interaction entity at each layer of the knowledge graph; Calculate the correlation between the historically retrieved interaction entities and the neighbor entities at each layer, and multiply the correlation with the neighbor entities at the previous layer to obtain the entity response at each layer; The hierarchical neighborhood summation is performed on the entity responses of each layer to obtain the retrieval interest features.

6. A method for retrieving and recommending scientific and technological innovation project data according to claim 2, characterized in that: After weighted calculation of the first prediction result and the second prediction result, the prediction result deviation value is fused with the weighted calculation result, and the final project data retrieval recommendation result is output according to the fused prediction value, specifically: Among them, F represents the predicted value after fusion; represents the weight value, f1 and f2 represent the first prediction result and the second prediction result respectively; w represents the deviation value of the prediction result.

7. A scientific and technological innovation project data retrieval and recommendation system, characterized in that: The system is used in a method for retrieving and recommending scientific and technological innovation project data as described in any one of claims 1 to 6, and the system comprises: The knowledge graph construction module is used to obtain scientific and technological innovation project data and construct a knowledge graph using each entity in the scientific and technological innovation project data as a graph node; A first prediction module is used to extract the search entity in the search information, determine the mapping entity corresponding to the search entity in the knowledge graph, calculate the association strength between the mapping entity and the entity having a relationship in the knowledge graph, construct a neighborhood representation of the mapping entity based on the association strength, and perform feature expansion on the mapping entity according to the neighborhood representation, input the mapping entity after feature expansion into a pre-constructed first prediction model, and calculate and obtain a first prediction result; The second prediction module is used to propagate the historical retrieval interaction entity in the knowledge graph according to the entity connection relationship, obtain the neighbor entity of the historical retrieval interaction entity in each layer of the knowledge graph, calculate the correlation between the historical retrieval interaction entity and the neighbor entity of each layer, and aggregate all the calculated correlations to obtain the retrieval interest feature, input the retrieval interest feature into the pre-built second prediction model, and calculate the second prediction result; A retrieval recommendation module is used to calculate the relative entropy between the first prediction result and the second prediction result to obtain a prediction result deviation value, perform weighted calculation on the first prediction result and the second prediction result, fuse the prediction result deviation value with the weighted calculation result, and output a final project data retrieval recommendation result according to the fused prediction value; Among them, the first prediction model is the KGCN model, and the second prediction model adopts the RippleNet model; the first prediction result refers to the data retrieval prediction result of the project data retrieval prediction based on the mapping entity after feature expansion using the KGCN model; the second prediction result refers to the data retrieval prediction result of the project data retrieval prediction based on the retrieval interest feature using the RippleNet model.

8. A computer device comprising a system memory and a processor, wherein the system memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising instructions, characterized in that When the instructions are executed by a computer device cluster, the computer device cluster executes the method according to any one of claims 1 to 6.

Images