Data processing method based on secure multi-party computing

Abstract

The invention relates to the technical field of data processing, particularly discloses a data processing method based on secure multi-party computing, and particularly relates to construction of a knowledge graph based on secure multi-party computing. Through a secure multi-party computing technology, a plurality of participants can jointly construct the knowledge graph on the premise of not exposing respective private data, and the cross-source data is verified in the process of constructing the knowledge graph. The constructed knowledge graph can be used for model training, entity classification, graph rule reasoning and the like.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data processing method based on secure multi-party computing. Background Art

[0002] With the development of big data and artificial intelligence technology, knowledge graphs, as structured graph data used to represent entities and their relationships, have been widely used in many fields (such as intelligent search, recommendation systems, natural language processing, etc.). However, building high-quality knowledge graphs often requires the collaboration of multiple data sources, and these data usually belong to different privacy holders. Traditional knowledge graph construction methods cannot solve the problems of data privacy protection and cross-domain collaboration.

[0003] Therefore, a method is needed that enables multiple parties to process data while ensuring privacy and jointly build a knowledge graph. Summary of the invention

[0004] The purpose of the present invention is to provide a data processing method based on secure multi-party computing to solve the following technical problems: In the prior art, when constructing a knowledge graph based on secure multi-party computing, although multiple parties can jointly construct a knowledge graph without exposing their respective private data, it is difficult to verify cross-source data in the process of constructing the knowledge graph, and the efficiency of constructing the knowledge graph is low.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A data processing method based on secure multi-party computing, the method being executed by a first party participating in secure multi-party computing, comprising the following steps: Decomposing the first private data held by the first party to obtain a plurality of data shards; wherein the plurality of data shards include a first data shard and other data shards, the first data shard is held by the first party, and the other data shards are held by other parties participating in the secure multi-party computation; Receiving a second data shard from another party participating in the secure multi-party computation; the second data shard is obtained by the other party decomposing the second private data owned by the other party; Based on the first data shard and the second data shard, collaborate with other parties participating in the secure multi-party computation to build an initial knowledge graph; Based on the first data shard and the second data shard, local cross-source data verification is performed locally on the first party to obtain a first shard of the verification result; Based on the first shard of the verification result, the other shards of the verification result held by other parties of the secure multi-party computation are coordinated to obtain a complete verification result; wherein the other shards of the verification result held by the other parties are obtained by the other parties through local calculation; Based on the complete verification result, updating the nodes and / or edges of the initial knowledge graph to obtain a target knowledge graph; Execute data processing tasks based on the target knowledge graph; wherein the data processing tasks include entity classification, prediction of relationships between entities, graph rule reasoning, and entity set mining.

[0006] As a further solution of the present invention, the method of performing local cross-source data verification locally on the first party based on the first data fragment and the second data fragment to obtain a first fragment of the verification result includes: Based on the first data shard and the second data shard, a first shard of the verification result is calculated by a preset first calculation formula; the preset first calculation formula is:

[0007] Among them, L Eij represents the jth shard of the verification result calculated by the jth participant based on the data shards he holds, e 1ij represents the i-th data shard of the private data E1 in the first private data held by the j-th participant, e 2ij Represents the i-th data shard of the private data E2 in the second private data held by the j-th participant.

[0008] As a further solution of the present invention, the first shard based on the verification result cooperates with other shards of the verification result held by other parties of the secure multi-party computation to obtain a complete verification result, including: The first party sends the first fragment of the verification result to the other party; receiving other shards of the verification result from the other party; Based on the first fragment of the verification result and the other fragments of the verification result, the complete verification result is obtained by using a preset second calculation formula; the preset second calculation formula is:

[0009] in, represents the least squares optimization value of private data E1 and private data E2, and n represents the total number of data shards.

[0010] As a further solution of the present invention, the first shard based on the verification result cooperates with other shards of the verification result held by other parties of the secure multi-party computation to obtain a complete verification result, including: The first party sends the first fragment of the verification result to the other party; receiving other shards of the verification result from the other party; Based on the first fragment of the verification result and the other fragments of the verification result, the complete verification result is obtained by using a preset third calculation formula; the preset third calculation formula is:

[0011] in, is the value of the preset third calculation formula, B(E,R) is a balance term for measuring the consistency between entity E and relationship R, and λ is a weight coefficient; the entity E is an entity corresponding to the privacy data E1 and the privacy data E2, and the relationship R is used to express the association relationship between the privacy data E1 and the privacy data E2.

[0012] As a further solution of the present invention, the updating of the nodes and / or edges of the initial knowledge graph based on the complete verification result to obtain the target knowledge graph includes: Based on the complete verification result, determining whether the difference between the private data E1 and the private data E2 exceeds a preset value; In response to exceeding the preset value, determining the priority of the first private data and the second private data based on the contribution of each of the first private data and the second private data to constructing the knowledge graph; The nodes and / or edges of the initial knowledge graph are updated based on the priority.

[0013] As a further solution of the present invention, the contribution of the first private data and the second private data to the construction of the knowledge graph is determined by: Determine the data proportion of the data sources corresponding to the first private data and the second private data in the total data used to construct the knowledge graph; The degree of contribution is determined based on the size of the data proportion.

[0014] As a further solution of the present invention, the data type of the node of the knowledge graph includes at least one of text data, voice data and video data.

[0015] A data processing system based on secure multi-party computing, the system is applied to a first party participating in secure multi-party computing, the system comprising: a data decomposition module, configured to decompose the first private data held by the first party to obtain a plurality of data shards; wherein the plurality of data shards include a first data shard and other data shards, the first data shard is held by the first party, and the other data shards are held by other parties participating in the secure multi-party computation; A data receiving module, configured to receive a second data shard from another party participating in the secure multi-party computation; the second data shard is obtained by the other party decomposing the second private data owned by the other party; A first collaboration module, configured to collaborate with other parties participating in the secure multi-party computation to construct an initial knowledge graph based on the first data shard and the second data shard; A data verification module, configured to perform local cross-source data verification on the first party based on the first data shard and the second data shard, to obtain a first shard of the verification result; A second collaboration module is configured to collaborate with other shards of the verification result held by other parties of the secure multi-party computation based on the first shard of the verification result to obtain a complete verification result; wherein the other shards of the verification result held by the other parties are obtained by the other parties through local calculation; A graph updating module, used to update the nodes and / or edges of the initial knowledge graph based on the complete verification result to obtain a target knowledge graph; An execution module is used to perform data processing tasks based on the target knowledge graph; wherein the data processing tasks include entity classification, prediction of relationships between entities, graph rule reasoning, and entity set mining.

[0016] Beneficial effects of the invention: The invention provides a more accurate and effective data processing solution. Through secure multi-party computing technology, it can realize the construction of high-quality and accurate knowledge graphs by multi-party collaboration under the premise of protecting the data privacy of the participants. At the same time, by cross-source verification of data from various participants, it can support the integration of multiple data types, enhance cross-organizational collaboration, and optimize the knowledge graph construction process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be further described below in conjunction with the accompanying drawings.

[0018] Figure 1 It is a flow chart of the data processing method based on secure multi-party computing of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] See also Figure 1 As shown, the present invention is a data processing method based on secure multi-party computing. In some embodiments, the operations shown in process 100 are performed by the first party participating in the secure multi-party computing. Specifically, it can be performed by a data processing system or processor based on secure multi-party computing.

[0021] Secure Multi-Party Computation (SMC) is an encrypted computing technology that allows multiple parties to jointly calculate the result of a function without disclosing their respective private data. Each party only knows its own input and final calculation result, and cannot obtain the input data of other parties. The first party described in the embodiment of the present invention can be any party among the participants in the secure multi-party computing, and the other parties can be any other party among the participants in the secure multi-party computing except the first party. Each party can collaborate with other parties to jointly build a knowledge graph by implementing the method described in the embodiment of the present invention. The first party refers to a specific party participating in the secure multi-party computing (for example, it can be an enterprise, organization or individual), which holds its own first private data and performs corresponding operations in the process of the present invention.

[0022] Step 101: decompose the first private data held by the first party to obtain multiple data fragments.

[0023] The multiple data shards include a first data shard and other data shards, the first data shard is held by the first party, and the other data shards are held by other parties participating in the secure multi-party computing.

[0024] Private data refers to sensitive information or data that is unwilling to disclose specific information about individuals, companies or other entities, such as personal identity information, financial data, health records, etc. First-party private data refers to private data held by the first party.

[0025] Data sharding refers to dividing private data into several small pieces (shards). The decomposed data shards can be held by different parties and processed in the secure multi-party computing process. Each data shard cannot independently restore the complete data to ensure data privacy.

[0026] The first data shard refers to the data shard obtained by decomposing the first party's private data and held by the first party. The other data shards refer to the other data shards other than the first data shard obtained by decomposing the first private data and held by other parties participating in the secure multi-party computing.

[0027] In some embodiments, the first private data can be decomposed into multiple encrypted data fragments by using an encryption algorithm (such as homomorphic encryption or a secret sharing protocol), and communication with other parties can be established through a secure multi-party computing protocol to send other data fragments to other parties.

[0028] Step 102: Receive a second data shard from other parties participating in the secure multi-party computation.

[0029] The second data fragment is obtained by decomposing the second private data owned by the other party.

[0030] In some embodiments, the first party may establish communication with other parties through a secure multi-party computing protocol to obtain the second data slice of the other party. For example, a secure channel (such as SSL / TLS protocol) may be used to ensure that the data is not leaked during the transmission process, and the second data slice is also encrypted to ensure that the privacy data of other parties can also be protected during the transmission process.

[0031] Step 103: Based on the first data shard and the second data shard, collaborate with other parties participating in the secure multi-party computation to build an initial knowledge graph.

[0032] Collaboration refers to performing calculations with other parties based on their respective data shards based on a secure multi-party computing protocol.

[0033] Data fusion: The first party fuses the first data shard it holds with the second data shard it receives, and uses encryption calculation methods (such as homomorphic encryption operations) to perform calculations to ensure that the privacy of the data is protected.

[0034] For example, the first party can provide the first data shard to the third party (which can be a neutral party or an executor agreed by multiple parties, such as a device, etc.) to merge with other data shards owned by other parties. At the same time, the first party will also provide the second data shard provided by other participating parties. In this way, the data shards of each party can be merged to jointly construct the initial knowledge graph.

[0035] The nodes of the initial knowledge graph represent entities, and the edges represent the relationships between entities. For example, if the first data shard and the second data shard both contain information about an entity (such as "company" and "employee"), the initial knowledge graph can generate corresponding entity nodes and relationship edges.

[0036] In some embodiments, an initial knowledge graph can be constructed based on the data after the data shards are fused using a graph construction algorithm (such as graph embedding technology, relationship extraction, etc.).

[0037] Step 104: Based on the first data shard and the second data shard, local cross-source data verification is performed locally on the first party to obtain a first shard of the verification result.

[0038] Data consistency verification refers to the first party comparing the first data shard it holds with the second data shard of the other party to verify whether the two shards are consistent in certain attributes (such as data type, value range, etc.).

[0039] In some embodiments, the first party may generate a first fragment of a local verification result based on the comparison result, and the fragment includes the first party's preliminary judgment on the consistency of the data.

[0040] In this step, since the local cross-source data verification is based on data sharding and is executed locally on the first party, the number of collaborations with other parties can be reduced and the efficiency of cross-source data verification can be improved.

[0041] In some embodiments, the locally performing cross-source data verification on the first party based on the first data shard and the second data shard to obtain a first shard of the verification result includes: Based on the first data shard and the second data shard, a first shard of the verification result is calculated by a preset first calculation formula (1). The preset first calculation formula is: (1) Among them, L Eij represents the jth shard of the verification result calculated by the jth participant based on the data shards he holds, e 1ij represents the i-th data shard of the private data E1 in the first private data held by the j-th participant, e 2ij Represents the i-th data shard of the private data E2 in the second private data held by the j-th participant.

[0042] For example, the first party may substitute the first data shard and the second data shard it owns into formula (1) to obtain the first shard of the verification result.

[0043] Among them, the private data E1 can be a certain data in the first private data or a part of the first private data, and the private data E2 can be a certain data in the second private data or a part of the second private data. For example, E1 can be the name of a certain technology company, and E2 can also be the name of a certain technology company. The difference is that E1 can be the full name of the technology company, and E2 is the abbreviation of the technology company. The purpose of cross-source data verification is to verify whether the private data E1 and the private data E2 refer to the same technology company, and after the verification is completed, the entity nodes corresponding to the private data E1 and the private data E2 in the initial knowledge graph are merged (for example, according to the priority of the entity node, the entity node attributes of the private data with higher priority are merged and identified) or the corresponding entity nodes are improved (for example, attributes are added to the entity nodes for distinction).

[0044] Step 105: Collaborate with other shards of the verification results held by other parties of the secure multi-party computation based on the first shard of the verification result to obtain a complete verification result.

[0045] The other shards of the verification result held by the other party are obtained by the other party's local calculation.

[0046] The first party can provide the first shard of the verification result to the third party or exchange the verification result shards held by each party with other parties. Each party can execute the secure multi-party computing protocol to merge the final verification results based on the verification result shards calculated locally.

[0047] By merging all verification result shards, a complete verification result is obtained. The complete verification result can reflect the relationship and consistency between the private data corresponding to all data shards.

[0048] In some embodiments, the first shard based on the verification result collaborates with other shards of the verification result held by other parties in the secure multi-party computation to obtain a complete verification result, which can be achieved through the following operations.

[0049] S10, the first party sends the first fragment of the verification result to the other party.

[0050] S11, receiving other fragments of the verification result from the other party.

[0051] The other shards of the verification results of other participants can be other shards (such as LE2, LE3, etc.) that calculate the verification results based on the private data shards they hold by the same method as the first party's local calculation.

[0052] S12, based on the first fragment of the verification result and the other fragments of the verification result, using the preset second calculation formula (2) to obtain the complete verification result. The preset second calculation formula (2) is: (2) in, represents the least squares optimization value of private data E1 and private data E2, and n represents the total number of data shards.

[0053] In this embodiment, the complete verification result (usually a numerical value or an optimized data structure) calculated using the least squares method can be used to generate the final complete verification result.

[0054] In some embodiments, the first shard based on the verification result collaborates with other shards of the verification result held by other parties of the secure multi-party computation to obtain a complete verification result, which can also be achieved in the following manner.

[0055] S20, the first party sends the first fragment of the verification result to the other party; S21, receiving other fragments of the verification result from the other party; S22, based on the first fragment of the verification result and the other fragments of the verification result, using a preset third calculation formula (3) to obtain the complete verification result. The preset third calculation formula (3) is: (3) in, is the value of the preset third calculation formula, B(E,R) is a balance term for measuring the consistency between entity E and relationship R, and λ is a weight coefficient; the entity E is an entity corresponding to the privacy data E1 and the privacy data E2, and the relationship R is used to express the association relationship between the privacy data E1 and the privacy data E2.

[0056] B(E,R) is a balance term that measures the consistency between entity E and relationship R. It can be used to measure the consistency between private data E1 and E2 during data processing.

[0057] Entity E refers to an object associated with private data E1 and E2. For example, private data E1 represents Zhang San, and private data E2 represents Zhang Ergou (Zhang Ergou is Zhang San's nickname). Entity E can be an abstract summary of private data E1 and private data E2, such as a person.

[0058] The relationship R is used to describe the association between private data E1 and E2. For example, E1 may be a person’s personal information, and E2 may be the person’s transaction record. The relationship between the two can be expressed as an association relationship, such as “ownership” or “transaction”.

[0059] The weight coefficient λ is used in the calculation to adjust the impact of the B(E, R) balance term on the final result.

[0060] Step 106: Based on the complete verification result, update the nodes and / or edges of the initial knowledge graph to obtain the target knowledge graph.

[0061] Knowledge graph updating refers to modifying inconsistent or erroneous nodes and edges in the initial knowledge graph based on the complete verification results. For example, if some entity nodes or relationship edges are confirmed to be invalid in multiple verification steps, these nodes or edges can be deleted or updated.

[0062] The target knowledge graph is the updated knowledge graph. The target knowledge graph can be the final knowledge graph or a knowledge graph that has been updated at least partially.

[0063] In some embodiments, updating the nodes and / or edges of the initial knowledge graph based on the complete verification result to obtain the target knowledge graph can be achieved through the following operations.

[0064] S30: Based on the complete verification result, determine whether the difference between the private data E1 and the private data E2 exceeds a preset value.

[0065] In some embodiments, the difference between the private data E1 and E2 can be calculated based on the complete verification result. The difference can be realized based on a distance metric (such as Euclidean distance) or an error calculation method. The calculated difference is compared with a preset value. If the difference is greater than the preset value, the subsequent steps are continued; if the difference is not greater than the preset value, there is no need to further update the knowledge graph.

[0066] S31: In response to exceeding the preset value, determining the priority of the first privacy data and the second privacy data based on the contribution of each of the first privacy data and the second privacy data to constructing the knowledge graph.

[0067] In some embodiments, by analyzing the private data E1 and E2, their respective contributions in the process of constructing the knowledge graph are determined. For example, a certain private data may provide more entity information or have higher reliability in the verification process. Based on the degree of contribution, a predefined rule or algorithm can be used to calculate the priority of E1 and E2. For example, if the contribution of E1 is higher, the priority of E1 is higher, and vice versa.

[0068] According to the evaluation results, each privacy data E1 and E2 can be assigned a priority value. Data with higher priority will occupy a more important position in subsequent knowledge graph updates.

[0069] In some embodiments, the degree of contribution of each of the first privacy data and the second privacy data to the construction of the knowledge graph is determined by: determining the data proportion of the data sources corresponding to each of the first privacy data and the second privacy data in the total data used to construct the knowledge graph; and determining the degree of contribution based on the size of the data proportion.

[0070] The amount of data can be directly provided by the participants of secure multi-party computing without involving specific privacy data. According to the proportion of data sources, the standard of contribution can be defined. For example, it can be set that the contribution of data sources exceeding a certain percentage is greater, and vice versa.

[0071] S32: Update the nodes and / or edges of the initial knowledge graph based on the priority.

[0072] Based on the priorities of E1 and E2, it can be determined which nodes need to be updated. Privacy data with higher priorities will preferentially affect nodes in the graph. For example, if E1 and E2 involve the same entity, then these data may be considered as the same node in the knowledge graph, or E1 may provide more information for a node, causing the node to be expanded or updated.

[0073] Edge updating refers to updating the edge in the graph that describes the relationship between E1 and E2. Depending on the strength, difference, or consistency of the relationship between E1 and E2, the weight of the edge is adjusted or a new relationship is established. For example, if the relationship between E1 and E2 is strong in a certain field, a stronger edge connecting them can be established in the knowledge graph.

[0074] Step 107, performing data processing tasks based on the target knowledge graph.

[0075] Among them, the data processing tasks include entity classification, prediction of relationships between entities, graph rule reasoning, and entity set mining.

[0076] Entity classification refers to using the target knowledge graph to classify the entity nodes in it, such as classifying different types of entities such as companies and individuals in the graph.

[0077] Relationship prediction refers to predicting possible new relationships by analyzing the relationships between entity nodes in the target graph. For example, predicting that a person may establish a working relationship with other companies in the future.

[0078] Rule reasoning refers to the derivation of new knowledge or rules based on existing relationships on the basis of the target graph.

[0079] Entity set mining refers to mining entity sets that meet specific criteria by analyzing entities in a graph, such as finding all companies in a graph that belong to a certain category.

[0080] In some embodiments, the execution of the data processing task can be implemented based on a graph neural network. The present invention does not limit the specific type of graph neural network.

[0081] The data type of the node of the knowledge graph of the present invention includes at least one of text data, voice data and video data. Text data refers to information expressed in written form, such as articles, descriptions, tags, etc. Voice data refers to information expressed by sound or voice. Video data refers to information expressed by video. Various types of data can be obtained by extracting business data from different business fields, for example, payment fields, construction fields, industrial fields, financial fields, etc.

[0082] Working principle of the invention: This patent uses secure multi-party computing technology to effectively protect the privacy of different data providers by processing data without exposing the privacy data of each party. The nodes of the knowledge graph support the integration of multiple data types such as text data, voice data, and video data, which enhances the diversity and applicability of graph construction. At the same time, although it is based on secure multi-party computing, the efficiency of data processing can be improved by performing local cross-source verification of data locally on each participant.

[0083] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A data processing method based on secure multi-party computing, characterized in that: The method is performed by a first party participating in a secure multi-party computation, and includes the following steps: Decomposing the first private data held by the first party to obtain a plurality of data shards; wherein the plurality of data shards include a first data shard and other data shards, the first data shard is held by the first party, and the other data shards are held by other parties participating in the secure multi-party computation; Receiving a second data shard from another party participating in the secure multi-party computation; the second data shard is obtained by the other party decomposing the second private data owned by the other party; Based on the first data shard and the second data shard, collaborate with other parties participating in the secure multi-party computation to build an initial knowledge graph; Based on the first data shard and the second data shard, local cross-source data verification is performed locally on the first party to obtain a first shard of the verification result; Based on the first shard of the verification result, the other shards of the verification result held by other parties of the secure multi-party computation are coordinated to obtain a complete verification result; wherein the other shards of the verification result held by the other parties are obtained by the other parties through local calculation; Based on the complete verification result, updating the nodes and / or edges of the initial knowledge graph to obtain a target knowledge graph; Execute data processing tasks based on the target knowledge graph; wherein the data processing tasks include entity classification, prediction of relationships between entities, graph rule reasoning, and entity set mining.

2. A data processing method based on secure multi-party computing according to claim 1, characterized in that: The locally performing local cross-source data verification on the first party based on the first data shard and the second data shard to obtain a first shard of the verification result includes: Based on the first data shard and the second data shard, a first shard of the verification result is calculated by a preset first calculation formula; the preset first calculation formula is: ； Among them, L Eij represents the jth shard of the verification result calculated by the jth participant based on the data shards he holds, e 1ij represents the i-th data shard of the private data E1 in the first private data held by the j-th participant, e 2ij Represents the i-th data shard of the private data E2 in the second private data held by the j-th participant.

3. A data processing method based on secure multi-party computing according to claim 2, characterized in that: The first shard based on the verification result cooperates with other shards of the verification result held by other parties of the secure multi-party computation to obtain a complete verification result, including: The first party sends the first fragment of the verification result to the other party; receiving other shards of the verification result from the other party; Based on the first fragment of the verification result and the other fragments of the verification result, the complete verification result is obtained by using a preset second calculation formula; the preset second calculation formula is: ； in, represents the least squares optimization value of private data E1 and private data E2, and n represents the total number of data shards.

4. The data processing method based on secure multi-party computing according to claim 2, characterized in that: The first shard based on the verification result cooperates with other shards of the verification result held by other parties of the secure multi-party computation to obtain a complete verification result, including: The first party sends the first fragment of the verification result to the other party; receiving other shards of the verification result from the other party; Based on the first fragment of the verification result and the other fragments of the verification result, the complete verification result is obtained by using a preset third calculation formula; the preset third calculation formula is: ； in, is the value of the preset third calculation formula, B(E,R) is a balance term for measuring the consistency between entity E and relationship R, and λ is a weight coefficient; the entity E is an entity corresponding to the privacy data E1 and the privacy data E2, and the relationship R is used to express the association relationship between the privacy data E1 and the privacy data E2.

5. A data processing method based on secure multi-party computing according to claim 3 or 4, characterized in that: The updating of the nodes and / or edges of the initial knowledge graph based on the complete verification result to obtain the target knowledge graph includes: Based on the complete verification result, determining whether the difference between the private data E1 and the private data E2 exceeds a preset value; In response to exceeding the preset value, determining the priority of the first private data and the second private data based on the contribution of each of the first private data and the second private data to constructing the knowledge graph; The nodes and / or edges of the initial knowledge graph are updated based on the priority.

6. A data processing method based on secure multi-party computing according to claim 5, characterized in that: The contribution of the first private data and the second private data to the construction of the knowledge graph is determined in the following manner: Determine the data proportion of the data sources corresponding to the first private data and the second private data in the total data used to construct the knowledge graph; The degree of contribution is determined based on the size of the data proportion.

7. The data processing method based on secure multi-party computing according to claim 1, characterized in that: The data type of the nodes of the knowledge graph includes at least one of text data, voice data and video data.

8. A data processing system based on secure multi-party computing, characterized in that: The system is applied to a first party participating in secure multi-party computing, and the system includes: a data decomposition module, configured to decompose the first private data held by the first party to obtain a plurality of data shards; wherein the plurality of data shards include a first data shard and other data shards, the first data shard is held by the first party, and the other data shards are held by other parties participating in the secure multi-party computation; A data receiving module, configured to receive a second data shard from another party participating in the secure multi-party computation; the second data shard is obtained by the other party decomposing the second private data owned by the other party; A first collaboration module, configured to collaborate with other parties participating in the secure multi-party computation to construct an initial knowledge graph based on the first data shard and the second data shard; A data verification module, configured to perform local cross-source data verification on the first party based on the first data shard and the second data shard, to obtain a first shard of the verification result; A second collaboration module is configured to collaborate with other shards of the verification result held by other parties of the secure multi-party computation based on the first shard of the verification result to obtain a complete verification result; wherein the other shards of the verification result held by the other parties are obtained by the other parties through local calculation; A graph updating module, used to update the nodes and / or edges of the initial knowledge graph based on the complete verification result to obtain a target knowledge graph; An execution module is used to perform data processing tasks based on the target knowledge graph; wherein the data processing tasks include entity classification, prediction of relationships between entities, graph rule reasoning, and entity set mining.

Images