CN-122001772-A - Distributed data privacy computing platform supporting multiparty collaboration and workflow construction method

CN122001772ACN 122001772 ACN122001772 ACN 122001772ACN-122001772-A

Abstract

The invention relates to the technical field of data processing, and particularly discloses a distributed data privacy computing platform supporting multiparty collaboration and a workflow construction method, wherein the platform comprises a workflow definition interface, a topology mapping module, a deployment compliance module, a dense state chain analysis module, an evidence return module and an arrangement audit module; and expanding a physical instance according to the metadata grouping of the transaction data set and the replication quantity parameter to form a connection relation, then checking and outputting a deployment execution plan after passing, completing identity alignment, link expansion and risk scoring under the condition of not converging the original transaction details, generating a suspicious link result and returning supervision to realize cross-domain compliance calculation, auditability and deployment.

Inventors

ZHANG ZHIWEN
LU WEN

Assignees

因正信息(上海)有限公司

Dates

Publication Date: 20260508
Application Date: 20260312

Claims (10)

1. A distributed data privacy computing platform supporting multi-party collaboration, comprising: The workflow definition interface is used for generating anti-suspicious transaction privacy calculation job description; The topology mapping module takes anti-suspicious transaction privacy calculation operation description as input, constructs a topology abstract graph and generates a topology configuration object; The deployment compliance module pre-checks the topological configuration object to form an instance deployment map, and then checks the instance deployment map to obtain a compliance check result and outputs a deployment execution plan; the secret state link analysis module adopts a link authentication cooperative model, takes a deployment execution plan as input, and outputs suspicious transaction link results; the evidence return module is used for packaging suspicious transaction link results into an evidence packet return supervision domain; The orchestration audit module is used for triggering deployment execution, orchestration operation by agents, audit solidification and resource recovery in compliance; The secret state link analysis module outputs suspicious transaction link results by adopting a link authentication collaborative model, and concretely comprises the steps of reading deployment and resource assignment information and channel enabling and communicating constraint information in a deployment execution plan to form an execution instruction list, extracting local characteristic elements for link analysis from the execution instruction list, splicing and packaging the local characteristic elements and standardized expression vectors to form an aligned input abstract, sending the aligned input abstract to a trans-organization identity aligned channel, outputting a transaction association key set, taking the transaction association key set as a link anchor point to form a candidate link set, extracting and solidifying link structure element abstracts for each candidate link, and generating suspicious transaction link results based on the link structure element abstracts by adopting an abstract driving link scoring method.
2. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein the summary-driven link scoring method is used for generating suspicious transaction link results based on link structure element summaries, and specifically comprises the steps of extracting structural feature quantities related to anti-suspicious transaction link identification, carrying out normalization and weighted summarization on the structural feature quantities according to pre-configured risk scoring indexes to obtain link risk scores, comparing the link risk scores with preset threshold conditions to determine suspicious sets, generating link identifiers for candidate links entering the suspicious sets, and outputting suspicious transaction link results containing the link identifiers and the link risk scores.
3. The distributed data privacy computing platform supporting multi-party collaboration of claim 2, wherein the structured feature quantities include link hops, recurring loop signs of the same node, degree of dispersion and pooling of funds among multiple opponents, degree of closeness of transaction time intervals, and fluctuation and split aggregate features of monetary sequences.
4. The distributed data privacy computing platform supporting multiparty collaboration of claim 2, wherein the determining the suspicious set specifically comprises marking the candidate link as suspicious and joining the suspicious set when the link risk score is not less than a preset primary threshold, marking the candidate link as a subset to be rechecked and joining the suspicious set when the link risk score is less than the preset primary threshold but not less than a preset classification threshold, while attaching a reason mark triggering the preset classification threshold, and marking the candidate link as non-suspicious and not entering the suspicious set when the link risk score is less than the preset classification threshold.
5. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein the workflow definition interface, when generating the anti-suspicious transaction privacy computing job description, comprises performing field completeness and parameter validity check on a field set and a data range, a payment mechanism identification, data set grouping information, a mechanism domain compliance boundary label, an identity alignment mode, a link expansion hop count range, a risk score index and a threshold condition, an evidence packet field set and a return target in job configuration, binding transaction data set metadata and a payment mechanism identification and forming a grouping description, solidifying the mechanism domain compliance boundary label into deployment admittance constraint, sorting the evidence packet field set and the return target into an evidence packet template and writing the evidence packet template and the return target into platform job storage.
6. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein when the topology mapping module builds a topology abstract graph, each participating payment mechanism is mapped to a payment mechanism data party role node, a supervision aggregation domain is mapped to a supervision aggregation party role node, and control and scheduling requirements are mapped to a collaborative orchestration role node, a cross-mechanism identity alignment channel edge and a dense intermediate result channel edge are established between the payment mechanism data party role node and the supervision aggregation party role node according to operation logic of 'first alignment, second extension and second return', and a suspicious link evidence return channel edge pointing from the supervision aggregation party role node to the collaborative orchestration role node or a supervision aggregation domain receiving end is established.
7. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein the topology configuration object comprises node attributes and edge attributes, wherein the node attributes comprise role type identification, affiliated institution domain identification, data consumer attributes, input data references, packet key descriptions, data set packet information, field completeness verification results, output aggregation location identification, evidence packet template references, backhaul destination identifications, control parameters, replication quantity parameters, role packet association lists, and compliance boundary labeling references, and the edge attributes comprise channel type identification, endpoint role identification, grouping rules, function mapping tags, channel backend selection parameters, message payload type constraints, channel endpoint packet connection relationships, and compliance boundary labeling references, and are stored in a unified field structure serialization write platform job.
8. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein the deployment and aggregation block performs pre-inspection by checking at least structural integrity of a role node set and a channel edge set, each role node has a data consumer identifier, a replication number parameter and grouping association information, each channel edge has an endpoint role identifier and a grouping rule, and transaction data set metadata comprises a payment mechanism identifier, field information for grouping and a field integrity check conclusion.
9. The distributed data privacy computing platform supporting multiparty collaboration according to claim 1, wherein the evidence back-pass module comprises generating a channel handle for a cross-organization identity alignment channel, a confidential intermediate result channel and an evidence back-pass channel and providing a unified channel operation interface to a physical instance according to channel types, endpoint role identifications, grouping rules, message load type constraints and channel back-end selection parameters in a topology configuration object, filling the suspicious transaction link results into an evidence packet according to a supervision report output structure, the evidence packet comprising a link identification, a link structure element abstract, a link risk score, a job identification, a time window identification, a participating organization identification set and an audit index field, generating an evidence packet back-pass instruction comprising a target supervision receiving end identification, channel selection information, a batch number or serial number and a sending acknowledgement policy, and driving the evidence packet to be back-passed to a convergence domain via the evidence back-pass channel.
10. A method for constructing a distributed data privacy workflow supporting multiparty collaboration, which is applied to the distributed data privacy computing platform supporting multiparty collaboration according to any one of claims 1-9, comprising the following steps: step 1, receiving anti-suspicious transaction privacy operation configuration, and checking the validity of fields and parameters; step 2, binding metadata of the data set and the organization identifier, fixing the rule boundary and the task parameter, and generating a job description; step 3, mapping three types of role nodes based on the job description and establishing three types of channel edges to form a topological abstract graph; Step 4, attributing and packaging the topology extraction graph and complementing the instantiation parameters to generate a topology configuration object; step 5, pre-checking the topology configuration object, expanding the examples and the connection according to the grouping and copying quantity parameters to form an example deployment diagram; And 6, after-checking that the instance deployment diagram meets the compliance boundary, outputting a deployment execution plan as a workflow product.

Description

Distributed data privacy computing platform supporting multiparty collaboration and workflow construction method Technical Field The invention relates to the technical field of data processing, in particular to a distributed data privacy computing platform supporting multiparty collaboration and a workflow construction method. Background The anti-suspicious transaction work of the cross-payment mechanism needs to carry out collaborative analysis on data scattered in different payment mechanisms so as to identify suspicious transaction links such as split transfer, circulating reflux, scattered collection, time-intensive transfer and the like of the cross-mechanism. The existing anti-suspicious transaction system is mainly used for carrying out rule matching or model detection in a single institution domain, and is used for monitoring anti-suspicious transactions around the interior of a financial institution, but is limited by inter-institution data barriers, so that a cross-institution link view is difficult to form, and the problem of compliance restriction, data security risk and operation cost is faced when the original transaction details are concentrated in a single center. The privacy calculation provides a technical path for cross-domain cooperation, but has the following difficulties in engineering landing that 1, multiparty data and calculation forces belong to different main bodies, a large amount of manual coordination is needed for deployment and operation and maintenance, the operation on-line period is long and errors are prone to occur, 2, complex cooperation topology frequently evolves between different mechanism domains and supervision domains, the existing system often lacks unified abstraction on a role-channel-instance layer, so that topology update needs to be improved in an intrusion mode, 3, the cross-domain communication link and compliance boundary are obviously different, a channel-level rear end configurable and audit traceable mechanism is lacking, and evidence materials meeting supervision requirements are difficult to stably output on the premise that the domain cannot be ensured. Therefore, a private computing platform and workflow construction scheme special for identifying anti-suspicious transaction links of a cross-payment mechanism are needed, so that the anti-suspicious transaction operations can be described in a unified topology abstraction and deployed automatically, and cross-mechanism identity alignment, secret state link expansion, risk scoring and evidence package generation are completed under the constraint of a compliance boundary, so that cross-mechanism anti-suspicious transaction collaborative analysis is realized. Disclosure of Invention In order to overcome the defects of the prior art, the invention provides a distributed data privacy computing platform supporting multiparty collaboration and a workflow construction method, which are characterized in that a topology configuration object is generated by taking inverse suspicious transaction privacy computing operation description as input, an example deployment diagram is formed by pre-checking the topology configuration object, a deployment execution plan is output by the post-checking example deployment diagram, a chain authentication collaboration model is adopted, a suspicious transaction link result is output by taking the deployment execution plan as input, and the suspicious transaction link result is packaged into an evidence package return supervision domain, so that the problems in the background technology are solved. In order to achieve the above purpose, the present invention provides the following technical solutions: A distributed data privacy computing platform supporting multi-party collaboration for identifying analysis scenarios across anti-suspicious transaction links of a payment authority, comprising: and the workflow definition interface is used for generating an anti-suspicious transaction privacy calculation job description. And the topology mapping module takes the anti-suspicious transaction privacy calculation job description as input, constructs a topology abstract graph and generates a topology configuration object. And the deployment compliance module pre-checks the topological configuration object to form an instance deployment map, and then checks the instance deployment map to obtain a compliance check result and outputs a deployment execution plan. And the dense state link analysis module adopts a link authentication cooperative model, takes the deployment execution plan as input, and outputs suspicious transaction link results. And the evidence return module is used for packaging suspicious transaction link results into an evidence packet return supervision domain. And the orchestration audit module is used for triggering deployment execution, proxy orchestration operation, audit solidification and resource recovery by compliance. The workflow definition interface is us