Search

CN-122023018-A - Investment decision data processing system for rule conflict detection

CN122023018ACN 122023018 ACN122023018 ACN 122023018ACN-122023018-A

Abstract

The invention discloses an investment decision data processing system for rule conflict detection, and relates to the technical field of computer-aided decision making. The system utilizes space cross comparison to identify attribute exclusion and logic deadlock among rules by converting investment rules into geometric closure areas in a multidimensional parameter space and traversing based on rule dependency graphs. When collision occurs, the system quantifies the collision dimension deviation, converts the rule unstructured attribute into a feature tensor and performs matrix weighted calculation to obtain a comprehensive decision priority score, and then overwrites the memory evaluation logic according to the winning rule to relieve the bottom computing deadlock. The invention thoroughly eliminates the system conflict caused by the interweaving of complex heterogeneous regulations and realizes the machine automatic arbitration and tamper-proof tracing of the whole evaluation flow.

Inventors

  • PAN YU
  • WANG CHUNBIN

Assignees

  • 浙江省发展规划研究院
  • 浙江工商大学

Dates

Publication Date
20260512
Application Date
20260414

Claims (10)

  1. 1. An investment decision data processing system for rule conflict detection, comprising: the rule mapping and constructing module is used for acquiring construction project investment decision rules, mapping judgment conditions in the rules into geometric closure areas in the multidimensional parameter space, and constructing rule dependency graphs based on logic association among the rules; The matching and collision detection module is used for receiving input data of the item to be evaluated, traversing and extracting an activated rule set based on the rule dependency graph, and executing space cross comparison on the geometric closure area corresponding to each rule in the rule set so as to identify conflict rules with space overlapping and attribute exclusion and corresponding conflict dimensions; The quantization and priority arbitration module is used for quantitatively calculating parameter deviation on the conflict dimension when the conflict rule is identified, converting unstructured attributes of the conflict rule into characteristic tensors, performing weighted calculation on the characteristic tensors by using a preset global weight vector to obtain priority scores, and determining winning rules according to the ordering of the priority scores; and the evaluation and output module is used for overwriting evaluation logic of the system according to the constraint parameters of the winning rule to remove the calculation deadlock, executing final project evaluation to generate an evaluation result message, and encrypting and solidifying a data link log generated in the evaluation process.
  2. 2. The investment decision data processing system of claim 1, wherein said mapping of the decision conditions in the rule into geometric closure regions in a multidimensional parameter space comprises: Establishing a multi-dimensional continuous parameter space data structure in a memory, extracting quantization judgment conditions in the rule, and mapping the quantization judgment conditions into corresponding feature coordinate axis dimensions in a multi-dimensional space; And analyzing a lower limit threshold and an upper limit threshold defined by the rule on each feature coordinate axis dimension, and generating a hyper-rectangular closure area surrounding an effective acting area of the rule in a multidimensional space by taking the lower limit threshold and the upper limit threshold as boundaries.
  3. 3. The investment decision data processing system of claim 2, wherein the mathematical resolution of the hyper-rectangular closure zone is characterized by: ; Wherein, the Represent the first The regular corresponding hyper-rectangular closure area, Representation of A continuous parameter space is maintained, As a total number of feature dimensions, Index dimension and value range To the point of , As a coordinate point in the parameter space, Is a coordinate point In the first place The values in the dimensions of the individual features, Is a rule In the first place A lower threshold specified in the dimension of each feature, Is a rule In the first place An upper threshold specified in the feature dimension.
  4. 4. A system for processing investment decision data for rule conflict detection according to claim 3, wherein said constructing a rule dependency graph based on logical associations between rules and performing a spatial cross-comparison of said geometric closure regions corresponding to each rule in said rule set comprises: Maintaining a directed acyclic graph containing a node set and an edge set in a system main memory as a rule dependency graph, calling a traversal algorithm to traverse the directed acyclic graph, and pushing a rule instruction meeting a precondition into a memory stack of an active set to be detected; the exclusive dispatching thread reads any activated first rule and second rule, and judges whether the corresponding hyper-rectangular closure area has volume overlapping and attribute constraint rejection in the multidimensional space; If the lower parameter limit of the first rule is larger than the upper parameter limit of the second rule on a certain feature coordinate axis dimension, judging that the geometric intersection is empty and generating a physical constraint deadlock, or outputting a collision interrupt signal when the logic gate array is used for performing exclusive or comparison to judge that the logic attributes are left.
  5. 5. The investment decision data processing system of claim 4, wherein said quantitatively calculating parameter bias in said conflicting dimensions comprises: a fixed point algorithm is executed to stack the data blocks corresponding to each parameter dimension in sequence, and the specific conflict dimension memory address with the intersection set being empty or left of the logic phase is locked; Extracting a constraint extraction threshold value of a collision rule triggering the collision interrupt signal in the collision dimension, and an actual calculation value of an input feature of an item to be evaluated in the collision dimension; and calling an arithmetic logic unit to calculate an absolute value deviation between the actual calculated value and the constraint extraction threshold, and calculating a relative deviation rate to generate a quantized deviation data packet.
  6. 6. The investment decision data processing system for rule conflict detection of claim 5 wherein the absolute value deviation and relative deviation rate is calculated as: ; ; Wherein, the Representing the absolute value deviation of the value, The relative rate of deviation is indicated as such, Inputting the actual dead reckoning of the feature in the conflict dimension for the item to be evaluated, Extracting a threshold for a constraint of the conflict rule in the conflict dimension, An absolute value of a threshold is extracted for the constraint.
  7. 7. The investment decision data processing system for rule conflict detection according to claim 1, wherein said converting unstructured properties of said conflict rule into feature tensors, performing a weighted calculation on said feature tensors using a preset global weight vector to obtain a priority score, in particular comprising: Accessing a metadata base of a rule main body with conflict, and converting the attribute of the metadata base into floating point scalar parameters including source hierarchy, release timeliness, professional field matching degree and forcing degree to form a feature tensor; The method comprises the steps of converting a difference value of a rule release time from a current system time into a scalar parameter of release timeliness through a time attenuation model, and calculating matching degree of a rule application field and an item actual field through a space vector distance similarity to serve as the scalar parameter of the professional field matching degree; And distributing dynamic weights for the source hierarchy, release timeliness, professional field matching degree and forced degree by adopting an analytic hierarchy process in combination with an information entropy computing mechanism to generate a global weight vector, scheduling a matrix multiplier of a central processing unit, and executing weighted computation of a characteristic tensor and the global weight vector to obtain a comprehensive arbitration priority score.
  8. 8. The investment decision data processing system of claim 7, wherein the specific mathematical calculation model of publication timeliness, professional domain matching, and comprehensive arbitration priority score is: the release timeliness And calculating by adopting an exponential time decay model, wherein the formula is as follows: ; Wherein, the Is a natural constant which is used for the production of the high-temperature-resistant ceramic material, For a time decay constant coefficient dynamically derived based on historical rule validity, The difference in time stamp from the current system physical time is issued for the rule, Scalar parameters of the release timeliness which are output after conversion; The degree of matching of the professional field And calculating by using a cosine similarity instruction, wherein the formula is as follows: ; Wherein, the Semantic feature vectors for the applicable fields set for the rules, For the feature vector of the actual field of the item to be evaluated, Representing the dot product of the two vectors, And (3) with Respectively representing the two-norm lengths of the two vectors; The calculation formula of the comprehensive decision priority score is as follows: ; Wherein, the Representation rules Is a comprehensive arbitration priority score of (1), , , , Dynamic global weight constants respectively calculated for source hierarchy, release timeliness, professional domain matching degree and mandatory degree and meeting And is also provided with , , , , Respectively express rules The source level, the release timeliness, the professional field matching degree and the forced degree of the floating point scalar parameter value corresponding to the converted floating point scalar parameter value.
  9. 9. The investment decision data processing system of rule conflict detection of claim 8, wherein said evaluating logic of the override system based on constraint parameters of the winning rule to relieve computing deadlocks, in particular comprises: The conflict rules are ordered in a descending order according to the numerical value of the priority score, the rule with the forefront ranking is extracted to be used as the winning rule, a write operation instruction is generated, and the constraint threshold corresponding to the winning rule is forcedly written into a working area of the evaluation main memory; And if the actual difference between the highest priority score and the next highest priority score is lower than the safety threshold dynamically deduced by the system, generating a hardware level abnormal interrupt signal to trigger manual review intervention.
  10. 10. The investment decision data processing system for rule conflict detection according to claim 1, wherein said encrypting and solidifying the data link log generated by the evaluation process comprises: Extracting node memory snapshot, original feature vector sequence set, decision tree execution topology path stack structure and weight feature score multiplication matrix calculation source data which are collided by utilizing a bypass sniffer recorder embedded in the system to form a merging log object; And the encryption chip adopts a secure hash protocol algorithm and a time stamp anti-counterfeiting signature to operate the combined log object to generate a digest, and the digest is solidified and written into a lasting audit hardware block with an anti-erasure attribute.

Description

Investment decision data processing system for rule conflict detection Technical Field The invention relates to the technical field of computer-aided decision making, in particular to an investment decision data processing system for rule conflict detection. Background Along with transformation upgrading of economic structures and refined development of investment management, the compliance supervision system facing investment decisions of construction projects presents complex characteristics of huge data volume, multiple dimensions, level nesting and the like. How to efficiently and accurately perform full-process automated compliance assessment of fixed asset investment projects has become an important research direction in the technical field of computer-aided decision making. Currently, existing computer-aided investment decision-making evaluation systems rely primarily on traditional rule engines (e.g., based on forward inference engines or state machine engines) or employ simple decision trees for pattern matching with graph databases. Such systems typically convert business rules into simple "If-Else" code logic or one-dimensional decision tables, which have a certain effect in handling flattened, single-dimensional compliance checks. However, the prior art has significant drawbacks in underlying data processing and hardware operating mechanisms in the face of a complex heterogeneous rule hierarchy crossing hierarchical, multiple departments. First, the prior art lacks the underlying multidimensional spatial algorithm support for multi-source rule automatic collision detection. When constraint rules in different dimensions generate topological intersection and contradiction in a multidimensional feature space, the system cannot scan the overlapping degree of the geometric range before compiling and executing, so that an arithmetic unit cannot output a uniquely determined Boolean value, and logic deadlock is extremely easy to fall into or a self-contradiction evaluation result is easily output. Secondly, the existing conflict positioning granularity is extremely coarse, the system can only give error reporting at the document level, and the rule cannot be disassembled into specific features in a multi-dimensional coordinate system, so that the coordinate fixed point identification at the parameter level and the machine level value deviation quantification cannot be realized. And when the rule conflict is processed, the existing system excessively depends on the hard coding absolute priority of the source code layer or the manual subjective intervention interruption, and can not convert unstructured attributes such as timeliness, hierarchy, forcing degree and the like of the rule into a feature matrix to carry out weighting operation, so that a mathematical automation arbitration model for multidimensional dynamic quantification is lacking. Finally, rule dependency topological relation and data flow in the existing evaluation flow are opaque, clear visual dependency path memory pointer chains cannot be extracted, and tamper-proof audit log closed loops containing hardware clocks and encryption features are absent. At present, a good technology is lacking, and the problems can be effectively solved. The information disclosed in this background section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person of ordinary skill in the art. Disclosure of Invention The invention aims to overcome the defects and provide an investment decision data processing system for rule conflict detection, which thoroughly solves the problem of calculation conflict caused by multi-source heterogeneous rule interweaving through the introduction of software and hardware cooperation and a bottom mathematical space algorithm. The technical scheme of the invention is as follows: the invention provides an investment decision data processing system for rule conflict detection, which comprises a rule mapping and constructing module, a matching and collision detection module, an evaluation and output module and a processing module, wherein the rule mapping and constructing module is used for acquiring construction project investment decision rules, mapping judging conditions in the rules into geometric closure areas in a multidimensional parameter space, constructing rule dependency graphs based on logic association among rules, receiving item input data to be evaluated, traversing and extracting an activated rule set based on the rule dependency graphs, executing space cross comparison on the geometric closure areas corresponding to each rule in the rule set so as to identify conflict rules with space overlapping and attribute exclusion and corresponding conflict dimensions, the quantization and priority judging module is used for quantizing and calculating paramet