Artificial intelligence-based power material procurement scheme review optimization method and system

Abstract

The application discloses an artificial intelligence-based power material procurement scheme review optimization method and system, and belongs to the technical field of artificial intelligence.The method comprises the following steps: constructing a logical directed graph based on input review requirements, wherein the logical directed graph is a set of three tuples of a source point, a directed edge and a sink point; acquiring a power resource procurement scheme; performing semantic analysis on text data in the power resource procurement scheme to extract keywords; searching for directed edges corresponding to the keywords in the logical directed graph; calling a word embedding model to perform semantic analysis on the directed edges; and outputting a review conclusion.The application introduces artificial intelligence on the basis of reserving an artificial review process, avoids the influence of human subjective factors, improves review efficiency, and greatly saves labor cost.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology, and more specifically, relates to a method and system for reviewing and optimizing power material procurement schemes based on artificial intelligence. Background Technology

[0002] In the review process of power material procurement plans, manual review first requires identifying the review clauses, then gradually reading through the power material procurement documents, filtering out the corresponding content in the documents for each review clause, and finally making a decision based on the corresponding content. However, manual review requires a significant amount of manpower, and the review results are easily influenced by subjective human factors, so the review efficiency needs further improvement.

[0003] Therefore, this invention proposes an artificial intelligence-based method and system for optimizing the review of power material procurement schemes. While retaining the manual review process, artificial intelligence is introduced to improve review efficiency and save labor costs. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to resolve the aforementioned deficiencies and propose an artificial intelligence-based method and system for reviewing and optimizing power material procurement schemes.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of this invention discloses an artificial intelligence-based method for reviewing and optimizing power material procurement schemes, the method comprising:

[0007] A logically directed graph is constructed based on the input review requirements. The logically directed graph is a set of triples consisting of source nodes, directed edges, and sink nodes.

[0008] Obtain power resource procurement plans and perform semantic analysis on the text data in the power resource procurement plans to extract keywords;

[0009] The directed edges corresponding to the keywords are retrieved from the logical directed graph, and the word embedding model is invoked to perform semantic analysis on the directed edges, outputting the review conclusion.

[0010] Furthermore, the input-based review requirements construct a logically directed graph, including:

[0011] The input review requirements are obtained, and semantic analysis is performed on the text corresponding to the review requirements to extract keywords as source points and sink points;

[0012] Using the source and sink points as indexes, determine the review rules between the source and sink points based on the review requirements;

[0013] Using the review rules as the directed edges, the logical directed graph is constructed by combining the source and sink points at both ends of the directed edges.

[0014] Furthermore, the logically directed graph also includes preset rejection nodes; the construction of the logically directed graph based on input review requirements also includes:

[0015] Obtain preset veto nodes and determine veto review rules between the veto nodes based on the review requirements;

[0016] Using the veto review rule as a directed edge, and combining the veto nodes at both ends of the veto review rule, the logical directed graph is constructed.

[0017] Furthermore, the process of obtaining the power resource procurement plan and performing semantic analysis on the text data within the power resource procurement plan to extract keywords includes:

[0018] Text data is extracted from the power resource procurement plan, and semantic analysis is performed on the text data line by line or paragraph by paragraph according to the text order to extract the professional terms of each line or paragraph of the text data as the keywords.

[0019] Based on the extracted technical terms, the source and sink nodes in the logical directed graph are retrieved, and when the corresponding source and sink nodes are retrieved, the directed edges corresponding to the technical terms are determined.

[0020] Furthermore, the step of retrieving the directed edge corresponding to the keyword in the logically directed graph, and calling the word embedding model to perform semantic analysis on the directed edge, and outputting the review conclusion, includes:

[0021] Determine whether a directed edge corresponding to the technical term exists in the logical directed graph; if so, then...

[0022] The word embedding model is invoked to perform semantic analysis on the directed edges corresponding to the technical terms; otherwise, semantic analysis is performed on the next line or the next paragraph of the current text data according to the text order.

[0023] Furthermore, the review conclusion includes a first conclusion, a second conclusion, and a third conclusion; the step of retrieving the directed edge corresponding to the keyword in the logically directed graph, and calling the word embedding model to perform semantic analysis on the directed edge and output the review conclusion further includes:

[0024] When the output review conclusion is the first conclusion, the review result of the current review text is determined to be a positive result;

[0025] When the output review conclusion is the second conclusion, the review result of the current review text is determined to be a negative result;

[0026] When the output review conclusion is the third conclusion, the review result of the current review text is determined to be an uncertain result, and the current review text is sent to the review terminal for manual review.

[0027] Furthermore, the method also includes:

[0028] Based on the selection relationship between nodes in the logically directed graph, the corresponding source or sink node is split from the logically directed graph to construct a selection bipartite graph based on the split source or sink node; the edge in the selection bipartite graph represents the constraint relationship between two nodes.

[0029] The weight of each node is calculated based on the constraint degree of each node in the selected bipartite graph. The weight is directly proportional to the degree of the node, inversely proportional to the degree of other nodes in the subgraph it belongs to, and directly proportional to the degree of adjacent nodes. The subgraph refers to a set of nodes of the same category.

[0030] Each node is reviewed in order of increasing weight.

[0031] The second aspect of this invention discloses an artificial intelligence-based power material procurement scheme review and optimization system, used to implement the artificial intelligence-based power material procurement scheme review and optimization method described in the first aspect, the system comprising:

[0032] A logical directed graph construction module is used to construct a logical directed graph based on the input review requirements. The logical directed graph is a set constructed from triples of source nodes, directed edges, and sink nodes.

[0033] The semantic analysis module is used to obtain power resource procurement plans and perform semantic analysis on the text data in the power resource procurement plans to extract keywords;

[0034] The review output module is used to retrieve the directed edges corresponding to the keywords in the logical directed graph, call the word embedding model to perform semantic analysis on the directed edges, and output the review conclusion.

[0035] A third aspect of the present invention discloses a terminal, including a processor and a storage medium; characterized in that:

[0036] The storage medium is used to store instructions;

[0037] The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0038] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the steps of the method described in the first aspect.

[0039] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention has the following advantages:

[0040] (1) The present invention constructs a logical directed graph between the review rules and the review clauses according to the review clause requirements, which facilitates the retrieval of keywords and the locking of review rules in the power material procurement plan during the subsequent review process, thereby improving the review efficiency.

[0041] (2) This invention introduces a word embedding model to perform semantic analysis on power material procurement schemes. Combined with a logical directed graph, it can extract keywords in power material procurement schemes in a timely manner and perform retrieval. The final review conclusions output to a certain extent avoid the influence of human subjective factors, have a high degree of intelligence, and save manpower costs. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the method for reviewing and optimizing power material procurement schemes based on artificial intelligence provided by the present invention.

[0043] Figure 2 This is a schematic diagram of the architecture of a directed logical graph in a specific embodiment provided by the present invention. Detailed Implementation

[0044] The present application will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and should not be construed as limiting the scope of protection of the present application.

[0045] like Figure 1 As shown in one embodiment, an artificial intelligence-based method for reviewing and optimizing power material procurement schemes includes the following steps:

[0046] Step S110: Construct a logical directed graph based on the input review requirements. The logical directed graph is a set of triples consisting of source nodes, directed edges, and sink nodes.

[0047] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention includes the following steps in step S110:

[0048] Step S111: Obtain the input review requirements and perform semantic analysis on the text corresponding to the review requirements to extract keywords as source points and sink points.

[0049] Step S112: Using the source and sink as indexes, determine the review rules between the source and sink based on the review requirements.

[0050] Step S113: Using the review rules as directed edges, construct a logical directed graph by combining the source and sink points at both ends of the directed edges.

[0051] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention further includes a preset veto node in the logical directed graph, and step S110 specifically includes the following steps:

[0052] Step S114: Obtain the preset veto nodes and determine the veto review rules between the veto nodes based on the review requirements.

[0053] Step S115: Using the veto review rule as a directed edge, and combining the veto nodes at both ends of the veto review rule, construct a logical directed graph.

[0054] In a specific embodiment, the AI-based power material procurement scheme review and optimization method provided by the present invention, to better illustrate the embodiments of the present invention, firstly provides three exemplary fragments of a power material procurement scheme document:

[0055] Segment 1: This procurement is for the expansion project of a 35kV substation, involving power equipment and accessories such as distribution transformers, box-type switchgear, 33kV high-voltage disconnect switches, 10kV cable junction boxes, surge arresters and supporting insulation materials. Procurement quantity: (1) 35kV oil-immersed transformers: 2 units, rated capacity 630kVA;

[0056] (2) 33kV high-voltage disconnect switches: 4 units; (3) 10kV cable junction boxes: 6 units. Among them, the technical parameters of the 10kV cable junction boxes include: (1) Voltage level: 10kV; (2) Conductor cross-section: single-core copper cable, cross-sectional area 120mm². 2 ~240mm 2 (3) Insulation layer: cross-linked polyethylene (XLPE), temperature resistance rating 90℃.

[0057] Segment 2: The purchaser can choose the best combination from three transformer technology routes: Option A (oil-immersed): 630kVA, 10 / 0.4kV; no-load loss ≤0.8%; delivery time 90 days; price 1.2 million; Option B (dry epoxy resin): 630kVA, 10 / 0.4kV; no-load loss ≤0.9%; delivery time 60 days; price 1.3 million; Option C (low-loss silicon steel sheet): 630kVA, 10 / 0.4kV; no-load loss ≤0.7%; delivery time 120 days; price 1.4 million; Evaluation rules: first sorted by no-load loss, then the price and delivery time are combined into a weighted score (weight: loss 50% / price 30% / delivery time 20%).

[0058] Segment 3: If a 10kV cable junction box meets the salt spray test level ≥96h, then the cable head mounting component H should be made of 316L stainless steel; the grounding terminal G must not be made of ordinary galvanized material (it must be made of stainless steel).

[0059] It should be noted that the above-mentioned fragments of the power material procurement plan document can be abstractly summarized as follows: Scenario 1: Satisfying a coupling relationship. In fragment 1, the technical term "10kV cable junction box" satisfies a constrained coupling relationship; for example, its technical parameter "voltage level" is 10kV. Scenario 2: Determining a selection relationship. In fragment 2, one of the three types of transformers can be selected. Scenario 3: Multiple technical parameters that appear completely unrelated (meaning they are far apart in the power material procurement plan document) should actually satisfy certain constraints. Therefore, Scenario 2 and Scenario 3 can actually be nested or combined.

[0060] It should be further clarified that coupling relationships can also include logical coupling and temporal coupling. Logical coupling mainly includes: agent coupling, restrictive coupling, and parallel coupling. Agent coupling refers to a verb or verb phrase performing operations such as "release," "detain," or "submit" on a noun term, forming a typical VO agent relationship. For example, the professional terms "return" and "quality assurance deposit": the action of "return" releases funds from "quality assurance deposit," constituting a VO agent relationship; the professional terms "audit" and "bid evaluation report": the action of "audit" acts on the noun "bid evaluation report," reflecting a VO agent relationship. Restrictive coupling refers to a noun term limiting another noun term; parallel coupling refers to multiple noun terms co-occurring at the same semantic or business level, jointly undertaking a certain function or effect.

[0061] In this embodiment, the temporal coupling type should also include: trigger coupling and inheritance coupling. Trigger coupling refers to a noun term acting as a trigger or prerequisite in the process, determining whether subsequent actions are initiated, manifested as an NV condition / constraint relationship. For example, the technical terms "post-qualification review result" and "entering the bidding evaluation": "Post-qualification review result" is a condition for "entering the bidding evaluation," constituting an NV trigger. Similarly, the technical terms "winning bid" and "contract signing": the generation of "winning bid" immediately triggers the "contract signing" process, thus constituting an NV trigger. Similar to trigger coupling, inheritance coupling refers to two noun terms succeeding, substituting, or mapping each other within the contract lifecycle, constituting an NN continuous / derived relationship. For example, the technical terms "investment guarantee deposit" and "performance guarantee deposit": "investment guarantee deposit" is transformed into "performance guarantee deposit" after the contract is signed, with the financial guarantee function being succeeded.

[0062] In this embodiment, based on two nodes (i.e., the source node and the sink node) as indexes, and using the preset rules corresponding to the edges (i.e., directed edges) as the judgment criteria, the power material procurement plan document is reviewed line by line or segment by segment. The direction of the directed edges can be uniformly set based on the above coupling relationship. The artificial intelligence-based power material procurement plan review optimization method provided by this invention includes steps 1 to 3:

[0063] Step 1: Based on the review requirements, construct a logical directed graph.

[0064] In this context, a logically directed graph is actually a set of triples consisting of {source vertex, directed edge, sink vertex}, in the form of... Figure 2 As shown in T1 to T7, "1, 2, 3..." all represent nodes, namely source and sink nodes, and "→" represents directed edges.

[0065] For example, delivery period review: Assuming the review standard is "all equipment is delivered within 90 days after the contract is signed", the professional terms "contract signing date" and "actual delivery completion date" can be extracted from it. The corresponding coupling relationship can be analyzed, and the resulting preset rule can be used as an edge (i.e., actual delivery completion date – contract signing date ≤ 90 days).

[0066] An excerpt from the power equipment procurement plan document reads: "The contract for this project was signed on June 1, 2025; the supplier was to complete the delivery of all equipment and bring it to the site for acceptance on September 5, 2025."

[0067] Extracting source and sink points:

[0068] Node 1: Contract signing date = 2025-06-01

[0069] Node 2: Actual delivery completion date = 2025-09-05

[0070] Calculate the time difference:

[0071] Time difference = 2025-09-05 – 2025-06-01 = 96 days

[0072] Applying edge rules: Determine if 96 days ≤ 90 days?

[0073] Review conclusion: Rejected.

[0074] For example, in the review of payment terms: assuming the review standard is "Party A shall complete the payment within 30 days from the date the supplier submits a qualified invoice", the professional terms "invoice receipt date" and "actual payment date" can be extracted from it. The corresponding coupling relationship can be analyzed, and the preset rule formed can be used as an edge (i.e.: actual payment date - invoice receipt date ≤ 30 days).

[0075] The excerpt from the power supply procurement plan document reads: "The supplier submitted a qualified invoice on July 15, 2025; the contract also stipulates that 'for high-value orders with a single amount exceeding RMB 1 million, the payment period can be extended to 45 days'; the actual payment date for Party A is August 28, 2025."

[0076] Extracting source and sink points:

[0077] Node 1: Invoice receipt date = 2025-07-15

[0078] Node 2: Actual Payment Date = 2025-08-28

[0079] Calculate the time difference:

[0080] Time difference = 2025-08-28 – 2025-07-15 = 44 days

[0081] Applying edge rules: Determine if 44 days ≤ 30 days? The "30 days" rule is not met; there is an exception in the document that "high-value orders can be extended to 45 days", that is: the time difference is 44 days ≤ 45 days, which meets the exception rule for "high-value orders".

[0082] Review conclusion: Agreed.

[0083] Therefore, a logical directed graph can contain some pre-defined veto nodes. Suppose the review rules clearly state that the use of device S of model S1 is not allowed, but the power material procurement plan document only shows device S of model S1, then this is directly marked as veto.

[0084] Step S120: Obtain the power resource procurement plan and perform semantic analysis on the text data in the power resource procurement plan to extract keywords.

[0085] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention includes the following steps in step S120:

[0086] Step S121: Extract text data from the power resource procurement plan, and perform semantic analysis on the text data line by line or paragraph by paragraph according to the text order to extract the professional terms of each line or paragraph of the text data as keywords.

[0087] Step S122: Based on the extracted technical terms, the source and sink nodes in the logical directed graph are retrieved, and when the corresponding source and sink nodes are retrieved, the directed edges corresponding to the technical terms are determined.

[0088] In a specific embodiment, the AI-based power material procurement plan review and optimization method provided by the present invention, step 2, analyzes each sentence or paragraph of the power material procurement plan document, extracts the professional terms as nodes, and then searches the logical directed graph to see if there are corresponding directed edges.

[0089] Step S130: Retrieve the directed edges corresponding to the keywords in the logical directed graph, call the word embedding model to perform semantic analysis on the directed edges, and output the review conclusion.

[0090] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention includes the following steps in step S130:

[0091] Step S131: Determine whether there are directed edges corresponding to technical terms in the logical directed graph.

[0092] Step S132: Call the word embedding model to perform semantic analysis on the directed edges corresponding to the technical terms.

[0093] Specifically, when the judgment result in step S131 is that there is a directed edge corresponding to a technical term in the logical directed graph, the word embedding model is called to perform semantic analysis on the directed edge corresponding to the technical term.

[0094] Step S133: Perform semantic analysis on the next line or paragraph of the current text data according to the text order.

[0095] Specifically, when the judgment result in step S132 is that there is no directed edge corresponding to the technical term in the logical directed graph, semantic analysis is performed on the next line or the next paragraph of the current text data according to the text order.

[0096] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention includes a first conclusion, a second conclusion, and a third conclusion in its review conclusions. Step S130 further includes the following steps:

[0097] Step S134: When the output review conclusion is the first conclusion, the review result of the current review text is determined to be a positive result.

[0098] Step S135: When the output review conclusion is the second conclusion, the review result of the current review text is determined to be a negative result.

[0099] Step S136: When the output review conclusion is the third conclusion, the review result of the current review text is determined to be an uncertain result, and the current review text is sent to the review terminal for manual review.

[0100] In a specific embodiment, the AI-based power material procurement scheme review optimization method provided by this invention, in step 3, if a corresponding directed edge exists, performs analysis based on semantic analysis (word embedding model, such as word2Vec and BERT) to obtain a review conclusion. If no corresponding directed edge exists, the method then analyzes the next sentence or paragraph of the power material procurement scheme document.

[0101] In some embodiments, the AI-based power material procurement scheme review and optimization method provided by the present invention further includes the following steps:

[0102] Step S210: Based on the selection relationships between nodes in the logically directed graph, the corresponding source or sink nodes are separated from the logically directed graph, and a selection bipartite graph is constructed based on the separated source or sink nodes. The edges in the selection bipartite graph represent the constraint relationships between two nodes.

[0103] Step S220: Calculate the weight of each node based on the constraint degree of each node in the selected bipartite graph; the weight is directly proportional to the degree of the node; inversely proportional to the degree of other nodes in the subgraph to which it belongs; and directly proportional to the degree of adjacent nodes; the subgraph refers to a set of nodes of the same category.

[0104] Step S230: Prioritize the review of each node according to its weight in ascending order.

[0105] In a specific embodiment, the AI-based power material procurement scheme review optimization method provided by this invention, in the application of directed logical graphs, combines scenarios 2 and 3 above, considering that the review conclusions obtained by AI mainly include three results: agreement, rejection, and uncertainty. Taking BERT as an example, the softmax probability corresponding to its Entailment type can be used as a confidence index to judge the accuracy of BERT itself. When the confidence index is too low, the review conclusion is uncertain, and the case is transferred to manual processing.

[0106] For example, from a review perspective, assume that equipment A can be selected from A1, A2, and A3, and equipment B can be selected from B1, B2, and B3; and that the power material procurement plan document also provides A1 to A3 and B1 to B2. Furthermore, assume that equipment A1 can only be matched with equipment B1; assume that equipment A2 can only be matched with equipment B2; and assume that equipment A3 can only be matched with equipment B3.

[0107] First, given the choice relationship, since the power material procurement plan document does not simultaneously provide option B3, how can we avoid reviewing the parameter information of A3 in the power material procurement plan document? Second, based on the number and complexity of uncertainties in the review conclusions, choosing the most optimized (equivalent to the most resource-efficient) review method is crucial. Assuming all complexity is the same, and equipment A1 and equipment B1 have uncertain review conclusions with 3 and 2 parameters respectively; while equipment A2 and equipment B2 have uncertain review conclusions with 1 and 0 parameters respectively. Therefore, when submitting to manual review, the two uncertain parameters of equipment A2 should be reviewed first: if they pass the manual review, the overall result is approved; otherwise, A1 and B1 should be reviewed next.

[0108] Therefore, in a specific embodiment, based on the initial selection of a bipartite graph, new bipartite graphs can be continuously extracted to determine the review method that saves the most review resources, including steps S1 to S4.

[0109] Step S1: Based on the selection relationship, construct a selection bipartite graph from the logically directed graph; its edges represent the constraint relationship between two nodes.

[0110] For example, A1, A2, and A3 can be used as nodes in a selected bipartite graph. It should be understood that all nodes have constraints. In a selected bipartite graph, these constraints are relaxed rather than restrictive. That is, if A1 can only match B1, then there is an edge between A1 and B1; while A2 can match B1 to B3, then there are edges between A2 and B1, B2, and B3. Therefore, in step S1, the corresponding nodes refer to nodes that have not been completely rejected: for a selected bipartite graph, the corresponding nodes that have been stripped are empty; while for subsequent bipartite graphs, the corresponding nodes that have been stripped refer to nodes without constraints, i.e., nodes with a degree of 0.

[0111] It should be noted that A1, A2, and A3 cannot have any constraints, i.e., no edges exist. Therefore, they essentially constitute a set subgraph (i.e., a single set node), referring to a set of nodes of the same category. Understandably, the nodes in the set subgraph have no connection relationships; the selection bipartite graph mentioned in this invention is actually not accurate enough. More accurately, it is a multipartite graph composed of multiple set subgraphs.

[0112] Step S2: Calculate the weight of each node and the weight of each edge based on the degree of each node.

[0113] Understandably, the weight of each node must satisfy: (1) be proportional to its own degree;

[0114] (2) It is inversely proportional to the degree of other nodes in the subgraph to which it is located; (3) It is directly proportional to the degree of the nodes adjacent to it, that is, the nodes in other subgraphs to which there are edges.

[0115] In some embodiments, a random traversal can be used to directly determine the weight of a node. First, the topological sort of all subgraphs is randomly determined, and then directed edges are constructed, that is, the directions of the edges of the nodes in the bipartite graph are determined. If the topological sort is random, the direction of the edges is also random. The reason for constructing directed edges is simply to ensure that each trajectory traverses each subgraph only once. Second, multiple random traversals are performed. The number of random traversals should ideally be more than 50 times E, where E is the number of edges selected in the bipartite graph. Finally, the weight of a node is determined by the number of branches corresponding to the node during the traversal; the weight of an edge is determined by the number of traversals. Specifically, the trajectory is generated based on depth-first search. Therefore, each trajectory actually corresponds one-to-one with a binary tree consisting of n nodes; where n is the number of subgraphs. Considering symmetry, and since the topological sort is random, the number of branches corresponding to a node can be defined as the number of left (or right) children of that node. Figure 2 For example, if the binary tree is T4, then the left child of node 4 is node 1; if the binary tree is T5, then the left children of node 4 are node 2 and node 1 respectively; if the binary tree is T6, then the left children of node 4 are node 3 and node 1 respectively. Then, the weight of a node is actually equal to the sum of the number of branches corresponding to that node in all traversals. Finally, the weights of the nodes in each set subgraph need to be normalized so that their sum equals 1. The weight of an edge is actually equal to the number of times that edge is traversed.

[0116] Step S3: Based on the current review conclusion, the rejected nodes are extracted from the selected bipartite graph, a bipartite graph is constructed, and the weights of the nodes in the bipartite graph are recalculated based on the weights of the nodes and edges in the selected bipartite graph.

[0117] Understandably, a rejected node refers to a node without any edges. Therefore, the states of edges in a binary graph are actually only agreement and uncertainty. At this point, the weight w of a node in the binary graph... ′ As shown in the following formula:

[0118]

[0119] Where w is the weight of the node in the chosen bipartite graph, and e,e ′ Let represent the weight of the edge and the weight of the uncertain edge, respectively; therefore, ∑e represents the sum of the weights of all edges associated with this node in the chosen bipartite graph; ∑e ′ This represents the sum of the weights of all uncertain edges associated with this node. Uncertain edges are edges whose state is uncertain.

[0120] Step S4: Determine the review priority according to the node weight from smallest to largest.

[0121] Understandably, the smaller the weight of a node, the higher its review priority; conversely, the larger the weight, the lower its review priority.

[0122] The following describes the AI-based power material procurement scheme review and optimization system provided by the present invention. The AI-based power material procurement scheme review and optimization system described below can be referred to in correspondence with the AI-based power material procurement scheme review and optimization method described above.

[0123] In one embodiment, an AI-based power material procurement scheme review and optimization system includes a logical directed graph construction module, a semantic analysis module, and a review output module.

[0124] The logical directed graph construction module is used to construct logical directed graphs based on the input review requirements. A logical directed graph is a set of triples consisting of source nodes, directed edges, and sink nodes.

[0125] The semantic analysis module is used to obtain power resource procurement plans and perform semantic analysis on the text data in the power resource procurement plans to extract keywords.

[0126] The review output module is used to retrieve the directed edges corresponding to keywords in the logical directed graph, call the word embedding model to perform semantic analysis on the directed edges, and output the review conclusion.

[0127] In this embodiment, the logical directed graph construction module of the AI-based power material procurement scheme review and optimization system provided by the present invention is specifically used for:

[0128] Obtain the input review requirements and perform semantic analysis on the text corresponding to the review requirements to extract keywords as source points and sink points.

[0129] Using source and sink as indexes, the review rules between source and sink are determined based on review requirements.

[0130] Using the review rules as directed edges, a logical directed graph is constructed by combining the source and sink nodes at both ends of the directed edges.

[0131] In this embodiment, the AI-based power material procurement scheme review and optimization system provided by the present invention further includes a preset veto node in its directed logical graph. The directed logical graph construction module is specifically used for:

[0132] Obtain the preset veto nodes and determine the veto review rules between the veto nodes based on the review requirements.

[0133] Using the veto review rules as directed edges, and combining the veto nodes at both ends of the veto review rules, a logical directed graph is constructed.

[0134] In this embodiment, the semantic analysis module of the AI-based power material procurement scheme review and optimization system provided by the present invention is specifically used for:

[0135] Text data is extracted from power resource procurement plans, and semantic analysis is performed line by line or paragraph by paragraph according to the text order to extract the professional terms in each line or paragraph as keywords.

[0136] Based on the extracted technical terms, the source and sink nodes in the logical directed graph are retrieved, and when the corresponding source and sink nodes are retrieved, the directed edges corresponding to the technical terms are determined.

[0137] In this embodiment, the review output module of the AI-based power material procurement scheme review and optimization system provided by the present invention is specifically used for:

[0138] Determine whether there are directed edges in a logical directed graph that correspond to technical terms.

[0139] If so, the word embedding model is invoked to perform semantic analysis on the directed edges corresponding to the technical terms.

[0140] If not, semantic analysis is performed on the next line or paragraph of the current text data according to the text order.

[0141] In this embodiment, the AI-based power material procurement scheme review and optimization system provided by the present invention provides review conclusions including a first conclusion, a second conclusion, and a third conclusion. The review output module is further used for:

[0142] When the output review conclusion is the first conclusion, the review result of the current review text is determined to be a positive result.

[0143] When the output review conclusion is the second conclusion, the review result of the current review text is determined to be a negative result.

[0144] When the output review conclusion is the third conclusion, the review result of the current review text is determined to be an uncertain result, and the current review text is sent to the review terminal for manual review.

[0145] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0146] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0147] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0148] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0149] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0150] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0151] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0152] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0153] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for reviewing and optimizing power material procurement schemes based on artificial intelligence, characterized in that, The method includes: A logically directed graph is constructed based on the input review requirements. The logically directed graph is a set of triples consisting of source nodes, directed edges, and sink nodes. Obtain power resource procurement plans and perform semantic analysis on the text data in the power resource procurement plans to extract keywords; The directed edges corresponding to the keywords are retrieved from the logical directed graph, and the word embedding model is called to perform semantic analysis on the directed edges, and the review conclusion is output. Based on the selection relationship between nodes in the logically directed graph, the corresponding source or sink node is split from the logically directed graph to construct a selection bipartite graph based on the split source or sink node; the edge in the selection bipartite graph represents the constraint relationship between two nodes. The weight of each node is calculated based on the constraint degree of each node in the selected bipartite graph. The weight is directly proportional to the degree of the node, inversely proportional to the degree of other nodes in the subgraph it belongs to, and directly proportional to the degree of adjacent nodes. The subgraph refers to a set of nodes of the same category. Based on the current review conclusions, the rejected nodes are extracted from the selected bipartite graph, a bipartite graph is constructed, and the weights of the nodes in the bipartite graph are recalculated based on the weights of the nodes and edges in the selected bipartite graph. Here, rejected nodes refer to nodes without edges; the weights of the nodes in the bipartite graph are... As shown in the following formula: in, Let this node be the weight in the chosen bipartite graph. Let these represent the weights of the edges and the weights of the uncertain edges, respectively; therefore, This represents the sum of the weights of all edges associated with this node in a selected bipartite graph; This represents the sum of the weights of all uncertain edges associated with this node; uncertain edges are those whose state is uncertain. Each node is reviewed in order of increasing weight.

2. The method for reviewing and optimizing power material procurement schemes based on artificial intelligence according to claim 1, characterized in that, The input-based review requirements construct a logical directed graph, including: The input review requirements are obtained, and semantic analysis is performed on the text corresponding to the review requirements to extract keywords as source points and sink points; Using the source and sink points as indexes, determine the review rules between the source and sink points based on the review requirements; Using the review rules as the directed edges, the logical directed graph is constructed by combining the source and sink points at both ends of the directed edges.

3. The method for reviewing and optimizing power material procurement schemes based on artificial intelligence according to claim 2, characterized in that, The directed logical graph also includes preset rejection nodes; the construction of the directed logical graph based on input review requirements also includes: Obtain preset veto nodes and determine veto review rules between the veto nodes based on the review requirements; Using the veto review rule as a directed edge, and combining the veto nodes at both ends of the veto review rule, the logical directed graph is constructed.

4. The method for reviewing and optimizing power material procurement schemes based on artificial intelligence according to claim 1, characterized in that, The process of obtaining a power resource procurement plan and performing semantic analysis on the text data within the plan to extract keywords includes: Text data is extracted from the power resource procurement plan, and semantic analysis is performed on the text data line by line or paragraph by paragraph according to the text order to extract the professional terms of each line or paragraph of the text data as the keywords. Based on the extracted technical terms, the source and sink nodes in the logical directed graph are retrieved, and when the corresponding source and sink nodes are retrieved, the directed edges corresponding to the technical terms are determined.

5. The method for reviewing and optimizing power material procurement schemes based on artificial intelligence according to claim 4, characterized in that, The process involves retrieving the directed edge corresponding to the keyword from the logically directed graph, calling a word embedding model to perform semantic analysis on the directed edge, and outputting a review conclusion, including: Determine whether a directed edge corresponding to the technical term exists in the logical directed graph; if so, then... Call the word embedding model to perform semantic analysis on the directed edges corresponding to the technical terms; otherwise, Perform semantic analysis on the next line or paragraph of the current text data according to the text order.

6. The method for reviewing and optimizing power material procurement schemes based on artificial intelligence according to claim 5, characterized in that, The review conclusions include a first conclusion, a second conclusion, and a third conclusion; the step of retrieving the directed edges corresponding to the keywords in the logically directed graph, calling the word embedding model to perform semantic analysis on the directed edges, and outputting the review conclusions also includes: When the output review conclusion is the first conclusion, the review result of the current review text is determined to be a positive result; When the output review conclusion is the second conclusion, the review result of the current review text is determined to be a negative result; When the output review conclusion is the third conclusion, the review result of the current review text is determined to be an uncertain result, and the current review text is sent to the review terminal for manual review.

7. A power material procurement scheme review and optimization system based on artificial intelligence, characterized in that, The system is used to implement the AI-based power material procurement scheme review and optimization method according to any one of claims 1 to 6, the system comprising: A logical directed graph construction module is used to construct a logical directed graph based on the input review requirements. The logical directed graph is a set constructed from triples of source nodes, directed edges, and sink nodes. The semantic analysis module is used to obtain power resource procurement plans and perform semantic analysis on the text data in the power resource procurement plans to extract keywords; The review output module is used to retrieve the directed edges corresponding to the keywords in the logical directed graph, call the word embedding model to perform semantic analysis on the directed edges, and output the review conclusion.

8. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-6.

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

Images