Data processing method and product

Abstract

The invention discloses a data processing method and a product. Comprising the steps of obtaining a to-be-processed text of a to-be-determined energy storage term, performing division to obtain at least one text block, and determining a first adaptation attribute between each text block and a preset energy storage term; for any text block, determining at least one to-be-screened energy storage clause corresponding to the text block based on the text block and a pre-constructed energy storage clause knowledge structure graph, and determining a second adaptive attribute according to the embedding vector of the text block and the clause embedding vector of the at least one to-be-screened energy storage clause; determining a third adaptation attribute of at least one to-be-displayed energy storage term associated with the text block according to the first adaptation attribute and the second adaptation attribute of the text block under the same to-be-displayed energy storage term; for the same to-be-displayed energy storage clause, weighting all the associated third adaptation attributes to obtain a target adaptation attribute relative to the to-be-processed text; and displaying the third adaptation attribute and the target adaptation attribute. And the accuracy of determining the energy storage terms is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data processing method and product. Background Technology

[0002] In the energy storage field, energy storage safety regulations span various standards, including battery safety (GB / T36276), fire protection (IEC62933-5-2), UL9540A thermal testing, and grid connection (Q / GDW11287). While manual comparison of clauses can be used to determine a project's safety regulatory requirements, this method is time-consuming and prone to overlooking overlapping regulations. Alternatively, keyword searches can be used to find relevant safety regulations, but this cannot address challenges such as long-distance dependencies, different version numbers, and semantic conflicts during regulation matching, making it difficult to obtain complete and accurate energy storage safety regulations and resulting in poor accuracy in determining the appropriate regulations. Summary of the Invention

[0003] This invention provides a data processing method and product to solve the problem of poor accuracy in determining the energy storage terms of the text to be processed in an energy storage system.

[0004] According to one aspect of the present invention, a data processing method is provided, comprising:

[0005] Obtain the text to be processed for the energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text;

[0006] The text to be processed is divided into at least one text block according to a preset rule, and at least one text block and a preset energy storage clause are input into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage clause.

[0007] For at least one text block, based on the text block and the pre-built knowledge structure graph of energy storage terms, at least one energy storage term to be screened corresponding to the text block is determined, and based on the embedding vector of the text block and the term embedding vector of at least one energy storage term to be screened, a second adaptation attribute between the text block and at least one energy storage term to be screened is determined.

[0008] For at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, the third adaptation attribute of at least one energy storage terms to be displayed associated with the text block is determined; wherein, at least one energy storage terms to be displayed is the union of preset energy storage terms and at least one energy storage terms to be screened.

[0009] For the same energy storage terms to be displayed, all third-party adaptation attributes associated with the energy storage terms to be displayed are weighted and processed to obtain the target adaptation attributes of the energy storage terms to be displayed relative to the text to be processed.

[0010] The third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and the target adaptation attribute of the energy storage terms to be displayed is displayed at the target position in the text to be processed.

[0011] Optionally, the knowledge structure graph of energy storage terms, including all preset energy storage terms, is determined as follows: Based on any two preset energy storage terms and pre-set logical constraints, it is determined whether the two preset energy storage terms are available; wherein, the logical constraints include at least the content inclusion constraint, content equivalence constraint, and content contradiction constraint; if the two preset energy storage terms are available, the two preset energy storage terms are respectively taken as root nodes, and the root attribute of the root node is determined according to the term attribute information of the energy storage terms; for the next-level child node associated with the root node, the first sub-attribute of the child node is determined according to the preset energy storage terms corresponding to the root node, wherein the first sub-attribute is related to the entity information associated with the preset energy storage terms; for the next-level leaf node associated with the child node, the term logic information is determined according to the preset energy storage terms of the root node associated with the leaf node, and the term logic information is taken as the node attribute of the leaf node.

[0012] Optionally, after determining the root node, child nodes, and leaf nodes corresponding to all available preset energy storage terms, the method further includes: obtaining the first sub-attribute of all child nodes, and removing redundancy from child nodes with the same first sub-attribute to obtain the updated energy storage term structure knowledge graph.

[0013] Optionally, after determining the available preset energy storage terms, the method further includes: inputting the preset energy storage terms into a pre-trained language model to extract entity information and logical word information from the preset energy storage terms based on the entity extraction layer in the language model; wherein, the logical word information is used to determine the logical information of the terms; and performing vector transformation processing on the entity information based on the terms vectorization layer in the language model to obtain a terms embedding vector that is compatible with the preset energy storage terms.

[0014] Optionally, after obtaining the clause embedding vector, the method further includes associating the clause identifier corresponding to the preset energy storage clause, the clause embedding vector of the preset energy storage clause, and the additional information of the preset energy storage clause, and caching them in the vector library.

[0015] Optionally, at least one text block and preset energy storage terms are input into a pre-trained target model to determine a first adaptation attribute between each text block and the preset energy storage terms, including: for at least one text block, inputting the text block and all preset energy storage terms into the target model so that the target model outputs the first adaptation attribute between the text block and each preset energy storage term.

[0016] Optionally, based on the text block and a pre-built knowledge structure graph of energy storage terms, at least one energy storage term to be screened corresponding to the text block is determined, and a second adaptation attribute between the text block and at least one energy storage term to be screened is determined according to the embedding vector of the text block and the term embedding vector of at least one energy storage term to be screened. This includes: determining the associated entities and logical relationships corresponding to the text block, and querying the knowledge structure graph of energy storage terms according to the associated entities and logical relationships to obtain at least one energy storage term to be screened associated with the text block; searching for the term embedding vector that is adapted to at least one energy storage term to be screened from the vector library according to the term identifier and / or additional information of at least one energy storage term to be screened; calculating the similarity between the term embedding vector of each energy storage term to be screened and the embedding vector of the text block to obtain the second adaptation attribute of the text block relative to each energy storage term to be screened.

[0017] Optionally, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, a third adaptation attribute of at least one energy storage terms associated with the text block is determined, including: obtaining the third adaptation attribute of the text block relative to the same energy storage terms by weighting the first and second adaptation attributes corresponding to the text block under the same energy storage terms.

[0018] Optionally, a third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and a target adaptation attribute of the energy storage terms to be displayed is displayed at a target position in the text to be processed. This includes: displaying a clause jump icon and a corresponding third adaptation attribute for at least some of the associated energy storage terms to be displayed at a preset position in each text block of the text to be processed; displaying clause jump icons and target adaptation attributes for all or at least some of the energy storage terms to be displayed at the target position in the text to be processed; wherein the clause jump icon is clickable to display the clause content of the energy storage terms to be displayed; at least some of the energy storage terms to be displayed are energy storage terms to be displayed whose target adaptation attributes satisfy a first preset condition, and / or, energy storage terms to be displayed whose third adaptation attributes satisfy a second preset condition.

[0019] According to another aspect of the present invention, a data processing apparatus is provided, comprising:

[0020] The pending text acquisition module is used to acquire the pending text of the energy storage terms to be determined, wherein the pending text includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text.

[0021] The first adaptation attribute determination module is used to divide the text to be processed into at least one text block according to a preset rule, and input at least one text block and preset energy storage terms into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms.

[0022] The second adaptation attribute determination module is used to determine, for at least one text block, at least one energy storage clause to be screened corresponding to the text block based on the text block and a pre-built knowledge structure graph of energy storage clauses, and to determine the second adaptation attribute between the text block and at least one energy storage clause to be screened based on the embedding vector of the text block and the clause embedding vector of at least one energy storage clause to be screened.

[0023] The third adaptation attribute determination module is used to determine, for at least one text block, the third adaptation attribute of at least one energy storage clause associated with the text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage clause to be displayed; wherein, at least one energy storage clause to be displayed is the union of preset energy storage clauses and at least one energy storage clause to be filtered.

[0024] The target adaptation attribute determination module is used to weight all third adaptation attributes associated with the same energy storage terms to be displayed, and obtain the target adaptation attribute of the energy storage terms to be displayed relative to the text to be processed.

[0025] The adaptation attribute display module is used to display the third adaptation attribute at a preset position in each text block of the text to be processed, and to display the target adaptation attribute of the energy storage terms to be displayed at the target position in the text to be processed.

[0026] According to another aspect of the present invention, embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the data processing method of any embodiment of the present invention.

[0027] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0028] At least one processor; and

[0029] A memory that is communicatively connected to at least one processor; wherein,

[0030] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the data processing method of any embodiment of the present invention.

[0031] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute and implement the data processing method of any embodiment of the present invention.

[0032] The technical solution of this invention involves obtaining the text to be processed for energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text input by the user in the front-end interactive interface and / or the imported project text; dividing the text to be processed into at least one text block according to preset rules, and inputting at least one text block and preset energy storage terms into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms; for at least one text block, determining at least one energy storage term to be screened corresponding to the text block based on the text block and a pre-built knowledge structure graph of energy storage terms, and determining the relationship between the text block and at least one preset energy storage term based on the embedding vector of the text block and the term embedding vector of the at least one preset energy storage term. The second adaptation attribute between the energy storage terms to be screened; for at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, the third adaptation attribute of at least one energy storage terms to be displayed associated with the text block is determined; wherein, at least one energy storage terms to be displayed is the union of the preset energy storage terms and at least one energy storage terms to be screened; for the same energy storage terms to be displayed, all the third adaptation attributes associated with the energy storage terms to be displayed are weighted to obtain the target adaptation attribute of the energy storage terms to be displayed relative to the text to be processed; the third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and the target adaptation attribute of the energy storage terms to be displayed is displayed at the target position in the text to be processed. This solution facilitates the collection of user-inputted or imported text from the front end, splits the text into blocks using preset rules, and rapidly generates the first matching attribute using a pre-trained target model. It then accurately filters the energy storage clauses to be processed based on a knowledge structure graph of energy storage clauses, calculates the second matching attribute using vector similarity, and performs two rounds of weighted fusion to obtain the third matching attribute and the target matching attribute. Finally, the results are presented visually. The entire process requires no manual intervention. It improves the efficiency and accuracy of clause matching by leveraging models and graphs, ensures the comprehensiveness of the matching range through the union of preset and filtered clauses, and considers the matching needs of text blocks and the overall text through hierarchical matching attribute calculation. The visual presentation makes the matching relationship intuitive and easy to understand, significantly reducing decision-making costs. Ultimately, it achieves high efficiency, accuracy, and convenience in determining energy storage clauses, solving the problem of poor accuracy in determining energy storage clauses in text to be processed in energy storage systems.

[0033] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart of a data processing method provided in Embodiment 1 of the present invention;

[0036] Figure 2 This is a flowchart of a data processing method provided in Embodiment 2 of the present invention;

[0037] Figure 3 This is a schematic diagram of the structure of a data processing device provided in Embodiment 3 of the present invention;

[0038] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the data processing method of the present invention. Detailed Implementation

[0039] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0040] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0041] Example 1

[0042] Figure 1 This is a flowchart of a data processing method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where data processing is performed to determine the energy storage terms corresponding to the text to be processed. This method can be executed by a data processing device, which can be implemented in hardware and / or software, and can be configured in electronic devices such as computers and servers. Figure 1 As shown, the method includes:

[0043] S110. Obtain the text to be processed for the energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text.

[0044] Specifically, the text to be processed can be understood as the text for which the appropriate energy storage terms need to be determined. It can be descriptive text entered by the user through the front-end interactive interface and / or imported project text. It is the basic data for extracting key information and deriving reasonable terms.

[0045] Specifically, a front-end interactive interface that supports text input and / or file import can be pre-built. Users can directly input text describing energy storage requirements manually, or upload project text through a preset file upload interface to obtain data to be processed.

[0046] In this embodiment, the front-end interaction design takes into account both direct input and file import methods, adapting to users' needs for quick and easy data entry and batch import of detailed project documents, simplifying the user operation process, lowering the threshold for text submission, and eliminating the need for users to additionally organize information formats, which helps to improve the overall efficiency of energy storage terms determination.

[0047] S120. Divide the text to be processed into at least one text block according to a preset rule, and input the at least one text block and the preset energy storage terms into the pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms.

[0048] Specifically, preset rules can be understood as standardized criteria pre-defined to achieve structured splitting of the text to be processed. Their core function is to ensure the consistency and relevance of text splitting, providing well-organized text blocks for subsequent adaptation calculations. Preset rules can include rules based on semantic logic, content length, format identifiers, etc. For example, semantic logic can include main components such as requirement descriptions and technical parameters; content length can be a preset fixed character threshold; and format identifiers can include paragraph separators, punctuation marks, list tags, and domain keywords. The pre-defined energy storage clauses can be understood as a set of standard clauses in the energy storage field that have been pre-organized and incorporated into the system. These clauses cover core scenarios such as cooperation fulfillment, technical specifications, and safety requirements, serving as the basic reference for text block adaptation and matching. This avoids omitting key general clauses during the matching process. The pre-defined energy storage clauses include, but are not limited to, clauses extracted from texts such as battery safety GB / T36276, fire protection IEC62933-5-2, UL9540A thermal testing, and grid access Q / GDW11287. The extraction method employed involves using pre-defined energy storage clause prompts and a pre-trained large language model to extract and process energy storage regulations. The target model can be understood as a large language model trained on specific data for energy storage clause matching scenarios. It possesses the ability to identify the semantic relationship between text blocks and clauses, and can quickly quantify the degree of fit between them, providing technical support for the batch and efficient generation of the first adaptation attribute. The first adaptation attribute can be understood as a quantitative representation of the degree of fit between the text block and the pre-defined energy storage clauses, which can be calculated and generated by the specially trained target model.

[0049] Specifically, the text to be processed is structurally split according to preset rules, forming at least one independent text block focused on core information. Then, all the split text blocks, along with preset energy storage clauses, are input into a specially trained target model. The model, through learned clause matching logic, determines the degree of fit between each text block and each preset energy storage clause, ultimately outputting a quantified matching result. This result represents the first fit attribute of each text block corresponding to each preset energy storage clause. For example, the first fit attribute may include the fit probability between the text block and the clause. Optionally, to better determine the fit degree between each text block and each preset energy storage clause through the target model, key information can be extracted from the text blocks first. Then, corresponding prompt words can be generated using the prompt word template needed to determine the fit degree, and these prompt words can be input into the target model to guide it in quickly determining the first fit attribute of each text block corresponding to each preset energy storage clause.

[0050] For example, for the splitting of the text to be processed, a dual strategy of semantic-driven segmentation as the main approach and length threshold segmentation as a supplementary approach can be adopted. For structured text to be processed, the semantic-driven segmentation strategy can be used to segment the text based on segment structure, punctuation marks, list markers, and domain key phrases. When the text to be processed has a poor structure or is noisy due to OCR recognition technology, the system introduces line number thresholds, character number thresholds, or token number thresholds for fallback segmentation. For example, every 3-5 lines of text can be divided into a block; every 300-500 characters can be divided into a block; and every 256 tokens can be used as a segmentation window.

[0051] For example, the determination of the first adaptation attribute can be obtained by a large language model based on joint reasoning between structured clause information and the text to be processed. Taking project text as the text to be processed as an example, the project text is split into multiple project text blocks, and these project text blocks are then processed. With pre-set energy storage terms The input to the large language model is combined to construct a decision task, such as: "Please determine whether the following item descriptions need to meet the following regulatory requirements, and output a probability value between 0 and 1." The large language model then processes this pair of tuples. Perform semantic understanding and logical matching, comprehensively considering applicable conditions (such as indoor installation, battery type, system capacity, etc.), behavior type (must / must not / may), and constraint objects, to give the probability of hit. To output the hit probability, the model's logits can be used via the Sigmoid function. Probabilistic processing is performed, where, The representation model calculates an original score based on the degree of fit between the text block and the energy storage clause, which is based on the input features.

[0052] In this embodiment, splitting text blocks according to preset rules makes complex text to be processed more targeted, avoids semantic confusion caused by overall text matching, and improves the accuracy of matching attribute calculation. By using a pre-trained target model to batch process the matching of text blocks with preset clauses, the generation efficiency of the first matching attribute is greatly improved, and the cost of manual intervention is reduced. At the same time, the model, based on the matching logic trained specifically, can stably output standardized quantitative results, laying a reliable foundation for the subsequent fusion of multi-dimensional matching attributes and helping to promote the efficient advancement of the energy storage clause matching process.

[0053] Optionally, at least one text block and preset energy storage terms are input into a pre-trained target model to determine a first adaptation attribute between each text block and the preset energy storage terms, including: for at least one text block, inputting the text block and all preset energy storage terms into the target model so that the target model outputs the first adaptation attribute between the text block and each preset energy storage term.

[0054] Specifically, at least one text block after splitting is extracted, and the complete set of preset energy storage terms is retrieved. A single text block and all preset energy storage terms are used as a set of input data and sequentially fed into the target model pre-trained with data from the energy storage field. The model calculates the matching degree between the two by parsing the core information of the text block and the semantic connotation of each preset energy storage term, and finally outputs the first adaptation attribute corresponding to the text block and each preset energy storage term. The first adaptation attribute may include quantified probability data. The adaptation attribute calculation of all text blocks and all preset energy storage terms is completed according to the same logic.

[0055] In this embodiment, a comprehensive comparison of the text block with all preset energy storage terms is achieved to avoid missing potential matching terms and ensure the integrity of the matching results. Furthermore, by using a model to batch process the matching calculations of a single text block and multiple terms, the matching efficiency is greatly improved, reducing the time spent on manual matching one by one. The obtained matching attributes can be accurately mapped to each term, providing refined data support for subsequent term selection and priority ranking, and helping to quickly optimize and determine energy storage terms.

[0056] S130. For at least one text block, based on the text block and the pre-built knowledge structure graph of energy storage terms, determine at least one energy storage term to be screened corresponding to the text block, and determine the second adaptation attribute between the text block and at least one energy storage term to be screened according to the embedding vector of the text block and the term embedding vector of at least one energy storage term to be screened.

[0057] The pre-constructed knowledge structure graph of energy storage clauses can be understood as a structured knowledge carrier integrating the core information of clauses in the energy storage field. Specifically, it consists of a three-layer knowledge graph constructed from clause-level information, concept-level information, and requirement-level information. The energy storage clauses to be screened can be understood as a subset of clauses highly relevant to the text block information, selected from a massive set of pre-defined clauses based on semantic retrieval and associative reasoning using the knowledge structure graph of energy storage clauses. This subset is used for subsequent precise adaptation. The second adaptation attribute is a quantitative result obtained by calculating the similarity between the embedding vector of the text block and the embedding vector of the energy storage clauses to be screened. Its core purpose is to accurately characterize the semantic fit between the text block and each clause to be screened, providing a basis for further screening and optimization of the clauses.

[0058] Specifically, for each segmented text block, relying on a pre-constructed knowledge structure graph of energy storage terms, the semantic retrieval and association reasoning functions of the graph are used to filter at least one energy storage term highly related to the text block information from a massive number of terms. Then, an encoding model is used to generate the embedding vector of the text block and the term embedding vector of each energy storage term to be filtered. By calculating the similarity between the two types of vectors, a second fitting attribute between the text block and each energy storage term to be filtered is determined, accurately quantifying the degree of semantic fit between the two. The calculation expression for the second fitting attribute is as follows:

[0059] ;

[0060] in, For project text blocks Embedded vector, is the embedding vector of the i-th legal clause (or the node attribute of the demand layer node in the knowledge graph).

[0061] In this embodiment, by combining the knowledge structure graph of energy storage terms to filter the terms to be screened, the matching scope is more focused on the core needs of the text block, avoiding interference from invalid terms; similarity calculation based on embedded vectors can deeply mine the potential semantic relationship between the text and the terms, improving the accuracy of the adaptation judgment; the collaboration of the two not only ensures the targeting of the terms screening, but also provides standardized data for subsequent fusion calculation through quantitative attributes, helping the energy storage terms matching process to balance efficiency and reliability.

[0062] Optionally, the knowledge structure graph of energy storage terms, including all preset energy storage terms, is determined as follows: Based on any two preset energy storage terms and pre-set logical constraints, it is determined whether the two preset energy storage terms are available; wherein, the logical constraints include at least the content inclusion constraint, content equivalence constraint, and content contradiction constraint; if the two preset energy storage terms are available, the two preset energy storage terms are respectively taken as root nodes, and the root attribute of the root node is determined according to the term attribute information of the energy storage terms; for the next-level child node associated with the root node, the first sub-attribute of the child node is determined according to the preset energy storage terms corresponding to the root node, wherein the first sub-attribute is related to the entity information associated with the preset energy storage terms; for the next-level leaf node associated with the child node, the term logic information is determined according to the preset energy storage terms of the root node associated with the leaf node, and the term logic information is taken as the node attribute of the leaf node.

[0063] Specifically, the clause logical constraints can be understood as a standardized set of rules used to verify the compatibility between pre-defined energy storage clauses. These rules aim to determine whether any two clauses can coexist in the knowledge structure graph, avoiding redundancy within the system. Clause logical constraints include, but are not limited to, clause content inclusion constraints, clause content equivalence constraints, and clause content contradiction constraints. It should be noted that clause content inclusion constraints refer to rules that determine whether the core provisions of one clause (such as performance standards or performance requirements) are fully covered by another. Clause content equivalence constraints can be specifically understood as rules that determine whether the key provisions of two clauses are completely identical and without substantial differences. Clause content contradiction constraints can be specifically understood as rules that identify conflicts between the core contents of two clauses (such as applicable scenarios or key indicators) and cannot be effective simultaneously. These three constraints together constitute the core basis for clause usability verification.

[0064] Specifically, based on all the pre-set energy storage clauses, and according to the three core logical constraints of clause content inclusion, equivalence, and contradiction, the usability of any two pre-set energy storage clauses is verified one by one to eliminate redundant clauses. After determining that two clauses are usable, they are set as root nodes of the knowledge structure graph, that is, as clause layer nodes. The root attributes of the root nodes are determined by combining the clause attribute information of the energy storage clauses. The root attributes of the root nodes include, but are not limited to, clause standard name, version number, clause number, clause original text, scope of application, and historical version information. For the next-level child nodes associated with the root node, i.e., concept layer nodes, based on the preset energy storage terms of the corresponding root node, the concept information associated with the terms is extracted as the first sub-attribute of the child node. For example, the concept information includes, but is not limited to, the concepts of equipment, location, operating conditions and physical quantities mentioned in the terms. Then, for the next-level leaf nodes associated with the child nodes, entity information is extracted according to the preset energy storage terms of their associated root nodes. The entity information includes, but is not limited to, subject, behavior, object, condition, logical words and logical types. This entity information is set as the node attribute of the leaf node, and finally, a knowledge structure graph of energy storage terms containing all preset energy storage terms is constructed.

[0065] For example, based on clause extraction, terminology recognition, and logical relationship mining, the expression form of the energy storage clause knowledge structure graph G, composed of three types of nodes: Clause, Concept, and Requirement, is shown below:

[0066] ;

[0067] in, Indicates a clause-level node; Represents a concept layer node; E represents the structured requirement layer node; E represents the set of edges relating to the relationships between different types of nodes.

[0068] For example, clause-level construction: First, the standard text is segmented at the clause level, breaking down the chapters, articles, and clauses in the standard into the smallest clause units. And create a Clause node for each clause. Each clause node stores the following metadata: standard name (e.g., "GB / T36276"); version number (e.g., "2018"); clause number (e.g., "5.4"); original clause text; scope of application, mandatory / recommended attributes, etc. Concept layer construction: For each clause, entity identification in the execution domain is performed, extracting concepts such as equipment, location, operating conditions, and physical quantities mentioned in the clause, and then standardizing the terminology (including synonym normalization, vector similarity aggregation, etc.) to form Concept nodes. To avoid rendering errors caused by the mixing of Chinese characters in a mathematical environment, each Concept node and its meaning can be displayed in a list format. Taking the example clause as an example, the following Concept nodes can be identified and normalized: Concept-ES: Energy Storage Battery System, Concept-Indoor: Indoor Installation, Concept-FP: Independent Fire Compartment, Concept-FD: Fire Detection System, Concept-EX: Explosion Suppression System; The system also establishes “mentions” relationship edges between the clause and the above concept nodes in the knowledge graph, for example: (Clause-C1, Concept-ES, “mentions”), (Clause-C1, Concept-Indoor, “mentions”), (Clause-C1, Concept-FP, “mentions”), (Clause-C1, Concept-FD, “mentions”), (Clause-C1, Concept-EX, “mentions”). This structure enables the clause text and term semantics to form a complete one-to-many connection relationship. Requirement Layer Construction: After processing by the LLM (Large Language Model) clause extraction layer, structured requirements can be extracted, including Subject, Verb, Object, Condition, LogicWord (such as "shall," "shall not," "may"), and LogicType (mandatory / prohibited / allowed / recommended). Let the clauses be... It can be broken down into several requirements Then, a corresponding Requirement node is created for each requirement, and relationship edges are established in the following manner: (Clause Requirement): Indicates the requirements given in the terms and conditions. Concept): Represents the subject object or constraint object that is required to be applied. For example, the clause is broken down into two structured requirements: Requirement-R1: When the energy storage battery system (Subject) is installed indoors (Condition), an independent fire compartment (Object) shall be set up (Verb); Requirement-R2: When the energy storage battery system (Subject) is installed indoors (Condition), a fire detection and explosion suppression system (Object) shall be configured (Verb). The following edges are established: (Clause-C1, Requirement-R1, "states"), (Clause-C1, Requirement-R2, "states"), (Requirement-R1, Concept-FP, "constraints"), (Requirement-R2, Concept-FD, "constraints"), (Requirement-R2, Concept-EX, "constraints"); and then a knowledge structure graph of energy storage clauses is formed based on the above-constructed connection relationship.

[0069] In this embodiment, the availability of clauses is verified through multiple logical constraints, eliminating contradictory clauses from the source and ensuring the consistency and compatibility of clauses in the graph. Following a hierarchical structure of root node-child node-leaf node, node attributes are defined layer by layer, combining clause attributes, entity information, and logical information. This makes the graph structure clear and semantically related, facilitating rapid retrieval and location of relevant clauses, and providing structured knowledge support for accurate matching of subsequent text blocks and clauses. The graph covers all preset energy storage clauses, ensuring comprehensive clause coverage. Simultaneously, the hierarchical attribute design improves knowledge reusability and matching efficiency, contributing to the standardization and accuracy of the energy storage clause matching process.

[0070] Optionally, after determining the root node, child nodes, and leaf nodes corresponding to all available preset energy storage terms, the method further includes: obtaining the first sub-attribute of all child nodes, and removing redundancy from child nodes with the same first sub-attribute to obtain the updated energy storage term structure knowledge graph.

[0071] Specifically, after constructing the root node, child node, and leaf node corresponding to all available preset energy storage terms, the system extracts the first sub-attribute of all child nodes in batches, that is, the attribute related to the entity information associated with the terms. Through attribute keyword comparison and core information deduplication algorithm, duplicate or highly homogeneous child nodes are screened and eliminated. Only one valid child node with the same first sub-attribute can be retained, while keeping the original association logic between the child node and the root node and leaf node unchanged, and finally forming the updated knowledge structure graph of energy storage terms.

[0072] In this embodiment, redundancy removal of child nodes simplifies the graph structure, reduces the storage resources occupied by duplicate nodes, and improves the response speed of graph retrieval and reasoning. It also avoids interference from child nodes with the same first sub-attribute to subsequent text block matching, ensuring the accuracy of clause screening and adaptation judgment. This makes the graph hierarchy clearer and the knowledge system simpler, reduces the cost of subsequent graph maintenance and optimization, and further improves the efficiency of the entire process of graph-based clause matching.

[0073] Optionally, after determining the available preset energy storage terms, the method further includes: inputting the preset energy storage terms into a pre-trained language model to extract entity information and logical word information from the preset energy storage terms based on the entity extraction layer in the language model; wherein, the logical word information is used to determine the logical information of the terms; and performing vector transformation processing on the entity information based on the terms vectorization layer in the language model to obtain a terms embedding vector that is compatible with the preset energy storage terms.

[0074] Specifically, after verifying the availability of the preset energy storage terms, all available terms are batch-input into a language model pre-trained with data from the energy storage field. First, the entity extraction layer built into the model accurately extracts entity information and logical word information from the terms. For example, entity information can be the model of energy storage equipment, the subject of performance, performance indicators, etc. Logical word information includes, but is not limited to, "includes," "equivalent to," and "must not," which are words used to define the logic of the terms. Then, with the help of the model's term vectorization layer, the extracted entity information is semantically encoded and vectorized to generate a term embedding vector that can represent the core connotation of the terms, providing a data foundation for subsequent adaptation attribute calculation.

[0075] In this embodiment, relying on the domain adaptation capability of the pre-trained language model, entity and logical word information is extracted efficiently and accurately, avoiding the errors and inefficiencies of manual extraction; entity information is transformed into standardized clause embedding vectors, making the semantics of clauses quantifiable, providing technical support for subsequent vector similarity calculation and accurate adaptation; logical word information is acquired simultaneously, providing supplementary basis for clarifying the logical relationship of clauses and optimizing the attributes of graph nodes, further improving the intelligence and standardization of the overall process.

[0076] Optionally, after obtaining the clause embedding vector, the method further includes associating the clause identifier corresponding to the preset energy storage clause, the clause embedding vector of the preset energy storage clause, and the additional information of the preset energy storage clause, and caching them in the vector library.

[0077] Specifically, after generating the clause embedding vectors corresponding to each preset energy storage clause, the unique clause identifier of each clause is extracted, such as the exclusive number and code, and its additional information is collected. The additional information includes, but is not limited to, the standard name of the document to which the clause belongs, the original clause text, version information, chapter level and applicable scenario. Then, a one-to-one correspondence relationship is established among the three, and the clause identifier, clause embedding vector and additional information of the same preset energy storage clause are integrated into a unified data unit. Then, batch caching is completed according to the storage specifications of the vector library to ensure that the data is stored in a structured manner and can be retrieved quickly.

[0078] In this embodiment, by associating and binding the core identifier of the clause, semantic vector, and supplementary information, integrated management of clause data is achieved, avoiding the confusion caused by scattered storage. The caching mechanism of the vector library greatly improves the speed of vector retrieval during subsequent adaptation calculations, reduces the time and resource consumption of repetitive data processing, and the complete retention of supplementary information provides support for subsequent clause adaptation verification and version tracking. Structured storage also reduces data maintenance costs and ensures the continuity and efficiency of the entire clause matching process.

[0079] S140. For at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, determine the third adaptation attribute of at least one energy storage terms to be displayed associated with the text block; wherein, at least one energy storage terms to be displayed is the union of preset energy storage terms and at least one energy storage terms to be screened.

[0080] Among them, the energy storage clauses to be displayed are a set of clauses covering the core agreed requirements of energy storage projects. They are composed of all preset energy storage clauses and at least one energy storage clause to be screened obtained through knowledge graph screening. This ensures the completeness of the clause scope and includes core clauses that are highly relevant to the text block. The third adaptation attribute is the comprehensive matching result of each text block with the same energy storage clause to be displayed. It is obtained by integrating the first adaptation attribute (the matching degree with preset clauses calculated by the model) and the second adaptation attribute (the fit degree with the clause to be screened represented by vector similarity) corresponding to the text block. The core purpose is to uniformly quantify the overall fit degree between the text block and the clause to be displayed, providing an intuitive basis for the sorting and screening of clauses.

[0081] Specifically, first, the set of energy storage terms to be displayed is defined. This set is the union of all preset energy storage terms and the energy storage terms to be filtered through a knowledge graph. For each text block, the first and second adaptation attributes determined under the same energy storage term to be displayed are extracted. For example, the first adaptation attribute can be the matching result between the text block and the preset terms calculated by the model, and the second adaptation attribute can be the fit between the text block and the term to be filtered represented by vector similarity. The two types of adaptation attributes are fused and calculated using a preset weighted fusion rule, and finally, a third adaptation attribute corresponding to each energy storage term to be displayed is output as the comprehensive judgment result of the matching degree between the text block and the term. For example, the preset weighted fusion rule can be the weights of the two types of attributes allocated according to scenario requirements. For example, the expression for determining the third adaptation attribute is as follows:

[0082] ;

[0083] in, Characterizes the third adaptation attribute. Characterizes the first adaptation attribute, Characterizes the second adaptation attribute.

[0084] In this embodiment, the terms to be displayed cover the core terms after preset and filtering, which ensures the integrity of the matching scope and focuses on highly relevant content, avoiding omission of valid terms or interference from redundant information. By integrating two types of adaptation attributes, the efficiency advantages of batch matching of the model and the semantic depth matching characteristics of vector similarity are taken into account, making the comprehensive results more comprehensive and reliable. At the same time, the unified and quantified third adaptation attribute provides a clear basis for subsequent term sorting and filtering, helping to improve the accuracy and efficiency of energy storage term determination.

[0085] Optionally, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, a third adaptation attribute of at least one energy storage terms associated with the text block is determined, including: obtaining the third adaptation attribute of the text block relative to the same energy storage terms by weighting the first and second adaptation attributes corresponding to the text block under the same energy storage terms.

[0086] Specifically, for each text block and the same energy storage clause to be displayed, the two types of attributes are quantitatively integrated and calculated based on the preset weight allocation ratio of the first and second adaptation attributes, and then through algorithms such as weighted summation. It should be noted that the sum of the weights must be 1. Finally, the third adaptation attribute of the text block and the corresponding energy storage clause to be displayed is output, forming a unified comprehensive matching judgment result.

[0087] In this embodiment, the weighted processing method can fully take into account the core value of the two types of adaptation attributes. It retains the efficiency of batch matching of the target model and incorporates the semantic depth matching advantage brought by vector similarity, making the third adaptation attribute more comprehensive and in line with actual needs. The weight ratio can be flexibly adjusted to adapt to the differentiated requirements of different project scenarios. At the same time, the unified quantification results provide a clear and unified judgment standard for subsequent clause sorting and screening, helping to quickly lock in highly adaptable clauses and improve the accuracy and efficiency of energy storage clause determination.

[0088] S150. For the same energy storage terms to be displayed, all third-party adaptation attributes associated with the energy storage terms to be displayed are weighted and processed to obtain the target adaptation attributes of the energy storage terms to be displayed relative to the text to be processed.

[0089] Specifically, the target adaptation attribute can be understood as a comprehensive result that quantitatively represents the degree of matching between a single energy storage clause to be displayed and the overall text to be processed. It is obtained by weighting and integrating the third adaptation attribute of all text blocks associated with the clause, combined with the weighting of the information importance of each text block. This not only takes into account the core value differences of different text blocks, but also coordinates the adaptation data of all related text blocks. It can intuitively reflect the overall fit between the energy storage clause to be displayed and the project requirements carried by the text to be processed, and is the core judgment basis for subsequent clause sorting, screening and final determination.

[0090] Specifically, for a single energy storage clause to be displayed, firstly, all the third adaptation attributes corresponding to the text blocks associated with it (i.e., the comprehensive matching quantitative result of each text block with the clause) are summarized. Then, according to the importance of the information of the text blocks (such as the core requirement description block and key parameter block having higher weight than the supplementary explanation block), the corresponding weights are preset. All third adaptation attributes are integrated and calculated through a weighted summation algorithm (the sum of the weights is 1). Finally, the target adaptation attribute of the energy storage clause to be displayed relative to the overall text to be processed is output, which quantitatively represents the overall fit between the clause and the text.

[0091] In this embodiment, weights are differentiated according to the importance of text blocks, which highlights the impact of core information on the adaptation results and makes the target adaptation attributes more in line with the actual needs of the project. The comprehensive adaptation data of all related text blocks is integrated to avoid the one-sidedness caused by a single text block and improve the comprehensiveness and reliability of the results. The unified and quantified target adaptation attributes provide a clear basis for subsequent clause sorting and screening, helping to quickly identify highly adaptable core clauses and greatly improve the efficiency and accuracy of energy storage clause determination.

[0092] S160. Display the third adaptation attribute at the preset position of each text block of the text to be processed, and display the target adaptation attribute of the energy storage terms to be displayed at the target position of the text to be processed.

[0093] Specifically, the system defines the preset display positions of text blocks (such as the end of the text block or the side annotation area) and the target display positions of the text to be processed (such as the summary area at the top of the text or the corresponding column in the clause list). The system retrieves the third adaptation attribute associated with each text block and the target adaptation attribute of each energy storage clause to be displayed from the data storage module. Through the front-end rendering logic, the third adaptation attribute is accurately mounted to the preset position of the corresponding text block. At the same time, the target adaptation attribute is displayed in the target position of the text to be processed according to the rules, ensuring that the two types of attributes correspond one-to-one with the text and clauses and are intuitive and easy to find.

[0094] In this embodiment, the fixed display position allows users to quickly associate text blocks with third-party adaptation attributes and overall text with target adaptation attributes, significantly reducing information verification costs. The visual presentation of adaptation attributes eliminates the need for additional query operations, allowing users to intuitively grasp the degree of matching between clauses and text, and text blocks, providing a clear reference for subsequent screening decisions. At the same time, the clear association display helps to quickly locate highly compatible clauses, further improving the efficiency and convenience of determining energy storage clauses.

[0095] Optionally, a third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and a target adaptation attribute of the energy storage terms to be displayed is displayed at a target position in the text to be processed. This includes: displaying a clause jump icon and a corresponding third adaptation attribute for at least some of the associated energy storage terms to be displayed at a preset position in each text block of the text to be processed; displaying clause jump icons and target adaptation attributes for all or at least some of the energy storage terms to be displayed at the target position in the text to be processed; wherein the clause jump icon is clickable to display the clause content of the energy storage terms to be displayed; at least some of the energy storage terms to be displayed are energy storage terms to be displayed whose target adaptation attributes satisfy a first preset condition, and / or, energy storage terms to be displayed whose third adaptation attributes satisfy a second preset condition.

[0096] Specifically, first, preset display positions for text blocks are set, such as the end and sidebar, and target display positions for the text to be processed, such as the top summary area and side clause bar. At least some of the energy storage clauses to be displayed are selected based on their target adaptation attributes meeting the first preset condition and their third adaptation attributes meeting the second preset condition. The first preset condition can refer to a high adaptation threshold of ≥0.8, and the second preset condition can refer to an adaptation threshold of ≥0.75. It should be noted that the first preset condition is greater than the second preset condition. Clickable clause jump indicators are configured for these clauses and all clauses to be displayed. Subsequently, the system renders the text at the preset position of each text block, displaying the corresponding jump indicators and third adaptation attributes of the associated clauses. At the target position, the system displays the jump indicators and target adaptation attributes of all or some of the aforementioned clauses. Users can click the jump indicators to view the complete clause content.

[0097] In this embodiment, highly compatible terms are selected and highlighted based on preset conditions to avoid information overload and allow users to quickly focus on the core matching content. The terms jump indicator is bound to the compatibility attribute, which not only intuitively conveys the degree of matching, but also supports one-click tracing to view the terms details, reducing the cost of information verification and validation. The two types of attributes are displayed precisely in different positions, taking into account both the matching perspective of text blocks and the overall text, which is convenient for checking local details and for global filtering decisions, greatly improving the efficiency and convenience of energy storage terms determination.

[0098] Optionally, after constructing the knowledge structure graph of energy storage provisions, i.e., the three-layer knowledge graph (Clause / Concept / Requirement), the SPARQL+LLM query layer provides a hybrid query capability for energy storage safety regulations: input query information, accurately retrieve structured knowledge based on the query information through SPARQL, and then use a large language model (LLM) to perform natural language completion and interpretation on the retrieval results, thereby achieving highly interpretable regulatory query and reasoning. For example, structured queries can use SPARQL (SPARQL Protocol and RDF Query Language) to query the knowledge structure graph of energy storage clauses. The Clause, Concept, and Requirement nodes and their relationships in the knowledge graph are all represented in the form of triples. Therefore, the following content can be accurately retrieved through SPARQL: (1) the hierarchical relationship between specified clauses (Clause); (2) the subject, behavior, object, and condition involved in a certain requirement (Requirement); (3) the set of clauses in which a certain concept (Concept) is located; (4) the logical relationship between clauses (inclusion, equivalence, contradiction); (5) all relevant clauses of a certain type of technical requirement (such as "fire detection", "ventilation requirements", "temperature rise limit"). Structured queries can extract regulatory information from knowledge graphs, including but not limited to: structured requirements of regulatory clauses, such as "whether a certain standard requires the configuration of a fire detection system under indoor installation conditions"; cross-standard clause alignment results, such as "which clauses in GB / T36276-2018 are equivalent to clause 5.4 in IEC62933-5-2"; logical relationship edges, such as "whether there are contradictory clauses"; complete set queries of technical elements, such as "the set of all clauses involving explosion suppression systems"; and condition matching, such as all Requirement nodes that meet the conditions of "indoor installation + lithium battery + fire requirements". Therefore, the essence of structured queries is to perform precise matching, filtering, and association queries on the three types of nodes and edges in the graph. After performing a query using SPARQL, the query results are converted into labeled structured prompts to provide a complete task description for LLMs (such as those based on the GPT series or large-scale instruction fine-tuning models) (e.g., "Please generate a Chinese explanation of the query results"). Instruction-following capabilities are used to summarize, complete, and explain the results. Natural language results are output, such as: "According to Clause 5.4 of GB / T36276-2018, indoor energy storage battery systems must be equipped with fire detection systems and should be set up with fire compartments." The generated natural language completion can be directly used for engineering audits, compliance reports, project reviews, and intelligent question answering.

[0099] The technical solution of this embodiment involves obtaining the text to be processed for the energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text input by the user in the front-end interactive interface and / or the imported project text; dividing the text to be processed into at least one text block according to preset rules, and inputting at least one text block and preset energy storage terms into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms; for at least one text block, determining at least one energy storage term to be screened corresponding to the text block based on the text block and the pre-built knowledge structure graph of energy storage terms, and determining the relationship between the text block and at least one energy storage term to be screened based on the embedding vector of the text block and the term embedding vector of the at least one energy storage term to be screened. The second adaptation attribute between the energy storage terms to be screened; for at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, determine the third adaptation attribute of at least one energy storage term to be displayed associated with the text block; wherein, at least one energy storage term to be displayed is the union of the preset energy storage terms and at least one energy storage term to be screened; for the same energy storage term to be displayed, weight all the third adaptation attributes associated with the energy storage term to be displayed are weighted to obtain the target adaptation attribute of the energy storage term to be displayed relative to the text to be processed; the third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and the target adaptation attribute of the energy storage term to be displayed is displayed at the target position in the text to be processed. This solution achieves precise matching of energy storage terms through multi-stage collaboration. Diverse text collection methods at the front end adapt to different user scenarios, and text block splitting improves the targeting of matching. The combination of target model with knowledge graph and vector calculation ensures both batch processing efficiency and improves semantic matching accuracy. The terms to be displayed are comprehensive and focused on the core, and two rounds of weighted fusion ensure that the matching results take into account both local and overall needs. Visual display reduces information verification costs, and intelligent operation throughout the process greatly reduces human intervention, achieving high efficiency, accuracy, and convenience in terms determination.

[0100] Example 2

[0101] Figure 2 This is a flowchart of a data processing method provided in Embodiment 2 of the present invention. The method in this embodiment is a further optimization of the method in the above embodiments. Optionally, it involves determining the associated entities and logical relationships corresponding to the text block, and querying the knowledge structure graph of energy storage terms based on the associated entities and logical relationships to obtain at least one energy storage term to be screened associated with the text block; based on the term identifier and / or additional information of the at least one energy storage term to be screened, searching the vector library for a term embedding vector that matches the at least one energy storage term to be screened; calculating the similarity between the term embedding vector of each energy storage term to be screened and the embedding vector of the text block to obtain a second adaptation attribute of the text block relative to each energy storage term to be screened. Figure 2 As shown, the method includes:

[0102] S210. Obtain the text to be processed for the energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text.

[0103] S220. Divide the text to be processed into at least one text block according to a preset rule, and input the at least one text block and the preset energy storage terms into the pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms.

[0104] S230. Determine the associated entities and logical relationships corresponding to the text block, and query the knowledge structure graph of energy storage terms based on the associated entities and logical relationships to obtain at least one energy storage term to be screened associated with the text block.

[0105] Specifically, by using entity recognition and logical relationship extraction models, related entities (such as energy storage equipment models, contracting entities, performance indicators, term requirements, etc.) and logical relationships between entities (such as requirements, adaptation, prohibition, inclusion, etc.) are accurately extracted from text blocks. Then, using these related entities and logical relationships as search conditions, semantic retrieval and association reasoning are performed on the pre-constructed knowledge structure graph of energy storage terms, and at least one energy storage term to be screened that highly matches the core information of the text block is selected, forming a targeted subset of terms.

[0106] For example, for a task that determines the associated entities and logical relationships corresponding to a text block, corresponding prompts are constructed according to a preset prompt template. The prompt template includes the following:

[0107] Role: You are a legal clause semantic analysis system, and you need to determine the logical relationship between two clauses.

[0108] Task: Your task is to synthesize the subject, action, object, logical connector, and condition of the two clauses and infer the relationship between them.

[0109] Available logical relationships: 1. Contain: For example, clause A's requirements are more comprehensive or stricter than clause B's, or B is a special case of A. 2. Equivalent: For example, the technical requirements, constraints, or logical terms expressed by the two clauses are consistent or equivalent. 3. Conflict: For example, the requirements of the two clauses conflict with each other, such as "must" vs. "must not," "must ≥ X" vs. "must ≤ X," or the conditions are incompatible.

[0110] Output format: Please output in the following JSON format:

[0111] {"relation":"contain|equivalent|conflict",

[0112] "reason": "Briefly explain the reasons for your judgment, no more than 50 words".

[0113] "dependency":{

[0114] "subject_match":true / false,

[0115] "action_match":true / false,

[0116] "object_match":true / false,

[0117] "logic_word_match":true / false,

[0118] "condition_conflict":true / false}

[0119] }

[0120] Based on the prompt word template, the text block containing the entities and logical relationships to be extracted, along with the prompt words, is input into a pre-trained large language model to obtain the entities and logical relationships corresponding to the text block. For example, if the text to be extracted is: "5.3.2 Energy storage systems should be equipped with bidirectional fuses, with rated current ≥ discharge peak value," the output will be:

[0121] {"Subject":"Energy Storage System",

[0122] "Verb":"Should be configured",

[0123] "Object":"Bidirectional fuse",

[0124] "Constraint":"I_rated≥I_peak",

[0125] "Norm":"mandatory"}.

[0126] In this embodiment, precise retrieval based on entities and logical relationships makes the association between the clauses to be filtered and the text blocks more targeted, avoiding interference from invalid information caused by broad matching; with the reasoning ability of the knowledge structure graph, core relevant clauses can be quickly located, greatly improving the filtering efficiency; the filtering results focus on the core needs of the text blocks, providing a high-quality foundation for subsequent adaptation attribute calculations, and further ensuring the accuracy of the final energy storage clause adaptation.

[0127] S240. Based on the clause identifier and / or additional information of at least one energy storage clause to be screened, find the clause embedding vector that matches the at least one energy storage clause to be screened from the vector library.

[0128] Specifically, the clause identifier can be understood as standardized identification information used to uniquely distinguish pre-defined energy storage clauses. It can exist in the form of numbers, codes, etc. Its core function is to achieve accurate positioning, association, and management of clauses, avoiding confusion between different clauses. Supplementary information can be understood as supplementary explanatory data for the clauses, mainly including original clause information and version association information. Version information includes, but is not limited to, the clause's publication version number, update time, and applicable version scope. Together, they provide comprehensive support for the retrieval, verification, and traceability of clauses.

[0129] Specifically, the key search criteria for at least one energy storage clause to be screened are first collected, including the unique clause identifier (such as a unique number or code) and / or additional information (including the original clause text, version information, etc.) for each clause. Then, using this information as search keywords, the precise query interface of the vector library is called. Through methods such as identifier matching and semantic association of additional information, the clause embedding vectors that are pre-associated and stored in the vector library with each energy storage clause to be screened are located and extracted, forming a vector dataset that corresponds one-to-one with the clauses to be screened, providing data support for subsequent adaptation attribute calculation.

[0130] In this embodiment, based on the uniqueness of the clause identifier and the supplementary verification of additional information, it is ensured that the retrieved clause embedding vector accurately matches the clause to be filtered, avoiding vector confusion or mismatch. The efficient retrieval capability of the vector library greatly shortens the time spent on vector retrieval, eliminating the need to repeatedly generate embedding vectors, which reduces the consumption of computing resources and improves the overall process efficiency. At the same time, the retrieval logic based on the identifier and additional information is compatible with the structured storage characteristics of the vector library, ensuring the stability and consistency of data retrieval, and laying a reliable foundation for subsequent similarity calculation.

[0131] S250. Calculate the similarity between the embedding vector of each energy storage clause to be screened and the embedding vector of the text block to obtain the second adaptation attribute of the text block relative to each energy storage clause to be screened.

[0132] Specifically, the embedding vectors of the text blocks (which have been generated by the encoding model to capture the core semantics of the text) and the embedding vectors of the clauses corresponding to each energy storage clause to be screened (precisely retrieved from the vector library) are obtained first. Then, a preset similarity calculation algorithm (such as cosine similarity, Euclidean distance, etc.) is used to calculate the semantic fit between the embedding vector of a single text block and the embedding vector of each clause to be screened. The quantified result obtained is directly used as the second adaptation attribute of the text block relative to the corresponding energy storage clause to be screened.

[0133] In this embodiment, vector similarity calculation can deeply mine the potential semantic relationship between text and clauses, which is more accurate than surface keyword matching and ensures the reliability of the adaptation results. The algorithm has high execution efficiency and can quickly complete batch vector comparison, improving the overall process speed. At the same time, the quantified results provide standardized data for subsequent fusion with the first adaptation attribute, helping to form a more scientific comprehensive adaptation judgment.

[0134] S260. For at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, determine the third adaptation attribute of at least one energy storage terms to be displayed associated with the text block; wherein, at least one energy storage terms to be displayed is the union of preset energy storage terms and at least one energy storage terms to be screened.

[0135] S270. For the same energy storage terms to be displayed, all third-party adaptation attributes associated with the energy storage terms to be displayed are weighted and processed to obtain the target adaptation attribute of the energy storage terms to be displayed relative to the text to be processed.

[0136] S280, Display the third adaptation attribute at the preset position of each text block of the text to be processed, and display the target adaptation attribute of the energy storage terms to be displayed at the target position of the text to be processed.

[0137] The technical solution of this embodiment collects user-input descriptive text and / or imported project text as the text to be processed for the energy storage terms to be determined through a front-end interactive interface. This text is then split into at least one text block according to preset rules. The text blocks and preset energy storage terms are input into a pre-trained target model to obtain the first matching attribute between each text block and the preset terms. Simultaneously, related entities and logical relationships are extracted from the text blocks. Based on this, the knowledge structure graph of energy storage terms is queried to filter out the energy storage terms to be screened. Then, based on the identifier and / or additional information of the terms to be screened, the corresponding term embedding vector is retrieved from the vector library. The similarity between this vector and the text block embedding vector is calculated to obtain the second matching attribute. The union of the preset terms and the terms to be screened is used as the energy storage terms to be displayed. The first and second matching attributes corresponding to the same term to be displayed are weighted and integrated to obtain the third matching attribute. Then, all the third matching attributes of the same term to be displayed are weighted according to the text block weight to generate the target matching attribute. Finally, the third matching attribute is displayed at a preset position in each text block, and the target matching attribute is displayed at the target position in the text to be processed, completing the entire process. It achieves accurate matching of energy storage terms through multi-dimensional collaboration. The front end conveniently collects text and splits it with preset rules, balancing input flexibility and processing standardization. The target model is combined with knowledge graphs and vector libraries, which not only ensures batch matching efficiency through model, but also deepens the accuracy of semantic matching through entity logic retrieval and vector similarity calculation. The terms to be displayed are comprehensive and focused on the core, and the two-round weighted fusion ensures that the matching attributes take into account both local text blocks and overall text requirements. The visualization display makes the matching relationship intuitive and verifiable. The intelligent operation throughout the process greatly reduces human intervention, making the determination of terms efficient, accurate and convenient.

[0138] Example 3

[0139] Figure 3 This is a schematic diagram of the structure of a data processing device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes:

[0140] The pending text acquisition module 310 is used to acquire the pending text of the energy storage terms to be determined, wherein the pending text includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text.

[0141] The first adaptation attribute determination module 320 is used to divide the text to be processed into at least one text block according to a preset rule, and input at least one text block and a preset energy storage clause into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage clause.

[0142] The second adaptation attribute determination module 330 is used to determine, for at least one text block, at least one energy storage clause to be screened corresponding to the text block based on the text block and a pre-built knowledge structure graph of energy storage clauses, and to determine the second adaptation attribute between the text block and at least one energy storage clause to be screened based on the embedding vector of the text block and the clause embedding vector of at least one energy storage clause to be screened.

[0143] The third adaptation attribute determination module 340 is used to determine, for at least one text block, the third adaptation attribute of at least one energy storage clause associated with the text block based on the first adaptation attribute and the second adaptation attribute corresponding to the text block under the same energy storage clause to be displayed; wherein, at least one energy storage clause to be displayed is the union of preset energy storage clauses and at least one energy storage clause to be filtered.

[0144] The target adaptation attribute determination module 350 is used to weight all the third adaptation attributes associated with the same energy storage terms to be displayed, and to obtain the target adaptation attribute of the energy storage terms to be displayed relative to the text to be processed.

[0145] The adaptation attribute display module 360 ​​is used to display the third adaptation attribute at a preset position in each text block of the text to be processed, and to display the target adaptation attribute of the energy storage terms to be displayed at the target position in the text to be processed.

[0146] The technical solution of this embodiment involves obtaining the text to be processed (the text to be processed) of the energy storage terms to be determined through a text acquisition module. The text to be processed includes at least the descriptive text input by the user in the front-end interactive interface and / or imported project text. A first adaptation attribute determination module divides the text to be processed into at least one text block according to preset rules, and inputs at least one text block and preset energy storage terms into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage terms. For each text block, a second adaptation attribute determination module, based on the text block and a pre-built knowledge structure graph of energy storage terms, determines at least one energy storage term to be screened corresponding to the text block, and determines the relationship between the text block and at least one preset energy storage term based on the embedding vector of the text block and the term embedding vector of the at least one preset energy storage term. The second adaptation attribute among the energy storage terms to be screened; the third adaptation attribute determination module, for at least one text block, determines the third adaptation attribute of at least one energy storage term to be displayed associated with the text block based on the first and second adaptation attributes corresponding to the text block under the same energy storage term to be displayed; wherein, at least one energy storage term to be displayed is the union of the preset energy storage terms and at least one energy storage term to be screened; the target adaptation attribute determination module, for the same energy storage term to be displayed, performs weighted processing on all the third adaptation attributes associated with the energy storage term to be displayed to obtain the target adaptation attribute of the energy storage term to be displayed relative to the text to be processed; the adaptation attribute display module displays the third adaptation attribute at a preset position in each text block of the text to be processed, and displays the target adaptation attribute of the energy storage term to be displayed at the target position in the text to be processed. This system enables convenient collection of user-inputted or imported text from the front end, text block splitting using preset rules, rapid generation of the first matching attribute using a pre-trained target model, accurate screening of energy storage clauses based on a knowledge structure graph of energy storage clauses, and obtaining the second matching attribute through vector similarity calculation. After two rounds of weighted fusion, the third matching attribute and the target matching attribute are obtained, and the results are finally presented visually. The entire process requires minimal manual intervention. It improves the efficiency and accuracy of clause matching by leveraging models and graphs, ensures comprehensive matching scope through the union of preset and screened clauses, and considers the matching needs of text blocks and the overall text through hierarchical matching attribute calculation. Visualization makes the matching relationship intuitive and easy to understand, significantly reducing decision-making costs and ultimately achieving efficient, accurate, and convenient determination of energy storage clauses.

[0147] Based on the above embodiments, optionally, the second adaptation attribute determination module 330 is specifically used to determine whether two preset energy storage terms are available based on any two preset energy storage terms and pre-set clause logical constraints; wherein, the clause logical constraints include at least clause content inclusion constraints, clause content equivalence constraints, and clause content contradiction constraints; if the two preset energy storage terms are available, the two preset energy storage terms are respectively used as root nodes, and the root attribute of the root node is determined according to the clause attribute information of the energy storage terms; for the next-level child node associated with the root node, the first sub-attribute of the child node is determined according to the preset energy storage terms corresponding to the root node, wherein the first sub-attribute is related to the entity information associated with the preset energy storage terms; for the next-level leaf node associated with the child node, the clause logical information is determined according to the preset energy storage terms of the root node associated with the leaf node, and the clause logical information is used as the node attribute of the leaf node.

[0148] Optionally, after determining the root node, child node, and leaf node corresponding to all available preset energy storage terms, the second adaptation attribute determination module 330 is specifically used to obtain the first sub-attribute of all child nodes and to remove redundancy from child nodes with the same first sub-attribute to obtain the updated energy storage term structure knowledge graph.

[0149] Optionally, after determining the available preset energy storage terms, the second adaptation attribute determination module 330 is further specifically used to input the preset energy storage terms into a pre-trained language model, so as to extract entity information and logical word information in the preset energy storage terms based on the entity extraction layer in the language model; wherein, the logical word information is used to determine the logical information of the terms; based on the terms vectorization layer in the language model, the entity information is subjected to vector transformation processing to obtain the terms embedding vector that is compatible with the preset energy storage terms.

[0150] Optionally, after obtaining the clause embedding vector, the second adaptation attribute determination module 330 is further specifically used to associate the clause identifier corresponding to the preset energy storage clause, the clause embedding vector of the preset energy storage clause, and the additional information of the preset energy storage clause, and cache them in the vector library.

[0151] Optionally, the first adaptation attribute determination module 320 is specifically used to input the text block and all preset energy storage terms into the target model for at least one text block, so that the target model outputs the first adaptation attribute between the text block and each preset energy storage term.

[0152] Optionally, the second adaptation attribute determination module 330 is specifically used to determine the associated entities and logical relationships corresponding to the text block, and query the knowledge structure graph of energy storage terms based on the associated entities and logical relationships to obtain at least one energy storage term to be screened associated with the text block; based on the term identifier and / or additional information of at least one energy storage term to be screened, search the vector library for a term embedding vector that is compatible with at least one energy storage term to be screened; calculate the similarity between the term embedding vector of each energy storage term to be screened and the embedding vector of the text block to obtain the second adaptation attribute of the text block relative to each energy storage term to be screened.

[0153] Optionally, the third adaptation attribute determination module 340 is specifically used to obtain the third adaptation attribute of the text block relative to the same energy storage clause by weighting the first adaptation attribute and the second adaptation attribute corresponding to the text block under the same energy storage clause.

[0154] Optionally, the adaptation attribute display module 360 ​​is specifically used to display, at least some of the associated energy storage terms to be displayed and the corresponding third adaptation attribute, at a preset position in each text block of the text to be processed; and to display, at a target position in the text to be processed, all or at least some of the energy storage terms to be displayed and the target adaptation attribute; wherein, the term jump mark is used to be clicked to display the term content of the energy storage terms to be displayed; at least some of the energy storage terms to be displayed are energy storage terms to be displayed whose target adaptation attribute satisfies the first preset condition, and / or, energy storage terms to be displayed whose third adaptation attribute satisfies the second preset condition.

[0155] The data processing apparatus provided in the embodiments of the present invention can execute the data processing method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0156] Example 4

[0157] Figure 4 This is a schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0158] like Figure 4As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0159] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0160] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as data processing methods.

[0161] In some embodiments, the data processing method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the data processing method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the data processing method by any other suitable means (e.g., by means of firmware).

[0162] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0163] Computer programs for implementing the data processing methods of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The computer programs can be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0164] Example 5

[0165] Embodiment 5 of the present invention also provides a computer-readable storage medium storing computer instructions for causing a processor to execute a data processing method, the method comprising:

[0166] Obtain the text to be processed for the energy storage terms to be determined, wherein the text to be processed includes at least the descriptive text entered by the user in the front-end interactive interface and / or the imported project text;

[0167] The text to be processed is divided into at least one text block according to a preset rule, and at least one text block and a preset energy storage clause are input into a pre-trained target model to determine the first adaptation attribute between each text block and the preset energy storage clause.

[0168] For at least one text block, based on the text block and the pre-built knowledge structure graph of energy storage terms, at least one energy storage term to be screened corresponding to the text block is determined, and based on the embedding vector of the text block and the term embedding vector of at least one energy storage term to be screened, a second adaptation attribute between the text block and at least one energy storage term to be screened is determined.

[0169] For at least one text block, based on the first and second adaptation attributes corresponding to the text block under the same energy storage terms to be displayed, the third adaptation attribute of at least one energy storage terms to be displayed associated with the text block is determined; wherein, at least one energy storage terms to be displayed is the union of preset energy storage terms and at least one energy storage terms to be screened.

[0170] For the same energy storage terms to be displayed, all third-party adaptation attributes associated with the energy storage terms to be displayed are weighted and processed to obtain the target adaptation attributes of the energy storage terms to be displayed relative to the text to be processed.

[0171] The third adaptation attribute is displayed at a preset position in each text block of the text to be processed, and the target adaptation attribute of the energy storage terms to be displayed is displayed at the target position in the text to be processed.

[0172] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0173] To provide interaction with an object, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the object; and a keyboard and pointing device (e.g., a mouse or trackball) through which the object provides input to the electronic device. Other types of devices can also be used to provide interaction with the object; for example, feedback provided to the object can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the object can be received in any form (including sound input, voice input, or tactile input).

[0174] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., a computer with a graphical user interface or web browser through which an item can interact with the implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0175] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0176] Example 6

[0177] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the data processing method of any embodiment of this invention.

[0178] In the implementation of a computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​as well as conventional procedural programming languages. The program code can be executed 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 remote computers, the remote computer can 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 can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0179] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0180] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A data processing method, characterized by, The method comprises the following steps: acquiring a to-be-processed text of a to-be-determined energy storage clause, wherein the to-be-processed text at least comprises a description text input by a user in a front-end interactive interface and / or a project text imported; dividing the to-be-processed text into at least one text block according to a preset rule, inputting the at least one text block and a preset energy storage clause into a target model pre-trained to determine a first adaptation attribute between each text block and the preset energy storage clause; for the at least one text block, determining at least one to-be-screened energy storage clause corresponding to the text block based on the text block and a pre-constructed energy storage clause knowledge structure graph, and determining a second adaptation attribute between the text block and the at least one to-be-screened energy storage clause according to an embedding vector of the text block and a clause embedding vector of the at least one to-be-screened energy storage clause; for the at least one text block, determining a third adaptation attribute of at least one to-be-displayed energy storage clause associated with the text block according to the first adaptation attribute and the second adaptation attribute of the text block corresponding to the same to-be-displayed energy storage clause, wherein the at least one to-be-displayed energy storage clause is a union of the preset energy storage clause and the at least one to-be-screened energy storage clause; for the same to-be-displayed energy storage clause, performing weighted processing on all third adaptation attributes associated with the to-be-displayed energy storage clause to obtain a target adaptation attribute of the to-be-displayed energy storage clause relative to the to-be-processed text; displaying the third adaptation attribute at a preset position of each text block of the to-be-processed text, and displaying the target adaptation attribute of the to-be-displayed energy storage clause at a target position of the to-be-processed text.

2. The method of claim 1, wherein, The energy storage clause knowledge structure graph comprising all preset energy storage clauses is determined by the following method: determining whether two preset energy storage clauses are usable based on the two preset energy storage clauses and a pre-set clause logical constraint condition, wherein the clause logical constraint condition at least comprises a clause content inclusion constraint condition, a clause content equivalence constraint condition and a clause content contradiction constraint condition; in the case that the two preset energy storage clauses are usable, taking the two preset energy storage clauses as root nodes respectively, and determining root attributes of the root nodes according to clause attribute information of the energy storage clauses; for a next-level child node associated with the root node, determining a first child attribute of the child node according to a preset energy storage clause corresponding to the root node, wherein the first child attribute is related to entity information associated with the preset energy storage clause; for a next-level leaf node associated with the child node, determining clause logical information according to a preset energy storage clause of a root node associated with the leaf node, and taking the clause logical information as a node attribute of the leaf node.

3. The method of claim 2, wherein, After determining the root nodes, child nodes and leaf nodes corresponding to all available preset energy storage clauses, the method further comprises: acquiring the first child attributes of all child nodes, and performing redundancy processing on child nodes with the same first child attribute to obtain an updated energy storage clause structure knowledge graph.

4. The method of claim 2, wherein, After determining the available preset energy storage clauses, the method further comprises: inputting the preset energy storage clause into a pre-trained language model, to extract entity information and logical word information in the preset energy storage clause based on an entity extraction layer in the language model; wherein the logical word information is used to determine the clause logical information; based on a clause vectorization layer in the language model, performing vector conversion processing on the entity information to obtain a clause embedding vector adapted to the preset energy storage clause.

5. The method of claim 4, wherein, After obtaining the clause embedding vector, the method further comprises: associating the clause identifier corresponding to the preset energy storage clause, the clause embedding vector of the preset energy storage clause, and additional information of the preset energy storage clause, and caching them into a vector library.

6. The method of claim 1, wherein, The inputting the at least one text block and the preset energy storage clause into the pre-trained target model to determine the first adaptation attribute between each of the text blocks and the preset energy storage clause comprises: For the at least one text block, inputting the text block and all preset energy storage clauses into the target model, so that the target model outputs the first adaptation attribute between the text block and each preset energy storage clause.

7. The method of claim 1, wherein, The determining the at least one to-be-screened energy storage clause corresponding to the text block based on the text block and the pre-constructed energy storage clause knowledge structure graph, and determining the second adaptation attribute between the text block and the at least one to-be-screened energy storage clause based on the embedding vector of the text block and the clause embedding vector of the at least one to-be-screened energy storage clause comprises: determining the associated entity and logical relationship corresponding to the text block, and querying the energy storage clause knowledge structure graph according to the associated entity and logical relationship to obtain at least one to-be-screened energy storage clause associated with the text block; According to the clause identifier and / or additional information of the at least one to-be-screened energy storage clause, searching the vector library for a clause embedding vector adapted to the at least one to-be-screened energy storage clause; calculating the similarity between the clause embedding vector of each to-be-screened energy storage clause and the embedding vector of the text block to obtain the second adaptation attribute of the text block relative to each to-be-screened energy storage clause.

8. The method of claim 1, wherein, The determining the third adaptation attribute of the at least one to-be-displayed energy storage clause associated with the text block based on the first adaptation attribute and the second adaptation attribute of the text block under the same to-be-displayed energy storage clause comprises: obtaining the third adaptation attribute of the text block relative to the same to-be-displayed energy storage clause by weighting the first adaptation attribute and the second adaptation attribute of the text block under the same energy storage clause.

9. The method of claim 1, wherein, The displaying the third adaptation attribute at the preset position of each text block of the to-be-processed text, and displaying the target adaptation attribute of the to-be-displayed energy storage clause at the target position of the to-be-processed text comprises: displaying the clause jump identifier and the corresponding third adaptation attribute of at least part of the to-be-displayed energy storage clause associated with each text block of the to-be-processed text at the preset position of the text block; displaying the clause jump identifier and the target adaptation attribute of all or at least part of the to-be-displayed energy storage clause at the target position of the to-be-processed text; The clause jump identifier is configured to be clicked to display the clause content of the to-be-displayed energy storage clause; the at least part of the to-be-displayed energy storage clause is a to-be-displayed energy storage clause whose target adaptation attribute meets a first preset condition and / or a to-be-displayed energy storage clause whose third adaptation attribute meets a second preset condition.

10. A computer program product, characterised in that, The computer program product comprises a computer program which, when executed by a processor, implements the data processing method according to any one of claims 1-9.

Images