Multi-Transform collaborative chemical reaction mining method and device, processor and storage medium

Abstract

The invention relates to the technical field of natural language processing, and particularly discloses a multi-Transform collaborative chemical reaction mining method and device, a processor and a storage medium, and the method comprises the steps: obtaining chemical reaction data and a pre-trained multi-Transform collaborative model; performing component standardization processing on the chemical reaction data to generate standardized chemical reaction components; carrying out reaction rule analysis on the chemical reaction data on the basis of an atom mapping rule, the chemical reaction components and the multi-Transform collaborative model to generate a component reaction rule; based on the multi-Transform collaborative model, carrying out structured step extraction on a target document to obtain structured reaction steps; and performing information mining on the structured reaction steps based on the component reaction rule to generate a chemical reaction after mining. By training the multi-Transform collaborative model, the chemical reaction in the literature is accurately and comprehensively identified and mined, the actual demand is met, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of natural language processing technology, and in particular to a multi-Transformer collaborative chemical reaction mining method, device, processor, and storage medium. Background Art

[0002] In the fields of chemistry, pharmaceuticals, and materials science, efficiently and accurately extracting chemical reaction information from vast amounts of literature is a fundamental and critical task. During research and experimentation, technicians often rely on extensive references to better inform their research and experiments. Currently, this process is largely manual, requiring technicians to review individual articles, identify key information such as reactants, products, catalysts, and solvents, and manually organize them into databases or analysis tools.

[0003] However, in actual application, manual mining has significant defects: on the one hand, the number of literature is huge and growing rapidly, resulting in heavy workload and low efficiency; on the other hand, the reaction descriptions in the literature often have problems such as vague expressions, inconsistent terminology or missing information (for example, "palladium carbon catalyst" does not indicate the dry and wet state, loading amount or particle size), which can easily lead to human misjudgment or omissions, affecting the reliability of the data.

[0004] In addition, the standardization of chemical reaction data faces severe challenges. The same chemical entity may have multiple names or specifications due to differences in experimental conditions or literature descriptions (for example, "palladium on carbon" includes different forms such as Pd / C (dry) and Pd / C (wet, containing 10% water)). There is also a lack of unified standards for defining the roles of reagents or solvents - some solvents (such as water and alcohols) may participate as reactants and provide atoms, while others only serve as inert media. This heterogeneous data not only increases the difficulty of manual merging and analysis, but also hinders the automated mining of chemical reaction knowledge, database construction, and the training and application of machine learning models. Summary of the Invention

[0005] In order to overcome the above-mentioned technical problems existing in the prior art, the embodiments of the present invention provide a multi-Transformer collaborative chemical reaction mining method, device, processor and storage medium. By training a multi-Transformer collaborative model, chemical reactions in the literature can be accurately and comprehensively identified and mined, meeting actual needs and improving work efficiency.

[0006] In order to achieve the above-mentioned purpose, an embodiment of the present invention provides a multi-Transformer collaborative chemical reaction mining method, which includes: obtaining chemical reaction data and a pre-trained multi-Transformer collaborative model; performing component standardization on the chemical reaction data to generate standardized chemical reaction components; performing reaction rule analysis on the chemical reaction data based on atom mapping rules, the chemical reaction components and the multi-Transformer collaborative model to generate component reaction rules; performing structured step extraction on the target document based on the multi-Transformer collaborative model to obtain structured reaction steps; performing information mining on the structured reaction steps based on the component reaction rules to generate mined chemical reactions.

[0007] Preferably, the method further includes: constructing a chemical entity recognition model based on SciBERT; constructing a chemical reaction recognition model based on a cross-module attention mechanism and a gated feature aggregation mechanism; determining optimal reaction data based on historical reaction data; optimizing the chemical reaction recognition model based on the optimal reaction data to generate an optimized model; constructing a product prediction model; and generating a pre-trained multi-Transformer collaborative model based on the chemical entity recognition model, the optimized model and the product prediction model.

[0008] Preferably, the chemical reaction recognition model is constructed based on the cross-module attention mechanism and the gated feature aggregation mechanism, including: constructing a relationship recognition Transformer based on the cross-attention mechanism; constructing a role classification Transformer based on the gated feature aggregation mechanism and the historical reaction data; obtaining a preset standard terminology library, and constructing a standardization Transformer based on the preset standard terminology library; generating a chemical reaction recognition model based on the relationship recognition Transformer, the role classification Transformer and the standardization Transformer.

[0009] Preferably, the component standardization processing of the chemical reaction data to generate standardized chemical reaction components includes: extracting chemical components from the chemical reaction data; performing standardized expression on the chemical components to generate standardized components; performing component labeling on the standardized components to generate standardized chemical reaction components, and the standardized chemical reaction components include reactants, products, reagents and reaction condition information.

[0010] Preferably, the reaction rule analysis of the chemical reaction data is performed based on the atomic mapping rules, the chemical reaction components and the multi-Transformer collaborative model to generate component reaction rules, including: performing atomic mapping analysis on the reactants and products in the chemical reaction data based on the atomic mapping rules to generate atomic mapping information; performing atomic matching on the atomic mapping information based on the multi-Transformer collaborative model to generate atomic matching information; determining unreacted components that have not been atomically mapped and reacted components that have been atomically mapped from the chemical reaction components based on the atomic matching information; determining the corresponding first component reaction rule based on the physical form of the unreacted component; determining the second component reaction rule of the reacted component based on the atomic matching information; and generating component reaction rules based on the first component reaction rule, the second component reaction rule and the reaction condition information.

[0011] Preferably, the structured step extraction of the target document based on the multi-Transformer collaborative model to obtain the structured reaction steps includes: extracting synthesis actions from the target document based on the multi-Transformer collaborative model; performing structured arrangement on the synthesis actions to generate structured synthesis information; and generating structured reaction steps based on the structured synthesis information.

[0012] Preferably, the information mining of the structured reaction step based on the component reaction rule to generate a post-mining chemical reaction includes: determining whether there is an expression defect in the structured reaction step; if so, obtaining the defect type; when the defect type is missing component, completing the component of the structured reaction step based on the component reaction rule to generate a completed chemical reaction; when the defect type is component mapping error, performing reaction correction on the structured reaction step based on the component reaction rule to generate a corrected chemical reaction; when the defect type is missing reaction conditions, completing the reaction conditions of the structured reaction step based on the component reaction rule to generate a supplemented chemical reaction; generating a post-mining chemical reaction based on the supplemented chemical reaction and / or the corrected chemical reaction and / or the supplemented chemical reaction.

[0013] Correspondingly, the present invention also provides a multi-Transformer collaborative chemical reaction mining device, which includes: an acquisition unit for acquiring chemical reaction data and a pre-trained multi-Transformer collaborative model; a processing unit for performing component standardization on the chemical reaction data to generate standardized chemical reaction components; an analysis unit for performing reaction rule analysis on the chemical reaction data based on atomic mapping rules, the chemical reaction components and the multi-Transformer collaborative model to generate component reaction rules; an extraction unit for performing structured step extraction on the target document based on the multi-Transformer collaborative model to obtain structured reaction steps; and a mining unit for performing information mining on the structured reaction steps based on the component reaction rules to generate mined chemical reactions.

[0014] On the other hand, the present invention further provides a processor for running a program, wherein the program, when run, is used to execute the method provided by an embodiment of the present invention.

[0015] On the other hand, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which implements the method provided by an embodiment of the present invention when the program is executed by a processor.

[0016] The technical solution provided by the present invention has at least the following technical effects:

[0017] By improving the existing literature reading method, pre-training multiple Transformer collaborative models, and analyzing publicly available standard chemical reaction data to generate standard component reaction rules, the multi-Transformer collaborative model and component reaction rules are used to automatically read the literature and automatically identify and mine chemical reaction information, thereby providing technical personnel with fast, accurate and complete chemical reaction information, improving user experience and meeting the actual needs of enterprises.

[0018] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following detailed description, they are used to explain the embodiments of the present invention, but do not constitute a limitation of the embodiments of the present invention. In the accompanying drawings:

[0020] Figure 1 This is a specific implementation flow chart of the multi-Transformer collaborative chemical reaction mining method provided by an embodiment of the present invention;

[0021] Figure 2is a schematic diagram of principle matching of atom mapping information provided by an embodiment of the present invention;

[0022] Figure 3 is a schematic diagram of extracting structured synthetic information from a target document based on a multi-Transformer collaborative model provided by an embodiment of the present invention;

[0023] Figure 4 is a schematic diagram of converting structured synthesis information into a chemical expression provided by an embodiment of the present invention;

[0024] Figure 5 Schematic diagram of chemical expression mining and defect improvement provided by an embodiment of the present invention;

[0025] Figure 6 Schematic diagram of the structure of a multi-Transformer collaborative chemical reaction mining device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following describes the specific implementation of the embodiment of the present invention in detail with reference to the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the embodiment of the present invention and is not used to limit the embodiment of the present invention.

[0027] The terms "system" and "network" in the embodiments of the present invention can be used interchangeably. "Multiple" refers to two or more. In view of this, "multiple" can also be understood as "at least two" in the embodiments of the present invention. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / ", unless otherwise specified, generally indicates that the previous and next associated objects are in an "or" relationship. In addition, it should be understood that in the description of the embodiments of the present invention, words such as "first" and "second" are only used to distinguish the purpose of description, and cannot be understood as indicating or implying relative importance, nor can they be understood as indicating or implying order.

[0028] See Figure 1 , an embodiment of the present invention provides a multi-Transformer collaborative chemical reaction mining method, the method comprising:

[0029] S10: Acquire chemical reaction data and pre-trained multi-Transformer collaborative model;

[0030] S20: performing component standardization processing on the chemical reaction data to generate standardized chemical reaction components;

[0031] S30: performing reaction rule analysis on the chemical reaction data based on the atom mapping rule, the chemical reaction components, and the multi-Transformer collaborative model to generate component reaction rules;

[0032] S40: extracting structured steps from the target document based on the multi-Transformer collaborative model to obtain structured reaction steps;

[0033] S50: performing information mining on the structured reaction steps based on the component reaction rules to generate a post-mining chemical reaction.

[0034] In one possible implementation, chemical reaction data and a pre-trained multi-Transformer collaborative model are first obtained. For example, in order to ensure the comprehensiveness and accuracy of chemical reaction data acquisition, chemical reaction data can be obtained from public compound databases such as SCIFINDER, ACS library, and USPTO, as well as journal article libraries. In recent years, natural language processing (NLP) technology, especially Transformer-based models (such as BERT, GPT, etc.), has demonstrated powerful capabilities in text information extraction tasks. However, a single Transformer model still has limitations when processing chemical reaction text: chemical reaction information involves entity recognition (such as reactants, products), relationship extraction (such as catalysis, solvation), role classification (such as solvents participating in / not participating in the reaction), and other subtasks. It is difficult for a single model to optimize all tasks at the same time, resulting in performance degradation. General pre-trained models (such as BERT) lack chemical domain knowledge and have insufficient semantic understanding of professional terms (such as "palladium carbon" and "DMF"), which can easily lead to misjudgments. Chemical reaction descriptions often contain complex conditions (such as temperature, pressure) and nested entities (such as "10% Pd / C"), and a single model is difficult to effectively capture such structured information.

[0035] In order to solve the above technical problems, in an embodiment of the present invention, the method also includes: constructing a chemical entity recognition model based on SciBERT; constructing a chemical reaction recognition model based on a cross-module attention mechanism and a gated feature aggregation mechanism; determining optimal reaction data based on historical reaction data; optimizing the chemical reaction recognition model based on the optimal reaction data to generate an optimized model; constructing a product prediction model; and generating a pre-trained multi-Transformer collaborative model based on the chemical entity recognition model, the optimized model and the product prediction model.

[0036] In one possible implementation, a chemical entity recognition module (NER Transformer) is first constructed, such as a domain-pre-trained Transformer (such as SciBERT, ChemBERTa), which specifically identifies key chemical entities such as reactants, products, catalysts, and solvents. During the training process, a SMILES string or reaction formula (such as C=O+[H]→CO) is used as input, and reactants, reagents, and conditions are encoded as dense vectors. The model extracts atomic / bond-level change features (such as functional group migration) to output corresponding information. Then, a chemical reaction recognition model is constructed based on a cross-module attention mechanism and a gated feature aggregation mechanism.

[0037] A multi-Transformer modular architecture for chemical reactions is constructed. In an embodiment of the present invention, the chemical reaction recognition model is constructed based on the cross-module attention mechanism and the gated feature aggregation mechanism, including: constructing a relationship recognition Transformer based on the cross-attention mechanism; constructing a role classification Transformer based on the gated feature aggregation mechanism and the historical reaction data; obtaining a preset standard terminology library, and constructing a standardization Transformer based on the preset standard terminology library; generating a chemical reaction recognition model based on the relationship recognition Transformer, the role classification Transformer and the standardization Transformer.

[0038] Specifically, a cross-attention mechanism is used to model the reaction relationship between entities (such as "A+B→C" and "Pd / C catalysis"), that is, a relationship recognition Transformer is constructed. Then, a role classification Transformer is created based on the gated feature aggregation mechanism and historical reaction data. Specifically, the Transformer is trained with data through the gated feature aggregation mechanism and historical reaction data to effectively distinguish the participation of solvents / reagents (such as "water as a reactant" vs. "DMF only as a solvent"). Finally, a standardized Transformer is constructed based on the preset standard terminology library. Specifically, it is trained using the standardized chemical reaction terms in the preset standard terminology library to map unstructured descriptions (such as "palladium carbon") to standard terms (such as "Pd / C (5% loading, dry)").

[0039] After constructing the chemical reaction recognition model, the optimal reaction data is further determined based on the historical reaction data. For example, in an embodiment of the present invention, a reaction condition optimizer can be further constructed, such as a reaction condition generator constructed using a chemical reaction recognition model. During training, the reactants, target products, and historical condition data are input to train the model to output data such as the numerical range of the optimal reagent / catalyst (such as "Pd / C") and reaction conditions (temperature, pH, etc.). The chemical reaction recognition model is optimized through the output content to obtain the optimized model. At this time, a product prediction model is further constructed, such as a hybrid architecture of a graph neural network (GNN) combined with a Transformer to construct a product prediction model, wherein the graph neural network part is used to process the molecular graph structure (such as MPNN), and the Transformer part is used to model long-range dependencies (such as atomic interactions). The atomic importance is dynamically adjusted through the attention mechanism (such as the higher weight of the reaction center atom). During training, the reactant SMILES and reaction rules are input to output the probability distribution of the candidate products (such as Top-k SMILES). Thus, the construction of the multi-Transformer collaborative model is completed.

[0040] For example, in one embodiment, a document records that "benzoic acid reacts with ethanol to form ethyl benzoate under Pd / C catalysis." The chemical entity recognition model recognizes the entities: Pd / C (catalyst), benzoic acid (reactant), ethanol (reactant), and ethyl benzoate (product). The relationship recognition transformer extracts the relationship: benzoic acid + ethanol → ethyl benzoate, Pd / C catalysis. The role classification transformer confirms that ethanol is the reactant (non-solvent). The normalization transformer normalizes "Pd / C" to "Pd / C (10% loading, dry)."

[0041] In an embodiment of the present invention, by training a multi-Transformer collaborative model, it is possible to effectively address the limitations of existing models in text processing in the chemical field, solve various defects in the process of chemical reaction identification in the chemical field, effectively improve the recognition accuracy of chemical reactions, and improve the comprehensiveness, accuracy and reliability of chemical reaction mining.

[0042] After completing model training and data collection, a standard database needs to be prepared in advance to achieve standardization in the subsequent identification and reaction mining of various documents.

[0043] In an embodiment of the present invention, the component standardization processing of the chemical reaction data to generate standardized chemical reaction components includes: extracting chemical components from the chemical reaction data; performing standardized expression on the chemical components to generate standardized components; performing component labeling on the standardized components to generate standardized chemical reaction components, and the standardized chemical reaction components include reactants, products, reagents and reaction condition information.

[0044] In one possible embodiment, chemical components are first extracted from the chemical reaction data, for example, information such as the name, chemical meaning, and type of each chemical component is extracted, and then the chemical components are standardized. Specifically, for chemical components that use non-standardized expressions such as abbreviations or abbreviations, they are standardized according to the IUPAC standard expressions to generate standardized components, and then further component annotations are performed on them, such as annotations of reactants, products, reagents, catalysts, and reaction conditions, to generate standardized reaction components.

[0045] Then, the reaction rules in the above chemical reaction data are further analyzed to generate component reaction rules, and combined with the above standardized chemical reaction components to jointly construct a standard reaction database for subsequent chemical reaction data mining.

[0046] In existing technologies, reaction rules are often extracted through manual analysis by technicians. However, these traditional analysis methods are inefficient, highly subjective, and subject to significant deviations, failing to meet practical needs. Directly extracting rules through natural language recognition can lead to significant recognition deviations due to semantic differences.

[0047] In order to solve the above technical problems, in an embodiment of the present invention, the reaction rule analysis of the chemical reaction data is performed based on the atomic mapping rules, the chemical reaction components and the multi-Transformer collaborative model to generate component reaction rules, including: performing atomic mapping analysis on the reactants and products in the chemical reaction data based on the atomic mapping rules to generate atomic mapping information; performing atomic matching on the atomic mapping information based on the multi-Transformer collaborative model to generate atomic matching information; determining unreacted components that have not been atomically mapped and reacted components that have been atomically mapped from the chemical reaction components based on the atomic matching information; determining the corresponding first component reaction rule based on the physical form of the unreacted component; determining the second component reaction rule of the reacted component based on the atomic matching information; and generating component reaction rules based on the first component reaction rule, the second component reaction rule and the reaction condition information.

[0048] In one possible implementation, to achieve more accurate chemical reaction rule analysis, the principle of principle mapping is employed to extract reaction rules. Specifically, atomic mapping analysis is first performed on the reactants and products in the chemical reaction data based on the atomic mapping rules to generate atomic mapping information. For example, each chemical reaction in the chemical reaction data is converted into a chemical reaction process expressed as a molecular formula. The mapping of each atom in the chemical reaction process is then analyzed using the molecular formula to generate atomic mapping information.

[0049] Then, the atom mapping information is further matched based on the multi-Transformer collaborative model to generate atom matching information. Specifically, the atoms are matched by finding the atoms with the largest attention weight in the self-attention layer to generate atom matching information. For example, see Figure 2 . At this time, the unreacted components and the reacted components are further determined from the chemical reaction components based on the atomic matching information. For example, in this embodiment, a compound without atomic mapping indicates that it does not participate in the reaction, such as a solvent, then it can be identified as an unreacted component, and its first component reaction rule in the chemical reaction process can be determined based on the physical form of the unreacted component. For example, in one embodiment, water as a solution does not participate in the reaction, then it is determined as a solution based on its physical form. At the same time, the second component reaction rule in the reaction process can be determined based on the atomic matching information of the reacted component, and finally the component reaction rule of the entire chemical reaction is generated. Based on this, a standard reaction rule database for a large number of chemical reactions is constructed. In the subsequent chemical reaction mining process, the reaction process can be standardized, comprehensive and rapid matched and improved based on the above-mentioned standard reaction rule database.

[0050] In an embodiment of the present invention, the structured step extraction of the target document based on the multi-Transformer collaborative model to obtain the structured reaction steps includes: extracting synthesis actions from the target document based on the multi-Transformer collaborative model; performing structured arrangement on the synthesis actions to generate structured synthesis information; and generating structured reaction steps based on the structured synthesis information.

[0051] In one possible implementation, an enterprise needs to quickly review and learn the chemical reactions in a document. Manual reading is inefficient, so the multi-Transformer collaborative model provided by the embodiment of the present invention is first used to extract the structured steps of the target document to extract the structured reaction steps. Specifically, the multi-Transformer collaborative model is used to identify the synthesis actions from the target document, while ignoring the contextual connection sentences and reaction effect sentences, thereby improving reading efficiency. The synthesis actions are then structured and arranged, and structured synthesis information is generated. For example, see Figure 3 .

[0052] For example, a chemical reaction process includes the following:

[0053] Reactant preparation: In a round-bottom flask were added 4-iodobenzoic acid (3.0 g, 12.1 mmol) and 100 ml of chloroform.

[0054] Add reagents: Sulfuryl chloride (5.0 mL) dissolved in 10 mL of chloroform and 2-3 drops of dimethylformamide (DMF) were added to the solution.

[0055] Reaction conditions: The mixture was heated under reflux for 2 hours while the reaction was monitored by oil bubbling.

[0056] Result: A clear solution of 4-iodobenzoyl chloride was obtained. After removal of volatile substances, a colorless oil was obtained, which solidified upon cooling.

[0057] After extracting the structured reaction steps, the following information can be obtained:

[0058] 1 Prepare solution: Dissolve 4-iodobenzoic acid (3.0 g, 12.1 mmol) in 100 mL of chloroform.

[0059] 2 Add sulfuryl chloride (SLN).

[0060] 3. Prepare sulfuryl chloride solution: Dissolve sulfuryl chloride (5.0 ml) in 10 ml of chloroform.

[0061] 4 Add sulfuryl chloride (SLN) again.

[0062] 5. Add excess dimethylformamide.

[0063] 6 Reflux for 2 hours.

[0064] 7 to obtain 4-iodobenzoyl chloride.

[0065] 8 Concentrate.

[0066] 9 A colorless oil was obtained.

[0067] Since traditional methods only process text into structured synthetic information and present it to technicians, the information may lack some compounds, which are difficult to find intuitively through text information alone. Therefore, in order to solve this technical problem, after generating structured synthetic information, it is further converted into standard chemical expressions for easy analysis and mining, for example, see Figure 4 .

[0068] Finally, after generating a chemical expression, the structured reaction step is subjected to information mining using the component reaction rules generated above. In an embodiment of the present invention, the information mining of the structured reaction step based on the component reaction rules to generate a post-mining chemical reaction includes: determining whether the structured reaction step has an expression defect; if so, obtaining the defect type; if the defect type is a missing component, completing the component of the structured reaction step based on the component reaction rules to generate a completed chemical reaction; if the defect type is a component mapping error, performing reaction correction on the structured reaction step based on the component reaction rules to generate a corrected chemical reaction; if the defect type is a missing reaction condition, completing the reaction condition of the structured reaction step based on the component reaction rules to generate a supplemented chemical reaction; generating a post-mining chemical reaction based on the supplemented chemical reaction and / or the corrected chemical reaction and / or the supplemented chemical reaction.

[0069] In one possible embodiment, when automatically identifying chemical reactions in the literature, the model can first generate preliminary identification information. Specifically, the preliminary identification information can be a preliminary generated reaction chemical formula or structured reaction steps, and then determine whether the structured reaction steps have expression defects. In an embodiment of the present invention, the expression defects include but are not limited to lack of reactants, lack of products, lack of reagents, inability to distinguish between reagents and solvents, lack of reaction conditions, etc.

[0070] If it is found that the chemical reaction recorded in the current literature has an expression defect, the defect type is immediately obtained and chemical reaction mining is performed. For example, in the first embodiment, it is detected that a chemical reaction has a defect of missing products, for example:

[0071] CC(C)S.CN(C)C=O.Fc1cccnc1F.O=C([O-])[O-].[K+].[K+]>>

[0072] Therefore, the product completion operation is automatically performed on it and the completed chemical reaction is generated. Based on the same mining logic, if other expression defects are found, they are automatically mined and defect improvement operations are performed on them according to the generated component reaction rules or related standard reaction databases, and the mined chemical reaction is generated. For example, see Figure 5 , which is a schematic diagram of the completed chemical reaction.

[0073] In an embodiment of the present invention, by adopting a multi-Transformer collaborative model to accurately identify chemical reactions in the literature and conduct detailed mining, various defects in the chemical reactions recorded in the literature can be effectively repaired, assisting technicians to better and more accurately obtain reference materials in the literature, improving user experience and work efficiency.

[0074] The following describes a multi-Transformer collaborative chemical reaction mining device provided by an embodiment of the present invention with reference to the accompanying drawings.

[0075] See Figure 6 Based on the same inventive concept, an embodiment of the present invention provides a multi-Transformer collaborative chemical reaction mining device, which includes: an acquisition unit for acquiring chemical reaction data and a pre-trained multi-Transformer collaborative model; a processing unit for performing component standardization on the chemical reaction data to generate standardized chemical reaction components; an analysis unit for performing reaction rule analysis on the chemical reaction data based on atom mapping rules, the chemical reaction components and the multi-Transformer collaborative model to generate component reaction rules; an extraction unit for performing structured step extraction on the target document based on the multi-Transformer collaborative model to obtain structured reaction steps; and a mining unit for performing information mining on the structured reaction steps based on the component reaction rules to generate mined chemical reactions.

[0076] Furthermore, an embodiment of the present invention also provides a processor for running a program, wherein the program, when run, is used to execute the method described in the embodiment of the present invention.

[0077] Furthermore, an embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the method described in the embodiment of the present invention when the program is executed by a processor.

[0078] The above describes in detail the optional implementation methods of the embodiments of the present invention in conjunction with the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above implementation methods. Within the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the scope of protection of the embodiments of the present invention.

[0079] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any appropriate manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not further describe various possible combinations.

[0080] Those skilled in the art will understand that all or part of the steps in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a program, which is stored in a storage medium and includes a number of instructions for causing a single-chip microcomputer, chip or processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program code.

[0081] In addition, various implementations of the embodiments of the present invention may be arbitrarily combined, and as long as they do not violate the concept of the embodiments of the present invention, they should also be regarded as the contents disclosed in the embodiments of the present invention.

Claims

1. A multi-Transformer collaborative chemical reaction mining method, characterized by: The method comprises: Obtain chemical reaction data and pre-trained multi-Transformer collaborative models; performing component standardization processing on the chemical reaction data to generate standardized chemical reaction components; Performing reaction rule analysis on the chemical reaction data based on atom mapping rules, the chemical reaction components, and the multi-Transformer collaborative model to generate component reaction rules; Extracting structured steps from the target document based on the multi-Transformer collaborative model to obtain structured reaction steps; Information mining is performed on the structured reaction steps based on the component reaction rules to generate a post-mining chemical reaction.

2. The method according to claim 1, characterized in that The method further comprises: Build a chemical entity recognition model based on SciBERT; Construct a chemical reaction recognition model based on cross-module attention mechanism and gated feature aggregation mechanism; Determine optimal response data based on historical response data; Optimizing the chemical reaction identification model based on the optimal reaction data to generate an optimized model; Build product prediction models; A pre-trained multi-Transformer collaborative model is generated based on the chemical entity recognition model, the optimized model and the product prediction model.

3. The method according to claim 2, characterized in that The chemical reaction recognition model is constructed based on the cross-module attention mechanism and the gated feature aggregation mechanism, including: Constructing a relation recognition Transformer based on the cross-attention mechanism; Constructing a role classification Transformer based on the gated feature aggregation mechanism and the historical response data; Obtaining a preset standard terminology library, and building a standardized Transformer based on the preset standard terminology library; A chemical reaction recognition model is generated based on the relationship recognition Transformer, the role classification Transformer, and the standardization Transformer.

4. The method according to claim 1, wherein The step of performing component standardization on the chemical reaction data to generate standardized chemical reaction components includes: extracting chemical components from the chemical reaction data; Performing standardized expression on the chemical components to generate standardized components; The standardized components are labeled to generate standardized chemical reaction components, which include reactants, products, reagents and reaction condition information.

5. The method according to claim 4, characterized in that The performing reaction rule analysis on the chemical reaction data based on the atom mapping rule, the chemical reaction components, and the multi-Transformer collaborative model to generate component reaction rules includes: performing atomic mapping analysis on reactants and products in the chemical reaction data based on atomic mapping rules to generate atomic mapping information; Performing atomic matching on the atomic mapping information based on the multi-Transformer collaborative model to generate atomic matching information; Determining, from the chemical reaction components, unreacted components that have not been atomically mapped and reacted components that have been atomically mapped based on the atomic matching information; Determining a corresponding first component reaction rule based on the physical form of the unreacted component; determining a second component reaction rule for the reacted component based on the atom matching information; A component reaction rule is generated based on the first component reaction rule, the second component reaction rule, and the reaction condition information.

6. The method according to claim 1, characterized in that The step of extracting structured steps from the target document based on the multi-Transformer collaborative model to obtain structured reaction steps includes: Extracting synthetic actions from the target document based on the multi-Transformer collaborative model; Arranging the synthesis actions in a structured manner to generate structured synthesis information; A structured reaction step is generated based on the structured synthesis information.

7. The method according to claim 1, characterized in that The performing information mining on the structured reaction steps based on the component reaction rules to generate a post-mining chemical reaction includes: Determining whether there is an expression defect in the structured reaction step; If yes, get the defect type; When the defect type is a lack of component, completing the components of the structured reaction step based on the component reaction rule to generate a completed chemical reaction; When the defect type is a component mapping error, performing reaction correction on the structured reaction step based on the component reaction rule to generate a corrected chemical reaction; When the defect type is missing reaction conditions, supplementing the reaction conditions of the structured reaction step based on the component reaction rule to generate a supplemented chemical reaction; A post-mining chemical reaction is generated based on the post-completion chemical reaction and / or the post-correction chemical reaction and / or the post-supplementation chemical reaction.

8. A multi-Transformer collaborative chemical reaction mining device, characterized by: The device comprises: Acquisition unit, used to obtain chemical reaction data and pre-trained multi-Transformer collaborative model; a processing unit, configured to perform component standardization processing on the chemical reaction data to generate standardized chemical reaction components; an analysis unit, configured to perform reaction rule analysis on the chemical reaction data based on atom mapping rules, the chemical reaction components, and the multi-Transformer collaborative model to generate component reaction rules; An extraction unit, configured to extract structured steps from the target document based on the multi-Transformer collaborative model to obtain structured reaction steps; A mining unit is used to perform information mining on the structured reaction steps based on the component reaction rules to generate a post-mining chemical reaction.

9. A processor, characterized in that: Used to run a program, wherein the program is used to execute the method according to any one of claims 1 to 7 when run.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images