Artificial intelligence-based business english writing structure automatic planning method

Abstract

This invention relates to the field of intelligent semantic processing, specifically to an artificial intelligence-based automatic planning method for business English writing structures. The method includes acquiring business English writing task information, hierarchical writing structure modeling, global business English context fusion, conditional business English structure generation, and outputting a business English writing structure planning scheme. This invention jointly models chapters, paragraphs, and functional units in the writing structure with contextual information, achieving deep integration of hierarchical structure and context, thereby ensuring that the generated writing structure can adapt to the needs of different audiences and communication scenarios. This invention introduces a global optimization mechanism and a conditional structure generation method, enhancing the logical coherence, information integrity, and pragmatic consistency of structural units. It constructs a conditional structure query vector, and, combined with business writing goals and contextual constraints, conditionally guides the generation of the globally optimized structure, forming multiple candidate business English writing structure schemes.

Description

Technical Field

[0001] This invention relates to the field of intelligent semantic processing, specifically to an automatic planning method for the structure of business English writing based on artificial intelligence. Background Technology

[0002] With the increasing frequency of global business exchanges, the demand for business English writing is growing year by year. However, existing technologies have the following main problems: (1) Lack of joint modeling ability for business context and writing objectives. Traditional business English writing assistance systems are mostly based on templates or fixed rules, which cannot fully consider the differences in different audiences, communication scenarios and pragmatic constraints, resulting in a lack of pertinence and flexibility in the generated writing structure.

[0003] (2) Lacking structural hierarchy optimization and conditional generation capabilities, existing methods often directly generate text or paragraphs, lacking hierarchical analysis and global optimization of chapters, paragraphs and functional units, and are unable to effectively screen generation schemes based on writing goals and contextual conditions, which easily leads to writing structures with incomplete information or pragmatic inconsistencies. Summary of the Invention

[0004] To address the aforementioned issues and overcome the shortcomings of existing technologies, this invention provides an AI-based automatic planning method for business English writing structures. Addressing the lack of joint modeling capabilities for business context and writing objectives, this invention acquires business English writing task information and performs semantic understanding to construct a business English context feature representation. Based on this context feature, this invention jointly models the chapters, paragraphs, and functional units in the writing structure with contextual information, achieving a deep integration of hierarchical structure and context. This ensures that the generated writing structure can adapt to the needs of different audiences and communication scenarios. Addressing the lack of structural hierarchy optimization and conditional generation capabilities, this invention introduces a global optimization mechanism and a conditional structure generation method. It performs global contextual propagation and optimization adjustments on the initial hierarchical writing structure, enhancing the logical coherence, information completeness, and pragmatic consistency of structural units. A conditional structure query vector is constructed, and combined with business writing objectives and contextual constraints, conditionally guides the generation of the globally optimized structure, forming multiple candidate business English writing structure schemes. A structure-aware evaluation method is used to screen the candidate schemes, selecting the most suitable structure for the current task, ensuring that the writing structure not only meets the task objectives but also considers audience characteristics and contextual constraints.

[0005] The technical solution adopted by this invention is as follows: The automatic planning method for business English writing structure based on artificial intelligence provided by this invention includes the following steps: Step S1: Obtain business English writing task information, receive business English writing task input information, including writing purpose, audience, scenario and content constraints, perform semantic understanding on business English writing task input information, and obtain English context features; Step S2: Hierarchical writing structure modeling. Based on the characteristics of the English context, the writing task is divided into hierarchical structural units, including chapters, paragraphs and functional units. The logical dependencies between the hierarchical structural units are established to obtain a preliminary hierarchical structural framework. Step S3: Global Business English Context Integration. The preliminary hierarchical structure and English context features are jointly modeled, and the hierarchical structure units are optimized to obtain a globally optimized hierarchical structure framework. Step S4: Conditional Business English Structure Generation. Based on the globally optimized hierarchical structure framework, specific writing structure schemes are generated under the constraints of business writing goals and context. The generated schemes are evaluated using condition-guided structure generation and structure perception evaluation methods to obtain candidate business English writing structure schemes. Step S5: Output the business English writing structure planning scheme, select the optimal structure based on the candidate business English writing structure schemes, and obtain the final business English writing structure planning scheme.

[0006] Furthermore, step S3 specifically includes the following steps: Step S31: Construction of Contextual Feature Vectors. Based on English contextual features, the writing purpose, audience type, communication scenario, and pragmatic constraints in the business English writing task are analyzed from multiple dimensions. The contextual elements of each dimension are then vectorized to construct a set of business English contextual feature vectors, represented as follows: ; in, Indicates the first Business English contextual feature vectors Represents the set of contextual feature vectors in business English; Step S32: Hierarchical Writing Structure. Based on the preliminary hierarchical structural framework, the chapter units, paragraph units, and functional units are abstracted into sets of structural units. For each structural unit, a set of business English contextual feature vectors is introduced to model the matching relationship between structural units and contextual constraints, generating a structure-context fusion representation. The calculated relationship is as follows: ; in, Indicates the first The structure-context fusion representation of individual structural units within the constraints of business English context. This is a mapping function that integrates structure and context. Represents a set of structural units; Step S33: Global Business English Contextual Dissemination. Based on the structure-context fusion representation, this step combines the dissemination and integration of business English contextual information between different structural levels within the hierarchical writing structure, resulting in a globally context-enhanced structural representation, as shown below: ; in, This represents the structure after global context enhancement. Indicates and A set of adjacent structural units that have dependencies. Indicates the contextual propagation weight between structural units. Indicates the first The structure-context fusion representation of individual structural units under the constraints of business English context; Step S34: Business English structure optimization and adjustment. Based on the structure representation enhanced by the global context, the structural units at each level are tested from the dimensions of pragmatic consistency, information integrity and politeness of expression in business English. When a structural unit is detected to be mismatched with the overall business writing goal or contextual constraints, the structure-context fusion representation of the corresponding structural unit is adjusted to obtain the optimized intermediate structure representation. Step S35: Output the globally optimized hierarchical business English structure representation, and integrate the optimized intermediate structure representation to generate the globally optimized hierarchical business English writing structure representation.

[0007] Furthermore, step S4 specifically includes the following steps: Step S41: Constructing a conditional structure query vector. Based on the globally optimized hierarchical business English structure representation, and combined with the business English writing objectives and English context features, a conditional structure query vector is constructed to guide structure generation, characterizing the overall constraint requirements of the current business English writing task on the writing structure. The representation is as follows: ; in, This represents the hierarchical structure of business English after global optimization. This represents the set of contextual feature vectors in business English. This is a conditional structured query mapping function. Represents a conditional structured query vector; Step S42: Generation of Business English Writing Structure Candidates. Based on the conditional structure query vector, the globally optimized hierarchical structure representation is conditionally guided to generate multiple candidate business English writing structure schemes under the conditions of business writing objectives, audience characteristics, and pragmatic constraints, forming a candidate structure set, as shown below: ; in, Indicates the first One candidate business English writing structure scheme For the candidate structure set; Step S43: Structure-aware conditional similarity evaluation. For each candidate business English writing structure, under the constraint of the conditional structure query vector, calculate the degree of structural matching with the business English writing target to obtain the structure-aware conditional similarity score. The calculation method is as follows: ; in, Indicates the candidate structure scheme The structural semantic representation function This represents the function for calculating structural semantic similarity. For normalized mapping functions, This indicates the overall suitability score of the candidate business English writing structure scheme under the current context and writing objectives; Step S44: Screening of candidate business English writing structures. Based on the conditional similarity score of structure perception, the candidate structure schemes are sorted and screened, and several structure schemes that meet the preset threshold or have the best ranking are selected as the set of candidate business English writing structure schemes.

[0008] The beneficial effects achieved by the present invention using the above solution are as follows: (1) In view of the lack of joint modeling ability of business context and writing goal, this invention obtains business English writing task information, performs semantic understanding on it, and constructs business English context feature representation. Based on the context feature, this invention jointly models the chapters, paragraphs and functional units in the writing structure with context information to achieve deep integration of hierarchical structure and context, thereby ensuring that the generated writing structure can adapt to the needs of different audiences and communication scenarios. (2) To address the lack of structural hierarchy optimization and conditional generation capabilities, this invention introduces a global optimization mechanism and a conditional structure generation method to conduct global contextual propagation and optimization adjustment of the preliminary hierarchical writing structure, thereby enhancing the logical coherence, information integrity and pragmatic consistency of the structural units. A conditional structure query vector is constructed, and combined with business writing objectives and contextual constraints, the globally optimized structure is conditionally guided to generate multiple candidate business English writing structure schemes. The structure-aware evaluation method is used to screen the candidate schemes and select the most suitable structural scheme for the current task, ensuring that the writing structure not only meets the task objectives but also takes into account the characteristics of the audience and contextual constraints. Attached Figure Description

[0009] Figure 1This is a flowchart illustrating the AI-based automatic structure planning method for business English writing provided by the present invention.

[0010] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0011] 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. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0012] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0013] Example 1, see Figure 1 The present invention provides an artificial intelligence-based method for automatically planning the structure of business English writing, which includes the following steps: Step S1: Obtain business English writing task information, receive business English writing task input information, including writing purpose, audience, scenario and content constraints, perform semantic understanding on business English writing task input information, and obtain English context features; Step S2: Hierarchical writing structure modeling. Based on the characteristics of the English context, the writing task is divided into hierarchical structural units, including chapters, paragraphs and functional units. The logical dependencies between the hierarchical structural units are established to obtain a preliminary hierarchical structural framework. Step S3: Global Business English Context Integration. The preliminary hierarchical structure and English context features are jointly modeled, and the hierarchical structure units are optimized to obtain a globally optimized hierarchical structure framework. Step S4: Conditional Business English Structure Generation. Based on the globally optimized hierarchical structure framework, specific writing structure schemes are generated under the constraints of business writing goals and context. The generated schemes are evaluated using condition-guided structure generation and structure perception evaluation methods to obtain candidate business English writing structure schemes. Step S5: Output the business English writing structure planning scheme, select the optimal structure based on the candidate business English writing structure schemes, and obtain the final business English writing structure planning scheme.

[0014] In this embodiment, the scenario is about drafting customer contract communication emails for a multinational company, and the specific steps include: Step S1: Obtain Business English Writing Task Information. The input information for the Business English Writing Task received by the system includes: (1) Purpose of writing: To confirm the contract terms with potential clients; (2) Target audience: Purchasing managers of medium-sized enterprises in Europe; (3) Scenario: Formal business email communication; (4) Content constraints: It must include a summary of the contract terms, payment terms and delivery time, and be written in a polite and professional tone; The system uses a natural language processing model to perform semantic understanding of the input information and extract key contextual features, including writing purpose vectors. Audience type vector Scene vectors and pragmatic constraint vector Integrate into a set of contextual features in business English ; Step S2: Hierarchical writing structure modeling. Based on the English context feature vector set, the email writing task is divided into hierarchical structural units: (1) Section Unit: The entire email; (2) Paragraph units: opening greeting paragraph, contract terms explanation paragraph, payment and delivery explanation paragraph, closing thanks paragraph; (3) Functional units: greeting, explaining terms, emphasizing payment terms, confirming delivery time, and polite closing; The system establishes logical dependencies between structural units, such as requiring the greeting paragraph to be at the beginning and the closing polite paragraph to be at the end. This results in a preliminary hierarchical structural framework. ; Step S3: Integrating the overall business English context, based on the initial structural framework. Based on this, each structural unit is associated with a set of contextual feature vectors. Joint modeling to generate structure-context fusion representations Global context propagation is performed on each structural unit, and the structural representation after global context enhancement is calculated: ; in, Representation and structural unit A set of adjacent units that have logical dependencies. This approach represents the contextual propagation weights between structural units. By detecting pragmatic consistency, information completeness, and politeness, it optimizes and adjusts the fusion representation of certain paragraphs, generating a globally optimized hierarchical business English structural representation. ; Step S4: Conditional Business English Structure Generation Based on the globally optimized structural representation and context feature vector set The system constructs conditional structured query vectors: ; Given the constraints of writing objectives and context, multiple candidate business English writing structures are generated. ; The candidate solutions differ in the order of payment terms, the length of the contract terms summary, and the level of politeness. A structure-aware conditional similarity calculation function is then used to score the candidate solutions. ; Select the solution with the highest score as the final candidate; Step S5: Output the business English writing structure plan. Based on the candidate plan scoring results, select the optimal structure and generate the final business English writing structure plan, including... (1) Email opening: Formal greeting; (2) Contract terms summary section: Highlighting key terms; (3) Payment and Delivery Instructions: Clearly state the terms; (4) Email closing: polite thanks.

[0015] Example 2, based on the above example, step S3 specifically includes the following steps: Step S31: Construction of Contextual Feature Vectors. Based on English contextual features, the writing purpose, audience type, communication scenario, and pragmatic constraints in the business English writing task are analyzed from multiple dimensions. The contextual elements of each dimension are then vectorized to construct a set of business English contextual feature vectors, represented as follows: ; in, Indicates the first Business English contextual feature vectors Represents the set of contextual feature vectors in business English; Step S32: Hierarchical Writing Structure. Based on the preliminary hierarchical structural framework, the chapter units, paragraph units, and functional units are abstracted into sets of structural units. For each structural unit, a set of business English contextual feature vectors is introduced to model the matching relationship between structural units and contextual constraints, generating a structure-context fusion representation. The calculated relationship is as follows: ; in, Indicates the first The structure-context fusion representation of individual structural units within the constraints of business English context. This is a mapping function that integrates structure and context. Represents a set of structural units; Step S33: Global Business English Contextual Dissemination. Based on the structure-context fusion representation, this step combines the dissemination and integration of business English contextual information between different structural levels within the hierarchical writing structure, resulting in a globally context-enhanced structural representation, as shown below: ; in, This represents the structure after global context enhancement. Indicates and A set of adjacent structural units that have dependencies. Indicates the contextual propagation weight between structural units. Indicates the first The structure-context fusion representation of individual structural units under the constraints of business English context; Step S34: Business English structure optimization and adjustment. Based on the structure representation enhanced by the global context, the structural units at each level are tested from the dimensions of pragmatic consistency, information integrity and politeness of expression in business English. When a structural unit is detected to be mismatched with the overall business writing goal or contextual constraints, the structure-context fusion representation of the corresponding structural unit is adjusted to obtain the optimized intermediate structure representation. Step S35: Output the globally optimized hierarchical business English structure representation, and integrate the optimized intermediate structure representation to generate the globally optimized hierarchical business English writing structure representation.

[0016] In this embodiment, the core code used is as follows: import numpy as np # ============================== # S31 Contextual Feature Vector Construction # ============================== def build_context_vector(purpose, audience, scene, constraint): """ Encode contextual elements into vectors """ c_purpose = np.array(purpose) c_audience = np.array(audience) c_scene = np.array(scene) c_constraint = np.array(constraint) C = np.concatenate([c_purpose, c_audience, c_scene, c_constraint]) return C # ============================== # S32 Structure-Context Fusion # ============================== def fusion_function(structure_vector, context_vector): """ Structure and context fusion mapping function F_j = f(s_j, C) """ # Simple linear fusion + nonlinear activation fusion = np.tanh(structure_vector + context_vector[:len(structure_vector)]) return fusion # ============================== # S33 Global Contextual Communication # ============================== def global_propagation(F_list, adjacency_matrix): """ G_j = Σ w_jk F_k adjacency_matrix represents the structural dependency weights. """ F_matrix = np.array(F_list) G_matrix = adjacency_matrix@F_matrix return G_matrix # ============================== # S34 Structural Optimization and Adjustment # ============================== def optimize_structure(G_matrix, threshold=0.2): """ Detecting pragmatic consistency and information integrity If the vector deviation of a certain structural unit is too large, a smoothing adjustment is performed. """ optimized = G_matrix.copy() mean_vector = np.mean(G_matrix, axis=0) for i in range(len(G_matrix)): deviation = np.linalg.norm(G_matrix[i] - mean_vector) if deviation > threshold: # Adjust towards the global mean optimized[i] = 0.7 G_matrix[i] + 0.3 mean_vector return optimized # ============================== # S35 Output Global Optimization Structure # ============================== def generate_global_structure(optimized_matrix): """ Output the final structure representation """ return optimized_matrix # ============================== # Example Run # ============================== if __name__ == "__main__": # Example context input (can be replaced according to the actual scenario) purpose = [0.8, 0.2] audience = [0.7, 0.3] scene = [0.9, 0.1] constraint = [0.85, 0.15] C = build_context_vector(purpose, audience, scene, constraint) # Assume there are 4 structural units structure_units = [ np.array([0.6, 0.4]), np.array([0.5, 0.5]), np.array([0.4, 0.6]), np.array([0.7, 0.3]) ] # Structure-Context Fusion F_list = [fusion_function(s, C) for s in structure_units] # Construct the structure dependency weight matrix (adjacency matrix) adjacency_matrix = np.array([ [0.5, 0.3, 0.1, 0.1], [0.2, 0.5, 0.2, 0.1], [0.1, 0.3, 0.5, 0.1], [0.1, 0.2, 0.2, 0.5] ]) # Global Propagation G_matrix = global_propagation(F_list, adjacency_matrix) # Optimization and Adjustment optimized_matrix = optimize_structure(G_matrix) # Output the final structure final_structure = generate_global_structure(optimized_matrix) print("The globally optimized structure representation:") print(final_structure).

[0017] Example 3, based on the above examples, specifically includes the following steps in step S4: Step S41: Constructing a conditional structure query vector. Based on the globally optimized hierarchical business English structure representation, and combined with the business English writing objectives and English context features, a conditional structure query vector is constructed to guide structure generation, characterizing the overall constraint requirements of the current business English writing task on the writing structure. The representation is as follows: ; in, This represents the hierarchical structure of business English after global optimization. This represents the set of contextual feature vectors in business English. This is a conditional structured query mapping function. Represents a conditional structured query vector; Step S42: Generation of Business English Writing Structure Candidates. Based on the conditional structure query vector, the globally optimized hierarchical structure representation is conditionally guided to generate multiple candidate business English writing structure schemes under the conditions of business writing objectives, audience characteristics, and pragmatic constraints, forming a candidate structure set, as shown below: ; in, Indicates the first One candidate business English writing structure scheme For the candidate structure set; Step S43: Structure-aware conditional similarity evaluation. For each candidate business English writing structure, under the constraint of the conditional structure query vector, calculate the degree of structural matching with the business English writing target to obtain the structure-aware conditional similarity score. The calculation method is as follows: ; in, Indicates the candidate structure scheme The structural semantic representation function This represents the function for calculating structural semantic similarity. For normalized mapping functions, This indicates the overall suitability score of the candidate business English writing structure scheme under the current context and writing objectives; Step S44: Screening of candidate business English writing structures. Based on the conditional similarity score of structure perception, the candidate structure schemes are sorted and screened, and several structure schemes that meet the preset threshold or have the best ranking are selected as the set of candidate business English writing structure schemes.

[0018] In this embodiment, the core code used is as follows: import numpy as np # =========================== # S41 Conditional Structured Query Vector Construction # =========================== def build_query_vector(G, C): """ Q = g(G, C) G: Globally optimized structure representation (n_units, dim) C: Contextual feature vector (dim,) """ # 1. Structural Hierarchical Attention Convergence attention_weights = np.softmax(np.mean(G, axis=1)) global_structure = np.sum(G attention_weights[:, None], axis=0) # 2. Context Mapping (Dimension Alignment) C_mapped = C[:len(global_structure)] # 3. Adaptive Fusion alpha = 0.6 Q = alpha global_structure + (1 - alpha) C_mapped return Q # NumPy doesn't have a softmax function, so implement your own. def softmax(x): e_x = np.exp(x - np.max(x)) return e_x / e_x.sum() np.softmax = softmax # =========================== # S42 Multipath Structure Candidate Generation # =========================== def generate_candidates(G, Q, num_candidates=5): """ Multiple candidate structures are generated based on the conditional query vector. """ candidates = [] for _ in range(num_candidates): # Introduce perturbation factors (to increase structural diversity) noise = np.random.normal(0, 0.05, G.shape) # Conditional generation guided_structure = G + noise + 0.3 Q candidates.append(guided_structure) return candidates # =========================== # Structural semantic encoding function # =========================== def structure_encoder(S_i): """ h(S_i) structural semantic representation function """ # Using global average pooling as the structural semantic representation return np.mean(S_i, axis=0) # =========================== # S43 Conditional Similarity Score # =========================== def similarity(vec1, vec2): """ Cosine similarity """ dot = np.dot(vec1, vec2) norm = np.linalg.norm(vec1) np.linalg.norm(vec2) return dot / (norm + 1e-8) def score_candidate(S_i, Q): """ Score(S_i) = σ(sim(h(S_i), Q)) Add structural penalty terms """ encoded = structure_encoder(S_i) sim_score = similarity(encoded, Q) # Penalty: If the internal variance of the structure is too large, it is considered unstable. penalty = np.var(S_i) raw_score = sim_score - 0.1 enalty # Normalization normalized = 1 / (1 + np.exp(-raw_score)) return normalized # =========================== # S44 Candidate Structure Screening # =========================== def select_top_k(candidates, Q, k=2): scored = [] for S_i in candidates: score = score_candidate(S_i, Q) scored.append((S_i, score)) # Sort by rating scored.sort(key=lambda x: x[1], reverse=True) top_structures = [item[0] for item in scored[:k]] return top_structures # =========================== # Example Run # =========================== if __name__ == "__main__": # Simulated Input G = np.random.rand(4, 8) # 4 structural units, 8-dimensional representation C = np.random.rand(8) # 8-dimensional context vector # S41 Q = build_query_vector(G, C) # S42 candidates = generate_candidates(G, Q, num_candidates=6) # S43 + S44 best_structures = select_top_k(candidates, Q, k=2) print("Number of candidate structures after filtering:", len(best_structures)).

[0019] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0020] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0021] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. An AI-based method for automatically planning the structure of business English writing, characterized by: The present invention provides an artificial intelligence-based method for automatically planning the structure of business English writing, which includes the following steps: Step S1: Obtain business English writing task information, receive business English writing task input information, perform semantic understanding on business English writing task input information, and obtain English context features; Step S2: Hierarchical writing structure modeling. Based on the characteristics of the English context, the writing task is divided into hierarchical structural units, and the logical dependencies between the hierarchical structural units are established to obtain a preliminary hierarchical structural framework. Step S3: Global Business English Context Integration. The preliminary hierarchical structure and English context features are jointly modeled, and the hierarchical structure units are optimized to obtain a globally optimized hierarchical structure framework. Step S4: Conditional Business English Structure Generation. Based on the globally optimized hierarchical structure framework, specific writing structure schemes are generated under the constraints of business writing goals and context. The generated schemes are evaluated using condition-guided structure generation and structure perception evaluation methods to obtain candidate business English writing structure schemes. Step S5: Output the business English writing structure planning scheme, select the optimal structure based on the candidate business English writing structure schemes, and obtain the final business English writing structure planning scheme.

2. The method for automatic planning of business English writing structure based on artificial intelligence according to claim 1, characterized in that: Step S3 specifically includes the following steps: Step S31: Construction of Contextual Feature Vectors. Based on English contextual features, the writing purpose, audience type, communication scenario, and pragmatic constraints in the business English writing task are analyzed from multiple dimensions. The contextual elements of each dimension are then vectorized to construct a set of business English contextual feature vectors, represented as follows: ； in, Indicates the first Business English contextual feature vectors Represents the set of contextual feature vectors in business English; Step S32: Hierarchical Writing Structure. Based on the preliminary hierarchical structural framework, the chapter units, paragraph units, and functional units are abstracted into sets of structural units. For each structural unit, a set of business English contextual feature vectors is introduced to model the matching relationship between structural units and contextual constraints, generating a structure-context fusion representation. The calculated relationship is as follows: ； in, Indicates the first The structure-context fusion representation of individual structural units within the constraints of business English context. This is a mapping function that integrates structure and context. Represents a set of structural units; Step S33: Global Business English Contextual Dissemination. Based on the structure-context fusion representation, this step combines the dissemination and integration of business English contextual information between different structural levels within the hierarchical writing structure, resulting in a globally context-enhanced structural representation, as shown below: ； in, This represents the structure after global context enhancement. Indicates and A set of adjacent structural units that have dependencies. Indicates the contextual propagation weight between structural units. Indicates the first The structure-context fusion representation of individual structural units under the constraints of business English context; Step S34: Business English structure optimization and adjustment. Based on the structure representation enhanced by the global context, the structural units at each level are tested from the dimensions of pragmatic consistency, information integrity and politeness of expression in business English. When a structural unit is detected to be mismatched with the overall business writing goal or contextual constraints, the structure-context fusion representation of the corresponding structural unit is adjusted to obtain the optimized intermediate structure representation. Step S35: Output the globally optimized hierarchical business English structure representation, and integrate the optimized intermediate structure representation to generate the globally optimized hierarchical business English writing structure representation.

3. The method for automatic planning of business English writing structure based on artificial intelligence according to claim 1, characterized in that: Step S4 specifically includes the following steps: Step S41: Constructing a conditional structure query vector. Based on the globally optimized hierarchical business English structure representation, and combined with the business English writing objectives and English context features, a conditional structure query vector is constructed to guide structure generation, characterizing the overall constraint requirements of the current business English writing task on the writing structure. The representation is as follows: ； in, This represents the hierarchical structure of business English after global optimization. This represents the set of contextual feature vectors in business English. This is a conditional structured query mapping function. Represents a conditional structured query vector; Step S42: Generation of Business English Writing Structure Candidates. Based on the conditional structure query vector, the globally optimized hierarchical structure representation is conditionally guided to generate multiple candidate business English writing structure schemes under the conditions of business writing objectives, audience characteristics, and pragmatic constraints, forming a candidate structure set, as shown below: ； in, Indicates the first One candidate business English writing structure scheme For the candidate structure set; Step S43: Structure-aware conditional similarity evaluation. For each candidate business English writing structure, under the constraint of the conditional structure query vector, calculate the degree of structural matching with the business English writing target to obtain the structure-aware conditional similarity score. The calculation method is as follows: ； in, Indicates the candidate structure scheme The structural semantic representation function This represents the function for calculating structural semantic similarity. For normalized mapping functions, This indicates the overall suitability score of the candidate business English writing structure scheme under the current context and writing objectives; Step S44: Screening of candidate business English writing structures. Based on the conditional similarity score of structure perception, the candidate structure schemes are sorted and screened, and several structure schemes that meet the preset threshold or have the best ranking are selected as the set of candidate business English writing structure schemes.

Images