A Method and System for Generating Engineering Management Reports Based on Natural Language Processing

Abstract

This invention relates to the field of text processing technology, and provides a method and system for generating engineering management reports based on natural language processing. This system enables in-depth analysis of construction logs, providing comprehensive and accurate decision-making support for engineering management. The method includes: acquiring a set of construction log texts generated during the engineering construction process; performing natural language processing on the construction log text set to extract key entities and progress events from each construction stage description unit, obtaining a construction feature set containing entity identification information and event sequence information; semantically aligning the construction feature set with a pre-set engineering specification knowledge base to generate specification association information between the construction stage description units and engineering specification requirements; and structurally integrating the specification association information based on a hierarchical template matching algorithm to generate a structured report text that meets engineering management requirements.

Description

Technical Field

[0001] This invention belongs to the field of text processing technology, specifically relating to a method and system for generating engineering management reports based on natural language processing. Background Technology

[0002] In the field of engineering management, the processing of construction logs often remains at the level of simple text recording and manual review. Traditionally, construction logs suffer from inconsistent formats, complex content, and a lack of systematic organization, making it difficult to quickly and accurately extract key information. Analyzing the compliance of construction progress with engineering specifications often relies on manual comparison based on experience, which is inefficient and prone to oversights. Furthermore, the generated reports are mostly fragmented and unstructured texts, lacking comprehensive integration of construction progress, specification compliance, and targeted recommendations, thus failing to provide effective decision support for engineering management. Therefore, existing technologies have failed to fully mine and analyze construction logs in depth, leading to information gaps and a lack of scientific decision-making in engineering management, making it difficult to ensure the efficient progress and standardized implementation of engineering projects. Summary of the Invention

[0003] This invention provides a method and system for generating engineering management reports based on natural language processing, which enables in-depth analysis of construction logs and provides comprehensive and accurate decision-making basis for engineering management.

[0004] In a first aspect, embodiments of the present invention provide a method for generating engineering management reports based on natural language processing, applied to an engineering management report generation system, the method comprising:

[0005] Obtain a set of construction log texts generated during the construction process. The set of construction log texts contains multiple continuously recorded construction stage description units. Each construction stage description unit includes at least one construction activity description statement and a corresponding site status description statement.

[0006] Natural language processing is performed on the construction log text set to extract key entities and progress events in each construction stage description unit, resulting in a construction feature set containing entity identification information and event timing information.

[0007] The construction feature set is semantically aligned with a pre-set engineering specification knowledge base to generate specification association information between the construction stage description unit and the engineering specification requirements.

[0008] The hierarchical template matching algorithm is used to perform structured integration of the specification-related information to generate a structured report text that meets the requirements of engineering management. The structured report text includes a description of the construction progress, specification compliance analysis, and recommendations.

[0009] Secondly, embodiments of the present invention provide an engineering management report generation system, which includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the above-described method.

[0010] Thirdly, embodiments of the present invention provide a computer-readable storage medium including a computer program, which, when run on an engineering management report generation system, causes the engineering management report generation system to perform the steps of the above-described method.

[0011] This invention, through obtaining a set of construction log texts containing detailed descriptions of construction stages and combining it with natural language processing technology for text mining, can accurately extract key entities and progress events to form a construction feature set. This breaks through the limitations of traditional manual analysis and greatly improves the accuracy and efficiency of information extraction. Semantic alignment of the construction feature set with the engineering specification knowledge base achieves a deep correlation between actual construction and specification requirements, enabling timely detection of potential specification non-compliance issues. Based on a hierarchical template matching algorithm, the structured integration generates a comprehensive structured report text covering construction progress, specification compliance, and recommendations. This provides a systematic, clear, and targeted decision-making basis for project management, enhancing the scientific, standardized, and forward-looking nature of project management and effectively promoting the efficient and compliant development of engineering construction projects. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating a method for generating engineering management reports based on natural language processing, as provided in an embodiment of the present invention.

[0013] Figure 2 This is a schematic diagram of the structure of an engineering management report generation system provided in an embodiment of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 technical solutions of the present invention, and not all embodiments. Based on the embodiments described in this document, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions of the present invention.

[0015] See Figure 1 This is an engineering management report generation method based on natural language processing provided in this embodiment of the invention. This method can be applied to an engineering management report generation system. The specific process is as follows: steps 110-140.

[0016] Step 110: Obtain the construction log text set generated during the construction process. The construction log text set contains multiple continuously recorded construction stage description units. Each construction stage description unit includes at least one construction activity description statement and a corresponding site status description statement.

[0017] To facilitate understanding of the overall technical solution, this embodiment of the invention uses a large commercial building project as an example for detailed explanation. This project encompasses multiple construction phases, from foundation construction to main structure erection and interior decoration. Construction logs are crucial records of the entire construction process, with construction workers recording daily activities and site conditions. In this project, the construction phase description units within the construction log text set detail the construction status at different stages. For example, during the foundation construction phase, the activity description might record "Conducting foundation excavation work, using excavators and loaders," while the corresponding site condition description might be "The ground is level, and warning signs have been set up around the site." These description units are recorded sequentially in chronological order, forming a complete chain of construction records.

[0018] As one implementation method, the acquisition of the construction log text set generated during the project construction process includes:

[0019] Step 111: Collect raw construction log data from the engineering management information system. The raw construction log data contains multiple unstructured text records indexed by timestamps.

[0020] In this commercial building project, an engineering management information system was used to record construction progress. This system allows construction workers to upload construction logs at any time. The raw construction log data is indexed by timestamps, making it easy to search and organize in chronological order. For example, in a record for a certain day, the timestamp might show "September 15, 2023, 9:00 AM." The corresponding unstructured text record might read, "Basement wall rebar tying work began today. Workers are operating in an orderly manner on site. The weather is sunny, and construction equipment is running normally." Although these text records do not have a fixed format, they contain rich construction information, and the timestamps clearly show the timeline of the construction.

[0021] Step 112: Perform data cleaning on the original construction log data, filtering out text segments containing duplicate content, punctuation marks with incorrect formatting, and redundant content unrelated to construction, to obtain preliminary cleaned candidate text of the construction log.

[0022] It's understandable that the original construction log data might have some issues and needs cleaning. For example, in this project, construction workers might sometimes mistakenly record duplicate content. Data cleaning begins with duplicate text detection of the original construction log data. By calculating the text similarity value between adjacent text segments, segments with similarity values ​​exceeding a preset threshold are identified as duplicate content, and only one of them is retained. For instance, if there are two text segments, "Concrete pouring work was carried out today, and the on-site workers are skilled" and "Concrete pouring operations were carried out today, and the on-site workers are skilled," and their similarity values ​​exceed the preset threshold, only one of them will be retained.

[0023] Simultaneously, the punctuation marks in the original construction log data underwent format verification. In project management, punctuation marks also need to be standardized, unifying full-width and half-width punctuation marks to the full-width punctuation format corresponding to project management, and correcting consecutive punctuation marks to a single standard punctuation mark. For example, "Today's work went very smoothly!!!" was corrected to "Today's work went very smoothly!".

[0024] In addition, the text content in the original construction log data undergoes topic relevance analysis. A keyword set is extracted from the text content and matched against a pre-defined construction topic keyword database. Text segments with keyword matching scores exceeding a preset threshold are selected as valid construction-related content. For example, the pre-defined construction topic keyword database includes keywords such as "building materials," "construction equipment," and "construction technology." If a text mentions "using new building materials for wall construction," after keyword matching, this text segment is considered valid construction-related content.

[0025] Finally, the detected non-duplicate text segments, the formatted text segments, and the valid text segments related to construction are merged to generate preliminary cleaned construction log candidate texts. These candidate texts do not contain duplicate content, incorrectly formatted punctuation, or redundant content unrelated to construction.

[0026] In a preferred embodiment, the data cleaning process performed on the original construction log data, filtering out text segments containing duplicate content, incorrectly formatted punctuation marks, and redundant content unrelated to construction, to obtain preliminary cleaned candidate text of the construction log, includes:

[0027] Step 1120: Perform duplicate text detection processing on the original construction log data, calculate the text similarity value between adjacent text segments, determine the text segments with similarity values ​​exceeding a preset threshold as duplicate content and retain one of them; perform format verification processing on the punctuation marks in the original construction log data, unify full-width and half-width symbols to the full-width punctuation format corresponding to engineering management, and correct multiple consecutive punctuation marks to a single standard punctuation mark; perform topic relevance analysis processing on the text content in the original construction log data, extract the keyword set of the text content, match the keyword set with a preset construction topic keyword library, and filter out text segments with keyword matching degree exceeding a preset threshold as valid content related to construction; merge the detected non-duplicate text segments, the format-corrected text segments, and the valid text segments related to construction to generate preliminary cleaned construction log candidate text, which does not contain duplicate content, incorrectly formatted punctuation, or redundant content unrelated to construction.

[0028] Continuing with the example of this commercial building project, in the duplicate text detection process, a text similarity calculation algorithm is used to calculate the similarity between each pair of adjacent texts. For example, one text describes "Some doors and windows were installed today, and the workers operated according to specifications," while another text states "Some doors and windows were installed today, and the workers operated in accordance with specifications." After calculation, their similarity values ​​exceed a preset threshold. In this case, the two texts are determined to be duplicate content, and only one of them is retained.

[0029] During the punctuation format verification process, the punctuation marks in the original construction log data were carefully examined. For instances where full-width and half-width characters were mixed, they were uniformly converted to the full-width punctuation format specified by project management. For example, "Today's work is completed, but there are some problems?" was changed to "Today's work is completed, but there are some problems?", and consecutive punctuation marks, such as "!!!", were corrected to a single standard punctuation mark "!".

[0030] In the topic relevance analysis, a set of keywords is extracted from each segment of the original construction log data. For example, for the text "At the construction site, a crane was used to lift building materials and simultaneously paint the walls," the extracted keyword set might be "crane," "building materials," and "wall painting." These keyword sets are then matched against a pre-defined construction topic keyword database. Only when the keyword match exceeds a preset threshold is the text segment considered valid content related to construction. Finally, the processed text segments are merged to obtain a pre-cleaned candidate text for the construction log, ensuring that the text does not contain duplicate content, formatted errors, punctuation errors, or redundant content unrelated to construction.

[0031] Step 113: The candidate text of the construction log is segmented according to a preset time interval. Text content describing the same construction stage within a continuous time range is merged into a construction stage description unit. The time range of the construction stage description unit is consistent with the stage division rules in the project construction plan.

[0032] In this commercial building project, the construction plan divides the entire project into multiple construction phases, such as the foundation construction phase, the main structure construction phase, and the interior decoration phase. The preset time intervals can be one week or one month. For example, using a one-week interval, the candidate text of the initially cleaned construction log is segmented. Within a certain week, the construction log records the work from Monday to Friday. Monday to Wednesday mainly involves the binding of foundation rebar, while Thursday and Friday begin the installation of foundation formwork. Therefore, the text describing the foundation rebar binding work and the text describing the foundation formwork installation work are merged into a single construction phase description unit. This unit corresponds to the foundation construction phase, and its time frame is consistent with the phase division rules in the construction plan, facilitating subsequent systematic analysis and summarization of the construction progress.

[0033] Step 114: Perform statement splitting on each construction stage description unit, separating the construction activity description statement that describes the construction activities and the site status description statement that records the site environment and equipment status, so that each construction stage description unit contains at least one construction activity description statement and one site status description statement.

[0034] Optionally, the merged basic construction stage description units can be split into separate statements. For example, a section of text in a construction stage description unit might read, "In the foundation construction area, from Monday to Wednesday, workers use rebar bending machines to tie the foundation rebars. The site environment is clean, and the equipment is operating normally." Through analysis, the construction activity description statement "workers use rebar bending machines to tie the foundation rebars" and the site status description statement "the site environment is clean, and the equipment is operating normally" can be separated. Thus, each construction stage description unit clearly includes information on both construction activities and site status, providing more accurate data for subsequent analysis.

[0035] Step 115: Perform text integrity verification on the split construction activity description statements and site status description statements, and generate a set of construction stage description units containing complete semantics as a construction log text set.

[0036] Furthermore, the separated construction activity description statements and site status description statements undergo text integrity verification. For example, for the construction activity description statement "Workers use a rebar bending machine to tie foundation rebar," it is checked whether it includes key information such as the construction subject (workers), construction equipment (rebar bending machine), construction object (foundation rebar), and construction behavior (tying work). For the site status description statement "The site environment is clean, and the equipment is operating normally," it is checked whether the site environment and equipment status are clearly described. Only when these statements contain complete semantics are they recombined into construction stage description units, ultimately generating a set of construction stage description units containing complete semantics, which serves as the construction log text set.

[0037] Step 120: Perform natural language processing on the construction log text set to extract key entities and progress events in each construction stage description unit, and obtain a construction feature set containing entity identification information and event timing information.

[0038] In this commercial building project, natural language processing was applied to the generated collection of construction log texts. For example, in a construction phase description unit during the foundation construction phase, the text reads, "On Tuesday, workers were mixing concrete in the foundation construction area using a concrete mixer. There was no standing water on the ground, and the equipment was operating normally." The goal of natural language processing was to extract key entities and progress events from this text.

[0039] In one implementation, the natural language processing performed on the construction log text set to extract key entities and progress events from each construction stage description unit yields a construction feature set containing entity identification information and event timing information, including:

[0040] Step 121: Perform word segmentation on the construction activity description statements and site status description statements in the construction log text set to obtain a word segmentation result set composed of basic vocabulary units.

[0041] The construction activity description statement "Workers are using a concrete mixer to mix concrete in the foundation construction area" and the site status description statement "There is no standing water on the ground, and the equipment is operating normally" in the above construction stage description unit are processed by word segmentation. The word segmentation algorithm will divide these statements into basic word units. For example, the construction activity description statement may be segmented into "workers", "in", "foundation construction area", "using", "concrete mixer", "mixing", and "concrete", and the site status description statement may be segmented into "site", "ground", "no standing water", "equipment", and "operating normally". These basic word units form the word segmentation result set.

[0042] Step 122: Call the pre-trained BERT language model to perform context semantic encoding on the word segmentation result set to generate a context embedding vector for each basic word unit. The context embedding vector contains the semantic association information of the word in the corresponding sentence.

[0043] In this step, the word segmentation results are input into the pre-trained BERT language model. Within the context of this commercial construction project, the BERT language model generates a context embedding vector for each basic lexical unit based on the contextual information of the words in the sentence. For example, for the word "concrete mixer," the model considers its context in the sentence "Workers use a concrete mixer to mix concrete in the foundation construction area," generating a context embedding vector that includes its semantic association information. This vector not only contains the semantics of "concrete mixer" itself but also its semantic relationships with other words such as "workers," "mixing," and "concrete."

[0044] Step 123: Perform entity recognition processing based on the context embedding vector to identify construction object entities, construction resource entities, and construction technology entities in the construction activity description statement, and to identify environmental status entities, equipment status entities, and personnel status entities in the site status description statement, thereby obtaining a set of entity identifiers containing entity type and entity location information.

[0045] Optionally, in the construction activity description statement, an entity recognition algorithm is used to identify the construction object entity "concrete," the construction resource entities "concrete mixer" and "worker," and the construction technology entity "mixing." In the site status description statement, the environmental status entity "site ground," the equipment status entity "equipment," and the personnel status entity (not explicitly mentioned in this statement but may appear in other related statements, such as "site management personnel") are identified. Simultaneously, the position information of each entity in the statement is recorded, such as the start and end positions of "concrete mixer" in the construction activity description statement, ultimately resulting in a set of entity identifiers containing entity type and location information.

[0046] As a design approach, entity recognition processing based on the context embedding vector is performed to identify construction object entities, construction resource entities, and construction technology entities in the construction activity description statement, and to identify environmental status entities, equipment status entities, and personnel status entities in the site status description statement, resulting in a set of entity identifiers containing entity type and entity location information, including:

[0047] Step 1230: Perform sequence labeling on the context embedding vector, labeling each basic lexical unit with an entity type label, including construction object entity labels, construction resource entity labels, construction technology entity labels, environmental status entity labels, equipment status entity labels, and personnel status entity labels; perform entity boundary recognition processing on the labeled lexical units, merging lexical units consecutively labeled with the same entity type label to form a complete entity name description; record the start and end positions of each entity name description in the construction activity description statement or site status description statement, generating a character index range representing the entity location; perform semantic consistency verification processing on the identified entity name descriptions, unifying different descriptions of the same entity into the standard entity name in the engineering specification knowledge base; integrate the entity type labels, standard entity names, and character index ranges to generate an entity identifier set containing entity type and entity location information.

[0048] In this commercial building project, the context embedding vectors are subjected to sequence labeling. For example, for lexical units in the construction activity description statements, based on their semantics and role in the statement, "concrete" is labeled with a construction object entity tag, "concrete mixer" with a construction resource entity tag, and "mixing" with a construction technology entity tag. Then, entity boundary recognition processing is performed on the labeled lexical units, merging lexical units labeled with the same entity type consecutively. For example, "concrete mixer" is treated as a single construction resource entity. Next, the start and end positions of each entity name expression in the statement are recorded, such as the character index range of "concrete mixer" in the statement. Simultaneously, semantic consistency verification processing is performed on the identified entity name expressions. For example, if "concrete mixer" ("concrete" is an abbreviation of "concrete") appears in other statements, it is unified as the standard entity name "concrete mixer" in the engineering specification knowledge base. Finally, the entity type tags, standard entity names, and character index ranges are integrated to form an entity identifier set containing entity type and entity location information.

[0049] Step 124: Perform temporal relationship analysis on the time adverbs and sequence conjunctions in the construction stage description unit to determine the execution order between the construction activity description statements and the temporal correspondence between the on-site status description statements and the construction activities, and generate event timing information that represents the execution order of construction activities and the timing of status changes.

[0050] In the construction phase description unit of the basic construction stage, time adverbs and sequence conjunctions are analyzed. For example, the construction phase description unit contains the description, "On Tuesday, workers first carried out rebar tying work, and then on Wednesday, they began installing formwork." By analyzing time adverbs and sequence conjunctions such as "Tuesday," "Wednesday," "first," and "then," it is determined that the rebar tying work was performed on Tuesday, the formwork installation work was performed on Wednesday, and the rebar tying work preceded the formwork installation work. Simultaneously, for the site status description statement, "On Tuesday, the site ground was level; on Wednesday, when the formwork was installed, there was a small amount of material piled up on site," the temporal correspondence between the site status and the construction activities is determined. Finally, event sequence information representing the execution order of construction activities and the timing of status changes is generated, clarifying the sequence of construction activities and the changes in site status over time.

[0051] Step 125: Associate and integrate the entity identifier set with the event timing information to generate a construction feature set containing entity identifier information and event timing information. The entity identifier information includes entity type, entity name and the position of the statement it is located in, and the event timing information includes the event execution order and the time point of the state change.

[0052] Optionally, the previously obtained entity identifier set and event timing information can be correlated and integrated. In this commercial building project, the entity identifier set contains information about construction resource entities such as "concrete mixer," and the event timing information contains information such as "rebar tying work is carried out on Tuesday, and formwork installation work is carried out on Wednesday." After integrating them, the generated construction feature set not only knows that the construction resource entity "concrete mixer" exists, but also knows that it may be used in the rebar tying work on Tuesday or the formwork installation work on Wednesday, clarifying the temporal association between the entity and the event. Thus, the construction feature set contains complete entity identifier information and event timing information.

[0053] Step 130: Perform semantic alignment processing between the construction feature set and the pre-set engineering specification knowledge base to generate specification association information between the construction stage description unit and the engineering specification requirements.

[0054] In commercial building projects, the engineering specification knowledge base contains various construction specifications and standards. The construction feature set contains entity and event information extracted from construction logs, and these need to be semantically aligned to determine the association between construction phase description units and engineering specification requirements.

[0055] In a preferred embodiment, the step of semantically aligning the construction feature set with a pre-set engineering specification knowledge base to generate specification association information between the construction stage description unit and engineering specification requirements includes:

[0056] Step 131: Extract a set of standard entries related to the construction object entity and construction resource entity from the engineering standard knowledge base. The set of standard entries includes construction technical requirements, resource allocation standards and environmental safety regulations.

[0057] From the engineering specification knowledge base, for the basic construction phase of commercial building projects, a set of specification entries related to the construction object entity such as "concrete" and the construction resource entity such as "concrete mixer" is extracted. For example, for "concrete," specification entries may include construction technical requirements such as concrete mix proportion requirements and pouring temperature requirements; for "concrete mixer," specification entries may involve resource allocation standards such as equipment periodic maintenance standards and operator qualification requirements, as well as environmental safety regulations such as safety protection regulations during equipment operation.

[0058] Step 132: Perform semantic feature extraction processing on the entity identification information in the construction feature set to generate entity semantic vectors representing entity type and entity attributes.

[0059] Furthermore, semantic feature extraction is performed on the entity identifier information in the construction feature set. For example, for the entity identifier information "concrete mixer", its entity type (construction resource entity) and its attributes, such as the equipment model and power, are analyzed. This information is encoded to generate an entity semantic vector representing the entity type and entity attributes. This vector can represent the semantic features of "concrete mixer" in the construction feature set, providing a basis for subsequent comparison with the specification semantic vector.

[0060] Step 133: Extract semantic features from the normative content in the set of normative entries to generate a normative semantic vector representing the normative requirements and applicable conditions.

[0061] Optionally, semantic features are extracted from the set of standard entries extracted from the engineering standard knowledge base. Taking the standard entry for "concrete mixer" as an example, the standard requirements may include that the equipment must undergo regular comprehensive inspections and that operators must possess the corresponding operating certificates; the applicable conditions may be under the corresponding construction environment and construction task. These standard requirements and applicable conditions are analyzed and encoded to generate standard semantic vectors representing the standard requirements and applicable conditions. These vectors reflect the semantic features of the standard entries, facilitating matching and comparison with entity semantic vectors.

[0062] Step 134: Calculate the semantic similarity value between the entity semantic vector and the specification semantic vector, and select specification entries whose semantic similarity value with the entity identification information in the construction feature set exceeds a preset threshold as associated specification entries.

[0063] In this step, the semantic similarity value between the entity semantic vector and the specification semantic vector can be calculated using the cosine similarity algorithm. In this commercial building project, taking the entity semantic vector of "concrete mixer" and its corresponding specification semantic vector as an example, the similarity between the two is calculated. After the calculation, a preset threshold is set, and specification entries with semantic similarity values ​​greater than this threshold are selected as candidate associated specification entries. For example, if the semantic vector of the specification entry "concrete mixer needs to undergo comprehensive maintenance every two weeks" has a similarity value exceeding the preset threshold with the entity semantic vector of "concrete mixer," then this specification entry becomes a candidate associated specification entry.

[0064] In a preferred embodiment, calculating the semantic similarity value between the entity semantic vector and the specification semantic vector, and selecting specification entries whose semantic similarity value with entity identifier information in the construction feature set exceeds a preset threshold as associated specification entries, includes:

[0065] Step 1340: Perform dimension alignment processing on the entity semantic vector and the canonical semantic vector; calculate the cosine value of the angle between the dimension-aligned entity semantic vector and the canonical semantic vector as the semantic similarity value; normalize the semantic similarity value to obtain a normalized similarity value; set a semantic similarity threshold, and filter out canonical entries with a normalized similarity value greater than the semantic similarity threshold as candidate associated canonical entries; perform redundancy filtering processing on the candidate associated canonical entries, remove duplicate canonical entries associated with the same entity identifier information, and retain the canonical entry with the highest semantic similarity value as the associated canonical entry.

[0066] In commercial building projects, the entity semantic vector of "concrete mixer" and its related specification semantic vector are first aligned dimensionally to ensure they are consistent in dimension for subsequent calculations. Next, the semantic similarity between them is determined by calculating the cosine of the angle between them. For example, the cosine of the angle between the entity semantic vector and the specification semantic vector is calculated; this value reflects the degree of similarity between the two vectors in direction, thus representing their semantic similarity.

[0067] Then, the obtained semantic similarity values ​​are normalized and mapped to a set range (e.g., [0, 1]) to obtain normalized similarity values, which facilitates comparison and threshold setting. A suitable semantic similarity threshold is set, such as 0.7. Standard entries with normalized similarity values ​​greater than 0.7 are selected as candidate association standard entries.

[0068] After obtaining candidate related specification entries, some duplicate or redundant entries may exist. For example, for "concrete mixer," there may be multiple specification entries involving equipment maintenance, but with slightly different descriptions. In this case, a redundancy screening process is performed on the candidate related specification entries, removing duplicate entries associated with the same entity identifier information, and retaining only the specification entry with the highest semantic similarity value as the related specification entry. This ensures that the related specification entries are both accurate and concise, providing more effective information for subsequent analysis.

[0069] Step 135: Combining the event sequence information in the construction feature set, analyze the applicable time range of the associated specification entries in the construction stage description unit, determine the time matching relationship between the execution sequence of construction activities and the time point of state change and the specification requirements, and generate specification association information containing associated specification entries and time matching relationships.

[0070] It is understandable that, regarding the event sequence information in the construction feature set, in the foundation construction phase of a commercial building project, the event sequence information "reinforcement tying work is carried out on Tuesday, formwork installation work is carried out on Wednesday, and the concrete mixer may be used during these two tasks" is known, along with the associated specification item "the concrete mixer needs to undergo comprehensive maintenance every two weeks." Analyzing the applicable time range of this associated specification item in the construction phase description unit, it is determined whether the time for comprehensive maintenance of the concrete mixer is reached during the reinforcement tying work on Tuesday and the formwork installation work on Wednesday. The timing matching relationship between the execution sequence of construction activities and the time points of status changes and the time requirements of the specification is determined. For example, if the specification requires maintenance after one week of continuous equipment use, and the concrete mixer in the current construction plan has been used since last week, nearly a week has passed by this Wednesday, then a timing matching relationship exists. Finally, specification association information containing associated specification items and timing matching relationships is generated.

[0071] Step 140: Based on the hierarchical template matching algorithm, the specification-related information is structurally integrated to generate a structured report text that meets the requirements of engineering management. The structured report text includes a description of the construction progress, specification compliance analysis, and recommendations.

[0072] In commercial building projects, hierarchical template matching algorithms are used to process standardized information to generate structured report texts that meet engineering management requirements, providing clear and accurate information for project management.

[0073] In an alternative embodiment, the hierarchical template matching algorithm is used to perform structured integration processing on the specification-related information to generate a structured report text that meets engineering management requirements. The structured report text includes a construction progress description, specification compliance analysis, and recommendations, including:

[0074] Step 141: Select a hierarchical template structure from the preset engineering management report template library. The hierarchical template structure includes a first-level heading layer, a second-level content layer, and a third-level detail layer. The first-level heading layer corresponds to the construction progress description, specification compliance analysis, and recommendations.

[0075] In commercial building projects, a pre-set project management report template library offers a variety of template structures to choose from. Selecting the hierarchical template structure, the first-level headings clearly correspond to three key sections: construction schedule description, code compliance analysis, and recommendations. For example, in this commercial building project, the construction schedule description clearly demonstrates the progress of each construction phase; the code compliance analysis assesses whether construction activities comply with engineering code requirements; and the recommendations section provides targeted improvement or optimization suggestions based on the analysis results. The second-level content layer and the third-level detail layer provide more detailed explanations and supporting information for these main sections.

[0076] Step 142: Map the event sequence information in the construction feature set to the construction progress description template area of ​​the first-level title layer, and organize the timeline description of construction activities according to the event execution order and the time point of state change.

[0077] In detail, the event sequence information from the construction feature set, such as "rebar tying work was carried out on Tuesday, and formwork installation work was carried out on Wednesday," is mapped to the construction progress description template area at the first-level heading level. The timeline description of construction activities is organized in the template area according to the order of event execution and the time points of status changes. In a commercial building project report, this might be presented as, "During the foundation construction phase, rebar tying work was carried out on Tuesday, and the construction status was normal; formwork installation work was carried out on Wednesday, and on-site construction progressed smoothly." This method clearly displays the time sequence and progress of construction activities, enabling managers to quickly understand the project's construction progress.

[0078] Step 143: Map the associated standard entries and time matching relationships in the standard association information to the standard compliance analysis template area of ​​the first-level heading layer, compare the execution status of construction activities with the time matching relationship of standard requirements, and generate an analysis conclusion on whether the construction activities comply with the standard requirements.

[0079] In detail, the relevant standard items and their time-related relationships in the standard information are mapped to the standard compliance analysis template area at the first-level heading level. For example, "The concrete mixer needs to undergo comprehensive maintenance every two weeks. Currently, the concrete mixer has been used continuously for one week and is expected to require maintenance next week." In this area, the time-related relationship between the execution of construction activities and the standard requirements is compared. For the foundation construction phase of a commercial building project, if the standard requires the concrete mixer to be inspected after one week of continuous use, and this time has been reached but no inspection has been performed, then the analysis concludes that the current construction activities do not meet the standard requirements regarding concrete mixer maintenance. Through the above comparison and analysis, a clear analytical conclusion on whether the construction activities comply with the standard requirements is generated, providing important decision-making basis for project management.

[0080] Step 144: Based on the analysis conclusions of the specification compliance analysis template area, extract improvement measures suggestions for non-compliance and optimization suggestions for compliance from the engineering specification knowledge base, and map them to the suggestion content template area of ​​the first-level heading layer.

[0081] Optionally, based on the analysis conclusions derived from the compliance analysis template area, in commercial building projects, if the analysis indicates that the maintenance of the concrete mixer does not meet the requirements of the specifications, improvement suggestions for this non-compliance are extracted from the engineering specification knowledge base, such as "Immediately arrange a comprehensive maintenance of the concrete mixer and establish a regular maintenance reminder mechanism." If the construction activities meet the requirements of the specifications, optimization suggestions are also extracted from the knowledge base, such as "Consider upgrading the equipment during maintenance to improve construction efficiency." These suggestions are mapped to the suggestion content template area at the first-level heading level, providing specific directions and measures for project improvement and optimization.

[0082] Step 145: Perform semantic description optimization on the second-level content layer and the third-level detail layer. Use a transfer learning strategy to adjust the general description statements in the template to optimized description statements that conform to the semantic style of the construction log text set, so that the language expression of the structured report text is consistent with the semantic output state of the original construction log, and generate a structured report text containing construction progress descriptions, specification compliance analysis and recommendations.

[0083] In detail, the semantic description optimization processing of the second-level content layer and the third-level detail layer involves using a transfer learning strategy to adjust the general description statements in the template to optimized description statements that conform to the semantic style of the construction log text set, so as to ensure that the language expression of the structured report text is consistent with the semantic output state of the original construction log, including:

[0084] Step 1450: Extract frequent term vocabulary and key sentence structures from the construction log text set to generate a language feature template representing the semantic style of the original construction log; select a pre-trained transfer learning model, use the language feature template as the target style for style transfer, and use the general descriptive statements in the language feature template as input text; fine-tune the transfer learning model to enable it to learn to convert general descriptive statements into optimized descriptive statements that conform to the language feature template; input the general descriptive statements from the second-level content layer and the third-level detail layer into the fine-tuned transfer learning model to generate optimized descriptive statements after style transfer; perform grammatical correctness verification and semantic coherence checks on the optimized descriptive statements, correct grammatical errors and semantic ambiguities generated during the transfer process, and obtain optimized descriptive statements that are consistent with the semantic output state of the original construction log.

[0085] In commercial building projects, frequent terminology sets, such as words like "conducting," "using," and "on-site," are extracted from construction log text sets, along with key sentence structures, such as "in the... area, ...conducting... work." This information is then integrated to generate a linguistic feature template representing the semantic style of the original construction logs.

[0086] Then, a pre-trained transfer learning model is selected, and the generated language feature template is used as the target style for style transfer. General descriptive statements in the language feature template, such as "carrying out specific work in a specific area," are used as input text. The transfer learning model is fine-tuned to learn how to convert such general descriptive statements into optimized descriptive statements that conform to the language feature template, such as "at the construction site, workers are carrying out specific construction work."

[0087] Furthermore, the general descriptive statements from the second-level content layer and the third-level detail layer are input into the fine-tuned transfer learning model. For example, the description of construction equipment in the second-level content layer, "the equipment is used on site," is transformed by the model into the optimized descriptive statement "the equipment is put into use at the construction site" after style transfer.

[0088] Next, the generated optimized description statements undergo syntactic correctness and semantic coherence checks. The statements are checked for syntactic errors, such as improper subject-verb-object collocation, and for semantic coherence and consistency with the logic of the construction scenario. Statements with syntactic errors or semantic ambiguity are corrected, ultimately resulting in optimized description statements that maintain consistency with the semantic output of the original construction log. This generates a structured report text containing construction progress descriptions, specification compliance analysis, and recommendations.

[0089] In an alternative embodiment, after generating the structured report text that meets engineering management requirements, the method further includes:

[0090] Step 210: Perform a content integrity check on the construction progress description report area in the structured report text to check whether it contains the event execution sequence and status change time points of all construction stage description units.

[0091] After generating the structured report text for a commercial building project, the construction progress description report section is validated. For example, during the foundation construction phase, the construction progress description should cover the sequence of events for each construction activity, from foundation excavation, rebar tying, formwork installation to concrete pouring, as well as the time points for status changes for each activity, such as start time, end time, and whether it was successfully completed. The report should be checked to ensure that this information is completely recorded. If any construction activity is missing or its time point is not fully recorded, it should be marked as incomplete and requires further supplementation and improvement.

[0092] Step 220: Perform a logical rationality check on the specification compliance analysis report area in the structured report text, and check whether the analysis conclusion is based on the associated specification entries and time matching relationships in the specification association information.

[0093] For the compliance analysis report section, examine the logical rationality of the analysis conclusions. In commercial building projects, if the analysis conclusion indicates that a certain construction activity does not meet the requirements of the specifications, check whether the conclusion is based on the relevant specification entries and time matching relationships in the specification information. For example, if the conclusion is that the concrete pouring temperature does not meet the specification requirements, it is necessary to confirm whether there is a corresponding specification entry that stipulates the temperature range for concrete pouring, and whether the concrete pouring time and actual temperature recorded in the construction log have been correctly compared and analyzed with the specification requirements. If the analysis conclusion is not reasonably based on the specification information, then the analysis process needs to be reviewed and revised.

[0094] Step 230: Perform a feasibility check on the suggested content reporting area in the structured report text, and check whether the improvement measures and optimization suggestions have corresponding standard basis in the engineering standard knowledge base.

[0095] In commercial building projects, the proposed content report area should be validated. Check whether the proposed improvement measures and optimization suggestions have corresponding regulatory basis in the engineering specifications knowledge base. For example, if a suggestion is to improve construction techniques to enhance construction quality, it's necessary to check whether there are relevant specifications and standards supporting this improvement in the engineering specifications knowledge base. If the suggestion lacks corresponding regulatory basis, its feasibility is questionable, and the content needs to be reassessed and adjusted to ensure that the suggestion is practically operable and compliant with regulations.

[0096] Step 240: Correct any missing information, logical contradictions, or unfounded suggestions found during the verification process in the report areas, supplement any missing progress information, adjust any contradictory analysis conclusions, and replace any unfounded suggestions.

[0097] Based on the verification results, the structured report text was revised. For missing progress information in the construction progress description report area, such as an omitted time record for a certain construction stage, the missing information was retrieved from the original construction log and supplemented. For logically contradictory analysis conclusions in the specification compliance analysis report area, the relevant specification information and actual construction conditions were re-examined, and the analysis conclusions were adjusted to ensure logical consistency. For unsubstantiated recommendations in the recommendation content report area, suitable evidence was sought from the engineering specification knowledge base, or substantiated recommendations were re-proposed to replace the original unsubstantiated recommendations, thus ensuring the accuracy and reliability of the report text.

[0098] Step 250: Generate an updated structured report text that has undergone content integrity verification and correction.

[0099] After the above verification and correction process, the corrected content is integrated to generate an updated structured report text. This updated report text is more accurate, complete, and reasonable in terms of construction progress description, specification compliance analysis, and recommendations, and can provide more effective decision support and reference for the engineering management of commercial building projects.

[0100] As a non-limiting embodiment, after generating the structured report text that meets the requirements of engineering management, the method further includes:

[0101] Obtain the semantic fluency assessment requirements for the structured report text, which include the formality requirements and coherence standards for engineering management documents.

[0102] The pre-trained NLP generation optimization model is invoked to perform language style conversion on the structured report text. The NLP generation optimization model is trained on a standard document corpus in the field of engineering management.

[0103] The NLP generation optimization model is used to convert colloquial expressions in the structured report text into written form, merge duplicate text, and add conjunctions to logically disjointed statements.

[0104] Extract the set of technical terms from the structured report text, perform consistency matching between the set of technical terms and the standard terminology database in the engineering specification knowledge base, and replace the non-standard terms in the set of technical terms with standard terms.

[0105] Generate a semantically optimized report text that has undergone language style conversion, duplicate text merging, logical connection supplementation, and terminology consistency correction.

[0106] Understandably, in commercial building projects, the first step is to clarify the semantic fluency assessment requirements of structured report texts. These requirements include the formality requirements of project management documents, such as rigorous and standardized language expression, avoiding colloquialisms and casual expressions; and the standards of expression coherence, that is, there should be reasonable logical connections between sentences to avoid logical jumps.

[0107] A pre-trained NLP generative optimization model, trained on a large corpus of standard documents in the field of engineering management, is used to perform language style conversion on structured report texts of commercial construction projects. For example, the colloquial expression "Let's start working today" is transformed into the formal "Construction work begins today".

[0108] Merge duplicate text in the report. For example, if "concrete pouring work is carried out on the construction site" is mentioned multiple times, merge it into "concrete pouring work is carried out on the construction site". For statements with logical jumps, add conjunctions to make them logically coherent. For example, "The construction workers tied the reinforcing bars, and then began the formwork installation", after adding conjunctions, it becomes "After the construction workers completed the reinforcing bar tying work, they immediately began the formwork installation work".

[0109] Extract specialized terminology sets from the structured report text, such as "concrete" and "scaffolding," and perform consistency matching between these terminology sets and the standard terminology database in the engineering specifications knowledge base. If the term "concrete" in the report does not match the standard terminology database, it is replaced with the standard term. After this processing, a semantically optimized report text is generated, making the report more in line with the requirements of engineering management documents in terms of language style, coherence, and the use of specialized terminology.

[0110] As a non-limiting embodiment, after generating the structured report text that meets the requirements of engineering management, the method further includes:

[0111] Extract the event sequence information from the construction progress description text set, the related standard items from the standard compliance analysis text set, and the improvement measures suggestions from the recommended content text set from the structured report text to form a set of report content elements;

[0112] Logical correlation analysis is performed on the set of elements in the report to identify whether there is a contradiction between the event execution order in the construction progress description and the time matching relationship in the standard compliance analysis, and to check whether there is a logical break between the improvement measures suggested in the recommendation content and the conclusions of the standard compliance analysis.

[0113] For the identified contradictions or breakpoints, the entity identification information and event sequence information in the construction feature set are traced back to verify whether the contradictions are caused by the original record deviation of the construction log text set or by the information mapping error of the hierarchical template matching algorithm, and the backtracking verification results are obtained.

[0114] Based on the backtracking verification results, the structured report text is partially modified, the description of the event execution order is adjusted to match the time matching relationship, and the logical derivation process between the suggested content and the analysis conclusion is supplemented.

[0115] Generate a logic verification report text that has undergone logical correlation analysis and local content correction.

[0116] In detail, in commercial building projects, the event sequence information of the construction progress description text set, such as "reinforcement binding is carried out on Tuesday and formwork is installed on Wednesday", the relevant specification items of the specification compliance analysis text set, such as "concrete mixers need to be maintained regularly", and the improvement measures suggestions of the recommendation content text set, such as "the frequency of daily inspections of concrete mixers should be increased", are extracted from the structured report texts. This information is then integrated to form a set of report content elements.

[0117] A logical correlation analysis was performed on the set of elements in the report. In this commercial building project, the focus was on identifying any contradictions between the event execution sequence in the construction schedule description and the time matching relationship in the code compliance analysis. For example, the construction schedule description showed that a concrete mixer was used continuously for construction during a certain period, but the code compliance analysis required maintenance of the concrete mixer during that period. However, no maintenance was recorded, indicating a contradiction in the time matching relationship. Simultaneously, the analysis was examined to check for any logical breaks between the improvement measures suggested in the recommendations and the conclusions of the code compliance analysis. For instance, if the code compliance analysis concluded that the concrete pouring temperature did not meet the code requirements, but the recommendations suggested increasing the concrete mixing intensity, this was clearly not logically related to the analysis conclusion and constituted a logical break.

[0118] For identified contradictions or breakpoints, the entity identification information and event sequence information in the construction feature set are traced back. For example, in the concrete mixer example above, the entity identification information related to the concrete mixer in the construction feature set, including equipment model, usage records, etc., and event sequence information, are examined to determine if there are discrepancies in the original records of the construction log text set. It's possible that construction personnel made a recording error, mistakenly recording normal use when maintenance was actually required; or it could be due to information mapping errors in the hierarchical template matching algorithm. For example, errors in time or entity association may have occurred when mapping construction feature information to the specification analysis template area. After detailed inspection and verification, the backtracking verification results are obtained.

[0119] Based on the backtesting results, partial content corrections are made to the structured report text. If the discrepancy stems from deviations in the original records of the construction log text set, such as a recording error, the description of the event execution sequence is adjusted to match the timeline, and the actual use and maintenance of the concrete mixer is clearly recorded in the report. If there is a logical disconnect between the recommendations and the analysis conclusions, the logical derivation process between the recommendations and the analysis conclusions is supplemented. For example, based on the recommendation to increase the concrete mixing intensity, further explanation is given regarding the relationship between the mixing intensity and the concrete temperature, and how adjusting the mixing intensity can affect the concrete pouring temperature to meet the specifications.

[0120] Through the above operations, a logical verification report text is generated after logical correlation analysis and partial content correction. This report text establishes a closer and more reasonable logical connection between construction progress, standard compliance, and recommendations, improving the accuracy and reliability of the report and providing more valuable information for the engineering management of commercial building projects.

[0121] As a non-limiting embodiment, after generating the structured report text that meets the requirements of engineering management, the method further includes:

[0122] Obtain the role identification information of the target reading user, including the management user role, the technical execution layer user role, and the external supervision user role;

[0123] Based on the role identification information, the set of information concerns corresponding to the role is extracted from the preset user demand feature library. The set of information concerns includes a summary of key indicators that the management focuses on, a description of operational details that the technical execution layer focuses on, and an explanation of the normative basis that external regulatory users focus on.

[0124] The structured report text is sorted by content priority, and the sorted content is then processed for information focus. An overview report containing key indicator summaries and conclusion overviews is generated for the management user role, an annotated report containing operational details and step descriptions is generated for the technical execution user role, and a compliance report containing regulatory basis explanations and compliance certificates is generated for the external regulatory user role.

[0125] Generate multi-version adapted report texts for different target reading user roles. The level of detail and information focus of the multi-version adapted report texts are matched with the information focus set of the corresponding role.

[0126] In commercial building projects, the first step is to obtain the role identification information of the target users. Different user roles have different needs and concerns regarding the reports. This embodiment of the invention includes management user roles, technical execution user roles, and external regulatory user roles.

[0127] In particular, management focuses on the overall progress and key performance indicators (KPIs) of the project to facilitate decision-making. Based on role identification information, a set of information points of concern to management is extracted from a pre-defined user demand feature library. This set includes summaries of key performance indicators, such as project progress completion rate, cost control, and quality compliance rate. For commercial building projects, management may be concerned with whether the foundation construction phase is completed on schedule, whether the cost is within budget, and whether the quality of key construction stages such as concrete pouring meets standards.

[0128] In addition, the technical execution layer primarily focuses on the details and steps of construction operations to ensure the accurate and error-free execution of construction tasks. A set of information concerns for the technical execution layer is extracted from the user requirement feature database, including detailed descriptions of operations, such as the specific operating procedures of a concrete mixer and the spacing requirements for rebar tying. In commercial building projects, technical execution personnel need a thorough understanding of the specific operational requirements of each construction stage to guide on-site construction.

[0129] Furthermore, external oversight users focus on whether the project complies with relevant regulations and standards, requiring clear explanations of the regulatory basis. We extract a set of information points of concern to external oversight users from the user demand feature database, including explanations of the regulatory basis, such as the engineering specifications and safety standards upon which various construction activities are based. For commercial building projects, external oversight users will check whether the foundation construction meets the requirements of regulations such as the building foundation engineering construction quality acceptance standards.

[0130] Next, the structured report text is prioritized based on content importance. The content is ordered according to the information focus of different user roles. For example, for management users, content related to key indicators is placed first; for technical execution users, content related to operational details is given higher priority; and for external regulatory users, content related to regulatory compliance is highlighted.

[0131] The sorted content undergoes information focusing processing. For management users, key indicator summaries and conclusion overviews are extracted from the structured report text to generate an overview report containing this information. For example, the overview report concisely presents key information such as the progress of the foundation construction phase, cost deviations, and overall quality evaluation, enabling management to quickly understand the core aspects of the project. For technical execution users, operational details and step-by-step instructions are extracted from the report to generate an annotation report. For example, the annotation report details the start-up, operation, and shutdown procedures of the concrete mixer, as well as the specific spacing and binding methods of the rebar, providing detailed construction guidance for technical execution personnel. For external regulatory users, the explanation of the applicable standards and compliance certificates are extracted to generate a compliance report. For example, the compliance report clearly indicates the engineering standard clauses followed by each stage of foundation construction and the testing methods used to prove that the construction activities comply with these standard requirements, providing clear compliance proof for external regulatory users.

[0132] Through the above processing, multiple versions of adapted report texts are finally generated for different target reading user roles. These report texts closely match the information focus of the corresponding roles in terms of content level and information focus, meeting the diverse needs of different user roles in commercial building project engineering management. This helps to improve the efficiency of project management and the scientific nature of decision-making, ensuring the smooth progress of the project and compliance with relevant standards and specifications.

[0133] In practical applications, text similarity calculation can be achieved by introducing ready-made similarity functions (such as cosine similarity or Jaccard similarity) from NLP toolkits (such as spaCy or scikit-learn). These functions can be used to calculate text segment similarity values ​​and apply preset thresholds to filter duplicate content. For word segmentation and contextual semantic encoding, word segmentation and embedding vector generation can be directly implemented using the interface of pre-trained BERT models (such as those provided by the HuggingFace library), without the need for custom training or in-depth structural analysis. For entity recognition, entity type labeling and boundary merging can be completed using standard sequence labeling model libraries (such as spaCy's entity recognition module or BERT-based NER), and pre-trained models can be used to process text data. For vector alignment and cosine similarity in semantic similarity calculation, vector similarity calculation functions from NLP libraries (such as NumPy's cosine similarity implementation) can be directly used to process dimensions and values.

[0134] More specifically, for transfer learning optimization, a transfer learning framework (such as TensorFlowHub or a pre-trained T5 model) is used for fine-tuning training, and general statements are adjusted by loading the base model and inputting style feature templates; for rules such as time interval division, the interval is automatically set with reference to the standard rule base for engineering management (such as the Primavera project management rule set); for knowledge base matching rules, the filtering of specification items is handled based on a SQL-like rule engine (such as Drools).

[0135] Furthermore, the logical relationship and hierarchy between the input construction log text set and the output structured report text are explained as follows: The input is the original construction log text set (containing descriptions of construction activities and on-site status). First, a data cleaning layer filters out redundant and erroneous content to form preliminary cleaned text. This layer focuses on the standardization and integrity of the original data. Next, a segmentation layer merges the text into construction stage description units and adds a time dimension structure according to time interval rules. Then, it enters the natural language processing layer. The word segmentation results are used by the BERT model to generate context embedding vectors, providing a semantic foundation for entity recognition and time series analysis. This layer... The input entity identifier set and event sequence information serve as intermediate outputs, forming the core elements of the construction feature set. Subsequently, through a semantic alignment processing layer, the construction feature set is matched with the engineering specification knowledge base to generate specification-related information, adding a new information layer of specification requirements and time matching relationships. Finally, through a hierarchical template matching processing layer, the information is integrated into a structured report text, incorporating construction progress descriptions, specification compliance analysis, and recommendations into the template mapping. Each processing layer deepens or expands the input data, and the output text not only inherits the semantic style of the input log but also achieves information refinement and standardization through a multi-layered structured process, reflecting a bottom-up hierarchical logic.

[0136] In summary, this invention, by acquiring a set of construction log texts containing detailed descriptions of construction stages and combining this with natural language processing technology for text mining, can accurately extract key entities and progress events to form a construction feature set. This breaks through the limitations of traditional manual analysis and greatly improves the accuracy and efficiency of information extraction. The semantic alignment of the construction feature set with the engineering specification knowledge base achieves a deep correlation between actual construction and specification requirements, enabling timely detection of potential specification non-compliance issues. The structured integration based on a hierarchical template matching algorithm generates a structured report text that comprehensively covers construction progress, specification compliance, and recommendations. This provides a systematic, clear, and targeted decision-making basis for project management, enhancing the scientific, standardized, and forward-looking nature of project management and effectively promoting the efficient and compliant development of engineering construction projects.

[0137] Based on the same inventive concept, embodiments of the present invention also provide an engineering management report generation system. See also... Figure 2 As shown, this is a schematic diagram of a possible engineering management report generation system provided in an embodiment of the present invention. Figure 2 The engineering management report generation system 200 includes a processor 210 and a memory 220. The memory 220 stores computer programs executable by the processor 210. By executing the instructions stored in the memory 220, the processor 210 can perform the steps of the above-described engineering management report generation method based on natural language processing.

[0138] Based on the same inventive concept, embodiments of the present invention provide a computer-readable storage medium including a computer program. When the computer program is run on an engineering management report generation system, it causes the engineering management report generation system to perform the steps of the aforementioned natural language processing-based engineering management report generation method. In some possible embodiments, various aspects of the natural language processing-based engineering management report generation method provided by the present invention can also be implemented as a program product including a computer program. When the program product is run on an engineering management report generation system, the computer program causes the engineering management report generation system to perform the steps of the aforementioned natural language processing-based engineering management report generation method. For example, the engineering management report generation system can perform actions such as... Figure 1 The steps are shown in the figure.

[0139] In the technical solutions involved in the above embodiments of the present invention, whether performing comparison calculations of multi-dimensional features or constructing composite parameters, if there are problems caused by significant differences in the number of dimensions, units of measurement, and semantic meanings of different features, those skilled in the art, based on their professional knowledge and past practical experience, can fully understand that these differences need to be properly handled so that the calculation results are accurate and comparable, and to avoid situations such as logical confusion and unclear mathematical meaning.

[0140] In detail, when faced with features of different numbers of dimensions, those skilled in the art can employ various strategies, including but not limited to feature selection, feature extraction, and kernel function processing, in order to accurately calculate the similarity, matching degree, or feature distance between different features.

[0141] In order to achieve comparability alignment of feature spaces when comparing multidimensional features, those skilled in the art can use a variety of existing and common technical means, including but not limited to standardization preprocessing, mapping transformation, and spatial projection.

[0142] In the process of constructing composite parameters (such as loss function values), different parameter terms often have different dimensions. Those skilled in the art can use normalization processing or an adaptive weight allocation mechanism based on distribution characteristics.

[0143] The aforementioned general techniques for solving feature matching and loss balance problems are all common knowledge in this field. These techniques have been fully verified and widely used in numerous practical applications, and those skilled in the art can skillfully and flexibly apply these methods to handle similar problems involving differences in dimensions.

[0144] The formulas and calculation processes involved in the embodiments of this invention, whether used for multidimensional feature comparison or composite loss function construction, strictly adhere to the principle of dimensional correspondence. Each variable in the formula has a clear and explicit physical meaning, and its operational logic fully conforms to basic mathematical and physical logic. The calculation results are necessarily the reasonable results expected by this invention. Those skilled in the art are capable of effectively solving various problems arising from the number of dimensions, dimensional differences, etc., in the multidimensional feature comparison calculation and composite loss function construction in the embodiments, based on specific data conditions and business needs, by comprehensively utilizing the above-mentioned general technical means, thus ensuring the accuracy, reliability, and implementability of the technical solution of this invention.

Claims

1. A method for generating engineering management reports based on natural language processing, characterized in that, The method includes: Obtain a set of construction log texts generated during the construction process. The set of construction log texts contains multiple continuously recorded construction stage description units. Each construction stage description unit includes at least one construction activity description statement and a corresponding site status description statement. Natural language processing is performed on the construction log text set to extract key entities and progress events in each construction stage description unit, resulting in a construction feature set containing entity identification information and event timing information. The construction feature set is semantically aligned with a pre-set engineering specification knowledge base to generate specification association information between the construction stage description unit and engineering specification requirements. This involves: extracting a set of specification entries related to construction object entities and construction resource entities from the engineering specification knowledge base, including construction technical requirements, resource allocation standards, and environmental safety regulations; extracting semantic features from the entity identifier information in the construction feature set to generate entity semantic vectors representing entity types and attributes; extracting semantic features from the specification content in the specification entry set to generate specification semantic vectors representing specification requirements and applicable conditions; calculating the semantic similarity value between the entity semantic vectors and the specification semantic vectors, and selecting specification entries whose semantic similarity value with the entity identifier information in the construction feature set exceeds a preset threshold as associated specification entries; and combining the event time sequence information in the construction feature set to analyze the applicable time range of the associated specification entries in the construction stage description unit, determining the time matching relationship between the execution sequence of construction activities and the time points of state changes and specification requirements, and generating specification association information containing associated specification entries and time matching relationships. The hierarchical template matching algorithm is used to perform structured integration of the specification-related information to generate a structured report text that meets the requirements of engineering management. The structured report text includes a description of the construction progress, specification compliance analysis, and recommendations.

2. The method for generating engineering management reports based on natural language processing according to claim 1, characterized in that, The collection of construction log texts generated during the project construction process includes: Raw construction log data is collected from the engineering management information system. The raw construction log data contains multiple segments of unstructured text records indexed by timestamps. The original construction log data is cleaned to filter out text segments containing duplicate content, punctuation marks with incorrect formatting, and redundant content unrelated to construction, resulting in preliminary cleaned candidate text of the construction log. The candidate text of the construction log is segmented according to a preset time interval, and the text content describing the same construction stage within a continuous time range is merged into a construction stage description unit. The time range of the construction stage description unit is consistent with the stage division rules in the engineering construction plan. Each construction stage description unit is split into two parts: a construction activity description statement describing the construction activities and a site status description statement recording the site environment and equipment status. This ensures that each construction stage description unit contains at least one construction activity description statement and one site status description statement. The text integrity of the split construction activity description statements and site status description statements is checked, and a set of construction stage description units containing complete semantics is generated as a construction log text set.

3. The method for generating engineering management reports based on natural language processing according to claim 1, characterized in that, The construction log text set is subjected to natural language processing to extract key entities and progress events from each construction stage description unit, resulting in a construction feature set containing entity identification information and event timing information, including: The construction activity description statements and site status description statements in the construction log text set are segmented into words to obtain a segmentation result set composed of basic vocabulary units. The pre-trained BERT language model is invoked to perform context semantic encoding on the word segmentation result set, generating a context embedding vector for each basic vocabulary unit. The context embedding vector contains the semantic association information of the word in the corresponding sentence. Entity recognition processing is performed based on the context embedding vector to identify construction object entities, construction resource entities, and construction technology entities in the construction activity description statement, and to identify environmental status entities, equipment status entities, and personnel status entities in the site status description statement, resulting in a set of entity identifiers containing entity type and entity location information. The temporal relationship analysis is performed on the time adverbs and sequence conjunctions in the construction stage description unit to determine the execution order between the construction activity description statements and the temporal correspondence between the on-site status description statements and the construction activities, and to generate event timing information that represents the execution order of construction activities and the timing of status changes. The entity identifier set is associated and integrated with the event timing information to generate a construction feature set containing entity identifier information and event timing information. The entity identifier information includes entity type, entity name and the position of the statement it is located in, and the event timing information includes the event execution order and the time point of the state change.

4. The method for generating engineering management reports based on natural language processing according to claim 1, characterized in that, The hierarchical template matching algorithm is used to perform structured integration processing on the specification-related information to generate a structured report text that meets the requirements of engineering management. The structured report text includes a construction progress description, specification compliance analysis, and recommendations, including: Select a hierarchical template structure from the preset engineering management report template library. The hierarchical template structure includes a first-level heading layer, a second-level content layer, and a third-level detail layer. The first-level heading layer corresponds to the construction progress description, specification compliance analysis, and recommendations. Map the event sequence information in the construction feature set to the construction progress description template area of ​​the first-level heading layer, and organize the timeline description of construction activities according to the event execution order and the time point of state change. The associated standard entries and time matching relationships in the aforementioned standard association information are mapped to the standard compliance analysis template area of ​​the first-level heading layer. The execution status of construction activities is compared with the time matching relationship of standard requirements to generate an analysis conclusion on whether the construction activities comply with the standard requirements. Based on the analysis conclusions of the specification compliance analysis template area, improvement measures suggestions for non-compliance and optimization suggestions for compliance are extracted from the engineering specification knowledge base and mapped to the suggestion content template area of ​​the first-level heading layer; Semantic description optimization is performed on the second-level content layer and the third-level detail layer. A transfer learning strategy is used to adjust the general description statements in the template to optimized description statements that conform to the semantic style of the construction log text set, so that the language expression of the structured report text is consistent with the semantic output state of the original construction log, and a structured report text containing construction progress description, specification compliance analysis and recommendations is generated. The semantic description optimization process for the second-level content layer and the third-level detail layer involves using a transfer learning strategy to adjust the general description statements in the template to optimized description statements that conform to the semantic style of the construction log text set. This ensures that the language expression of the structured report text remains consistent with the semantic output of the original construction log, including: Extract frequent term vocabulary and key sentence structures from the construction log text set to generate a language feature template representing the semantic style of the original construction log; A pre-trained transfer learning model is selected, the language feature template is used as the target style for style transfer, and the general descriptive statements in the language feature template are used as the input text. The transfer learning model is fine-tuned and trained so that it learns to convert general descriptive statements into optimized descriptive statements that conform to the language feature template. Input the general descriptive statements from the second-level content layer and the third-level detail layer into the fine-tuned transfer learning model to generate optimized descriptive statements after style transfer. The optimized description statement is subjected to syntax correctness verification and semantic coherence check to correct syntax errors and semantic ambiguities generated during the migration process, resulting in an optimized description statement that is consistent with the semantic output state of the original construction log.

5. The method for generating engineering management reports based on natural language processing according to claim 2, characterized in that, The original construction log data undergoes data cleaning processing, filtering out text segments containing duplicate content, incorrectly formatted punctuation marks, and redundant content unrelated to construction, resulting in preliminary cleaned candidate text for the construction log, including: The original construction log data is subjected to duplicate text detection processing. The text similarity value between adjacent text segments is calculated. Text segments with similarity values ​​exceeding a preset threshold are identified as duplicate content and one of them is retained. The punctuation marks in the original construction log data are formatted and verified. Full-width and half-width marks are unified to the full-width punctuation format corresponding to the project management, and multiple consecutive punctuation marks are corrected to a single standard punctuation mark. The text content in the original construction log data is subjected to topic relevance analysis, and a set of keywords is extracted from the text content. The set of keywords is matched with a preset construction topic keyword library, and text segments with keyword matching degree exceeding a preset threshold are selected as valid content related to construction. The detected non-duplicate text segments, formatted text segments, and valid text segments related to construction are merged to generate preliminary cleaned candidate construction log texts. These candidate construction log texts do not contain duplicate content, formatted punctuation errors, or redundant content unrelated to construction.

6. The method for generating engineering management reports based on natural language processing according to claim 3, characterized in that, The entity recognition processing based on the context embedding vector identifies construction object entities, construction resource entities, and construction technology entities in the construction activity description statement, and environmental status entities, equipment status entities, and personnel status entities in the site status description statement, resulting in an entity identifier set containing entity type and entity location information, including: The context embedding vector is subjected to sequence labeling, and each basic vocabulary unit is labeled with an entity type label. The entity type labels include construction object entity labels, construction resource entity labels, construction technology entity labels, environmental status entity labels, equipment status entity labels, and personnel status entity labels. Entity boundary recognition processing is performed on the labeled lexical units, and lexical units with consecutive labels of the same entity type are merged to form a complete entity name representation. Record the start and end positions of each entity name in the construction activity description statement or site status description statement, and generate a character index range representing the entity's position; The identified entity names are subjected to semantic consistency verification, and different representations of the same entity are unified into the standard entity names in the engineering specification knowledge base. Integrate entity type tags, standard entity names, and character index ranges to generate a set of entity identifiers that includes entity type and entity location information.

7. The method for generating engineering management reports based on natural language processing according to claim 1, characterized in that, The step of calculating the semantic similarity value between the entity semantic vector and the specification semantic vector, and selecting specification entries whose semantic similarity value with the entity identification information in the construction feature set exceeds a preset threshold as associated specification entries, includes: Perform dimension alignment processing on the entity semantic vector and the canonical semantic vector; The cosine of the angle between the entity semantic vector after dimensional alignment and the canonical semantic vector is used as the semantic similarity value. The semantic similarity values ​​are normalized to obtain normalized similarity values; Set a semantic similarity threshold and filter out normative entries with normalized similarity values ​​greater than the semantic similarity threshold as candidate associated normative entries; Redundancy filtering is performed on the candidate associated specification entries to remove duplicate specification entries associated with the same entity identifier information, and the specification entry with the highest semantic similarity value is retained as the associated specification entry.

8. The method for generating engineering management reports based on natural language processing according to claim 1, characterized in that, After generating the structured report text that meets the requirements of engineering management, the following is also included: Perform a content integrity check on the construction progress description report area in the structured report text to check whether it contains the event execution sequence and status change time points of all construction stage description units; The logical rationality of the compliance analysis report area in the structured report text is verified to check whether the analysis conclusions are based on the related standard entries and time matching relationships in the standard association information. The feasibility of the suggestions in the structured report text is verified, and it is checked whether the improvement measures and optimization suggestions have corresponding normative basis in the engineering specification knowledge base. Correct any missing information, logical inconsistencies, or unfounded suggestions found during the verification process in the report areas, supplement missing progress information, adjust contradictory analysis conclusions, and replace unfounded suggestions. Generate an updated structured report text that has undergone content integrity verification and correction.

9. An engineering management report generation system, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of any one of the methods described in claims 1 to 8.

Images