Data analysis method and device, equipment, medium and product

Abstract

The invention relates to the technical field of buildings, and discloses a data analysis method and device, equipment, a medium and a product, and the method comprises the steps: obtaining a to-be-analyzed building data list and a building knowledge graph; analyzing description information in the building data list one by one by utilizing a preset building description model to obtain a plurality of list descriptions corresponding to the building data list; utilizing the building description model to extract construction element information matched with each list description and element attributes corresponding to the construction element information from the building knowledge graph; and performing structured processing on the construction element information and the element attributes to obtain a structured analysis result of the building data list. According to the method, the building description model and the knowledge graph are utilized to perform data analysis on the building data list, and the structured analysis result of the building data list can be quickly and accurately obtained, so that the efficiency and the accuracy of data analysis are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of construction technologies, and particularly to a data parsing method, apparatus, device, medium and product. Background Art

[0002] In the construction industry, a construction data list (such as a construction consumables list) is closely related to the cost of a construction project, and is also the core basis for project cost control, risk assessment and decision-making analysis. Therefore, accurately, comprehensively and efficiently parsing and managing the list content plays a crucial role in ensuring the successful completion of a project on schedule, with high quality and within budget.

[0003] Since the content of the list is diverse, the structure is complex, and there are many special writing rules for users, it is necessary to structure the data for the list to realize the use of construction data. Currently, the main structured parsing method used is to first use a named entity recognition algorithm to identify construction elements such as components and materials and their corresponding attributes, and then use a rule-based method to standardize the named entities into structured information. However, this structured parsing method still faces some challenges and limitations, resulting in relatively low data parsing efficiency and accuracy. Summary of the Invention

[0004] In view of this, the present disclosure provides a data parsing method, apparatus, device, medium and product to solve the problem of low data parsing efficiency and accuracy.

[0005] In a first aspect, the present disclosure provides a data parsing method, which includes:

[0006] Obtain a construction data list to be parsed and a construction knowledge graph;

[0007] Use a preset construction description model to parse the description information in the construction data list item by item to obtain multiple list descriptions corresponding to the construction data list;

[0008] Use the construction description model to extract construction element information and corresponding element attributes of the construction element information that match each list description from the construction knowledge graph;

[0009] Perform structured processing on the construction element information and the element attributes to obtain a structured parsing result of the construction data list.

[0010] In an embodiment of the present disclosure, by obtaining a list of building data to be parsed and a building knowledge graph; using a preset building description model to parse the description information in the list of building data item by item to obtain multiple list descriptions corresponding to the list of building data; using the building description model to extract construction element information and the corresponding element attributes of the construction element information that match each list description from the building knowledge graph; performing structured processing on the construction element information and the element attributes to obtain a structured parsing result of the list of building data. Since the embodiment of the present disclosure uses the building description model and the knowledge graph to perform data parsing on the list of building data, it can quickly and accurately obtain the structured parsing result of the list of building data, thereby improving the efficiency and accuracy of data parsing.

[0011] In an alternative embodiment, using a preset building description model to parse the description information in the list of building data item by item to obtain multiple list descriptions corresponding to the list of building data includes:

[0012] Using the building description model to parse the first working data in the list of building data to generate the first working description information in the list of building data, where the first working description information includes working part description information and working object description information;

[0013] Using the building description model to filter the list of building data according to a preset filtering rule to obtain second working data;

[0014] Using the building description model to parse the third working data in the second working data to generate second working description information;

[0015] Concatenating the first working description information and the second working description information to generate a list description.

[0016] In an embodiment of the present disclosure, by using the building description model to parse the list of building data item by item and concatenating the parsing results, it is possible to accurately obtain a complete list description, thereby improving the accuracy and integrity of data parsing.

[0017] In an alternative embodiment, using the building description model to extract construction element information and the corresponding element attributes of the construction element information that match each list description from the building knowledge graph includes:

[0018] Comparing the similarity of each list description with the building knowledge graph and extracting construction element information that matches each list description from the building knowledge graph;

[0019] Using the building description model to parse the construction element information to obtain the target element corresponding to the list description and the element attributes corresponding to the target element.

[0020] In the embodiments of the present disclosure, by extracting construction element information and the corresponding element attributes of the construction element information from the building knowledge graph through similarity comparison, the parsing scope of the target element can be narrowed, thereby improving the efficiency of data parsing.

[0021] In an alternative embodiment, the similarity between each list description and the building knowledge graph is compared, and the construction element information matching each list description is extracted from the building knowledge graph, including:

[0022] For any list description, determine the similarity between the list description and each data text in the building knowledge graph;

[0023] According to the similarity, extract a preset number of candidate construction text information with the highest similarity to the list description from the building knowledge graph. The candidate construction text information includes candidate construction work information and candidate construction part information;

[0024] Use the building description model to screen the candidate construction text information to obtain the construction element information, which includes the target construction work and the target construction part.

[0025] In the embodiments of the present disclosure, by extracting a preset number of candidate construction text information with the highest similarity from the building knowledge graph according to the similarity between the list description and each data text in the building knowledge graph, the parsing scope of the target element can be narrowed, thereby improving the efficiency of data parsing.

[0026] In an alternative embodiment, use the building description model to parse the construction element information to obtain the target element corresponding to the list description and the element attributes corresponding to the target element, including:

[0027] Obtain multiple candidate elements corresponding to the target construction work;

[0028] Use the building description model to identify the candidate elements and determine the target elements existing in the list description;

[0029] Obtain multiple candidate attribute names corresponding to the target element;

[0030] Use the building description model to identify the candidate attribute names and determine the element attributes existing in the list description.

[0031] In the embodiments of the present disclosure, by using the building description model to identify the candidate elements and candidate attribute names, the target elements and element attributes existing in the list description can be quickly and accurately determined, improving the efficiency and accuracy of data parsing.

[0032] In an alternative embodiment, it further includes:

[0033] Obtain application feedback data for the structured parsing result and the error reason corresponding to the application feedback data;

[0034] Update the building knowledge graph and / or the building description model according to the error reason.

[0035] In the embodiments of the present disclosure, by optimizing and updating the knowledge graph and / or the building description model according to the error reason of the application feedback data, the coverage rate of the knowledge graph for building data can be improved, and the parsing ability of the building description model for the building data list can be enhanced, thereby improving the quality of the structured parsing result of the building data list.

[0036] In an alternative embodiment, updating the building knowledge graph and / or the building description model according to the error reason includes:

[0037] If the error reason has nothing to do with the building knowledge graph, modify the prompt information of the building description model;

[0038] Optimize the building description model according to the modified prompt information;

[0039] And / or,

[0040] If the error reason is related to the building knowledge graph, update the node elements of the building knowledge graph;

[0041] Optimize the building knowledge graph according to the modified node elements.

[0042] In the embodiments of the present disclosure, by optimizing the building description model according to the modified prompt information, the parsing ability of the building description model for the building data list can be enhanced, and by optimizing the building knowledge graph according to the modified node elements, the coverage rate of the knowledge graph for building data can be improved, thereby improving the quality of the structured parsing result of the building data list.

[0043] In an alternative embodiment, obtaining the building knowledge graph includes:

[0044] Obtain building engineering data;

[0045] Parse the building engineering data to determine the first building elements at different levels;

[0046] Based on the hierarchical positions of the first building elements, determine the second building elements and the third building elements attached to the first building elements, and the second building elements do not have attribute information;

[0047] Based on the hierarchical positions of the third building elements, determine the attribute information of the third building elements;

[0048] Perform knowledge fusion based on the first building elements, the second building elements, the third building elements, and the attribute information to generate a knowledge graph.

[0049] In the disclosed embodiment, by gradually parsing and integrating the construction project data, a knowledge graph with a clear structure and rich information can be generated, thereby improving the accuracy and efficiency of data parsing.

[0050] In a second aspect, the present disclosure provides a data parsing device, the device comprising:

[0051] The first acquisition module is used to obtain the list of building data to be parsed and the building knowledge graph;

[0052] A parsing module, used to parse the description information in the building data list one by one by using a preset building description model to obtain multiple list descriptions corresponding to the building data list;

[0053] An extraction module is used to extract construction element information matching each list description and element attributes corresponding to the construction element information from the building knowledge graph using the building description model;

[0054] The module is used to perform structured processing on the construction element information and element attributes to obtain the structured parsing result of the building data list.

[0055] In a third aspect, the present disclosure provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the data parsing method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0056] In a fourth aspect, the present disclosure provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the data parsing method of the first aspect or any corresponding embodiment thereof.

[0057] In a fifth aspect, the present disclosure provides a computer program product, including computer instructions, which are used to enable a computer to execute the data analysis method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the specific embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0059] Figure 1 is a flowchart of a data parsing method according to an embodiment of the present disclosure;

[0060] Figure 2 is a schematic diagram of a knowledge graph according to an embodiment of the present disclosure;

[0061] Figure 3 is a schematic flowchart of a data parsing method according to another embodiment of the present disclosure;

[0062] Figure 4 is a schematic flowchart of a data parsing method according to still another embodiment of the present disclosure;

[0063] Figure 5 is a schematic flowchart of a data parsing method according to yet another embodiment of the present disclosure;

[0064] Figure 6 is a block diagram of the structure of a data parsing device according to an embodiment of the present disclosure;

[0065] Figure 7 is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present disclosure. Detailed implementation manners

[0066] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.

[0067] In the construction industry, a building data list (such as a building consumables list) is closely related to the cost of a building construction project and is also the core basis for project cost control, risk assessment, and decision-making analysis. Therefore, accurately, comprehensively, and efficiently parsing and managing the list content is crucial for ensuring the smooth completion of the project on schedule, with high quality, and within budget.

[0068] Due to the diverse content, complex structure, and many special writing rules of users of the list, it is necessary to perform data structuring on the list to realize the use of building data. Currently, the main structured parsing method used is to first use a named entity recognition algorithm to identify building elements such as components and materials and their corresponding attributes, and then use a rule-based method to standardize the named entities into structured information.

[0069] However, this structured parsing method still faces some challenges and limitations, resulting in relatively low data parsing efficiency and accuracy. First, the named entity recognition algorithm requires a large amount of labeled data to train the model, and it is necessary to maintain the thesaurus and regular expressions of the feature library, which will consume a large amount of labor and time costs. Second, the named entity recognition algorithm requires a large number of complex manual rules to correct the entity results, and these rules may conflict with each other, restricting the accuracy. Third, when the named entity recognition algorithm model and manual rules encounter new data, the accuracy of the output results cannot be guaranteed. In addition, the feature library cannot cover all the content expressed in the list, such as construction work, etc.

[0070] To solve the above problems, according to an embodiment of the present disclosure, an embodiment of a data parsing method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0071] In this embodiment, a data parsing method is provided, as Figure 1 shown Figure 1 is a schematic flowchart of a data parsing method according to an embodiment of the present disclosure. This process can be applied to computer devices such as servers, and includes the following steps:

[0072] Step S101, obtain a building data list to be parsed and a building knowledge graph.

[0073] The building data list is usually a list of some actual engineering data, such as: cost list, progress list, quality list, etc. Specifically, the user can create a building data list in the corresponding style in the computer device according to actual needs and save it. When it is necessary to parse the building data list, the computer device can access the storage location of the building data list and obtain the corresponding building data list from this storage location. Of course, it can also be obtained in real time. For example, a fixed time point is set, and when the time reaches this fixed time point, the currently obtained building data list is automatically pushed to the computer device, and then the computer device can automatically obtain this building data list. There is no specific limitation here.

[0074] Among them, the building data list can include three types of content: list superior, list name, and list features. The list superior can include division information, sub-item information, and location information. For example: Civil Engineering \ Building 1 in XX Community \ Wall and Column Surface Decoration Project. The list name is a summary of the entire list description content and can include the work object and work content. For example: Interior 1: Cement Mortar Wall Surface (Combustion Performance Class A). The list features are the detailed content of the list and can include information such as work object, work content, work materials, and work location. For example: Residential bedrooms, living rooms, dining rooms, attic, kitchens, bathrooms, electrical and plumbing wells, machine rooms, storage rooms, inside elevator shafts, inside louvers, inner sides of parapet walls, duty rooms, 1.5 mm thick 1:2.5 cement mortar for plastering, 2.8 mm thick 1:1:6 cement-lime plaster mortar for undercoating and scratching or scoring lines, 3.3 mm thick special mortar with admixture for undercoating and roughing or a special interface agent for one-time splashing and fluffing. For details, refer to 05J909-NQ16-8D (indicating that construction workers can refer to the specific content of 8D in the interior wall finish practice of NQ16 in the 05J909 atlas for corresponding construction operations).

[0075] In addition, as Figure 2 shown, the building knowledge graph is a knowledge graph in the construction field constructed based on construction project data. This construction project data is usually some actual project data, such as: cost list, cost item, schedule plan, etc. The building knowledge graph can be constructed by a method based on rules and templates, or by a model architecture based on machine learning, or by integrating a method based on rules and templates and a model architecture based on machine learning. No specific limitation is made here.

[0076] Step S102: Use a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list.

[0077] The preset building description model is used to implement the natural language description of the building data list. Specifically, the building description model can be trained by a model architecture based on a large language model, or by a model architecture based on machine learning, or by a model architecture based on a neural network. No specific limitation is made here.

[0078] Taking the building description model trained with a large language model-based model architecture as an example, the computer device first inputs the building data list and text extraction instructions (prompt words for describing the extraction of text information in the building data list) into the large language model, uses the large language model to perform overall semantic understanding of the building data list, extracts various text information in the building data list according to the text extraction instructions and outputs it, and then inputs the various text information and text splicing instructions (prompt words for describing the splicing of various text information) into the large language model, uses the large language model to splice the various text information according to the text splicing instructions and output it item by item, and obtains multiple list descriptions corresponding to the building data list.

[0079] Step S103: Use the building description model to extract construction element information that matches each list description from the building knowledge graph, as well as the element attributes corresponding to the construction element information.

[0080] Such as Figure 2 shown, the construction element information may include construction work (such as decoration engineering, wall waterproofing, spraying of wall waterproof materials, etc.), construction location (such as space - bathroom), work object (such as wall), work content (such as spraying of waterproof coating), and work materials (such as waterproof coating), etc. The element attributes correspond to the construction element information and include attribute names and attribute values (such as wall - wall material - cast - in - place concrete, waterproof coating - thickness - 1mm, etc.).

[0081] Taking the building description model trained with a large language model-based model architecture as an example, the computer device traverses each list description, inputs each list description and construction element extraction instructions (prompt words for describing the extraction of construction elements in the list description) into the large language model respectively, uses the large language model to extract multiple construction element information of each list description according to the construction element instructions and outputs it, and then inputs the multiple construction element information and element attribute extraction instructions (prompt words for describing the extraction of element attributes in the construction element information) into the large language model, uses the large language model to extract the element attributes of the multiple construction element information according to the element attribute extraction instructions and outputs it, and obtains construction element information and element attributes that match each list description.

[0082] Step S104: Perform structured processing on the construction element information and element attributes to obtain a structured parsing result of the building data list.

[0083] After the computer device obtains the construction element information and element attributes, it integrates the construction element information and element attributes corresponding to each list description in a structured manner, and stores the integrated result in a corresponding structured database (such as a relational database, a non - relational database, a graph database, etc.) to obtain the final structured parsing result of the building data list.

[0084] In an embodiment of the present disclosure, by obtaining a building data list to be parsed and a building knowledge graph; using a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list; using the building description model to extract construction element information and the corresponding element attributes of the construction element information that match each list description from the building knowledge graph; performing a structured process on the construction element information and the element attributes to obtain a structured parsing result of the building data list. Obtain the building data list to be parsed and the building knowledge graph. Since the embodiment of the present disclosure uses the building description model and the knowledge graph to perform data parsing on the building data list, it can quickly and accurately obtain the structured parsing result of the building data list, thereby improving the efficiency and accuracy of data parsing.

[0085] In some alternative embodiments, this embodiment provides a data parsing method, as Figure 3 shown, Figure 3 is a schematic flowchart of a data parsing method according to another embodiment of the present disclosure. This process can be applied to computer devices, such as servers, etc., and includes the following steps:

[0086] Step S201, obtain a building data list to be parsed and a building knowledge graph. For details, please refer to Figure 1 step S101 of the embodiment shown, which will not be elaborated here.

[0087] Step S202, use a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list.

[0088] Specifically, the above step S202 includes:

[0089] Step S2021, use the building description model to parse the first work data in the building data list to generate the first work description information in the building data list. The first work description information includes work location description information and work object description information.

[0090] The first work data includes work location data and work object data. The first work description information includes work location description information (such as the third bid section of the building construction project (FJSG-3)\YY toll station\comprehensive building\main body of the comprehensive building\stairwell, electrical room) and work object description information (such as wall surface).

[0091] Specifically, the computer device inputs the building data list and text parsing instructions (prompt words used to describe the parsing of the first work data in the building data list) into the large language model, uses the large language model to perform overall semantic understanding of the building data list, parses and outputs the first work data in the building data list according to the text parsing instructions, and obtains the first work description information in the building data list.

[0092] In a specific example, the computer device inputs the building data list and text parsing instructions obtained above: "View the list superior, list name, and list features, list all text descriptions regarding the work area without omission", "Combine the list superior, list name, and list features to write the work object of this list item" into the large language model. By parsing and outputting the work part data and work object data through the large language model, work part description information and work object description information can be obtained: "Bedrooms, living rooms, dining rooms, attics, kitchens, bathrooms, water and electricity well rooms, storage rooms, inside elevator shafts, inside wall surfaces of louvers, inner sides of parapets, inner wall surfaces of duty rooms (combustion performance class A) in Building 1 of XX Community".

[0093] Step S2022: Use the building description model to filter the building data list according to the preset filtering rules to obtain the second work data.

[0094] The second work data is the work data after filtering the building data list. The preset filtering rules are filtering rules preset according to the actual application scenario. For example, in the description in the list features: "4. 1.5 - mm thick polymer cement waterproof coating (Type II) (listed separately)", where "listed separately" means that this item is not actually included in the current building data list, so it needs to be filtered out.

[0095] Specifically, the computer device inputs the building data list and the preset filtering rules into the large language model, uses the large language model to filter the building data list according to the preset filtering rules and output, and obtains the second work data in the building data list.

[0096] Step S2023: Use the building description model to parse the third work data in the second work data to generate the second work description information.

[0097] The third work data in the second work data refers to the specific work descriptions in the filtered building data list. The second work description information can include work content description information, work material description information, etc.

[0098] Specifically, the computer device inputs the second working data and a text parsing instruction (a prompt word used to describe the parsing of the third working data in the second working data) into the large language model, and uses the large language model to parse and output the third working data in the second working data item by item according to the text parsing instruction, obtaining multiple pieces of second working description information in the building data list.

[0099] In a specific example, the computer device inputs the filtered second working data and the text parsing instruction into the large language model, and the large language model parses and outputs specific working descriptions such as working content description and working material description item by item, obtaining multiple pieces of specific working description information: "smoothing with cement mortar", "plastering the base layer with cement-lime mortar and sweeping it rough", and "plastering the base layer with special mortar and scraping it rough".

[0100] Step S2024: Concatenate the first working description information and the second working description information to generate a list description.

[0101] The computer device inputs the first working description information, each piece of second working description information, and a text concatenation instruction (a prompt word used to describe the concatenation of the first working description information and each piece of second working description information respectively) into the large language model, and uses the large language model to concatenate and output the first working description information and each piece of second working description information item by item according to the text concatenation instruction, obtaining multiple list descriptions corresponding to the building data list.

[0102] In a specific example, the computer device inputs the first working description information, multiple pieces of second working description information, and the text concatenation instruction into the large language model. By concatenating the first working description information and multiple pieces of second working description information through the large model, multiple list descriptions corresponding to the building data list can be obtained: "Interior walls of bedrooms, living rooms, dining rooms, attics, kitchens, bathrooms, electrical and water pump rooms, storage rooms, elevator shafts, inside shutters, inner sides of parapets, and inside the duty room (combustion performance A level) of Building 1 in XX Community \\ Smoothing with cement mortar", "Interior walls of bedrooms, living rooms, dining rooms, attics, kitchens, bathrooms, electrical and water pump rooms, storage rooms, elevator shafts, inside shutters, inner sides of parapets, and inside the duty room (combustion performance A level) of Building 1 in XX Community \\ Plastering the base layer with cement-lime mortar and sweeping it rough", and "Interior walls of bedrooms, living rooms, dining rooms, attics, kitchens, bathrooms, electrical and water pump rooms, storage rooms, elevator shafts, inside shutters, inner sides of parapets, and inside the duty room (combustion performance A level) of Building 1 in XX Community \\ Plastering the base layer with special mortar and scraping it rough".

[0103] Step S203: Use the building description model to extract construction element information and the corresponding element attributes that match each list description from the building knowledge graph. For details, please refer to Figure 1Step S103 of the illustrated embodiment will not be elaborated here.

[0104] Step S204: Structurally process the construction element information and element attributes to obtain the structural analysis result of the building data list. For details, please refer to Figure 1 Step S104 of the illustrated embodiment will not be elaborated here.

[0105] In the embodiments of the present disclosure, by parsing each item in the building data list using the building description model and splicing the parsing results, the complete list description can be accurately obtained, thereby improving the accuracy and integrity of data parsing.

[0106] In some alternative embodiments, the present embodiment provides a data parsing method, as Figure 4 shown, Figure 4 is a schematic flowchart of a data parsing method according to another embodiment of the present disclosure. This process can be applied to computer devices such as servers and includes the following steps:

[0107] Step S301: Obtain the building data list to be parsed and the building knowledge graph. For details, please refer to Figure 3 Step S201 of the illustrated embodiment will not be elaborated here.

[0108] Step S302: Use the preset building description model to parse each description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list. For details, please refer to Figure 3 Step S202 of the illustrated embodiment will not be elaborated here.

[0109] Step S303: Use the building description model to extract the construction element information and the corresponding element attributes of the construction element information that match each list description from the building knowledge graph.

[0110] Specifically, the above step S303 includes:

[0111] Step S3031: Compare the similarity of each list description with the building knowledge graph, and extract the construction element information that matches each list description from the building knowledge graph.

[0112] Among them, the construction element information includes construction work information (such as decoration engineering, wall waterproofing, spraying of wall waterproof materials, etc.) and construction location information (such as space - bathroom, etc.).

[0113] Specifically, before comparing the similarity between each list description and the construction knowledge graph, the computer device first uses a text embedding model to convert each list description and the data text in the construction knowledge graph into numerical vectors. Among them, the text embedding model can be a traditional model such as a bag-of-words model or a term frequency-inverse document frequency model, or a model trained based on a neural network, or a pre-trained language model, and no specific limitation is made here.

[0114] Next, the computer device calculates the similarity between each list description vector and the construction knowledge graph vector respectively, and then compares the magnitudes of the similarities to obtain the construction element information that matches each list description. Among them, the similarity calculation method can be cosine similarity or Euclidean distance. The similarity can be character similarity or semantic similarity.

[0115] In some optional embodiments, the above step S3031 includes:

[0116] Step a1, for any list description, determine the similarity between the list description and each data text in the construction knowledge graph.

[0117] Step a2, extract a preset number of candidate construction text information with the highest similarity to the list description from the construction knowledge graph according to the similarity. The candidate construction text information includes candidate construction work information and candidate construction part information.

[0118] Step a3, use the construction description model to screen the candidate construction text information to obtain the construction element information, where the construction element information includes the target construction work and the target construction part.

[0119] The computer device first calculates the similarity between each list description and each data text in the construction knowledge graph, and respectively obtains a preset number of candidate construction text information with the highest similarity to each data text in the construction knowledge graph according to the magnitude relationship of the similarities. Among them, the preset number refers to the number of candidate construction text information, and the candidate construction text information includes candidate construction work information and candidate construction part information.

[0120] Then, the computer device traverses each list description, and inputs each list description, multiple candidate construction text information, and a text screening instruction (used to describe the prompt words for screening the elements existing in each list description in the candidate construction text information) into the large language model, and uses the large language model to screen the elements existing in each list description in the multiple candidate construction text information according to the text screening instruction and output, to obtain the construction element information. Among them, the construction element information includes the target construction work and the target construction part.

[0121] For example, for the bill of quantities description: "Decoration project of Building 1-1# in Plot A02 \\ Toilets in public buildings and residential parts \\ Roughcast wall surface (waterproof) (aerated block wall surface) \\ Thinned M15 cement mortar", the computer device calculates the similarity between the bill of quantities description and each data text in the building knowledge graph to obtain multiple candidate construction text information, and then inputs the bill of quantities description, multiple candidate construction text information, and text screening instructions into the large language model for screening, obtaining the target construction parts: "Single building (Building 1-1#)" and "Space (toilet)".

[0122] In the above implementation, by extracting the preset number of candidate construction text information with the highest similarity from the building knowledge graph according to the similarity between the bill of quantities description and each data text in the building knowledge graph, the parsing range of the target elements can be narrowed, thereby improving the efficiency of data parsing.

[0123] Step S3032, use the building description model to parse the construction element information to obtain the target elements corresponding to the bill of quantities description and the element attributes corresponding to the target elements.

[0124] Among them, the target elements are the other elements in the bill of quantities description except for the construction element information.

[0125] Specifically, the computer device traverses each bill of quantities description, and inputs each bill of quantities description, construction element information, and text parsing instructions (prompt words used to describe the parsing of the other elements and corresponding element attributes in each bill of quantities description except for the construction element information) into the large language model, and uses the large language model to parse and output the other elements and corresponding element attributes in each bill of quantities description except for the construction element information according to the text parsing instructions, obtaining the target elements corresponding to each bill of quantities description and the element attributes corresponding to the target elements.

[0126] In some optional implementation manners, the above step S3032 includes:

[0127] Step b1, obtain multiple candidate elements corresponding to the target construction work.

[0128] Step b2, use the building description model to identify the candidate elements to determine the target elements existing in the bill of quantities description.

[0129] Step b3, obtain multiple candidate attribute names corresponding to the target elements.

[0130] Step b4, use the building description model to identify the candidate attribute names to determine the element attributes existing in the bill of quantities description.

[0131] The computer device first obtains multiple candidate elements corresponding to the target construction work. For example, if the target construction work is the concrete pouring of a structural wall, the corresponding candidate elements may include the work object (structural wall), the work content (concrete pouring), and the work material (concrete), etc.

[0132] Then, the computer device traverses each list description, and inputs each list description, the multiple candidate elements, and the text recognition instruction (used to describe the prompt words for recognizing the elements existing in each list description among the multiple candidate elements) into the large language model. The large language model is used to recognize the elements existing in each list description among the multiple candidate elements according to the text recognition instruction and output them to obtain the target elements.

[0133] Next, the computer device obtains multiple candidate attribute names corresponding to the target elements. For example, if the target element is the concrete in the work material, the corresponding candidate attribute name may be the strength grade.

[0134] After that, the computer device traverses each list description, and inputs each list description, the multiple candidate attribute names, and the text recognition instruction (used to describe the prompt words for recognizing the attribute names and their corresponding attribute values existing in each list description among the multiple candidate attribute names) into the large language model. The large language model is used to recognize the attribute names and their corresponding attribute values existing in each list description among the multiple candidate attribute names according to the text recognition instruction and output them to obtain the element attributes corresponding to the target elements. Among them, the element attributes corresponding to the target elements include the target attribute name and the target attribute value. For example, if the target element is the concrete in the work material, the corresponding target attribute name may be the strength grade, and the attribute value may be C30.

[0135] In the above embodiments, by using the building description model to recognize the candidate elements and candidate attribute names, the target elements and element attributes existing in the list description can be quickly and accurately determined, improving the efficiency and accuracy of data parsing.

[0136] Step S304: Perform structured processing on the construction element information and element attributes to obtain the structured parsing result of the building data list. For details, please refer to Figure 3 Step S204 of the illustrated embodiment, which will not be elaborated here.

[0137] Step S305: Obtain the application feedback data for the structured parsing result and the error reason corresponding to the application feedback data.

[0138] After obtaining the structured parsing result of the building data list, the computer device applies it to downstream tasks (such as data connection, etc.) and receives application feedback data. Then, by analyzing the application feedback data through business experts, comparing the structured parsing result of the building data list with the expected result of the downstream task, it is possible to locate which type of elements are deviated. Further checking the output results of each previous step, it can be found which step has deviated, and the error reason corresponding to the application feedback data can be obtained.

[0139] Step S306, update the building knowledge graph and / or the building description model according to the error reason.

[0140] The error reasons include errors related to the building knowledge graph, errors related to the large language model, and errors unrelated to both the building knowledge graph and the large language model.

[0141] Specifically, for errors related to the building knowledge graph, the computer device updates the building knowledge graph. For errors related to the large language model, the computer device optimizes the building description model. It should be noted that for errors unrelated to both the building knowledge graph and the large language model, no processing is performed in this embodiment.

[0142] In some optional implementation manners, the above step S306 includes:

[0143] Step c1, if the error reason is unrelated to the building knowledge graph, modify the prompt information of the building description model.

[0144] Step c2, optimize the building description model according to the modified prompt information.

[0145] When the error reason is unrelated to the building knowledge graph, the computer device can modify the prompt words of the large language model or use retrieval-augmented generation technology to add parsing prompts under specific user text conditions to obtain the modified prompt information.

[0146] Specifically, the computer device can adjust the content of the prompt words to describe the task in more detail. For example: "Please accurately extract the specific processes, materials used, and detailed location information of the working parts in the building data list, and pay attention to distinguishing the part descriptions at different levels." It can also add constraint conditions to the prompt words according to common error cases. For example: "Avoid misjudging the remarks information unrelated to the construction work as the work content, and only extract the key information related to the actual construction."

[0147] In addition, the computer device can also use retrieval-augmented generation technology to provide more targeted parsing hints under specific user text conditions. First, the computer device retrieves relevant information from a large amount of text data in the construction field (such as construction specification documents, text of previous construction project cases, etc.) according to the construction data list input by the user. Then, for a list containing descriptions of special construction techniques, the computer device retrieves relevant technique explanation documents through retrieval-augmented generation technology, and uses the retrieved information as hints, which are input into the large language model together with the original list text, providing more background knowledge and context information for the large language model to help it better understand the list text.

[0148] Accordingly, the computer device can retrain the building description model in a fine-tuning manner based on the modified prompt words or the added retrieved information, optimize the parameters of the building description model, and obtain an optimized building description model.

[0149] Step c3, if the error reason is related to the construction knowledge graph, update the node elements of the construction knowledge graph.

[0150] Step c4, optimize the construction knowledge graph according to the modified node elements.

[0151] In the case where the error reason is related to the construction knowledge graph, the business expert judges the missing nodes in the construction knowledge graph to obtain the corresponding nodes of the construction knowledge graph that need to be updated, and adds the nodes to be updated to the construction knowledge graph.

[0152] Accordingly, the computer device can respond to the addition operation of the nodes, add the corresponding nodes of the construction knowledge graph to be updated to the corresponding positions of the construction knowledge graph, update the construction knowledge graph, and obtain an optimized construction knowledge graph.

[0153] In the above embodiments, by optimizing the building description model according to the modified hint information, the parsing ability of the building description model for the construction data list can be improved. By optimizing the construction knowledge graph according to the modified node elements, the coverage rate of the knowledge graph for construction data can be increased, thereby improving the quality of the structured parsing results of the construction data list.

[0154] In the embodiments of the present disclosure, by extracting construction element information and the corresponding element attributes of the construction element information from the construction knowledge graph through similarity comparison, the parsing scope of the target element can be narrowed, thereby improving the efficiency of data parsing. By optimizing and updating the knowledge graph and / or the building description model according to the error reasons of the application feedback data, the coverage rate of the knowledge graph for construction data can be increased, and the parsing ability of the building description model for the construction data list can be improved, thereby improving the quality of the structured parsing results of the construction data list.

[0155] In some alternative embodiments, the present embodiment provides a data parsing method. As Figure 5 shown, Figure 5 FIG. 4 is a schematic flowchart of a data parsing method according to another embodiment of the present disclosure. This process can be applied to a computer device, such as a server, and includes the following steps:

[0156] Step S401: Obtain a list of building data to be parsed and a building knowledge graph.

[0157] Specifically, obtaining the building knowledge graph in step S401 includes:

[0158] Step S4011: Obtain construction project data.

[0159] Construction project data are usually some actual project data, such as: cost list, cost item, progress plan, etc.

[0160] Specifically, a user can create corresponding styles of construction project data in a computer device according to actual needs and save them. When it is necessary to parse the construction project data, the computer device can access the storage location of the construction project data and obtain the corresponding construction project data from that storage location. Of course, it can also be obtained in real time. For example, a fixed time point is set, and when the time reaches this fixed time point, the currently obtained construction project data is automatically pushed to the computer device, and then the computer device can automatically obtain this construction project data. There is no specific limitation here.

[0161] Step S4012: Parse the construction project data to determine first building elements at different levels.

[0162] The first building elements refer to construction work, and there is a hierarchical relationship among the construction work. As Figure 2 shown, the construction work can include decoration engineering, wall waterproofing, and wall waterproof coating spraying. Among them, decoration engineering is the superior of wall waterproofing, and wall waterproofing is the superior of wall waterproof coating spraying.

[0163] Specifically, the computer device can use natural language processing technology, data mining technology, or semantic analysis technology to parse the construction work of the construction project data and the relationships between different construction works to obtain multiple first building elements at different levels.

[0164] Step S4013: Based on the hierarchical positions of the first building elements, determine second building elements and third building elements attached to the first building elements. The second building elements do not have attribute information.

[0165] The second building elements include construction parts, etc. Among them, the construction parts include monomers, floors, spaces, etc. AsFigure 2 For the "space - toilet" shown, the third building element includes the object of work (such as a wall surface), the content of work (such as spraying waterproof coating), the materials of work (such as waterproof coating), etc. Among them, the second building element does not have attribute information, and the third building element or the accessory elements of the third building element have attribute information.

[0166] Specifically, the computer device can use natural language processing technology, data mining technology or semantic analysis technology to analyze elements such as the construction location, object of work, content of work, materials of work in the construction project data, the relationships between these elements, and the relationships between these elements and the first building element, so as to obtain the second building element and the third building element attached to the first building element.

[0167] Step S4014, determine the attribute information of the third building element based on the hierarchical position of the third building element.

[0168] The attribute information includes an attribute name and an attribute value (such as wall surface - belonging wall material - cast - in - place concrete, waterproof coating - thickness - 1mm, etc.), as Figure 2 shown.

[0169] Specifically, the computer device can use natural language processing technology, data mining technology or semantic analysis technology to analyze the attribute name and attribute value of the third building element or the accessory elements of the third building element in the construction project data, so as to obtain the attribute information.

[0170] In some specific examples, for the element of the object of work such as a structural wall, its attribute name can be thickness, and the attribute value can be "numerical value + mm". For the element of the materials of work such as ordinary concrete, its attribute name can be strength grade, and the attribute values can be C10, C15, C20, etc.

[0171] Step S4015, perform knowledge fusion based on the first building element, the second building element, the third building element and the attribute information to generate a knowledge graph.

[0172] The construction knowledge graph can be constructed by a rule - and - template - based method, or by a machine - learning - based model architecture, or by integrating a rule - and - template - based method and a machine - learning - based model architecture. There is no specific limitation here.

[0173] Specifically, based on any of the above methods or other methods, the computer device constructs corresponding nodes in the knowledge graph according to the first building element, the second building element, and the third building element, constructs the relationships between the corresponding nodes in the knowledge graph according to the relationships between the first building element, the second building element, and the third building element, and constructs the attributes of the corresponding nodes in the knowledge graph according to the attribute information, and finally obtains the building knowledge graph.

[0174] Step S402: Use a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list. For details, please refer to Figure 4 Step S302 of the embodiment shown, which will not be elaborated here.

[0175] Step S403: Use the building description model to extract the construction element information that matches each list description from the building knowledge graph and the element attributes corresponding to the construction element information. For details, please refer to Figure 4 Step S303 of the embodiment shown, which will not be elaborated here.

[0176] Step S404: Perform structured processing on the construction element information and the element attributes to obtain the structured parsing result of the building data list. For details, please refer to Figure 4 Step S304 of the embodiment shown, which will not be elaborated here.

[0177] In the embodiments of the present disclosure, by gradually parsing and integrating the building engineering data, a knowledge graph with clear structure and rich information can be generated, improving the accuracy and efficiency of data parsing.

[0178] In this embodiment, a data parsing device is further provided. This device is used to implement the above embodiments and preferred implementation manners, and those that have been described will not be elaborated again. As used below, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0179] This embodiment provides a data parsing device, as Figure 6 shown, including:

[0180] A first acquisition module 601, configured to acquire a building data list to be parsed and a building knowledge graph;

[0181] An analysis module 602, configured to use a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list;

[0182] An extraction module 603 is configured to extract construction element information that matches each list description and the element attributes corresponding to the construction element information from the building knowledge graph by using a building description model;

[0183] A obtaining module 604 is configured to perform structured processing on the construction element information and the element attributes to obtain a structured parsing result of the building data list.

[0184] In an embodiment of the present disclosure, by obtaining a building data list to be parsed and a building knowledge graph; using a preset building description model to parse the description information in the building data list item by item to obtain multiple list descriptions corresponding to the building data list; using the building description model to extract construction element information that matches each list description and the element attributes corresponding to the construction element information from the building knowledge graph; performing structured processing on the construction element information and the element attributes to obtain a structured parsing result of the building data list. Since the embodiment of the present disclosure uses the building description model and the knowledge graph to perform data parsing on the building data list, it can quickly and accurately obtain the structured parsing result of the building data list, thereby improving the efficiency and accuracy of data parsing.

[0185] In some optional implementation manners, the parsing module 602 includes:

[0186] A generating sub-module is configured to parse the first working data in the building data list by using the building description model to generate first working description information in the building data list, where the first working description information includes working position description information and working object description information;

[0187] A filtering sub-module is configured to filter the building data list according to a preset filtering rule by using the building description model to obtain second working data;

[0188] A first parsing sub-module is configured to parse the third working data in the second working data by using the building description model to generate second working description information;

[0189] A splicing sub-module is configured to splice the first working description information and the second working description information to generate a list description.

[0190] In some optional implementation manners, the extraction module 603 includes:

[0191] An extraction sub-module is configured to compare the similarity of each list description with the building knowledge graph and extract construction element information that matches each list description from the building knowledge graph;

[0192] A second parsing sub-module is configured to parse the construction element information by using the building description model to obtain target elements corresponding to the list description and the element attributes corresponding to the target elements.

[0193] In some alternative embodiments, the extraction sub-module includes:

[0194] A first determination unit configured to determine the similarity between any list description and each data text in the building knowledge graph;

[0195] An extraction unit configured to extract a preset number of candidate construction text information with the highest similarity to the list description from the building knowledge graph according to the similarity. The candidate construction text information includes candidate construction work information and candidate construction part information;

[0196] A screening unit configured to screen the candidate construction text information by using a building description model to obtain construction element information. The construction element information includes a target construction work and a target construction part.

[0197] In some alternative embodiments, the second parsing sub-module includes:

[0198] A first acquisition unit configured to acquire a plurality of candidate elements corresponding to the target construction work;

[0199] A second determination unit configured to identify the candidate elements by using a building description model to determine the target elements existing in the list description;

[0200] A second acquisition unit configured to acquire a plurality of candidate attribute names corresponding to the target elements;

[0201] A third determination unit configured to identify the candidate attribute names by using a building description model to determine the element attributes existing in the list description.

[0202] In some alternative embodiments, the apparatus further includes:

[0203] A second acquisition module configured to acquire application feedback data for the structured parsing result and the error reason corresponding to the application feedback data;

[0204] An update module configured to update the building knowledge graph and / or the building description model according to the error reason.

[0205] In some alternative embodiments, the update module includes:

[0206] A modification sub-module configured to modify the prompt information of the building description model if the error reason has nothing to do with the building knowledge graph;

[0207] A first optimization sub-module configured to optimize the building description model according to the modified prompt information;

[0208] An update sub-module configured to update the node elements of the building knowledge graph if the error reason is related to the building knowledge graph;

[0209] The second optimization sub-module is used to optimize the building knowledge graph according to the modified node elements.

[0210] In some alternative embodiments, the first acquisition module 601 includes:

[0211] An acquisition sub-module for acquiring construction project data;

[0212] A first determination sub-module for parsing the construction project data to determine the first building elements at different levels;

[0213] A second determination sub-module for determining the second building elements and the third building elements attached to the first building elements based on the hierarchical positions of the first building elements, where the second building elements do not have attribute information;

[0214] A third determination sub-module for determining the attribute information of the third building elements based on the hierarchical positions of the third building elements;

[0215] A fusion sub-module for performing knowledge fusion based on the first building elements, the second building elements, the third building elements, and the attribute information to generate a knowledge graph.

[0216] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding above embodiments, and will not be elaborated here.

[0217] The data parsing device in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0218] This disclosure embodiment also provides a computer device having the above-mentioned Figure 6 shown data parsing device.

[0219] Please refer to Figure 7 , Figure 7 which is a schematic structural diagram of a computer device provided by an alternative embodiment of this disclosure. As shown in Figure 7As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting the components, including a high-speed interface and a low-speed interface. Each component communicates with each other using different buses and can be installed on a common motherboard or in other ways as needed. The processor can process instructions executed within the computer device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some alternative embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as an array of computer devices, a set of blade computer devices, or a multi-processor system). Figure 7 Taking one processor 10 as an example in

[0220] The processor 10 can be a central processing unit, a network processor, or a combination thereof. Among them, the processor 10 can further include a hardware chip. The above hardware chip can be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The above programmable logic device can be a complex programmable logic device, a field programmable gate array, a generic array logic, or any combination thereof.

[0221] Among them, the memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiments.

[0222] The memory 20 can include a program storage area and a data storage area. Among them, the program storage area can store an operating system and application programs required for at least one function; the data storage area can store data created according to the use of the computer device, etc. In addition, the memory 20 can include a high-speed random access memory and can also include a non-transitory memory, such as at least one disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some alternative embodiments, the memory 20 can optionally include a memory remotely set relative to the processor 10, and these remote memories can be connected to the computer device through a network. Examples of the above network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0223] The memory 20 can include a volatile memory, such as a random access memory; the memory can also include a non-volatile memory, such as a flash memory, a hard disk, or a solid-state drive; the memory 20 can also include a combination of the above types of memories.

[0224] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0225] Embodiments of the present disclosure also provide a computer-readable storage medium. The methods according to the embodiments of the present disclosure can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the methods described herein can be processed by such software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by the computer, the processor, or the hardware, the methods shown in the above embodiments are implemented.

[0226] A part of the present disclosure can be applied as a computer program product, for example, computer program instructions. When executed by a computer, through the operation of the computer, the methods and / or technical solutions according to the present disclosure can be invoked or provided. Those skilled in the art should be able to understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Herein, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible by the computer.

[0227] Although the embodiments of the present disclosure have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present disclosure, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A data analysis method, characterized in that: The method comprises: Obtain the list of building data to be parsed and the building knowledge graph; Analyzing the description information in the building data list one by one by using a preset building description model to obtain a plurality of list descriptions corresponding to the building data list; Extracting construction element information matching each of the list descriptions and element attributes corresponding to the construction element information from the building knowledge graph using the building description model; The construction element information and the element attributes are structured to obtain a structured parsing result of the building data list.

2. The method according to claim 1, characterized in that The method of using a preset building description model to parse the description information in the building data list one by one to obtain multiple list descriptions corresponding to the building data list includes: Parsing the first work data in the building data list using the building description model to generate first work description information in the building data list, wherein the first work description information includes work part description information and work object description information; Using the building description model, filtering the building data list according to a preset filtering rule to obtain second working data; parsing the third work data in the second work data using the building description model to generate second work description information; The first work description information and the second work description information are concatenated to generate the list description.

3. The method according to claim 1, characterized in that The step of extracting construction element information matching each of the list descriptions and element attributes corresponding to the construction element information from the building knowledge graph using the building description model includes: Comparing the similarity between each of the list descriptions and the building knowledge graph, and extracting the construction element information matching each of the list descriptions from the building knowledge graph; The construction element information is parsed using the building description model to obtain the target element corresponding to the list description and the element attribute corresponding to the target element.

4. The method according to claim 3, characterized in that The similarity comparison between each of the list descriptions and the building knowledge graph is performed, and the construction element information matching each of the list descriptions is extracted from the building knowledge graph, including: For any one of the list descriptions, determine the similarity between the list description and each data text in the building knowledge graph; Extracting a preset number of candidate construction text information having the highest similarity to the list description from the building knowledge graph according to the similarity, the candidate construction text information including candidate construction work information and candidate construction location information; The candidate construction text information is screened using the building description model to obtain the construction element information, which includes target construction work and target construction location.

5. The method according to claim 4, characterized in that The using the building description model to parse the construction element information to obtain the target element corresponding to the list description and the element attribute corresponding to the target element includes: Acquire multiple candidate elements corresponding to the target construction work; Using the building description model to identify the candidate elements, determining the target elements present in the inventory description; Obtain multiple candidate attribute names corresponding to the target element; The candidate attribute names are identified using the building description model to determine the element attributes present in the inventory description.

6. The method according to claim 1, characterized in that Also includes: Obtaining application feedback data for the structured parsing result and an error cause corresponding to the application feedback data; The building knowledge graph and / or the building description model is updated according to the cause of the error.

7. The method according to claim 6, characterized in that The updating of the building knowledge graph and / or the building description model according to the error cause includes: If the cause of the error is not related to the building knowledge graph, modify the prompt information of the building description model; Optimizing the building description model according to the modified prompt information; and / or, If the cause of the error is related to the building knowledge graph, updating the node elements of the building knowledge graph; The architectural knowledge graph is optimized according to the modified node elements.

8. The method according to claim 1, characterized in that Obtaining the building knowledge graph includes: Obtain construction project data; Parsing the construction engineering data to determine the first construction elements at different levels; Based on the hierarchical position of the first building element, determining a second building element and a third building element attached to the first building element, wherein the second building element has no attribute information; Determining attribute information of the third building element based on the hierarchical position of the third building element; The knowledge graph is generated by performing knowledge fusion based on the first building element, the second building element, the third building element and the attribute information.

9. A data analysis device, characterized in that: The device comprises: The first acquisition module is used to obtain the list of building data to be parsed and the building knowledge graph; A parsing module, used to parse the description information in the building data list one by one by using a preset building description model to obtain a plurality of list descriptions corresponding to the building data list; An extraction module, used to extract construction element information matching each of the list descriptions and element attributes corresponding to the construction element information from the building knowledge graph using the building description model; The obtaining module is used to perform structured processing on the construction element information and the element attributes to obtain the structured parsing result of the building data list.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the data parsing method according to any one of claims 1 to 8 by executing the computer instructions.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the data analysis method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the data analysis method according to any one of claims 1 to 8.

Images