Building internet data storage, construction strategy adjustment and building task retrieval method

Abstract

The invention relates to the technical field of building industry internet data management, in particular to a building internet data storage, construction strategy adjustment and building task retrieval method. The invention provides a building internet data storage method. Building internet data is stored in a database in a knowledge graph structure; tense labels are embedded through attributes, and cross-domain semantic association is established through a knowledge graph technology, so that ordered storage of building internet data is facilitated, and subsequent data management, retrieval and calling are facilitated. According to the method, data are stored in a complex graph structure, and efficient management, intelligent service and dynamic evolution of large-scale engineering element data are supported.

Description

Technical Field

[0001] The present invention relates to the technical field of building industrial Internet data governance, and in particular to a method for storing building Internet data, adjusting construction strategies, and retrieving building tasks. Background Art

[0002] With the deepening of the digital transformation of the building industry, data has become the core element driving the whole life cycle management and decision-making of projects. However, traditional data governance technologies are difficult to match the complex scenario requirements of the industry, revealing the following key bottlenecks:

[0003] First, at the level of information closed-loop feedback, the building industry involves multiple links such as design, construction, and operation and maintenance. The data sources are heterogeneous (such as BIM models, Internet of Things sensors, business system logs, etc.), and their dynamic relevance and feedback mechanism are lacking, resulting in obstacles to the collaborative service of the entire domain and the optimization of the value chain, and prominent information island problems.

[0004] Second, in the face of a large amount of engineering data (such as high-precision point clouds, spatio-temporal trajectories, component attribute topological relationships, etc.), traditional relational databases are limited by rigid architectures and computing paradigms, and are difficult to efficiently support complex semantic queries (such as spatio-temporal topological reasoning, multi-modal data fusion analysis) and real-time decision-making requirements, restricting the in-depth mining of data value.

[0005] In addition, the building scenario has high dynamics, with frequent demand changes and complex data temporal evolution. The existing data space lacks the ability of adaptive adjustment and cannot realize the intelligent iteration of service logic and knowledge model based on uncertain conditions (such as the propagation of design changes, resource scheduling disturbances), resulting in insufficient system robustness and flexibility.

[0006] The above problems severely limit the effectiveness of intelligent construction and digital delivery, and there is an urgent need to achieve breakthroughs through new data governance architectures and technical paths. Summary of the Invention

[0007] In order to overcome the defect of difficult intelligent management of building industry data in the above-mentioned prior art, the present invention proposes a method for storing building Internet data, which stores data in a complex graph structure to support the efficient management, intelligent services, and dynamic evolution of large-scale engineering element data.

[0008] A method for storing building Internet data proposed by the present invention stores building Internet data in a knowledge graph structure in a database; the knowledge graph is represented as: entity - relationship - attribute; where the entity includes sensors, building components, and equipment; the relationships include: installation location, structural association, and construction involvement; the attributes include sensor detection values, installation time, entity type, construction progress, construction start time, and end time.

[0009] Preferably, in the knowledge graph, the entities further include construction personnel and / or logical entities. The logical entities include BIM (Building Information Modeling), schedule plan, progress report, quality inspection report, and environmental inspection area. The relationships corresponding to the logical entities include one or more of structural association, data flow, and temporal dependence.

[0010] A construction strategy adjustment method proposed by the present invention first trains a construction strategy adjustment model on a construction data set. The samples of the construction data set are the knowledge graphs of the entities involved in the construction tasks, and the labels are the adjustment strategies including the construction speed index, construction start time, and end time of each entity. The progress plan is adjusted by combining the construction adjustment strategy with the existing construction data.

[0011] Preferably, the construction strategy adjustment model is linked with a database that stores building Internet data represented in the form of a knowledge graph structure. The construction strategy adjustment model performs strategy adjustment based on the database data, and the database updates the attribute information based on the adjustment strategy.

[0012] A building task retrieval method proposed by the present invention first constructs a database and a knowledge base. The database stores building Internet data represented in the form of a knowledge graph structure, and the knowledge base stores industry specifications. The database and the knowledge base are respectively retrieved based on the input question to obtain the database query result and the knowledge base query result. Data enhancement is performed by combining the input question, the database query result, and the knowledge base query to obtain the final answer.

[0013] Preferably, a retrieval model is first trained on a question-answering data set {input question, answer}. The retrieval model includes three parts. The first part retrieves the database based on the input question to obtain the database query result. The second part retrieves the knowledge base based on the input question to obtain the knowledge base query result. The third part generates the final answer based on the input question, the database query result, and the knowledge base query result. When querying, the input question is substituted into the retrieval model, and the retrieval model generates an answer based on the database and the knowledge base.

[0014] Preferably, the method for constructing the database is as follows: first, a knowledge base is constructed, and then the obtained building Internet data is cleaned based on the knowledge base. The cleaned data is stored in the database in the form of a knowledge graph structure.

[0015] Preferably, the building Internet data cleaning method is as follows: calculate the similarity between each piece of building Internet data and each knowledge text in the knowledge base, and use the maximum similarity as the evaluation value of the building Internet data. Delete the building Internet data whose evaluation value is lower than the set similarity threshold.

[0016] Alternatively, a clustering algorithm is used to cluster the building Internet data and the knowledge text, and the building Internet data that cannot be classified into the class where the knowledge text is located is deleted.

[0017] A construction data management system proposed by the present invention includes: a data acquisition module, a data storage and indexing module, an intelligent query and analysis module, an adaptive evolution engine module, and an evaluation module;

[0018] The data acquisition module is used to collect building Internet data and organize it into a knowledge graph;

[0019] The data storage and indexing module is connected to the data acquisition module, and the data storage and indexing module is used to store the building Internet data in the form of a knowledge graph structure into a database;

[0020] The intelligent query and analysis module is used to query the database according to the input question;

[0021] The adaptive evolution engine module is used to adjust the construction plan in the database by combining a construction strategy adjustment model.

[0022] Preferably, it further includes an evaluation module. The evaluation module is connected to the adaptive evolution engine module. The evaluation module is used to evaluate the gap between the current state and the goal of the construction; the adaptive evolution engine module makes construction strategy adjustments according to the feedback information of the evaluation module.

[0023] The advantages of the present invention are as follows:

[0024] (1) A method for storing building Internet data proposed by the present invention stores the building Internet data in a database in the form of a knowledge graph structure; by embedding temporal tags in attributes and establishing cross-domain semantic associations through knowledge graph technology, it facilitates the orderly storage of building Internet data and subsequent data management and retrieval calls.

[0025] (2) A method for adjusting construction strategies proposed by the present invention generates adjustment strategies according to changes in sensor monitoring data through a construction strategy adjustment model, thereby ensuring real-time adjustment of the entire construction plan. Through the linkage between the construction strategy adjustment model and the database, feedback control at the construction site and adaptive adjustment of the construction plan are realized.

[0026] (3) The present invention combines a complex graph structure with an adaptive evolution mechanism to solve the problem of dynamic governance of building data, realizes full life cycle traceability of data through temporal logic and snapshot technology, and integrates mobile agents and deep learning to improve the real-time performance and robustness of intelligent services.

[0027] (4) A method for retrieving building tasks proposed by the present invention combines a knowledge graph and a reinforcement learning model to realize a complex graph incremental evolution model and a multi-network graph data cross-correlation model, and realizes dynamic mapping and historical traceability of engineering elements through entity association and temporal logic technology.

[0028] (5) Through the semantic intelligent query technology, the present invention can achieve multi-dimensional fuzzy query and precise matching based on the fuzzy theory and semantic retrieval algorithm; and through the graph structure optimization and distributed computing, the query efficiency of large-scale data is improved.

[0029] (6) The construction data management system proposed by the present invention adopts a demand-driven adaptive evolution framework, combines machine learning and deep learning, constructs an intelligent service dynamic optimization model, and responds to the changes in the data space in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a diagram showing the construction speed;

[0031] Figure 2 It is a flowchart of a building task retrieval method;

[0032] Figure 3 It is an entity display, where the red ones are sensors, the blue ones are building components, and the green ones are construction tasks;

[0033] Figure 4 It is a module connection diagram of the construction data management system. DETAILED DESCRIPTION OF THE INVENTION

[0034] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0035] A method for storing building Internet data proposed by the present invention stores the building Internet data in a database in the form of a knowledge graph; the building Internet data consists of multi-source heterogeneous standardized data, and the data sources include BIM, GIS, IoT, etc.

[0036] Table 1: Partial building Internet data

[0037]

[0038] Among them, Si represents the i-th sensor, Bi represents the i-th beam, Bi-j represents the j-th column of the i-th type, and Ti represents the i-th construction task; i and j are numerical serial numbers.

[0039] Referring to Table 2, Figure 3 , the knowledge graph is expressed as: entity - relationship - attribute; among them, the entities include sensors, building components and equipment; the relationships include: installation location, structural association and construction involvement; the attributes include sensor detection values, installation time, entity type, construction progress, construction start time and end time, etc.

[0040] The sensors include stress sensors, temperature and humidity sensors, etc. The building components include beams, columns, slabs, etc. The equipment includes tower cranes, concrete pumps, etc. In a specific embodiment, part of the knowledge graph structure is shown in Table 2 below. In the relationships, "installation location" can refer to "sensor - installation location", "structural association" can refer to "component - constituent material", and "construction involved" can refer to "construction task - construction object".

[0041] Table 2 Partial Knowledge Graph Structure

[0042]

[0043]

[0044] Among them, Si represents the i-th sensor, Bi represents the i-th beam, Bi-j represents the j-th column of the i-th type, and Ti represents the i-th construction task; i and j are numerical serial numbers.

[0045] In Table 2, the timestamp uses the construction start time, and "installed in" and "timestamp" can be placed in the "attributes" of the knowledge graph.

[0046] In the knowledge graph, the entities can further include construction personnel. Let sensors, building components, and construction personnel be collectively referred to as physical entities. Specifically in implementation, BIM (Building Information Model), schedule plan, progress report, quality inspection report, and environmental detection area, etc. can also be used as logical entities for knowledge graph structure storage. The relationships corresponding to the logical entities are all structural associations. Among them, BIM is associated with physical entities (i.e., sensors and building components), the quality inspection report is associated with the detection object, and the environmental detection area is associated with the construction location.

[0047] In further implementation, the relationships corresponding to the logical entities can also be defined to include: data flow and temporal dependency, etc. The "data flow" relationship can be manifested as "sensor - data storage" and "equipment - control instruction", etc. The "temporal dependency" relationship can be manifested as "construction task - prerequisite task" and "schedule plan - actual completion", etc.; the actual completion can be specifically seen in the progress report.

[0048] In this way, in this embodiment, through data modeling and association, the building Internet data (design, construction, operation and maintenance) is abstracted into an "entity - relationship - attribute" graph structure. Temporal tags are embedded through attributes, and cross - domain semantic associations are established through knowledge graph technology, facilitating the orderly storage of building Internet data and facilitating subsequent data management and retrieval calls.

[0049] Specifically, in this embodiment, the distributed storage technology is adopted in the database. Combining the range sharding and hash sharding strategies, load balancing is achieved through consistent hashing to ensure efficient data storage and retrieval. In this embodiment, a graph-structured database is established through the knowledge graph to implement a complex graph incremental evolution model and a multi-network graph data cross-correlation model. Through entity association and temporal logic technology, the dynamic mapping and historical traceability of engineering elements are realized.

[0050] The present invention further proposes a construction strategy adjustment method, including the following steps:

[0051] S1. Construct a construction data set, where the sample is the knowledge graph of the entities involved in the construction task, and the label is the adjustment strategy including the construction speed index, construction start time, and end time of each entity; construct a neural network model as the basic model;

[0052] S2. Use the method of reinforcement learning to train the basic model on the construction data set to obtain the converged basic model as the construction strategy adjustment model.

[0053] During the training process of the construction strategy adjustment model, the policy gradient method can be used to construct the basic model so that it can automatically find the optimal strategy when facing a changing environment. In this embodiment, the online learning technology is used to enable the model to continuously adjust parameters according to the latest data, adapt to environmental changes, and achieve the ability of "learning while running"; use the Q-learning (reinforcement learning) algorithm. At each moment, according to the environmental state, select an action, observe the feedback reward, and then update the Q value using the following update formula:

[0054] Q(s t ,a t )←Q(s t ,a t )+α[r t +γQ(s t+1 ,a t+1 )-Q(s t ,a t )]

[0055] where α is the learning rate and γ is the discount factor; Q(s t ,a t ) is the Q value at time slot t, r t is the reward at time slot t, and Q(s t+1 ,a t+1 ) is the Q value at time slot t + 1.

[0056] Specifically, during implementation, the construction speed in the adjustment strategy can be expressed as Figure 1As shown; where equal to 1 indicates normal construction, a positive number less than 1 indicates slow construction progress; a value less than 0 indicates suspension of construction, and the absolute value of the number indicates the relevant number of expected shutdown time.

[0057] In this embodiment, the building Internet data is stored in the database in the form of a knowledge graph structure. The detection data in the sensor association attributes is real-time data. The construction strategy adjustment model generates adjustment strategies according to the changes in the sensor detection data, so as to ensure the real-time adjustment of the entire construction plan.

[0058] In this way, by combining sensors, data input, and real-time monitoring, external environmental changes (such as temperature fluctuations, user behavior patterns, network attacks, etc.) and internal state anomalies can be perceived, and the construction strategy can be adjusted in real time through the construction strategy adjustment model, and the database can be updated. In this way, based on the database to calculate evaluation indicators (such as efficiency, energy consumption, safety), the real-time monitoring of the gap between the current state and the target state can be realized.

[0059] Through the linkage between the construction strategy adjustment model and the database, the feedback control of the construction site and the adaptive adjustment of the construction plan are realized. After each decision is executed, feedback data is collected through the monitoring system of the construction site to evaluate the quality of the decision according to the actual operation effect; historical snapshots are saved during the evolution process to ensure that in case of wrong decisions or unforeseen situations, it can quickly roll back to a stable state, thus ensuring the continuity and security of the system.

[0060] In this embodiment, the construction strategy adjustment model is regularly updated based on the latest data to ensure that the model can timely reflect the changes in the external environment and internal state; in a new scenario, through transfer learning technology, the existing model experience can be quickly transferred to a new task, reducing the time for model retraining. Fault tolerance and redundancy design are also adopted in this embodiment. A fault tolerance mechanism is designed in the model architecture to ensure that the model can still operate normally and gradually recover in case of partial node failures or data anomalies;

[0061] Refer to Figure 2 , this embodiment also proposes a building task retrieval method, including the following steps:

[0062] St1. Collect the industry specifications related to the building task to construct a knowledge base.

[0063] Specifically, the knowledge base is generated based on industry specifications using RAG technology, so as to realize fuzzy query and precise matching in multiple dimensions (time, space, semantics) based on fuzzy theory and semantic retrieval algorithms when called later.

[0064] St2. Obtain and clean the building Internet data, and store the cleaned data in the database in the form of a knowledge graph structure;

[0065] The method for cleaning construction Internet data is as follows: calculate the similarity between each piece of construction Internet data and each knowledge text in the knowledge base, and use the maximum similarity as the evaluation value of the construction Internet data; delete the construction Internet data whose evaluation value is lower than the set similarity threshold.

[0066] The similarity between construction Internet data and knowledge text can be calculated based on vector distance, and vector similarity can be specifically adopted.

[0067] In specific implementation, an embedding model can be used to parse the standardized construction Internet data file to obtain the embedding vector of the construction Internet data. The embedding model converts construction Internet data and industry knowledge text from natural language into high-dimensional vectors, then captures the semantic information behind the data through vectors, and is mapped to a farther vector space through deep learning, and the similarity relationship between different texts is calculated accordingly.

[0068] The method for cleaning construction Internet data can also adopt a clustering algorithm to cluster construction Internet data and knowledge text, so as to delete the construction Internet data that cannot be classified into the class where the knowledge text is located, and store the remaining construction Internet data in the database in the form of a knowledge graph structure.

[0069] St3. Construct a retrieval model and a question-answering data set {input question, answer}. The retrieval model consists of three parts. The first part retrieves the database based on the input question to obtain the database query result; the second part retrieves the knowledge base based on the input question to obtain the knowledge base query result; the third part generates the final answer based on the input question, the database query result, and the knowledge base query result.

[0070] The retrieval model can specifically adopt LLM models such as GPT and Llama to generate the final answer based on the enhanced context.

[0071] St4. Use the reinforcement learning method to train the retrieval model on the question-answering data set until the model converges.

[0072] In specific implementation, the first part and the second part of the retrieval model can adopt existing query means, such as vector comparison, semantic retrieval, etc.; in this way, only the third part of the retrieval model needs to be trained and updated during the training process. For the third part, a natural language model, a text enhancement model, etc. can be specifically adopted.

[0073] St5. When a query task is required, input the question into the retrieval model, and the retrieval model generates an answer by combining the question, the database, and the knowledge base.

[0074] In specific implementation, both the construction data set and the Q&A data set are from construction data. Distributed sensors, logging systems, and API interfaces can be deployed at the construction site and monitoring system to collect key data at various stages (design, construction, operation and maintenance) of the building project in real time; the monitoring system uses a big data platform to track indicators such as system operation status, resource load, and service response in real time, providing data support for subsequent construction decisions.

[0075] Specifically, the collected construction data can first go through anomaly detection. Through statistical analysis and anomaly detection algorithms, data fluctuations and anomalies can be discovered in a timely manner, so as to screen normal and safe construction data to build a data set, providing a feedback basis for the evolution engine.

[0076] Refer to Figure 4 , this embodiment also proposes a construction data management system, including: a data collection module, a data storage and indexing module, an intelligent query and analysis module, an adaptive evolution engine module, and an evaluation module.

[0077] The data collection module is used to collect building Internet data and organize it into a knowledge graph, thereby abstracting building industry information into a knowledge graph structure of "entity-relationship-attribute" and embedding temporal tags. Building industry information includes design information, construction information, operation and maintenance information, etc. The data collection module can combine data modeling for the transformation of the knowledge graph. In the knowledge graph, entities are spatially associated, as Figure 3 shown, where the red ones are sensors, the blue ones are building components, and the green ones are construction tasks.

[0078] The data collection module can directly collect standardized data of multi-source heterogeneous data; among them, the sources of multi-source data include BIM, GIS, IoT, etc. The data collection module cleans the collected data and then transforms it into a knowledge graph structure.

[0079] The data storage and indexing module is used to store the building Internet data in the knowledge graph structure into the database. The data storage and indexing module adopts distributed storage technology and combines range sharding and hash sharding strategies for the storage of knowledge graph data, so as to achieve load balancing through consistent hashing and ensure efficient data storage and retrieval.

[0080] The intelligent query and analysis module is used to query the database according to the input question and obtain the query result. The intelligent query and analysis module can specifically integrate tools such as natural language processing, graph computing, and semantic retrieval, and use inverted index, time series, and spatial index technologies to optimize the query efficiency of large-scale data.

[0081] Specifically, the intelligent query and analysis module can query specific types of construction data from the database according to the query type, such as sensors, building components, and construction tasks. A retrieval model is stored in the intelligent query and analysis module, and the final answer to the input question is generated by combining the database and the knowledge base through the retrieval model. The knowledge base can specifically use the FAISS vector database to store text knowledge, and use database retrieval instructions for similarity search to obtain the knowledge base query results.

[0082] The adaptive evolution engine module is used to obtain abnormal situations and adjust the construction strategy in combination with the construction strategy adjustment model to overcome abnormal situations. Abnormal situations include external environment changes and internal state abnormalities. External environment changes include temperature fluctuations, user behavior patterns, and cyberattacks, etc., and internal state abnormalities include conflicts in construction plan times and disordered construction relationships, etc. External environment changes can be specifically perceived through sensors, data input, or real-time monitoring.

[0083] The adaptive evolution engine module can introduce an AI training unit to dynamically monitor data changes, optimize the data structure and service process in real time based on reinforcement learning and mobile agent mechanisms, and ensure that the system has a high degree of adaptability.

[0084] The evaluation module is used to calculate evaluation indicators and analyze the gap between the current state and the target. The evaluation indicators can specifically select efficiency, energy consumption, and safety, etc.

[0085] Specifically in implementation, it can be set that the adaptive evolution engine module adjusts the construction strategy according to the feedback information of the evaluation module.

[0086] This system combines a knowledge graph and a reinforcement learning model to implement a complex graph incremental evolution model and a multi-network graph data cross-correlation model, and realizes the dynamic mapping and historical traceability of engineering elements through entity association and temporal logic technology.

[0087] The construction strategy decision model adopted by this system supports snapshot storage, rollback operations, and temporal data space management to ensure the controllability of the data evolution process. This system adopts a demand-driven adaptive evolution framework, combines machine learning and deep learning, constructs an intelligent service dynamic optimization model, and responds to data space changes in real time. This system also adopts mobile agent technology to realize the autonomous migration of service nodes and resource collaboration, and complete closed-loop feedback.

[0088] The database adopted by this system supports high-order semantic intelligent query technology, and can realize fuzzy query and precise matching in multiple dimensions (time, space, semantics) based on fuzzy theory and semantic retrieval algorithms; and through graph structure optimization and distributed computing, the query efficiency of large-scale data is improved.

[0089] Of course, for those skilled in the art, the present invention is not limited to the details of the above-described exemplary embodiments, but also includes the same or similar structures that can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference numerals in the claims should not be construed as limiting the claims involved.

[0090] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment contains only one independent technical solution. This narrative style of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

[0091] The technologies, shapes, and structures not detailed in the present invention are all well-known technologies.

Claims

1. A method for storing building Internet data, characterized in that, Store the building Internet data in a database in the form of a knowledge graph; the knowledge graph is represented as: entity - relationship - attribute; among them, the entities include sensors, building components, and equipment; the relationships include: installation location, structural association, and construction involvement; the attributes include sensor detection values, installation time, entity type, construction progress, construction start time, and end time.

2. The building Internet data storage method according to claim 1, wherein, In the knowledge graph, the entities also include construction workers and / or logical entities, and the logical entities include BIM (Building Information Modeling), schedule plan, progress report, quality inspection report, and environmental inspection area; the relationships corresponding to the logical entities include one or more of: structural association, data flow, and temporal dependence.

3. A construction strategy adjustment method, characterized in that, First, train a construction strategy adjustment model on the construction data set. The samples of the construction data set are the knowledge graphs of the entities involved in the construction tasks, and the labels are the adjustment strategies including the construction speed indicators, construction start time, and end time of each entity; adjust the schedule plan by combining the construction adjustment strategy with the existing construction data.

4. The construction strategy adjustment method according to claim 3, wherein, The construction strategy adjustment model is linked to the database storing the building Internet data represented in the form of a knowledge graph. The construction strategy adjustment model makes strategy adjustments based on the database data, and the database updates the attribute information based on the adjustment strategy.

5. A building task retrieval method, characterized in that, First, construct a database and a knowledge base. The database stores the building Internet data represented in the form of a knowledge graph, and the knowledge base stores industry specifications; respectively retrieve the database and the knowledge base based on the input question to obtain the database query result and the knowledge base query result; Combine the input question, the database query result, and the knowledge base query to perform data enhancement to obtain the final answer.

6. The building task retrieval method according to claim 5, wherein, First, train a retrieval model on the question-answering data set {input question, answer}. The retrieval model includes three parts. The first part retrieves the database based on the input question to obtain the database query result; the second part retrieves the knowledge base based on the input question to obtain the knowledge base query result; The third part generates the final answer based on the input question, the database query result, and the knowledge base query result; during query, substitute the input question into the retrieval model, and the retrieval model generates an answer based on the database and the knowledge base.

7. The building task retrieval method according to claim 5, wherein The construction method of the database is: first construct a knowledge base, and then clean the obtained building Internet data based on the knowledge base, and store the cleaned data in the database in the form of a knowledge graph.

8. The building task retrieval method according to claim 7, characterized in that, The building Internet data cleaning method is: calculate the similarity between each piece of building Internet data and each knowledge text in the knowledge base, and use the maximum similarity as the evaluation value of the building Internet data; delete the building Internet data whose evaluation value is lower than the set similarity threshold; Or, use a clustering algorithm to cluster the building Internet data and the knowledge text, and delete the building Internet data that cannot be classified into the class where the knowledge text is located.

9. A construction data management system, characterized in that, Including: Data acquisition module, data storage and indexing module, intelligent query and analysis module, adaptive evolution engine module, and evaluation module; The data acquisition module is used to collect building Internet data and organize it into a knowledge graph; The data storage and indexing module is connected to the data acquisition module. The data storage and indexing module is used to store the building Internet data in the form of a knowledge graph into the database; The intelligent query and analysis module is used to query the database according to the input question; The adaptive evolution engine module is used to adjust the construction plan in the database by combining the construction strategy adjustment model.

10. The construction data management system according to claim 9, characterized in that, It also includes an evaluation module. The evaluation module is connected to the adaptive evolution engine module. The evaluation module is used to evaluate the gap between the current state of the construction and the goal; the adaptive evolution engine module makes construction strategy adjustments according to the feedback information of the evaluation module.

Images