Building barrier breaking knowledge graph generation method, question and answer method and system
By combining text vector models and architectural knowledge models, an architectural obstacle avoidance knowledge graph is generated, which solves the problem of low accuracy in existing technologies and achieves higher quality knowledge graph construction.
Patent Information
- Application Number
- CN202510917930.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-03
- Publication Date
- 2025-11-21
AI Technical Summary
Existing technologies have low accuracy in constructing knowledge graphs of building obstacle breaches, and cannot effectively extract entities and relationships in the field of building obstacle breaches, resulting in information loss.
By acquiring architectural knowledge models and obstacle avoidance data, and using text vector models and architectural knowledge models to generate vector representations, entity intent recognition is performed to generate an architectural obstacle avoidance knowledge graph.
It improves the accuracy of knowledge graphs, ensures the integrity and consistency of entity intents, and enhances the construction quality of knowledge graphs.
Smart Images
Figure CN120994804A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of natural language processing technology, and in particular to a method, question-answering method and system for generating a knowledge graph of building obstacle avoidance. Background Technology
[0002] As society's demands for information gradually increase, in the field of building obstacle avoidance, constructing clear and effective knowledge graphs based on existing building obstacle avoidance information has gradually become one of the key focuses for relevant personnel.
[0003] Currently, the relevant technologies typically extract entities (such as materials, components, engineering terms, etc.) and relationships in the field of building obstacle destruction directly from the original obstacle destruction building dataset through reasoning models, and then construct a knowledge graph based on triples. However, the accuracy of the knowledge graphs generated in this way is relatively low.
[0004] Therefore, the problems with the relevant technologies still need to be solved and optimized. Summary of the Invention
[0005] The purpose of this invention is to at least partially solve one of the technical problems existing in the related art.
[0006] Therefore, one objective of this invention is to provide a method, question-and-answer method, and system for generating a knowledge graph of building obstacle avoidance, wherein the generation method can effectively improve the accuracy of the generated knowledge graph.
[0007] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of this application include:
[0008] In a first aspect, embodiments of this application provide a method for generating a knowledge graph of building obstacle avoidance, including:
[0009] Acquire architectural knowledge models and architectural obstacle breach data;
[0010] The building obstacle breach data is input into a text vector model to obtain a first vector representation output by the text vector model. The first vector representation is a word vector corresponding to the text semantic features of the building obstacle breach data.
[0011] The building obstacle breach data is input into the building knowledge model to obtain a second vector representation, which is used to represent the vector representation of the geometric attribute data corresponding to the building obstacle breach data.
[0012] Based on the first vector representation and the second vector representation, entity intent recognition processing is performed to obtain the entity intent of building obstacle breaching;
[0013] Based on the intent of the building obstacle breaching entity, a building obstacle breaching knowledge graph is generated.
[0014] In addition, the method according to the above embodiments of this application may also have the following additional technical features:
[0015] Furthermore, in one embodiment of this application, the step of inputting the building obstacle breach data into a text vector model to obtain a first vector representation output by the text vector model includes:
[0016] Text character extraction is performed on the building obstacle breach data to obtain several building obstacle breach characters;
[0017] All the aforementioned building obstacle breach characters are bidirectionally vector-encoded to obtain the first vector representation.
[0018] Furthermore, in one embodiment of this application, the step of inputting the building obstacle breach data into the building knowledge model to obtain a second vector representation includes:
[0019] Obtain the geometric attribute data corresponding to the building obstacle breach data;
[0020] The geometric attribute data is structured and encoded to obtain several intermediate vector representations;
[0021] The second vector representation is obtained by concatenating all the intermediate vector representations.
[0022] Furthermore, in one embodiment of this application, the step of performing entity intent recognition processing based on the first vector representation and the second vector representation to obtain the building obstacle breach entity intent includes:
[0023] The first vector representation and the second vector representation are concatenated to obtain the third vector representation;
[0024] Feature extraction is performed on the third vector representation to obtain building obstacle breach features;
[0025] The building obstacle breach features are labeled and identified to obtain the building obstacle breach entity intent.
[0026] Furthermore, in one embodiment of this application, the step of extracting features from the third vector representation to obtain building obstacle breach features includes:
[0027] Forward feature extraction is performed on the third vector representation to obtain the forward hidden layer vector representation;
[0028] Backward feature extraction is performed on the third vector representation to obtain the backward hidden layer vector representation;
[0029] Based on the forward hidden layer vector representation, the corresponding backward hidden layer vector representation is concatenated to obtain the building obstacle breach feature.
[0030] Furthermore, in one embodiment of this application, the annotation and recognition processing of the building obstacle breach features is performed to obtain the building obstacle breach entity intent;
[0031] The building obstacle breach features are subjected to conditional discrimination processing to obtain a global label sequence corresponding to the building obstacle breach features;
[0032] The global label sequence is merged to obtain the building obstacle breaching entity intent.
[0033] Secondly, embodiments of this application provide a question-and-answer method for a knowledge graph of architectural obstacle avoidance, including:
[0034] Retrieve unanswered questions about building obstacle avoidance;
[0035] The question about building obstacle breaches is processed by entity and intent recognition to obtain entity relationship information;
[0036] Based on the entity relationship information, a knowledge query is performed on the aforementioned building obstacle breach knowledge graph to obtain the building obstacle breach response result corresponding to the building obstacle breach question.
[0037] Thirdly, embodiments of this application provide a system for generating a knowledge graph of building obstacle avoidance, comprising:
[0038] The first processing unit is used to acquire architectural knowledge models and architectural obstacle data;
[0039] The second processing unit is used to input the building obstacle breach data into a text vector model to obtain a first vector representation output by the text vector model. The first vector representation is a word vector corresponding to the text semantic features of the building obstacle breach data.
[0040] The third processing unit is used to input the building obstacle breach data into the building knowledge model to obtain a second vector representation, which is used to represent the vector representation of the geometric attribute data corresponding to the building obstacle breach data.
[0041] The fourth processing unit is used to perform entity intent recognition processing based on the first vector representation and the second vector representation to obtain the building obstacle-breaking entity intent;
[0042] The fifth processing unit is used to generate a building obstacle breach knowledge graph based on the building obstacle breach entity's intent.
[0043] Fourthly, embodiments of this application also provide an electronic device, including:
[0044] At least one processor;
[0045] At least one memory for storing at least one program;
[0046] When the at least one program is executed by the at least one processor, the at least one processor performs the method described above.
[0047] Fifthly, embodiments of this application also provide a computer-readable storage medium storing a processor-executable program, which, when executed by the processor, is used to implement the above-described method.
[0048] The advantages and beneficial effects of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application:
[0049] This application discloses a method, question-answering method, and system for generating a knowledge graph of architectural obstacle avoidance. The method involves acquiring an architectural knowledge model and architectural obstacle avoidance data; inputting the architectural obstacle avoidance data into a text vector model to obtain a first vector representation output by the text vector model, where the first vector representation is a character vector corresponding to the textual semantic features of the architectural obstacle avoidance data; inputting the architectural obstacle avoidance data into the architectural knowledge model to obtain a second vector representation, where the second vector representation is used to characterize the geometric attribute data corresponding to the architectural obstacle avoidance data; performing entity intent recognition processing based on the first vector representation and the second vector representation to obtain architectural obstacle avoidance entity intent; and generating an architectural obstacle avoidance knowledge graph based on the architectural obstacle avoidance entity intent. This method, based on the first vector representation and the second vector representation output by the architectural knowledge model, determines the corresponding architectural obstacle avoidance entity intent and generates an architectural obstacle avoidance knowledge graph based on this intent, which can effectively improve the accuracy of the generated knowledge graph. Attached Figure Description
[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following description is provided with accompanying drawings of the relevant technical solutions in the embodiments of this application or the prior art. It should be understood that the accompanying drawings described below are only for the purpose of clearly illustrating some embodiments of the technical solutions in this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.
[0051] Figure 1 A flowchart illustrating a method for generating a knowledge graph of building obstacle avoidance provided in an embodiment of this application;
[0052] Figure 2 A schematic diagram of the framework of a knowledge graph generation system for building obstacle avoidance provided in an embodiment of this application;
[0053] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0054] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. The step numbers in the following embodiments are set only for ease of explanation, and there is no limitation on the order between the steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.
[0055] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.
[0056] The technical terms used in the embodiments of this application are explained below:
[0057] Building breaching refers to the rapid and effective destruction of key buildings, either entirely or partially, through blasting, mechanical demolition, or other means during engineering operations in order to open up passageways or achieve specific objectives.
[0058] Currently, related technologies typically extract entities (such as materials, components, and engineering terms) and relationships in the building obstacle domain directly from the original obstacle-prone building dataset using inference models, and then construct a knowledge graph based on triples. However, due to the complexity and variability of data in the building obstacle domain, and the need for inference models to have a certain generalization ability, the entities and relationships extracted by the inference models may not be accurate, and information loss is likely to occur. For example, if a text statement in the original obstacle-prone building dataset states "The structural material of reinforced concrete buildings is C30 concrete," the inference model (such as the BERT model) may easily treat "C30" as ordinary data information during inference, failing to connect it with concrete and losing the meaning of representing the type of concrete, resulting in low accuracy of the subsequently generated knowledge graph.
[0059] It should be noted that the aforementioned related technologies are only used to assist in understanding the technical solutions of this application and do not mean that they belong to the publicly disclosed prior art.
[0060] In view of this, embodiments of the present invention provide a method, question-answering method, and system for generating a knowledge graph of building obstacle breaches. The method determines the intent of the corresponding building obstacle breach entity based on a first vector representation and a second vector representation output by a building knowledge model. Specifically, it utilizes the geometric attribute knowledge in the building knowledge model to supplement the first vector representation output by the text vector model with knowledge in the field of building obstacle breaches, making the knowledge information contained in the vector representation involved in the recognition process more complete, thereby effectively improving the accuracy of the subsequently generated knowledge graph.
[0061] Reference Figure 1 In this embodiment of the application, a method for generating a knowledge graph of building obstacle avoidance includes:
[0062] Step 110: Obtain the building knowledge model and building obstacle breach data;
[0063] In this embodiment, the building knowledge model can be a digital building model created using the BIM tool Revit. This building knowledge model contains geometric attribute information of the building, specifically including geometric information, construction attributes, engineering data, etc. The building obstacle data can be text data related to building obstacle, which can be publicly available data on the Internet. For example, the original building obstacle data can be retrieved from the Internet based on at least one of the following keywords: "steel structure," "reinforced concrete structure," "brick-concrete structure," "brick-wood structure," "wood structure," "building structure," "building characteristics," "building demolition," etc. Then, the Easydl platform is used to annotate the original obstacle data with entities and relationships to obtain the final building obstacle data.
[0064] Step 120: Input the building obstacle breach data into the text vector model to obtain the first vector representation output by the text vector model. The first vector representation is a word vector corresponding to the text semantic features of the building obstacle breach data.
[0065] In this embodiment of the application, building obstacle breach data can be input into a text vector model, which can be a BERT model. Based on the context information in the building obstacle breach data, the text vector model vectorizes each character in the building obstacle breach data to obtain a first vector representation.
[0066] In some embodiments, inputting the building obstacle breach data into a text vector model to obtain a first vector representation output by the text vector model includes:
[0067] Text character extraction is performed on the building obstacle breach data to obtain several building obstacle breach characters;
[0068] All the aforementioned building obstacle breach characters are bidirectionally vector-encoded to obtain the first vector representation.
[0069] In this embodiment, text character extraction can be performed by a text vector model to process building obstacle data to form several building obstacle characters; then, bidirectional vector encoding can be performed by inputting each building obstacle character into a bidirectional Transformer encoder, and superimposing three layers of vectors—Token embeddings, Segment embeddings, and Positionembeddings—to obtain a word vector that integrates the context of the building obstacle data and has the semantic features of the building obstacle text, denoted as the first vector representation.
[0070] Specifically, if one of the text statements in the building obstacle breach data is "the structural material of the steel-concrete structure building is C30 concrete", then the corresponding first vector representation can be (X steel, X concrete, ..., X C X3, X0, X-mixed, X-coated, X-soil).
[0071] Step 130: Input the building obstacle breach data into the building knowledge model to obtain a second vector representation, which is used to represent the vector representation of the geometric attribute data corresponding to the building obstacle breach data;
[0072] In this embodiment of the application, building obstacle data can be input into a building knowledge model for retrieval, thereby obtaining a vector representation of geometric attribute data that matches all or part of the characters of the building obstacle data.
[0073] In some embodiments, inputting the building obstacle breach data into the building knowledge model to obtain a second vector representation includes:
[0074] Obtain the geometric attribute data corresponding to the building obstacle breach data;
[0075] The geometric attribute data is structured and encoded to obtain several intermediate vector representations;
[0076] The second vector representation is obtained by concatenating all the intermediate vector representations.
[0077] In this embodiment of the application, geometric attribute data corresponding to building obstacle data can be obtained through a building knowledge model. Specifically, for the character combination "C30 concrete" in the building obstacle data, the geometric attribute data of the character combination can be obtained through a building knowledge model. The geometric attribute data specifically includes the strength grade, material type, and durability of C30 concrete.
[0078] Understandably, structured coding can convert geometric attribute data into vector form; for example, intensity levels in geometric attribute data can be converted into vector form R. 强度 Let R be the first intermediate vector representation; the material type in the geometric property data can be converted into vector form R. 类型 Let R be the second intermediate vector representation; and the durability in the geometric attribute data can be converted into vector form R. 耐久性 Let be the third intermediate vector.
[0079] It is worth noting that after obtaining all intermediate vector representations, all intermediate vector representations can be integrated based on a concatenation operation, and the overall form of all intermediate vector representations can be determined as the second vector representation. Furthermore, the geometric attribute data in the examples of this application are for illustrative purposes only and are not intended to limit this application. For example, geometric attribute data may also include compressive strength, slump, etc., which will not be elaborated upon here.
[0080] Step 140: Based on the first vector representation and the second vector representation, perform entity intent recognition processing to obtain the building obstacle breach entity intent;
[0081] In this embodiment of the application, entity relationship extraction and intent recognition can be performed on the first vector representation and the second vector representation to obtain the building obstacle breach entity intent corresponding to the building obstacle breach data.
[0082] In some embodiments, the step of performing entity intent recognition processing based on the first vector representation and the second vector representation to obtain the building obstacle breach entity intent includes:
[0083] The first vector representation and the second vector representation are concatenated to obtain the third vector representation;
[0084] In this embodiment of the application, the first vector can be represented as (Xsteel, Xmixed, ..., Xconcrete). C X3, X0, Xmix, Xconcrete, Xsoil) and the second vector representation (R) 强度 R 类型 R 耐久性 Perform vector concatenation to obtain the third vector representation (Fsteel, Fmix, ..., Fconcrete). C F3, F0, F-mixed, F-concrete, F-soil). Among them:
[0085]
[0086] Feature extraction is performed on the third vector representation to obtain building obstacle breach features;
[0087] Further, the feature extraction from the third vector representation to obtain building obstacle breach features includes:
[0088] Forward feature extraction is performed on the third vector representation to obtain the forward hidden layer vector representation;
[0089] Backward feature extraction is performed on the third vector representation to obtain the backward hidden layer vector representation;
[0090] Based on the forward hidden layer vector representation, the corresponding backward hidden layer vector representation is concatenated to obtain the building obstacle breach feature.
[0091] In this embodiment, feature information contained in the third vector representation can be extracted using a neural network model to obtain building obstacle features. Specifically, this embodiment uses a Bidirectional Long Short-Term Memory (BiLSTM) network model as an example. This BiLSTM model includes a forward LSTM network and a backward LSTM network. Forward feature extraction can be performed by inputting the third vector representation into the forward LSTM network to obtain the forward hidden layer vector representation corresponding to the third vector representation. Backward feature extraction can be performed by inputting the third vector representation into the backward LSTM network to obtain the backward hidden layer vector representation. Then, the forward and backward hidden layer vector representations with corresponding relationships are concatenated to obtain the building obstacle features.
[0092] For example, for F in the third vector representation 混 F 凝 and F 土 After being fed into a forward LSTM network, the forward hidden layer vector output by the forward LSTM network can be represented as {h}. L0 ,h L1 ,h L2}; and after being input into the reverse LSTM network, the backward hidden layer vector output by the reverse LSTM network is represented as {h R0 ,h R1 ,h R2}; The architectural obstacle breach features obtained by splicing can be [L 混 ,L 凝 ,L 土 ].in:
[0093]
[0094] L 混 To be with F 混 Corresponding building obstacle breach characteristics; h L0 To be with F 混 The corresponding forward hidden layer vector; h R2 To be with F 混 The corresponding backward hidden layer vector. L 凝 To be with F 凝 Corresponding building obstacle breach characteristics; h L1 To be with L凝 The corresponding forward hidden layer vector; h R1 To be with L 凝 The corresponding backward hidden layer vector. L 土 To be with F 土 Corresponding building obstacle breach characteristics; h L2 To be with F 土 The corresponding forward hidden layer vector; h R0 To be with F 土 The corresponding backward hidden layer vector.
[0095] The building obstacle breach features are labeled and identified to obtain the building obstacle breach entity intent.
[0096] Furthermore, the annotation and recognition processing of the building obstacle-breaking features is performed to obtain the building obstacle-breaking entity intent;
[0097] The building obstacle breach features are subjected to conditional discrimination processing to obtain a global label sequence corresponding to the building obstacle breach features;
[0098] The global label sequence is merged to obtain the building obstacle breaching entity intent.
[0099] In this embodiment, after obtaining the building obstacle breach features containing contextual information, the entity and type label corresponding to the building obstacle breach data can be determined based on the building obstacle breach features. Specifically, the conditional discrimination process can involve inputting the building obstacle breach features into a conditional random field (CRF) model, and outputting a global label sequence corresponding to the feature sequence of the building obstacle breach features. This global label sequence includes several labels, and adjacent labels in the global label sequence may be the same or different. Then, consecutive and identical labels in the global label sequence are merged to obtain an entity with a type label, denoted as the building obstacle breach entity intent.
[0100] For example, for building barrier characteristics (L steel, L concrete, ..., L... C The corresponding global label sequence can be (B-building structure, I-building structure, I-building structure, I-building structure, I-building structure, I-building structure, I-building structure, O, O, O, O, O, O, B-building features, I-building features, I-building features, I-building features, I-building features, I-building features, I-building features). Then, continuous and identical labels are merged to obtain the building obstacle breach entity intent. The building obstacle breach entity intent specifically includes "steel-concrete structure building" which belongs to the building structure and "C30 concrete" which belongs to the building features.
[0101] Step 150: Generate a knowledge graph of building obstacle breaching based on the intent of the building obstacle breaching entity.
[0102] In this embodiment, based on the intent of the building obstacle breach entity, the corresponding building obstacle breach data can be organized into a knowledge graph using entity-relationship-entity triples, thereby obtaining a building obstacle breach knowledge graph. Furthermore, in practical applications, multiple different building obstacle breach data can be acquired, and the building obstacle breach entity intent corresponding to each data point can be obtained. Then, based on all the acquired building obstacle breach entity intents, a building obstacle breach knowledge graph can be generated.
[0103] This application provides a question-and-answer method for a knowledge graph of building obstacle avoidance, including:
[0104] Retrieve unanswered questions about building obstacle avoidance;
[0105] The question about building obstacle breaches is processed by entity and intent recognition to obtain entity relationship information;
[0106] Based on the entity relationship information, a knowledge query is performed on the aforementioned building obstacle breach knowledge graph to obtain the building obstacle breach response result corresponding to the building obstacle breach question.
[0107] In this embodiment of the application, the question to be answered regarding building obstacle removal can be a natural question provided by the user; entity and intent recognition processing can be performed by extracting entities and relationships from the user's natural question using several models used to construct the knowledge graph, thereby obtaining entity relationship information; then, based on the entity relationship information, a query operation is performed in the building obstacle removal knowledge graph to obtain the answer corresponding to the building obstacle removal question, which is recorded as the building obstacle removal response result.
[0108] The following describes in detail, with reference to the accompanying drawings, a system for generating a knowledge graph of building obstacle avoidance according to an embodiment of this application.
[0109] Reference Figure 2 The system for generating a knowledge graph of building obstacle avoidance proposed in this application includes:
[0110] The first processing unit 101 is used to acquire building knowledge models and building obstacle data;
[0111] The second processing unit 102 is used to input the building obstacle data into a text vector model to obtain a first vector representation output by the text vector model. The first vector representation is a word vector corresponding to the text semantic features of the building obstacle data.
[0112] The third processing unit 103 is used to input the building obstacle data into the building knowledge model to obtain a second vector representation, the second vector representation being used to characterize the vector representation of the geometric attribute data corresponding to the building obstacle data;
[0113] The fourth processing unit 104 is used to perform entity intent recognition processing based on the first vector representation and the second vector representation to obtain the building obstacle-breaking entity intent;
[0114] The fifth processing unit 105 is used to generate a building obstacle breach knowledge graph based on the building obstacle breach entity's intent.
[0115] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.
[0116] Reference Figure 3 This application also provides an electronic device, including:
[0117] At least one processor 201;
[0118] At least one memory 202 is used to store at least one program;
[0119] When the at least one program is executed by the at least one processor 201, the at least one processor 201 implements the method embodiment described above.
[0120] Similarly, it can be understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0121] This application also provides a computer-readable storage medium storing a program executable by a processor 201, which, when executed by the processor 201, is used to implement the above-described method embodiments.
[0122] Similarly, the content of the above method embodiments is applicable to the present computer-readable storage medium embodiments. The specific functions implemented by the present computer-readable storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.
[0123] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.
[0124] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods.
[0125] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this application are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.
[0126] Furthermore, although this application is described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding this application. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional technology for an engineer. Therefore, those skilled in the art can implement the application set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of this application, which is determined by the full scope of the appended claims and their equivalents.
[0127] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods in the embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0128] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0129] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0130] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0131] In the foregoing description of this specification, the references to terms such as "one embodiment," "another embodiment," or "some embodiments," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0132] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.
[0133] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.
Claims
1. A method for generating a building breach knowledge graph, characterized in that, The method comprises the following steps: obtaining a building knowledge model and building obstacle data; inputting the building obstacle data into a text vector model to obtain a first vector representation output by the text vector model, the first vector representation being a word vector corresponding to a text semantic feature of the building obstacle data; inputting the building obstacle data into the building knowledge model to obtain a second vector representation, the second vector representation being a vector representation of geometric attribute data corresponding to the building obstacle data; performing entity intention recognition processing according to the first vector representation and the second vector representation to obtain a building obstacle entity intention; generating a building obstacle knowledge graph according to the building obstacle entity intention.
2. The method of claim 1, wherein, The step of inputting the building obstacle data into the text vector model to obtain the first vector representation output by the text vector model comprises the following steps: performing text character extraction on the building obstacle data to obtain a plurality of building obstacle characters; performing bidirectional vector coding on all the building obstacle characters to obtain the first vector representation.
3. The method of claim 1, wherein, The step of inputting the building obstacle data into the building knowledge model to obtain the second vector representation comprises the following steps: obtaining geometric attribute data corresponding to the building obstacle data; performing structured coding on the geometric attribute data to obtain a plurality of intermediate vector representations; performing vector splicing on all the intermediate vector representations to obtain the second vector representation.
4. The method according to any one of claims 1 to 3, characterized in that, The step of performing entity intention recognition processing according to the first vector representation and the second vector representation to obtain a building obstacle entity intention comprises the following steps: performing vector splicing on the first vector representation and the second vector representation to obtain a third vector representation; performing feature extraction on the third vector representation to obtain a building obstacle feature; performing label recognition processing on the building obstacle feature to obtain the building obstacle entity intention.
5. The method of claim 4, wherein, The step of performing feature extraction on the third vector representation to obtain a building obstacle feature comprises the following steps: performing forward feature extraction on the third vector representation to obtain a forward hidden layer vector representation; performing backward feature extraction on the third vector representation to obtain a backward hidden layer vector representation; performing feature splicing on the corresponding backward hidden layer vector representation according to the forward hidden layer vector representation to obtain the building obstacle feature.
6. The method of claim 4, wherein, The step of performing label recognition processing on the building obstacle feature to obtain the building obstacle entity intention comprises the following steps: performing conditional discrimination processing on the building obstacle feature to obtain a global label sequence corresponding to the building obstacle feature; performing label merging on the global label sequence to obtain the building obstacle entity intention. 7.A method for answering questions of a building barrier knowledge graph, characterized in that, The method comprises the following steps: obtaining a building obstacle question to be answered; performing entity and intention recognition processing on the building obstacle question to obtain entity relationship information; performing knowledge query on the building obstacle knowledge graph according to the entity relationship information to obtain a building obstacle response result corresponding to the building obstacle question. 8.A system for generating a building breach knowledge graph, characterized in that, The method comprises the following steps: a first processing unit is configured to obtain a building knowledge model and building obstacle data; a second processing unit, configured to input the building damage data into a text vector model to obtain a first vector representation output by the text vector model, the first vector representation being a word vector corresponding to a text semantic feature of the building damage data; a third processing unit, configured to input the building damage data into the building knowledge model to obtain a second vector representation, the second vector representation being a vector representation of geometric attribute data corresponding to the building damage data; a fourth processing unit, configured to perform entity intention recognition processing according to the first vector representation and the second vector representation to obtain a building damage entity intention; a fifth processing unit, configured to generate a building damage knowledge graph according to the building damage entity intention.
9. An electronic device, comprising: comprising: at least one processor; at least one memory for storing at least one program; when the at least one program is executed by the at least one processor, the at least one processor implements the method of any one of claims 1-6 or the method of claim 7.
10. A computer readable storage medium having stored therein a program which is executable by a processor, characterized in that, The program executable by the processor, when executed by the processor, is used to implement the method of any one of claims 1-6 or the method of claim 7.