Intelligent management method and system for engineering cost data based on Internet search service

Abstract

The present application relates to the field of intelligent data management, and specifically to an intelligent management method and system for engineering cost data based on an Internet search service. It collects engineering project retrieval inputs input by users, and captures engineering cost project data based on an Internet search service, and then uses a large language model to perform semantic expression perfection on the engineering project retrieval input, and then sends it together with the engineering cost project data into a data processing and semantic understanding model based on artificial intelligence and natural language processing technology for processing, thereby capturing the fine-grained semantic features and retrieval demand semantic features of the engineering cost project, and using the retrieval demand semantic features as query features to perform fine-grained semantic query matching on the engineering cost project data, so as to decide whether to return the engineering cost project data based on the degree of matching between the two. In this way, the smooth progress of the engineering project and the rational allocation of resources can be ensured.

Description

Technical Field

[0001] The present application relates to the field of intelligent data management, and specifically to an intelligent management method and system for engineering cost data based on Internet search services. Background Art

[0002] Project cost data management is a crucial link in the construction industry, which involves project budget, cost control, project planning, financial analysis and decision support. With the acceleration of urbanization and the continuous expansion of the construction market, the scale and complexity of engineering projects are also increasing, resulting in an increasingly urgent need for the management of engineering cost data. Effective engineering cost management and project planning can not only help companies make wise decisions in the early stages of a project to control costs and avoid unnecessary cost increases in the later stages, but also ensure the smooth progress of the project and the rational allocation of resources.

[0003] In traditional engineering cost data management, engineering cost data mainly comes from the company's past project experience, supplier quotations, standard price lists issued by industry associations, etc. This method has the problem of information islands, which makes the coverage of engineering cost and project data relatively narrow, making it difficult to fully reflect the latest market conditions and the specific characteristics of the project, and unable to flexibly respond to changes in the market and project requirements. In addition, the traditional engineering cost data management method usually relies on manual retrieval and manual experience to query engineering cost project data and make engineering project decisions. This method has a low level of intelligence, is not only inefficient, but also difficult to fully utilize the large amount of semantics in engineering cost and project data, resulting in large deviations in the management and decision-making of engineering cost project data, which in turn affects the financial health of the project and the timely completion of the project.

[0004] Therefore, an optimized engineering cost data management solution is desired. Summary of the invention

[0005] This application is made in consideration of the above problems. One purpose of this application is to provide a method and system for intelligent management of engineering cost data based on Internet search services.

[0006] The embodiment of the present application provides a method for intelligent management of engineering cost data based on Internet search services, which includes:

[0007] Obtaining engineering project search input entered by a user;

[0008] Performing semantic expression improvement processing on the engineering project search input and then performing semantic encoding to obtain a semantic encoding feature vector of the engineering project search input;

[0009] Capture the first engineering cost project data based on Internet search services;

[0010] Performing word-granularity-based semantic coding of the cost item data of the first engineering cost item to obtain a sequence of word-granularity semantic coding feature vectors of the first engineering cost item data;

[0011] Taking the engineering project retrieval input semantic encoding feature vector as the query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as the sequence of library feature vectors, inputting the query feature vector and the sequence of library feature vectors into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector;

[0012] Based on the retrieval input-cost item cross-domain query matching feature vector, a management result is generated, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result.

[0013] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, the engineering project search input is semantically expressed and then semantically encoded to obtain the engineering project search input semantic encoding feature vector, including:

[0014] Passing the engineering project search input through a search expression semantic improver based on an AIGC model to obtain a perfected engineering project search input;

[0015] The improved engineering project retrieval input is semantically encoded to obtain a semantic encoding feature vector of the engineering project retrieval input.

[0016] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, the first engineering cost item data is semantically encoded based on word granularity to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors, including:

[0017] After word segmentation processing is performed on the first engineering cost item data, a sequence of the first engineering cost item word granularity semantic encoding feature vectors is obtained by passing it through a cost item semantic encoder including a word embedding layer.

[0018] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, the engineering project retrieval input semantic encoding feature vector is used as a query feature vector and the sequence of the first engineering cost project word granularity semantic encoding feature vector is used as a sequence of library feature vectors, and the query feature vector and the sequence of the library feature vector are input into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost project cross-domain query matching feature vector, including:

[0019] Performing a linear transformation on each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a first engineering cost item word granularity semantic coding feature vector after each linear transformation;

[0020] Calculating the feature distribution difference energy coefficients between the engineering project retrieval input semantic encoding feature vector and each row vector in the initial key matrix to obtain a set of first engineering cost item word granularity semantic feature distribution difference energy coefficients;

[0021] Compare each first engineering cost item word granularity semantic feature distribution difference energy coefficient in the set of the first engineering cost item word granularity semantic feature distribution difference energy coefficient with a predetermined threshold value, and in response to the first engineering cost item word granularity semantic feature distribution difference energy coefficient being less than the predetermined threshold value, use the mean vector between the row vector in the initial key matrix corresponding to the first engineering cost item word granularity semantic feature distribution difference energy coefficient and the engineering project retrieval input semantic encoding feature vector as the updated row vector to obtain an updated key matrix;

[0022] Using a value embedding matrix and a value embedding vector to process the engineering project retrieval input semantic encoding feature vector to obtain an engineering project retrieval input semantic value vector;

[0023] Taking the engineering project retrieval input semantic encoding feature vector as a query vector, each updated row vector in the update key matrix as a key vector and the engineering project retrieval input semantic value vector as a value vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of first engineering cost item word granularity semantic cross-domain query attention vectors;

[0024] The position-wise mean vector of the sequence of the first engineering cost item word granularity semantic cross-domain query attention vector is calculated to obtain the retrieval input-cost item cross-domain query matching feature vector.

[0025] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vector is linearly transformed to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​the first engineering cost item word granularity semantic coding feature vector after each linear transformation, including:

[0026] Each first engineering cost item word granularity semantic encoding feature vector in the sequence of the first engineering cost item word granularity semantic encoding feature vector is multiplied by the corresponding linear weight matrix and then added by position with the linear bias vector to obtain a sequence of the first engineering cost item word granularity semantic encoding feature vector after linear transformation;

[0027] Each of the first engineering cost item word granularity semantic coding feature vectors after linear transformation in the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation is used as a row vector of a matrix, so as to arrange the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation in a matrix to obtain the initial key matrix.

[0028] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, the feature distribution difference energy coefficient between the engineering project retrieval input semantic encoding feature vector and each row vector in the initial key matrix is ​​calculated to obtain a set of first engineering cost project word granularity semantic feature distribution difference energy coefficients, including:

[0029] Respectively calculating the positional difference, positional dot product and positional addition between the engineering project retrieval input semantic encoding feature vector and the row vector in the initial key matrix to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantics difference feature vector, the engineering project retrieval input semantics-first engineering cost item word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost item word granularity semantics sum feature vector;

[0030] The engineering project retrieval input semantics-first engineering cost item word granularity semantics differential feature vector, the engineering project retrieval input semantics-first engineering cost item word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost item word granularity semantics sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantics feature association representation vector;

[0031] Extracting the maximum value of the feature in the engineering project retrieval input semantics-first engineering cost item word granularity semantic feature association representation vector to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantic association maximum representation value;

[0032] Calculating the mean and variance of the row vectors in the initial key matrix to obtain the first engineering cost item word granularity semantic feature mean and the first engineering cost item word granularity semantic feature variance;

[0033] Adding the first engineering cost item word granularity semantic feature variance to the hyperparameter to obtain the first feature distribution difference energy coefficient;

[0034] Calculate the square of the difference between the maximum representation value of the engineering project search input semantics-first engineering cost item word granularity semantic association and the mean value of the first engineering cost item word granularity semantic feature to obtain a semantic feature difference value;

[0035] The first engineering cost item word granularity semantic feature variance is multiplied by a constant of two and then added to the semantic feature difference value and the hyperparameter to obtain a second feature distribution difference energy coefficient;

[0036] The first feature distribution difference energy coefficient is divided by the second feature distribution difference energy coefficient to obtain the first engineering cost item word granularity semantic feature distribution difference energy coefficient.

[0037] For example, according to the intelligent management method of engineering cost data based on Internet search service in an embodiment of the present application, the engineering project retrieval input semantic encoding feature vector is processed using a value embedding matrix and a value embedding vector to obtain an engineering project retrieval input semantic value vector, including:

[0038] The matrix multiplication between the engineering project retrieval input semantic coding feature vector and the value embedding matrix is ​​calculated, and then the matrix multiplication is added to the value embedding vector by position to obtain the engineering project retrieval input semantic value vector.

[0039] For example, according to the intelligent management method of engineering cost data based on Internet search service of an embodiment of the present application, the engineering project retrieval input semantic encoding feature vector is used as a query vector, each updated row vector in the update key matrix is ​​used as a key vector, and the engineering project retrieval input semantic value vector is used as a value vector, and the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of the first engineering cost project word granularity semantic cross-domain query attention vector, including:

[0040] After calculating the product between the query vector and the transposed vector of the key vector, dividing the obtained query-key feature association value by the square root of the length of the key vector to obtain a query-key feature association coefficient;

[0041] Inputting the query-key feature association coefficient into a softmax function to obtain a query-key feature association weight value;

[0042] The query-key feature association weight value is used as a weight, and the transposed vector of the value vector is weighted to obtain a first engineering cost item word granularity semantic cross-domain query attention vector.

[0043] For example, according to the intelligent management method of engineering cost data based on Internet search service in an embodiment of the present application, a management result is generated based on the retrieval input-cost item cross-domain query matching feature vector, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result, including:

[0044] The retrieval input-cost item cross-domain query matching feature vector is input into a classifier-based intelligent management result generator to obtain the management result, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result.

[0045] The embodiment of the present application also provides an intelligent management system for engineering cost data based on Internet search services, which includes:

[0046] A search input acquisition module is used to acquire the engineering project search input input by the user;

[0047] A retrieval input semantic coding module, used for performing semantic expression improvement processing on the engineering project retrieval input and then performing semantic coding to obtain a semantic coding feature vector of the engineering project retrieval input;

[0048] A project data capture module, used for capturing the first engineering cost project data based on an Internet search service;

[0049] A cost item semantic encoding module, used for performing word granularity-based cost item semantic encoding on the first engineering cost item data to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors;

[0050] A local association strengthening module is used to use the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as a sequence of library feature vectors, and input the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association strengthening to obtain a retrieval input-cost item cross-domain query matching feature vector;

[0051] A management result generation module is used to generate a management result based on the retrieval input-cost item cross-domain query matching feature vector, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result.

[0052] According to the intelligent management method and system of engineering cost data based on Internet search service of the embodiment of the present application, it collects the engineering project search input input by the user, and captures the engineering cost project data based on the Internet search service, and then uses the large language model to perform semantic expression improvement on the engineering project search input, and then sends it together with the engineering cost project data into the data processing and semantic understanding model based on artificial intelligence and natural language processing technology for processing, so as to capture the fine-grained semantic features and search demand semantic features of the engineering cost project, and use the search demand semantic features as query features to perform fine-grained semantic query matching on the engineering cost project data, so as to decide whether to return the engineering cost project data according to the matching degree between the two. In this way, the smooth progress of the engineering project and the rational allocation of resources can be ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0054] Figure 1 A schematic diagram of the application architecture of the intelligent management method for engineering cost data based on Internet search service in an embodiment of the present application is shown;

[0055] Figure 2 A flowchart of an intelligent management method for engineering cost data based on Internet search services in an embodiment of the present application is shown;

[0056] Figure 3 A flowchart of sub-step S550 of the method for intelligent management of engineering cost data based on Internet search service in an embodiment of the present application is shown;

[0057] Figure 4 A schematic diagram of the structure of an intelligent management system for engineering cost data based on Internet search services in an embodiment of the present application is shown; and

[0058] Figure 5 An application scenario diagram of the intelligent management method of engineering cost data based on Internet search services in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0059] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0060] The terms used in this specification are those common terms currently widely used in the art in consideration of the functions of the present application, but these terms may vary according to the intentions of those skilled in the art, precedents, or new technologies in the art. In addition, specific terms may be selected, and in this case, their detailed meanings will be described in the detailed description of the present application. Therefore, the terms used in the specification should not be understood as simple names, but rather as a general description based on the meaning of the terms and the present application.

[0061] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0062] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. At the same time, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0063] Figure 1 A schematic diagram of the application architecture of an intelligent management method for engineering cost data based on Internet search services in an embodiment of the present application is shown, including a server 100 and a terminal device 200.

[0064] The terminal device 200 and the server 100 can be connected via the Internet to achieve mutual communication. Optionally, the above-mentioned Internet uses standard communication technology and / or protocol. The Internet is usually the Internet, but it can also be any network, including but not limited to any combination of a local area network (LAN), a metropolitan area network (MAN), a wide area network (WAN), a mobile, wired or wireless network, a dedicated network or a virtual private network. In some embodiments, the data exchanged through the network is represented by technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec) can also be used to encrypt all or some links. In other embodiments, customized and / or dedicated data communication technologies can also be used to replace or supplement the above-mentioned data communication technologies.

[0065] The server 100 can provide various network services for the terminal device 200, wherein the server 100 can be a single server, a server cluster consisting of several servers, or a cloud computing center. Specifically, the server 100 may include a processor 110 (Center Processing Unit, CPU), a memory 120, an input device 130, and an output device 140, etc. The input device 130 may include a keyboard, a mouse, a touch screen, etc., and the output device 140 may include a display device, such as a liquid crystal display (Liquid Crystal Display, LCD), a cathode ray tube (Cathode Ray Tube, CRT), etc.

[0066] The memory 120 may include a read-only memory (ROM) and a random access memory (RAM), and provides the processor 110 with program instructions and data stored in the memory 120. In the embodiment of the present application, the memory 120 may be used to store the program of the intelligent management method for engineering cost data based on Internet search service in the embodiment of the present application.

[0067] The processor 110 calls the program instructions stored in the memory 120, and the processor 110 is used to execute the steps of any one of the engineering cost data intelligent management methods based on Internet search services in the embodiments of the present application according to the obtained program instructions.

[0068] In addition, the application architecture diagram in the embodiment of the present application is intended to more clearly illustrate the technical solution in the embodiment of the present application, and does not constitute a limitation on the technical solution provided in the embodiment of the present application. Of course, for other application architectures and business applications, the technical solution provided in the embodiment of the present application is also applicable to similar problems.

[0069] The following is a non-restrictive explanation of the intelligent management method of engineering cost data based on Internet search services provided by at least one embodiment of the present application through several examples or embodiments. As described below, different features in these specific examples or embodiments can be combined with each other without conflicting with each other to obtain new examples or embodiments, and these new examples or embodiments also fall within the scope of protection of the present application.

[0070] In response to the above technical problems, in the technical solution of the present application, a method for intelligent management of engineering cost data based on Internet search services is proposed, which can use the massive data resources on the Internet to automatically retrieve engineering cost project data that meets the user's engineering project needs, thereby improving the intelligence level of engineering cost data management, avoiding the information islands and artificial inefficiency and low accuracy problems of traditional solutions, so as to ensure the smooth progress of engineering projects and the rational allocation of resources.

[0071] Specifically, the technical concept of the present application is to collect the engineering project search input input by the user, and capture the engineering cost project data based on the Internet search service, and then use the large language model to perform semantic expression improvement on the engineering project search input, and then send it together with the engineering cost project data into the data processing and semantic understanding model based on artificial intelligence and natural language processing technology for processing, so as to capture the fine-grained semantic features and retrieval demand semantic features of the engineering cost project, and use the retrieval demand semantic features as query features to perform fine-grained semantic query matching on the engineering cost project data, so as to decide whether to return this engineering cost project data based on the degree of matching between the two. In this way, the relevant engineering cost project data can be matched according to the user's engineering project retrieval needs, thereby realizing more intelligent engineering cost data management, ensuring the smooth progress of engineering projects and the rational allocation of resources.

[0072] Figure 2 The flowchart of the intelligent management method of engineering cost data based on Internet search service in the embodiment of the present application is shown. For example, the intelligent management method of engineering cost data based on Internet search service can be executed by a server, which can be Figure 1 The server 100 shown in FIG. Figure 2 As shown, according to the embodiment of the present application, the intelligent management method of engineering cost data based on Internet search service includes the following steps: S510, obtaining an engineering project search input input by a user; S520, performing semantic expression improvement processing on the engineering project search input and then performing semantic encoding to obtain a semantic encoding feature vector of the engineering project search input; S530, capturing first engineering cost project data based on Internet search service; S540, performing word granularity-based cost project semantic encoding on the first engineering cost project data to obtain a sequence of word granularity semantic encoding feature vectors of the first engineering cost project; S5 50, using the semantic encoding feature vector of the engineering project retrieval input as the query feature vector and the sequence of the semantic encoding feature vectors of the first engineering cost item word granularity as the sequence of library feature vectors, inputting the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector; S560, generating a management result based on the retrieval input-cost item cross-domain query matching feature vector, the management result being used to indicate whether to return the first engineering cost item data as the best matching result.

[0073] More specifically, in the technical solution of the present application, first, the engineering project retrieval input input by the user is obtained, and the engineering project retrieval input is passed through the retrieval expression semantic improver based on the AIGC model to obtain the perfected engineering project retrieval input. It is worth mentioning that the AIGC model, namely the Artificial Intelligence Generated Content model, is a model that uses artificial intelligence technology to automatically generate text, images, audio and other types of content. In the intelligent management method of engineering cost data, the reasons and effects of using the retrieval expression semantic improver based on the AIGC model include the following points: 1. Improve semantic understanding ability: The AIGC model can understand the deep meaning of the engineering project retrieval input input by the user, so as to more accurately capture the user's query intention. 2. Enhance the integrity of expression: The user input may be incomplete or vague. The AIGC model can make the retrieval input more complete and accurate by predicting and completing the user's potential needs. 3. Improve retrieval efficiency: Through the improvement of semantic expression, invalid retrieval caused by unclear input can be reduced, and the efficiency and accuracy of retrieval can be improved. 4. Strong adaptability: The AIGC model can adapt to different query habits and expressions, and provide users with more personalized retrieval services. 5. Handle complex queries: For queries that contain multiple concepts or complex logical relationships, the AIGC model can better understand and process them, providing more accurate retrieval results. 6. Reduce manual intervention: The automated semantic improvement process reduces manual intervention in retrieval input, reduces operational complexity and error rates. 7. Continuous learning and optimization: The AIGC model has the ability to learn, and can continuously optimize based on user feedback and usage habits to improve retrieval quality. 8. Support cross-domain retrieval: The AIGC model can handle cross-domain semantic encoding, so that the retrieval of engineering cost data is not limited to specific fields, increasing the breadth and depth of retrieval. Therefore, by using a retrieval expression semantic improver based on the AIGC model, the intelligent management system for engineering cost data can provide more intelligent, efficient and accurate retrieval services, thereby improving the quality and efficiency of the entire engineering cost data management.

[0074] Next, considering that the engineering project retrieval information input by the user is usually given in the form of natural language, and natural language has ambiguity and ambiguity, different expressions may refer to the same meaning. Although the retrieval expression semantic perfector can semantically perfect the expression of the engineering project retrieval input input by the user, it is not deep enough in semantic understanding. Therefore, in order to be able to more comprehensively and meticulously semantically understand and analyze the engineering project retrieval requirements input by the user, so as to capture the user's true engineering project retrieval intention, so as to subsequently return the engineering cost project data that meets the user's needs, in the technical solution of the present application, the improved engineering project retrieval input is further semantically encoded to extract the semantic features in the improved engineering project retrieval input to obtain the engineering project retrieval input semantic encoding feature vector.

[0075] Correspondingly, in step S520, the engineering project retrieval input is semantically perfected and then semantically encoded to obtain a semantic encoding feature vector of the engineering project retrieval input, including: passing the engineering project retrieval input through a retrieval expression semantic perfector based on the AIGC model to obtain a perfected engineering project retrieval input; and semantically encoding the perfected engineering project retrieval input to obtain a semantic encoding feature vector of the engineering project retrieval input.

[0076] It should be understood that there are a large number of public data sources on the Internet, including government websites, industry associations, academic papers, industry reports, etc. These data sources can provide rich data information related to engineering cost and project planning. Therefore, in order to be able to search for engineering cost project data that meets user needs from the Internet, in the technical solution of this application, it is necessary to capture the first engineering cost project data based on the Internet search service. Through the Internet search service, cost data related to engineering projects can be automatically captured from these data sources on the Internet, expanding the coverage and timeliness of the data.

[0077] Then, after word segmentation, the first engineering cost item data is encoded in a cost item semantic encoder including a word embedding layer to extract the semantic coding features based on word granularity in the first engineering cost item data, thereby obtaining a sequence of the first engineering cost item word granularity semantic coding feature vectors.

[0078] Correspondingly, in step S540, the first engineering cost item data is subjected to word granularity-based cost item semantic encoding to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors, including: performing word segmentation processing on the first engineering cost item data and then passing it through a cost item semantic encoder including a word embedding layer to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors.

[0079] Furthermore, considering that each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vector contains semantic coding features based on word granularity in the first engineering cost item data, and each word granularity semantic feature has a relevant semantic connection and potential influence. However, traditional context coding methods can usually only focus on the association between the query vector and the key vector, while ignoring the potential correlation between each first engineering cost item word granularity semantic coding feature vector as a key vector. If the implicit correlation information contained in each key vector in the key matrix composed of these first engineering cost item word granularity semantic coding feature vectors can be utilized, the accuracy of cross-domain query matching coding can be significantly improved. Based on this, in the technical solution of the present application, the engineering project retrieval input semantic encoding feature vector is further used as a query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector is used as a sequence of library feature vectors, and the sequence of the query feature vector and the library feature vector are input into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector.

[0080] Specifically, the cross-domain attention gate query encoder based on key matrix local association enhancement first performs a linear transformation on each library feature vector in the sequence of library feature vectors to adjust its feature representation dimension, so as to align it with the feature dimension of the query feature vector, so as to ensure that the sequence of the engineering project retrieval input semantic encoding feature vector and the first engineering cost project word granularity semantic encoding feature vector can be compared in the same feature space.

[0081] Next, the feature distribution difference between the query feature vector and each row vector (key vector) in the initial key matrix is ​​calculated to obtain a set of difference energy coefficients. That is, the semantic coding feature of the engineering project retrieval input is used as a soft anchor reference point and the feature distribution difference between the query feature vector and each key vector in the initial key matrix is ​​measured by the feature distribution energy difference coefficient to obtain a set of difference energy coefficients. It should be understood that, considering that the query feature vector is a common soft anchor reference point, the set of feature distribution energy difference coefficients as a whole also contains the local correlation information between the semantic coding features of the first engineering cost project word granularity. Therefore, the obtained difference energy coefficients reflect the similarity between the query vector and the key vector, and at the same time contain the local correlation information between the key vectors.

[0082] Then, the feature distribution difference energy coefficient between the engineering project retrieval input semantic coding features and the first engineering cost project word granularity semantic coding features is compared with a predetermined threshold. For the case where the feature distribution difference is less than the threshold, the row vector in the key matrix is ​​updated using the mean vector of the query feature vector and the corresponding row vector. Here, the update operation strengthens the library feature vector that is more similar to the query feature vector, that is, strengthens the features and semantics with higher relevance to the engineering project retrieval semantics in the sequence of the first engineering cost project word granularity semantic coding feature vector. Through the local association strengthening mechanism, the pertinence of the key matrix and the quality of subsequent semantic matching are improved, which helps to more accurately judge the degree of matching between the engineering project retrieval input semantic coding features and the first engineering cost project word granularity semantic coding features, thereby providing a basis for the subsequent task of returning the best engineering cost project data.

[0083] Furthermore, the semantic encoding feature vector of the engineering project retrieval input is used as the query vector, each updated row vector in the update key matrix is ​​used as the key vector, and the engineering project retrieval input semantic value vector is used as the value vector. The cross-domain attention gate query module based on the Transformer structure is used to combine the query vector, the key vector and the value vector to perform feature self-attention calculation to achieve cross-domain information integration. The effect of the cross-domain attention mechanism is to allow the model to flow and integrate information between different semantic feature domains, improve the richness of the semantic feature representation of engineering project retrieval semantics and engineering cost project word granularity, and the accuracy of the matching between engineering project retrieval semantics and the first engineering cost project word granularity semantics. Finally, by calculating the mean of the sequence of the first engineering cost project word granularity semantic cross-domain query attention vector by position, the final retrieval input-cost project cross-domain query matching feature vector is obtained. The purpose is to comprehensively consider the information of all cross-domain query attention vectors in the sequence and generate a comprehensive feature representation, which helps to improve the overall understanding and response ability of the model to complex engineering project retrieval semantic queries input by different users.

[0084] Accordingly, in step S550, if Figure 3As shown, the engineering project retrieval input semantic coding feature vector is used as the query feature vector and the sequence of the first engineering cost item word granularity semantic coding feature vector is used as the sequence of library feature vectors, and the query feature vector and the sequence of the library feature vector are input into a cross-domain attention gate query encoder based on the local association enhancement of the key matrix to obtain a retrieval input-cost item cross-domain query matching feature vector, including: S551, linearly transforming each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vector to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a first engineering cost item word granularity semantic coding feature vector after each linear transformation; S552, calculating the feature distribution difference energy coefficient between the engineering project retrieval input semantic coding feature vector and each row vector in the initial key matrix to obtain a set of first engineering cost item word granularity semantic feature distribution difference energy coefficients; S553, comparing each first engineering cost item word granularity semantic feature distribution difference energy coefficient in the set of the first engineering cost item word granularity semantic feature distribution difference energy coefficient with a predetermined threshold The values ​​are compared, and in response to the difference energy coefficient of the semantic feature distribution of the first engineering cost project word granularity is less than a predetermined threshold, the row vector in the initial key matrix corresponding to the difference energy coefficient of the semantic feature distribution of the first engineering cost project word granularity and the mean vector between the engineering project retrieval input semantic encoding feature vector are used as the updated row vector to obtain the updated key matrix; S554, the engineering project retrieval input semantic encoding feature vector is processed using the value embedding matrix and the value embedding vector to obtain the engineering project retrieval input semantic value vector; S555, the engineering project retrieval input semantic encoding feature vector is used as the query vector, each updated row vector in the updated key matrix is ​​used as the key vector, and the engineering project retrieval input semantic value vector is used as the value vector, and the query vector, the key vector and the value vector are input into the cross-domain attention gate query module based on the Transformer structure to obtain the sequence of the first engineering cost project word granularity semantic cross-domain query attention vector; S556, the positional mean vector of the sequence of the first engineering cost project word granularity semantic cross-domain query attention vector is calculated to obtain the retrieval input-cost project cross-domain query matching feature vector.

[0085] Wherein, in step S551, each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vector is linearly transformed to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​each first engineering cost item word granularity semantic coding feature vector after linear transformation, including: multiplying each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vector with the corresponding linear weight matrix respectively and then adding them with the linear bias vector by position to obtain a sequence of first engineering cost item word granularity semantic coding feature vectors after linear transformation; using each first engineering cost item word granularity semantic coding feature vector after linear transformation in the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation as the row vector of the matrix, so as to arrange the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation in a matrix to obtain the initial key matrix.

[0086] Among them, in step S552, the feature distribution difference energy coefficient between the engineering project retrieval input semantic coding feature vector and each row vector in the initial key matrix is ​​calculated to obtain a set of the first engineering cost item word granularity semantic feature distribution difference energy coefficients, including: respectively calculating the position difference, position dot multiplication and position addition between the engineering project retrieval input semantic coding feature vector and the row vector in the initial key matrix to obtain the engineering project retrieval input semantic-first engineering cost item word granularity semantic difference feature vector, the engineering project retrieval input semantic-first engineering cost item word granularity semantic difference feature vector. The engineering project retrieval input semantics-first engineering cost project word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost project word granularity semantics sum feature vector; the engineering project retrieval input semantics-first engineering cost project word granularity semantics differential feature vector, the engineering project retrieval input semantics-first engineering cost project word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost project word granularity semantics sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain the engineering project retrieval input semantics-first engineering cost project word granularity semantics sum feature vector. The engineering cost project word granularity semantic feature association representation vector is extracted; the feature maximum value in the engineering project retrieval input semantic-first engineering cost project word granularity semantic feature association representation vector is extracted to obtain the engineering project retrieval input semantic-first engineering cost project word granularity semantic association maximum representation value; the mean and variance of the row vector in the initial key matrix are calculated to obtain the first engineering cost project word granularity semantic feature mean and the first engineering cost project word granularity semantic feature variance; the first engineering cost project word granularity semantic feature variance is added to the hyperparameter to obtain the first feature distribution difference energy coefficient; the square of the difference between the engineering project retrieval input semantic-first engineering cost project word granularity semantic association maximum representation value and the first engineering cost project word granularity semantic feature mean is calculated to obtain the semantic feature difference value; the first engineering cost project word granularity semantic feature variance is multiplied by a constant two and then added to the semantic feature difference value and the hyperparameter to obtain the second feature distribution difference energy coefficient; the first feature distribution difference energy coefficient is divided by the second feature distribution difference energy coefficient to obtain the first engineering cost project word granularity semantic feature distribution difference energy coefficient.

[0087] Among them, in step S554, the engineering project retrieval input semantic coding feature vector is processed using a value embedding matrix and a value embedding vector to obtain an engineering project retrieval input semantic value vector, including: calculating the matrix multiplication between the engineering project retrieval input semantic coding feature vector and the value embedding matrix, and then adding it to the value embedding vector by position to obtain the engineering project retrieval input semantic value vector.

[0088] Among them, in step S555, the engineering project retrieval input semantic encoding feature vector is used as the query vector, each updated row vector in the update key matrix is ​​used as the key vector, and the engineering project retrieval input semantic value vector is used as the value vector. The query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of first engineering cost project word granularity semantic cross-domain query attention vectors, including: after calculating the product between the query vector and the transposed vector of the key vector, the obtained query-key feature association value is divided by the square root of the length of the key vector to obtain a query-key feature association coefficient; the query-key feature association coefficient is input into a softmax function to obtain a query-key feature association weight value; the query-key feature association weight value is used as a weight to weight the transposed vector of the value vector to obtain the first engineering cost project word granularity semantic cross-domain query attention vector.

[0089] In a specific example, the engineering project retrieval input semantic encoding feature vector is used as a query feature vector and the sequence of the first engineering cost project word granularity semantic encoding feature vector is used as a sequence of library feature vectors, and the query feature vector and the sequence of the library feature vector are input into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost project cross-domain query matching feature vector, including: using the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost project word granularity semantic encoding feature vector as a sequence of library feature vectors, and the query feature vector and the sequence of the library feature vector are input into the cross-domain attention gate query encoder based on key matrix local association enhancement to obtain the retrieval input-cost project cross-domain query matching feature vector using the following cross-domain attention gate query formula; wherein, the cross-domain attention gate query formula is:

[0090] K={k 1 ,k 2 ,...,k n}

[0091] M o ={k 1 ′; k 2 ′; ...; k n ′}

[0092] k i ′=f(k i ,W k ) = k i W k +b k

[0093]

[0094] v v =v q W q +b q

[0095]

[0096] Where K is the sequence of the semantic encoding feature vectors of the first engineering cost item word granularity, k 1 ,k 2 ,...,k n are the first engineering cost item word granularity semantic coding feature vectors in the sequence of the first engineering cost item word granularity semantic coding feature vectors, k i is the i-th first engineering cost item word granularity semantic encoding feature vector in the sequence of the first engineering cost item word granularity semantic encoding feature vectors, W k is the linear weight matrix, b k is the linear bias vector, f(k i ,W k ) is a linear transformation, k i ′ is the first engineering cost item word granularity semantic coding feature vector after linear transformation corresponding to the i-th first engineering cost item word granularity semantic coding feature vector, k 1 ′; k 2 ′; ...; k n ′ are the first engineering cost item word granularity semantic coding feature vectors after linear transformation in the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation, M o is the initial key matrix obtained by matrixing the semantic encoding feature vectors of the first engineering cost item word granularity after each linear transformation, v q is the query feature vector, Θ, ⊙ and They represent positional subtraction, positional multiplication, and positional addition, respectively. conv1D is a one-dimensional convolution operation. MaxPool is a maximum pooling operation. R i is the feature correlation coefficient corresponding to the semantic encoding feature vector of the i-th first engineering cost item word granularity, max(·) represents the maximum value operation, μ and σ 2 are the feature mean and feature variance corresponding to the i-th first engineering cost item word granularity semantic encoding feature vector, λ is a hyperparameter, e(R i ) is the feature distribution difference energy coefficient corresponding to the i-th first engineering cost item word granularity semantic encoding feature vector, θ is the predetermined threshold, k i ″ is the updated first engineering cost item word granularity semantic coding feature vector corresponding to the i-th first engineering cost item word granularity semantic coding feature vector, Mi is the updated key matrix obtained by matrixing the semantic encoding feature vectors of the first engineering cost item word granularity after each update, W q is the value embedding matrix, b q is the value embedding vector, v v is the value vector corresponding to the query feature vector, softmax(·) is the softmax function, d is the length of the first engineering cost item word granularity semantic encoding feature vector after each update, v pi is the cross-domain query attention vector corresponding to the i-th first engineering cost item word granularity semantic encoding feature vector, n is the number of vectors in the sequence of the first engineering cost item word granularity semantic encoding feature vector, v p The retrieval input is a cross-domain query matching feature vector of the cost item.

[0097] Then, the retrieval input-cost project cross-domain query matching feature vector is input into the classifier-based intelligent management result generator to obtain a management result, and the management result is used to indicate whether to return the first engineering cost project data as the best matching result. That is, the cross-domain query matching feature information between the retrieval input semantic coding feature of the engineering project and the first engineering cost project word granularity semantic coding feature is used for classification processing, so as to decide whether to return the engineering cost project data based on the degree of matching between the two. In this way, the relevant engineering cost project data can be matched according to the user's engineering project retrieval requirements, thereby realizing more intelligent engineering cost data management, ensuring the smooth progress of the engineering project and the rational allocation of resources.

[0098] Correspondingly, in step S560, a management result is generated based on the retrieval input-cost item cross-domain query matching feature vector, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result, including: inputting the retrieval input-cost item cross-domain query matching feature vector into a classifier-based intelligent management result generator to obtain the management result, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result.

[0099] It should be understood that the role of the classifier is to use the given categories and known training data to learn classification rules and classifiers, and then classify (or predict) unknown data. Logistic regression and SVM are often used to solve binary classification problems. For multi-class classification problems, logistic regression or SVM can also be used, but multiple binary classifications are required to form a multi-classification. However, this is prone to errors and is not efficient. Commonly used multi-classification methods include the Softmax classification function.

[0100] Preferably, the retrieval input-cost item cross-domain query matching feature vector is input into a classifier-based intelligent management result generator to obtain the management result, including: optimizing the retrieval input-cost item cross-domain query matching feature vector to obtain an optimized retrieval input-cost item cross-domain query matching feature vector; inputting the optimized retrieval input-cost item cross-domain query matching feature vector into the classifier-based intelligent management result generator to obtain the management result.

[0101] Among them, the retrieval input-cost item cross-domain query matching feature vector is optimized to obtain an optimized retrieval input-cost item cross-domain query matching feature vector, including: subtracting the 0 norm of the retrieval input-cost item cross-domain query matching feature vector from the length of the retrieval input-cost item cross-domain query matching feature vector to obtain a retrieval input-cost item cross-domain query matching isolated representation value; calculating a power function with each eigenvalue of the retrieval input-cost item cross-domain query matching feature vector as the base, the difference of the retrieval input-cost item cross-domain query matching isolated representation value minus one as the exponent, and multiplying it by the retrieval input-cost item cross-domain query matching isolated representation value plus one and then multiplying it by the retrieval input-cost item cross-domain query matching isolated representation value plus one. The base-2 logarithm of the product of the isolated representation value of the cross-domain query match is obtained to obtain the leading value of the cross-domain query match of the retrieval input-cost item; after multiplying each eigenvalue of the cross-domain query match feature vector of the retrieval input-cost item by the difference of the isolated representation value of the cross-domain query match of the retrieval input-cost item minus one and dividing it by the isolated representation value of the cross-domain query match of the retrieval input-cost item, an exponential function with a natural constant as the base is calculated to obtain the bias value of the cross-domain query match of the retrieval input-cost item; the leading value of the cross-domain query match of the retrieval input-cost item is added to the bias value of the cross-domain query match of the retrieval input-cost item to obtain each eigenvalue of the optimized cross-domain query match feature vector of the retrieval input-cost item.

[0102] Specifically, the retrieval input-cost item cross-domain query matching feature vector, for example, each feature value v of V is recorded as i , where v i The optimization process of ∈V is expressed as follows:

[0103]

[0104] n=L-‖V‖ 0

[0105] Wherein, V represents the cross-domain query matching feature vector of the retrieval input-cost item, v i represents the ith eigenvalue of the retrieval input-cost item cross-domain query matching feature vector, ‖V‖ 0represents the zero norm of the retrieval input-cost item cross-domain query matching feature vector, L represents the length of the retrieval input-cost item cross-domain query matching feature vector, n represents the retrieval input-cost item cross-domain query matching isolated representation value, exp(·) represents the exponential operation of the value, the exponential operation of the value represents the calculation of the natural exponential function value with the value as the power, log[·] represents the logarithmic function value with base 2, v i ′ represents the i-th eigenvalue of the optimized retrieval input-cost item cross-domain query matching feature vector.

[0106] Here, the engineering project retrieval input semantic encoding feature vector represents the text semantic features of the improved engineering project retrieval input after the engineering project retrieval input is expressed and improved by the AIGC model, and the sequence of the first engineering cost item word granularity semantic encoding feature vectors represents the word granularity text semantic context association features of the first engineering cost item data determined based on word segmentation. When the engineering project retrieval input semantic encoding feature vector is used as the query feature vector and the sequence of the first engineering cost project word granularity semantic encoding feature vectors is used as the sequence of library feature vectors, and the query feature vector and the sequence of the library feature vectors are input into the cross-domain attention gate query encoder based on the key matrix local association enhancement, considering the inconsistency of the text semantic compactness caused by the difference in the source text content length between the engineering project retrieval input semantic encoding feature vector and each first engineering cost project word granularity semantic encoding feature vector in the sequence of the first engineering cost project word granularity semantic encoding feature vector, this will cause the obtained retrieval input-cost project cross-domain query matching feature vector to have a cross-domain attention query offset. Therefore, it is expected to further improve the global regression comprehensibility of the cross-domain attention query matching of the retrieval input-cost project cross-domain query matching feature vector, thereby improving the accuracy of the management results obtained by its input into the classifier-based intelligent management result generator.

[0107] Therefore, for the high-dimensional feature manifold of the retrieval input-cost project cross-domain query matching feature vector, the eigenvalues ​​of the feature set are used as the vector field representation of the aggregation dimension. The value of the derivative of the vector field of the retrieval input-cost project cross-domain query matching feature vector at the isolated zero position is used as the order information to fix the local position of the eigenvalues ​​of its feature set, and a bias for the reversibility of the feature regression distribution of the retrieval input-cost project cross-domain query matching feature vector is added as a reward to achieve the mapping target tracking of the regression distribution of the retrieval input-cost project cross-domain query matching feature vector for the eigenvalue position, so that the feature set of the retrieval input-cost project cross-domain query matching feature vector can perceive the mapping migration to the aggregate distribution, thereby improving the accuracy of the management result obtained by the classifier-based intelligent management result generator input by the retrieval input-cost project cross-domain query matching feature vector by improving the global regression comprehensibility of the cross-domain focus query matching of the retrieval input-cost project cross-domain query matching feature vector. In this way, the massive data resources on the Internet can be used to automatically retrieve engineering cost project data that meets the user's engineering project needs, thereby improving the intelligence level of engineering cost data management, avoiding the information islands and artificial inefficiency and low accuracy problems of traditional solutions, and ensuring the smooth progress of engineering projects and the rational allocation of resources.

[0108] Based on the above embodiments, see Figure 4The figure is a schematic diagram of the structure of an intelligent management system 800 for engineering cost data based on Internet search service in an embodiment of the present application. The intelligent management system 800 for engineering cost data based on Internet search service includes: a search input acquisition module 810, which is used to obtain the engineering project search input input by the user; a search input semantic encoding module 820, which is used to perform semantic expression improvement processing on the engineering project search input and then perform semantic encoding to obtain the engineering project search input semantic encoding feature vector; a project data capture module 830, which is used to capture the first engineering cost project data based on Internet search service; a cost project semantic encoding module 840, which is used to perform word granularity-based cost project semantic encoding on the first engineering cost project data to obtain the first engineering cost project word granularity semantic encoding feature vector sequence; a local association enhancement module 850, which is used to use the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost project word granularity semantic encoding feature vector as a sequence of library feature vectors, and input the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost project cross-domain query matching feature vector; a management result generation module 860, which is used to generate a management result based on the retrieval input-cost project cross-domain query matching feature vector, and the management result is used to indicate whether to return the first engineering cost project data as the best matching result.

[0109] Here, those skilled in the art can understand that the specific functions and operations of each module in the above-mentioned engineering cost data intelligent management system 800 based on Internet search service have been referred to above. Figure 2 to Figure 3 The invention has been introduced in detail in the description of the intelligent management method of engineering cost data based on Internet search service, and therefore, its repeated description will be omitted.

[0110] Figure 5 FIG. 1 is an application scenario diagram of the intelligent management method of engineering cost data based on Internet search service according to an embodiment of the present application. Figure 5 As shown, in this application scenario, first, the engineering project search input entered by the user is obtained (for example, Figure 5 D1) and capture the first engineering cost project data based on an Internet search service (e.g., Figure 5 Then, the engineering project search input and the first engineering cost project data are input to a server (for example, Figure 5In S) as shown in , the server is able to use the engineering cost data intelligent management algorithm based on the Internet search service to process the engineering project retrieval input and the first engineering cost project data to obtain a management result indicating whether to return the first engineering cost project data as the best matching result.

[0111] Based on the above embodiments, another exemplary embodiment of an electronic device is also provided in the embodiments of the present application. In some possible implementations, the electronic device in the embodiments of the present application may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor can implement the steps of the intelligent management method for engineering cost data based on Internet search service in the above embodiments when executing the program.

[0112] For example, in the case of electronic equipment Figure 1 Taking the server 100 in the example as an example, the processor in the electronic device is the processor 110 in the server 100, and the memory in the electronic device is the memory 120 in the server 100.

[0113] The embodiment of the present application also provides a computer-readable storage medium, on which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, the intelligent management method for engineering cost data based on Internet search service according to the embodiment of the present application described with reference to the above figures can be executed. The computer-readable storage medium includes, but is not limited to, for example, volatile memory and / or non-volatile memory. The volatile memory may, for example, include a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may, for example, include a read-only memory (ROM), a hard disk, a flash memory, etc.

[0114] The embodiment of the present application also provides a computer program product or a computer program, which includes computer executable instructions, and the computer executable instructions are stored in a computer-readable storage medium. The processor of the computer device reads the computer executable instructions from the computer-readable storage medium, and the processor executes the computer executable instructions, so that the computer device executes the intelligent management method of engineering cost data based on Internet search service according to the embodiment of the present application.

[0115] Those skilled in the art will appreciate that the contents disclosed in this application may be subject to various variations and improvements. For example, the various devices or components described above may be implemented by hardware, or by software, firmware, or a combination of some or all of the three.

[0116] In addition, although the present application makes various references to certain units in the system according to embodiments of the present application, any number of different units can be used and run on the client and / or server. The units are only illustrative, and different aspects of the system and method can use different units.

[0117] Those skilled in the art will appreciate that all or part of the steps in the above method can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, such as a read-only memory, a disk or an optical disk. Optionally, all or part of the steps in the above embodiment can also be implemented using one or more integrated circuits. Accordingly, each module / unit in the above embodiment can be implemented in the form of hardware or in the form of software function modules. The present application is not limited to any particular form of combination of hardware and software.

[0118] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an idealized or extremely formal sense, unless explicitly defined as such herein.

[0119] The above is an explanation of the present application and should not be considered as limiting thereof. Although several exemplary embodiments of the present application have been described, those skilled in the art will readily appreciate that many modifications may be made to the exemplary embodiments without departing from the novel teachings and advantages of the present application.

Claims

1. A method for intelligent management of engineering cost data based on Internet search services, characterized in that: include: Obtaining engineering project search input entered by a user; Performing semantic expression improvement processing on the engineering project search input and then performing semantic encoding to obtain a semantic encoding feature vector of the engineering project search input; Capture the first engineering cost project data based on Internet search services; Performing word-granularity-based semantic coding of the cost item data of the first engineering cost item to obtain a sequence of word-granularity semantic coding feature vectors of the first engineering cost item data; Taking the engineering project retrieval input semantic encoding feature vector as the query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as the sequence of library feature vectors, inputting the query feature vector and the sequence of library feature vectors into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector; Based on the search input-cost item cross-domain query matching feature vector, generating a management result, the management result is used to indicate whether to return the first engineering cost item data as the best matching result; The method uses the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as a sequence of library feature vectors, and inputs the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector, including: Performing a linear transformation on each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a first engineering cost item word granularity semantic coding feature vector after each linear transformation; Calculating the feature distribution difference energy coefficients between the engineering project retrieval input semantic encoding feature vector and each row vector in the initial key matrix to obtain a set of first engineering cost item word granularity semantic feature distribution difference energy coefficients; Compare each first engineering cost item word granularity semantic feature distribution difference energy coefficient in the set of the first engineering cost item word granularity semantic feature distribution difference energy coefficient with a predetermined threshold value, and in response to the first engineering cost item word granularity semantic feature distribution difference energy coefficient being less than the predetermined threshold value, use the mean vector between the row vector in the initial key matrix corresponding to the first engineering cost item word granularity semantic feature distribution difference energy coefficient and the engineering project retrieval input semantic encoding feature vector as the updated row vector to obtain an updated key matrix; Using a value embedding matrix and a value embedding vector to process the engineering project retrieval input semantic encoding feature vector to obtain an engineering project retrieval input semantic value vector; Taking the engineering project retrieval input semantic encoding feature vector as a query vector, each updated row vector in the update key matrix as a key vector and the engineering project retrieval input semantic value vector as a value vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of first engineering cost item word granularity semantic cross-domain query attention vectors; The position-wise mean vector of the sequence of the first engineering cost item word granularity semantic cross-domain query attention vector is calculated to obtain the retrieval input-cost item cross-domain query matching feature vector.

2. The intelligent management method of engineering cost data based on Internet search service according to claim 1 is characterized in that: The engineering project search input is semantically perfected and then semantically encoded to obtain a semantic encoding feature vector of the engineering project search input, including: Passing the engineering project search input through a search expression semantic improver based on an AIGC model to obtain a perfected engineering project search input; The improved engineering project retrieval input is semantically encoded to obtain a semantic encoding feature vector of the engineering project retrieval input.

3. The intelligent management method of engineering cost data based on Internet search service according to claim 2 is characterized in that: The first engineering cost item data is semantically encoded based on word granularity to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors, including: After word segmentation processing is performed on the first engineering cost item data, a sequence of the first engineering cost item word granularity semantic encoding feature vectors is obtained by passing it through a cost item semantic encoder including a word embedding layer.

4. The intelligent management method of engineering cost data based on Internet search service according to claim 3 is characterized in that: Performing a linear transformation on each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a first engineering cost item word granularity semantic coding feature vector after each linear transformation, including: Each first engineering cost item word granularity semantic encoding feature vector in the sequence of the first engineering cost item word granularity semantic encoding feature vector is multiplied by the corresponding linear weight matrix and then added by position with the linear bias vector to obtain a sequence of the first engineering cost item word granularity semantic encoding feature vector after linear transformation; Each of the first engineering cost item word granularity semantic coding feature vectors after linear transformation in the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation is used as a row vector of a matrix, so as to arrange the sequence of the first engineering cost item word granularity semantic coding feature vectors after linear transformation in a matrix to obtain the initial key matrix.

5. The intelligent management method of engineering cost data based on Internet search service according to claim 4 is characterized in that: Calculating the feature distribution difference energy coefficients between the engineering project retrieval input semantic encoding feature vector and each row vector in the initial key matrix to obtain a set of first engineering cost item word granularity semantic feature distribution difference energy coefficients, including: Respectively calculating the positional difference, positional dot product and positional addition between the engineering project retrieval input semantic encoding feature vector and the row vector in the initial key matrix to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantics difference feature vector, the engineering project retrieval input semantics-first engineering cost item word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost item word granularity semantics sum feature vector; The engineering project retrieval input semantics-first engineering cost item word granularity semantics differential feature vector, the engineering project retrieval input semantics-first engineering cost item word granularity semantics dot product feature vector and the engineering project retrieval input semantics-first engineering cost item word granularity semantics sum feature vector are cascaded and processed through a one-dimensional convolution layer, and the obtained feature vector is subjected to maximum pooling processing to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantics feature association representation vector; Extracting the maximum value of the feature in the engineering project retrieval input semantics-first engineering cost item word granularity semantic feature association representation vector to obtain the engineering project retrieval input semantics-first engineering cost item word granularity semantic association maximum representation value; Calculating the mean and variance of the row vectors in the initial key matrix to obtain the first engineering cost item word granularity semantic feature mean and the first engineering cost item word granularity semantic feature variance; Adding the first engineering cost item word granularity semantic feature variance to the hyperparameter to obtain the first feature distribution difference energy coefficient; Calculate the square of the difference between the maximum representation value of the engineering project search input semantics-first engineering cost item word granularity semantic association and the mean value of the first engineering cost item word granularity semantic feature to obtain a semantic feature difference value; The first engineering cost item word granularity semantic feature variance is multiplied by a constant of two and then added to the semantic feature difference value and the hyperparameter to obtain a second feature distribution difference energy coefficient; The first feature distribution difference energy coefficient is divided by the second feature distribution difference energy coefficient to obtain the first engineering cost item word granularity semantic feature distribution difference energy coefficient.

6. The intelligent management method of engineering cost data based on Internet search service according to claim 5 is characterized in that: The engineering project retrieval input semantic encoding feature vector is processed using a value embedding matrix and a value embedding vector to obtain an engineering project retrieval input semantic value vector, including: The matrix multiplication between the engineering project retrieval input semantic coding feature vector and the value embedding matrix is ​​calculated, and then the matrix multiplication is added to the value embedding vector by position to obtain the engineering project retrieval input semantic value vector.

7. The intelligent management method of engineering cost data based on Internet search service according to claim 6 is characterized in that: The engineering project retrieval input semantic encoding feature vector is used as a query vector, each updated row vector in the update key matrix is ​​used as a key vector, and the engineering project retrieval input semantic value vector is used as a value vector, and the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of first engineering cost project word granularity semantic cross-domain query attention vectors, including: After calculating the product between the query vector and the transposed vector of the key vector, dividing the obtained query-key feature association value by the square root of the length of the key vector to obtain a query-key feature association coefficient; Inputting the query-key feature association coefficient into a softmax function to obtain a query-key feature association weight value; The query-key feature association weight value is used as a weight, and the transposed vector of the value vector is weighted to obtain a first engineering cost item word granularity semantic cross-domain query attention vector.

8. The intelligent management method of engineering cost data based on Internet search service according to claim 7 is characterized in that: Based on the search input-cost item cross-domain query matching feature vector, a management result is generated, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result, including: The retrieval input-cost item cross-domain query matching feature vector is input into a classifier-based intelligent management result generator to obtain the management result, and the management result is used to indicate whether to return the first engineering cost item data as the best matching result.

9. An intelligent management system for engineering cost data based on Internet search services, characterized in that: include: A search input acquisition module is used to acquire the engineering project search input input by the user; A retrieval input semantic coding module, used for performing semantic expression improvement processing on the engineering project retrieval input and then performing semantic coding to obtain a semantic coding feature vector of the engineering project retrieval input; A project data capture module, used for capturing the first engineering cost project data based on an Internet search service; A cost item semantic encoding module, used for performing word granularity-based cost item semantic encoding on the first engineering cost item data to obtain a sequence of first engineering cost item word granularity semantic encoding feature vectors; A local association strengthening module is used to use the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as a sequence of library feature vectors, and input the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association strengthening to obtain a retrieval input-cost item cross-domain query matching feature vector; A management result generation module, used to generate a management result based on the search input-cost item cross-domain query matching feature vector, wherein the management result is used to indicate whether to return the first engineering cost item data as the best matching result; The method uses the engineering project retrieval input semantic encoding feature vector as a query feature vector and the sequence of the first engineering cost item word granularity semantic encoding feature vector as a sequence of library feature vectors, and inputs the query feature vector and the sequence of the library feature vector into a cross-domain attention gate query encoder based on key matrix local association enhancement to obtain a retrieval input-cost item cross-domain query matching feature vector, including: Performing a linear transformation on each first engineering cost item word granularity semantic coding feature vector in the sequence of the first engineering cost item word granularity semantic coding feature vectors to obtain an initial key matrix, wherein each row vector in the initial key matrix is ​​a first engineering cost item word granularity semantic coding feature vector after each linear transformation; Calculating the feature distribution difference energy coefficients between the engineering project retrieval input semantic encoding feature vector and each row vector in the initial key matrix to obtain a set of first engineering cost item word granularity semantic feature distribution difference energy coefficients; Compare each first engineering cost item word granularity semantic feature distribution difference energy coefficient in the set of the first engineering cost item word granularity semantic feature distribution difference energy coefficient with a predetermined threshold value, and in response to the first engineering cost item word granularity semantic feature distribution difference energy coefficient being less than the predetermined threshold value, use the mean vector between the row vector in the initial key matrix corresponding to the first engineering cost item word granularity semantic feature distribution difference energy coefficient and the engineering project retrieval input semantic encoding feature vector as the updated row vector to obtain an updated key matrix; Using a value embedding matrix and a value embedding vector to process the engineering project retrieval input semantic encoding feature vector to obtain an engineering project retrieval input semantic value vector; Taking the engineering project retrieval input semantic encoding feature vector as a query vector, each updated row vector in the update key matrix as a key vector and the engineering project retrieval input semantic value vector as a value vector, the query vector, the key vector and the value vector are input into a cross-domain attention gate query module based on a Transformer structure to obtain a sequence of first engineering cost item word granularity semantic cross-domain query attention vectors; The position-wise mean vector of the sequence of the first engineering cost item word granularity semantic cross-domain query attention vector is calculated to obtain the retrieval input-cost item cross-domain query matching feature vector.

Images