CN117669582A - Engineering consultation processing method and device based on deep learning and electronic equipment - Google Patents

Abstract

This application provides an engineering consulting processing method, device and electronic equipment based on deep learning, which relates to the technical field of data processing. In this method, a consultation request sent by the user device is received, and the consultation request is used to represent a consultation request for the target project; a semantic report is generated according to the consultation request; the semantic report is encrypted to obtain an encrypted semantic report; and the encrypted semantic report is input to In the preset processing model, the response result is obtained; the response result is sent to the user device. Implementing the technical solution provided by this application has the technical effect of improving the efficiency of processing engineering consultation.

Description

An engineering consulting processing method, device and electronic equipment based on deep learning

Technical field

This application relates to the technical field of data processing, and specifically relates to an engineering consulting processing method, device and electronic equipment based on deep learning.

Background technique

Engineering consulting is an intellectual service that follows the principles of independence, science, and impartiality. It aims to provide consulting activities for government departments, project owners, and other types of customers in the decision-making and management of engineering construction projects. These consulting activities cover the early project establishment stage, survey and design stage, construction stage, and evaluation after commissioning or delivery.

Usually, when users need engineering consulting services, they will communicate offline with the staff who provide engineering consulting services. However, due to the large volume of engineering consultation, staff need to repeatedly query relevant information, which undoubtedly prolongs the processing cycle of engineering consultation and results in low efficiency in processing engineering consultation.

Therefore, there is an urgent need for an engineering consulting processing method, device and electronic equipment based on deep learning.

Contents of the invention

This application provides an engineering consultation processing method, device and electronic equipment based on deep learning, which has the technical effect of improving the processing efficiency of engineering consultation.

In the first aspect of this application, a deep learning-based engineering consultation processing method is provided. The method includes: receiving a consultation request sent by a user device, where the consultation request is used to represent a consultation request for a target project; according to the Consult the request and generate a semantic report; encrypt the semantic report to obtain an encrypted semantic report; input the encrypted semantic report into the preset processing model to obtain a response result; send the response result to the user equipment.

By adopting the above technical solution, the processing efficiency is greatly improved by automatically receiving consultation requests sent by user devices and automatically generating semantic reports. By encrypting the generated semantic reports, the security of user data can be protected and data leakage can be prevented. By inputting the encrypted semantic report into the preset processing model, the response results can be obtained quickly, further improving the processing efficiency. Finally, by sending the response results to the user's device, the user can obtain feedback information in time to better understand the consultation results. As a result, the process can achieve automated processing, data encryption protection, preset processing models and timely feedback of response results, providing users with more efficient, secure and convenient consulting services. Compared with related technologies, manual intervention is no longer required and can Automatically complete the processing of engineering consultation, which has the technical effect of improving the efficiency of processing engineering consultation.

Optionally, generating a semantic report based on the consultation request specifically includes: obtaining text data in the consultation request; using a text fingerprint algorithm to match the text data with a preset keyword library; if the key is confirmed If the word matching is successful, intent recognition is performed based on the keywords to obtain the intent consulting field; the semantic report is generated based on the intent consulting field.

By adopting the above technical solution, by obtaining the text data in the consultation request and processing the data, subsequent matching and identification can be made more accurate. By using the text fingerprint algorithm to match the text data with the preset keyword library, keywords related to the text data can be quickly found, providing a reference for subsequent intent recognition. If it is confirmed that the keyword matching is successful, intent recognition will be performed based on the keywords to obtain the intent consultation area, which will facilitate the understanding of the user's intention based on the text data input by the user, thereby providing more accurate consulting services. By generating semantic reports based on the intended consultation field, users' consultation requests can be converted into standardized semantic reports to facilitate subsequent processing and analysis. This makes it easy to effectively process text data, perform keyword matching and intent recognition, and generate semantic reports based on the intent recognition results, thereby providing more accurate and efficient consulting services.

Optionally, the encrypted semantic report includes a semantic tag value, and encrypting the semantic report to obtain the encrypted semantic report specifically includes: using a secret sharing algorithm to encrypt the semantic report to generate an encrypted semantic report; The encrypted semantic report is split into multiple semantic sub-reports, and corresponding ciphertext and random numbers are generated for each semantic sub-report; based on the cipher text and random numbers of each semantic sub-report, the tag value of each semantic sub-report is calculated; multiple The mean value of the tag values of the semantic sub-report is used as the semantic tag value of the encrypted semantic report.

By adopting the above technical solution and encrypting the generated semantic reports, the security of user data can be protected and data leakage can be prevented. Using a secret sharing algorithm to encrypt semantic reports can enhance data security and make the data more difficult to crack during transmission. By splitting the encrypted semantic report into multiple semantic sub-reports, the privacy and integrity of each semantic sub-report can be better protected, making each semantic sub-report difficult to crack individually. By generating corresponding ciphertext and random numbers for each semantic sub-report, the difficulty of cracking can be increased, making it impossible for attackers to deduce the original semantic information through ciphertext and random numbers. By calculating the tag value of each semantic subreport based on the ciphertext and random number of each semantic subreport, the content and meaning of each semantic subreport can be better protected, making it impossible for attackers to deduce the original semantic information through the tag value. . By using the average tag value of multiple semantic sub-reports as the semantic tag value of the encrypted semantic report, the privacy and integrity of the entire encrypted semantic report can be better protected, making it impossible for attackers to deduce the original semantic information through the average tag value. .

Optionally, inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: extracting a first feature word from the encrypted semantic report, where the first feature word is a preset dimension. Corresponding keywords, the preset dimensions include project category, project content, project cycle, project budget and construction environment; compare the first feature word with any preset encrypted semantic report in the preset processing model Hamming similarity is calculated to obtain the target encrypted semantic report; and the response result is obtained according to the target encrypted semantic report.

By adopting the above technical solution, the first feature words are extracted from the encrypted semantic report, and these feature words are keywords corresponding to the preset dimensions. By extracting these feature words, users' consultation requests can be better understood and more accurately matched and responded to. The target encrypted semantic report is obtained by performing Hamming similarity calculation on the extracted first feature word and any preset encrypted semantic report in the preset processing model. By performing Hamming similarity calculation, the degree of similarity between two texts can be quantified to find the preset encrypted semantic report that best matches the user's consultation request. By encrypting semantic reports according to the target, the response results are obtained. By matching the most similar preset encrypted semantic reports, the response results most relevant to the user's consultation request can be quickly obtained, thus improving the efficiency and accuracy of consultation.

Optionally, performing Hamming similarity calculation on the first feature word and any one of the preset encrypted semantic reports in the preset processing model to obtain the target encrypted semantic report specifically includes: extracting any one of the preset encrypted semantic reports. Assume the second feature word in the encrypted semantic report; calculate the Hamming distance between the first feature word and the second feature word; compare the size relationship between the Hamming distance and the preset Hamming distance; If the Hamming distance is less than or equal to the preset Hamming distance, it is confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report.

By adopting the above technical solution, by extracting the second feature words in any preset encrypted semantic report in the preset processing model, these feature words are keywords corresponding to the preset dimensions. By extracting these second feature words, they can be compared and matched with the first feature words in the user's consultation request. This is a method of measuring the similarity between two texts by calculating the Hamming distance between the first feature word and the second feature word. By calculating the Hamming distance, the difference between two texts can be quantified to determine the degree of similarity between them. By comparing the relationship between the Hamming distance and the preset Hamming distance, the preset encryption semantic report that best matches the user's consultation request can be filtered out. If the Hamming distance is less than or equal to the preset Hamming distance, it is confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report. This means that the preset encrypted semantic report that is most similar to the user's consultation request will be selected as the response result, thus improving the accuracy and efficiency of consultation.

Optionally, inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: obtaining the topic semantics in any historical response result, and there are multiple preset processing models stored in advance. historical response results; calculate the response label value corresponding to each topic semantics; determine the similarity between the response label value and the semantic label value; if it is confirmed that the similarity is greater than or equal to the preset similarity threshold, then determine The historical response result corresponding to the response tag value is the response result.

By adopting the above technical solution, the topic semantics in any historical response results are obtained, and these historical response results are pre-stored in the preset processing model. By leveraging historical response results, users' consultation requests can be better understood and more accurate and comprehensive responses can be provided. By calculating the response tag value corresponding to each topic semantics. These response tag values are calculated based on the topic semantics of historical response results and can reflect the importance of different topics in historical response results. By judging the similarity between the response tag value and the semantic tag value. By comparing the similarity of two tag values, the degree of correlation between them can be quantified, thereby determining the degree of matching between the user's consultation request and historical response results. If it is confirmed that the similarity is greater than or equal to the preset similarity threshold, the historical response result corresponding to the response tag value is determined to be the response result, so as to improve the accuracy and efficiency of consultation.

Optionally, the step of inputting the encrypted semantic report into a preset processing model and training the preset processing model before obtaining the response result; the training of the preset processing model specifically includes: obtaining training information, The training information includes encrypted semantic reports and response results; the training information is input into the adaptive feature fusion network for training to obtain the first training result; the first training result and the training information are superimposed and standardized After processing, the second training result is obtained; the second training result is input into the adaptive feature fusion network for processing, and a third training result is obtained; the third training result is compared with the second training result. Superposition and standardization are performed until the training information similarity matrix is output, and the training information similarity matrix satisfies the preset logistic regression condition.

By adopting the above technical solution and training the preset processing model, the ability to match and respond to user consultation requests can be improved. By obtaining training information, including encrypted semantic reports and response results. This training information is used to train the preset processing model so that it can better understand the user's consultation request and give accurate response results. By inputting the training information into the adaptive feature fusion network for training, the first training result is obtained. The adaptive feature fusion network can automatically learn and extract features based on the input training information, thereby improving the efficiency and accuracy of training. The second training result is obtained by superimposing and standardizing the first training result and the training information. This processing can make the training results more stable and reliable, and improve the generalization ability of the model. By inputting the second training result into the adaptive feature fusion network again for processing, the third training result is obtained, and the third training result and the second training result are superimposed and standardized. This multiple superposition and standardization processing can further improve the accuracy and stability of the model. By outputting the training information similarity matrix, the matrix can be used to measure the similarity between different encrypted semantic reports to meet the preset logistic regression conditions. This similarity matrix can be used in subsequent matching and response processes to improve processing efficiency and accuracy.

In the second aspect of this application, an engineering consultation processing device based on deep learning is provided. The processing device includes a receiving module and a processing module, wherein the receiving module is used to receive a consulting request sent by the user equipment, and the The consultation request is used to represent a consultation request for the target project; the processing module is used to generate a semantic report according to the consultation request; the processing module is also used to encrypt the semantic report to obtain an encrypted semantic report; The processing module is also used to input the encrypted semantic report into a preset processing model to obtain a response result; the processing module is also used to send the response result to the user equipment.

In a third aspect of the present application, an electronic device is provided. The electronic device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions, and the user interface and the network interface are both used to store instructions. To communicate with other devices, the processor is used to execute instructions stored in the memory, so that the electronic device performs the method as described above.

In a fourth aspect of the present application, a computer-readable storage medium is provided. The computer-readable storage medium stores instructions. When the instructions are executed, the method as described above is performed.

To sum up, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

1. By automatically receiving consultation requests sent by user devices and automatically generating semantic reports, processing efficiency is greatly improved. By encrypting the generated semantic reports, the security of user data can be protected and data leakage can be prevented. By inputting the encrypted semantic report into the preset processing model, the response results can be obtained quickly, further improving the processing efficiency. Finally, by sending the response results to the user's device, the user can obtain feedback information in time to better understand the consultation results. As a result, the process can achieve automated processing, data encryption protection, preset processing models and timely feedback of response results, providing users with more efficient, secure and convenient consulting services. Compared with related technologies, manual intervention is no longer required and can Automatically complete the processing of engineering consultation, which has the technical effect of improving the efficiency of processing engineering consultation;

2. By calculating the Hamming distance, the difference between two texts can be quantified to determine the degree of similarity between them. By comparing the relationship between the Hamming distance and the preset Hamming distance, the preset encryption semantic report that best matches the user's consultation request can be filtered out. If the Hamming distance is less than or equal to the preset Hamming distance, it is confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report. This means that the preset encrypted semantic report most similar to the user's consultation request will be selected as the response result, thus improving the accuracy and efficiency of consultation;

3. By utilizing historical response results, we can better understand users’ consultation requests and provide them with more accurate and comprehensive responses. By calculating the response tag value corresponding to each topic semantics. These response tag values are calculated based on the topic semantics of historical response results and can reflect the importance of different topics in historical response results. By judging the similarity between the response tag value and the semantic tag value. By comparing the similarity of two tag values, the degree of correlation between them can be quantified, thereby determining the degree of matching between the user's consultation request and historical response results. If it is confirmed that the similarity is greater than or equal to the preset similarity threshold, the historical response result corresponding to the response tag value is determined to be the response result, so as to improve the accuracy and efficiency of consultation.

Description of drawings

Figure 1 is a schematic flowchart of an engineering consulting processing method based on deep learning provided by an embodiment of the present application.

Figure 2 is a schematic module diagram of a deep learning-based engineering consultation processing device provided by an embodiment of the present application.

FIG. 3 is a schematic structural diagram of an electronic device provided by an embodiment of the present application.

Explanation of reference signs: 21. Receiving module; 22. Processing module; 31. Processor; 32. Communication bus; 33. User interface; 34. Network interface; 35. Memory.

Detailed ways

In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described The embodiments are only some of the embodiments of this application, not all of them.

In the description of the embodiments of this application, words such as "for example" or "for example" are used to represent examples, illustrations or illustrations. Any embodiment or design described as "such as" or "for example" in the embodiments of the present application shall not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "for example" or "for example" is intended to present the concept in a concrete manner.

In the description of the embodiments of this application, the term “plurality” means two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Therefore, features defined as "first" and "second" may explicitly or implicitly include one or more of these features. The terms “including,” “includes,” “having,” and variations thereof all mean “including but not limited to,” unless otherwise specifically emphasized.

Engineering planning is a certain application of science and mathematics. Through this application, the properties of matter and energy in nature can be implemented through various structures, machines, products, systems and processes in the shortest time and with the smallest amount of manpower. To produce something that is efficient, reliable and useful to humanity. So the concept of engineering came into being, and it gradually developed into an independent discipline and skill. In real life, it is closely related to engineering everywhere, but the implementation of engineering requires relatively professional knowledge reserves. However, if the professional knowledge of the construction personnel cannot meet the requirements during the actual construction process, engineering consultation is required.

Engineering consulting is a very important intellectual service that follows the principles of independence, science, and impartiality, and aims to provide consulting activities for government departments, project owners, and other various customers in the decision-making and management of engineering construction projects. These consulting activities cover the early project establishment stage, survey and design stage, construction stage, and evaluation after commissioning or delivery. The core of engineering consulting is to use professional knowledge and experience to provide customers with high-quality, comprehensive consulting services, help them make more reasonable and scientific decisions, and improve the quality and efficiency of projects.

Usually, when users need engineering consulting services, they will communicate offline with the staff who provide engineering consulting services. However, due to the large volume of engineering consultation, staff need to repeatedly query relevant information, which undoubtedly prolongs the processing cycle of engineering consultation and results in low efficiency in processing engineering consultation. In addition, due to the lack of effective information management methods, the quality and efficiency of engineering consulting work have been affected to a certain extent. Therefore, how to improve the processing efficiency and quality of engineering consulting is an important issue facing the current engineering consulting industry.

In order to solve the above technical problems, this application provides an engineering consulting processing method based on deep learning. Refer to Figure 1 , which is a schematic flow chart of an engineering consulting processing method based on deep learning provided by an embodiment of this application. The processing method is applied to the server and includes steps S110 to S150. The above steps are as follows:

S110. Receive a consultation request sent by the user device. The consultation request is used to represent a consultation request for the target project.

Specifically, when the user has an engineering consulting problem, he or she can input the problem by using an engineering consulting application on the user's device or accessing an engineering consulting website, where the consulting request includes the consulting problem. In this embodiment of the present application, the user equipment is a smartphone, and the server is used to provide background services for the user equipment. The server can be one server, a server cluster composed of multiple servers, or a cloud computing service center. The server can communicate with user devices over wired or wireless networks. Types of user devices include but are not limited to: Android system devices, mobile operating system (iOS) devices developed by Apple, personal computers (PC), WorldWide Web (web) devices, virtual reality (Virtual Reality, VR) equipment, augmented reality (AugmentedReality, AR) equipment and other equipment.

S120. Generate a semantic report according to the consultation request.

Specifically, after the server receives the consultation request, it will use natural language processing technology to perform semantic analysis on the consultation request. This process can include steps such as lexical analysis, syntactic analysis, and semantic understanding to extract key information in user consultation requests, such as project name, consultation questions, time, etc.

In one possible implementation, the text data in the consultation request is obtained; a text fingerprint algorithm is used to match the text data with a preset keyword library; if the keyword matching is confirmed to be successful, intent recognition is performed based on the keywords to obtain the intent Consulting field; generate semantic reports based on the intended consulting field.

Specifically, after the server receives the consultation request, it will extract text data from it. These text data contain users' consultation needs and questions about the target project. The text fingerprint algorithm is a technology used for text matching. It extracts feature information from the text and generates a unique fingerprint to determine whether two texts are similar. The server matches the extracted text data with the keywords in the preset keyword library to determine whether it contains the same keywords. If the match is successful, it means that the text data contains keywords in the preset keyword library, and then intent recognition can be performed based on these keywords. Intent recognition refers to extracting the user's intention from the text, that is, what information the user mainly wants to know. Through intention recognition, the intention consultation area can be obtained, that is, the target engineering area that the user mainly wants to consult. Finally, based on the intent query field, the server can generate corresponding semantic reports. This report is usually the result of a semantic understanding of the user's consultation request. It can help the server better understand the user's consultation needs, thereby providing a more accurate and effective response.

For example, suppose the user sends a consultation request: "Excuse me, what is the progress of the construction project of the commercial building on Chaoyang Road?" After the server receives the consultation request, it will extract text data from it: "Chaoyang Road Commercial Building The progress of the construction project." Then, the server will match it with the keywords in the preset keyword database and find that keywords such as "Chaoyang Road", "Commercial Building", and "Construction Project" are all matched successfully. Next, the server will perform intent recognition based on these keywords and determine that the user's intent is to learn about the progress of the Chaoyang Road commercial building construction project. Finally, based on this intention, the server can generate corresponding semantic reports, including project name, location, consultation questions, time and other information. This semantic report can help the server better understand the user's consultation needs and provide support for subsequent responses.

S130. Encrypt the semantic report to obtain the encrypted semantic report.

Specifically, the report generated by the server based on the user's consultation request contains the semantic understanding results of the user's consultation questions. This report is usually a structured or semi-structured data format, such as XML, JSON, etc. The information is then converted into an unreadable format through specific algorithms and keys, thereby protecting the confidentiality and integrity of the information. The encrypted semantic report is an encrypted semantic report. It is a protected data format that prevents unauthorized access and tampering.

In a possible implementation, the encrypted semantic report includes semantic tag values, and the semantic report is encrypted to obtain the encrypted semantic report, which specifically includes: using a secret sharing algorithm to encrypt the semantic report to generate the encrypted semantic report; and converting the encrypted semantic report into Split into multiple semantic sub-reports, and generate corresponding ciphertext and random numbers for each semantic sub-report; calculate the tag value of each semantic sub-report based on the cipher text and random numbers of each semantic sub-report; combine multiple semantic sub-reports The mean value of the label value is used as the semantic label value of the encrypted semantic report.

Specifically, in the encrypted semantic report, in addition to the encrypted content of the report itself, the tag value of each semantic sub-report is also included. The tag value is a summary or description of the content of each semantic subreport, which can facilitate subsequent query and processing. The secret sharing algorithm is an algorithm that divides the secret key into several parts and stores them in multiple participating terminals. Only the terminal holding the corresponding shares can reassemble the original secret through these shares. Here, a secret sharing algorithm is used to encrypt the semantic report, which can increase the difficulty of cracking and improve the security of the data. To better manage and process encrypted semantic reports, you can split the report into multiple smaller subreports. Each semantic subreport contains a portion of the original semantic data and the corresponding ciphertext. After splitting the encrypted semantic report, the corresponding ciphertext and random numbers need to be generated for each semantic sub-report. These random numbers can be used as identifiers for each subreport to facilitate subsequent processing and querying. Based on the ciphertext and random number of each semantic subreport, the tag value of each subreport can be calculated. These tag values are a summary or description of the content of the subreport and can be used as an identifier for each subreport. Finally, the tag values of multiple semantic sub-reports are averaged as the semantic tag value of the entire encrypted semantic report. This tag value can be used as an overall identifier to facilitate subsequent processing and query of encrypted semantic reports.

For example, suppose there is a semantic report, including "Project name: Chaoyang Road Commercial Building; Location: Chaoyang District, Beijing; Consulting question: How is the project progress?" and it is encrypted. The process of generating an encrypted semantic report is as follows: First , using a secret sharing algorithm to encrypt semantic reports and generate encrypted semantic reports. This process encrypts the original data into a ciphertext string. Second, split the encrypted semantic report into multiple semantic sub-reports. For example, it can be split into three subreports: "Project name": "Chaoyang Road Commercial Building", "Location": "Chaoyang District, Beijing", "Consultation question": "How is the project progress?" Next, the corresponding ciphertext and random numbers are generated for each semantic subreport. For example, you can generate a random number for each subreport and calculate the corresponding ciphertext. Then, based on the ciphertext and random number of each semantic subreport, the label value of each subreport is calculated. For example, you can use a hash function such as MD5 to hash the content of each subreport to obtain the corresponding tag value. The tag values of multiple semantic sub-reports are averaged as the semantic tag value of the entire encrypted semantic report. Furthermore, if the tag values of the three sub-reports are 0x123456, 0xabcdef, and 0x789abc respectively, the average of the three tag values corresponding to the three sub-reports can be used as the semantic tag value of the entire encrypted semantic report, that is, (0x123456+0xabcdef+0x789abc )/3. The final encrypted semantic report includes information such as ciphertext, random numbers, and tag values for each subreport, which can be stored and used. When you need to query and process encrypted semantic reports, you can quickly locate and retrieve relevant subreport information through semantic tag values.

S140. Input the encrypted semantic report into the preset processing model to obtain the response result.

S150. Send the response result to the user equipment.

Specifically, by automatically receiving consultation requests sent by user devices and automatically generating semantic reports, processing efficiency is greatly improved. By encrypting the generated semantic reports, the security of user data can be protected and data leakage can be prevented. By inputting the encrypted semantic report into the preset processing model, the response results can be obtained quickly, further improving the processing efficiency. Finally, by sending the response results to the user's device, the user can obtain feedback information in time to better understand the consultation results. As a result, the process can achieve automated processing, data encryption protection, preset processing models and timely feedback of response results, providing users with more efficient, secure and convenient consulting services. Compared with related technologies, manual intervention is no longer required and can Automatically complete the processing of engineering consultation, which has the technical effect of improving the efficiency of processing engineering consultation.

In one possible implementation, inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: extracting a first feature word from the encrypted semantic report, where the first feature word is a key corresponding to the preset dimension. words, the preset dimensions include project category, project content, project cycle, project budget and construction environment; perform Hamming similarity calculation on the first feature word and any preset encrypted semantic report in the preset processing model to obtain the target Encrypted semantic report; according to the target encrypted semantic report, the response result is obtained.

Specifically, the server extracts the first feature words from the encrypted semantic report, and these feature words are keywords corresponding to the preset dimensions. The preset dimensions include project category, project content, project cycle, project budget and construction environment. These feature words can be regarded as key descriptions of the report content, which are helpful for subsequent processing and understanding. Next, the server compares the extracted first feature word with the preset encrypted semantic report in the preset processing model, and calculates the Hamming similarity between them. Hamming similarity is a method of measuring the similarity between two strings. The larger the value, the more similar the two strings are. By calculating the similarity, the preset encryption semantic report that is most similar to the input encryption semantic report can be found. Through the calculation in the previous step, the most similar preset encryption semantics report can be obtained. This report is the target encryption semantics report. This target report is an instance of the preset processing model and can be used for subsequent processing and analysis. The last step is to obtain the response results based on the target encryption semantic report. This result may be a direct answer to the user's consultation request, or it may be a further explanation or suggestion for the user's request. This result depends on the settings of the default processing model and the content of the target encryption semantic report.

In a possible implementation, Hamming similarity calculation is performed between the first feature word and any preset encrypted semantic report in the preset processing model to obtain the target encrypted semantic report, which specifically includes: extracting any preset Encrypt the second feature word in the semantic report; calculate the Hamming distance between the first feature word and the second feature word; compare the Hamming distance with the preset Hamming distance; if the Hamming distance is less than or equal to If the Hamming distance is preset, it is confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report.

Specifically, the server extracts any preset encrypted semantic report from the preset processing model and extracts the second feature word therefrom. These second feature words are keywords corresponding to the first feature words and are used to calculate Hamming similarity. Next, the server calculates the Hamming distance between the first feature word and the second feature word. Hamming distance is the number of unmatched characters between two strings and is used to measure how similar two strings are. By calculating the Hamming distance, the similarity between two feature words can be quantitatively evaluated. Finally, the server compares the calculated Hamming distance with the preset Hamming distance. The preset Hamming distance is a preset threshold used to determine whether two feature words are similar enough. If the calculated Hamming distance is less than or equal to the preset Hamming distance, it means that the two feature words are similar enough, and the corresponding preset encrypted semantic report can be confirmed to be the target encrypted semantic report. If the calculated Hamming distance is less than or equal to the preset Hamming distance, then it can be confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report. This target report will be used as input for subsequent processing and generate the final response results.

In one possible implementation, inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: obtaining the topic semantics in any historical response result. The preset processing model has multiple presets stored in it. Historical response results; calculate the response label value corresponding to each topic semantics; determine the similarity between the response label value and the semantic label value; if it is confirmed that the similarity is greater than or equal to the preset similarity threshold, determine the historical response corresponding to the response label value The result is the response result.

Specifically, the server first obtains the topic semantics in any historical response result. These topic semantics are the main content and meaning of historical response results, and are used for subsequent similarity calculation and judgment. Among them, multiple historical response results are pre-stored in the preset processing model, and these results are used to provide reference and assist in generating new response results. By calculating the corresponding response tag value for each topic semantics of each historical response result. These label values can be the results of classification or labeling based on the content and meaning of historical response results, and are used for subsequent similarity comparison and judgment. The server compares the calculated response tag value with the semantic tag value to determine the similarity between them. Similarity can be calculated by a specific algorithm or model and is used to measure the similarity between two label values. If the calculated similarity is greater than or equal to the preset similarity threshold, it can be confirmed that the historical response result corresponding to the response tag value is the most similar to the currently input encrypted semantic report. This historical response result will be used as the final response result. Among them, the method of calculating similarity can use cosine similarity or Euclidean distance, etc., which will not be described again here.

In a possible implementation, the encrypted semantic report is input into the preset processing model, and before the response result is obtained, the preset processing model is trained; training the preset processing model specifically includes: obtaining training information, and the training information includes encrypted semantics Report and response results; input the training information into the adaptive feature fusion network for training to obtain the first training result; superimpose and standardize the first training result and the training information to obtain the second training result; combine the second training result with The results are input into the adaptive feature fusion network for processing to obtain the third training result; the third training result and the second training result are superimposed and standardized until the training information similarity matrix is output, and the training information similarity matrix meets the preset Logistic regression conditions.

Specifically, the server first obtains training information, including encrypted semantic reports and response results. This information is used to train the adaptive feature fusion network to obtain a model that can process encrypted semantic reports and generate response results. The training information is then input into the adaptive feature fusion network, and the first training result is obtained through training. The adaptive feature fusion network is a deep learning model that can automatically learn and extract features, and generate corresponding results based on the input information. Secondly, the server superimposes and standardizes the first training result and the original training information to obtain the second training result. This step can be regarded as the first evaluation and adjustment of the model, helping the model to better learn and understand the input information. Next, the server inputs the second training result into the adaptive feature fusion network again for processing, and obtains the third training result. This processing can be regarded as the second evaluation and adjustment of the model to further optimize the performance of the model. Finally, the server superimposes and standardizes the third training result and the second training result. This step can be regarded as the third evaluation and adjustment of the model, making the model more adaptable and closer to actual needs. After multiple iterations and adjustments, a training information similarity matrix is finally output, which satisfies the preset logistic regression conditions. Logistic regression is a common classification algorithm that can be used for both prediction and classification tasks. The preset logistic regression conditions here may refer to classification or prediction criteria set based on specific tasks and needs.

This application also provides an engineering consultation processing device based on deep learning. Refer to Figure 2 , which is a schematic module diagram of an engineering consultation processing device based on deep learning provided in an embodiment of this application. The processing device is a server. The server includes a receiving module 21 and a processing module 22. The receiving module 21 is used to receive a consulting request sent by the user equipment, and the consulting request is used to represent a consulting request for the target project; the processing module 22 is used to According to the consultation request, a semantic report is generated; the processing module 22 is also used to encrypt the semantic report to obtain an encrypted semantic report; the processing module 22 is also used to input the encrypted semantic report into the preset processing model to obtain the response result; processing Module 22 is also used to send the response result to the user equipment.

In one possible implementation, generating a semantic report based on the consultation request specifically includes: the receiving module 21 obtains the text data in the consultation request; the processing module 22 uses the text fingerprint algorithm to match the text data with the preset keyword database; If the processing module 22 confirms that the keyword matching is successful, it will perform intent recognition based on the keywords and obtain the intent consulting field; the processing module 22 will generate a semantic report based on the intent consulting field.

In a possible implementation, the encrypted semantic report includes semantic tag values, and the processing module 22 encrypts the semantic report to obtain the encrypted semantic report, which specifically includes: the processing module 22 uses a secret sharing algorithm to encrypt the semantic report and generates encrypted semantics. Report; the processing module 22 splits the encrypted semantic report into multiple semantic sub-reports, and generates corresponding ciphertext and random numbers for each semantic sub-report; the processing module 22 calculates each semantic according to the ciphertext and random numbers of each semantic sub-report. The tag value of the subreport; the processing module 22 uses the average value of the tag values of multiple semantic subreports as the semantic tag value of the encrypted semantic report.

In a possible implementation, the processing module 22 inputs the encrypted semantic report into the preset processing model to obtain the response result, which specifically includes: the processing module 22 extracts the first feature word from the encrypted semantic report, and the first feature word is Keywords corresponding to the preset dimensions. The preset dimensions include project category, project content, project cycle, project budget and construction environment; the processing module 22 compares the first feature word with any preset encrypted semantic report in the preset processing model. Hamming similarity is calculated to obtain the target encrypted semantic report; the processing module 22 obtains the response result according to the target encrypted semantic report.

In a possible implementation, the processing module 22 performs Hamming similarity calculation on the first feature word and any preset encrypted semantic report in the preset processing model to obtain the target encrypted semantic report, which specifically includes: a processing module 22. Extract the second feature word in any preset encrypted semantic report; the processing module 22 calculates the Hamming distance between the first feature word and the second feature word; the processing module 22 compares the Hamming distance with the preset Hamming distance. The size relationship between; if the Hamming distance is less than or equal to the preset Hamming distance, the processing module 22 confirms that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report.

In a possible implementation, the processing module 22 inputs the encrypted semantic report into the preset processing model to obtain the response result. Specifically, it further includes: the receiving module 21 obtains the topic semantics in any historical response result, and the preset processing model Multiple historical response results are pre-stored in it; the processing module 22 calculates the response tag value corresponding to each topic semantics; the processing module 22 determines the similarity between the response tag value and the semantic tag value; if the processing module 22 confirms that the similarity is greater than or equal to If the similarity threshold is preset, the historical response result corresponding to the response tag value is determined to be the response result.

In a possible implementation, the processing module 22 inputs the encrypted semantic report into the preset processing model, and before obtaining the response result, trains the preset processing model; training the preset processing model specifically includes: the receiving module 21 obtains the training information , the training information includes encrypted semantic reports and response results; the processing module 22 inputs the training information into the adaptive feature fusion network for training, and obtains the first training result; the processing module 22 superimposes and standardizes the first training result and the training information. Then, the second training result is obtained; the processing module 22 inputs the second training result into the adaptive feature fusion network for processing, and obtains the third training result; the processing module 22 superimposes and standardizes the third training result and the second training result. Process until the training information similarity matrix is output, and the training information similarity matrix meets the preset logistic regression conditions.

It should be noted that when the devices provided in the above embodiments implement their functions, only the division of the above functional modules is used as an example. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the equipment The internal structure is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be described again here.

The present application also provides an electronic device. Refer to FIG. 3 , which is a schematic structural diagram of an electronic device provided by an embodiment of the present application. The electronic device may include: at least one processor 31, at least one network interface 34, user interface 33, memory 35, and at least one communication bus 32.

Among them, the communication bus 32 is used to realize connection communication between these components.

Among them, the user interface 33 may include a display screen (Display) and a camera (Camera), and the optional user interface 33 may also include a standard wired interface and a wireless interface.

Among them, the network interface 34 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

Among them, the processor 31 may include one or more processing cores. The processor 31 uses various interfaces and lines to connect various parts of the entire server, and executes the server by running or executing instructions, programs, code sets or instruction sets stored in the memory 35, and calling data stored in the memory 35. Various functions and processing data. Optionally, the processor 31 can use at least one of digital signal processing (Digital Signal Processing, DSP), field-programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). implemented in hardware form. The processor 31 may integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processor (Graphics Processing Unit, GPU), a modem, etc. Among them, the CPU mainly handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content that needs to be displayed on the display; and the modem is used to handle wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 31 and may be implemented by a separate chip.

The memory 35 may include random access memory (RAM) or read-only memory (Read-Only Memory). Optionally, the memory 35 includes a non-transitory computer-readable storage medium. Memory 35 may be used to store instructions, programs, codes, sets of codes, or sets of instructions. The memory 35 may include a program storage area and a data storage area, where the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), Instructions, etc., used to implement each of the above method embodiments; the storage data area can store data, etc. involved in each of the above method embodiments. The memory 35 may optionally be at least one storage device located remotely from the aforementioned processor 31 . As shown in FIG. 3 , the memory 35 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program of an engineering consulting processing method based on deep learning.

In the electronic device shown in Figure 3, the user interface 33 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 31 can be used to call the memory 35 to store an engineering consulting process based on deep learning. The application program of the method, when executed by one or more processors, causes the electronic device to execute one or more of the methods in the above embodiments.

It should be noted that for the sake of simple description, the foregoing method embodiments are expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence. Because in accordance with this application, certain steps may be performed in other orders or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily necessary for this application.

This application also provides a computer-readable storage medium, which stores instructions. When executed by one or more processors, the electronic device is caused to perform the method described in one or more of the above embodiments.

In the above embodiments, each embodiment is described with its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

In the several embodiments provided in this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into Another system, or some features can be ignored, or not implemented. Another point is that the coupling or direct coupling or communication connection between each other shown or discussed may be through some service interfaces, and the indirect coupling or communication connection of the devices or units may be in electrical or other forms.

A unit described as a separate component may or may not be physically separate. A component shown as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or software functional units.

Integrated units may be stored in a computer-readable memory when implemented as software functional units and sold or used as independent products. Based on this understanding, the technical solution of the present application is essentially or contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory, It includes several instructions to cause a computer device (which can be a personal computer, a server or a network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned memory includes: U disk, mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

The above are only exemplary embodiments of the present disclosure and do not limit the scope of the present disclosure. That is to say, all equivalent changes and modifications made based on the teachings of this disclosure are still within the scope of this disclosure. Other embodiments of the present disclosure will readily occur to those skilled in the art, upon consideration of the specification and disclosure of practical truths. This application is intended to cover any variations, uses, or adaptations of the disclosure that follow the general principles of the disclosure and include common knowledge or customary technical means in the technical field that are not described in the disclosure. . It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being defined by the following claims.

Claims (10)

1. An engineering consulting processing method based on deep learning, characterized in that the method includes: Receive a consultation request sent by the user equipment, where the consultation request is used to represent a consultation request for the target project; Generate semantic reports based on the consultation request; Encrypt the semantic report to obtain an encrypted semantic report; Input the encrypted semantic report into the preset processing model to obtain the response result; Send the response result to the user equipment. 2. The engineering consulting processing method based on deep learning according to claim 1, characterized in that generating a semantic report based on the consulting request specifically includes: Obtain the text data in the consultation request; Using a text fingerprint algorithm to match the text data with a preset keyword library; If it is confirmed that the keyword matching is successful, intent identification is performed based on the keywords and the intent consultation field is obtained; The semantic report is generated according to the intent consultation field. 3. The engineering consulting processing method based on deep learning according to claim 1, characterized in that the encrypted semantic report includes a semantic tag value, and the encrypted semantic report is encrypted to obtain an encrypted semantic report, which specifically includes: Use a secret sharing algorithm to encrypt semantic reports and generate encrypted semantic reports; Split the encrypted semantic report into multiple semantic sub-reports, and generate corresponding ciphertext and random numbers for each semantic sub-report; Calculate the tag value of each semantic subreport based on the ciphertext and random number of each semantic subreport; The mean value of the tag values of multiple semantic sub-reports is used as the semantic tag value of the encrypted semantic report. 4. The engineering consulting processing method based on deep learning according to claim 1, characterized in that said inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: Extract first feature words from the encrypted semantic report, where the first feature words are keywords corresponding to preset dimensions, and the preset dimensions include project category, project content, project cycle, project budget, and construction environment; Perform Hamming similarity calculation on the first feature word and any preset encrypted semantic report in the preset processing model to obtain the target encrypted semantic report; According to the target encrypted semantic report, the response result is obtained. 5. The engineering consulting processing method based on deep learning according to claim 4, characterized in that, the first feature word is Hamming with any one of the preset encrypted semantic reports in the preset processing model. Similarity calculation to obtain the target encryption semantic report, including: Extract the second feature word in any preset encrypted semantic report; Calculate the Hamming distance between the first feature word and the second feature word; Compare the size relationship between the Hamming distance and the preset Hamming distance; If the Hamming distance is less than or equal to the preset Hamming distance, it is confirmed that the preset encrypted semantic report corresponding to the second feature word is the target encrypted semantic report. 6. The engineering consulting processing method based on deep learning according to claim 3, characterized in that the step of inputting the encrypted semantic report into a preset processing model to obtain a response result specifically includes: Obtain the topic semantics in any historical response result, and multiple historical response results are pre-stored in the preset processing model; Calculate the response tag value corresponding to each topic semantics; Determine the similarity between the response tag value and the semantic tag value; If it is confirmed that the similarity is greater than or equal to the preset similarity threshold, it is determined that the historical response result corresponding to the response tag value is the response result. 7. The engineering consulting processing method based on deep learning according to claim 1, characterized in that, before inputting the encrypted semantic report into a preset processing model and obtaining the response result, the preset processing model is trained. ; The training of the preset processing model specifically includes: Obtain training information, which includes encrypted semantic reports and response results; Input the training information into the adaptive feature fusion network for training, and obtain the first training result; After superimposing and standardizing the first training result and the training information, a second training result is obtained; Input the second training result into the adaptive feature fusion network for processing to obtain a third training result; The third training result and the second training result are superimposed and standardized until the training information similarity matrix is output, and the training information similarity matrix satisfies the preset logistic regression condition. 8. An engineering consultation processing device based on deep learning, characterized in that the processing device includes a receiving module (21) and a processing module (22), wherein, The receiving module (21) is used to receive a consulting request sent by the user equipment, where the consulting request is used to represent a consulting request for the target project; The processing module (22) is used to generate semantic reports according to the consultation request; The processing module (22) is also used to encrypt the semantic report to obtain an encrypted semantic report; The processing module (22) is also used to input the encrypted semantic report into a preset processing model to obtain a response result; The processing module (22) is also used to send the response result to the user equipment. 9. An electronic device, characterized in that the electronic device includes a processor (31), a memory (35), a user interface (33) and a network interface (34), and the memory (35) is used to store instructions, The user interface (33) and the network interface (34) are both used to communicate with other devices, and the processor (31) is used to execute instructions stored in the memory (35) so that the electronic device The method as claimed in any one of claims 1 to 7 is carried out. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.