Knowledge enhancement method and system based on capital construction large model

Abstract

The invention discloses a knowledge enhancement method and system based on an infrastructure large model, and the method comprises the steps: collecting massive online power infrastructure question and answer data, and extracting power infrastructure question and answer data keywords which comprise question data provided by a user, power infrastructure project data and power answer data; and generating a sub-neural network tree for the question data proposed by the user, the power infrastructure project data and the power answer data. By establishing the first association mechanism, establishing the second association mechanism, rechecking the power capital construction question and answer data, analyzing the power capital construction question and answer data generated according to the big data, and calculating the conditional probability of occurrence of the common keywords of the power capital construction question and answer data, the accuracy of the power answer data can be effectively enhanced, and the user experience is improved. And the matching performance of electric power answering results can be enhanced, questions proposed by users can be automatically answered in real time, convenience and rapidness are achieved, and the requirements of workers are met.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a knowledge enhancement method and system based on a large infrastructure model. Background Art

[0002] With the continuous advancement of computer technology and artificial intelligence, the power infrastructure field has gradually begun to adopt big data analysis technology to improve the efficiency and quality of power infrastructure services. In terms of power infrastructure Q&A, power infrastructure Q&A based on big data analysis has gradually received attention and attention, becoming a new trend in the development of the power infrastructure industry. The application of big data analysis in the power infrastructure Q&A system, the power infrastructure Q&A system based on big data analysis, can provide users with more accurate, comprehensive, and personalized Q&A services by collecting, integrating, analyzing and mining a large amount of Q&A data.

[0003] However, the accuracy of the answer data and the matching of the diagnostic results obtained by the existing analysis methods are both low and need to be enhanced. It is difficult to meet the needs of the staff and is not conducive to promotion and use. Summary of the invention

[0004] In order to solve the above technical problems, a knowledge enhancement method and system based on a large infrastructure model is provided. This technical solution solves the problem that the accuracy of the answer data and the matching of the diagnostic results obtained by the existing analysis method proposed in the above background technology are low and need to be enhanced.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A knowledge enhancement method based on a large infrastructure model, comprising:

[0007] Collect massive amounts of online power infrastructure Q&A data and extract keywords from the power infrastructure Q&A data. Keywords include user-asked questions, power infrastructure project data, and power answer data.

[0008] Generate sub-neural network trees for question data, power infrastructure project data and power answer data raised by users, and merge sub-neural network trees generated from question data, power infrastructure project data and power answer data raised by the same user into a total neural network tree;

[0009] Count the common keywords that appear in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords appearing in the power infrastructure question and answer;

[0010] According to the judgment rule, the relevance between the common keywords of the power infrastructure question and answer and the common keywords appearing in the other sub-neural network trees is determined, and a first relevance mechanism is established;

[0011] Calculate the conditional probability of common keywords appearing in power answers under the condition of power infrastructure questions and answers;

[0012] According to the judgment rules, the relevance of commonly used keywords in power answers and commonly used keywords in power infrastructure questions and answers is determined, and a second relevance mechanism is established;

[0013] Get real-time power infrastructure Q&A data from users, review the power answer data, determine its accuracy, and provide expert corrections;

[0014] As the total neural network tree is continuously updated, the association mechanism is supplemented and updated. The above total neural network tree and association mechanism are stored in the central processing library.

[0015] Preferably, the generation of a sub-neural network tree from the user's question data, power infrastructure project data and power answer data specifically includes the following steps:

[0016] According to the keywords in the question data raised by users, the power infrastructure project data and the power answer data, duplicate keywords are deleted and merged;

[0017] Generate a keyword tree under the respective keywords in the question data raised by the user, the power infrastructure project data and the power answer data;

[0018] Map the keyword tree to the corresponding user question data, power infrastructure project data and power answer data;

[0019] Sub-neural network trees are formed for the user's question data, power infrastructure project data and power answer data.

[0020] Preferably, merging the sub-neural network trees into a total neural network tree specifically comprises the following steps:

[0021] Users generate personal electricity accounts in the system;

[0022] The sub-neural network trees of the user's question data, power infrastructure project data and power answer data are mapped to the power personal account;

[0023] Form the total neural network tree.

[0024] Preferably, the counting of common keywords appearing in each sub-neural network tree in the total neural network tree specifically includes the following steps:

[0025] Count the keywords of each sub-neural network tree in all the total neural network trees;

[0026] Calculate the frequency of occurrence of a single keyword;

[0027] If the frequency of occurrence is greater than 0.5%, it is a common keyword. If the frequency of occurrence does not exceed 0.5%, it is not a common keyword.

[0028] Preferably, the calculation of the conditional probability of occurrence of common keywords in the electric power infrastructure question and answer specifically includes the following steps:

[0029] Take any common keyword A that appears in each sub-neural network tree in all the total neural network trees;

[0030] The frequency N of the common keyword A appearing is searched in each sub-neural network tree;

[0031] Take any one of the commonly used keywords in the power infrastructure question and answer B;

[0032] In each sub-neural network tree, the frequency M of common keyword B appearing when common keyword A appears is counted;

[0033] Then the conditional probability of the common keyword B appearing is P=M / N.

[0034] Preferably, the determination rule is as follows:

[0035] If the conditional probability P is greater than 80%, it is judged to be related;

[0036] If the conditional probability does not exceed P and is greater than 80%, it is judged that there is no correlation.

[0037] Preferably, the establishing of the first association mechanism specifically comprises the following steps:

[0038] If the commonly used keywords of power infrastructure questions and answers are related to the commonly used keywords appearing in other sub-neural network trees, a conditional reflex is established between the two, that is, when the commonly used keywords in other sub-neural network trees appear, the commonly used keywords of power infrastructure questions and answers with related keywords are returned.

[0039] Preferably, the review of the power solution data specifically includes the following steps:

[0040] Extract common keywords from the power answer data, and pair them in groups of two;

[0041] Count the frequencies K and T of the common keywords in the common keyword groups in each sub-neural network tree, count the frequencies Z of the paired common keyword groups appearing simultaneously in each sub-neural network tree, and calculate the probability L=Z / (K+T);

[0042] If the probability is higher than 0.1%, it means that the power solution is normal. If the probability does not exceed 0.1%, it means that the power solution is abnormal and experts will correct it.

[0043] In addition, the present invention also provides a knowledge enhancement system based on a large infrastructure model, comprising:

[0044] A collection module, which is used to collect massive online power infrastructure question and answer data and extract keywords from the power infrastructure question and answer data, the keywords including question data raised by users, power infrastructure project data and power answer data;

[0045] A neural network tree generation module, the neural network tree generation module is used to generate sub-neural network trees for question data raised by users, power infrastructure project data and power answer data, and merge sub-neural network trees generated from question data raised by the same user, power infrastructure project data and power answer data into a total neural network tree;

[0046] A computing module, wherein the computing module internally integrates a first computing unit and a second computing unit;

[0047] An establishment module, wherein the internal integration of the establishment module includes a first establishment unit and a second establishment unit;

[0048] A review module, which is used to obtain real-time power infrastructure question and answer data from users, review the power answer data, determine its accuracy, and provide expert corrections;

[0049] A first calculation unit, the first calculation unit is used to count the common keywords appearing in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords of the power infrastructure question and answer appearing;

[0050] A second calculation unit, the second calculation unit is used to calculate the conditional probability of the common keywords of the power answer appearing under the condition of power infrastructure question and answer;

[0051] A first establishing unit, the first establishing unit is used to determine the relevance of common keywords of the power infrastructure question and answer and common keywords appearing in other sub-neural network trees according to the determination rule, and establish a first association mechanism;

[0052] The second establishing unit is used to determine the relevance of the commonly used keywords for power answers and the commonly used keywords for power infrastructure questions and answers according to the determination rule, and establish a second association mechanism.

[0053] Compared with the prior art, the present invention provides a knowledge enhancement method and system based on a large infrastructure model, which has the following beneficial effects:

[0054] The present invention establishes a first association mechanism, establishes a second association mechanism, and reviews the power infrastructure question and answer data. It analyzes the power infrastructure question and answer data generated by big data and calculates the conditional probability of the occurrence of common keywords in the power infrastructure question and answer data. This can effectively enhance the accuracy of the power answer data and the matching of the power answer results, automatically provide real-time answers to questions raised by users, is convenient and fast, and meets the needs of staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Schematic diagram of the knowledge enhancement method in the present invention;

[0056] Figure 2 A schematic diagram of a method for generating a sub-neural network tree for question data raised by a user, power infrastructure project data and power answer data in the present invention;

[0057] Figure 3 A schematic diagram of a method for merging sub-neural network trees into a total neural network tree according to the present invention;

[0058] Figure 4 Schematic diagram of the method for counting common keywords appearing in each sub-neural network tree in all total neural network trees in the present invention. DETAILED DESCRIPTION

[0059] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0060] Example 1

[0061] Please refer to Figure 1-Figure 4 As shown, a knowledge enhancement method and system based on a large infrastructure model includes:

[0062] Collect massive amounts of online power infrastructure Q&A data and extract keywords from the power infrastructure Q&A data. Keywords include user-asked questions, power infrastructure project data, and power answer data.

[0063] Generate sub-neural network trees for question data, power infrastructure project data and power answer data raised by users, and merge sub-neural network trees generated from question data, power infrastructure project data and power answer data raised by the same user into a total neural network tree;

[0064] Count the common keywords that appear in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords appearing in the power infrastructure question and answer;

[0065] According to the judgment rule, the relevance between the common keywords of the power infrastructure question and answer and the common keywords appearing in the other sub-neural network trees is determined, and a first relevance mechanism is established;

[0066] Calculate the conditional probability of common keywords appearing in power answers under the condition of power infrastructure questions and answers;

[0067] According to the judgment rules, the relevance of commonly used keywords in power answers and commonly used keywords in power infrastructure questions and answers is determined, and a second relevance mechanism is established;

[0068] Get real-time power infrastructure Q&A data from users, review the power answer data, determine its accuracy, and provide expert corrections;

[0069] As the total neural network tree is continuously updated, the association mechanism is supplemented and updated. The above total neural network tree and association mechanism are stored in the central processing library.

[0070] It can be understood by those skilled in the art that the present invention, by establishing a first association mechanism, establishing a second association mechanism, and reviewing the power infrastructure question and answer data, analyzes the power infrastructure question and answer data generated by big data, and calculates the conditional probability of the occurrence of common keywords in the power infrastructure question and answer data, thereby effectively enhancing the accuracy of the power answer data and the matching of the power answer results, automatically providing real-time answers to questions raised by users, which is convenient and quick, and meets the needs of the staff.

[0071] The generation of a sub-neural network tree for the user's question data, power infrastructure project data and power answer data specifically includes the following steps:

[0072] According to the keywords in the question data raised by users, the power infrastructure project data and the power answer data, duplicate keywords are deleted and merged;

[0073] Generate a keyword tree under the respective keywords in the question data raised by the user, the power infrastructure project data and the power answer data;

[0074] Map the keyword tree to the corresponding user question data, power infrastructure project data and power answer data;

[0075] Sub-neural network trees are formed for the user's question data, power infrastructure project data and power answer data.

[0076] The merging of sub-neural network trees into a total neural network tree specifically includes the following steps:

[0077] Users generate personal electricity accounts in the system;

[0078] The sub-neural network trees of the user's question data, power infrastructure project data and power answer data are mapped to the power personal account;

[0079] Form the total neural network tree.

[0080] The specific steps of counting the common keywords appearing in each sub-neural network tree in all the total neural network trees include:

[0081] Count the keywords of each sub-neural network tree in all the total neural network trees;

[0082] Calculate the frequency of occurrence of a single keyword;

[0083] If the frequency of occurrence is greater than 0.5%, it is a common keyword. If the frequency of occurrence does not exceed 0.5%, it is not a common keyword.

[0084] Calculating the conditional probability of common keywords appearing in power infrastructure questions and answers specifically includes the following steps:

[0085] Take any common keyword A that appears in each sub-neural network tree in all the total neural network trees;

[0086] The frequency N of the common keyword A appearing is searched in each sub-neural network tree;

[0087] Take any one of the commonly used keywords in the power infrastructure question and answer B;

[0088] In each sub-neural network tree, the frequency M of common keyword B appearing when common keyword A appears is counted;

[0089] Then the conditional probability of the common keyword B appearing is P=M / N.

[0090] The judgment rules are as follows:

[0091] If the conditional probability P is greater than 80%, it is judged to be related;

[0092] If the conditional probability does not exceed P and is greater than 80%, it is judged that there is no correlation.

[0093] Those skilled in the art can understand that the purpose of calculating the conditional probability is mainly to find the connection between the common keywords of the question data raised by the user, the power infrastructure project data and the common keywords of the power answer data, that is, to link the question data raised by the user, the power infrastructure project data and the power answer data to each other, so that when encountering the same question data, the corresponding power answer data can be automatically obtained through computer analysis. Taking the common keyword C of the question data as an example, when the common keyword D of the power answer data is calculated to appear, if the probability of the common keyword C appearing is higher than 80%, it means that the appearance of the common keyword C will lead to the answer of the common keyword D, because if the common keyword C will not lead to the answer of the common keyword D, then based on the statistics of big data, when there is the answer of the common keyword D, the possibility of the common keyword C appearing cannot be so high, on the contrary, it should be lower than 20%, or even lower. Therefore, through this conditional probability algorithm, the common keywords of the question data raised by the user corresponding to the common keywords that lead to each answer can be obtained.

[0094] Establishing the first association mechanism specifically includes the following steps:

[0095] If the commonly used keywords of power infrastructure questions and answers are related to the commonly used keywords appearing in other sub-neural network trees, a conditional reflex is established between the two, that is, when the commonly used keywords in other sub-neural network trees appear, the commonly used keywords of power infrastructure questions and answers with related keywords are returned.

[0096] The review of power solution data includes the following steps:

[0097] Extract common keywords from the power answer data, and pair them in groups of two;

[0098] Count the frequencies K and T of the common keywords in the common keyword groups in each sub-neural network tree, count the frequencies Z of the paired common keyword groups appearing simultaneously in each sub-neural network tree, and calculate the probability L=Z / (K+T);

[0099] If the probability is higher than 0.1%, it means that the power solution is normal. If the probability does not exceed 0.1%, it means that the power solution is abnormal and experts will correct it.

[0100] It is understandable to those skilled in the art that, based on statistics from big data, when one of two commonly used keywords appears, the probability of both appearing at the same time is extremely low, less than 0.1%. This means that in a normal environment, it is difficult for two commonly used keywords to appear on the same answer at the same time. If they do appear, it means that there may be a problem with the question data, which may result in errors in the filling of these two keywords. Therefore, a review is required.

[0101] In addition, the present invention also provides a knowledge enhancement system based on a large infrastructure model, comprising:

[0102] The collection module is used to collect massive online power infrastructure question and answer data and extract keywords from the power infrastructure question and answer data. The keywords include question data raised by users, power infrastructure project data, and power answer data.

[0103] A neural network tree generation module, the neural network tree generation module is used to generate sub-neural network trees for question data raised by users, power infrastructure project data and power answer data, and merge sub-neural network trees generated from question data raised by the same user, power infrastructure project data and power answer data into a total neural network tree;

[0104] A computing module, wherein the computing module internally includes a first computing unit and a second computing unit;

[0105] A building module, wherein the internal integration of the building module includes a first building unit and a second building unit;

[0106] The review module is used to obtain the real-time power infrastructure question and answer data from users, review the power answer data, determine its accuracy, and provide expert corrections;

[0107] A first calculation unit, the first calculation unit is used to count the common keywords appearing in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords of the power infrastructure question and answer appearing;

[0108] The second calculation unit is used to calculate the conditional probability of the common keywords appearing in the power answer under the condition of power infrastructure question and answer;

[0109] A first establishing unit, the first establishing unit is used to determine the relevance of the common keywords of the electric power infrastructure question and answer and the common keywords appearing in the other sub-neural network trees according to the determination rule, and establish a first association mechanism;

[0110] The second establishing unit is used to determine the relevance of the commonly used keywords for power answers and the commonly used keywords for power infrastructure questions and answers according to the determination rule, and establish a second association mechanism.

[0111] To summarize, the present invention establishes a first association mechanism, establishes a second association mechanism, and reviews the power infrastructure question and answer data, analyzes the power infrastructure question and answer data generated by big data, and calculates the conditional probability of the occurrence of common keywords in the power infrastructure question and answer data. This can effectively enhance the accuracy of the power answer data and the matching of the power answer results, automatically provide real-time answers to questions raised by users, is convenient and fast, and meets the needs of staff.

[0112] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A knowledge enhancement method based on a large infrastructure model, characterized in that: include: Collect massive amounts of online power infrastructure Q&A data and extract keywords from the power infrastructure Q&A data. Keywords include user-asked questions, power infrastructure project data, and power answer data. Generate sub-neural network trees for question data, power infrastructure project data and power answer data raised by users, and merge sub-neural network trees generated from question data, power infrastructure project data and power answer data raised by the same user into a total neural network tree; Count the common keywords that appear in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords appearing in the power infrastructure question and answer; According to the judgment rule, the relevance between the common keywords of the power infrastructure question and answer and the common keywords appearing in the other sub-neural network trees is determined, and a first relevance mechanism is established; Calculate the conditional probability of common keywords appearing in power answers under the condition of power infrastructure questions and answers; According to the judgment rules, the relevance of commonly used keywords in power answers and commonly used keywords in power infrastructure questions and answers is determined, and a second relevance mechanism is established; Get real-time power infrastructure Q&A data from users, review the power answer data, determine its accuracy, and provide expert corrections; As the total neural network tree is continuously updated, the association mechanism is supplemented and updated. The above total neural network tree and association mechanism are stored in the central processing library.

2. The knowledge enhancement method based on the infrastructure large model according to claim 1 is characterized in that: The generation of a sub-neural network tree for the user's question data, power infrastructure project data and power answer data specifically includes the following steps: According to the keywords in the question data raised by users, the power infrastructure project data and the power answer data, duplicate keywords are deleted and merged; Generate a keyword tree under the respective keywords in the question data raised by the user, the power infrastructure project data and the power answer data; Map the keyword tree to the corresponding user question data, power infrastructure project data and power answer data; Sub-neural network trees are formed for the user's question data, power infrastructure project data and power answer data.

3. The knowledge enhancement method based on the infrastructure large model according to claim 2 is characterized in that: The merging of the sub-neural network trees into the total neural network tree specifically comprises the following steps: Users generate personal electricity accounts in the system; The sub-neural network trees of the user's question data, power infrastructure project data and power answer data are mapped to the power personal account; Form the total neural network tree.

4. The knowledge enhancement method based on infrastructure large model according to claim 3 is characterized in that: The method of counting the common keywords appearing in each sub-neural network tree in the total neural network tree specifically comprises the following steps: Count the keywords of each sub-neural network tree in all the total neural network trees; Calculate the frequency of occurrence of a single keyword; If the frequency of occurrence is greater than 0.5%, it is a common keyword. If the frequency of occurrence does not exceed 0.5%, it is not a common keyword.

5. The knowledge enhancement method based on infrastructure large model according to claim 4 is characterized in that: The calculation of the conditional probability of the common keywords appearing in the power infrastructure question and answer specifically includes the following steps: Take any common keyword A that appears in each sub-neural network tree in all the total neural network trees; The frequency N of the common keyword A appearing is searched in each sub-neural network tree; Take any one of the commonly used keywords in the power infrastructure question and answer B; In each sub-neural network tree, the frequency M of common keyword B appearing when common keyword A appears is counted; Then the conditional probability of the common keyword B appearing is P=M / N.

6. The knowledge enhancement method based on infrastructure large model according to claim 5 is characterized in that: The determination rules are as follows: If the conditional probability P is greater than 80%, it is judged to be related; If the conditional probability does not exceed P and is greater than 80%, it is judged that there is no correlation.

7. The knowledge enhancement method based on infrastructure large model according to claim 6 is characterized in that: The establishment of the first association mechanism specifically includes the following steps: If the commonly used keywords of power infrastructure questions and answers are related to the commonly used keywords appearing in other sub-neural network trees, a conditional reflex is established between the two, that is, when the commonly used keywords in other sub-neural network trees appear, the commonly used keywords of power infrastructure questions and answers with related keywords are returned.

8. The knowledge enhancement method based on infrastructure large model according to claim 7 is characterized in that: The review of the power solution data specifically includes the following steps: Extract common keywords from the power answer data, and pair them in groups of two; Count the frequencies K and T of the common keywords in the common keyword groups in each sub-neural network tree, count the frequencies Z of the paired common keyword groups appearing simultaneously in each sub-neural network tree, and calculate the probability L=Z / (K+T); If the probability is higher than 0.1%, it means that the power solution is normal. If the probability does not exceed 0.1%, it means that the power solution is abnormal and experts will correct it.

9. A knowledge enhancement system based on a large infrastructure model, used to implement a knowledge enhancement method based on a large infrastructure model as described in any one of claims 1 to 8, characterized in that: include: A collection module, which is used to collect massive online power infrastructure question and answer data and extract keywords from the power infrastructure question and answer data, the keywords including question data raised by users, power infrastructure project data and power answer data; A neural network tree generation module, the neural network tree generation module is used to generate sub-neural network trees for question data raised by users, power infrastructure project data and power answer data, and merge sub-neural network trees generated from question data raised by the same user, power infrastructure project data and power answer data into a total neural network tree; A computing module, wherein the computing module internally integrates a first computing unit and a second computing unit; An establishment module, wherein the internal integration of the establishment module includes a first establishment unit and a second establishment unit; The review module is used to obtain the user's real-time power infrastructure question and answer data, review the power answer data, determine its accuracy, and provide expert corrections.

10. The knowledge enhancement system based on infrastructure large model according to claim 1 is characterized in that: include: A first calculation unit, the first calculation unit is used to count the common keywords appearing in each sub-neural network tree in all the total neural network trees, and calculate the conditional probability of the common keywords of the power infrastructure question and answer appearing; A second calculation unit, the second calculation unit is used to calculate the conditional probability of the common keywords of the power answer appearing under the condition of power infrastructure question and answer; A first establishing unit, the first establishing unit is used to determine the relevance of common keywords of the power infrastructure question and answer and common keywords appearing in other sub-neural network trees according to the determination rule, and establish a first association mechanism; The second establishing unit is used to determine the relevance of the commonly used keywords for power answers and the commonly used keywords for power infrastructure questions and answers according to the determination rule, and establish a second association mechanism.

Images