An intelligent data-driven rapid construction optimization system for power production scenarios

Abstract

The present invention discloses an intelligent query rapid construction and optimization system for power production scenarios, which relates to the field of artificial intelligence technology. The system includes: a word vector management module, a question management module, a semantic extraction module, a semantic offset management module, a semantic offset management module, a feature word screening module, and a feature word compensation module. By dynamically adjusting semantic weight values, accurately judging semantic offsets, strengthening key feature words, and compensating for word vector offsets, the accuracy of user intent tracking and the matching degree of query results in multi-round conversations are significantly improved. This application can effectively avoid query deviations caused by semantic offsets or improper allocation of feature word weights, and meet the needs of efficient and accurate data queries in power production scenarios.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent data-driven rapid construction optimization system for power production scenarios. Background Art

[0002] The Intelligent Questioning System is an intelligent question-answering tool that combines dynamic adjustment, context modeling, and domain knowledge enhancement technologies. The system obtains the characteristics of user questions and conversation scenarios, matches the associated word model, and independently designs associated words to guide user questions, thereby making up for the shortcomings of manual natural language input. It can also recognize and process free-form natural language input to improve the accuracy of data query tasks.

[0003] Existing intelligent question-answering systems have limitations in terms of the depth of semantic understanding and the accuracy of intent capture. Although specific algorithms can improve the efficiency of keyword extraction, the system's ability to predict changes in user intent over multiple rounds in continuous conversation scenarios still needs to be improved. Furthermore, when semantic drift occurs in user questions, the system's real-time detection and compensation mechanisms need further optimization, potentially leading to discrepancies between query results and user needs. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent data rapid construction optimization system for power production scenarios to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent data-driven rapid construction optimization system for power production scenarios.

[0006] The system includes: a word vector management module, a question management module, a semantic extraction module, a semantic offset management module, a semantic offset management module, a feature word screening module and a feature word compensation module.

[0007] The word vector management module is used to collect historical records of continuous dialogue between the questioner and the agent, extract characteristic words from the question information, and establish a correspondence between characteristic words and word vectors through word vector tags;

[0008] The question management module is used to collect the question records of the questioner to the agent in each round of dialogue scenarios, and compile the questions raised by the questioner into a question sequence;

[0009] The semantic extraction module is used to classify the feature words in the question, use the classification results as semantic features, identify and label the semantic features, and obtain the feature words under each semantic feature;

[0010] Semantic offset management module: used to calculate the change value of the semantic features in the question information, including the semantic weight value of the target semantic feature and the semantic change evaluation value of the same semantic feature in two consecutive rounds of questions;

[0011] Semantic drift management module: used to calculate the semantic drift threshold and determine whether the questioner's question intention has a semantic drift based on the semantic drift threshold;

[0012] The feature word screening module is used to screen the feature words that need to be strengthened in the question content, collect the feature words that need to be strengthened, and obtain a feature word set;

[0013] The feature word compensation module is used to calculate the vector offset and compensate the word vector to generate a new vector. It also obtains the feature word corresponding to the new vector based on the correspondence between the vector and the feature word, and outputs the feature word to the questioner for selection.

[0014] Furthermore, in the word vector management module, when collecting the historical records of the questioner in the continuous dialogue scenario with the intelligent agent, the feature words in the question information are analyzed through natural language processing technology, and the word embedding algorithm is used to convert the feature words into word vectors in a high-dimensional space, and a mapping relationship between the feature words and the word vectors is established.

[0015] Furthermore, in the question management module, when collecting the question records of the questioner to the intelligent agent in each round, the question content is obtained through voice recognition or text input, and the question content is arranged in chronological order to form a question sequence; the question sequence is indexed by timestamp, and supports retrieval and analysis of question distribution patterns in different time periods by time range.

[0016] Furthermore, in the semantic extraction module, when classifying the feature words in the question, the feature words are divided into equipment status class, operating parameter class, fault diagnosis class and scheduling strategy class based on the domain knowledge base and semantic network model.

[0017] Furthermore, the semantic offset management module includes: a semantic feature value calculation unit and a semantic change management unit;

[0018] The semantic feature value calculation unit is used to calculate the i-th semantic feature in the t-th round of questions as the target semantic feature and calculate the semantic weight value W of the target semantic feature. t,i , , where q ti represents the number of feature words in the i-th semantic feature in the t-th round of questions, Q t Indicates the number of feature words included in the t-th round of questions;

[0019] The semantic change management unit is used to calculate the semantic change evaluation value of the same semantic feature in two adjacent rounds of questions, and obtain the semantic weight value W of the target semantic feature in the t-1 round of questions. t-1,i , calculate the semantic change value R of the target semantics in the tth round of questions t , .

[0020] The semantic offset management module dynamically analyzes the semantic weight value to ensure that the importance of the target semantic features in each round of questioning can be accurately captured.

[0021] Furthermore, the semantic shift management module includes: a semantic change value prediction unit, a shift threshold management unit and a shift judgment unit;

[0022] The semantic change value prediction unit is used to calculate the semantic change prediction value, obtain the semantic change value of the target semantic feature in the nth round of questioning, n ≥ 3, where n rounds of questioning end with the tth round of questioning, and calculate the semantic change prediction value R of the target semantic feature in the t+1th round by function fitting. t+1,p ;

[0023] The offset threshold management unit is used to calculate the semantic offset threshold, and calculate the semantic offset threshold S of the target semantics in the t+1 round of questioning. t+1 , , α represents the weight value, satisfying the condition: 0<α<1, where ω t-1,t Indicates the similarity between the word vectors of all feature words in the t-1th round of questions and the tth round of questions;

[0024] The deviation judgment unit is used to judge whether the questioner's question intention is deviated and obtain the true value R of the semantic change of the question in the t+1 round. t+1,a , when R t+1,a >S t+1 When , it is judged that in the t+1 round of questions, the target semantics has a semantic shift.

[0025] The semantic shift management module combines the semantic change prediction value and context similarity to calculate the semantic shift threshold, effectively judging whether the semantics have shifted, thereby avoiding query deviation caused by semantic shift.

[0026] Furthermore, the feature word screening module includes: a word frequency acquisition unit, a word evaluation value management unit and a feature word set management unit;

[0027] The word frequency acquisition unit is used to obtain the word frequency of each feature word in the target semantics in the t+1 round of questions, and the number of question rounds in which each feature word appears in the previous t+1 rounds. In the t+1 round of questions, the word frequency of the kth feature word in the target semantics in the t+1 round of questions is recorded as TF k , in the first t+1 rounds of questions, the number of rounds in which the kth feature word appears is recorded as Nk ;

[0028] The word evaluation value management unit is used to calculate the word evaluation value B of the kth feature word k , ;

[0029] The feature word set management unit is used to arrange the feature words of the target semantics from high to low according to the word evaluation values, and put the first m feature words into the feature word set.

[0030] The feature word screening module screens key feature words by calculating word evaluation values to ensure that the feature word set can accurately reflect the core content of the target semantics.

[0031] Furthermore, the feature word compensation module includes: a vector offset calculation unit, an offset coefficient calculation unit, a vector correction unit and a feature word output unit;

[0032] The vector offset calculation unit is used to calculate the vector offset. The vector offset is the average value of the word vectors of the feature words included in all semantic features in the t+1th round of questions, which is recorded as avg;

[0033] The offset coefficient calculation unit is used to calculate the offset coefficient γ, ;

[0034] The vector correction unit is used to calculate the correction vector of the word vector, where the correction vector p of the kth feature word is calculated k , , where v k Table k-th feature word corresponding to the word vector;

[0035] The feature word output unit is used to collect the feature words corresponding to the correction vector and output the feature vocabulary set corresponding to the correction vector to the questioner.

[0036] The feature word compensation module generates new feature words by calculating the correction vector to compensate for the word vector offset problem caused by semantic offset or improper distribution of feature word weights, thereby improving the matching degree between query results and user needs.

[0037] Compared with existing technologies, the present invention significantly improves the accuracy of user intent tracking and query result matching in multi-round conversations by dynamically adjusting semantic weights, accurately determining semantic shifts, strengthening key feature words, and compensating for word vector shifts. This application effectively avoids query bias caused by semantic shifts or improperly assigned feature word weights, meeting the demand for efficient and accurate data queries in power production scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a structural diagram of an intelligent data-driven rapid optimization system for power production scenarios according to the present invention. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] Example: Figure 1 As shown, the present invention provides a technical solution, an intelligent data rapid construction optimization system for power production scenarios.

[0041] In power generation scenarios, equipment status monitoring is a typical application of intelligent question-and-answer systems. First, historical records of continuous dialogues between the questioner and the agent are collected, and natural language processing techniques are used to analyze the characteristic words in the questions. For example, when a questioner inquires about the operating status of a generator, the question may contain keywords such as "generator," "temperature," and "vibration." These keywords are converted into word vectors in a high-dimensional space using a word embedding algorithm, and a mapping relationship is established between the characteristic words and the word vectors. The dimensionality of the word vectors is determined by a pretrained language model, and the update frequency is linked to the real-time nature of the question information. This real-time update mechanism ensures that the word vectors can dynamically adapt to changes in the question content, thereby improving the system's response accuracy.

[0042] The system includes: a word vector management module, a question management module, a semantic extraction module, a semantic offset management module, a semantic offset management module, a feature word screening module and a feature word compensation module.

[0043] The word vector management module is used to collect historical records of continuous dialogue between the questioner and the agent, extract characteristic words from the question information, and establish a correspondence between characteristic words and word vectors through word vector tags;

[0044] Among them, in the word vector management module, when collecting the historical records of the questioner in the continuous dialogue scenario with the intelligent agent, the feature words in the question information are analyzed through natural language processing technology, and the word embedding algorithm is used to convert the feature words into word vectors in high-dimensional space, and establish a mapping relationship between feature words and word vectors.

[0045] The question management module is used to collect the question records of the questioner to the agent in each round of dialogue scenarios, and compile the questions raised by the questioner into a question sequence;

[0046] Among them, in the question management module, when collecting the question records of the questioner to the intelligent agent in each round, the question content is obtained through voice recognition or text input, and the question content is arranged in chronological order to form a question sequence; the question sequence is indexed by timestamp, and supports retrieval and analysis of question distribution patterns in different time periods by time range.

[0047] The semantic extraction module is used to classify the feature words in the question, use the classification results as semantic features, identify and label the semantic features, and obtain the feature words under each semantic feature;

[0048] Among them, in the semantic extraction module, when classifying the feature words in the question, the feature words are divided into equipment status class, operating parameter class, fault diagnosis class and scheduling strategy class based on the domain knowledge base and semantic network model.

[0049] Semantic offset management module: used to calculate the change value of the semantic features in the question information, including the semantic weight value of the target semantic feature and the semantic change evaluation value of the same semantic feature in two consecutive rounds of questions;

[0050] The semantic offset management module includes: a semantic feature value calculation unit and a semantic change management unit.

[0051] The semantic feature value calculation unit is used to calculate the i-th semantic feature in the t-th round of questions as the target semantic feature and calculate the semantic weight value W of the target semantic feature. t,i , , where q ti represents the number of feature words in the i-th semantic feature in the t-th round of questions, Q t Indicates the number of feature words included in the t-th round of questions;

[0052] The semantic change management unit is used to calculate the semantic change evaluation value of the same semantic feature in two adjacent rounds of questions, and obtain the semantic weight value W of the target semantic feature in the t-1 round of questions. t-1,i , calculate the semantic change value R of the target semantics in the tth round of questions t , .

[0053] Semantic drift management module: used to calculate the semantic drift threshold and determine whether the questioner's question intention has a semantic drift based on the semantic drift threshold;

[0054] The semantic shift management module includes: a semantic change value prediction unit, a shift threshold management unit and a shift judgment unit.

[0055] The semantic change value prediction unit is used to calculate the semantic change prediction value, obtain the semantic change value of the target semantic feature in the nth round of questioning, n ≥ 3, where n rounds of questioning end with the tth round of questioning, and calculate the semantic change prediction value R of the target semantic feature in the t+1th round by function fitting. t+1,p ;

[0056] The offset threshold management unit is used to calculate the semantic offset threshold, and calculate the semantic offset threshold S of the target semantics in the t+1 round of questioning. t+1 , , α represents the weight value, satisfying the condition: 0<α<1, where ω t-1,t Indicates the similarity between the word vectors of all feature words in the t-1th round of questions and the tth round of questions;

[0057] The deviation judgment unit is used to judge whether the questioner's question intention is deviated and obtain the true value R of the semantic change of the question in the t+1 round. t+1,a , when R t+1,a >S t+1 When , it is judged that in the t+1 round of questions, the target semantics has a semantic shift.

[0058] The feature word screening module is used to screen the feature words that need to be strengthened in the question content, collect the feature words that need to be strengthened, and obtain a feature word set;

[0059] The feature word screening module includes: a word frequency acquisition unit, a word evaluation value management unit and a feature word set management unit.

[0060] The word frequency acquisition unit is used to obtain the word frequency of each feature word in the target semantics in the t+1 round of questions, and the number of question rounds in which each feature word appears in the previous t+1 rounds. In the t+1 round of questions, the word frequency of the kth feature word in the target semantics in the t+1 round of questions is recorded as TF k , in the first t+1 rounds of questions, the number of rounds in which the kth feature word appears is recorded as N k ;

[0061] The word evaluation value management unit is used to calculate the word evaluation value B of the kth feature word k , ;

[0062] The feature word set management unit is used to arrange the feature words of the target semantics from high to low according to the word evaluation values, and put the first m feature words into the feature word set.

[0063] The feature word compensation module is used to calculate the vector offset and compensate the word vector to generate a new vector. Based on the correspondence between the vector and the feature word, the feature word corresponding to the new vector is obtained and output to the questioner for selection.

[0064] The feature word compensation module includes: a vector offset calculation unit, an offset coefficient calculation unit, a vector correction unit and a feature word output unit.

[0065] The vector offset calculation unit is used to calculate the vector offset. The vector offset is the average value of the word vectors of the feature words included in all semantic features in the t+1th round of questions, which is recorded as avg.

[0066] The offset coefficient calculation unit is used to calculate the offset coefficient γ, ;

[0067] The vector correction unit is used to calculate the correction vector of the word vector, where the correction vector p of the kth feature word is calculated k , , where v k Table k-th feature word corresponding to the word vector;

[0068] The feature word output unit is used to collect the feature words corresponding to the correction vector and output the feature vocabulary set corresponding to the correction vector to the questioner.

[0069] In practical applications, the method of this invention is applicable to multi-turn dialogue systems in power production scenarios, such as those used to query equipment operating status, retrieve fault diagnosis information, and collect production data. For example, at a power company, operations and maintenance personnel need to obtain historical operating data or real-time monitoring information for specific equipment by asking natural language questions. During this process, the system must be able to accurately capture changes in user intent and dynamically adjust query strategies to ensure that query results closely match user needs.

[0070] First, in the semantic extraction module, the user's question information is transmitted to the processing unit inside the module through the interface. The processing unit splits the question content into multiple feature words according to the preset word segmentation algorithm and identifies the target semantic features therein. The determination of the target semantic features is based on the power production-related terms and their semantic classification rules stored in the domain knowledge base. For example, when the user asks "Query the temperature change trend of the main transformer A-phase winding in the past week", the processing unit will extract feature words such as "main transformer", "A-phase winding", "temperature", and "change trend", and classify them as target semantic features. Subsequently, the processing unit calculates the semantic weight value W of the target semantic feature in the tth round of questions t,i , whose formula is W t,i = q ti / Q t Among them, q ti Indicates the number of feature words contained in the target semantic feature, Q tIndicates the total number of feature words in the current question. For example, if the current question contains 10 feature words and the target semantic feature contains 3 feature words, the semantic weight value is 0.3. This weight value is stored in the semantic weight database and can be used by subsequent modules.

[0071] Next, the offset management module extracts the semantic weight values of the same semantic feature in two adjacent rounds of questions from the semantic weight database and calculates the semantic change evaluation value R t The formula for semantic change evaluation is R t = (W t,i -W t-1,i ) / W t-1,i .

[0072] For example, if the weight values of a semantic feature in the fourth round of questions are 0.8, 0.75, 0.6, and 0.4, where the weight value in the tth round is 0.4, the semantic change assessment values are -0.0625, -0.2, and -0.33, respectively.

[0073] The offset management module generates a semantic change dataset based on historical data and uses a linear regression algorithm to fit and predict the semantic change prediction value R of the next round of questions. t+1,p , is -0.4525. At the same time, the module also calculates the similarity ω of the word vectors corresponding to all feature words in the first two rounds of questions t-1,t , whose value ranges from 0 to 1. The offset management module combines the semantic change prediction value R t+1,p Similarity with word vector ω t-1,t Calculate the semantic deviation threshold S t+1 ,

[0074] The formula is S t+1 = α × R t+1,p + (1-α) × ω t-1,t . Among them, α is the weight coefficient, which can be set to 0.6 according to actual needs. For example, if R t+1,p is -0.4525 and ω t-1,t is 0.6, then the semantic shift threshold S t+1 0.6 × (-0.4525) + 0.4 × 0.6 = -0.0315.

[0075] After completing the calculation of the semantic shift threshold, the shift management module obtains the true value R of the semantic change of the question in the t+1 round t+1,a , and with the semantic offset threshold S t+1 For comparison. t+1,a Greater than S t+1, then the target semantics are judged to have experienced a semantic shift and a semantic shift flag is generated. For example, if the weight of a semantic feature in the t+1 round of questions suddenly increases from 0.4 to 0.5, the true value of the semantic change is (0.5 - 0.4) / 0.4 = 0.25, which exceeds the semantic shift threshold of 0.2815. At this time, the shift management module generates a semantic shift flag and transmits it to the feature word screening module.

[0076] After receiving the semantic offset flag, the feature word screening module counts the word frequency TF of each feature word in the target semantics in the t+1 round of questions k and the number of rounds N that appeared in the first t+1 rounds of questions k The module sorts the feature words according to their evaluation values from high to low, and selects the top m feature words to form a feature word set. For example, if m is set to 3, the three feature words with the highest evaluation values are selected as the feature word set and transmitted to the feature word compensation module.

[0077] After receiving the feature word set, the feature word compensation module calculates the average value avg of the word vectors corresponding to all feature words in the t+1 round of questions. The module combines the semantic offset threshold S t+1 and semantic change true value R t+1,a Calculate the offset coefficient γ, the formula is γ = S t+1 / R t+1,a .

[0078] For example, if the semantic shift threshold is -0.0315 and the true value of semantic change is 0.25, the shift coefficient γ is (-0.0315) / 0.25 ≈ -0.126. The module calculates the corrected vector p for each feature word based on the shift coefficient γ and the average word vector avg k , whose formula is p k =0.5×(v k + (1-γ) × avg). Where, v k Represents the original word vector corresponding to the k-th feature word.

[0079] For example, if the original word vector of a feature word is [0.1, 0.2, 0.3] and the average word vector is [0.4, 0.5, 0.6], then its corrected vector is 0.5×([0.1, 0.2, 0.3] + 1.126×[0.17, 0.28, 0.36])=[0.145, 0.257, 0.352]. Fine-tuning of the corresponding original word vector has been achieved. The feature word compensation module generates a new feature word based on the corrected vector and outputs the new feature word to the user interface for user selection.

[0080] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. An intelligent data-driven rapid construction optimization system for power production scenarios, characterized by: The system includes: Word vector management module, question management module, semantic extraction module, semantic offset management module, semantic offset management module, feature word screening module and feature word compensation module; The word vector management module is used to collect historical records of continuous dialogue between the questioner and the agent, extract characteristic words from the question information, and establish a correspondence between characteristic words and word vectors through word vector tags; The question management module is used to collect the question records of the questioner to the agent in each round of dialogue scenarios, and compile the questions raised by the questioner into a question sequence; The semantic extraction module is used to classify the feature words in the question, use the classification results as semantic features, identify and label the semantic features, and obtain the feature words under each semantic feature; Semantic offset management module: used to calculate the change value of the semantic features in the question information, including the semantic weight value of the target semantic feature and the semantic change evaluation value of the same semantic feature in two consecutive rounds of questions; Semantic drift management module: used to calculate the semantic drift threshold and determine whether the questioner's question intention has a semantic drift based on the semantic drift threshold; The semantic shift management module includes: a semantic change value prediction unit, a shift threshold management unit and a shift judgment unit; The semantic change value prediction unit is used to calculate the semantic change prediction value, obtain the semantic change value of the target semantic feature in the nth round of questioning, n≥3, wherein the nth round of questioning ends with the tth round of questioning, and calculate the semantic change prediction value R of the target semantic feature in the t+1th round by function fitting. t+1,p ; The offset threshold management unit is used to calculate the semantic offset threshold, and calculate the semantic offset threshold S of the target semantics in the t+1 round of questioning. t+1 , S t+1 =α×R t+1，p +(1-α)×ω t-1，t , α represents the weight value, satisfying the condition: 0<α<1, where ω t-1,t Indicates the similarity between the word vectors of all feature words in the t-1th round of questions and the tth round of questions; The deviation judgment unit is used to judge whether the questioner's question intention is deviated and obtain the true value R of the semantic change of the question in the t+1 round. t+1,a , when R t+1,a >S t+1 When , it is judged that in the t+1 round of questions, the target semantics has a semantic shift; The feature word screening module is used to screen the feature words that need to be strengthened in the question content, collect the feature words that need to be strengthened, and obtain a feature word set; The feature word compensation module is used to calculate the vector offset and compensate the word vector to generate a new vector, and obtain the feature word corresponding to the new vector based on the correspondence between the vector and the feature word, and output the feature word to the questioner for selection.

2. The intelligent data rapid construction optimization system for power production scenarios according to claim 1 is characterized by: In the word vector management module, when collecting the historical records of the questioner in the continuous dialogue scenario with the intelligent agent, the feature words in the question information are analyzed through natural language processing technology, and the word embedding algorithm is used to convert the feature words into word vectors in a high-dimensional space, and a mapping relationship between the feature words and the word vectors is established.

3. The intelligent data rapid construction optimization system for power production scenarios according to claim 1 is characterized by: In the question management module, when collecting the question records of the questioner to the intelligent agent in each round, the question content is obtained through voice recognition or text input, and the question content is arranged in chronological order to form a question sequence; the question sequence is indexed by timestamp and supports retrieval and analysis of question distribution patterns in different time periods by time range.

4. The intelligent data rapid construction optimization system for power production scenarios according to claim 1 is characterized by: In the semantic extraction module, when classifying the feature words in the question, the feature words are divided into equipment status class, operation parameter class, fault diagnosis class and scheduling strategy class based on the domain knowledge base and semantic network model.

5. The intelligent data rapid construction optimization system for power production scenarios according to claim 1 is characterized by: The semantic offset management module includes: a semantic feature value calculation unit and a semantic change management unit; The semantic feature value calculation unit is used to calculate the i-th semantic feature in the t-th round of questions as the target semantic feature and calculate the semantic weight value W of the target semantic feature. t,i , Among them, q ti represents the number of feature words in the i-th semantic feature in the t-th round of questions, Q t Indicates the number of feature words included in the t-th round of questions; The semantic change management unit is used to calculate the semantic change evaluation value of the same semantic feature in two adjacent rounds of questions, and obtain the semantic weight value W of the target semantic feature in the t-1 round of questions. t-1,i , calculate the semantic change value R of the target semantics in the tth round of questions t , 6. The intelligent data rapid construction optimization system for power production scenarios according to claim 5 is characterized by: The feature word screening module includes: a word frequency acquisition unit, a word evaluation value management unit and a feature word set management unit; The word frequency acquisition unit is used to obtain the word frequency of each feature word in the target semantics in the t+1 round of questions, and the number of question rounds in which each feature word appears in the previous t+1 rounds. In the t+1 round of questions, the word frequency of the kth feature word in the target semantics in the t+1 round of questions is recorded as TF k In the first t+1 rounds of questions, the number of rounds in which the k-th feature word appears is recorded as N k ; The word evaluation value management unit is used to calculate the word evaluation value B of the kth feature word k , The feature word set management unit is used to arrange the feature words of the target semantics from high to low according to the word evaluation values, and put the first m feature words into the feature word set.

7. The intelligent data rapid construction optimization system for power production scenarios according to claim 6 is characterized by: The feature word compensation module includes: a vector offset calculation unit, an offset coefficient calculation unit, a vector correction unit and a feature word output unit; The vector offset calculation unit is used to calculate the vector offset, where the vector offset is the average value of the word vectors of the feature words included in all semantic features in the t+1th round of questions, recorded as avg; The offset coefficient calculation unit is used to calculate the offset coefficient γ, The vector correction unit is used to calculate the correction vector of the word vector, wherein the correction vector p of the k-th feature word is calculated k , where v k The word vector corresponding to the k-th feature word in the table; The feature word output unit is used to collect the feature words corresponding to the correction vector and output the feature vocabulary set corresponding to the correction vector to the questioner.

Images