Intelligent query system and method for power transmission and distribution production data based on natural language interactionTechnical Field
The invention relates to the technical field of knowledge maps, in particular to an intelligent query system and method for power transmission and distribution production data based on natural language interaction.
Background
Along with the continuous expansion of the power grid scale and the improvement of the informatization degree of the power system, massive data are generated in the power transmission and distribution production process, including equipment operation data, power grid topology data, maintenance records, fault reports and the like. The data are scattered in different systems and databases, the formats are various, and the traditional query mode is difficult to quickly and accurately acquire the required information. The power transmission and distribution business relates to a plurality of professional fields and links, and when a worker performs data query, a plurality of factors and conditions often need to be comprehensively considered. For example, a plurality of data sources such as a device account, real-time monitoring data, historical fault records and the like are required to be related to the running state of a specific type of device in a certain area, traditional query sentences are complicated to write, and the technical requirements on business personnel are high. With the continuous progress of artificial intelligence technologies such as natural language processing, knowledge graph, deep learning and the like, a new thought and method are provided for solving the problem of data query in power transmission and distribution production. NLP technology is capable of understanding human natural language, converting a user's query intent into computer-executable instructions.
However, when the user query intention is analyzed by using a natural language processing technology and the user data query process is simplified, when the user question is fuzzy and clear keywords are not available, the accuracy of intelligent query is greatly affected, and different databases in power transmission and distribution production data are difficult to communicate with each other, and when the user queries, the error of the generated result is larger.
Disclosure of Invention
The invention aims to provide an intelligent query system and method for power transmission and distribution production data based on natural language interaction, so as to solve the problems in the prior art.
In order to achieve the above purpose, the present invention provides the following technical solutions:
an intelligent query method for power transmission and distribution production data based on natural language interaction comprises the following steps:
S100, collecting a multi-source database in power transmission and distribution production, standardizing all data in the multi-source database, extracting three metadata of field names, data types and constraint conditions of the data in the multi-source database, constructing a basic metadata unit, calculating statistical characteristics and constructing a metadata base;
further, the specific steps of constructing the basic metadata unit and calculating the statistical characteristics are as follows:
S101, acquiring a multi-source database in power transmission and distribution production, wherein the multi-source database comprises a relational database, a time sequence database and a GIS space database, scanning data in the three databases, extracting field names, data types and constraint conditions of the data in the three databases by using an SQL query algorithm, and constructing a basic metadata unit Mi={namei,typei,constraintsi }, wherein Mi represents a basic metadata unit of the ith data, namei represents the field name of the extracted ith data, typei represents the data type of the extracted ith data, and constraintsi represents the constraint conditions of the extracted ith data;
S102, calculating the average value, variance and skewness of each datum, taking the average value, variance and skewness of the data as the statistical characteristics of each datum, constructing enhancement metadata as Mi*=Mi⋃{ui,σ2i,skewnessi by using the statistical characteristics, wherein Mi* represents the ith enhancement metadata, ui represents the average value of the ith data, sigma2i represents the variance of the ith data, skewnessi represents the skewness of the ith data, outputting the enhancement metadata of each datum, normalizing the basic metadata unit and the enhancement metadata, and combining to construct a metadata base.
By standardizing all data in the multi-source database, the data format and specification can be unified, data inconsistency and errors are reduced, and data accuracy and integrity are improved, so that a high-quality data base is provided for subsequent data analysis and application.
Metadata such as field names, data types and constraint conditions are extracted, and a metadata base is constructed, so that the structure and the attribute of the data are clearer and more definite, data management staff and developers can understand the data conveniently, maintenance, update and management of the data are performed, and meanwhile, quick starting of a new data user is facilitated.
The statistical features are calculated to provide a basis for deep analysis of the data, so that the user can be helped to know the distribution, trend and other features of the data, powerful support is provided for decision making, for example, in power transmission and distribution production, the rule of the running data of the equipment can be found through statistical analysis, and maintenance and fault prevention can be performed in advance.
S200, setting a hard rule and a soft rule for a metadata base, establishing semantic association across database fields by using the two rules, and constructing an initial dynamic data map by using data in the metadata base and the semantic association;
Further, the specific steps of constructing an initial dynamic data map by utilizing the data in the metadata base and semantic association are as follows:
S201, setting hard rules, namely searching field names in a metadata base by using a power field professional dictionary, traversing the field names in the metadata base, and when the professional terms which are the same as the field names in the metadata base are searched in the power field professional dictionary, marking the field names in the metadata base by using the data types and the attributes of the professional terms in the power field professional dictionary;
s202, setting a soft rule, namely, word segmentation is carried out on field names in a metadata base, the field names are converted into a vector format, cosine similarity of vectors with different field names in the metadata base is calculated, and the formula is as follows:
;
In the formula, Stext represents cosine similarity between the ith field name and the jth field name, vi represents the ith field name vector, vj represents the jth field name vector, the JS divergence is utilized to calculate distribution similarity between different field names, and the cosine similarity and the distribution similarity are combined to calculate comprehensive similarity between different field names, wherein the formula is as follows:
;
In the formula, Sij represents the comprehensive similarity of the ith field name and the jth field name, JS (pi||pj) represents the distribution similarity of the ith field name and the jth field name, a similarity threshold value Sy is empirically set, and when Sij is more than or equal to Sy, an association is established between the ith field name and the jth field name;
S203, combining the hard rules and the soft rules to construct the association of field names in different multi-source databases, setting priority levels, wherein the priority levels of the hard rules are larger than that of the soft rules, preferentially selecting the association of the hard rules when the association constructed for the hard rules and the soft rules of the same field name is different, taking metadata in different multi-source databases as nodes, and constructing an initial dynamic data map by taking the association in a hard rule association list and the association in a soft rule association matrix as edges.
The method has the advantages that the hard rules and the soft rules are set to establish semantic association of fields crossing the database, data originally scattered in different databases are associated, data islands are broken, fusion and sharing of the data are achieved, and the utilization value of the data is improved.
The initial dynamic data map can be constructed to intuitively display the relation between the data in a graphical mode, so that a user can conveniently and quickly understand the relation and dependence between the data, find potential modes and rules, and provide a more comprehensive view angle for problem analysis and decision making in power transmission and distribution production.
S300, calculating the relation strength between different nodes in the dynamic data map, and giving weight to the dynamic data map edges by using the relation strength;
further, the specific step of giving weight to the dynamic data map edge by using the relation strength is as follows:
s301, collecting working logs in power transmission and distribution production within the past 24 hours in real time, wherein the co-occurrence times of different field names represent the times of the simultaneous occurrence of the different field names in the working logs, calculating the distance between the different field names according to GIS data in a GIS database, and calculating the relation strength of each side in a dynamic data map, wherein the formula is as follows:
;
In the formula, Rij represents the relation strength of the edge between the ith node and the jth node, C (Ni,Nj) represents the co-occurrence times of the ith field name and the jth field name, and Lij represents the distance between the ith node and the jth node field name;
and assigning a weight to each side in the dynamic data spectrum by using the calculated relation strength, and outputting the dynamic data spectrum with the weight.
The relationship strength among different nodes in the dynamic data map is calculated, the edge weight is given, the important relationship among the data can be highlighted, so that a user can pay more attention to the key association when looking up the map, the efficiency and pertinence of data analysis are improved, and important equipment and data related to faults can be rapidly positioned in fault investigation.
S400, monitoring metadata increment, and updating the dynamic data map according to metadata increment setting updating strategies;
further, the specific steps of updating the dynamic data map according to the metadata increment setting updating strategy are as follows:
s401, three metadata in a basic metadata unit in a metadata base and statistical features in enhanced metadata are monitored in real time, and when field addition and statistical feature offset in the metadata base are monitored, the association and association strength update dynamic data map is recalculated in the dynamic data map.
The metadata increment is monitored, the dynamic data map is updated according to the increment setting updating strategy, the change of the data can be timely reflected, the timeliness of the data is guaranteed, the map is always consistent with the actual data, accurate information is provided for a user, and for example, in power transmission and distribution production, the real-time updating of equipment operation data can help operation and maintenance personnel to timely master the equipment state.
S500, inputting a user real-time query statement, extracting an entity in the query statement, judging ambiguity, mapping the entity to a dynamic data map, and formulating an active guiding strategy to obtain a user query intention;
Further, the specific steps for obtaining the user query intention are as follows:
S501, collecting real-time query sentences of a user, extracting query sentence entities, matching the query sentence entities in a dynamic data map by using field names to obtain corresponding map nodes, and calculating the ambiguity of the query sentences of the user, wherein the formula is as follows:
;
In the formula, fuzz represents the ambiguity of a user query statement, Tno represents a user query statement entity which is not found when being matched In a dynamic data map, inc represents the information entropy of the user query statement, and Tz represents the total number of nodes In the dynamic data map;
And S502, when judging that Fuzz > Fs, triggering the system to conduct active guidance, wherein Fs is an ambiguity threshold value, setting according to historical query experience, generating an active guidance statement by using a natural language processing algorithm to inquire a user, determining a user query target range according to map nodes successfully matched, and enumerating all targets in the user query target range by an enumeration method to determine the user query intention.
The entity in the query statement is extracted, the ambiguity judgment is carried out, the entity is mapped into the dynamic data map to formulate the active guiding strategy, the query intention of the user can be more accurately understood, the inaccuracy of the query result caused by unclear expression of the user or incomplete query statement is avoided, and the success rate of the query of the user is improved.
The active guiding strategy can help the user to express own demands more accurately, provide the query result which meets the user's expectations, reduce the query operation steps of the user, improve the query efficiency, improve the user experience and enhance the satisfaction degree of the user to the system.
S600, reasoning in the dynamic data map according to the query intention of the user to obtain a query result, and constructing a visual chart by using the query result to display the query result to the user.
Further, the specific steps of constructing a visual chart by using the query result to display the query result to the user are as follows:
S601, searching in a dynamic data map according to user query intention to obtain corresponding nodes, and extracting all user query intention nodes and metadata associated with the user query intention nodes from the current dynamic data map to form a data subgraph;
s602, constructing the generated user query result into a visual chart, and displaying the visual chart to a user.
The query result is obtained by reasoning in the dynamic data map according to the query intention of the user, and a visual chart is constructed by utilizing the query result and is displayed to the user, so that the data can be presented in an intuitive mode, the user can understand the query result more easily, and key information can be acquired quickly, so that decisions can be made more effectively, for example, in the power transmission and distribution production management, the visual chart can help management personnel to know the production running condition quickly, and make reasonable decisions.
The intelligent query system for the power transmission and distribution production data based on natural language interaction comprises a data collection module, a metadata base construction module, a dynamic data map module, a map updating module, a query statement analysis module and a result generation module;
the data collection module is used for collecting data in the multi-source database and standardizing all the data in the multi-source database;
The metadata base construction module is used for extracting three metadata of field names, data types and constraint conditions of data in the multi-source database, constructing a basic metadata unit, calculating statistical characteristics and constructing a metadata base;
The dynamic data map module is used for setting a hard rule and a soft rule for a metadata base, establishing semantic association across database fields by utilizing the two rules, constructing an initial dynamic data map by utilizing data in the metadata base and the semantic association, and giving weight to map edges;
the map updating module is used for monitoring the metadata increment and updating the dynamic data map according to the metadata increment setting updating strategy;
the query statement analysis module is used for inputting a user real-time query statement, extracting an entity in the query statement, judging ambiguity, mapping the entity into a dynamic data map, and formulating an active guiding strategy to obtain a user query intention;
the result generation module is used for carrying out reasoning in the dynamic data map according to the query intention of the user to obtain a query result, and constructing a visual chart by utilizing the query result to display the query result to the user.
The dynamic data map module comprises a rule setting unit and a weight calculating unit;
The rule setting unit is used for setting hard rules and soft rules for the metadata base, and establishing semantic association of fields of the cross-database by utilizing the two rules;
The weight calculation unit is used for calculating the relation strength between different nodes in the dynamic data map, and the relation strength is utilized to give weight to the dynamic data map edges.
The query statement analysis module comprises a fuzzy judgment unit and an active guiding unit;
The fuzzy judgment unit is used for calculating the ambiguity of the user real-time query statement and judging whether the ambiguity is larger than an ambiguity threshold value or not to obtain whether the user real-time query statement is ambiguous or not;
The active guiding unit is used for generating an active guiding sentence to inquire the user by using a natural language processing algorithm after judging that the real-time inquiry sentence of the user is fuzzy, so as to obtain the inquiry intention.
Compared with the prior art, the invention has the beneficial effects that:
1. The invention sets the hard rules and the soft rules to establish semantic association of fields crossing the database, and links the data originally scattered in different databases, thereby breaking the data island, realizing the fusion and sharing of the data and improving the utilization value of the data.
2. The invention carries out deep reasoning on the user query intention through the dynamic data map, obtains and outputs the result of the user query, provides the query result which is more in line with the user expectation, can reduce the query operation steps of the user and improves the query efficiency.
3. According to the invention, the entity in the query statement is extracted, the ambiguity judgment is carried out, and the entity is mapped into the dynamic data map to formulate the active guiding strategy, so that the query intention of the user can be more accurately understood, the inaccurate query result caused by unclear user expression or incomplete query statement is avoided, and the success rate of the user query is improved.
Drawings
FIG. 1 is a block diagram of an intelligent query system for power transmission and distribution production data based on natural language interaction;
Fig. 2 is a schematic diagram of steps of the intelligent query method for power transmission and distribution production data based on natural language interaction.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Embodiment as shown in fig. 1-2, the present invention provides a technical solution,
An intelligent query method for power transmission and distribution production data based on natural language interaction comprises the following steps:
S100, collecting a multi-source database in power transmission and distribution production, standardizing all data in the multi-source database, extracting three metadata of field names, data types and constraint conditions of the data in the multi-source database, constructing a basic metadata unit, calculating statistical characteristics and constructing a metadata base;
the specific steps of constructing a basic metadata unit and calculating statistical characteristics are as follows:
S101, acquiring a multi-source database in power transmission and distribution production, wherein the multi-source database comprises a relational database, a time sequence database and a GIS space database, scanning data in the three databases, extracting field names, data types and constraint conditions of the data in the three databases by using an SQL query algorithm, and constructing a basic metadata unit Mi={namei,typei,constraintsi }, wherein Mi represents a basic metadata unit of the ith data, namei represents the field name of the extracted ith data, typei represents the data type of the extracted ith data, and constraintsi represents the constraint conditions of the extracted ith data;
S102, calculating the average value, variance and skewness of each datum, taking the average value, variance and skewness of the data as the statistical characteristics of each datum, constructing enhancement metadata as Mi*=Mi⋃{ui,σ2i,skewnessi by using the statistical characteristics, wherein Mi* represents the ith enhancement metadata, ui represents the average value of the ith data, sigma2i represents the variance of the ith data, skewnessi represents the skewness of the ith data, outputting the enhancement metadata of each datum, normalizing the basic metadata unit and the enhancement metadata, and combining to construct a metadata base.
By standardizing all data in the multi-source database, the data format and specification can be unified, data inconsistency and errors are reduced, and data accuracy and integrity are improved, so that a high-quality data base is provided for subsequent data analysis and application.
Metadata such as field names, data types and constraint conditions are extracted, and a metadata base is constructed, so that the structure and the attribute of the data are clearer and more definite, data management staff and developers can understand the data conveniently, maintenance, update and management of the data are performed, and meanwhile, quick starting of a new data user is facilitated.
The statistical features are calculated to provide a basis for deep analysis of the data, so that the user can be helped to know the distribution, trend and other features of the data, powerful support is provided for decision making, for example, in power transmission and distribution production, the rule of the running data of the equipment can be found through statistical analysis, and maintenance and fault prevention can be performed in advance.
S200, setting a hard rule and a soft rule for a metadata base, establishing semantic association across database fields by using the two rules, and constructing an initial dynamic data map by using data in the metadata base and the semantic association;
the specific steps of constructing an initial dynamic data map by utilizing data in a metadata base and semantic association are as follows:
S201, setting hard rules, namely searching field names in a metadata base by using a power field professional dictionary, traversing the field names in the metadata base, and when the professional terms which are the same as the field names in the metadata base are searched in the power field professional dictionary, marking the field names in the metadata base by using the data types and the attributes of the professional terms in the power field professional dictionary;
s202, setting a soft rule, namely, word segmentation is carried out on field names in a metadata base, the field names are converted into a vector format, cosine similarity of vectors with different field names in the metadata base is calculated, and the formula is as follows:
;
In the formula, Stext represents cosine similarity between the ith field name and the jth field name, vi represents the ith field name vector, vj represents the jth field name vector, the JS divergence is utilized to calculate distribution similarity between different field names, and the cosine similarity and the distribution similarity are combined to calculate comprehensive similarity between different field names, wherein the formula is as follows:
;
In the formula, Sij represents the comprehensive similarity of the ith field name and the jth field name, JS (pi||pj) represents the distribution similarity of the ith field name and the jth field name, a similarity threshold value Sy is empirically set, and when Sij is more than or equal to Sy, an association is established between the ith field name and the jth field name;
S203, combining the hard rules and the soft rules to construct the association of field names in different multi-source databases, setting priority levels, wherein the priority levels of the hard rules are larger than that of the soft rules, preferentially selecting the association of the hard rules when the association constructed for the hard rules and the soft rules of the same field name is different, taking metadata in different multi-source databases as nodes, and constructing an initial dynamic data map by taking the association in a hard rule association list and the association in a soft rule association matrix as edges.
The method has the advantages that the hard rules and the soft rules are set to establish semantic association of fields crossing the database, data originally scattered in different databases are associated, data islands are broken, fusion and sharing of the data are achieved, and the utilization value of the data is improved.
The initial dynamic data map can be constructed to intuitively display the relation between the data in a graphical mode, so that a user can conveniently and quickly understand the relation and dependence between the data, find potential modes and rules, and provide a more comprehensive view angle for problem analysis and decision making in power transmission and distribution production.
S300, calculating the relation strength between different nodes in the dynamic data map, and giving weight to the dynamic data map edges by using the relation strength;
The specific steps of giving weight to the dynamic data map edge by using the relation strength are as follows:
s301, collecting working logs in power transmission and distribution production within the past 24 hours in real time, wherein the co-occurrence times of different field names represent the times of the simultaneous occurrence of the different field names in the working logs, calculating the distance between the different field names according to GIS data in a GIS database, and calculating the relation strength of each side in a dynamic data map, wherein the formula is as follows:
;
In the formula, Rij represents the relation strength of the edge between the ith node and the jth node, C (Ni,Nj) represents the co-occurrence times of the ith field name and the jth field name, and Lij represents the distance between the ith node and the jth node field name;
and assigning a weight to each side in the dynamic data spectrum by using the calculated relation strength, and outputting the dynamic data spectrum with the weight.
The relationship strength among different nodes in the dynamic data map is calculated, the edge weight is given, the important relationship among the data can be highlighted, so that a user can pay more attention to the key association when looking up the map, the efficiency and pertinence of data analysis are improved, and important equipment and data related to faults can be rapidly positioned in fault investigation.
S400, monitoring metadata increment, and updating the dynamic data map according to metadata increment setting updating strategies;
the method for updating the dynamic data map according to the metadata increment setting updating strategy comprises the following specific steps:
s401, three metadata in a basic metadata unit in a metadata base and statistical features in enhanced metadata are monitored in real time, and when field addition and statistical feature offset in the metadata base are monitored, the association and association strength update dynamic data map is recalculated in the dynamic data map.
The metadata increment is monitored, the dynamic data map is updated according to the increment setting updating strategy, the change of the data can be timely reflected, the timeliness of the data is guaranteed, the map is always consistent with the actual data, accurate information is provided for a user, and for example, in power transmission and distribution production, the real-time updating of equipment operation data can help operation and maintenance personnel to timely master the equipment state.
S500, inputting a user real-time query statement, extracting an entity in the query statement, judging ambiguity, mapping the entity to a dynamic data map, and formulating an active guiding strategy to obtain a user query intention;
the specific steps for obtaining the user query intention are as follows:
S501, collecting real-time query sentences of a user, extracting query sentence entities, matching the query sentence entities in a dynamic data map by using field names to obtain corresponding map nodes, and calculating the ambiguity of the query sentences of the user, wherein the formula is as follows:
;
In the formula, fuzz represents the ambiguity of a user query statement, Tno represents a user query statement entity which is not found when being matched In a dynamic data map, inc represents the information entropy of the user query statement, and Tz represents the total number of nodes In the dynamic data map;
And S502, when judging that Fuzz > Fs, triggering the system to conduct active guidance, wherein Fs is an ambiguity threshold value, setting according to historical query experience, generating an active guidance statement by using a natural language processing algorithm to inquire a user, determining a user query target range according to map nodes successfully matched, and enumerating all targets in the user query target range by an enumeration method to determine the user query intention.
The entity in the query statement is extracted, the ambiguity judgment is carried out, the entity is mapped into the dynamic data map to formulate the active guiding strategy, the query intention of the user can be more accurately understood, the inaccuracy of the query result caused by unclear expression of the user or incomplete query statement is avoided, and the success rate of the query of the user is improved.
The active guiding strategy can help the user to express own demands more accurately, provide the query result which meets the user's expectations, reduce the query operation steps of the user, improve the query efficiency, improve the user experience and enhance the satisfaction degree of the user to the system.
S600, reasoning in the dynamic data map according to the query intention of the user to obtain a query result, and constructing a visual chart by using the query result to display the query result to the user.
The specific steps of constructing a visual chart by using the query result to display the query result to the user are as follows:
S601, searching in a dynamic data map according to user query intention to obtain corresponding nodes, and extracting all user query intention nodes and metadata associated with the user query intention nodes from the current dynamic data map to form a data subgraph;
s602, constructing the generated user query result into a visual chart, and displaying the visual chart to a user.
The query result is obtained by reasoning in the dynamic data map according to the query intention of the user, and a visual chart is constructed by utilizing the query result and is displayed to the user, so that the data can be presented in an intuitive mode, the user can understand the query result more easily, and key information can be acquired quickly, so that decisions can be made more effectively, for example, in the power transmission and distribution production management, the visual chart can help management personnel to know the production running condition quickly, and make reasonable decisions.
The intelligent query system for the power transmission and distribution production data based on natural language interaction comprises a data collection module, a metadata base construction module, a dynamic data map module, a map updating module, a query statement analysis module and a result generation module;
the data collection module is used for collecting data in the multi-source database and standardizing all the data in the multi-source database;
The metadata base construction module is used for extracting three metadata of field names, data types and constraint conditions of data in the multi-source database, constructing a basic metadata unit, calculating statistical characteristics and constructing a metadata base;
The dynamic data map module is used for setting a hard rule and a soft rule for a metadata base, establishing semantic association across database fields by utilizing the two rules, constructing an initial dynamic data map by utilizing data in the metadata base and the semantic association, and giving weight to map edges;
the map updating module is used for monitoring the metadata increment and updating the dynamic data map according to the metadata increment setting updating strategy;
the query statement analysis module is used for inputting a user real-time query statement, extracting an entity in the query statement, judging ambiguity, mapping the entity into a dynamic data map, and formulating an active guiding strategy to obtain a user query intention;
the result generation module is used for carrying out reasoning in the dynamic data map according to the query intention of the user to obtain a query result, and constructing a visual chart by utilizing the query result to display the query result to the user.
The dynamic data map module comprises a rule setting unit and a weight calculating unit;
The rule setting unit is used for setting hard rules and soft rules for the metadata base, and establishing semantic association of fields of the cross-database by utilizing the two rules;
The weight calculation unit is used for calculating the relation strength between different nodes in the dynamic data map, and the relation strength is utilized to give weight to the dynamic data map edges.
The query statement analysis module comprises a fuzzy judgment unit and an active guiding unit;
The fuzzy judgment unit is used for calculating the ambiguity of the user real-time query statement and judging whether the ambiguity is larger than an ambiguity threshold value or not to obtain whether the user real-time query statement is ambiguous or not;
The active guiding unit is used for generating an active guiding sentence to inquire the user by using a natural language processing algorithm after judging that the real-time inquiry sentence of the user is fuzzy, so as to obtain the inquiry intention.
Embodiment 1 extraction and association of transformer temperature fields;
the field temp, type being flow, is extracted from the transducer table of InfluxDB, constrained to be non-null.
Calculating statistical characteristics, namely, an average value of 85.3 ℃, a variance of 4.1 and a skewness of 2, and normalizing field names into a transducer_temp;
And calculating the similarity by using a soft rule to obtain the comprehensive similarity of the field names of the device_temp with the relation library as 0.72, setting the similarity threshold as 0.6, and judging that the association exists between the field names of the device_temp and the device_temp.
Embodiment 2, user query statement is "which is the most recently failed region;
Matching keywords such as 'fault', 'region' and the like to map nodes, calculating the ambiguity as 0.64, setting the ambiguity threshold as 0.5, and judging the query ambiguity of a user;
Starting active guidance, obtaining a user inquiry target range of 'Huadong, certain city, one month, one week and the like' according to user inquiry map nodes, generating an active guidance problem of 'please select area range of A, huadong district of B, certain city power distribution network' by using a natural language processing algorithm, sequentially inquiring by using an enumeration method, and obtaining the user inquiry intention of 'counting the past 7 days of failure times of the Huadong district'.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.