Weather bulletin material automatic generation method, apparatus and device, and storage medium

Abstract

The invention discloses a weather bulletin material automatic generation method and device, equipment and a storage medium, and the method comprises the steps: obtaining weather forecast live data and a bulletin type of a national weather bulletin, processing the weather forecast live data through a data service interface according to the bulletin type, and obtaining weather information; analyzing the meteorological information according to an analysis service interface to obtain meteorological service text data meeting a preset specification; and inputting the meteorological service text data into an automatic bulletin generation service model, and outputting a comprehensive bulletin material through a material organization interface, so that the compiling efficiency of the service material can be effectively improved, the automatically generated forecast text is correct in content, smooth in statement and correct in semantics and grammar, and the text style conforms to the writing habit of a predictor. The correctness, consistency and normalization of the document are guaranteed, errors possibly caused by manual operation are avoided, and the automatic generation speed and efficiency of the weather bulletin material are improved.

Description

Technical Field

[0001] The present invention relates to the field of meteorological service technology, and in particular to a method, device, equipment and storage medium for automatically generating weather bulletin materials. Background Art

[0002] The national meteorological bulletin refers to the various weather bulletins currently issued daily by the National Meteorological Center (Central Meteorological Observatory). It is the authoritative release of forecast and warning service materials for various types of disastrous weather such as typhoons, heavy rains, cold waves, high temperatures, sandstorms, and heavy fog on the official website of the Central Meteorological Observatory. The service materials are mainly described in text language, supplemented by the basic writing principles of graphics or tables.

[0003] Although some research has been done on the automatic generation of meteorological text service materials at home and abroad, a systematic, real-time business system that can be applied in business has not yet been formed. At present, weather bulletins are still mainly written manually by forecasters. Generally, data or pictures are opened through the system platform, and the data and pictures are compared to analyze the weather that needs to be described in detail in each zone, and then they are compiled into documents. This business process is inefficient, especially in complex weather conditions, when timeliness requirements are often not met. In addition, different forecasters have different habits, which can easily lead to deviations in text content. Summary of the invention

[0004] The main purpose of the present invention is to provide a method, device, equipment and storage medium for automatically generating weather bulletin materials, aiming to solve the technical problems in the prior art that weather bulletins are mainly written manually, such business process is inefficient, especially in complex weather conditions, the timeliness is poor, and the text content is prone to deviations, lack of accuracy and standardization.

[0005] In a first aspect, the present invention provides a method for automatically generating weather bulletin materials, the method comprising the following steps: Acquire live weather forecast data and the bulletin type of the national weather bulletin, and process the live weather forecast data through a data service interface according to the bulletin type to obtain weather information; Analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The meteorological service text data is input into the bulletin automatic generation service model, and the comprehensive bulletin materials are output through the material organization interface.

[0006] Optionally, the acquiring of the weather forecast live data and the bulletin type of the national weather bulletin, and processing the weather forecast live data through a data service interface according to the bulletin type to obtain the weather information includes: Obtaining subjective and objective weather forecast data and real-time weather observation data, and using the subjective and objective weather forecast data and the real-time weather observation data as weather forecast real-time data; Get the bulletin type of the national weather bulletin; According to the bulletin type, the weather forecast live data is converted into vectorized point, line and surface data stored in a GIS standard format through a data service interface; Perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

[0007] Optionally, the converting of the weather forecast live data into vectorized point, line and surface data stored in a GIS standard format through a data service interface according to the bulletin type includes: According to the bulletin type, the weather forecast live data is converted into grid forecast data, fall zone forecast and warning data and target live data of various meteorological elements stored in a GIS standard format through a data service interface; According to the forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids, the grid forecast data, the fall area forecast and warning data and the target actual data are converted to obtain vectorized point, line and surface data.

[0008] Optionally, performing data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes includes: The spatial topology and attribute information corresponding to the vectorized point, line and surface data are obtained, a data topological relationship between the spatial topology and the attribute information is established, and meteorological information with topological attributes is obtained.

[0009] Optionally, analyzing the meteorological information according to the analysis service interface to obtain meteorological service text data that meets preset specifications includes: According to the analysis service interface, the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information are clipped, intersected, superimposed and merged to obtain meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area.

[0010] Optionally, the step of inputting the meteorological service text data into a bulletin automatic generation service model and outputting comprehensive bulletin materials through a material organization interface includes: Acquire historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning; Using a preset neural network model and word segmentation system to conduct deep learning on the target historical text service material, and obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments; Perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship combined with season and scope size to obtain a bulletin automatic generation service model; The meteorological range text data is input into the bulletin automatic generation service model, and the comprehensive bulletin material is output through the material organization interface.

[0011] Optionally, the step of inputting the meteorological range text data into the bulletin automatic generation service model and outputting comprehensive bulletin materials through a material organization interface includes: Perform statistics on historical bulletin documents, create a template document, and fill the variable fields in the template document with actual data through Aspose.Words components according to preset stable setting instructions to obtain a preset bulletin document; Inputting the meteorological range text data into the bulletin automatic generation service model, and outputting comprehensive bulletin materials conforming to the preset bulletin document format through the material organization interface; Upon receiving the subjective revision opinion, the comprehensive communiqué material is subjectively revised to obtain the communiqué material after subjective revision, and the communiqué material after subjective revision is compared with the comprehensive communiqué material before subjective revision by using the Jaccard similarity coefficient to obtain the Jaccard distance; When the Jaccard distance is less than a preset distance, the subjectively revised communiqué material is used as the updated comprehensive communiqué material.

[0012] In a second aspect, in order to achieve the above-mentioned purpose, the present invention further provides a device for automatically generating weather bulletin materials, the device for automatically generating weather bulletin materials comprising: A data processing module, used to obtain weather forecast live data and the bulletin type of the national weather bulletin, and process the weather forecast live data through the data service interface according to the bulletin type to obtain weather information; An information analysis module, used to analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The material generation module is used to input the meteorological service text data into the bulletin automatic generation service model and output the comprehensive bulletin material through the material organization interface.

[0013] In the third aspect, in order to achieve the above-mentioned purpose, the present invention also proposes a device for automatically generating weather bulletin materials, which includes: a memory, a processor, and a weather bulletin material automatic generation program stored in the memory and executable on the processor, and the weather bulletin material automatic generation program is configured to implement the steps of the weather bulletin material automatic generation method as described above.

[0014] In a fourth aspect, in order to achieve the above-mentioned purpose, the present invention also proposes a storage medium, on which a program for automatically generating weather bulletin materials is stored, and when the program for automatically generating weather bulletin materials is executed by a processor, the steps of the method for automatically generating weather bulletin materials as described above are implemented.

[0015] The method for automatically generating weather bulletin materials proposed in the present invention obtains actual weather forecast data and the bulletin type of the national weather bulletin, processes the actual weather forecast data through a data service interface according to the bulletin type, and obtains meteorological information; analyzes the meteorological information according to an analysis service interface, and obtains meteorological service text data that meets preset specifications; inputs the meteorological service text data into an automatic bulletin generation service model, and outputs comprehensive bulletin materials through a material organization interface, which can effectively improve the writing efficiency of service materials, and the automatically generated forecast text has correct content and is consistent with the original data, without omitting important meteorological elements. Compared with the forecast text manually written by the forecaster, the automatically generated forecast text has a high degree of consistency, and the sentences of the automatically generated forecast text are fluent, semantically and grammatically correct, and the text style conforms to the writing habits of the forecaster, thereby ensuring the correctness, consistency and standardization of the document, avoiding errors that may be caused by manual operation, and improving the speed and efficiency of the automatic generation of weather bulletin materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention; Figure 2 A schematic diagram of the flow chart of the first embodiment of the method for automatically generating weather bulletin materials according to the present invention; Figure 3 It is a schematic diagram of the overall logic in the method for automatically generating weather bulletin materials of the present invention; Figure 4 A schematic diagram of the flow chart of the second embodiment of the method for automatically generating weather bulletin materials according to the present invention; Figure 5 A schematic diagram of meteorological data conversion in the method for automatically generating weather bulletin materials of the present invention; Figure 6 A schematic diagram of a flow chart of a third embodiment of a method for automatically generating weather bulletin materials according to the present invention; Figure 7 A schematic diagram of a fourth embodiment of the method for automatically generating weather bulletin materials according to the present invention; Figure 8 A schematic diagram of describing keywords in the method for automatically generating weather bulletin materials of the present invention; Fig. 9 A schematic diagram of generating a bulletin text service model in the method for automatically generating weather bulletin materials of the present invention; Fig.10A schematic diagram of text planning in the method for automatically generating weather bulletin materials of the present invention; Fig.11 A schematic diagram of text verification and evaluation in the method for automatically generating weather bulletin materials of the present invention; Fig.12 This is a functional module diagram of the first embodiment of the automatic generation device for weather bulletin materials of the present invention.

[0017] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] The solution of the embodiment of the present invention is mainly as follows: by obtaining the weather forecast live data and the bulletin type of the national weather bulletin, the weather forecast live data is processed through the data service interface according to the bulletin type to obtain weather information; the weather information is analyzed according to the analysis service interface to obtain weather service text data that meets the preset specifications; the weather service text data is input into the bulletin automatic generation service model, and the comprehensive bulletin material is output through the material organization interface, which can effectively improve the writing efficiency of the service material, the automatically generated forecast text content is correct, consistent with the original data, no important meteorological elements are omitted, and the forecast text manually written by the forecaster has a high degree of consistency. The automatically generated forecast text has smooth sentences, correct semantics and grammar, and the text style conforms to the forecaster's writing habits, which ensures the correctness, consistency and standardization of the document, avoids errors that may be caused by manual operation, improves the speed and efficiency of the automatic generation of weather bulletin materials, and solves the problem that the weather bulletin in the prior art is mainly written by humans, the business process is inefficient, especially the timeliness is poor under complex weather conditions, and it is easy to have text content deviations, insufficient accuracy and standardization.

[0020] Reference Figure 1 , Figure 1 The figure is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0021] like Figure 1As shown, the device may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed RAM memory, or a stable memory (Non-Volatile Memory), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0022] Those skilled in the art will understand that Figure 1 The device structure shown in the figure does not constitute a limitation of the device, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0023] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating device, a network communication module, a user interface module, and a weather bulletin material automatic generation program.

[0024] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and performs the following operations: Acquire live weather forecast data and the bulletin type of the national weather bulletin, and process the live weather forecast data through a data service interface according to the bulletin type to obtain weather information; Analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The meteorological service text data is input into the bulletin automatic generation service model, and the comprehensive bulletin materials are output through the material organization interface.

[0025] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: Obtaining subjective and objective weather forecast data and real-time weather observation data, and using the subjective and objective weather forecast data and the real-time weather observation data as weather forecast real-time data; Get the bulletin type of the national weather bulletin; According to the bulletin type, the weather forecast live data is converted into vectorized point, line and surface data stored in a GIS standard format through a data service interface; Perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

[0026] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: According to the bulletin type, the weather forecast live data is converted into grid forecast data, fall zone forecast and warning data and target live data of various meteorological elements stored in a GIS standard format through a data service interface; According to the forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids, the grid forecast data, the fall area forecast and warning data and the target actual data are converted to obtain vectorized point, line and surface data.

[0027] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: The spatial topology and attribute information corresponding to the vectorized point, line and surface data are obtained, a data topological relationship between the spatial topology and the attribute information is established, and meteorological information with topological attributes is obtained.

[0028] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: According to the analysis service interface, the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information are clipped, intersected, superimposed and merged to obtain meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area.

[0029] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: Acquire historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning; Using a preset neural network model and word segmentation system to conduct deep learning on the target historical text service material, and obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments; Perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship combined with season and scope size to obtain a bulletin automatic generation service model; The meteorological range text data is input into the bulletin automatic generation service model, and the comprehensive bulletin material is output through the material organization interface.

[0030] The device of the present invention calls the weather bulletin material automatic generation program stored in the memory 1005 through the processor 1001, and also performs the following operations: Perform statistics on historical bulletin documents, create a template document, and fill the variable fields in the template document with actual data through Aspose.Words components according to preset stable setting instructions to obtain a preset bulletin document; Inputting the meteorological range text data into the bulletin automatic generation service model, and outputting comprehensive bulletin materials conforming to the preset bulletin document format through the material organization interface; Upon receiving the subjective revision opinion, the comprehensive communiqué material is subjectively revised to obtain the communiqué material after subjective revision, and the communiqué material after subjective revision is compared with the comprehensive communiqué material before subjective revision by using the Jaccard similarity coefficient to obtain the Jaccard distance; When the Jaccard distance is less than a preset distance, the subjectively revised communiqué material is used as the updated comprehensive communiqué material.

[0031] This embodiment adopts the above scheme to obtain the actual weather forecast data and the bulletin type of the national weather bulletin, and processes the actual weather forecast data through the data service interface according to the bulletin type to obtain meteorological information; analyzes the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; inputs the meteorological service text data into the bulletin automatic generation service model, and outputs the comprehensive bulletin materials through the material organization interface, which can effectively improve the efficiency of writing service materials. The automatically generated forecast text has correct content and is consistent with the original data, without missing any important meteorological elements. Compared with the forecast text manually written by the forecaster, the automatically generated forecast text has a high degree of consistency, and the sentences are fluent, the semantics and grammar are correct, and the text style conforms to the forecaster's writing habits, which ensures the correctness, consistency and standardization of the document, avoids errors that may be caused by manual operation, and improves the speed and efficiency of automatic generation of weather bulletin materials.

[0032] Based on the above hardware structure, an embodiment of the method for automatically generating weather bulletin materials of the present invention is proposed.

[0033] Reference Figure 2 , Figure 2 It is a flow chart of the first embodiment of the method for automatically generating weather bulletin materials of the present invention.

[0034] In a first embodiment, the method for automatically generating weather bulletin materials comprises the following steps: Step S10: Acquire the weather forecast live data and the bulletin type of the national weather bulletin, and process the weather forecast live data through the data service interface according to the bulletin type to obtain weather information.

[0035] It should be noted that the weather forecast live data is the current real-time data of the weather forecast. The types of bulletins of the national weather bulletin include short-term weather forecast bulletins, for example: forecast information on weather conditions within the next 1-3 days, generally including specific weather phenomena, such as sunny, cloudy, rainfall (light rain, moderate rain, heavy rain, etc.), snowfall (light snow, moderate snow, heavy snow), thunderstorms, etc.; the approximate range of temperature, such as the highest temperature and the lowest temperature; elements such as wind direction and wind force; it can also include: medium-term weather forecast bulletins, for example: weather trend forecasts usually covering the next 4-10 days; it can also include: long-term weather forecast bulletins, for example: weather outlook involving the next month or even a quarter; of course, it can also include other types of weather bulletins, such as: disastrous weather warning bulletins, the content of which generally focuses on disastrous weather such as typhoons, heavy rains, blizzards, cold waves, strong winds, sandstorms, high temperatures, droughts, lightning, hail, frost, and fog; it can also be a climate monitoring bulletin, an environmental meteorological bulletin, etc., which is not limited in this embodiment.

[0036] It is understandable that after obtaining the live weather forecast data and the bulletin type of the national weather bulletin, the live weather forecast data can be processed through the data service interface according to the bulletin type to obtain meteorological information.

[0037] Step S20: Analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets preset specifications.

[0038] It should be understood that the meteorological information can be analyzed through the analysis service interface to obtain meteorological service text data that meets pre-set preset specifications, that is, to form standardized and structured text data.

[0039] Step S30: input the meteorological service text data into the bulletin automatic generation service model, and output the comprehensive bulletin material through the material organization interface.

[0040] It is understandable that after the meteorological service text data is input into the preset bulletin automatic generation service model, the comprehensive bulletin material can be output through the material organization interface.

[0041] In the specific implementation, see Figure 3 , Figure 3 FIG. 1 is a schematic diagram of the overall logic of the method for automatically generating weather bulletin materials of the present invention, such as Figure 3As shown, the bulletin type is selected to process the actual data of weather forecast, meteorological information analysis and mining is carried out through the data service interface, analysis is carried out through the analysis service interface, and comprehensive bulletin materials are output through the bulletin automatic generation service model and material organization interface. Generally, the system can be composed of four modules: text service material automatic generation model building module, meteorological data real-time processing module, comprehensive service material automatic organization module, and quantitative objective inspection and evaluation module. It can realize the automatic generation and distribution of various meteorological bulletin service materials, and the corresponding data interfaces between modules are connected. Each module is encapsulated in the form of microservices. It is only necessary to ensure that the standards of the input and output interfaces are consistent. The upgrade of a single algorithm module can be realized without affecting the architecture and operation of the overall system.

[0042] This embodiment adopts the above scheme to obtain the actual weather forecast data and the bulletin type of the national weather bulletin, and processes the actual weather forecast data through the data service interface according to the bulletin type to obtain meteorological information; analyzes the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; inputs the meteorological service text data into the bulletin automatic generation service model, and outputs the comprehensive bulletin materials through the material organization interface, which can effectively improve the efficiency of writing service materials. The automatically generated forecast text has correct content and is consistent with the original data, without missing any important meteorological elements. Compared with the forecast text manually written by the forecaster, the automatically generated forecast text has a high degree of consistency, and the sentences are fluent, the semantics and grammar are correct, and the text style conforms to the forecaster's writing habits, which ensures the correctness, consistency and standardization of the document, avoids errors that may be caused by manual operation, and improves the speed and efficiency of automatic generation of weather bulletin materials.

[0043] Further, Figure 4 FIG. 1 is a flow chart of a second embodiment of the method for automatically generating weather bulletin materials according to the present invention. Figure 4 As shown, a second embodiment of the method for automatically generating weather bulletin materials of the present invention is proposed based on the first embodiment. In this embodiment, the step S10 specifically includes the following steps: Step S11, obtaining subjective and objective weather forecast data and actual weather observation data, and using the subjective and objective weather forecast data and the actual weather observation data as actual weather forecast data.

[0044] It should be noted that the text description information sources in weather bulletins usually have two main aspects: one is the real-time information; the other is the subjective and objective forecast data. After obtaining the subjective and objective weather forecast data and the real-time weather observation data, the subjective and objective weather forecast data and the real-time weather observation data can be used as the real-time weather forecast data.

[0045] Step S12: Obtain the bulletin type of the national weather bulletin.

[0046] Understandably, there are many different types of national weather bulletins.

[0047] Step S13: converting the weather forecast live data into vectorized point, line and surface data stored in a GIS standard format through a data service interface according to the bulletin type.

[0048] It should be understood that various forms of raw meteorological data can be converted into meteorological description data in the form of structured text, that is, according to the type of the bulletin, the actual meteorological forecast data can be converted into vectorized point, line and surface data stored in the standard format of the Geographic Information System (GIS) through the data service interface.

[0049] Furthermore, the step S13 specifically includes the following steps: According to the bulletin type, the weather forecast live data is converted into grid forecast data, fall zone forecast and warning data and target live data of various meteorological elements stored in a GIS standard format through a data service interface; According to the forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids, the grid forecast data, the fall area forecast and warning data and the target actual data are converted to obtain vectorized point, line and surface data.

[0050] In the specific implementation, see Figure 5 , Figure 5 Schematic diagram of meteorological data conversion in the method for automatically generating weather bulletin materials of the present invention, such as Figure 5 As shown, the data is vectorized, and various formats of subjective and objective forecasts and actual observation data are converted into point, line, surface, and raster data stored in the GIS standard format. The grid forecast data of various meteorological elements, the forecast and warning data of the falling area, and the actual data are converted into corresponding vectorized point, line, and surface data in accordance with the different formats of the data sets through a variety of processing methods such as forecast points, forecast lines, forecast surfaces, forecast grids, observation sites, and observation grids.

[0051] Step S14: Perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

[0052] It can be understood that after performing data topology on the vectorized point, line and surface data, meteorological information with topological properties can be obtained.

[0053] Furthermore, the step S14 specifically includes the following steps: The spatial topology and attribute information corresponding to the vectorized point, line and surface data are obtained, a data topological relationship between the spatial topology and the attribute information is established, and meteorological information with topological attributes is obtained.

[0054] It should be understood that if Figure 5 As shown, the spatial topology and attribute information of the vectorized point, line and surface data after the corresponding transformation are input and processed, and the corresponding data topological relationship is established to form various forecast and actual data with topological attributes, that is, meteorological information with topological attributes is obtained.

[0055] This embodiment adopts the above scheme, by acquiring subjective and objective weather forecast data and actual weather observation data, and using the subjective and objective weather forecast data and the actual weather observation data as actual weather forecast data; acquiring the bulletin type of the national weather bulletin; converting the actual weather forecast data into vectorized point, line and surface data stored in the GIS standard format through the data service interface according to the bulletin type; performing data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes, so as to obtain meteorological information quickly and accurately, and effectively improve the efficiency and accuracy of writing service materials.

[0056] Further, Figure 6 FIG. 1 is a flow chart of a third embodiment of the method for automatically generating weather bulletin materials according to the present invention. Figure 6 As shown, a third embodiment of the method for automatically generating weather bulletin materials of the present invention is proposed based on the first embodiment. In this embodiment, the step S20 specifically includes the following steps: Step S21: according to the analysis service interface, the meteorological information corresponding to the spatial distribution field of the meteorological area and the geographical area are clipped, intersected, superimposed and merged to obtain meteorological service text data of various weather phenomena in the superimposed geographical area that meets the preset specifications.

[0057] It should be noted that, according to the analysis service interface, the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information can be clipped, intersected, superimposed and fused, so as to obtain meteorological service text data that meets the preset specifications for various weather phenomena in the superimposed geographical area.

[0058] In the specific implementation, Figure 5 As shown in the figure, the spatial distribution field of the meteorological zone is clipped, intersected, superimposed and fused with the multi-level geographical areas to obtain information such as the level, direction and proportion of various weather phenomena in the superimposed geographical areas. Among them, the division of geographical areas combines the 3-5 level national-provincial-municipal meteorological geographical zoning standards, and also takes into account the description habits of place names, rivers, plains, mountains and plateaus commonly used by forecasters and the public, fully reflecting the characteristics of natural language in algorithm applications.

[0059] Through the above scheme, this embodiment obtains meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area by clipping, intersecting, superimposing and merging the spatial distribution field of the meteorological area and the geographical area corresponding to the meteorological information according to the analysis service interface, and can quickly obtain meteorological service text data that meets load specifications, thereby improving the speed and efficiency of automatic generation of weather bulletin materials.

[0060] Further, Figure 7 FIG. 4 is a flow chart of a fourth embodiment of the method for automatically generating weather bulletin materials according to the present invention. Figure 7 As shown, a fourth embodiment of the method for automatically generating weather bulletin materials of the present invention is proposed based on the first embodiment. In this embodiment, the step S30 specifically includes the following steps: Step S31: Acquire historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning.

[0061] It should be noted that after obtaining the historical text service materials, data cleaning can be performed on the historical text service materials to filter out noise and irrelevant information therein, thereby obtaining the target historical text service materials after data cleaning.

[0062] Step S32: Use a preset neural network model and word segmentation system to perform deep learning on the target historical text service material to obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments.

[0063] It is understandable that, by using the pre-set neural network model and word segmentation system, the target historical text service material can be deeply learned to obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments.

[0064] In the specific implementation, see Figure 8 , Figure 8 Schematic diagram of keyword description in the method for automatically generating weather bulletin materials of the present invention, such as Figure 8 As shown, deep learning is performed on the cleaned data using neural network technology (the association relationship of words adopts a fully linked deep neural network, the phrases adopt a long short-term memory network method, the sentence and paragraph structure adopt a recurrent neural network, etc., which are not limited in this embodiment) and a word segmentation system (maximum matching method, reverse maximum matching method, bidirectional matching method, best matching method, minimum word segmentation method, word grid algorithm, etc., which are not limited in this embodiment); through deep learning of text materials, the word frequency and association relationship of the corresponding description keywords in different semantic environments can be obtained.

[0065] Step S33: perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship in combination with season and scope size to obtain a bulletin automatic generation service model.

[0066] It should be understood that phrase planning, sentence planning, paragraph planning and chapter planning are performed according to the word frequency and the keyword association relationship in combination with the season and range size, so as to obtain a bulletin automatic generation service model.

[0067] In the specific implementation, see Fig. 9 , Fig. 9 The schematic diagram of the bulletin text service model generation in the weather bulletin material automatic generation method of the present invention is as follows: Fig. 9 As shown, model generation mainly consists of several processes: data standardization training processing, in-depth training of historical materials, and final output model rule construction. First, the meteorological service text data needs to be cleaned. This embodiment cleans and organizes the data from the aspects of noise removal and material classification standardization. For the noise removal of meteorological service text data, some symbols, labels, pictures and the like in the document that have no direct relationship with the content expression can be uniformly judged as noise for text materials, such as document alignment and indentation. By comparing the materials before and after noise removal, the information such as styles, pictures, and headers that affect the text expression can be removed as noise, which is more conducive to the subsequent text material training work.

[0068] Regarding the normalization of material classification, this embodiment normalizes and classifies text materials according to weather phenomena (precipitation, temperature, severe convection, etc.) and weather levels (light, medium, heavy, violent, moderate, light, etc.). The analysis result classifies similar weather descriptions into a sequence. For example, weather descriptions with only light rain are classified separately, while descriptions with heavy rain are classified into another sequence. The materials processed according to this classification rule are used as the basic data set for the next step of training.

[0069] It should be noted that the original meteorological data is mainly structured and vectorized vector data, including points, grids, lines, surfaces and other forms, covering various meteorological elements (rain, snow, temperature, wind, strong weather, environment), etc., and the timeliness includes short-term, short-term, medium-term and long-term timeliness. The description of meteorological information is in the form of unstructured information description, which is formed by a document described in text and a table in the form of a data list. Then how to establish the relationship between the two is the most important technical point to realize the expression from meteorological data to text language. This embodiment proposes the following basic ideas for this problem: convert various forms of meteorological original data into meteorological description data in the form of structured text (this process uses GIS-related overlay At the same time, the final descriptive text information corresponding to the original meteorological data is extracted to form a standardized original meteorological data-structured text-finally published descriptive text association relationship; based on many years of historical meteorological text service products, corpus data cleaning, text feature extraction, and machine learning training are carried out to establish an automatic generation model for meteorological service text descriptions. The core idea of ​​this model is to extract meteorological and geographic information feature variables through deep learning of historical materials, analyze high-frequency phrases and short sentences, establish contextual associations, and combine the forecaster's material writing experience to form a multi-dimensional linear analysis model of data → text, and deeply apply natural language technology to realize the automatic generation of various types of text service materials.

[0070] In the specific implementation, see Fig.10 , Fig.10 Schematic diagram of text planning in the method for automatically generating weather bulletin materials of the present invention, such as Fig.10 As shown in the figure, after the deep training, the frequency of weather description rules in different situations is analyzed according to the weather phenomena and weather levels, and it can be further classified according to the season and range size; according to the relationship between weather phenomena and linking words, the frequency of corresponding linking words in different contexts is also obtained; through the word frequency statistics of keywords and linking words in different contexts, it can provide basic support for the construction of the subsequent automatic analysis model; combined with the basis of the previous works, the construction of the text automatic analysis model is realized, and on the basis of the word frequency and association relationship training results, the phrase description, sentence expression, and chapter organization of the meteorological text material are planned, and finally a model library with different text descriptions in different scenarios is formed; in the phrase planning, the structure of the phrase and the meteorological and geographical location variables are sorted out respectively; in the sentence planning, the text description habits of different weather phenomena are studied and trained, and a sentence template library describing various weather phenomena is formed in combination with the regional geographical information; finally, in the chapter planning, the phrase library and sentence library as well as different weather type description rules are combined to form a rich paragraph template library and chapter template library.

[0071] Step S34: input the meteorological range text data into the bulletin automatic generation service model, and output the comprehensive bulletin material through the material organization interface.

[0072] It can be understood that after the meteorological range text data is input into the bulletin automatic generation service model, the comprehensive bulletin material can be output through the material organization interface.

[0073] Furthermore, the step S34 specifically includes the following steps: Perform statistics on historical bulletin documents, create a template document, and fill the variable fields in the template document with actual data through Aspose.Words components according to preset stable setting instructions to obtain a preset bulletin document; Inputting the meteorological range text data into the bulletin automatic generation service model, and outputting comprehensive bulletin materials conforming to the preset bulletin document format through the material organization interface; Upon receiving the subjective revision opinion, the comprehensive communiqué material is subjectively revised to obtain the communiqué material after subjective revision, and the communiqué material after subjective revision is compared with the comprehensive communiqué material before subjective revision by using the Jaccard similarity coefficient to obtain the Jaccard distance; When the Jaccard distance is less than a preset distance, the subjectively revised communiqué material is used as the updated comprehensive communiqué material.

[0074] It should be noted that, through the statistics of historical bulletin documents, a template document is established, and the areas with unchanged content, style and position are defined as constant fields, and the areas with changing content are defined as variable fields. The general format is customized, and the template document can be called through the Aspose.Words component, and then its variable fields are filled with actual data and finally exported as the target bulletin document; Aspose technology is used to customize templates for comprehensive organization of text, graphics and other service material contents, to achieve automatic generation of text paragraphs, automatic insertion of thematic maps, dynamic changes of time parameters, automatic adaptation of formats, automatic updating of duty information and publication flags, etc., and finally generate a meteorological service material document with unified style and standardized format.

[0075] It can be understood that when subjective correction opinions are received, the comprehensive communiqué materials are subjectively revised to obtain subjectively revised communiqué materials, and the two communiqué materials are compared through the Jaccard similarity coefficient. When the Jaccard distance is less than the preset distance, the subjectively revised communiqué materials can be used as updated comprehensive communiqué materials.

[0076] In the specific implementation, see Fig.11 , Fig.11 Schematic diagram of text verification evaluation in the method for automatically generating weather bulletin materials of the present invention. Fig.11 As shown, the Jaccard similarity coefficient is used to compare the similarities and differences between finite sample sets. The larger the Jaccard coefficient value, the higher the sample similarity. Minimum hash is an indicator related to the Jaccard coefficient, which we call the Jaccard distance. The Jaccard distance is mainly used to describe the similarity between sets. The larger the Jaccard distance, the lower the sample similarity; this embodiment tests these two indicators separately according to the characteristics of the material, and can achieve quantitative and qualitative evaluation of the entire article, paragraphs, different seasons, and different elements, basically meeting the requirements of quantitative evaluation, and providing strong support for subsequent algorithm iterations.

[0077] This embodiment uses the above scheme to obtain historical text service materials, clean the historical text service materials, and obtain the target historical text service materials after data cleaning; use the preset neural network model and word segmentation system to perform deep learning on the target historical text service materials, and obtain the word frequency and keyword association relationship of the corresponding description keywords under different semantic environments; perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship in combination with the season and range size to obtain a bulletin automatic generation service model; input the meteorological range text data into the bulletin automatic generation service model, and output the comprehensive bulletin material through the material organization interface; it can effectively improve the writing efficiency of service materials, the automatically generated forecast text content is correct, consistent with the original data, no important meteorological elements are omitted, compared with the forecast text manually written by the forecaster, the degree of consistency is high, the automatically generated forecast text has smooth sentences, correct semantics and grammar, and the text style conforms to the forecaster's writing habits, which ensures the correctness, consistency and standardization of the document, avoids errors that may be caused by manual operation, and improves the speed and efficiency of automatic generation of weather bulletin materials.

[0078] Accordingly, the present invention further provides a device for automatically generating weather bulletin materials.

[0079] Reference Fig.12 , Fig.12 This is a functional module diagram of the first embodiment of the automatic generation device for weather bulletin materials of the present invention.

[0080] In the first embodiment of the weather bulletin material automatic generation device of the present invention, the weather bulletin material automatic generation device comprises: The data processing module 10 is used to obtain the weather forecast live data and the bulletin type of the national weather bulletin, and to process the weather forecast live data through the data service interface according to the bulletin type to obtain meteorological information.

[0081] The information analysis module 20 is used to analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets preset specifications.

[0082] The material generation module 30 is used to input the meteorological service text data into the bulletin automatic generation service model and output the comprehensive bulletin material through the material organization interface.

[0083] The data processing module 10 is also used to obtain subjective and objective weather forecast data and actual weather observation data, and use the subjective and objective weather forecast data and the actual weather observation data as actual weather forecast data; obtain the bulletin type of the national weather bulletin; convert the actual weather forecast data into vectorized point, line and surface data stored in a GIS standard format through a data service interface according to the bulletin type; perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

[0084] The data processing module 10 is also used to convert the meteorological forecast real-time data into grid forecast data, fall area forecast and warning data and target real-time data of various meteorological elements stored in GIS standard format through the data service interface according to the bulletin type; and convert the grid forecast data, the fall area forecast and warning data and the target real-time data according to forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids to obtain vectorized point, line and surface data.

[0085] The data processing module 10 is also used to obtain the spatial topology and attribute information corresponding to the vectorized point, line and surface data, establish a data topological relationship between the spatial topology and the attribute information, and obtain meteorological information with topological attributes.

[0086] The information analysis module 20 is also used to cut, intersect, superimpose and merge the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information according to the analysis service interface, so as to obtain meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area.

[0087] The material generation module 30 is also used to obtain historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning; use a preset neural network model and a word segmentation system to perform deep learning on the target historical text service materials to obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments; perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship combined with the season and range size to obtain a bulletin automatic generation service model; input the meteorological range text data into the bulletin automatic generation service model, and output comprehensive bulletin materials through the material organization interface.

[0088] The material generation module 30 is also used to collect statistics on historical bulletin documents, establish template documents, fill the variable fields in the template documents with actual data through the Aspose.Words component according to preset stable setting instructions, and obtain a preset bulletin document; input the meteorological range text data into the bulletin automatic generation service model, and output comprehensive bulletin materials that conform to the preset bulletin document format through the material organization interface; when subjective correction opinions are received, subjectively revise the comprehensive bulletin materials to obtain subjectively revised bulletin materials, and compare the subjectively revised bulletin materials with the comprehensive bulletin materials before the subjective revision through the Jaccard similarity coefficient to obtain the Jaccard distance; when the Jaccard distance is less than the preset distance, the subjectively revised bulletin materials are used as updated comprehensive bulletin materials.

[0089] Among them, the steps implemented by each functional module of the weather bulletin material automatic generation device can refer to the various embodiments of the weather bulletin material automatic generation method of the present invention, and will not be repeated here.

[0090] In addition, an embodiment of the present invention further provides a storage medium, on which a weather bulletin material automatic generation program is stored. When the weather bulletin material automatic generation program is executed by a processor, the following operations are implemented: Acquire live weather forecast data and the bulletin type of the national weather bulletin, and process the live weather forecast data through a data service interface according to the bulletin type to obtain weather information; Analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The meteorological service text data is input into the bulletin automatic generation service model, and the comprehensive bulletin materials are output through the material organization interface.

[0091] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: Obtaining subjective and objective weather forecast data and real-time weather observation data, and using the subjective and objective weather forecast data and the real-time weather observation data as weather forecast real-time data; Get the bulletin type of the national weather bulletin; According to the bulletin type, the weather forecast live data is converted into vectorized point, line and surface data stored in a GIS standard format through a data service interface; Perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

[0092] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: According to the bulletin type, the weather forecast live data is converted into grid forecast data, fall zone forecast and warning data and target live data of various meteorological elements stored in a GIS standard format through a data service interface; According to the forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids, the grid forecast data, the fall area forecast and warning data and the target actual data are converted to obtain vectorized point, line and surface data.

[0093] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: The spatial topology and attribute information corresponding to the vectorized point, line and surface data are obtained, a data topological relationship between the spatial topology and the attribute information is established, and meteorological information with topological attributes is obtained.

[0094] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: According to the analysis service interface, the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information are clipped, intersected, superimposed and merged to obtain meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area.

[0095] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: Acquire historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning; Using a preset neural network model and word segmentation system to conduct deep learning on the target historical text service material, and obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments; Perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship combined with season and scope size to obtain a bulletin automatic generation service model; The meteorological range text data is input into the bulletin automatic generation service model, and the comprehensive bulletin material is output through the material organization interface.

[0096] Furthermore, when the weather bulletin material automatic generation program is executed by the processor, the following operations are also implemented: Perform statistics on historical bulletin documents, create a template document, and fill the variable fields in the template document with actual data through Aspose.Words components according to preset stable setting instructions to obtain a preset bulletin document; Inputting the meteorological range text data into the bulletin automatic generation service model, and outputting comprehensive bulletin materials conforming to the preset bulletin document format through the material organization interface; Upon receiving the subjective revision opinion, the comprehensive communiqué material is subjectively revised to obtain the communiqué material after subjective revision, and the communiqué material after subjective revision is compared with the comprehensive communiqué material before subjective revision by using the Jaccard similarity coefficient to obtain the Jaccard distance; When the Jaccard distance is less than a preset distance, the subjectively revised communiqué material is used as the updated comprehensive communiqué material.

[0097] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned implementation method can be accomplished by instructing related hardware through a program, and the program is stored in a storage medium, including a number of instructions for enabling a device (which may be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps of the method described in each implementation method of the present application; and the aforementioned storage medium is a computer-readable storage medium, including but not limited to: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, and other media that can store program codes.

[0098] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0099] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0100] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for automatically generating weather bulletin materials, characterized in that: The method for automatically generating weather bulletin materials comprises: Acquire live weather forecast data and the bulletin type of the national weather bulletin, and process the live weather forecast data through a data service interface according to the bulletin type to obtain weather information; Analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The meteorological service text data is input into the bulletin automatic generation service model, and the comprehensive bulletin materials are output through the material organization interface.

2. The method for automatically generating weather bulletin materials according to claim 1, characterized in that: The method of acquiring the weather forecast live data and the bulletin type of the national weather bulletin, and processing the weather forecast live data through the data service interface according to the bulletin type to obtain the weather information includes: Obtaining subjective and objective weather forecast data and real-time weather observation data, and using the subjective and objective weather forecast data and the real-time weather observation data as weather forecast real-time data; Get the bulletin type of the national weather bulletin; According to the bulletin type, the weather forecast live data is converted into vectorized point, line and surface data stored in a GIS standard format through a data service interface; Perform data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes.

3. The method for automatically generating weather bulletin materials according to claim 2, characterized in that: The step of converting the weather forecast live data into vectorized point, line and surface data stored in a GIS standard format through a data service interface according to the bulletin type includes: According to the bulletin type, the weather forecast live data is converted into grid forecast data, fall zone forecast and warning data and target live data of various meteorological elements stored in a GIS standard format through a data service interface; According to the forecast points, forecast lines, forecast surfaces, forecast grids, observation sites and observation grids, the grid forecast data, the fall area forecast and warning data and the target actual data are converted to obtain vectorized point, line and surface data.

4. The method for automatically generating weather bulletin materials according to claim 2, characterized in that: The performing data topology on the vectorized point, line and surface data to obtain meteorological information with topological attributes includes: The spatial topology and attribute information corresponding to the vectorized point, line and surface data are obtained, a data topological relationship between the spatial topology and the attribute information is established, and meteorological information with topological attributes is obtained.

5. The method for automatically generating weather bulletin materials according to claim 1, characterized in that: The step of analyzing the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specification includes: According to the analysis service interface, the spatial distribution field of the meteorological zone and the geographical area corresponding to the meteorological information are clipped, intersected, superimposed and merged to obtain meteorological service text data that meets preset specifications for various weather phenomena in the superimposed geographical area.

6. The method for automatically generating weather bulletin materials according to claim 1, characterized in that: The step of inputting the meteorological service text data into the bulletin automatic generation service model and outputting the comprehensive bulletin material through the material organization interface includes: Acquire historical text service materials, perform data cleaning on the historical text service materials, and obtain target historical text service materials after data cleaning; Using a preset neural network model and word segmentation system to conduct deep learning on the target historical text service material, and obtain the word frequency and keyword association relationship of corresponding description keywords in different semantic environments; Perform phrase planning, sentence planning, paragraph planning and chapter planning according to the word frequency and the keyword association relationship combined with season and scope size to obtain a bulletin automatic generation service model; The meteorological range text data is input into the bulletin automatic generation service model, and the comprehensive bulletin material is output through the material organization interface.

7. The method for automatically generating weather bulletin materials according to claim 6, characterized in that: The step of inputting the meteorological range text data into the bulletin automatic generation service model and outputting comprehensive bulletin materials through the material organization interface includes: Perform statistics on historical bulletin documents, create a template document, and fill the variable fields in the template document with actual data through the Aspose.Words component according to the preset stable setting instructions to obtain the preset bulletin document; Inputting the meteorological range text data into the bulletin automatic generation service model, and outputting comprehensive bulletin materials conforming to the preset bulletin document format through the material organization interface; Upon receiving the subjective revision opinion, the comprehensive communiqué material is subjectively revised to obtain the communiqué material after subjective revision, and the communiqué material after subjective revision is compared with the comprehensive communiqué material before subjective revision by using the Jaccard similarity coefficient to obtain the Jaccard distance; When the Jaccard distance is less than a preset distance, the subjectively revised communiqué material is used as the updated comprehensive communiqué material.

8. A device for automatically generating weather bulletin materials, characterized in that: The weather bulletin material automatic generation device comprises: A data processing module, used to obtain weather forecast live data and the bulletin type of the national weather bulletin, and process the weather forecast live data through the data service interface according to the bulletin type to obtain weather information; An information analysis module, used to analyze the meteorological information according to the analysis service interface to obtain meteorological service text data that meets the preset specifications; The material generation module is used to input the meteorological service text data into the bulletin automatic generation service model and output the comprehensive bulletin material through the material organization interface.

9. A weather bulletin material automatic generation device, characterized in that: The weather bulletin material automatic generation device includes: a memory, a processor, and a weather bulletin material automatic generation program stored in the memory and executable on the processor, wherein the weather bulletin material automatic generation program is configured to implement the steps of the weather bulletin material automatic generation method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a weather bulletin material automatic generation program, which, when executed by a processor, implements the steps of the weather bulletin material automatic generation method as described in any one of claims 1 to 7.

Images