Land utilization information extraction method and system based on historical soil survey text

Abstract

The invention discloses a historical soil survey text-based land utilization information extraction method and system. The method comprises the following steps of: establishing a plant name dictionary of three land utilization types of paddy fields, dry lands and other forest lands in a research area; reading a vegetation description text Vi of a sampling point from a historical soil survey text of a research area, judging whether the Vi contains plants in a plant name dictionary or not, if so, assigning a land utilization type corresponding to the plant to a related variable of a species name, assigning a coverage value corresponding to full coverage to the coverage, storing the coverage in a species information list, and if not, assigning the coverage value corresponding to the full coverage to the species information list; otherwise, using a regular expression and an information extraction method to obtain names and coverage degrees of main species, and storing the names and the coverage degrees in a species information list; for each element in the species information list, extracting a corresponding life type from a preset plant website by using a web crawler technology; and inferring the land utilization type of the sampling point according to the life type and the corresponding coverage degree. The method is high in efficiency.

Description

Technical Field

[0001] The present invention relates to the field of natural language information processing, and in particular to a method and system for extracting land use information based on historical soil survey texts. Background Art

[0002] The extraction of land use information contained in historical soil survey texts is crucial for matching soil entity description characteristics with soil spatial distribution characteristics and improving the spatial positioning quality of historical survey sample points. However, in many historical soil surveys, the land use information of survey sample points has not been accurately and standardizedly recorded. It is of great significance and application value to study how to effectively use the original soil survey data and scientifically infer the land use type information in the specific survey scenario of the historical survey time and sample survey location.

[0003] Some researchers have tried to use the vegetation survey information in the original soil census data to artificially infer the land use type information at the historical survey time and sample survey location, but the relevant processing methods are relatively inefficient and have a low degree of automation, which cannot meet the requirements of rapid inference of a large number of land use types. This inefficiency is mainly caused by the heterogeneity and unstructured characteristics of historical vegetation description information and the non-standardization of vegetation naming. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide an efficient land use information extraction method and system based on historical soil survey texts.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0006] A land use information extraction method based on historical soil survey text, the method comprising the following steps:

[0007] (1) Establish a dictionary of plant names for the three land use types of paddy fields, dry land and other woodlands in the study area;

[0008] (2) Read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i Does it contain plants in the plant name dictionary? If so, execute (3); otherwise, execute (4);

[0009] (3) Assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1)Store the species information list PlantList i and execute step (5);

[0010] (4) Use regular expressions to describe the vegetation text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract the species name key of each plant ij , and the corresponding coverage value ij , forming species information (key ij ,value ij )Store the species information list PlantList i and execute step (5);

[0011] (5) For the species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i ;

[0012] (6) According to Life List i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i ;

[0013] (7) Repeat steps (2) to (6) until all sampling points are processed.

[0014] Furthermore, step (1) specifically includes:

[0015] (1.1) Obtain the species names of plants in paddy fields, dry land and other woodlands in the study area through historical experience or data;

[0016] (1.2) The species names of plants whose land use types belong to paddy fields, dry lands and other woodlands are collected as plant name dictionaries for paddy fields, dry lands and other woodlands respectively.

[0017] Furthermore, step (4) specifically includes:

[0018] (4-1) Based on regular expressions, the vegetation description text V i Replace all the conjunctions in the sentence with commas;

[0019] (4-2) The vegetation description text V processed from step (4-1) iIn the example, it is determined whether there is a description text of the main vegetation. If so, the description text of all the main vegetation is extracted based on the regular expression and stored in the temporary variable main_plant. Otherwise, the vegetation description information V is deleted based on the regular expression. i The remaining text is stored in the temporary variable main_plant;

[0020] (4-3) According to the Chinese stop word list, delete the stop words in the temporary variable main_plant, and use the deleted text as the main vegetation description text I i ;

[0021] (4-4) Main vegetation description text I i Input information extraction model, extract I i The species names of all plants in and the corresponding coverage description text; wherein the information extraction model is a trained artificial intelligence model;

[0022] (4-5) Based on the regular expression, the coverage description text is normalized to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i},n i For I i Number of species names.

[0023] Further, step (5) includes:

[0024] (5-1) For the species information list PlantList i , if the species name key i1 is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, execute step (5-2);

[0025] (5-2) For the species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ;

[0026] (5-3) For any URL encoding C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ;

[0027] (5-4) Access and parse plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , and execute step (5-6); otherwise, execute step (5-5);

[0028] (5-5) Find the URL N' corresponding to the botanical name of the current species suggested by the website ij , and access and parse N' ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None;

[0029] (5-6) Loop through steps (5-2) to (5-5) until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i}, n i PlantList is a list of species information i Number of species names.

[0030] Further, step (6) comprises:

[0031] (6-1) From the Life List i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ;

[0032] (6-2) If there is at least one element among all elements that contains "tree" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling pointi The value is assigned as “sparse woodland”, and δ is the preset threshold;

[0033] (6-3) If all elements do not contain "trees" and there is at least one element containing "shrubs", then the land use type U corresponding to the current sampling point i Assign the value to "shrubbery";

[0034] (6-4) If all elements do not contain "trees" and "shrubs", and at least one element contains "herbs", then the land use type U corresponding to the current sampling point is i Assign the value to "grass".

[0035] A land use information extraction system based on historical soil survey texts, including:

[0036] The dictionary building module is used to build a dictionary of plant names for three land use types: paddy fields, dry land, and other woodlands in the study area;

[0037] The species name judgment module is used to read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i whether the plant in the plant name dictionary is included, if yes, execute (3) the first species information list forming module, otherwise execute (4) the second species information list forming module;

[0038] The first species information list forming module is used to assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1 )Store the species information list PlantList i In, and execute the life style list formation module;

[0039] The second species information list forming module is used to use regular expressions to generate vegetation description text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract each species name and the corresponding coverage value, and assign each species name to the species name key ij , the corresponding coverage value is assigned to the coverage value ij , forming species information (key ij ,value ij )Store the species information list PlantList iIn, and execute the life style list formation module;

[0040] Life form list generation module, used for species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i ;

[0041] Land use type inference module, used to infer land use types based on life type list LifeList i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i ; and return to execute the species name judgment module until the processing of all sampling points is completed.

[0042] Furthermore, the dictionary building module specifically includes:

[0043] The vegetation acquisition unit is used to obtain the species names of plants in paddy fields, dry land and other woodlands in the study area through historical experience or data, as well as the land use type corresponding to each plant, whether it belongs to paddy fields, dry land or other woodlands;

[0044] The dictionary building unit is used to collect the species names of plants whose land use types belong to paddy fields, dry lands and other woodlands as plant name dictionaries of paddy fields, dry lands and other woodlands respectively.

[0045] Furthermore, the second species information list forming module specifically includes:

[0046] The first regular expression processing unit is used to convert the vegetation description text V into i Replace all the conjunctions in the sentence with commas;

[0047] The main text extraction unit is used to extract the vegetation description text V from the regular expression processing unit. i In the example, it is determined whether there is a description text of the main vegetation. If so, the description text of all the main vegetation is extracted based on the regular expression and stored in the temporary variable main_plant. Otherwise, the vegetation description information V is deleted based on the regular expression. i The remaining text is stored in the temporary variable main_plant;

[0048] The stop word deletion unit is used to delete the stop words in the temporary variable main_plant according to the Chinese stop word list, and the deleted text is used as the main vegetation description text I i ;

[0049] Species information extraction unit, used to extract the main vegetation description text I i Input information extraction model, extract I i The species names of all plants and the corresponding coverage description texts in the data; wherein the information extraction model is a trained artificial intelligence model;

[0050] The second regular expression processing unit is used to normalize the coverage description text based on the regular expression to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i},n i For I i Number of species names.

[0051] Furthermore, the lifestyle list forming module specifically includes:

[0052] Life form acquisition unit, used for species information list PlantList i , if the species name key i1 is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, the encoding unit is executed;

[0053] Coding unit for species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ;

[0054] URL acquisition unit, used to encode any URL C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ;

[0055] The first life-type search unit is used to access and parse the plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyleij , and execute the loop execution unit; otherwise, execute the second life type search unit in step;

[0056] The second life form search unit is used to search for the URL N' corresponding to the botanical name of the current species prompted by the website ij , and access and parse N' ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None;

[0057] Loop through the execution unit and return to the execution encoding unit until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i}, n i PlantList is a list of species information i Number of species names.

[0058] Furthermore, the land use type inference module includes:

[0059] The first land use type extraction unit is used to extract the land use type from the life type list LifeList i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ;

[0060] The second land use type extraction unit is used to extract the land use type if at least one element among all elements contains "trees" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling point i The value is assigned as “sparse woodland”, and δ is the preset threshold;

[0061] The third land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and at least one element contains "shrubs". i Assign the value to "shrubbery";

[0062] The fourth land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and "shrubs" and at least one element contains "herbs". i Assign the value to "grass".

[0063] Compared with the prior art, the invention has the following beneficial effects: the invention has a high degree of automation, greatly reducing the input of human resources and time. In addition, the method is universal and provides a new way to restore and analyze the missing historical land use type information in various applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A flow chart of a method for extracting land use information based on historical soil survey texts provided by an embodiment of the present invention;

[0065] Figure 2 A flow chart of extracting species names and their coverage in the method of the present invention;

[0066] Figure 3 A flowchart of crawling lifestyle information in the method of the present invention;

[0067] Figure 4 The figure is a flow chart for inferring land use type in the method of the present invention. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0069] Embodiment 1

[0070] The embodiment of the present invention provides a method for extracting land use information based on historical soil survey texts. Figure 1 As shown, the method comprises the following steps:

[0071] (1) Establish a dictionary of plant names for the three land use types of paddy fields, dry land and other woodlands in the study area.

[0072] The specific steps include: (1.1) obtaining the species names of plants in paddy fields, dry lands and other woodlands in the study area through historical experience or data, as well as whether the land use type of each plant belongs to paddy fields, dry lands or other woodlands; (1.2) collecting the species names of plants whose land use types belong to paddy fields, dry lands and other woodlands as plant name dictionaries for paddy fields, dry lands and other woodlands, respectively.

[0073] For example, the data of Jiangxi Province in the second soil survey in China (1979-1986) is used as experimental data, as shown in Table 1 below.

[0074] Table 1 Data of Jiangxi Province in the Second Soil Survey in China (1979-1986)

[0075]

[0076] The plant name dictionary for paddy field P, dry land D and other woodland O is shown in Table 2;

[0077] Table 2 Plant name dictionary

[0078]

[0079] (2) Read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i Does it contain plants in the plant name dictionary? If so, execute (3); otherwise, execute (4).

[0080] (3) Assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1 )Store the species information list PlantList i and execute step (5). Figure 2 shown.

[0081] For example, in this embodiment, V 195 "One season of early rice or late rice, winter leisure", because it contains "early rice" and "late rice" in P, the key in this sampling point 195,1 Assign the value to "paddy field", the coverage value corresponding to full coverage is 100, then the coverage value 195,1 Assign a value of 100. V6 is "planted with sweet potatoes and wheat crops, two crops a year", because it contains "sweet potatoes" and "wheat" in D, key 6,1 Assigned value to "dry land", value 6,1 Assign a value of 100; V3 is "Kumquat, Tea", because it contains "Kumquat" and "Tea" in O, 3,1 Assigned value to "other forest land", value 3,1 Assign a value of 100.

[0082] (4) Use regular expressions to describe the vegetation text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract the species name key of each plant ij, and the corresponding coverage value ij , to form species information (key ij , value ij ) and store it in the species information list PlantList i , and execute step (5).

[0083] As Figure 2 shown, this step specifically includes:[[]]

[0084] (4-1) Based on regular expressions, replace all juxtaposed conjunctive words in the vegetation description text V i with commas; for example, replace conjunctions such as "and", "and", "with" with ",";

[0085] (4-2) From the vegetation description text V processed in step (4-1) i , determine whether there is a description text of the main vegetation. If so, extract all description texts of the main vegetation based on regular expressions and store them in the temporary variable main_plant. Otherwise, delete the description text of the secondary vegetation in the vegetation description information V i , and store the remaining text in the temporary variable main_plant; among them, the description text of the main vegetation will have description words such as "mainly...", "more...", etc., and the description text of the secondary vegetation will have description words such as "interspersed with...". For example, V1 is "mainly fir and pine, and some are camellia oleifera and chestnut", containing "mainly...". Based on regular expressions, "fir and pine" is extracted and stored in the temporary variable main_plant; V2 is "sparse masson pine, interspersed with a large area of grassland of the Gramineae family", containing the secondary vegetation information of "interspersed with...". Based on regular expressions, "interspersed with a large area of grassland of the Gramineae family" is deleted, and the remaining "sparse masson pine" is stored in main_plant;

[0086] (4-3) According to the Chinese stop word list, delete the stop words in the temporary variable main_plant, and use the text after deletion as the main vegetation description text I i ; among them, the Chinese stop word list is formed by adding some specific stop words, such as "have", "grow", "plant", etc., based on the Chinese stop word library of Harbin Institute of Technology according to the experimental data of this embodiment;

[0087] (4-4) Input the main vegetation description text I i into the information extraction model, and extract I iAll species names and corresponding coverage description text cover in ; wherein the information extraction model is a trained artificial intelligence model, such as ChatGPT, which is extracted by inputting a command word prompt, for example, the command word can be: extract species names and corresponding density information from the following text: {I i}, please output in the following format: species name - density information. The model output is: "Pinus massoniana - sparse; Sapium sebiferum - dense; Dicranopteris dichotoma - density not clearly mentioned; Imperata cylindrica - density not clearly mentioned". Call the API multiple times and collect valid results. The valid result with the most occurrences is the final result extracted;

[0088] (4-5) Based on the regular expression, the coverage description text is normalized to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i},n i For I i The number of species names in the data; the normalization processing is to process the vague descriptive information such as "sparse" and "sporadic" into specific coverage. For example, by constructing a density mapping table of density_mapping = {'sporadic': 5, 'a small amount': 10, 'sparse': 10, 'medium': 50, 'large area': 70, 'dense': 90, 'density not clearly mentioned': 30}, the final list of all species information is shown in Table 3 below.

[0089] Table 3 Species information list of all sampling points

[0090]

[0091] (5) For the species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i .

[0092] like Figure 3 As shown, this step specifically includes:

[0093] (5-1) For the species information list PlantList i , if the species name key i1is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, execute step (5-2); for example, Sample 195 Species name key 195,1 is "paddy field", which belongs to the land use type corresponding to the plant name dictionary, then l 195,1 Assign the value to "paddy field". Similarly, Sample 6's l 6,1 Assigned to "dry land", Sample3's l 3,1 The value is assigned as “other forest land”;

[0094] (5-2) For the species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ; For example, the PlantList4 of Sample4 is {'Pinus massoniana': 10,'Rosa multiflora': 30,'Cynodon dactylon': 30,'Dicranopteris dichotoma': 30}, for key 4,1 ='Massonia pine', which can convert 'Massonia pine' into the corresponding URL encoding C 4,1 ='%E9%A9%AC%E5%B0%BE%E6%9D%BE';

[0095] (5-3) For any URL encoding C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ; The default plant website is https: / / www.iplant.cn / info / . For C 4,1 , after splicing N 4,1 = "https: / / www.iplant.cn / info / %E9%A9%AC%E5%B0%BE%E6%9D%BE"; The default plant website address can be set by the user;

[0096] (5-4) Access and parse plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , and execute step (5-6); otherwise, execute step (5-5); specifically, according to the label And the attribute search element "lifestyle" with id desc26 and class desctxt;

[0097] (5-5) Find the URL N' corresponding to the botanical name of the current species suggested by the website ij , and access and parse N' ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None;

[0098] Specifically, according to the label And attribute search with class spantxt Link, that is, the URL corresponding to the botanical name of the species indicated on the website N' ij , if the URL N' is found ij , then use BeautifulSoup to parse N' ij HTML, and according to the tag And the attribute search element "lifestyle" with id desc26 and class desctxt, for example, when key 4,4 When searching for Dicranopteris dicranopteris, the life form was not directly found, but the website corresponding to its scientific name Dicranopteris dicranopteris was found, namely, https: / / www.iplant.cn / info / Dicranopteris%20pedata?id=CA658DCEFDB17D4A. The website was visited and the life form was searched, and the result was "ferns".

[0099] (5-6) Loop through steps (5-2) to (5-5) until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i }, n i PlantList is a list of species information i Number of species names.

[0100] For example, LifeList4 of Sample4 is obtained as {'tree', 'climbing shrub', 'perennial low herb', 'fern'}, wherein the intermediate data is specifically shown in Table 4 below, and the life list LifeList of all sampling points is specifically shown in Table 5.

[0101] Table 4 Intermediate data of crawled life style information of Sample 4

[0102] <![CDATA[PlantList4(key)]]> URL Encoding Lifestyle Information Scientific Name Website Lifestyle Information <![CDATA[‘ 马尾松 ’]]> <![CDATA[ ％E9％A9％AC％E5％B0％BE％E6％9D％BE ]]> <![CDATA[ 乔木 ]]> / / 'Wild rose' %E9%87%8E%E8%94%B7%E8%96%87 Climbing shrubs / / 'Cynodon dactylon' %E7%8B%97%E7%89%99%E6%A0%B9 Perennial low herb / / 'Dicranopteris dichotoma' %E9%93%81%E8%8A%92%E8%90%81 Not found exist fern

[0103] Table 5 Life List of all sampling points

[0104]

[0105]

[0106] (6) According to Life List i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i .

[0107] like Figure 4 As shown, this step specifically includes:

[0108] (6-1) From the Life List i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ; For example, Sample 195 LifeList 195 There is only one element "paddy field", which belongs to the land use type, then U 195 The value of U6 in Sample 6 is "paddy field", and the value of U3 in Sample 3 is "other forest land".

[0109] (6-2) If there is at least one element among all elements that contains "tree" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling point i The value is assigned as "sparse woodland", δ is a preset threshold, which is 30 in this embodiment; for example, Sample 1's l 1,1 and l 1,2 All contain "arbor", and value 1,1 and value 1,2 Both are 30, so U1 is "forested land"; while Sample2 and Sample4 both contain "trees" in their life form lists, and their coverage is 10, which is less than δ, so U2 and U4 are "sparse woodlands";

[0110] (6-3) If all elements do not contain "trees" and there is at least one element containing "shrubs", then the land use type U corresponding to the current sampling point i Assign the value to "shrub forest"; for example, the life form in the life form list of Sample5 does not contain "tree", and the element contains "shrub", so U5 is "shrub forest";

[0111] (6-4) If all elements do not contain "trees" and "shrubs", and at least one element contains "herbs", then the land use type U corresponding to the current sampling point is i Assign the value to "grassland". For example, the first life form in Sample7 is None, and the second life form only contains "herbs", so U7 is "grassland".

[0112] (7) Repeat steps (2) to (6) until all sampling points are processed.

[0113] Table 6 Land use type inference process and final results

[0114]

[0115]

[0116] Embodiment 2

[0117] The embodiment of the present invention provides a land use information extraction system based on historical soil survey text, including:

[0118] The dictionary building module is used to build a dictionary of plant names for three land use types: paddy fields, dry land, and other woodlands in the study area;

[0119] The species name judgment module is used to read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i whether the plant in the plant name dictionary is included, if yes, execute (3) the first species information list forming module, otherwise execute (4) the second species information list forming module;

[0120] The first species information list forming module is used to assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1 )Store the species information list PlantList i In, and execute the life style list formation module;

[0121] The second species information list forming module is used to use regular expressions to generate vegetation description text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract the species name key of each plant ij , and the corresponding coverage value ij , forming species information (key ij ,value ij )Store the species information list PlantList i In, and execute the life style list formation module;

[0122] Life form list generation module, used for species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i ;

[0123] Land use type inference module, used to infer land use types based on life type list LifeList i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i ; and return to execute the species name judgment module until the processing of all sampling points is completed.

[0124] The dictionary building module specifically includes:

[0125] The vegetation acquisition unit is used to obtain the species names of plants in paddy fields, dry land and other woodlands in the study area through historical experience or data, as well as the land use type corresponding to each plant, whether it belongs to paddy fields, dry land or other woodlands;

[0126] The dictionary building unit is used to collect the species names of plants whose land use types belong to paddy fields, dry lands and other woodlands as plant name dictionaries of paddy fields, dry lands and other woodlands respectively.

[0127] Wherein, the second species information list forming module specifically includes:

[0128] The first regular expression processing unit is used to convert the vegetation description text V into i Replace all the conjunctions in the sentence with commas;

[0129] The main text extraction unit is used to extract the vegetation description text V from the regular expression processing unit. i In the example, it is determined whether there is a description text of the main vegetation. If so, the description text of all the main vegetation is extracted based on the regular expression and stored in the temporary variable main_plant. Otherwise, the vegetation description information V is deleted based on the regular expression. i The remaining text is stored in the temporary variable main_plant;

[0130] The stop word deletion unit is used to delete the stop words in the temporary variable main_plant according to the Chinese stop word list, and the deleted text is used as the main vegetation description text I i ;

[0131] Species information extraction unit, used to extract the main vegetation description text I i Input information extraction model, extract I i The species names of all plants in and the corresponding coverage description text; wherein the information extraction model is a trained artificial intelligence model;

[0132] The second regular expression processing unit is used to normalize the coverage description text based on the regular expression to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i },n i For I i Number of species names.

[0133] The life style list forming module specifically includes:

[0134] Life form acquisition unit, used for species information list PlantList i , if the species name key i1 is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, the encoding unit is executed;

[0135] Coding unit for species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ;

[0136] URL acquisition unit, used to encode any URL C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ;

[0137] The first life-type search unit is used to access and parse the plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , and execute the loop execution unit; otherwise, execute the second life type search unit in step;

[0138] The second life form search unit is used to search for the URL N' corresponding to the botanical name of the current species prompted by the website ij , and access and parse N' ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None;

[0139] Loop through the execution unit and return to the execution encoding unit until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i }, n i PlantList is a list of species information i Number of species names.

[0140] Wherein, the land use type inference module includes:

[0141] The first land use type extraction unit is used to extract the land use type from the life type list LifeList i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ;

[0142] The second land use type extraction unit is used to extract the land use type if at least one element among all elements contains "trees" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling point i The value is assigned as "sparse woodland", and δ is the preset threshold;

[0143] The third land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and at least one element contains "shrubs". i Assign the value to "shrub forest";

[0144] The fourth land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and "shrubs" and at least one element contains "herbs". i Assign the value to "grass".

[0145] The system provided in the embodiment of the present invention can be used to execute the method provided in the first embodiment of the present invention, and has the corresponding functions and beneficial effects of the execution method. For details not detailed, please refer to the first embodiment and will not be repeated.

[0146] It is worth noting that in the embodiment of the above system, the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0147] The embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, or of course, it can be implemented only by hardware, as long as the function or effect can be achieved.

[0148] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. A land use information extraction method based on historical soil survey texts, It is characterized in that The method comprises the following steps: (1) Establish a dictionary of plant names for the three land use types of paddy fields, dry land and other woodlands in the study area; (2) Read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i Does it contain plants in the plant name dictionary? If so, execute (3); otherwise, execute (4); (3) Assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1 )Store the species information list PlantList i and execute step (5); (4) Use regular expressions to describe the vegetation text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract the species name key of each plant ij , and the corresponding coverage value ij , forming species information (key ij ,value ij )Store the species information list PlantList i and execute step (5); (5) For the species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i ; (6) Based on Life List i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i ; (7) Repeat steps (2) to (6) until all sampling points are processed.

2. The method for extracting land use information based on historical soil survey text according to claim 1, It is characterized in that Step (1) specifically includes: (1.1) Obtain the species names of plants in paddy fields, dry land and other woodlands in the study area through historical experience or data; (1.2) The species names of plants whose land use types belong to paddy fields, dry lands and other woodlands are collected as plant name dictionaries for paddy fields, dry lands and other woodlands respectively.

3. The method for extracting land use information based on historical soil survey text according to claim 1, It is characterized in that Step (4) specifically includes: (4-1) Based on regular expressions, the vegetation description text V i Replace all the conjunctions in the sentence with commas; (4-2) The vegetation description text V processed from step (4-1) i In the example, it is determined whether there is a description text of the main vegetation. If so, the description text of all the main vegetation is extracted based on the regular expression and stored in the temporary variable main_plant. Otherwise, the vegetation description information V is deleted based on the regular expression. i The remaining text is stored in the temporary variable main_plant; (4-3) According to the Chinese stop word list, delete the stop words in the temporary variable main_plant, and use the deleted text as the main vegetation description text I i ; (4-4) Main vegetation description text I i Input information extraction model, extract I i The species names of all plants in and the corresponding coverage description text; wherein the information extraction model is a trained artificial intelligence model; (4-5) Based on the regular expression, the coverage description text is normalized to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i },n i For I i Number of species names.

4. The method for extracting land use information based on historical soil survey text according to claim 1, It is characterized in that Step (5) comprises: (5-1) For the species information list PlantList i , if the species name key i1 is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, execute step (5-2); (5-2) For the species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ; (5-3) For any URL encoding C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ; (5-4) Access and parse plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , and execute step (5-6); otherwise, execute step (5-5); (5-5) Find the URL N corresponding to the botanical name of the current species suggested by the website ’ ij , and access and parse N ’ ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None; (5-6) Loop through steps (5-2) to (5-5) until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i }, n i PlantList is a list of species information i Number of species names in the table.

5. The method for extracting land use information based on historical soil survey text according to claim 1, It is characterized in that Step (6) comprises: (6-1) From the Life List i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ; (6-2) If there is at least one element among all elements that contains "tree" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling point i The value is assigned as "sparse woodland", and δ is the preset threshold; (6-3) If all elements do not contain "trees" and there is at least one element containing "shrubs", then the land use type U corresponding to the current sampling point i Assign the value to "shrub forest"; (6-4) If all elements do not contain "trees" and "shrubs", and at least one element contains "herbs", then the land use type U corresponding to the current sampling point is i Assign the value to "grass".

6. A land use information extraction system based on historical soil survey texts, It is characterized in that include: The dictionary building module is used to build a dictionary of plant names for three land use types: paddy fields, dry land, and other woodlands in the study area; The species name judgment module is used to read the vegetation description text V of any sampling point from the historical soil survey text of the study area i , determine the vegetation description text V i Does it contain plants in the plant name dictionary? If so, execute (3); otherwise, execute (4); The first species information list forming module is used to assign the land use type corresponding to the plant name dictionary to the species name key i1 , assign the coverage value corresponding to full coverage to coverage value i1 , forming species information (key i1 ,value i1 )Store the species information list PlantList i In, and execute the life style list formation module; The second species information list forming module is used to use regular expressions to generate vegetation description text V i Preprocessing is performed to obtain the main vegetation description text I i , and then from I i Extract the species name key of each plant ij , and the corresponding coverage value ij , forming species information (key ij ,value ij )Store the species information list PlantList i In, and execute the life style list formation module; Life form list generation module, used for species information list PlantList i For each element in the life form, we use web crawler technology to extract the corresponding life form from the preset plant website and store it in the life form list LifeList i ; Land use type inference module, used to infer land use types based on life type list LifeList i And the species information list PlantList i The coverage in the inference of the land use type U of the current sampling point i ; and return to execute the species name judgment module until the processing of all sampling points is completed.

7. The land use information extraction system based on historical soil survey text according to claim 6, It is characterized in that The dictionary building module specifically includes: The vegetation acquisition unit is used to obtain the species names of plants in paddy fields, dry land and other woodlands in the study area through historical experience or data, as well as the land use type corresponding to each plant, whether it belongs to paddy fields, dry land or other woodlands; The dictionary building unit is used to collect the species names of plants whose land use types belong to paddy fields, dry lands and other woodlands as plant name dictionaries of paddy fields, dry lands and other woodlands respectively.

8. The land use information extraction system based on historical soil survey text according to claim 6, It is characterized in that The second species information list forming module specifically includes: The first regular expression processing unit is used to convert the vegetation description text V into i Replace all the conjunctions in the sentence with commas; The main text extraction unit is used to extract the vegetation description text V from the regular expression processing unit. i In the example, it is determined whether there is a description text of the main vegetation. If so, the description text of all the main vegetation is extracted based on the regular expression and stored in the temporary variable main_plant. Otherwise, the vegetation description information V is deleted based on the regular expression. i The remaining text is stored in the temporary variable main_plant; The stop word deletion unit is used to delete the stop words in the temporary variable main_plant according to the Chinese stop word list, and the deleted text is used as the main vegetation description text I i ; Species information extraction unit, used to extract the main vegetation description text I i Input information extraction model, extract I i The species names of all plants and the corresponding coverage description texts in the data; wherein the information extraction model is a trained artificial intelligence model; The second regular expression processing unit is used to normalize the coverage description text based on the regular expression to obtain the coverage value of each plant. ij , forming species information (key ij ,value ij ) Add species information list PlantList i Finally, we get PlantList i ={(key ij ,value ij )|j=1,2,…,n i },n i For I i Number of species names.

9. The land use information extraction system based on historical soil survey text according to claim 6, It is characterized in that The life style list forming module specifically includes: Life form acquisition unit, used for species information list PlantList i , if the species name key i1 is the land use type corresponding to the plant name dictionary, then the corresponding life form l i1 Assign value to key i1 , stored in the life list LifeList i Otherwise, the encoding unit is executed; Coding unit for species information list PlantList i Any species name in key ij , key the species name ij Convert to the corresponding URL encoding C ij ; URL acquisition unit, used to encode any URL C ij , concatenate it to the preset plant URL to get the plant URL N of the current species ij ; The first life-type search unit is used to access and parse the plant website N ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , and execute the loop execution unit; otherwise, execute the second life type search unit in step; The second life form search unit is used to search for the URL N corresponding to the botanical name of the current species prompted by the website ’ ij , and access and parse N ’ ij HTML, find out whether there is a life element in it, if so, assign the value of the found life element to key ij Corresponding lifestyle ij , otherwise, l ij The value of is assigned to None; Loop through the execution unit and return to the execution encoding unit until all URL encodings are completed. ij Traverse to get the life list LifeList i = {l ij |j=1,2,…,n i }, n i PlantList is a list of species information i Number of species names.

10. The land use information extraction system based on historical soil survey text according to claim 6, It is characterized in that The land use type inference module includes: The first land use type extraction unit is used to extract the land use type from the life type list LifeList i Extract all elements from , if there is only one element and the element is the land use type corresponding to the plant name dictionary, then the land use type U corresponding to the current sampling point is i Assigned to l ij ; The second land use type extraction unit is used to extract the land use type if at least one element among all elements contains "trees" and its corresponding coverage value ij >=δ, then the land use type U corresponding to the current sampling point i Assign the value "forested land" if at least one element among all elements contains "trees" and its corresponding coverage has value ij <δ, then the land use type U corresponding to the current sampling point i The value is assigned as "sparse woodland", and δ is the preset threshold; The third land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and at least one element contains "shrubs". i Assign the value to "shrub forest"; The fourth land use type extraction unit is used to extract the land use type U corresponding to the current sampling point if all elements do not contain "trees" and "shrubs" and at least one element contains "herbs". i Assign the value to "grass".

Images