Hydrological forecasting model recommendation method and device based on literature bibliography and product

Abstract

The embodiment of the invention provides a hydrological forecasting model recommendation method, device and product based on literature bibliography, and relates to the technical field of hydrological forecasting. The method comprises the steps that multiple pieces of literature information related to hydrological forecasting models are acquired, and a target hydrological forecasting model corresponding to each piece of literature information is determined; key words corresponding to the multiple pieces of literature information are obtained, and corresponding similar words are determined; taking keywords and similar words respectively corresponding to the multiple pieces of literature information as training samples, taking target hydrological forecasting models respectively corresponding to the multiple pieces of literature information as labels, and training the initial classification model to obtain a trained classification model; determining related information of a to-be-modeled unfamiliar drainage basin and determining similar drainage basin information; and inputting the related information of the to-be-modeled unfamiliar watershed and / or the similar watershed information into the trained classification model to obtain a hydrological forecasting model recommended for the to-be-modeled unfamiliar watershed, so that the to-be-modeled unfamiliar watershed improves the scientificity of the selected model in the hydrological forecasting process.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of hydrological forecasting, and in particular to a method, device and product for recommending a hydrological forecasting model based on literature citations. Background Art

[0002] With the development of hydrological forecasting technology, hundreds of hydrological models have been proposed. Due to the complexity of the natural environment, there is no hydrological model that can perfectly simulate the runoff generation and confluence process of all river basins in the world. Each hydrological model has its own advantages and disadvantages, and can only be applied to a certain type of river basins with common characteristics.

[0003] At present, researchers or practitioners in the field of hydrological forecasting can only rely on experience and knowledge reserves to decide which model to use in a certain basin. For unfamiliar basins, they can only rely on experience and judgment, and it is difficult to accurately find a hydrological model suitable for the basin, which affects the forecast accuracy; or for unfamiliar basins, using multiple models for comparison is more labor-intensive and difficult to find the optimal model. Summary of the invention

[0004] Based on the above technical problems, the embodiments of the present invention provide a method, device and product for recommending a hydrological forecast model based on literature citations, aiming to construct a classifier, and recommend a hydrological forecast model suitable for an unfamiliar river basin to be modeled to users through a trained classification model, thereby improving the scientific nature of the model selected in the hydrological forecast process, saving user energy and improving user work efficiency on the basis of improving the accuracy of hydrological forecasts.

[0005] A first aspect of an embodiment of the present invention provides a method for recommending a hydrological forecast model based on literature citations, the method comprising:

[0006] Acquire multiple literature information related to the hydrological forecast model, and determine the target hydrological forecast model corresponding to each piece of literature information;

[0007] Obtain keywords corresponding to a plurality of the document information respectively;

[0008] Based on the characteristics of each watershed, similar words corresponding to the keyword are determined; the similar words corresponding to the keyword include: words that are semantically similar to the keyword, and / or words that are similar to the keyword in terms of watershed characteristics;

[0009] Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the initial classification model is trained to obtain a trained classification model;

[0010] Identify relevant information about the unfamiliar watershed to be modeled;

[0011] Based on the characteristics of each of the river basins, similar river basin information similar to the relevant information of the unfamiliar river basin to be modeled is determined; the similar river basin information includes: river basin information semantically similar to the relevant information of the unfamiliar river basin to be modeled, and / or river basin information similar to the relevant information of the unfamiliar river basin to be modeled in terms of river basin characteristics;

[0012] The relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information is input into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

[0013] Optionally, the keywords corresponding to the plurality of document information and similar words corresponding to the keywords are used as training samples, and the target hydrological forecast models corresponding to the plurality of document information are used as labels to train the initial classification model to obtain the trained classification model, including:

[0014] Dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a first training sample set and a second test sample set;

[0015] Using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label, training the initial classification model to obtain an intermediate classification model;

[0016] Using the first test sample set as input of the intermediate classification model, and determining the accuracy of the intermediate classification model based on the target hydrological prediction model corresponding to the first test sample set and the sample hydrological prediction model output by the intermediate classification model;

[0017] When the accuracy of the intermediate classification model is greater than or equal to the accuracy threshold, determining the intermediate classification model as the trained classification model;

[0018] When the accuracy of the intermediate classification model is less than the accuracy threshold, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords are re-divided into at least two parts, a second training sample set and a second test sample set, and the intermediate classification model is continuously trained until the trained classification model is obtained.

[0019] Optionally, the obtaining of keywords corresponding to the plurality of document information respectively includes:

[0020] Perform word segmentation and text vectorization processing on the plurality of document information respectively to obtain keywords corresponding to the plurality of document information respectively; wherein the keywords include at least one of the following: basin name, basin location, basin topographic features, and forecasting method;

[0021] The relevant information of the strange river basin to be modeled includes at least one of the following: the name of the strange river basin to be modeled, the location of the strange river basin to be modeled, the topographic features of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled.

[0022] Optionally, determining the target hydrological forecast model corresponding to each piece of literature information includes:

[0023] Marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to the level threshold by manual marking;

[0024] The used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold is determined as the target hydrological forecast model corresponding to each piece of literature information.

[0025] Optionally, the step of marking the hydrological forecast model used in each document information and / or the hydrological forecast model with an effect level greater than or equal to a level threshold by manual marking includes:

[0026] By manually annotating by multiple persons, a plurality of annotation results are obtained for each document information, each annotation result including: a hydrological forecast model used in each document information and / or a hydrological forecast model with an effect level greater than or equal to a level threshold;

[0027] A cross-validation is performed based on the multiple annotation results to obtain a final annotation result, and the final annotation result is determined as the target hydrological forecast model corresponding to each document information.

[0028] Optionally, the method further comprises:

[0029] Regularly search scientific and technological literature to obtain multiple new literature information related to hydrological forecast models, and determine the new target hydrological forecast model corresponding to each new literature information;

[0030] The keywords corresponding to the multiple new document information and similar words corresponding to the keywords are used as training samples, the new target hydrological forecast models corresponding to the multiple new document information are used as labels, and the model parameters of the trained classification model are adjusted to obtain an optimized classification model.

[0031] Optionally, the initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model; the initial classification model is trained using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and the target hydrological forecast models corresponding to the plurality of document information as labels, to obtain a trained classification model, including:

[0032] Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the support vector machine initial classification model is trained to obtain a trained support vector machine classification model;

[0033] Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the neural network initial classification model is trained to obtain a trained neural network classification model;

[0034] Evaluating the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results;

[0035] Based on the evaluation result, the trained classification model is determined from the trained support vector machine classification model and the trained neural network classification model.

[0036] A second aspect of an embodiment of the present invention provides a hydrological forecast model recommendation device based on literature citations, the device comprising:

[0037] A document acquisition module is used to acquire multiple document information related to the hydrological forecast model and determine the target hydrological forecast model corresponding to each document information;

[0038] A keyword acquisition module, used to acquire keywords corresponding to the plurality of document information respectively;

[0039] A similar word determination module is used to determine similar words corresponding to the keyword based on the characteristics of each watershed; the similar words corresponding to the keyword include: words with similar semantics to the keyword, and / or words similar to the keyword in terms of watershed characteristics;

[0040] A model training module is used to train the initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained classification model;

[0041] An information determination module, used to determine relevant information of an unfamiliar watershed to be modeled;

[0042] A similar information determination module is used to determine similar river basin information similar to the relevant information of the strange river basin to be modeled based on the characteristics of each river basin; the similar river basin information includes: river basin information with semantic similarity to the relevant information of the strange river basin to be modeled, and / or river basin information with similar river basin characteristics to the relevant information of the strange river basin to be modeled;

[0043] The model recommendation module is used to input the relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

[0044] Optionally, the model training module includes:

[0045] A sample division module, used for dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a first training sample set and a second test sample set;

[0046] An intermediate model training module, used to train the initial classification model using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label to obtain an intermediate classification model;

[0047] A sample testing module, used to use the first test sample set as the input of the intermediate classification model, and determine the accuracy of the intermediate classification model based on the target hydrological prediction model corresponding to the first test sample set and the sample hydrological prediction model output by the intermediate classification model;

[0048] A model determination module, configured to determine that the intermediate classification model is the trained classification model when the accuracy of the intermediate classification model is greater than or equal to an accuracy threshold;

[0049] The model acquisition module is used to, when the accuracy of the intermediate classification model is less than the accuracy threshold, re-divide the keywords corresponding to the multiple document information and the similar words corresponding to the keywords into at least two parts, a second training sample set and a second test sample set, and continue to train the intermediate classification model until the trained classification model is obtained.

[0050] Optionally, the keyword acquisition module includes:

[0051] A document keyword acquisition module is used to perform word segmentation and text vectorization processing on the plurality of document information respectively, and obtain keywords corresponding to the plurality of document information respectively; wherein the keywords include at least one of the following: basin name, basin location, basin terrain characteristics, and forecasting method;

[0052] The relevant information of the strange river basin to be modeled includes at least one of the following: the name of the strange river basin to be modeled, the location of the strange river basin to be modeled, the topographic features of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled.

[0053] Optionally, the document acquisition module includes:

[0054] A manual marking module, used for marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to a level threshold by manual marking;

[0055] The target model determination module is used to determine the used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold as the target hydrological forecast model corresponding to each document information.

[0056] Optionally, the manual labeling module includes:

[0057] A multi-person annotation module, used for obtaining multiple annotation results for each document information by manual annotation by multiple persons, each annotation result including: a hydrological forecast model used in each document information and / or a hydrological forecast model with an effect level greater than or equal to a level threshold;

[0058] The cross-validation module is used to perform cross-validation based on the multiple annotation results to obtain a final annotation result, and determine the final annotation result as the target hydrological forecast model corresponding to each document information.

[0059] Optionally, the device further comprises:

[0060] A regular acquisition module is used to regularly search scientific and technological literature, obtain multiple new literature information related to the hydrological forecast model, and determine the new target hydrological forecast model corresponding to each new literature information;

[0061] The model optimization module is used to use the keywords corresponding to the multiple new document information and the similar words corresponding to the keywords as training samples, and the new target hydrological forecast models corresponding to the multiple new document information as labels, to adjust the model parameters of the trained classification model to obtain the optimized classification model.

[0062] Optionally, the initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model; the model training module includes:

[0063] The first model training module is used to train the support vector machine initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained support vector machine classification model;

[0064] The second model training module is used to train the neural network initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained neural network classification model;

[0065] A model performance evaluation module is used to evaluate the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results;

[0066] A classification model determination module is used to determine the trained classification model from the trained support vector machine classification model and the trained neural network classification model based on the evaluation result.

[0067] A third aspect of an embodiment of the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the method for recommending a hydrological forecast model based on literature citations as described in the first aspect of an embodiment of the present invention is implemented.

[0068] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for recommending a hydrological forecast model based on literature citations according to the first aspect of an embodiment of the present invention is implemented.

[0069] According to the method for recommending a hydrological forecast model based on literature titles of an embodiment of the present invention, first, a plurality of literature information related to the hydrological forecast model is obtained, and a target hydrological forecast model corresponding to each literature information is determined; secondly, keywords corresponding to the plurality of literature information are obtained; then, based on the characteristics of each river basin, similarities corresponding to the keywords are determined; then, using keywords corresponding to the plurality of literature information and similar words corresponding to the keywords as training samples, and using the target hydrological forecast models corresponding to the plurality of literature information as labels, an initial classification model is trained to obtain a trained classification model; finally, relevant information of an unfamiliar river basin to be modeled is determined; based on the characteristics of each river basin, similar river basin information similar to the relevant information of the unfamiliar river basin to be modeled is determined, and the relevant information and / or similar river basin information of the unfamiliar river basin to be modeled is input into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

[0070] In this embodiment, the target hydrological forecast model is determined based on the literature information, and then the initial classification model is trained based on the keywords corresponding to the literature information, the similar words corresponding to the keywords and the target hydrological forecast model to obtain a trained classification model capable of recommending hydrological forecast models. The trained classification model is used to accurately recommend hydrological forecast models suitable for unfamiliar river basins to be modeled to users, thereby improving the scientific nature of the model selection in the hydrological forecast process to improve the accuracy of hydrological forecasts. In addition, there is no need to use multiple models for comparison for unfamiliar river basins, which saves user energy and improves the adaptability of hydrological forecast models while improving user work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0072] Figure 1 is a flow chart of a hydrological forecast model recommendation method based on literature citations according to an embodiment of the present invention;

[0073] Figure 2 It is a structural block diagram of a hydrological forecast model recommendation device based on literature citations provided by an embodiment of the present invention;

[0074] Figure 3 It is a schematic diagram of an electronic device shown in an embodiment of the present invention. DETAILED DESCRIPTION

[0075] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0076] Please refer to Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for recommending a hydrological forecast model based on literature citations according to an embodiment of the present invention. Figure 1 As shown, the method may include steps S11 to S14:

[0077] Step S11: Acquire multiple literature information related to the hydrological forecast model, and determine the target hydrological forecast model corresponding to each piece of literature information.

[0078] In this embodiment, multiple document information related to the hydrological forecast model can be obtained, and the document information can be retrieved from scientific and technological documents based on keywords related to the hydrological forecast model. The keywords related to the hydrological forecast model in this embodiment may include one or more of the following: hydrological forecast, hydrological model, flood forecast; scientific and technological documents include but are not limited to: papers, journals, and forums; document information includes but is not limited to: document titles and document abstracts of scientific and technological documents matching the keywords. In an optional implementation, the multiple document information obtained can be stored.

[0079] In this embodiment, after obtaining multiple document information, the target hydrological forecast model corresponding to each document information in the multiple document information can be determined, wherein the target hydrological forecast model is a hydrological forecast model used or recommended in the scientific and technological document corresponding to the document information.

[0080] Step S12: Obtain keywords corresponding to a plurality of the document information.

[0081] In this embodiment, multiple pieces of document information may be processed to obtain keywords corresponding to each piece of document information in the multiple pieces of document information.

[0082] Step S13: Based on the characteristics of each watershed, determine similar words corresponding to the keyword; the similar words corresponding to the keyword include: words that are semantically similar to the keyword, and / or words that are similar to the keyword in terms of watershed characteristics.

[0083] In this embodiment, after obtaining the keywords corresponding to the document information, similar words similar to the keywords corresponding to the document information can be determined based on the characteristics of each watershed. Among them, the word vector of the keyword can be determined, and then the characteristics of each watershed are used as similarity calculation factors to determine similar words with a similarity between the word vector that is higher than a first threshold. In this embodiment, the similar words corresponding to the keywords corresponding to the document information include: words that are semantically similar to the keywords, and / or words that are similar to the keywords in terms of watershed characteristics.

[0084] Step S14: using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, to train the initial classification model and obtain a trained classification model.

[0085] In this embodiment, after obtaining the keywords corresponding to multiple document information and the similar words corresponding to the keywords, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords can be used as training samples, and the target hydrological forecast models corresponding to the multiple document information can be used as labels to train the initial classification model to obtain a trained classification model.

[0086] Among them, the keywords corresponding to one document information and the similar words corresponding to the keywords in multiple document information can be used as training samples, and the target hydrological forecast model corresponding to the document information can be used as a label to form a group of training data, so as to form multiple groups of training data based on the keywords corresponding to multiple document information, the similar words corresponding to the keywords and the target hydrological forecast model, and obtain a training data set for the initial classification model, and train the initial classification model based on the training data set to obtain a trained classification model, which is used to recommend a hydrological forecast model according to the watershed information.

[0087] Step S14: Determine relevant information of the unfamiliar watershed to be modeled.

[0088] In this embodiment, relevant information of the unfamiliar river basin with modeling can be determined. In this embodiment, the unfamiliar river basin to be modeled is an unfamiliar river basin that needs to use a relevant hydrological forecast model for research or analysis; wherein, the unfamiliar river basin in this embodiment can refer to a river basin that the user is not familiar with in a subjective sense, or can refer to a newly discovered river basin in an objective sense, and this embodiment does not limit this.

[0089] Step S15: Based on the characteristics of each of the river basins, determine similar river basin information that is similar to the relevant information of the strange river basin to be modeled; the similar river basin information includes: river basin information that is semantically similar to the relevant information of the strange river basin to be modeled, and / or river basin information that is similar to the relevant information of the strange river basin to be modeled in terms of river basin characteristics.

[0090] In this embodiment, similar basin information similar to the relevant information of the strange basin to be modeled can be determined based on the characteristics of each basin. Wherein, similar to the above-mentioned step S13, the relevant information of the strange basin to be modeled can be subjected to text segmentation processing to determine the word vector of the relevant information of the strange basin to be modeled, and then the characteristics of each basin are used as similarity calculation factors to determine similar words whose similarity between the word vectors of the relevant information of the strange basin to be modeled is higher than the second threshold as similar basin information. In this embodiment, similar basin information similar to the relevant information of the strange basin to be modeled includes: basin information that is semantically similar to the relevant information of the strange basin to be modeled, and / or basin information that is similar to the relevant information of the strange basin to be modeled in terms of basin characteristics.

[0091] Step S16: inputting the relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

[0092] In this embodiment, after obtaining the trained classification model, for the unfamiliar river basin to be modeled, the relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information similar to the relevant information of the unfamiliar river basin to be modeled can be input into the trained classification model to obtain the hydrological forecast model recommended for the unfamiliar river basin to be modeled output by the trained classification model. The relevant information and / or similar river basin information of the unfamiliar river basin to be modeled in this embodiment corresponds to the training samples of the trained classification model (keywords corresponding to the literature information and their corresponding similar words).

[0093] In this embodiment, on the basis of scientific literature, the target hydrological forecast model is determined based on the literature information, and the initial classification model is trained based on the literature information and the target hydrological forecast model to obtain a trained classification model, so as to screen out the hydrological forecast models used on the same or similar river basins as the unfamiliar river basin to be modeled through the trained classification model, and recommend them to the modeling personnel, thereby improving the scientific nature of the model selection in the hydrological forecast process to improve the accuracy of the hydrological forecast, and there is no need to use multiple models for comparison for the unfamiliar river basin to be modeled, which saves the user's energy. On the basis of improving the user's work efficiency, the adaptability of the hydrological forecast model is improved.

[0094] In an optional embodiment, based on the multiple document information obtained, a first number (freely settable, such as 1,000 articles) of document information can be randomly selected, and the first number of document information can be processed to obtain keywords corresponding to each document information in the first number of document information. Then, based on the keywords corresponding to each document information in the first number of document information and similar words corresponding to the keywords, and the target hydrological forecast model corresponding to each document information in the first number of document information, an initial classification model can be trained to obtain a trained classification model.

[0095] In combination with the above embodiments, in one implementation, the present invention further provides a hydrological forecast model recommendation method based on literature citations. In this method, the above step S13 may specifically include steps S21 to S25:

[0096] Step S21: divide the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts: a first training sample set and a second test sample set.

[0097] In this embodiment, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords can be divided into at least two parts: a first training sample set and a second test sample set. The first training sample set is a sample set for training the classification model, and the second test sample set is a sample set for testing and verifying the classification model.

[0098] The ratio of the document information in the first training sample set and the second test sample set can be set arbitrarily, such as taking the keywords corresponding to half of the document information in the multiple document information as the first training sample set, and taking the keywords corresponding to the remaining half of the document information as the second test sample set; or taking the keywords corresponding to two-thirds of the document information in the multiple document information as the first training sample set, and taking the keywords corresponding to the remaining one-third of the document information as the second test sample set, and there is no restriction on this.

[0099] Step S22: using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label, training the initial classification model to obtain an intermediate classification model.

[0100] In this embodiment, after obtaining the first training sample set and the second test sample set, the first training sample set can be used as the training sample of the initial classification model, and the target hydrological forecast model corresponding to the first training sample set can be used as a label to train the initial classification model to obtain an intermediate classification model. The intermediate model is a classification model obtained by training based on the first training sample set.

[0101] For example, in the first training sample set, a keyword corresponding to a document information and a similar word corresponding to the keyword can be used as a training sample, and the target hydrological forecast model corresponding to the document information can be used as a label to form a group of training data. Thus, based on the keywords corresponding to the keywords and the similar words corresponding to the keywords and the target hydrological forecast model respectively corresponding to the first training sample set, multiple groups of training data can be formed to train the initial classification model to obtain a trained classification model.

[0102] Step S23: using the first test sample set as the input of the intermediate classification model, and determining the accuracy of the intermediate classification model based on the target hydrological forecast model corresponding to the first test sample set and the sample hydrological forecast model output by the intermediate classification model.

[0103] In this embodiment, after obtaining the intermediate classification model, the first test sample set can be used as the input of the intermediate classification model and input into the intermediate classification model to obtain the output of the intermediate classification model: a sample hydrological forecast model, which is a hydrological forecast model predicted by the intermediate classification model based on the first test sample set.

[0104] Based on the target hydrological forecast model corresponding to the first test sample set and the sample hydrological forecast model output by the intermediate classification model, the accuracy of the intermediate classification model is determined. For example, the keywords corresponding to each document information in the first test sample set and the similar words corresponding to the keywords can also be input into the intermediate classification model to obtain the sample hydrological forecast model corresponding to each document information in the first test sample set output by the intermediate classification model, and then based on the target hydrological forecast model corresponding to each document information in the first test sample set and the sample hydrological forecast model corresponding to each document information in the first test sample set, the accuracy of the intermediate classification model is determined. For example, the accuracy of the intermediate classification model can be determined by comparing the similarity, overlap rate, etc. between the target hydrological forecast model and the sample hydrological forecast model, and the accuracy represents the prediction accuracy of the intermediate classification model.

[0105] Step S24: when the accuracy of the intermediate classification model is greater than or equal to the accuracy threshold, determine that the intermediate classification model is the trained classification model.

[0106] In this embodiment, after obtaining the accuracy of the intermediate classification model, the size of the intermediate classification model and the accuracy threshold can be compared. When it is determined that the accuracy of the intermediate classification model is greater than or equal to the accuracy threshold, the intermediate classification model can be determined to be a trained classification model. In this embodiment, the accuracy threshold can be set in advance, and the specific value of the accuracy threshold can be freely set according to demand, for example, the accuracy threshold is 80%, 90%, etc., and there is no limitation on this.

[0107] Step S25: When the accuracy of the intermediate classification model is less than the accuracy threshold, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords are re-divided into at least two parts, a second training sample set and a second test sample set, and the intermediate classification model is continued to be trained until the trained classification model is obtained.

[0108] In this embodiment, when it is determined that the accuracy of the intermediate classification model is less than the accuracy threshold, it is characterized that the prediction accuracy of the intermediate classification model is low and the model training needs to be continued. At this time, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords can be re-divided into at least two parts: a second training sample set and a second test sample set. Then, the second training sample set is used as the first training sample set in the above steps, and the second test sample set is used as the first training sample set in the above steps. The above steps S22-S25 are repeated to continue training the intermediate classification model obtained in the previous round until the trained classification model is obtained.

[0109] In this embodiment, the classification model can be trained using the divided training data and verified on the divided verification data, and a classification model with higher classification accuracy can be obtained through methods such as model comparison and parameter tuning.

[0110] In addition, in combination with the above embodiments, in an optional implementation, a second number of document information may be randomly selected from a plurality of acquired document information related to the hydrological forecast model to train the classification model, such as using keywords corresponding to a portion of the document information in the second number of document information and similar words corresponding to the keywords as training samples, and using a portion of the target hydrological forecast model corresponding to the document information as a label to train an intermediate classification model, and then using keywords corresponding to the remaining part of the document information in the second number of document information and similar words corresponding to the keywords as inputs to the intermediate classification model to obtain a sample hydrological forecast model output by the intermediate classification model, and then verifying the intermediate classification model using the target hydrological forecast model and the sample hydrological forecast model corresponding to the remaining part of the document information in the second number of document information to obtain a trained classification model. Next, among multiple literature information related to the hydrological forecast model, the keywords corresponding to the remaining literature information after the second number of literature information are randomly selected and the similar words corresponding to the keywords are input into the trained classification model, and the output results of the trained classification model are sampled and checked to determine whether the recommended hydrological forecast model is the target hydrological forecast model corresponding to the literature information. If the accuracy is insufficient, repeat the above steps until a classification model that meets the conditions is obtained.

[0111] In combination with the above embodiments, in one implementation, the present invention further provides a hydrological forecast model recommendation method based on literature citations. In the method, the above step S12 may specifically include step S31:

[0112] Step S31: performing word segmentation and text vectorization processing on the plurality of document information to obtain keywords corresponding to the plurality of document information respectively.

[0113] In this embodiment, the document titles or document abstracts in the retrieved scientific and technological documents generally include: basin name, basin location, basin terrain features, forecasting methods, etc. After obtaining multiple document information, the multiple document information can be segmented and vectorized to obtain keywords corresponding to the multiple document information. Among them, the keywords corresponding to the document information in this embodiment include at least one of the following: basin name, basin location, basin terrain features, forecasting methods.

[0114] Among them, in this embodiment, the relevant information of the strange river basin to be modeled includes at least one or more of the following: the river basin name of the strange river basin to be modeled, the river basin location of the strange river basin to be modeled, the river basin terrain characteristics of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled (such as flood forecasting, water resources forecasting, etc.), so as to obtain the strange river basin to be modeled recommended by the trained classification model and / or the hydrological forecasting model applicable to the forecasting method.

[0115] In combination with the above embodiments, the present invention further provides a method for recommending a hydrological forecast model based on literature titles. In the method, the "determining a target hydrological forecast model corresponding to each literature information" in the above step S11 may specifically include steps S41 to S42:

[0116] Step S41: manually marking the hydrological forecast model used in each document information and / or the hydrological forecast model with an effect level greater than or equal to a level threshold.

[0117] In this embodiment, after obtaining the literature information, the hydrological forecast model used in each literature information and / or the hydrological forecast model with an effect level greater than or equal to the level threshold can be marked by manual annotation. The hydrological forecast model used is the hydrological forecast model used for the relevant basin recorded in the literature information.

[0118] In this embodiment, the hydrological forecast model can be divided into effect levels to characterize the use effect of the hydrological forecast model. The level threshold of this embodiment is the minimum effect level that characterizes the good effect of the hydrological forecast model, which can be freely set. The hydrological forecast model with an effect level greater than or equal to the level threshold in the literature information of this embodiment is a hydrological forecast model with good use effect for the relevant basin recorded in the literature information.

[0119] Step S42: Determine the used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold as the target hydrological forecast model corresponding to each document information.

[0120] In this embodiment, after determining the hydrological forecast model used in each document information and / or the hydrological forecast model with an effect level greater than or equal to the level threshold, the hydrological forecast model used in each document information and / or the hydrological forecast model with an effect level greater than or equal to the level threshold can be determined as the target hydrological forecast model corresponding to each document information.

[0121] In combination with the above embodiments, in one implementation, the present invention further provides a hydrological forecast model recommendation method based on literature citations. In this method, the above step S41 may specifically include steps S51 to S52:

[0122] Step S51: multiple people manually annotate each document information to obtain multiple annotation results.

[0123] In this embodiment, the hydrological forecast model can be annotated for each document information by multiple people manually annotating, and multiple annotation results corresponding to each document information can be obtained. In this embodiment, each of the multiple annotation results includes: the hydrological forecast model used in each document information and / or the hydrological forecast model with an effect level greater than or equal to a level threshold.

[0124] Step S52: performing cross-validation based on the multiple annotation results to obtain a final annotation result, and determining the final annotation result as the target hydrological forecast model corresponding to each document information.

[0125] In this embodiment, a plurality of annotation results corresponding to each document information are cross-validated to obtain a final annotation result, and the final annotation result is thus determined as the target hydrological forecast model corresponding to each document information.

[0126] In this embodiment, multiple experts or annotators independently annotate to reduce subjective errors. In addition, in one embodiment, when randomly extracting document information, the diversity of document sources of the document information is ensured as much as possible, and document information from different regions, different research institutions and different time periods can be included to increase the generalization ability of the classification model.

[0127] In combination with the above embodiments, in one implementation, the present invention further provides a hydrological forecast model recommendation method based on literature citations. In the method, in addition to the above steps, steps S61 and S62 may also be included:

[0128] Step S61: regularly search scientific and technological literature to obtain a plurality of new literature information related to the hydrological forecast model, and determine a new target hydrological forecast model corresponding to each new literature information.

[0129] In this embodiment, the search cycle can be freely set, and the latest scientific and technological documents are regularly searched based on the search cycle to obtain multiple new document information related to the hydrological forecast model, and the new target hydrological forecast model corresponding to each new document information is determined. Among them, the latest scientific and technological documents are multiple scientific and technological documents from recent to distant in terms of publication time, the new document information is the newly obtained document information, and the new target hydrological forecast model is the target hydrological forecast model corresponding to the new document information.

[0130] Step S62: Using keywords corresponding to the multiple new document information and similar words corresponding to the keywords as training samples, using new target hydrological forecast models corresponding to the multiple new document information as labels, adjusting the model parameters of the trained classification model to obtain an optimized classification model.

[0131] In this embodiment, after obtaining the new document information, the keywords corresponding to the new document can be determined, and then the keywords corresponding to multiple new document information and similar words corresponding to the keywords are used as training samples, and the new target hydrological forecast models corresponding to multiple new document information are used as labels. The trained classification model is optimized, and the model parameters of the trained classification model are adjusted to obtain the optimized classification model.

[0132] In this embodiment, as new documents are continuously published, the document database and classification model are regularly updated to maintain the timeliness and accuracy of the classification model and recommend a more suitable hydrological forecast model for the unfamiliar watershed to be modeled.

[0133] In combination with the above embodiments, in one implementation, the present invention further provides a hydrological forecast model recommendation method based on literature citations. In this method, the above step S14 may specifically include steps S71 to S74:

[0134] Step S71: Using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using the target hydrological forecast models corresponding to the plurality of document information as labels, the support vector machine initial classification model is trained to obtain a trained support vector machine classification model.

[0135] In this embodiment, the initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model. The keywords corresponding to the plurality of document information and the similar words corresponding to the keywords can be used as training samples, and the target hydrological forecast models corresponding to the plurality of document information can be used as labels to train the support vector machine initial classification model to obtain a trained support vector machine classification model.

[0136] Step S72: Using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using the target hydrological forecast models corresponding to the plurality of document information as labels, the neural network initial classification model is trained to obtain a trained neural network classification model.

[0137] In this embodiment, the keywords corresponding to multiple document information and similar words corresponding to each keyword can be used as training samples, and the target hydrological forecast models corresponding to the multiple document information can be used as labels to train the neural network initial classification model to obtain a trained neural network classification model.

[0138] Step S73: Evaluate the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results.

[0139] In this embodiment, the classification model is trained by two different classification algorithms: a neural network classification algorithm and a support vector machine classification algorithm, so as to obtain a trained support vector machine classification model and a trained neural network classification model. After obtaining the trained support vector machine classification model and the trained neural network classification model, the model performance of the trained support vector machine classification model can be evaluated by cross-validation, and the model performance of the trained neural network classification model can be evaluated to obtain an evaluation result.

[0140] Step S74: Based on the evaluation result, determine the trained classification model from the trained support vector machine classification model and the trained neural network classification model.

[0141] In this embodiment, based on the evaluation results obtained by cross-validation, the classification model with better model performance corresponding to the evaluation results can be determined as the trained classification model from the trained support vector machine classification model and the trained neural network classification model. For example, if the evaluation results of the cross-validation show that the model performance of the trained support vector machine classification model is better, then the trained support vector machine classification model is determined to be the trained classification model; if the evaluation results of the cross-validation show that the model performance of the trained neural network classification model is better, then the trained neural network classification model is determined to be the trained classification model.

[0142] The embodiment of the present invention takes into account that selecting a suitable hydrological forecast model for a river basin is very important for the hydrological forecast accuracy of the river basin, but the current methods for selecting hydrological forecast models mostly rely on experience, and for unfamiliar river basins, only experience can be relied on. Based on this, this embodiment, based on scientific literature, constructs a classifier to screen out hydrological models used in the same or similar river basins as the unfamiliar river basin to be modeled, and recommends them to modelers (staff who use hydrological forecast models for research, analysis or other related business work), thereby improving the scientific nature of the model selected in the hydrological forecast process and helping to improve the accuracy of hydrological forecasts.

[0143] It should be noted that, for the sake of simplicity, the method embodiments are described as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present invention are not limited by the order of the actions described, because according to the embodiments of the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily required by the embodiments of the present invention.

[0144] Based on the same inventive concept, an embodiment of the present invention provides a hydrological forecast model recommendation device 200 based on literature titles. Figure 2 , Figure 2 FIG. 1 is a structural block diagram of a hydrological forecast model recommendation device based on literature citations provided by an embodiment of the present invention. Figure 2 As shown, the device 200 includes:

[0145] The document acquisition module 201 is used to acquire multiple document information related to the hydrological forecast model and determine the target hydrological forecast model corresponding to each document information;

[0146] A keyword acquisition module 202, used to acquire keywords corresponding to the plurality of document information;

[0147] The similar word determination module 203 is used to determine similar words corresponding to the keyword based on the characteristics of each watershed; the similar words corresponding to the keyword include: words with similar semantics to the keyword, and / or words similar to the keyword in terms of watershed characteristics;

[0148] The model training module 204 is used to train the initial classification model using the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained classification model;

[0149] An information determination module 205 is used to determine relevant information of the unfamiliar watershed to be modeled;

[0150] The similar information determination module 206 is used to determine similar river basin information similar to the relevant information of the unfamiliar river basin to be modeled based on the characteristics of each river basin; the similar river basin information includes: river basin information semantically similar to the relevant information of the unfamiliar river basin to be modeled, and / or river basin information similar to the relevant information of the unfamiliar river basin to be modeled in terms of river basin characteristics;

[0151] The model recommendation module 207 is used to input the relevant information of the unfamiliar river basin to be modeled into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

[0152] Optionally, the model training module 204 includes:

[0153] A sample division module, used for dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a first training sample set and a second test sample set;

[0154] An intermediate model training module, used to train the initial classification model using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label to obtain an intermediate classification model;

[0155] A sample testing module, used to use the first test sample set as the input of the intermediate classification model, and determine the accuracy of the intermediate classification model based on the target hydrological prediction model corresponding to the first test sample set and the sample hydrological prediction model output by the intermediate classification model;

[0156] A model determination module, configured to determine that the intermediate classification model is the trained classification model when the accuracy of the intermediate classification model is greater than or equal to an accuracy threshold;

[0157] The model acquisition module is used to, when the accuracy of the intermediate classification model is less than the accuracy threshold, re-divide the keywords corresponding to the multiple document information and the similar words corresponding to the keywords into at least two parts, a second training sample set and a second test sample set, and continue to train the intermediate classification model until the trained classification model is obtained.

[0158] Optionally, the keyword acquisition module 202 includes:

[0159] A document keyword acquisition module is used to perform word segmentation and text vectorization processing on the plurality of document information respectively, and obtain keywords corresponding to the plurality of document information respectively; wherein the keywords include at least one of the following: basin name, basin location, basin terrain characteristics, and forecasting method;

[0160] The relevant information of the strange river basin to be modeled includes at least one of the following: the name of the strange river basin to be modeled, the location of the strange river basin to be modeled, the topographic features of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled.

[0161] Optionally, the document acquisition module 201 includes:

[0162] A manual marking module, used for marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to a level threshold by manual marking;

[0163] The target model determination module is used to determine the used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold as the target hydrological forecast model corresponding to each document information.

[0164] Optionally, the manual labeling module includes:

[0165] A multi-person annotation module, used for obtaining multiple annotation results for each document information by manual annotation by multiple persons, each annotation result including: a hydrological forecast model used in each document information and / or a hydrological forecast model with an effect level greater than or equal to a level threshold;

[0166] The cross-validation module is used to perform cross-validation based on the multiple annotation results to obtain a final annotation result, and determine the final annotation result as the target hydrological forecast model corresponding to each document information.

[0167] Optionally, the device 200 further includes:

[0168] A regular acquisition module is used to regularly search scientific and technological literature, obtain multiple new literature information related to the hydrological forecast model, and determine the new target hydrological forecast model corresponding to each new literature information;

[0169] The model optimization module is used to use the keywords corresponding to the multiple new document information and the similar words corresponding to the keywords as training samples, and the new target hydrological forecast models corresponding to the multiple new document information as labels, to adjust the model parameters of the trained classification model to obtain the optimized classification model.

[0170] Optionally, the initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model; the model training module 204 includes:

[0171] The first model training module is used to train the support vector machine initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained support vector machine classification model;

[0172] The second model training module is used to train the neural network initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained neural network classification model;

[0173] A model performance evaluation module is used to evaluate the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results;

[0174] A classification model determination module is used to determine the trained classification model from the trained support vector machine classification model and the trained neural network classification model based on the evaluation result.

[0175] Based on the same inventive concept, another embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the hydrological forecast model recommendation method based on literature citations as described in any of the above embodiments of the present invention are implemented.

[0176] Based on the same inventive concept, another embodiment of the present invention provides an electronic device 300, such as Figure 3 shown. Figure 330 is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device includes a memory 302, a processor 301, and a computer program stored in the memory and executable on the processor, and when executed by the processor, the steps in the hydrological forecast model recommendation method based on literature citations described in any of the above embodiments of the present invention are implemented.

[0177] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0178] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0179] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, embodiments of the present invention may take the form of complete hardware embodiments, complete software embodiments, or embodiments combining software and hardware. Furthermore, embodiments of the present invention may take the form of computer program products implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0180] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0181] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0182] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0183] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0184] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal 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 terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0185] The above is a detailed introduction to a hydrological forecast model recommendation method, device and product based on literature citations provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A hydrological forecast model recommendation method based on literature citations, characterized in that: The method comprises: Acquire multiple literature information related to the hydrological forecast model, and determine the target hydrological forecast model corresponding to each piece of literature information; Obtain keywords corresponding to a plurality of the document information respectively; Based on the characteristics of each watershed, similar words corresponding to the keyword are determined; the similar words corresponding to the keyword include: words that are semantically similar to the keyword, and / or words that are similar to the keyword in terms of watershed characteristics; Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the initial classification model is trained to obtain a trained classification model; Identify relevant information about the unfamiliar watershed to be modeled; Based on the characteristics of each of the river basins, similar river basin information similar to the relevant information of the unfamiliar river basin to be modeled is determined; the similar river basin information includes: river basin information semantically similar to the relevant information of the unfamiliar river basin to be modeled, and / or river basin information similar to the relevant information of the unfamiliar river basin to be modeled in terms of river basin characteristics; The relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information is input into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

2. The hydrological forecast model recommendation method based on literature citation according to claim 1, characterized in that: The method uses the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords as training samples, uses the target hydrological forecast models corresponding to the plurality of document information as labels, trains the initial classification model, and obtains the trained classification model, including: Dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a first training sample set and a second test sample set; Using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label, training the initial classification model to obtain an intermediate classification model; Using the first test sample set as input of the intermediate classification model, and determining the accuracy of the intermediate classification model based on the target hydrological prediction model corresponding to the first test sample set and the sample hydrological prediction model output by the intermediate classification model; When the accuracy of the intermediate classification model is greater than or equal to the accuracy threshold, determining the intermediate classification model as the trained classification model; When the accuracy of the intermediate classification model is less than the accuracy threshold, the keywords corresponding to the multiple document information and the similar words corresponding to the keywords are re-divided into at least two parts, a second training sample set and a second test sample set, and the intermediate classification model is continuously trained until the trained classification model is obtained.

3. The hydrological forecast model recommendation method based on literature citation according to claim 1, characterized in that: The obtaining of keywords corresponding to the plurality of document information respectively includes: Perform word segmentation and text vectorization processing on the plurality of document information respectively to obtain keywords corresponding to the plurality of document information respectively; wherein the keywords include at least one of the following: basin name, basin location, basin topographic features, and forecasting method; The relevant information of the strange river basin to be modeled includes at least one of the following: the name of the strange river basin to be modeled, the location of the strange river basin to be modeled, the topographic features of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled.

4. The hydrological forecast model recommendation method based on literature citation according to claim 1, characterized in that: Determining the target hydrological forecast model corresponding to each piece of literature information includes: Marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to the level threshold by manual marking; The used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold is determined as the target hydrological forecast model corresponding to each piece of literature information.

5. The hydrological forecast model recommendation method based on literature citation according to claim 4, characterized in that: The step of manually marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to the level threshold includes: By manually annotating by multiple persons, a plurality of annotation results are obtained for each document information, each annotation result including: a hydrological forecast model used in each document information and / or a hydrological forecast model with an effect level greater than or equal to a level threshold; A cross-validation is performed based on the multiple annotation results to obtain a final annotation result, and the final annotation result is determined as the target hydrological forecast model corresponding to each document information.

6. The hydrological forecast model recommendation method based on literature citation according to any one of claims 1 to 5, characterized in that: The method further comprises: Regularly search scientific and technological literature to obtain multiple new literature information related to hydrological forecast models, and determine the new target hydrological forecast model corresponding to each new literature information; The keywords corresponding to the multiple new document information and similar words corresponding to the keywords are used as training samples, the new target hydrological forecast models corresponding to the multiple new document information are used as labels, and the model parameters of the trained classification model are adjusted to obtain an optimized classification model.

7. The hydrological forecast model recommendation method based on literature citation according to claim 1, characterized in that: The initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model; the keywords corresponding to the plurality of document information and similar words corresponding to the keywords are used as training samples, and the target hydrological forecast models corresponding to the plurality of document information are used as labels to train the initial classification model, and obtain a trained classification model, including: Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the support vector machine initial classification model is trained to obtain a trained support vector machine classification model; Using keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using target hydrological forecast models corresponding to the plurality of document information as labels, the neural network initial classification model is trained to obtain a trained neural network classification model; Evaluating the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results; Based on the evaluation result, the trained classification model is determined from the trained support vector machine classification model and the trained neural network classification model.

8. A hydrological forecast model recommendation device based on literature citation, characterized in that: The device comprises: A document acquisition module is used to acquire multiple document information related to the hydrological forecast model and determine the target hydrological forecast model corresponding to each document information; A keyword acquisition module, used to acquire keywords corresponding to the plurality of document information respectively; A similar word determination module is used to determine similar words corresponding to the keyword based on the characteristics of each watershed; the similar words corresponding to the keyword include: words with similar semantics to the keyword, and / or words similar to the keyword in terms of watershed characteristics; A model training module is used to train the initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples, and using the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained classification model; An information determination module, used to determine relevant information of an unfamiliar watershed to be modeled; A similar information determination module is used to determine similar river basin information similar to the relevant information of the strange river basin to be modeled based on the characteristics of each river basin; the similar river basin information includes: river basin information with semantic similarity to the relevant information of the strange river basin to be modeled, and / or river basin information with similar river basin characteristics to the relevant information of the strange river basin to be modeled; The model recommendation module is used to input the relevant information of the unfamiliar river basin to be modeled and / or the similar river basin information into the trained classification model to obtain a hydrological forecast model recommended for the unfamiliar river basin to be modeled.

9. The hydrological forecast model recommendation device based on literature citation according to claim 8, characterized in that: The model training module includes: A sample set division module, used for dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a first training sample set and a second test sample set; An intermediate model training module, used to train the initial classification model using the first training sample set as training samples and the target hydrological forecast model corresponding to the first training sample set as a label to obtain an intermediate classification model; A model testing module, used to use the first test sample set as the input of the intermediate classification model, and determine the accuracy of the intermediate classification model based on the target hydrological prediction model corresponding to the first test sample set and the sample hydrological prediction model output by the intermediate classification model; A first module determination module is used to determine that the intermediate classification model is the trained classification model when the accuracy of the intermediate classification model is greater than or equal to the accuracy threshold; The second module determines a module for re-dividing the keywords corresponding to the plurality of document information and the similar words corresponding to the keywords into at least two parts, a second training sample set and a second test sample set, when the accuracy of the intermediate classification model is less than the accuracy threshold, and continuing to train the intermediate classification model until the trained classification model is obtained.

10. The hydrological forecast model recommendation device based on literature citation according to claim 8, characterized in that: The keyword acquisition module includes: A document keyword acquisition module is used to perform word segmentation and text vectorization processing on the plurality of document information respectively, and obtain keywords corresponding to the plurality of document information respectively; wherein the keywords include at least one of the following: basin name, basin location, basin terrain characteristics, and forecasting method; The relevant information of the strange river basin to be modeled includes at least one of the following: the name of the strange river basin to be modeled, the location of the strange river basin to be modeled, the topographic features of the strange river basin to be modeled, and the forecasting method of the strange river basin to be modeled.

11. The hydrological forecast model recommendation device based on literature citation according to claim 8, characterized in that: The document acquisition module comprises: A manual marking module, used for marking the hydrological forecast model used in each document information and / or the hydrological forecast model whose effect level is greater than or equal to a level threshold by manual marking; The target model determination module is used to determine the used hydrological forecast model and / or the hydrological forecast model with an effect level greater than or equal to a level threshold as the target hydrological forecast model corresponding to each document information.

12. The hydrological forecast model recommendation device based on literature citation according to claim 11, characterized in that: The manual annotation module comprises: A multi-person annotation module, used for obtaining multiple annotation results for each document information by manual annotation by multiple persons, each annotation result including: a hydrological forecast model used in each document information and / or a hydrological forecast model with an effect level greater than or equal to a level threshold; The cross-validation module is used to perform cross-validation based on the multiple annotation results to obtain a final annotation result, and determine the final annotation result as the target hydrological forecast model corresponding to each document information.

13. The hydrological forecast model recommendation device based on literature citation according to any one of claims 8 to 12, characterized in that: The device also includes: A regular acquisition module is used to regularly search scientific and technological literature, obtain multiple new literature information related to the hydrological forecast model, and determine the new target hydrological forecast model corresponding to each new literature information; The model optimization module is used to use the keywords corresponding to the multiple new document information and the similar words corresponding to the keywords as training samples, and the new target hydrological forecast models corresponding to the multiple new document information as labels, to adjust the model parameters of the trained classification model to obtain the optimized classification model.

14. The hydrological forecast model recommendation device based on literature citation according to claim 8, characterized in that: The initial classification model at least includes: a support vector machine initial classification model and a neural network initial classification model; the model training module includes: The first model training module is used to train the support vector machine initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained support vector machine classification model; The second model training module is used to train the neural network initial classification model using the keywords corresponding to the plurality of document information and similar words corresponding to the keywords as training samples and the target hydrological forecast models corresponding to the plurality of document information as labels to obtain a trained neural network classification model; A model performance evaluation module is used to evaluate the model performance of the trained support vector machine classification model and the trained neural network classification model respectively by cross-validation to obtain evaluation results; A classification model determination module is used to determine the trained classification model from the trained support vector machine classification model and the trained neural network classification model based on the evaluation result.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by the processor, the method for recommending a hydrological forecast model based on literature citations as described in any one of claims 1 to 7 is implemented.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for recommending a hydrological forecast model based on literature citations as described in any one of claims 1 to 7 is implemented.

Images