A multi-dimensional evaluation method, system, medium and equipment for recreational evaluation of forest parks

Abstract

The present invention discloses a multi-dimensional evaluation method, system, medium and equipment for evaluating the recreation degree of a forest park. The method comprehensively considers three key dimensions, namely environment, service and society, deeply analyzes the factors affecting the recreation degree from multiple angles, integrates multiple data sources such as environmental image information, service information and social information, establishes a comprehensive evaluation system, and adopts a PLS-SEM model for comprehensive evaluation. The method can simultaneously evaluate the reliability of indicators in each dimension and the influence relationship between potential variables. According to the model analysis results, improvement suggestions such as the second environmental feature, the second service feature and the second social feature are put forward in a targeted manner, the supporting services and management framework of the forest park are optimized, the level of cultural services is improved, the supply and demand relationship is periodically evaluated and adjusted, and the emotional scheduling effect of product services in the recreation degree of the entire forest park is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tourism industry, and in particular to a method, system, medium and equipment for evaluating the recreation degree of a forest park in multiple dimensions. Background Art

[0002] Cultural services are an important part of the park's multiple functions, including aesthetics, recreation and ecotourism, education, society, spirit and religion. From the supply side, park cultural services are jointly produced by the park's natural ecosystem, management agencies and tourism service industry, and provide a series of cultural services to visitors; from the demand side, they mostly refer to the form of welfare benefits that humans obtain from the park ecosystem.

[0003] However, there are many problems in the current delivery of park cultural services from the supply side to the demand side: First, due to the characteristics of park ecological protection and visitor capacity restrictions, there is a spatial configuration mismatch between the supply side and the demand side of cultural services; second, the supply side and the demand side involve multiple stakeholders such as the government, the market, and the society. Due to the different interests of the participants and the complex and constantly changing relationships between them, the supply and demand validity does not match, which may lead to the adverse consequence of value co-destruction; third, the lack of innovation in cultural service products has led to low quality of the entire cultural service product system, especially the low tourism experience, which has limited contribution to enhancing the value of park ecological products. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a method, system, medium and equipment for evaluating the recreation degree of a forest park in multiple dimensions.

[0005] In order to achieve the above technical objectives, the technical solution adopted by the present invention is:

[0006] In a first aspect, the present invention provides a multi-dimensional method for evaluating the recreation degree of a forest park, comprising:

[0007] Acquire environmental image information of the current forest park, the environmental image information includes panoramic image information and local image information, the local image information is an image containing part of the environment, and generate a first environmental feature according to the environmental image information;

[0008] and obtaining a plurality of service information of the current forest park, each service information including a service category, a service group, a service address and service item information of the current service point, the service item information including a service commodity, a service item and a service price of the current service point, and generating a first service feature according to the plurality of service information;

[0009] and, obtaining social information of the current forest park, the social information including visitor comment information, management change information and park information of the current forest park, and generating a first social feature of the current forest park according to the social information;

[0010] Construct a PLS-SEM model, which includes a measurement model and a structural model. The environmental factor is used as the first latent variable of the PLS-SEM model, the service factor is used as the second latent variable of the PLS-SEM model, the social factor is used as the third latent variable of the PLS-SEM model, the first environmental characteristic is used as the first observed variable corresponding to the first latent variable, the first service characteristic is used as the second observed variable corresponding to the second latent variable, and the first social characteristic is used as the third observed variable corresponding to the third latent variable to construct the measurement model.

[0011] The first hypothesis relationship between the first latent variable and the recreation degree is set, the second hypothesis relationship between the second latent variable and the recreation degree is set, and the third hypothesis relationship between the third latent variable and the recreation degree is set. The structural model is constructed according to the first hypothesis relationship, the second hypothesis relationship and the third hypothesis relationship.

[0012] quantifying the first latent variable according to the first observed variable in the measurement model to obtain a first quantification result, quantifying the second latent variable according to the second observed variable in the measurement model to obtain a second quantification result, and quantifying the third latent variable according to the third observed variable in the measurement model to obtain a third quantification result;

[0013] The first quantitative result, the second quantitative result and the third quantitative result are input into the structural model, and the goodness of fit index of the first latent variable, the second latent variable and the third latent variable to the recreation degree is calculated;

[0014] The recreation degree evaluation results and improvement suggestions of the current forest park are obtained. The recreation degree evaluation results include the quantitative contribution of the first environmental characteristics, the first service characteristics and the first social characteristics to the recreation degree. The improvement suggestions include the second environmental characteristics, the second service characteristics and the second social characteristics.

[0015] In some embodiments, generating the first environmental feature according to the environmental image information includes:

[0016] Performing image preprocessing on the environmental image information, the image preprocessing includes image correction and image enhancement, to obtain a first image information group and a second image information group, the first image information group includes a plurality of first image information, the plurality of first image information is generated after the panoramic image information is preprocessed, and the second image information group includes a plurality of second image information, the plurality of second image information is generated after the local image information is preprocessed;

[0017] Performing image segmentation on the first image information, and inputting the segmentation result of the first image information into an image classifier for classification and labeling, to obtain a plurality of first tourist spot information corresponding to the current forest park, wherein the first tourist spot information includes geographic coordinate information of the current first tourist spot and a first image feature;

[0018] and performing image segmentation on the second image information, and inputting the segmentation result of the second image information into an image classifier for classification and labeling, so as to obtain a plurality of second tourist spot information corresponding to the current forest park, wherein the second tourist spot information includes geographic coordinate information of the current second tourist spot and second image features;

[0019] Matching the second tourist spot information with the first tourist spot information according to the geographical location information;

[0020] If the second play point information matches the first play point information successfully, the first play point is recorded as the third play point, and the third play point information is generated, and the third play point information includes the first image feature and the second image feature of the current third play point and the geographic coordinate information of the third play point;

[0021] If the second play point information is not matched with the first play point information, the second play point is recorded as a fourth play point, and the fourth play point information is generated, where the fourth play point information includes the second image feature and the geographical location coordinate information of the fourth play point;

[0022] The third tourist spot information and the fourth tourist spot information are input into the trained neural network model to obtain the first environmental characteristics, which include ecological aesthetics, ecological safety and ecological cleanliness.

[0023] In some embodiments, the neural network model is trained by the following steps:

[0024] Obtain the target image information of the play point in the sample database, input the target image information of the play point into the neural network model for feature extraction, obtain the visual feature vector corresponding to each target image information of the play point and the initial model parameters of the current neural network model, and input the target image information of the play point into the neural network model for feature extraction. Formula (1) is as follows:

[0025] ;

[0026] In formula (1), For the visual feature vectors, For the Game point target image information, It is the feature extraction function of the CNN-based neural network model;

[0027] The original classifier layer of the last layer of the neural network model is replaced with the newly added classifier layer, and ecological attribute labels are defined, where the ecological attribute labels include ecological beauty labels, ecological safety labels, and ecological cleanliness labels, and each ecological attribute label has a preset label threshold;

[0028] The newly introduced additional classifier layer is fine-tuned through the visual feature vector and the optimization function to obtain the final model parameters corresponding to the neural network model, which is expressed by formula (2):

[0029] ;

[0030] In formula (2), To optimize the function, are the final model parameters, are the initial model parameters, For the The sample prediction value of ecological attribute labels, For the The preset label threshold of ecological attribute labels, is the regularization coefficient;

[0031] The optimized neural network model is the trained neural network model, which is expressed by formula (3). Formula (3) is as follows:

[0032] ;

[0033] In formula (3), is the trained neural network model. is the expression of the neural network model using the final model parameters;

[0034] The output results of the neural network model include the predicted values ​​corresponding to the ecological attribute labels, which are expressed by formulas (4) to (6). Formula (4) is as follows:

[0035] ;

[0036] In formula (4), is the predicted value of ecological beauty, is the standard deviation function;

[0037] Formula (5) is as follows:

[0038] ;

[0039] In formula (5), is the predicted value of ecological safety, is the weight vector, is the vector norm;

[0040] Formula (6) is as follows:

[0041] ;

[0042] In formula (6), is the predicted value of ecological cleanliness, is the mean function.

[0043] In some embodiments, generating the first service feature according to the service information includes:

[0044] Perform DBSCAN algorithm clustering on multiple service information to obtain multiple service scenario clustering labels;

[0045] Inputting multiple service scenario clustering labels and multiple service information into the LDA model to obtain a first service feature, where the first service feature includes a service group feature of each service point, and each service group feature includes service group information and service preference information;

[0046] Input multiple service scenario clustering labels and multiple service information into the LDA model to obtain a first service feature, the first service feature includes a service group feature of each service point, each service group feature includes service group information and service preference information including:

[0047] The words appearing in each service information are sorted according to the service points to form a vocabulary list;

[0048] Count the frequency of words in each vocabulary, and generate a vocabulary matrix based on the vocabulary and service information;

[0049] Obtain the service scenario clustering label associated with the current vocabulary matrix and use it as the input feature of the context information of the LDA model;

[0050] The vocabulary matrix is ​​input into the LDA model to obtain the service group characteristics corresponding to the current vocabulary matrix, which is expressed by formula (7). Formula (7) is as follows:

[0051] ;

[0052] In formula (7), To serve the group characteristics, is the vocabulary of the current service information, is the hidden topic output by the LDA model, for Under the corresponding hidden topic The vocabulary distribution parameter of the corresponding vocabulary, i.e., the quantitative value of the service preference information, is a set of hyperparameters for the LDA model, for The corresponding global topic distribution parameters under the hidden topic, A collection of multiple service information. For current service information The corresponding distribution probability under the hidden topic, For current service information The corresponding topic distribution parameters under the hidden topic are the quantitative values ​​of the service group information.

[0053] In some embodiments, the first social feature includes a first emotional feature and a first time feature, and generating the first social feature of the current forest park according to the social information includes:

[0054] Collecting tourist comment information within a first preset time period, and performing data cleaning on the tourist comment information to obtain a first cleansed text, wherein the first cleansed text includes a plurality of comment information;

[0055] The first cleaned text is input into the BERT model to obtain the initial sentiment features corresponding to each comment information, which is expressed by formula (8). Formula (8) is as follows:

[0056] ;

[0057] In formula (8), is the initial emotional feature, To review the information, For emotional labels, To describe the semantic features of comment information in the BERT model, is the mapping matrix between sentiment labels and semantic features, is the bias parameter of the BERT model, is the activation function;

[0058] Concatenate multiple initial emotion features to obtain a first emotion feature;

[0059] and, obtaining management change information within a second preset time period;

[0060] Construct the time series data of current forest parks based on management change information and park information;

[0061] The time series data is input into the Prophet model to obtain the first time characteristics. The first time characteristics include trend characteristics, seasonal characteristics and holiday effect characteristics, which are expressed by formula (9). Formula (9) is as follows:

[0062] ;

[0063] In formula (9), For the first time feature, As a trend feature, It is a seasonal feature. It is the holiday effect characteristic. is the random error term, For time;

[0064] The first emotional feature is integrated with the first time feature to obtain the first social feature, which is expressed by formula (10). Formula (10) is as follows:

[0065] ;

[0066] In formula (10), The first social characteristic.

[0067] In some embodiments, the measurement model is represented by formula (11), which is as follows:

[0068] ;

[0069] In formula (11), , For the observed variables, For the The observed variables and The quantitative results between latent variables are For the potential variables, For the The measurement error of each observed variable;

[0070] The structural model is expressed by formula (12), which is as follows:

[0071] ;

[0072] In formula (12), is the goodness-of-fit index of recreation degree, For the The path coefficient corresponding to the hypothesized relationship is For the potential variables, is the structural error term.

[0073] In some embodiments, the method further comprises:

[0074] In the process of constructing the measurement model, the combined reliability and convergent validity of the measurement model are calculated;

[0075] Until the combined reliability meets the preset combined reliability threshold, and the convergent validity meets the preset convergent validity threshold;

[0076] The combined reliability is expressed by formula (13), which is as follows:

[0077] ;

[0078] In formula (13), For the The combined reliability of latent variables is For the The measurement error variance of the observed variables;

[0079] Convergent validity is expressed by formula (14), which is as follows:

[0080] ;

[0081] In formula (14), For the Convergent validity of latent variables.

[0082] In a second aspect, the present invention provides a multi-dimensional evaluation system for the recreation degree of a forest park, which is applicable to the multi-dimensional evaluation method for the recreation degree of a forest park described in the first aspect. The system includes an information processing module and an evaluation module. The information processing module is used to obtain environmental image information of the current forest park, the environmental image information includes panoramic image information and local image information, the local image information is an image containing part of the environment, and a first environmental feature is generated according to the environmental image information; and, multiple service information of the current forest park is obtained, each service information includes a service category, a service group, a service address, and service item information of the current service point, the service item information includes service commodities, service items, and service prices of the current service point, and a first service feature is generated according to the multiple service information; and, social information of the current forest park is obtained, the social information includes visitor comment information, management change information, and park information of the current forest park, and a first social feature of the current forest park is generated according to the social information;

[0083] The evaluation module is used to construct a PLS-SEM model. The PLS-SEM model includes a measurement model and a structural model. The environmental factor is used as the first latent variable of the PLS-SEM model, the service factor is used as the second latent variable of the PLS-SEM model, and the social factor is used as the third latent variable of the PLS-SEM model. The first environmental feature is used as the first observed variable corresponding to the first latent variable, the first service feature is used as the second observed variable corresponding to the second latent variable, and the first social feature is used as the third observed variable corresponding to the third latent variable to construct the measurement model; the first hypothesis relationship between the first latent variable and the recreation degree is set, the second hypothesis relationship between the second latent variable and the recreation degree is set, and the third hypothesis relationship between the third latent variable and the recreation degree is set. According to the first hypothesis relationship, the second hypothesis relationship and the third hypothesis relationship, the evaluation module is used to construct a PLS-SEM model. The system constructs a structural model; quantifies the first latent variable according to the first observed variable in the measurement model to obtain a first quantitative result, and quantifies the second latent variable according to the second observed variable in the measurement model to obtain a second quantitative result, and quantifies the third latent variable according to the third observed variable in the measurement model to obtain a third quantitative result; inputs the first quantitative result, the second quantitative result and the third quantitative result into the structural model, and calculates the goodness of fit index of the first latent variable, the second latent variable and the third latent variable to the recreation degree; obtains the recreation degree evaluation results and improvement suggestions of the current forest park, the recreation degree evaluation results include the quantitative contribution of the first environmental characteristics, the first service characteristics and the first social characteristics to the recreation degree, and the improvement suggestions include the second environmental characteristics, the second service characteristics and the second social characteristics.

[0084] In a third aspect, the present invention further provides a computer-readable storage medium storing computer program instructions, which implement the method described in the first aspect when executed by a processor.

[0085] In a fourth aspect, the present invention further provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.

[0086] By adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art:

[0087] Different from the existing technology, the above technical solution comprehensively considers the three key dimensions of environment, service and society, deeply analyzes the factors affecting recreation degree from multiple angles, integrates multiple data sources such as environmental image information, service information and social information, establishes a comprehensive evaluation system, and uses the PLS-SEM model for comprehensive evaluation. It can simultaneously evaluate the reliability of indicators in each dimension and the influence relationship between potential variables. According to the results of model analysis, targeted improvement suggestions such as the second environmental characteristics, the second service characteristics and the second social characteristics are put forward to optimize the supporting services and management structure of the forest park, improve the level of cultural services, periodically evaluate and adjust the supply and demand relationship, and improve the emotional scheduling effect of product services in the entire forest park recreation degree. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0089] Figure 1 It is a schematic diagram of steps S101 to S106 of the method for multi-dimensionally evaluating the recreation degree of a forest park;

[0090] Figure 2 It is a schematic diagram of steps S201 to S206 of the method for multi-dimensionally evaluating the recreation degree of a forest park;

[0091] Figure 3 It is a schematic diagram of the recreation degree system for evaluating forest parks in multiple dimensions;

[0092] Figure 4 It is a schematic diagram of the layout of the recreation system for multi-dimensional evaluation of forest parks.

[0093] Reference numerals:

[0094] 1. Multi-dimensional evaluation system of recreation degree of forest parks;

[0095] 11. Information processing module;

[0096] 12. Evaluation module. DETAILED DESCRIPTION

[0097] The present invention will be further described in detail below in conjunction with the accompanying drawings and examples. It is particularly noted that the following examples are only used to illustrate the present invention, but are not intended to limit the scope of the present invention. Similarly, the following examples are only partial embodiments of the present invention rather than all embodiments, and all other embodiments obtained by those of ordinary skill in the art without creative work are within the scope of protection of the present invention.

[0098] See also Figure 1 In a first aspect, this embodiment provides a multi-dimensional method for evaluating the recreation degree of a forest park, including:

[0099] S101, obtaining environmental image information of the current forest park, the environmental image information including panoramic image information and local image information, the local image information being an image containing part of the environment, and generating a first environmental feature according to the environmental image information; and, obtaining multiple service information of the current forest park, each service information including the service category, service group, service address and service item information of the current service point, the service item information including the service commodity, service item and service price of the current service point, and generating a first service feature according to the multiple service information; and, obtaining social information of the current forest park, the social information including visitor comment information, management change information and park information of the current forest park, and generating a first social feature of the current forest park according to the social information;

[0100] S102, constructing a PLS-SEM model, the PLS-SEM model includes a measurement model and a structural model, taking environmental factors as the first latent variable of the PLS-SEM model, service factors as the second latent variable of the PLS-SEM model, social factors as the third latent variable of the PLS-SEM model, taking the first environmental feature as the first observed variable corresponding to the first latent variable, taking the first service feature as the second observed variable corresponding to the second latent variable, taking the first social feature as the third observed variable corresponding to the third latent variable, and constructing the measurement model;

[0101] S103, setting a first hypothesis relationship between the first latent variable and the recreation degree, setting a second hypothesis relationship between the second latent variable and the recreation degree, setting a third hypothesis relationship between the third latent variable and the recreation degree, and constructing a structural model based on the first hypothesis relationship, the second hypothesis relationship, and the third hypothesis relationship;

[0102] S104, quantifying the first latent variable according to the first observed variable in the measurement model to obtain a first quantization result, quantifying the second latent variable according to the second observed variable in the measurement model to obtain a second quantization result, and quantifying the third latent variable according to the third observed variable in the measurement model to obtain a third quantization result;

[0103] S105, inputting the first quantification result, the second quantification result and the third quantification result into the structural model, and calculating the goodness of fit index of the first latent variable, the second latent variable and the third latent variable to the recreation degree;

[0104] S106. Obtain the recreation degree evaluation result and improvement suggestions of the current forest park. The recreation degree evaluation result includes the quantitative contribution of the first environmental characteristic, the first service characteristic and the first social characteristic to the recreation degree. The improvement suggestions include the second environmental characteristic, the second service characteristic and the second social characteristic.

[0105] In step S101, the environmental image information includes panoramic image information and local image information, wherein the panoramic image information may be remote sensing image information, and the local image information may be obtained by collecting pictures related to the current forest park on the Internet, and such pictures may be pictures of tourists in the past period of time, or pictures taken by staff when inspecting the forest park, and this embodiment does not limit this. The local image information and the panoramic image information are organized into environmental image information, and then the first environmental feature is generated based on the environmental image information for subsequent use. Furthermore, the process of generating the first environmental feature from the environmental image information includes image processing and recognition of the environmental image information, and then extracting the first environmental feature of the current forest park, that is, the first environmental feature includes image features and other feature information obtained based on the image, such as the ecological aesthetics, ecological safety, and ecological cleanliness described later.

[0106] At the same time, step S101 also includes the acquisition of service information. Each piece of service information corresponds to a service point in the current forest park. The service point can be understood as the service facilities or service items provided by the forest park to tourists, collectively referred to as service categories. Specific service categories include catering services, entertainment services, and cultural services, etc. The service group can be understood as the main service population of the current service point. For example, it can be divided according to age into infant groups, children groups, youth groups, and elderly groups. The service address is the address information of the current service point in the forest park, and can also contain geographic coordinate information to facilitate more accurate statistics of service information. The service item information is the specific content of the current service point, including service commodities, service items, service prices, etc. The first service feature can be extracted from the service information. Specifically, the first service feature can provide a reference for the service characteristics and cultural communication status of the current forest park. At the same time, the first service feature can indirectly reflect the comfort level of tourists in the current forest park.

[0107] Similarly, step S101 also includes the feature collection and extraction of social information, and the social information includes tourist comment information, management change information and park information. It should be noted that the first social feature extracted from the social information tends to express the current forest park's positioning in the society and the park portrait. The first social feature includes the emotional characteristics of the current forest park in the tourist comment information, as well as the inherent landscape, art and cultural heritage of the current forest park. At the same time, the park's management change information can indirectly reflect the advantages and disadvantages of the current forest park in terms of management.

[0108] In step S101, the first environmental feature, the first service feature and the first social feature are obtained, which can comprehensively collect the factors affecting the recreation degree of the current forest park, facilitate subsequent feature analysis and data processing, and thus help improve the accuracy and objectivity of the subsequent output evaluation results.

[0109] In step S102, the PLS-SEM model is used to evaluate the recreation degree of the above three characteristics. The PLS-SEM model is a model based on structural equations, which is used to explore the relationship between complex variables. In terms of the recreation degree of forest parks, there are many factors affecting the recreation degree, and the influence relationship between each factor is not clear enough, such as environmental factors, service factors and social factors described later. At the same time, the data information of forest parks belongs to small sample data, and the use of the PLS-SEM model is more in line with the needs of evaluating the recreation degree of a single forest park. Specifically, the PLS-SEM model includes a measurement model and a structural model. The complex variables mentioned in the PLS-SEM model can be understood as latent variables. The measurement model can be understood as a quantitative operation model of latent variables. Through the quantification between latent variables and observed variables, the influence of observed variables on latent variables can be obtained. The structural model is the spatial relationship between latent variables and output results, and the output results can be obtained through the structural model. In this embodiment, the latent variables include environmental factors, service factors and social factors. For the convenience of distinction, the environmental factors are recorded as the first latent variables, the service factors are recorded as the second latent variables, and the social factors are recorded as the third latent variables. The first environmental feature is used as the first observed variable of the first latent variable, the first service feature is used as the second observed variable of the second latent variable, and the first social feature is used as the third observed variable of the third latent variable. Furthermore, the output result is the recreation degree feature shown at present. For the convenience of description, the recreation degree feature is visualized as the goodness of fit index of recreation degree.

[0110] In step S103, a first hypothesized relationship between the first latent variable and the recreation degree is defined, a second hypothesized relationship between the second latent variable and the recreation degree is defined, and a third hypothesized relationship between the third latent variable and the recreation degree is defined, thereby constructing a structural model.

[0111] In step S104, the first quantitative result, the second quantitative result and the third quantitative result are first obtained through the measurement model. The first quantitative result is the influence of the first observed variable on the first latent variable. It can also be understood as obtaining the quantitative value of the first latent variable on the recreation degree from the first observed variable. Similarly, the second quantitative result and the third quantitative result can be understood with reference to the first quantitative result.

[0112] In step S105, after obtaining the first quantitative result, the second quantitative result and the third quantitative result, they are input into the structural model. During the simulation process, the structural model calculates the influence path between the first latent variable and the recreation degree under the first hypothetical relationship based on the first quantitative result to obtain the first path coefficient, and calculates the influence path between the second latent variable and the recreation degree under the second hypothetical relationship based on the second quantitative result to obtain the second path coefficient, and calculates the influence path between the third latent variable and the recreation degree under the third hypothetical relationship based on the third quantitative result to obtain the third path coefficient. The first path coefficient, the second path coefficient and the third path coefficient are also the goodness-of-fit indicators of the first latent variable, the second latent variable and the third latent variable for the recreation degree.

[0113] In step S106, after obtaining the goodness of fit index of recreation degree, the recreation degree evaluation result and improvement suggestions of the current forest park can be further generated. It should be noted that the recreation degree evaluation result includes the quantitative contribution of the first environmental feature to the recreation degree, that is, the influence of the first environmental feature on the recreation degree, the quantitative contribution of the first service feature to the recreation degree, that is, the influence of the first service feature on the recreation degree, and the quantitative contribution of the first social feature to the recreation degree, that is, the influence of the first social feature on the recreation degree.

[0114] Specifically, the impact of the first environmental feature on recreation degree may also include evaluation indicators of the first environmental feature from three perspectives: ecological aesthetics, ecological cleanliness and ecological safety. The evaluation indicators may be expressed in text form, such as "good, poor, excellent, medium" or "qualified, unqualified", etc. Corresponding to the impact of the first environmental feature on recreation degree, the second environmental feature in the improvement suggestion may be generated synchronously. The second environmental feature is an improvement indicator for the environment of the current forest park in the future. For example, the first environmental feature shows that the evaluation indicator of the ecological cleanliness of a certain recreational spot is "unqualified", then the second environmental feature is the need to improve the evaluation indicator of the ecological cleanliness of the recreational spot to "qualified" in the future, as well as the improvement suggestions or strategies required to achieve the second environmental feature, for reference by forest park managers. For example, the improvement suggestions or strategies may be to add cleaners, increase the frequency of cleaning, etc. This embodiment does not limit this.

[0115] Similarly, the impact of the first service feature on recreation degree can be reflected in the positive and negative impacts of the current service point, including whether the setting of the service point is reasonable, whether the service categories of the service point are fully utilized, etc. In the end, the comprehensive impact index corresponding to each service point can be obtained, and the advantages and disadvantages of each service point can be reflected. Furthermore, the corresponding second service feature is generated according to the recreation degree evaluation result. The second service feature includes improvement suggestions for each service point and improvement cost assessment. For example, a service point has been losing money for many years. From the recreation degree evaluation report, it can be found that the current service point has a mismatch between the needs of the service group and the supply of service items. In this case, the second service feature can record the specific matters that need to be adjusted at the current service point and the cost assessment involved in the adjustment.

[0116] Similarly, the first social characteristic reflects the social portrait attribute of the forest park. The impact of the first social characteristic on recreation can be used to evaluate the advantages and disadvantages of the current forest park from a management perspective. Based on this perspective, the second social characteristic can be further generated. The second social characteristic is the proposed changes and innovations made at the management level, which can specifically include the optimization of the forest park management system, the allocation of management personnel, the guidance and training of park culture, etc., to enhance the comfort of tourists and the cultural atmosphere of the park from a management perspective.

[0117] This embodiment collects data from three dimensions: environment, service and society, forming an objective basis for evaluation and avoiding the limitations brought by a single perspective. The advanced PLS-SEM model is used for analysis and modeling. Through the construction of measurement models and structural models, the influence of each dimension on recreation can be quantified, and targeted improvement suggestions such as the second environmental characteristics, the second service characteristics and the second social characteristics can be put forward, providing a reference direction for park managers to formulate specific optimization measures, forming a sustainable evaluation system, which is conducive to the continuous improvement of park management level.

[0118] See also Figure 2 In some embodiments, generating the first environmental feature according to the environmental image information includes:

[0119] S201, performing image preprocessing on environmental image information, the image preprocessing including image correction and image enhancement, to obtain a first image information group and a second image information group, the first image information group including a plurality of first image information, the plurality of first image information being panoramic image information generated after image preprocessing, and the second image information group including a plurality of second image information, the plurality of second image information being local image information generated after image preprocessing;

[0120] S202, performing image segmentation on the first image information, and inputting the segmentation result of the first image information into an image classifier for classification and labeling, to obtain a plurality of first tourist spot information corresponding to the current forest park, the first tourist spot information including the geographic coordinate information of the current first tourist spot and the first image feature; and performing image segmentation on the second image information, and inputting the segmentation result of the second image information into an image classifier for classification and labeling, to obtain a plurality of second tourist spot information corresponding to the current forest park, the second tourist spot information including the geographic coordinate information of the current second tourist spot and the second image feature;

[0121] S203, matching the second tourist spot information with the first tourist spot information according to the geographical location information;

[0122] S204: if the second play point information matches the first play point information successfully, the first play point is recorded as a third play point, and the third play point information is generated, where the third play point information includes the first image feature and the second image feature of the current third play point and the geographic coordinate information of the third play point;

[0123] S205: if the second play point information is not matched with the first play point information, the second play point is recorded as a fourth play point, and the fourth play point information is generated, where the fourth play point information includes the second image feature and the geographical location coordinate information of the fourth play point;

[0124] S206: Input the third tourist spot information and the fourth tourist spot information into the trained neural network model to obtain the first environmental characteristics, where the first environmental characteristics include ecological beauty, ecological safety, and ecological cleanliness.

[0125] In step S201, the environmental image information is preprocessed to improve the subsequent recognition accuracy. Specifically, the image preprocessing steps include image correction and image enhancement, wherein image correction is mainly used to eliminate geometric distortion and radiation distortion in the image to obtain more accurate spatial information. Specifically, image correction can reduce geometric deformation, brightness distortion and other problems in the environmental image information through affine transformation, geometric correction and other forms, so that the spatial positioning of the environmental image information is more accurate. Image enhancement includes steps such as edge sharpening, median filtering, and pseudo-color transformation, which can improve the contrast and edge features in the environmental image information and remove noise in the environmental image information, so as to facilitate image recognition operations.

[0126] For ease of expression, the processed environmental image information is divided into a first image information group and a second image information group, wherein the first image information group is a plurality of first image information obtained after image preprocessing of the panoramic image information, and the second image information group is a plurality of second image information obtained after image preprocessing of the local image information.

[0127] In step S202, the first image information and the second image information are respectively subjected to image segmentation and image classification, and finally the first tourist attraction point information and the second tourist attraction point information can be obtained. The two processing steps are the same, and the processing steps of the first image information are taken as an example for specific description, and the processing steps of the second image information can be understood with reference to the processing steps of the first image information. In this embodiment, the first image information is subjected to image segmentation, and the image segmentation divides each first image information into several regions. The specific image segmentation principle can be gray value segmentation, edge segmentation, etc. The first image information includes a panoramic image of the forest park. After image segmentation, a rough object image can be obtained, including trails, railings, hillsides, valleys, rivers, recreational facilities, etc. Further, the segmented object image is recorded as the first object image, and this group of first object images is input into the image classifier, and the image classifier classifies the first object image so that a mapping relationship is established between the first object image and the corresponding image label. For example, trails, paths, and bridges can be classified into road categories, and so on. The image classifier can also be understood as an image classification algorithm, such as a support vector machine, random forest or other algorithm model. Pre-training the image classifier can improve the classification accuracy of the image classifier. This process can also be understood as an image recognition process. On this basis, the image features in the first object image can be further extracted. The specific extraction factors include image texture, image shape and image features, etc., and the first tourist point information is obtained by sorting. The first tourist point information includes geographic coordinate information and first image features. It should be noted that the first image information is sorted according to the tourist point after image segmentation, classification and labeling to obtain the first tourist point information. The tourist point is a tourist point pre-set in the current forest park. On this basis, the first image feature in the first tourist point information is the image feature of the current tourist point, and the geographic coordinate information is the geographic coordinate information or area information of the current tourist point. Similarly, after image segmentation, classification and labeling, the second image information is also sorted according to the pre-set tourist spots in the forest park to obtain the second tourist spot information. The difference from the first tourist spot information is that the second tourist spot information is more detailed because the second image information is based on local image features. The same image segmentation, classification and labeling steps reflect more accurate content. The second tourist spot information can serve as a supplement to the first tourist spot information.

[0128] On this basis, in step S203, the second play point information needs to be matched with the first play point information, that is, the play points in the second play point information are matched with the play points in the first play point information. The specific matching method can be based on the matching of geographic coordinate information. Within the same geographic coordinate information or the same area, it can be indicated that the second play point information and the first play point information are the information content of the same play point.

[0129] In step S204, if the match is successful, that is, it means that the current first play point information and the current second play point information are the information content of the same play point, then the second play point information and the first play point information are aggregated to form the third play point information, and in order to facilitate the distinction from the other play points mentioned above, the play point is recorded as the third play point;

[0130] In step S205, if the match is unsuccessful, it means that the current first play point information and the current second play point information are not the information content of the same play point, and the second play point information does not match the remaining first play point information, which means that the second play point information may be newly developed, or a play point formed by tourists' independent excavation. In order to distinguish it from the aforementioned play points, it is recorded as the fourth play point. That is, the third play point is the current forest park's established play point, and the fourth play point is the current forest park's temporary play point. There is a certain instability, which is specifically reflected in that the fourth play point may be seasonal, may have safety hazards, and there are many uncontrollable factors.

[0131] In step S206, the third tourist spot information and the fourth tourist spot information are input into the trained neural network model to obtain the first environmental feature. This embodiment uses the form of a neural network model to implement feature extraction based on tourist spot information to obtain the first environmental feature. In this embodiment, the first environmental feature is divided into three dimensions: ecological beauty, ecological safety, and ecological cleanliness.

[0132] Further, in some embodiments, the neural network model is trained by the following steps:

[0133] Obtain the target image information of the play point in the sample database, input the target image information of the play point into the neural network model for feature extraction, obtain the visual feature vector corresponding to each target image information of the play point and the initial model parameters of the current neural network model, and input the target image information of the play point into the neural network model for feature extraction. Formula (1) is as follows:

[0134] ;

[0135] In formula (1), For the visual feature vectors, For the Game point target image information, It is the feature extraction function of the CNN-based neural network model;

[0136] The original classifier layer of the last layer of the neural network model is replaced with the newly added classifier layer, and ecological attribute labels are defined, where the ecological attribute labels include ecological beauty labels, ecological safety labels, and ecological cleanliness labels, and each ecological attribute label has a preset label threshold;

[0137] The newly introduced additional classifier layer is fine-tuned through the visual feature vector and the optimization function to obtain the final model parameters corresponding to the neural network model, which is expressed by formula (2):

[0138] ;

[0139] In formula (2), To optimize the function, are the final model parameters, are the initial model parameters, For the The sample prediction value of ecological attribute labels, For the The preset label threshold of ecological attribute labels, is the regularization coefficient;

[0140] The optimized neural network model is the trained neural network model, which is expressed by formula (3). Formula (3) is as follows:

[0141] ;

[0142] In formula (3), is the trained neural network model. is the expression of the neural network model using the final model parameters;

[0143] The output results of the neural network model include the predicted values ​​corresponding to the ecological attribute labels, which are expressed by formulas (4) to (6). Formula (4) is as follows:

[0144] ;

[0145] In formula (4), is the predicted value of ecological beauty, is the standard deviation function;

[0146] Formula (5) is as follows:

[0147] ;

[0148] In formula (5), is the predicted value of ecological safety, is the weight vector, is the vector norm;

[0149] Formula (6) is as follows:

[0150] ;

[0151] In formula (6), is the predicted value of ecological cleanliness, is the mean function.

[0152] In this embodiment, the neural network model can select a pre-trained CNN neural network model. On this basis, the classifier layer of the last layer of the neural network model is selected for replacement and fine-tuning optimization, so that the neural network model has the conversion function of converting image features into first environment features based on image feature extraction.

[0153] Specifically, we first select the target image information of the play point from the sample database, and initialize the CNN neural network model, which is used as a feature extractor to extract the visual feature vector corresponding to the target image information of the play point. The visual feature vector here is also the aforementioned image feature, including image texture, image color and image shape, etc. During the training process, the neural network model has initial model parameters. The entire feature extraction process can be expressed by formula (1). It should be noted that here is the serial number of the tourist spot extracted from the sample database. The total number of tourist spots in each forest park is different. This embodiment does not set a limitation on this. It can be understood that the tourist spot target image information extracted from the sample database should be the tourist spot target image information of the same sample forest park.

[0154] On this basis, the classifier layer of the last layer of the neural network model is replaced. For ease of expression, the classifier layer before replacement is recorded as the original classifier layer, and the classifier layer after replacement is recorded as the newly added classifier layer. The newly added classifier layer is used to optimize the neural network model. At the same time, it is necessary to predefine the ecological attribute label. In this embodiment, the ecological attribute label corresponds to the aforementioned ecological beauty, ecological cleanliness and ecological safety, that is, the ecological beauty label, the ecological cleanliness label and the ecological safety label. Each ecological attribute label has a preset label threshold, which can be understood as follows: the size of the preset label threshold is predefined. When the predicted value of the corresponding ecological attribute label exceeds the preset label threshold, it indicates that the ecological attribute is qualified, or that the ecological attribute is acceptable. For example, the ecological beauty label has an ecological beauty preset label threshold. When the predicted value of the ecological beauty output by the current neural network model is greater than the ecological beauty prediction label threshold, it means that the ecological beauty of the tourist spot reaches the acceptable level for tourists.

[0155] On this basis, we use the optimization function to perform fine-tuning optimization. Combining it with formula (2), we can know that: is the sequence number of the ecological attribute label. Specifically, there are three ecological attribute labels. Then, the ecological attribute label corresponding to the ecological aesthetics is the first ecological attribute label, that is, , the ecological attribute label corresponding to the ecological security is the second ecological attribute label, that is, , the ecological attribute label corresponding to ecological cleanliness is the third ecological attribute label, that is, .

[0156] The optimized neural network model, that is, the trained neural network model, after inputting the information of the third and fourth tourist spots into this neural network model, will output the predicted values ​​of the ecological attributes corresponding to the third and fourth tourist spots, and further organize them to obtain the first environmental characteristics.

[0157] Furthermore, based on the target image information of the tourist spots provided by the aforementioned sample database, this embodiment provides an expression corresponding to the ecological attribute label, as shown in Formula (4) to Formula (6). The specific function expression can be selected according to the different forest parks, and this embodiment does not limit this.

[0158] This embodiment realizes intelligent extraction of environmental features by fusing and analyzing panoramic image information and local image information, and adopts computer vision technologies such as image segmentation and image classification, which reduces the burden of manual operation, improves analysis efficiency, and can more comprehensively reflect the environmental features of forest parks. By training and optimizing the neural network model, the automatic evaluation of environmental features is realized, and the first environmental features can be quickly generated; wherein, the neural network model adopts a model training method based on a sample database, so that this environmental feature extraction method has certain scalability, and the model can be migrated and optimized according to the conditions of different forest parks, so as to achieve a wider range of applications. The use of neural network models for accurate evaluation of ecological beauty, ecological safety, and ecological cleanliness improves the objectivity and credibility of environmental feature extraction, and can be well applied to different types of forest parks.

[0159] In some embodiments, generating the first service feature according to the service information includes:

[0160] Perform DBSCAN algorithm clustering on multiple service information to obtain multiple service scenario clustering labels;

[0161] Inputting multiple service scenario clustering labels and multiple service information into the LDA model to obtain a first service feature, where the first service feature includes a service group feature of each service point, and each service group feature includes service group information and service preference information;

[0162] Input multiple service scenario clustering labels and multiple service information into the LDA model to obtain a first service feature, the first service feature includes a service group feature of each service point, each service group feature includes service group information and service preference information including:

[0163] The words appearing in each service information are sorted according to the service points to form a vocabulary list;

[0164] Count the frequency of words in each vocabulary, and generate a vocabulary matrix based on the vocabulary and service information;

[0165] Obtain the service scenario clustering label associated with the current vocabulary matrix and use it as the input feature of the context information of the LDA model;

[0166] The vocabulary matrix is ​​input into the LDA model to obtain the service group characteristics corresponding to the current vocabulary matrix, which is expressed by formula (7). Formula (7) is as follows:

[0167] ;

[0168] In formula (7), To serve the group characteristics, is the vocabulary of the current service information, is the hidden topic output by the LDA model, for Under the corresponding hidden topic The vocabulary distribution parameter of the corresponding vocabulary, i.e., the quantitative value of the service preference information, is a set of hyperparameters for the LDA model, for The corresponding global topic distribution parameters under the hidden topic, A collection of multiple service information. For current service information The corresponding distribution probability under the hidden topic, For current service information The corresponding topic distribution parameters under the hidden topic are the quantitative values ​​of the service group information.

[0169] In this embodiment, the DBSCAN algorithm is used in advance to cluster the service information to obtain multiple service scenario clustering labels, which can roughly divide the service information of multiple service points. On this basis, the service scenario clustering labels are used as context information input of the LDA model, so that the LDA model can be used as an influencing factor in the process of processing service information and discover potential topics under different service scenarios.

[0170] Specifically, the LDA model is a topic model algorithm that can automatically discover potential topics from a large amount of text data and represent the relationship between topics and documents, thereby obtaining the required feature information. In this embodiment, the document can be understood as service information, the topic is the hidden topic of the LDA model during the operation process, and the feature information is also the service group feature.

[0171] Specifically, multiple service information is regarded as "documents", and the words appearing in each service information are sorted according to the service points to form a vocabulary list. On this basis, the frequency of words in each vocabulary list is counted, and a vocabulary matrix is ​​generated according to the vocabulary list and service information. Furthermore, the service scenario clustering label associated with the vocabulary matrix is ​​obtained, and the vocabulary matrix and the service scenario clustering label are input into the LDA model together to obtain the service group characteristics corresponding to this vocabulary matrix.

[0172] In formula (7), Indicates that the topic is hidden The current service information vocabulary The probability distribution of the service group is the service group characteristics. It should be noted that the service group information and service preference information shown in this embodiment, the service group information is also the distribution information of the service group of the service point. If it is distinguished by the aforementioned age groups, the service proportion of "infants, children, young people, and the elderly" in the service point can be obtained. The service preference information can be understood as the service content that the service point can provide. For example, the service preference information can include the proportion of "eating, playing, and leisure" in the service point, so as to know the overall characteristic content of the service point.

[0173] By combining the characteristics of multiple service groups, we can obtain the first service characteristic, which is the potential variable reflecting the service factors of the current forest park.

[0174] This embodiment obtains service scenario labels through DBSCAN clustering, which provides important context information for the LDA model and helps LDA to more accurately discover potential service topic features. Using the topic modeling capability of the LDA model, service group features and service preference information can be automatically extracted from a large amount of service information, providing a comprehensive and objective basis for the construction of the first service feature.

[0175] In some embodiments, the first social feature includes a first emotional feature and a first time feature, and generating the first social feature of the current forest park according to the social information includes:

[0176] Collecting tourist comment information within a first preset time period, and performing data cleaning on the tourist comment information to obtain a first cleansed text, wherein the first cleansed text includes a plurality of comment information;

[0177] The first cleaned text is input into the BERT model to obtain the initial sentiment features corresponding to each comment information, which is expressed by formula (8). Formula (8) is as follows:

[0178] ;

[0179] In formula (8), is the initial emotional feature, To review the information, For emotional labels, To describe the semantic features of comment information in the BERT model, is the mapping matrix between sentiment labels and semantic features, is the bias parameter of the BERT model, is the activation function;

[0180] Concatenate multiple initial emotion features to obtain a first emotion feature;

[0181] and, obtaining management change information within a second preset time period;

[0182] Construct the time series data of current forest parks based on management change information and park information;

[0183] The time series data is input into the Prophet model to obtain the first time characteristics. The first time characteristics include trend characteristics, seasonal characteristics and holiday effect characteristics, which are expressed by formula (9). Formula (9) is as follows:

[0184] ;

[0185] In formula (9), For the first time feature, As a trend feature, It is a seasonal feature. It is the holiday effect characteristic. is the random error term, For time;

[0186] The first emotional feature is integrated with the first time feature to obtain the first social feature, which is expressed by formula (10). Formula (10) is as follows:

[0187] ;

[0188] In formula (10), The first social characteristic.

[0189] In this embodiment, the first social feature is divided into a first emotional feature and a first temporal feature, wherein the first emotional feature is the quantitative embodiment of the emotional value provided by the current forest park to tourists, and the first temporal feature includes the periodic change law of the current forest park in the management and social impression due to time change factors, such as the holiday effect and seasonal effect described later, while taking into account forest park categories such as Wuyishan Forest Park and Xiangshan Park that are obviously affected by seasonal ecological changes.

[0190] Specifically, in this embodiment, the first sentiment feature is extracted based on the tourist comment information, using the BERT model, which is a pre-trained language model based on Transformer and is good at tasks such as text classification. The BERT model is used to perform sentiment analysis on the collected tourist comment information, predict the sentiment tendency (positive, negative or neutral) of each comment, and count the proportion of positive, negative and neutral emotions in all tourist comments. These proportion indicators can reflect the overall experience of tourists.

[0191] In this embodiment, it is necessary to predefine emotion labels, which can be set according to actual needs. The more detailed the emotion labels are, the more helpful it is for the BERT model to classify emotion features more accurately. The specific activation function and bias parameters can be appropriately adjusted and selected during the training phase of BERT to improve the recognition accuracy of the initial emotion features. Finally, multiple initial emotion features are spliced ​​to form the first emotion feature.

[0192] It should be noted that different first preset time periods of the tourist comment information shown in this embodiment will also form different first emotional features. For example, tourist comment information of a long time segment can be selected to extract the emotional features of the current forest park in a long time, which are recorded as long-term emotional features. Further, tourist comment information of a short time segment can be selected to extract the emotional features of the current forest park in the most recent time period, which are recorded as short-term emotional features. By splicing long-term emotional features with short-term emotional features, the tourist experience output by the forest park at different levels can be more accurately reflected. For example, for some forest parks that are significantly affected by seasonal ecological changes, short-term emotional features will be briefly improved in some seasons, and long-term emotional features are more inclined to the basic experience and feeling properties of the current forest park for tourists. This method can achieve multi-level expression of the first emotional feature.

[0193] In this embodiment, the first time feature is obtained by extracting management change information and park information. Specifically, the second preset time period may be the same as or different from the first preset time period, and this embodiment does not limit this. Management change information includes multiple changes based on management level, such as personnel changes, service changes, and changes in park opening hours. Each change record corresponds to a change text and change time information.

[0194] In this embodiment, time series data is constructed based on the management change information and the park information. That is, the management change information is sorted in chronological order. At the same time, the seasonal change information in the park information is matched with the management change information in combination with the seasonal change information reflected in the park information. For example, a forest park has beautiful fallen leaves in autumn. This beautiful fallen leaves, as a park feature of the forest park in the park information, is directly related to the appropriate increase in personnel control and cleaning frequency in autumn in the management change information.

[0195] The sorted time series data is input into the Prophet model. The Prophet model is a time series prediction model that is mainly used to process time series data with trends and seasonality. Formula (9) can be used to express the first time feature. By decomposing the information in the time series data according to time, the changing pattern of the forest park over time can be more accurately described.

[0196] The first time feature and the first emotion feature are fused to form the first social feature. The specific fusion method may be principal component analysis or feature fusion based on an attention mechanism, which is not limited in this embodiment.

[0197] This embodiment shows the specific generation process of the first social characteristic. Through the analysis of the first emotional characteristic and the first time characteristic, the first social characteristic of the current forest park at the social level can be given from both subjective and objective perspectives, which is convenient for subsequent recreational evaluation.

[0198] In some embodiments, the measurement model is represented by formula (11), which is as follows:

[0199] ;

[0200] In formula (11), , For the observed variables, For the The observed variables and The quantitative results between latent variables are For the potential variables, For the The measurement error of each observed variable;

[0201] The structural model is expressed by formula (12), which is as follows:

[0202] ;

[0203] In formula (12), is the goodness-of-fit index of recreation degree, For the The path coefficient corresponding to the hypothesized relationship is For the potential variables, is the structural error term.

[0204] In some embodiments, the method further comprises:

[0205] In the process of constructing the measurement model, the combined reliability and convergent validity of the measurement model are calculated;

[0206] Until the combined reliability meets the preset combined reliability threshold, and the convergent validity meets the preset convergent validity threshold;

[0207] The combined reliability is expressed by formula (13), which is as follows:

[0208] ;

[0209] In formula (13), For the The combined reliability of latent variables is For the The measurement error variance of the observed variables;

[0210] Convergent validity is expressed by formula (14), which is as follows:

[0211] ;

[0212] In formula (14), For the Convergent validity of latent variables.

[0213] In this embodiment, formula (11) is a specific expression of the measurement model. hour, , which is the first observed variable, , that is, the quantitative result between the first observed variable and the first latent variable, recorded as the first quantitative result, , which is the first latent variable, , which is the measurement error corresponding to the first observed variable. Similarly, the second observed variable, the second latent variable, and the second quantitative result correspond to When the formula (11) is used, the third observed variable, the third latent variable, and the third quantitative result correspond to The formula (11) is used when

[0214] In this embodiment, formula (12) is a specific expression of the structural model.

[0215] Furthermore, in the PLS-SEM model, the fitting accuracy of the current measurement model is measured by evaluating the combined reliability and convergent validity in the measurement model. Furthermore, the evaluation indicators of the structural model include the coefficient of determination, effect size, etc. By evaluating the measurement model and the structural model, the quality of the output results of the PLS-SEM model can be ensured, and the pertinence and accuracy of the recreation evaluation results and improvement suggestions can be improved.

[0216] See also Figure 3 In the second aspect, the present embodiment provides a multi-dimensional evaluation system 1 for evaluating the recreation degree of a forest park, which is applicable to the multi-dimensional evaluation method for the recreation degree of a forest park described in the first aspect. The system includes an information processing module 11 and an evaluation module 12. The information processing module 11 is used to obtain environmental image information of the current forest park, the environmental image information includes panoramic image information and local image information, the local image information is an image containing part of the environment, and a first environmental feature is generated according to the environmental image information; and, multiple service information of the current forest park is obtained, each service information includes a service category, a service group, a service address, and service item information of the current service point, the service item information includes service commodities, service items, and service prices of the current service point, and a first service feature is generated according to the multiple service information; and, social information of the current forest park is obtained, the social information includes visitor comment information, management change information, and park information of the current forest park, and a first social feature of the current forest park is generated according to the social information;

[0217] The evaluation module 12 is used to construct a PLS-SEM model, which includes a measurement model and a structural model. The environmental factor is used as the first latent variable of the PLS-SEM model, the service factor is used as the second latent variable of the PLS-SEM model, and the social factor is used as the third latent variable of the PLS-SEM model. The first environmental feature is used as the first observed variable corresponding to the first latent variable, the first service feature is used as the second observed variable corresponding to the second latent variable, and the first social feature is used as the third observed variable corresponding to the third latent variable to construct the measurement model; a first hypothesis relationship between the first latent variable and the recreation degree is set, a second hypothesis relationship between the second latent variable and the recreation degree is set, and a third hypothesis relationship between the third latent variable and the recreation degree is set. According to the first hypothesis relationship, the second hypothesis relationship and the third hypothesis relationship, The structural model is constructed based on the relationship between the first observed variable and the second observed variable in the measurement model; the first latent variable is quantified according to the first observed variable in the measurement model to obtain the first quantitative result, and the second latent variable is quantified according to the second observed variable in the measurement model to obtain the second quantitative result, and the third latent variable is quantified according to the third observed variable in the measurement model to obtain the third quantitative result; the first quantitative result, the second quantitative result and the third quantitative result are input into the structural model, and the goodness of fit index of the first latent variable, the second latent variable and the third latent variable to the recreation degree is calculated; the recreation degree evaluation results and improvement suggestions of the current forest park are obtained, the recreation degree evaluation results include the quantitative contribution of the first environmental characteristics, the first service characteristics and the first social characteristics to the recreation degree, and the improvement suggestions include the second environmental characteristics, the second service characteristics and the second social characteristics.

[0218] The multi-dimensional evaluation system for the recreation degree of a forest park shown in this embodiment can be understood by referring to the aforementioned method, and no unnecessary elaboration is given here.

[0219] For further information, see Figure 4 This embodiment also provides a specific application example based on the above method steps:

[0220] This example sets up four modules: cultural ecosystem services, tourism experience, symbolic interaction, comfort objects, and service quality, and constructs two circular analysis frameworks. It helps tourists recognize and perceive the spiritual expression of cultural ecosystem services through symbolic interaction, and uses comfort objects to obtain tourists' feedback on cultural ecosystem services, thus forming a virtuous feedback and cycle and achieving the sustainable development of the cultural ecosystem.

[0221] On this basis, the categories and products of comfort objects are improved through service quality and tourism experience, and the service quality is further adjusted by tourism experience. A closed loop is formed between the three, achieving a multi-level win-win effect. The comfort embodiment generation process under the synergy between the internal comfort experience of recreationists and the external comfort elements in a specific time and space. The comfort experience finally formed by recreationists is not a simple psychological perception, but the result of the interaction of a series of related factors such as perception, body and environment during the tourism process. Obviously, the internal comfort experience and external comfort elements jointly create the overall comfort of recreationists and affect their recreational behavior. The internal action mechanism of the recreation comfort model is a systematic embodied perception model, the core of which lies in the four aspects of body subjectivity, immersion experience, meaning experience and social adaptability.

[0222] This example is guided by the balance of supply and demand of park cultural services and establishes a model of the supply and demand relationship of cultural services. It proposes the concept of "comfort experience" to evaluate the co-creation effect and co-creation space characteristics of national park tourism experience value in which tourists participate. Through the adaptation of four aspects of expected services, experience services, functional services and tourism services, it constructs a co-creation relationship between park cultural services and tourism experience value, and realizes a multi-dimensional evaluation of the recreation degree of forest parks.

[0223] In a third aspect, this embodiment further provides a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method described in the first aspect is implemented.

[0224] The computer program involved in this embodiment can be stored in a computer device readable storage medium, which includes but is not limited to a disk, a tape, a magnetic card, a floppy disk, a flash memory, an optical disk, an optical card, a read-only memory (ROM), a random access memory (RAM), an erasable programmable ROM (EPROM) and an electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, protein and other units with information storage capabilities. In a specific embodiment, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiment can be stored in a single medium in a centralized manner, or it can be stored in multiple media in a distributed manner. The memory containing the computer device readable storage medium can be a non-volatile memory or a random access memory. These computer device readable storage media can be built into the device, or they can be connected to the device involved in the embodiment as an external device or a part of an external device. In some embodiments, a memory with a computer device readable storage medium is deployed locally; in other embodiments, a solution of deploying the memory away from the processor may also be adopted, such as a network attached memory accessed via an RF circuit or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), etc., or a suitable combination thereof, as long as the computer device can access the memory. In addition, the computer program involved in the embodiment may be stored in plaintext / ciphertext form, or may be designed as training data, which may be integrated and reorganized through model training and implicitly stored in the parameter state of a deep neural network or other machine learning model.

[0225] In a fourth aspect, this embodiment further provides an electronic device, comprising a memory and a processor, wherein the memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method described in the first aspect.

[0226] The processor described in this embodiment can be implemented by hardware, firmware, software or a combination thereof, and can use a circuit, a single or multiple application-specific integrated circuits (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, at least one of a microprocessor, and also includes other physical, biological or chemical structures that can achieve similar or equivalent functions to the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of the present application, or any combination of the steps mentioned therein.

[0227] Different from the existing technology, the above technical solution comprehensively considers the three key dimensions of environment, service and society, deeply analyzes the factors affecting recreation degree from multiple angles, integrates multiple data sources such as environmental image information, service information and social information, establishes a comprehensive evaluation system, and uses the PLS-SEM model for comprehensive evaluation. It can simultaneously evaluate the reliability of indicators in each dimension and the influence relationship between potential variables. According to the results of model analysis, targeted improvement suggestions such as the second environmental characteristics, the second service characteristics and the second social characteristics are put forward to optimize the supporting services and management structure of the forest park, improve the level of cultural services, periodically evaluate and adjust the supply and demand relationship, and improve the emotional scheduling effect of product services in the entire forest park recreation degree.

[0228] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0229] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0230] The above descriptions are only some embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Any equivalent device or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A multi-dimensional method for evaluating the recreational degree of a forest park, characterized in that: include: Acquire environmental image information of the current forest park to generate a first environmental feature, where the environmental image information includes panoramic image information and local image information, where the local image information is an image containing part of the environment; Acquire multiple service information of the current forest park and generate a first service feature; Acquire the social information of the current forest park and generate the first social feature of the current forest park; Constructing a PLS-SEM model, the PLS-SEM model includes a measurement model and a structural model, taking environmental factors as the first latent variable of the PLS-SEM model, service factors as the second latent variable of the PLS-SEM model, social factors as the third latent variable of the PLS-SEM model, taking the first environmental feature as the first observed variable corresponding to the first latent variable, taking the first service feature as the second observed variable corresponding to the second latent variable, taking the first social feature as the third observed variable corresponding to the third latent variable, and constructing the measurement model; Setting a first hypothesis relationship between the first latent variable and the recreation degree, setting a second hypothesis relationship between the second latent variable and the recreation degree, setting a third hypothesis relationship between the third latent variable and the recreation degree, and constructing the structural model according to the first hypothesis relationship, the second hypothesis relationship and the third hypothesis relationship; quantifying the first latent variable according to the first observed variable to obtain a first quantization result, quantifying the second latent variable according to the second observed variable to obtain a second quantization result, and quantifying the third latent variable according to the third observed variable to obtain a third quantization result; Inputting the first quantification result, the second quantification result and the third quantification result into the structural model to generate a goodness of fit index for recreation degree; Obtain the recreation evaluation results and improvement suggestions of the current forest park; Generating a first environmental feature according to the environmental image information includes: Performing image preprocessing on the environment image information to obtain a first image information group and a second image information group, wherein the first image information group includes a plurality of first image information, which is panoramic image information generated after image preprocessing, and the second image information group includes a plurality of second image information, which is local image information generated after image preprocessing; Performing image segmentation on the first image information, and inputting the segmentation result of the first image information into an image classifier for classification and labeling, to obtain a plurality of first tourist spot information corresponding to the current forest park, wherein the first tourist spot information includes geographic coordinate information of the current first tourist spot and a first image feature; Performing image segmentation on the second image information, and inputting the segmentation result of the second image information into an image classifier for classification and labeling, to obtain a plurality of second tourist spot information corresponding to the current forest park, wherein the second tourist spot information includes geographic coordinate information of the current second tourist spot and second image features; Matching the second tourist spot information with the first tourist spot information according to geographical location information; If the match is successful, the first play point is recorded as the third play point, and the third play point information is generated, wherein the third play point information includes the first image feature and the second image feature of the current third play point and the geographic coordinate information of the third play point; If the match is unsuccessful, the second play point is recorded as the fourth play point, and the fourth play point information is generated, where the fourth play point information includes the second image feature and the geographical location coordinate information of the fourth play point; The third tourist spot information and the fourth tourist spot information are input into the trained neural network model to obtain the first environmental characteristics, which include ecological beauty, ecological safety and ecological cleanliness.

2. The multi-dimensional evaluation method of recreation degree of forest parks according to claim 1 is characterized in that: The neural network model is trained by the following steps: Obtain the target image information of the play point in the sample database, input the target image information of the play point into the neural network model for feature extraction, obtain the visual feature vector corresponding to each target image information of the play point and the initial model parameters of the current neural network model, and input the target image information of the play point into the neural network model for feature extraction. Formula (1) is as follows: ； In formula (1), For the visual feature vectors, For the Game point target image information, It is the feature extraction function of the CNN-based neural network model; The original classifier layer of the last layer of the neural network model is replaced with a newly added classifier layer, and ecological attribute labels are defined, wherein the ecological attribute labels include an ecological beauty label, an ecological safety label, and an ecological cleanliness label, and each of the ecological attribute labels has a preset label threshold; The newly introduced additional classifier layer is fine-tuned by using the visual feature vector and the optimization function to obtain the final model parameters corresponding to the neural network model, which are expressed by formula (2): ； In formula (2), is the optimization function, are the final model parameters, are the initial model parameters, For the The sample prediction value of ecological attribute labels, For the The preset label threshold of ecological attribute labels, is the regularization coefficient; The optimized neural network model is the trained neural network model, which is expressed by formula (3). Formula (3) is as follows: ； In formula (3), is the trained neural network model, is the expression of the neural network model using the final model parameters; The output result of the neural network model includes the predicted value corresponding to the ecological attribute label, which is expressed by formula (4) to formula (6). Formula (4) is as follows: ； In formula (4), is the predicted value of ecological beauty, is the standard deviation function; The formula (5) is as follows: ； In formula (5), is the predicted value of ecological safety, is the weight vector, is the vector norm; The formula (6) is as follows: ； In formula (6), is the predicted value of ecological cleanliness, is the mean function.

3. The multi-dimensional evaluation method of recreation degree of forest parks according to claim 1 is characterized in that: Generating a first service feature according to the service information includes: Performing DBSCAN algorithm clustering on the plurality of service information to obtain a plurality of service scenario clustering labels; Inputting the plurality of service scenario clustering labels and the plurality of service information into the LDA model to obtain the first service feature, wherein the first service feature includes a service group feature of each service point, and each service group feature includes service group information and service preference information, including: Arranging the words appearing in each of the service information according to the service points to form a vocabulary list; Counting the frequency of occurrence of words in each of the vocabulary lists, and generating a vocabulary matrix according to the vocabulary lists and the service information; Obtaining the service scenario clustering label associated with the current vocabulary matrix, and using it as an input feature of the context information of the LDA model; The vocabulary matrix is ​​input into the LDA model to obtain the service group characteristics corresponding to the current vocabulary matrix, which is expressed by formula (7). The formula (7) is as follows: ； In formula (7), The characteristics of the service group are: is the vocabulary of the service information currently described, is the hidden topic output by the LDA model, for Under the corresponding hidden topic The vocabulary distribution parameter of the corresponding vocabulary, i.e., the quantitative value of the service preference information, is a set of hyperparameters for the LDA model, for The corresponding global topic distribution parameters under the hidden topic, is a collection of multiple service information, For the current service information The corresponding distribution probability under the hidden topic, For the current service information The corresponding topic distribution parameters under the hidden topic are the quantitative values ​​of the service group information.

4. The multi-dimensional evaluation method of recreation degree of forest parks according to claim 1 is characterized in that: The first social feature includes a first emotional feature and a first time feature. Generating the first social feature of the current forest park according to the social information includes: Collecting tourist comment information within a first preset time period, and performing data cleaning on the tourist comment information to obtain a first cleansed text, wherein the first cleansed text includes a plurality of comment information; The first cleaned text is input into the BERT model to obtain the initial sentiment feature corresponding to each of the review information, which is expressed by formula (8). The formula (8) is as follows: ； In formula (8), is the initial emotional feature, To review the information, For emotional labels, is the semantic feature of the review information in the BERT model, is the mapping matrix between sentiment labels and semantic features, is the bias parameter of the BERT model, is the activation function; Concatenate the multiple initial emotion features to obtain the first emotion feature; and, obtaining management change information within a second preset time period; Constructing the time series data of the current forest park according to the management change information and the park information; The time series data is input into the Prophet model to obtain the first time feature, which includes trend feature, seasonal feature and holiday effect feature, and is expressed by formula (9). The formula (9) is as follows: ； In formula (9), For the first time feature, As a trend feature, It is a seasonal feature. It is the holiday effect characteristic. is the random error term, For time; The first emotional feature is fused with the first time feature to obtain the first social feature, which is expressed by formula (10). The formula (10) is as follows: ； In formula (10), This is the first social characteristic.

5. The multi-dimensional evaluation method of recreation degree of forest parks according to claim 1 is characterized in that: The measurement model is expressed by formula (11), which is as follows: ； In formula (11), , For the observed variables, For the The observed variables and The quantitative results between latent variables are For the potential variables, For the The measurement error of each observed variable; The structural model is expressed by formula (12), which is as follows: ； In formula (12), is the goodness-of-fit index of recreation degree, For the The path coefficient corresponding to the hypothesized relationship is For the potential variables, is the structural error term.

6. The multi-dimensional evaluation method of recreation degree of forest parks according to claim 5 is characterized in that: The method further comprises: In the process of constructing the measurement model, calculating the combined reliability and convergent validity of the measurement model; Until the combined reliability meets a preset combined reliability threshold, and the convergent validity meets a preset convergent validity threshold; The combined reliability is expressed by formula (13), which is as follows: ； In formula (13), For the The combined reliability of latent variables is For the The measurement error variance of the observed variables; The convergent validity is expressed by formula (14), which is as follows: ； In formula (14), For the Convergent validity of latent variables.

7. A multi-dimensional evaluation system for the recreation degree of forest parks, characterized in that: The method for evaluating the recreation degree of a forest park in multiple dimensions according to any one of claims 1 to 6 is applicable to the system comprising: An information processing module, the information processing module is used to obtain environmental image information of the current forest park, the environmental image information includes panoramic image information and local image information, the local image information is an image containing part of the environment, a first environmental feature is generated according to the environmental image information, and, obtain multiple service information of the current forest park, each of the service information includes the service category, service group, service address and service item information of the current service point, the service item information includes the service commodity, service item and service price of the current service point, a first service feature is generated according to the multiple service information, and, obtain social information of the current forest park, the social information includes visitor comment information, management change information and park information of the current forest park, and a first social feature of the current forest park is generated according to the social information; An evaluation module is used to construct a PLS-SEM model, wherein the PLS-SEM model includes a measurement model and a structural model, wherein environmental factors are used as the first latent variables of the PLS-SEM model, service factors are used as the second latent variables of the PLS-SEM model, and social factors are used as the third latent variables of the PLS-SEM model; the first environmental feature is used as the first observed variable corresponding to the first latent variable, the first service feature is used as the second observed variable corresponding to the second latent variable, and the first social feature is used as the third observed variable corresponding to the third latent variable to construct the measurement model; a first hypothesis relationship between the first latent variable and the recreation degree is set, a second hypothesis relationship between the second latent variable and the recreation degree is set, and a third hypothesis relationship between the third latent variable and the recreation degree is set; and according to the first hypothesis relationship, the second hypothesis relationship and the third The structural model is constructed based on the assumed relationship; the first latent variable is quantified according to the first observed variable in the measurement model to obtain a first quantification result, and the second latent variable is quantified according to the second observed variable in the measurement model to obtain a second quantification result, and the third latent variable is quantified according to the third observed variable in the measurement model to obtain a third quantification result; the first quantification result, the second quantification result and the third quantification result are input into the structural model to calculate the goodness of fit index of the first latent variable, the second latent variable and the third latent variable to the recreation degree; the recreation degree evaluation result and improvement suggestions of the current forest park are obtained, the recreation degree evaluation result includes the quantitative contribution of the first environmental feature, the first service feature and the first social feature to the recreation degree, and the improvement suggestions include the second environmental feature, the second service feature and the second social feature.

8. A computer-readable storage medium storing computer program instructions, characterized in that: The computer program instructions implement the method according to any one of claims 1 to 6 when executed by a processor.

9. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method according to any one of claims 1 to 6.

Images