Urban sound scene perception extraction method and system based on text big data

Abstract

The invention discloses an urban sound scene perception extraction method and system based on text big data, and the method comprises the steps: obtaining historical text data with a geographic position mark and a timestamp, and carrying out the preprocessing of the historical text data, and obtaining a sound related data set; sampling and labeling the sound related data set to obtain sound scene multi-dimensional data, and constructing a sound scene perception model according to the sound scene multi-dimensional data; and obtaining current text data, inputting the current text data into the sound scene perception model, and outputting an urban sound scene perception prediction result. According to the method, large-scale historical text data from social media can be efficiently processed by utilizing natural language processing and machine learning technologies, and text information related to sound scenes is automatically screened, so that the urban sound environment data processing efficiency is greatly improved, the high cost of a traditional monitoring means is reduced, and the urban sound environment data processing efficiency is improved. And the accuracy of acquiring the urban sound environment monitoring result is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, system, terminal and computer-readable storage medium for extracting urban soundscape perception based on text big data. Background Art

[0002] Current research on urban soundscapes focuses on the monitoring and control of noise pollution. The main method is sound pressure level analysis based on physical acoustics, which can identify the quality of urban sound environment and reduce noise pollution to a certain extent.

[0003] However, existing technologies are overly dependent on the collection of physical data. For example, traditional noise monitoring requires the deployment of a large number of sensor devices, which is costly and has limited data coverage. At the same time, the data collected by the sensors are mostly local noise information, making it difficult to comprehensively monitor the sound environment level in a large area, resulting in inaccurate urban soundscape monitoring results, and staff are unable to adjust and optimize the urban soundscape according to the actual urban soundscape conditions.

[0004] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0005] The main purpose of the present invention is to provide a method, system, terminal and computer-readable storage medium for extracting urban soundscape perception based on text big data, aiming to solve the problem that the existing technology overly relies on the collection of physical data, ignores the subjective perception of users, and cannot comprehensively monitor the sound environment level in a larger area, resulting in the obtained urban soundscape monitoring results being inaccurate, and the staff cannot adjust and optimize the urban soundscape according to the actual urban sound environment.

[0006] To achieve the above object, the present invention provides a method for extracting urban soundscape perception based on text big data, and the method for extracting urban soundscape perception based on text big data comprises the following steps: Acquire historical text data with geographic location tags and timestamps, and preprocess the historical text data to obtain a sound-related data set; Performing sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and constructing a soundscape perception model based on the multi-dimensional soundscape data; The current text data is acquired, and the current text data is input into the soundscape perception model, and the city soundscape perception prediction result is output.

[0007] Optionally, the urban soundscape perception extraction method based on text big data, wherein the step of obtaining historical text data with geographic location tags and timestamps and preprocessing the historical text data to obtain a sound-related data set specifically includes: Acquire historical text data with geographic location tags and timestamps, and construct an initial data set based on the historical text data, wherein the initial data set includes user comment information, geographic location information, release time information, and text information; Preprocessing the initial data set to obtain a feature vector; City soundscape keywords are determined, and sound-related data extraction processing is performed on the feature vector according to the city soundscape keywords to obtain a sound-related data set.

[0008] Optionally, the urban soundscape perception extraction method based on text big data, wherein the preprocessing of the initial data set to obtain a feature vector specifically includes: Performing data cleaning and text formatting processing on the initial data set to obtain initial text data; The initial text data is subjected to word segmentation processing and semantic analysis processing to obtain the feature vector.

[0009] Optionally, in the urban soundscape perception extraction method based on text big data, the sampling and annotation processing includes initial sampling and annotation processing and soundscape feature sampling and annotation processing; The sampling and labeling process of the sound-related data set to obtain multi-dimensional soundscape data specifically includes: Using natural language processing technology to perform initial sampling and annotation processing on the sound-related data set to obtain an urban soundscape data set; The natural language processing technology is used to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain multi-dimensional soundscape data.

[0010] Optionally, the urban soundscape perception extraction method based on text big data, wherein the initial sampling and annotation processing of the sound-related data set using natural language processing technology to obtain the urban soundscape data set specifically includes: Performing initial sampling and annotation processing on the sound-related data set using the natural language processing technology to obtain a first annotated data set, and dividing the first annotated data set into a first training set and a first test set; Determine a first preset model, perform model training on the first preset model according to the first training set, and perform model fine-tuning on the first preset model according to the first test set to obtain an urban soundscape model; The sound-related data set is input into the urban soundscape model to obtain an urban soundscape data set.

[0011] Optionally, the urban soundscape perception extraction method based on text big data, wherein the step of using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain soundscape multi-dimensional data, specifically includes: Using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data, and dividing the urban soundscape specific dimension data into a second training set and a second test set; Determine a second preset model, perform model training on the second preset model according to the second training set, and perform model fine-tuning on the second preset model according to the second test set to obtain a soundscape specific dimensional model, and obtain a multi-dimensional model of urban soundscape after multiple iterations of model training and model fine-tuning; The urban soundscape dataset is input into the soundscape multidimensional model to obtain soundscape multidimensional data.

[0012] Optionally, the urban soundscape perception extraction method based on text big data, wherein the step of using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data, specifically includes: Performing preset sound source labeling processing on the urban soundscape dataset according to the sound source type, and extracting data belonging to the preset sound source type in the urban soundscape dataset to obtain preset sound source type data; The preset sound source type data is subjected to specific dimension data labeling processing, and the specific dimension data in the preset sound source type data is extracted to obtain the specific dimension data of the urban soundscape.

[0013] In addition, to achieve the above-mentioned purpose, the present invention also provides an urban soundscape perception extraction system based on text big data, wherein the urban soundscape perception extraction system based on text big data comprises: A data set construction module is used to obtain historical text data with geographic location tags and time stamps, and pre-process the historical text data to obtain a sound-related data set; A soundscape perception model generation module is used to perform sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and to construct a soundscape perception model based on the multi-dimensional soundscape data; The prediction result generation module is used to obtain current text data, input the current text data into the soundscape perception model, and output the urban soundscape perception prediction result.

[0014] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and an urban soundscape perception extraction program based on text big data stored in the memory and executable on the processor, wherein the urban soundscape perception extraction program based on text big data implements the steps of the urban soundscape perception extraction method based on text big data when executed by the processor.

[0015] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an urban soundscape perception extraction program based on text big data, and when the urban soundscape perception extraction program based on text big data is executed by a processor, the steps of the urban soundscape perception extraction method based on text big data are implemented as described above.

[0016] In the present invention, historical text data with geographic location tags and timestamps are obtained, and the historical text data are preprocessed to obtain a sound-related data set; the sound-related data set is sampled and labeled to obtain multi-dimensional soundscape data, and a soundscape perception model is constructed based on the multi-dimensional soundscape data; current text data is obtained, and the current text data is input into the soundscape perception model to output the city soundscape perception prediction result. By obtaining historical text data with geographic location tags and timestamps, the present invention uses natural language processing and machine learning technology to efficiently process historical text data from social media and automatically filter text information related to soundscapes, which can greatly improve the efficiency of urban soundscape data processing and effectively reduce the high cost problem of traditional urban soundscape data monitoring methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of a preferred embodiment of the urban soundscape perception and extraction method based on text big data of the present invention; Figure 2 It is a schematic diagram of the automatic screening process of text big data of a preferred embodiment of the urban soundscape perception and extraction method based on text big data of the present invention; Figure 3 It is a schematic diagram of the recognition and classification process of multi-dimensional features of soundscapes in a preferred embodiment of the urban soundscape perception and extraction method based on text big data of the present invention; Figure 4 It is a dynamic generation and prediction flow chart of a soundscape perception model of a preferred embodiment of the urban soundscape perception extraction method based on text big data of the present invention; Figure 5 It is a structural diagram of a preferred embodiment of the urban soundscape perception and extraction system based on text big data of the present invention; Figure 6 It is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0019] At present, the research on urban soundscape is mostly focused on the monitoring and control of noise pollution. The main method is sound pressure level analysis based on physical acoustics. This method can identify the quality of urban sound environment and reduce noise pollution to a certain extent, but it also has the following shortcomings: 1. Dependence on the physical environment and neglect of subjective perception: Existing sound environment monitoring technologies, such as noise maps and sound pressure level analysis, are usually based on physical data and are difficult to reflect residents' subjective perception of the sound environment. Although this method can accurately describe the physical distribution of the sound environment, it fails to capture residents' subjective experience of the urban sound environment, resulting in poor results in practical applications. 2. High-cost data limits the development of large-scale urban soundscape research: Traditional noise monitoring requires a large number of sensor equipment to be deployed, which is costly and has limited data coverage. At the same time, the data collected by the sensor is mostly local noise information, making it difficult to comprehensively monitor the sound environment level in a large area. 3. Limited capabilities for refined analysis and noise management: Existing technologies usually rely on the measurement of overall noise levels, lack accurate identification of different noise sources, and are mostly identified indirectly through methods such as machine learning. This method makes it difficult to provide specific and targeted intervention measures for urban noise management, resulting in noise control strategies that are often too general and cannot be effectively adjusted and optimized according to actual conditions.

[0020] In order to solve the problems existing in the prior art, such as data ignoring subjective perception, difficulty in effectively conducting large-scale soundscape research, and difficulty in directly identifying sound sources, the present invention proposes an urban soundscape perception extraction method based on text big data. The method utilizes large-scale text data obtained from social media, etc., combines natural language processing and machine learning technology, automatically screens out text information related to soundscape perception, and accurately extracts residents' subjective perception of the urban sound environment. Based on this, multi-dimensional element extraction of the urban sound environment and dynamic urban soundscape mapping can be continued in the future.

[0021] The urban soundscape perception extraction method based on text big data described in the preferred embodiment of the present invention is as follows: Figure 1 As shown, the urban soundscape perception extraction method based on text big data includes the following steps: Step S10: Acquire historical text data with geographic location tags and time stamps, and preprocess the historical text data to obtain a sound-related data set.

[0022] Specifically, historical text data with geographic location tags and timestamps are obtained, and an initial data set is constructed according to the historical text data, wherein the initial data set includes user comment information, geographic location information, release time information and text information; data cleaning processing and text formatting processing are performed on the initial data set to obtain initial text data; word segmentation processing and semantic analysis processing are performed on the initial text data to obtain the feature vector; urban soundscape keywords are determined, and sound-related data extraction processing is performed on the feature vector according to the urban soundscape keywords to obtain a sound-related data set.

[0023] like Figure 2 As shown, the automatic screening process for text big data in the present invention is mainly completed by a data collection module, a data preprocessing module and a data screening module, wherein the data collection module: obtains text data with geographic location tags and timestamps by calling the API interface of the social media platform to form an initial data set, wherein the collected data includes user comment information, geographic location information, release time information and text information. Data preprocessing module: cleans the collected data, removes invalid or noise information, performs text formatting, and then performs word segmentation and semantic analysis on the text data to generate a feature vector. Furthermore, keywords related to urban soundscapes (such as noise, tranquility, etc.) are extracted through preset keyword extraction software (the preset keyword extraction software generates a keyword library and then filters the text data), and data that may be related to city-related urban sounds are extracted based on this keyword data set.

[0024] In the specific implementation process, the soundscape perception extraction process based on social media data is: obtain text data with geographic location tags through social media platforms, use natural language processing technology to process the text, so as to extract information related to soundscape perception. This implementation method is suitable for data collection on open social platforms and is suitable for large-scale urban soundscape analysis.

[0025] The present invention can be used as a supplement to the collection of traditional sensor data: the present invention focuses on perceiving urban soundscapes through big data, which can effectively supplement the deficiencies of traditional sensors in capturing soundscape data. While some sensor data physically quantifies the environmental soundscape, the present invention can use big data such as social media to perform subjective soundscape perception analysis, thus improving the single method of traditional soundscape research that relies on physical measurement. Through this supplementary mechanism, the present invention can not only improve the accuracy of noise source identification, but also provide more comprehensive and complete data support for the dynamic prediction of soundscapes, thus promoting the in-depth development of soundscape research.

[0026] In addition, the present invention also provides data collection based on mobile crowdsourcing perception: through mobile applications or public platforms, residents are encouraged to participate in the collection of noise data, and crowdsourcing perception technology is used to obtain a large amount of user feedback data related to noise. This method is suitable for supplementary data collection in local areas in the city, and public feedback is combined with social media data to further improve the accuracy of the soundscape perception model.

[0027] The automatic screening and perception extraction process based on text big data in the present invention: The present invention proposes an automatic screening and perception extraction method based on text big data, which can identify text data related to urban soundscape perception through natural language processing (NLP) and machine learning technology.

[0028] Based on this, the present invention can also be further implemented through the following combination schemes: Scheme 1: Combine keyword extraction and large language model to filter out user feedback data related to soundscape perception through text semantic analysis. Scheme 2: Through the hierarchical classification model, the text is screened from top to bottom at multiple levels, first identifying sound-related data, and then further extracting sound source information and noise evaluation information.

[0029] The present invention constructs a low-cost, efficient large-scale analysis framework for soundscape perception: the present invention reduces the high cost of relying on sensor equipment in traditional noise monitoring through social media big data resources, and greatly improves data coverage and real-time performance. It can automatically capture urban social media data and conduct large-scale dynamic analysis of urban soundscapes.

[0030] Based on this, the present invention can also be further implemented through the following combination schemes: Scheme 1: Combined with mobile crowdsourcing perception technology, the soundscape perception model is updated in real time through public feedback data to form a dynamic soundscape map. Scheme 2: Use machine learning models to fuse historical data and real-time data to generate a time series analysis of urban soundscape perception and predict future soundscape change trends.

[0031] Step S20: performing sampling and annotation processing on the sound-related data set to obtain soundscape multi-dimensional data, and constructing a soundscape perception model according to the soundscape multi-dimensional data, wherein the sampling and annotation processing includes initial sampling and annotation processing and soundscape feature sampling and annotation processing.

[0032] Specifically, the natural language processing technology is used to perform initial sampling and labeling processing on the sound-related data set to obtain a first labeled data set, and the first labeled data set is divided into a first training set and a first test set; a first preset model is determined, and the first preset model is trained according to the first training set, and the first preset model is fine-tuned according to the first test set to obtain an urban soundscape model; the sound-related data set is input into the urban soundscape model to obtain an urban soundscape data set.

[0033] like Figure 2 As shown, a data screening module is provided in the present invention: based on natural language processing technology, the present invention classifies texts (by means of initial sampling and labeling) and screens out valid texts related to soundscape perception.

[0034] Further, the urban soundscape dataset is subjected to preset sound source labeling processing according to the sound source type, and data belonging to the preset sound source type in the urban soundscape dataset is extracted to obtain preset sound source type data; specific dimension data labeling processing is performed on the preset sound source type data, and specific dimension data in the preset sound source type data is extracted to obtain urban soundscape specific dimension data, and the urban soundscape specific dimension data is divided into a second training set and a second test set; a second preset model is determined, model training is performed on the second preset model according to the second training set, and model fine-tuning is performed on the second preset model according to the second test set to obtain a soundscape specific dimension model, and a multi-dimensional model of urban soundscape is obtained after multiple iterations of model training and model fine-tuning; the urban soundscape dataset is input into the soundscape multi-dimensional model to obtain soundscape multi-dimensional data.

[0035] Among them, the multi-dimensional data of soundscape mainly refers to the multi-dimensional characteristics of soundscape, including but not limited to sound source, emotional tendency, etc. The specific dimensional data of urban soundscape refers to the characteristic data of a specific dimensionality in the multi-dimensional characteristics of soundscape (machine learning can only filter one dimension at a time).

[0036] like Figure 3 As shown in the figure, for the identification and classification process of multi-dimensional features of soundscape: the present invention sets up a hierarchical screening module, which further classifies the initially screened text, subdivides it according to soundscape dimensions such as sound source type, combines time and geographical information, and generates multi-dimensional data tags. Figure 3 The invention describes how to use this method to extract urban noise data (this is an example, urban noise data is a specific dimension data in the urban soundscape, and the invention mainly introduces noise data as a representative of multi-dimensional data).

[0037] like Figure 4As shown, for the dynamic generation and prediction process of the soundscape perception model, a dynamic perception map generation module and a prediction model module are set in the present invention.

[0038] Among them, the dynamic perception map generation module: crawls and filters social media big data in real time, maps text content to urban space, and generates real-time soundscape perception maps, which can reflect the soundscape perception status of different areas of the city. At the same time, the soundscape perception map can also show the distribution of sound sources and their impact.

[0039] Prediction model module: Use time series analysis methods to process the processed historical data to predict the changing trend of future urban soundscapes and generate future soundscape heat maps to provide decision support for urban sound environment planning.

[0040] Regarding soundscape perception analysis combined with geographic information system (GIS): Based on text data, the present invention further combines geographic information system technology to analyze the distribution and change trends of sound environments in different geographical areas. This implementation method is suitable for the refined management of urban sound environments and can provide urban planners with more intuitive noise heat maps and future predictions.

[0041] Accurate identification and classification mechanism for multiple noise sources: The present invention constructs a direct and accurate identification mechanism for different noise sources through social media text, overcoming the limitation of the prior art that only relies on the measurement of the overall noise level.

[0042] Based on this, the present invention also provides a combination scheme: Scheme 1: By integrating geographic information and text perception data, different noise sources such as traffic, machinery, and nature are classified and identified, and time and space tags are introduced into the data to achieve dynamic monitoring of noise sources. Scheme 2: Use deep learning models (such as convolutional neural networks CNN, random forests, etc.) to train noise source data to achieve automatic classification and hierarchical evaluation of multiple noise sources.

[0043] Regarding the personalized urban sound environment perception mechanism and future soundscape prediction: the present invention introduces the subjective feedback of the public and incorporates personalized noise perception into the urban noise management system through the mining of social media data, thus solving the problem that the existing technology ignores individual needs and public participation.

[0044] Based on this, the present invention also provides a combined solution: establishing a user perception scoring system based on social media feedback, generating a subjective perception map of urban noise for different groups of people, collecting subjective soundscape perception data in real time, and constructing a future soundscape prediction model.

[0045] Step S30: acquiring current text data, inputting the current text data into the soundscape perception model, and outputting a city soundscape perception prediction result.

[0046] After the soundscape perception model is constructed, the present invention can perform noise data analysis on the current text data collected in real time. The specific process is: obtaining the current text data through various media channels, and inputting the current text data into the constructed soundscape perception model, and the soundscape perception model will automatically analyze the noise data of the current text data, and automatically generate a real-time soundscape perception map and a future soundscape heat map, so as to predict the urban soundscape.

[0047] In summary, the present invention proposes a method for extracting urban soundscape perception based on text big data, which uses natural language processing and machine learning technology to automatically screen perceptual information related to urban soundscape from large-scale text data, thereby providing a scientific basis for urban planning and sound environment management. The technical content includes the following parts: 1. Automatic screening and extraction method for urban soundscape perception based on big data, including keyword extraction, semantic analysis and multi-level screening model. 2. Identification and classification method for multi-dimensional features of soundscape, especially a solution for accurately identifying multi-dimensional feature sources of soundscape by combining social media data and geographic information. 3. A low-cost, large-scale soundscape perception analysis framework, which uses big data and machine learning technology to dynamically monitor and predict urban soundscape. 3. Personalized noise perception mechanism and future soundscape prediction model, allowing the public to feedback noise perception through social media and other platforms, and use these feedbacks for future soundscape predictions and incorporate them into noise management solutions.

[0048] Technical effects of the present invention: 1. Efficient large-scale text data processing: The present invention uses natural language processing and machine learning technologies to efficiently process large-scale text data from social media and automatically filter text information related to soundscapes, which greatly improves the efficiency of data processing and reduces the high cost of traditional noise monitoring methods.

[0049] 2. Dynamic soundscape perception and prediction: By combining geographic information and time series analysis, it is possible to generate real-time urban soundscape perception maps and predict future changes in the sound environment, providing accurate and real-time data support for urban planners.

[0050] 3. Effects of big text data screening: Through the combination of keyword extraction and large language models, data related to urban soundscape perception can be accurately screened, significantly improving the relevance and accuracy of the data. Compared with traditional methods, this invention can process large-scale data in a short time and avoid the inefficiency of manual screening.

[0051] 4. Effect of multi-dimensional information recognition of soundscape: By combining text and multi-source data, the classification and recognition of sound sources is more accurate. Compared with traditional sound environment monitoring methods, the present invention can distinguish the different dimensional characteristics of soundscape and formulate targeted strategies to effectively improve the urban sound environment.

[0052] 5. The effect of the public feedback mechanism: Through the real-time collection and processing of public feedback, more targeted sound environment management solutions can be provided to urban planners, ensuring that the management measures can meet the needs of different groups and improving the scientificity and flexibility of urban sound environment management.

[0053] In addition, in order to better protect the core technology of the present invention and prevent possible circumvention design solutions adopted by competitors, possible design changes or alternative solutions of the present invention can be considered from the following aspects: 1. Alternative or deformation of machine learning algorithms: Alternative solution: The present invention relies on specific machine learning algorithms (such as BERT, LDA, etc.) to screen text data and identify noise sources. Competitors may use different machine learning algorithms, such as support vector machines (SVM), convolutional neural networks (CNN) or other deep learning algorithms, to achieve similar noise source identification functions, thereby circumventing the core technical solution of the present invention. Countermeasures: Expand the innovation of the present invention to multiple algorithms, and the protection is not limited to specific machine learning models, but covers the overall framework and technical ideas of text classification, sound source identification and prediction.

[0054] 2. Alternative solutions for personalized sound environment perception feedback: The present invention uses public feedback in social media combined with natural language processing technology to extract personalized sound environment perception information in real time. Competitors may collect users’ subjective perception feedback on the urban sound environment through different public participation channels, such as through mobile applications, online crowdsourcing platforms or IoT devices, and build similar personalized sound environment perception mechanisms. Countermeasures: Protect the technical details of public feedback in the present invention, including the collection, processing and application of feedback, to ensure that no matter what form of public participation is used, it falls within the technical protection scope of the present invention.

[0055] 3. Maintain the flexibility of the algorithm to ensure that no matter what machine learning or deep learning algorithm is used, the technical ideas of text screening, sound source recognition and perception extraction are within the technical protection scope of the present invention.

[0056] 4. Strengthen patent protection for public feedback mechanisms, covering different channels of public participation to counter competitors’ alternatives of collecting user perception information through other means.

[0057] Furthermore, if Figure 5As shown, based on the above-mentioned urban soundscape perception extraction method based on text big data, the present invention also provides an urban soundscape perception extraction system based on text big data, wherein the urban soundscape perception extraction system based on text big data includes: A data set construction module 51 is used to obtain historical text data with geographic location tags and time stamps, and pre-process the historical text data to obtain a sound-related data set; A soundscape perception model generation module 52 is used to perform sampling and annotation processing on the sound-related data set to obtain soundscape multi-dimensional data, and to construct a soundscape perception model based on the soundscape multi-dimensional data; The prediction result generation module 53 is used to obtain current text data, input the current text data into the soundscape perception model, and output the city soundscape perception prediction result.

[0058] Furthermore, if Figure 6 As shown, based on the above-mentioned urban soundscape perception extraction method and system based on text big data, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 6 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0059] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, the memory 20 stores a city soundscape perception extraction program 40 based on text big data, and the city soundscape perception extraction program 40 based on text big data can be executed by the processor 10, thereby realizing the city soundscape perception extraction method based on text big data in the present application.

[0060] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, and is used to run the program code or process data stored in the memory 20, such as executing the urban soundscape perception extraction method based on text big data.

[0061] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The components 10-30 of the terminal communicate with each other via a system bus.

[0062] In one embodiment, when the processor 10 executes the urban soundscape perception extraction program 40 based on text big data in the memory 20, the following steps are implemented: Acquire historical text data with geographic location tags and timestamps, and preprocess the historical text data to obtain a sound-related data set; Performing sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and constructing a soundscape perception model based on the multi-dimensional soundscape data; The current text data is acquired, and the current text data is input into the soundscape perception model, and the city soundscape perception prediction result is output.

[0063] The step of obtaining historical text data with geographic location tags and timestamps and preprocessing the historical text data to obtain a sound-related data set specifically includes: Acquire historical text data with geographic location tags and timestamps, and construct an initial data set based on the historical text data, wherein the initial data set includes user comment information, geographic location information, release time information, and text information; Preprocessing the initial data set to obtain a feature vector; City soundscape keywords are determined, and sound-related data extraction processing is performed on the feature vector according to the city soundscape keywords to obtain a sound-related data set.

[0064] The preprocessing of the initial data set to obtain a feature vector specifically includes: Performing data cleaning and text formatting processing on the initial data set to obtain initial text data; The initial text data is subjected to word segmentation processing and semantic analysis processing to obtain the feature vector.

[0065] The sampling and annotation processing includes initial sampling and annotation processing and soundscape feature sampling and annotation processing; The sampling and labeling process of the sound-related data set to obtain multi-dimensional soundscape data specifically includes: Using natural language processing technology to perform initial sampling and annotation processing on the sound-related data set to obtain an urban soundscape data set; The natural language processing technology is used to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain multi-dimensional soundscape data.

[0066] The process of using natural language processing technology to perform initial sampling and annotation processing on the sound-related data set to obtain an urban soundscape data set specifically includes: Performing initial sampling and annotation processing on the sound-related data set using the natural language processing technology to obtain a first annotated data set, and dividing the first annotated data set into a first training set and a first test set; Determine a first preset model, perform model training on the first preset model according to the first training set, and perform model fine-tuning on the first preset model according to the first test set to obtain an urban soundscape model; The sound-related data set is input into the urban soundscape model to obtain an urban soundscape data set.

[0067] The process of using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain multi-dimensional soundscape data specifically includes: Using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data, and dividing the urban soundscape specific dimension data into a second training set and a second test set; Determine a second preset model, perform model training on the second preset model according to the second training set, and perform model fine-tuning on the second preset model according to the second test set to obtain a soundscape specific dimensional model, and obtain a multi-dimensional model of urban soundscape after multiple iterations of model training and model fine-tuning; The urban soundscape dataset is input into the soundscape multidimensional model to obtain soundscape multidimensional data.

[0068] The process of using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data specifically includes: Performing preset sound source labeling processing on the urban soundscape dataset according to the sound source type, and extracting data belonging to the preset sound source type in the urban soundscape dataset to obtain preset sound source type data; The preset sound source type data is subjected to specific dimension data labeling processing, and the specific dimension data in the preset sound source type data is extracted to obtain the specific dimension data of the urban soundscape.

[0069] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an urban soundscape perception extraction program based on text big data, and when the urban soundscape perception extraction program based on text big data is executed by a processor, the steps of the urban soundscape perception extraction method based on text big data are implemented as described above.

[0070] In summary, the present invention provides a method for extracting urban soundscape perception based on text big data and related equipment, the method comprising: obtaining historical text data with geographic location tags and timestamps, and preprocessing the historical text data to obtain a sound-related data set; performing sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and constructing a soundscape perception model based on the multi-dimensional soundscape data; obtaining current text data, and inputting the current text data into the soundscape perception model, and outputting urban soundscape perception prediction results. The present invention can efficiently process historical text data from social media and automatically filter text information related to soundscape by obtaining historical text data with geographic location tags and timestamps, and by using natural language processing and machine learning technologies, which can greatly improve the efficiency of urban soundscape data processing and effectively reduce the high cost problem of traditional urban soundscape data monitoring methods.

[0071] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or terminal including the element.

[0072] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.

[0073] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for extracting urban soundscape perception based on text big data, characterized in that: The urban soundscape perception extraction method based on text big data includes: Acquire historical text data with geographic location tags and timestamps, and preprocess the historical text data to obtain a sound-related data set; Performing sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and constructing a soundscape perception model based on the multi-dimensional soundscape data; The current text data is acquired, and the current text data is input into the soundscape perception model, and the city soundscape perception prediction result is output.

2. The urban soundscape perception extraction method based on text big data according to claim 1 is characterized in that: The acquiring of historical text data with geographic location tags and timestamps, and preprocessing of the historical text data to obtain a sound-related data set specifically includes: Acquire historical text data with geographic location tags and timestamps, and construct an initial data set based on the historical text data, wherein the initial data set includes user comment information, geographic location information, release time information, and text information; Preprocessing the initial data set to obtain a feature vector; City soundscape keywords are determined, and sound-related data extraction processing is performed on the feature vector according to the city soundscape keywords to obtain a sound-related data set.

3. The urban soundscape perception extraction method based on text big data according to claim 2 is characterized in that: The preprocessing of the initial data set to obtain a feature vector specifically includes: Performing data cleaning and text formatting processing on the initial data set to obtain initial text data; The initial text data is subjected to word segmentation processing and semantic analysis processing to obtain the feature vector.

4. The urban soundscape perception extraction method based on text big data according to claim 1 is characterized in that: The sampling and annotation processing includes initial sampling and annotation processing and soundscape feature sampling and annotation processing; The sampling and labeling process of the sound-related data set to obtain multi-dimensional soundscape data specifically includes: Using natural language processing technology to perform initial sampling and annotation processing on the sound-related data set to obtain an urban soundscape data set; The natural language processing technology is used to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain multi-dimensional soundscape data.

5. The urban soundscape perception extraction method based on text big data according to claim 4 is characterized in that: The initial sampling and labeling processing of the sound-related data set by using natural language processing technology to obtain the urban soundscape data set specifically includes: Performing initial sampling and annotation processing on the sound-related data set using the natural language processing technology to obtain a first annotated data set, and dividing the first annotated data set into a first training set and a first test set; Determine a first preset model, perform model training on the first preset model according to the first training set, and perform model fine-tuning on the first preset model according to the first test set to obtain an urban soundscape model; The sound-related data set is input into the urban soundscape model to obtain an urban soundscape data set.

6. The urban soundscape perception extraction method based on text big data according to claim 4 is characterized in that: The adopting of the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain multi-dimensional soundscape data specifically includes: Using the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data, and dividing the urban soundscape specific dimension data into a second training set and a second test set; Determine a second preset model, perform model training on the second preset model according to the second training set, and perform model fine-tuning on the second preset model according to the second test set to obtain a soundscape specific dimensional model, and obtain a multi-dimensional model of urban soundscape after multiple iterations of model training and model fine-tuning; The urban soundscape dataset is input into the soundscape multidimensional model to obtain soundscape multidimensional data.

7. The urban soundscape perception extraction method based on text big data according to claim 6 is characterized in that: The adopting of the natural language processing technology to perform soundscape feature sampling and annotation processing on the urban soundscape dataset to obtain urban soundscape specific dimension data specifically includes: Performing preset sound source labeling processing on the urban soundscape dataset according to the sound source type, and extracting data belonging to the preset sound source type in the urban soundscape dataset to obtain preset sound source type data; The preset sound source type data is subjected to specific dimension data labeling processing, and the specific dimension data in the preset sound source type data is extracted to obtain the specific dimension data of the urban soundscape.

8. An urban soundscape perception and extraction system based on text big data, characterized in that: The urban soundscape perception and extraction system based on text big data includes: A data set construction module is used to obtain historical text data with geographic location tags and time stamps, and pre-process the historical text data to obtain a sound-related data set; A soundscape perception model generation module is used to perform sampling and annotation processing on the sound-related data set to obtain multi-dimensional soundscape data, and to construct a soundscape perception model based on the multi-dimensional soundscape data; The prediction result generation module is used to obtain current text data, input the current text data into the soundscape perception model, and output the urban soundscape perception prediction result.

9. A terminal, characterized in that: The terminal includes: a memory, a processor, and a city soundscape perception extraction program based on text big data stored in the memory and executable on the processor, wherein the city soundscape perception extraction program based on text big data, when executed by the processor, implements the steps of the city soundscape perception extraction method based on text big data according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a city soundscape perception extraction program based on text big data, and when the city soundscape perception extraction program based on text big data is executed by a processor, the steps of the city soundscape perception extraction method based on text big data are implemented as described in any one of claims 1 to 7.

Images