Intelligent community life service recommendation system and method based on big data analysis

Abstract

The invention relates to the technical field of community life management, in particular to an intelligent community life service recommendation system and method based on big data analysis, and the system comprises a resident demand analysis module, a provider service resource module, a matching degree acquisition and correction module, and a resident demand dynamic matching monitoring module. The resident demand analysis module comprehensively collects and analyzes individual demands of residents and provides a data basis for recommendation service; the provider service resource module effectively integrates and quantifies resource information of service providers, and matching and evaluation are facilitated; the matching degree acquisition and correction module calculates and analyzes the matching degree and corrects in combination with the position information to ensure efficient preliminary screening of the supplier and the demander; and the resident demand dynamic matching monitoring module realizes dynamic deep matching between resident demands and service providers by using real-time context awareness and data fusion technologies. The technical problem that the current community life management effect is poor is solved.

Description

Technical Field

[0001] The present invention relates to the field of community life management technology, and in particular to an intelligent community life service recommendation system and method based on big data analysis. Background Art

[0002] The existing community life management has the following problems: asymmetric supply and demand information, lack of effective docking and communication channels between residents' demand information and service providers, resulting in mismatch between demand and service; there are many and scattered life service providers in the community, but they are often scattered and lack unified management and integration, which makes residents feel confused and inconvenient when choosing services; lack of dynamic matching and monitoring mechanism, the current community life management cannot achieve dynamic matching and real-time monitoring between residents' needs and service providers, affecting the timeliness and effectiveness of services. To this end, the present invention proposes an intelligent community life service recommendation system and method based on big data analysis. Summary of the invention

[0003] The purpose of the present invention is to solve the problems in the background technology and to propose an intelligent community life service recommendation system and method based on big data analysis.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] The intelligent community life service recommendation system based on big data analysis includes: resident demand analysis module, provider service resource module, matching degree acquisition and correction module, and resident demand dynamic matching monitoring module;

[0006] Resident demand analysis module: obtains residents' personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and extracts features, and constructs residents' demand feature vectors based on the extracted features;

[0007] Provider service resource module: organizes service provider information including service type, service scope and service price, constructs service resource quantitative vector, and calculates service resource quantitative evaluation value;

[0008] Matching degree acquisition and correction module: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening;

[0009] Resident demand dynamic matching monitoring module: establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

[0010] Furthermore, the process of obtaining residents' personalized demand information and extracting features by the resident demand analysis module includes:

[0011] Collect and pre-process each resident's personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records;

[0012] The resident information record includes age and gender. For the resident information record, the actual age value of resident i is obtained, and the resident age is divided into different intervals and marked as ra. i =(1,2,3,4); Get the resident's gender information rg i =(0,1), with males as 1 and females as 0; the different age ranges of residents include adolescents, young adults, middle-aged people, and the elderly, and are coded with numbers as 1, 2, 3, and 4 respectively;

[0013] For dietary preference records, create a 1 List of dietary combinations, where m 1 is the number of dietary combinations; for resident i, the fuzzy logic method is used to quantify dietary preferences as follows:

[0014] Step U1: Define μ i(ι) is the membership of resident i to the ιth diet combination, where ι is the index of the diet combination type, and ι=1,2,…,m 1 , the value range of ι is [0,1];

[0015] Step U2: Integrate to obtain the dietary preference feature vector of resident i:

[0016]

[0017] Where, υ is the dimension of dietary preference feature vector;

[0018] For household goods consumption records, determine m 2 Types of household items, including m 2 is the number of household items; collects the consumption amount of resident i on the kth household item in multiple time periods t Where k is the index of household products. The future consumption trend is predicted by time series analysis method as follows:

[0019] Step V1: Consumption amount Considered as a time series, the prediction equation is obtained using the autoregressive moving average model:

[0020]

[0021] In the formula, rc′ i(k) is the predicted consumption amount of household goods by resident i for the kth household product, p, q are the orders of the autoregressive moving average model, J is the index of the order of the autoregressive moving average model, θ J are the autoregressive moving average model parameters, is a white noise sequence;

[0022] Step V2: Calculate the characteristic value rh of household goods consumption by resident i i :

[0023]

[0024] In the formula, w k is the weight of the kth household item, is the average consumption amount of all residents on the kth household product;

[0025] For the cleaning and maintenance records of home appliances, a preset period is set to check the cleaning and maintenance status of home appliances of resident i in the period. If there is a cleaning and maintenance record, it is recorded as 1, otherwise it is recorded as 0. At the same time, the number of types of home appliances of resident i is counted, and the type distribution of repaired home appliances is analyzed to obtain the characteristic vector of home appliance cleaning and maintenance frequency

[0026]

[0027] In the formula, represents the proportion of the number of repairs of the first type of household appliances of resident i, the proportion of the number of repairs of the second type of household appliances, ..., The proportion of repair times of each type of household appliances, is the dimension of the feature vector of the frequency of cleaning and maintenance of household appliances;

[0028] Feature extraction is performed based on the preprocessed personalized demand information of residents: the low-dimensional resident features are mapped to the high-dimensional feature space using the Gaussian kernel function, where the Gaussian kernel function K(rx,ry) formula is:

[0029]

[0030] In the formula, rx,ry is the combination of elements in the resident's age, gender, dietary preference feature vector, household goods consumption feature value, and household appliance cleaning and maintenance frequency feature vector, and σ is a hyperparameter;

[0031] For resident i, its characteristics after kernel mapping are expressed as:

[0032]

[0033] Furthermore, the process of constructing the resident demand feature vector based on the extracted features by the resident demand analysis module includes:

[0034] Use a deep neural network structure to construct a resident demand feature vector: use the kernel-mapped features as the input layer, and perform feature extraction and transformation through multiple hidden layers; set n 1 ,n 2 ,…,n h There are neurons in the hidden layer, where n 1 ,n 2 ,…,n h is the number of neurons in the first hidden layer, the number of neurons in the second hidden layer, ..., the number of neurons in the hth hidden layer, where h is the neuron index of the hidden layer; set the output of the last hidden layer for:

[0035]

[0036] Then the demand characteristic vector of resident i is

[0037]

[0038] Furthermore, the process of collating the service provider information by the provider service resource module includes:

[0039] Get the information of each service provider j;

[0040] For the service type, establish a list of all possible service types, determine the list of pm possible service types, and for each service provider, if service provider j provides the lth service type pt j(l) , then a vector of service types is formed

[0041]

[0042] Among them, pm represents the number of service types, and l represents the service type index;

[0043] For the service scope, in order to determine the service scope ps of service provider j j, expressed in terms of geographic area; based on the radius of service coverage, the coverage radius of the service provider is clarified, and the service range is approximately regarded as a circle with the community as the center; the coverage area of ​​the service provider is calculated using the circle area formula; if the maximum serviceable distance is used, the value of the distance is directly recorded;

[0044] For service price pp j , which records the average price of the main services provided by service provider j.

[0045] Furthermore, the process of constructing a service resource quantization vector and calculating a service resource quantization evaluation value by the provider service resource module includes:

[0046] Combine the service type vector, service scope, and service price into a service resource quantification vector Right now

[0047]

[0048] Calculate the quantitative evaluation value of service resources pe j :

[0049]

[0050] Furthermore, the process of the matching degree acquisition and correction module calculating the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector includes:

[0051] By calculating the characteristic vector of residents' demand and service provider resource quantization vector The cosine similarity between them is used to obtain the basic matching degree MD between resident i and service provider j ij :

[0052]

[0053] in,

[0054]

[0055] Furthermore, the matching degree acquisition and correction module acquires the location information of the residents and the service provider, and corrects the basic matching degree in combination with the service scope of the service provider. The process of completing the preliminary screening includes:

[0056] The location information of residents is expressed in coordinate form RL i =(x i ,y i ), and the location coordinates of service provider j are RL j =(x j ,y j ), and combined with the service provider's service scope psj , set the distance influence factor λ;

[0057] Calculate the distance LD between resident i and service provider j ij :

[0058]

[0059] If the distance between residents and service providers is within the service range, that is, |RL i -RL j |≤ps j , then the basic matching degree remains unchanged, MD' ij =MD ij ;

[0060] If the distance exceeds the service range, |RL i -RL j |>ps j , then the basic matching degree is corrected according to the correction formula based on the exceeded distance:

[0061] Residents and service providers are ranked according to the corrected basic matching scores, and the ones with the highest basic matching scores are selected as the final matching pairs.

[0062] Furthermore, the process of establishing a dynamic monitoring mechanism based on event-driven and time window-based dynamic matching monitoring module for residents' needs includes:

[0063] Real-time data is collected using smart sensor networks and IoT devices, combined with interactive information from the community life service platform. If a resident generates a driving behavior that triggers a monitoring event, the resident demand capture process is immediately initiated.

[0064] Set a time window and adjust the dynamic monitoring interval according to the activity of residents and the frequency of change of demand in different time periods: mark the activity of residents at time τ as A(τ), and the frequency of change of residents' demand at time τ as F(τ); define the monitoring interval function I(τ), if the activity of residents A(τ) ≥ α 1 And the frequency of change of residents' demand F(τ)≥β 1 , it is the peak time in the morning and evening, the monitoring frequency is increased, otherwise it is relaxed; where α 1 , β 1 They are respectively the pre-set thresholds for residents’ activity level and the frequency of changes in residents’ needs.

[0065] Furthermore, the resident demand dynamic matching monitoring module collects multimodal information and combines it with the resident demand feature vector using data fusion technology to generate a new resident demand feature vector. The process includes:

[0066] Through natural language processing technology, the interactive information of the community life service platform is analyzed to obtain multimodal information;

[0067] The collected residents’ personalized demand information and multimodal information are weighted and integrated to obtain a new resident demand feature vector: Obtaining the resident demand feature vector Assume the multimodal vector is Define the weight vector w respectively rx and w az ; Weighted fusion of residents’ demand feature vector and multimodal vector is used to obtain new residents’ demand feature vector

[0068]

[0069] In the formula, It represents the multiplication of corresponding elements, that is, the vector is multiplied by the corresponding weight; |||| represents the norm of the vector.

[0070] Furthermore, the resident demand dynamic matching monitoring module obtains the adaptive monitoring factor through processing and analysis, and calculates the deep matching degree between the resident and the service provider based on the adaptive monitoring factor, including:

[0071] Set different seasonal factors according to the season: Set the season to ms, and for refrigeration-related services, define the seasonal factor function S cool (ms) is:

[0072]

[0073] In the formula, γ 1 , γ 2 is a seasonally defined constant, and γ 1 >γ 2 , if S cool (ms) = γ 1 , it is determined to be summer and refrigeration related services are implemented;

[0074] For heating related services, define the seasonal factor function S heat (ms) is:

[0075]

[0076] In the formula, γ 3 , γ 4 is a seasonally defined constant, and γ 3 >γ 4 , if S heat (ms) = γ 3 , it is determined to be winter, and heating-related services are implemented;

[0077] Incorporate real-time contextual factors, including weather, location, time, environmental data, and social information. Environmental data is collected through environmental monitoring devices in the community, including air quality and noise levels. Social information includes community event announcements and surrounding traffic conditions.

[0078] The real-time situation factors are processed to obtain the real-time situation factors: the real-time situation factors are set to include weather lw, location ll, time lt, environmental data le=(le air ,le noise ) and social information ls=(ls event ,ls traffic ), where le air ,le noise are air quality and noise level, respectively, ls event ,ls traffic They are community activity announcements and surrounding traffic conditions respectively; the real-time situational factors are processed by function Φ to obtain the real-time situational factor LC:

[0079] LC=Φ(lw,ll,lt,le,ls);

[0080] The seasonal factor S(ms) = S cool / heat (ms) is organically integrated with the real-time situation factor LC to obtain the adaptive monitoring factor AMF:

[0081] amf=ψS(ms)+(1-ψ)LC,

[0082] Where ψ is the weight coefficient used to balance the seasonal factor and the real-time situational factor, and ψ∈[0,1];

[0083] Get the new demand characteristic vector of resident i and service provider j resource quantization vector

[0084] Calculate the cosine similarity between the resident demand feature vector and the service provider resource quantization vector as the deep matching degree MB ij , and update the service resource quantitative evaluation value pe j ; Among them, the depth matching degree MB ij The calculation formula is:

[0085]

[0086] Adaptive monitoring factor amf and depth matching degree MB ij There is a linear adjustment relationship between them:

[0087]

[0088] In the formula, MB ij' is the adjusted depth matching degree; η 1 ,η 2 ,η 3 ,η 4 is an adjustable parameter; δ is a positive number used to prevent the denominator from being zero.

[0089] The intelligent community life service recommendation method based on big data analysis includes:

[0090] Step 1: Obtain residents' personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and perform feature extraction, and construct residents' demand feature vectors based on the extracted features;

[0091] Step 2: Organize the service provider information including service type, service scope and service price, construct the service resource quantification vector, and calculate the service resource quantification evaluation value;

[0092] Step 3: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening;

[0093] Step 4: Establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

[0094] Compared with the existing technology, the advantages of the intelligent community life service recommendation system and method based on big data analysis provided by the present invention are:

[0095] 1. The present invention obtains residents' personalized demand information and performs feature extraction, and constructs residents' demand feature vectors based on the extracted features, which provides a data basis for the subsequent matching process, making the recommendation more accurate and personalized; by collating service provider information, constructing service resource quantification vectors, and calculating service resource quantification evaluation values, the capabilities of service providers can be objectively evaluated, providing a basis for the matching process;

[0096] 2. The present invention calculates the basic matching degree according to the resident demand feature vector and the service provider resource quantification vector, obtains the location information of the resident and the service provider, and corrects the basic matching degree in combination with the service scope of the service provider to complete the preliminary screening, thereby realizing the preliminary screening and matching of both the supply and demand sides and ensuring the quality and efficiency of the recommended services;

[0097] 3. The present invention establishes a dynamic monitoring mechanism based on the combination of event-driven and time windows; collects multimodal information, uses data fusion technology to combine it with the resident demand feature vector, and generates a new resident demand feature vector; obtains adaptive monitoring factors through processing and analysis, and calculates the deep matching degree between residents and service providers based on the adaptive monitoring factors, which can capture the changes in residents' needs in real time, improve the timeliness and response speed of matching, and realize more intelligent recommendation services. BRIEF DESCRIPTION OF THE DRAWINGS

[0098] Figure 1 This is a module diagram of the intelligent community life service recommendation system based on big data analysis proposed in the present invention.

[0099] Figure 2 This is a flow chart of the smart community life service recommendation method based on big data analysis proposed in the present invention. DETAILED DESCRIPTION

[0100] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0101] Reference Figure 1 , an intelligent community life service recommendation system based on big data analysis, which includes a resident demand analysis module, a provider service resource module, a matching degree acquisition and correction module, and a resident demand dynamic matching monitoring module;

[0102] Resident demand analysis module: obtains residents' personalized demand information, including records of residents' information (age and gender), dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and performs feature extraction, and constructs residents' demand feature vectors based on the extracted features;

[0103] Provider service resource module: organizes service provider information including service type, service scope and service price, constructs service resource quantitative vector, and calculates service resource quantitative evaluation value;

[0104] Matching degree acquisition and correction module: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening;

[0105] Resident demand dynamic matching monitoring module: establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

[0106] The resident demand analysis module obtains the personalized demand information of residents and performs feature extraction. The steps of constructing the resident demand feature vector based on the extracted features include:

[0107] Step 101: Collect and pre-process the personalized demand information of each resident, including resident information records (age and gender), dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records;

[0108] Step 102: For the resident information record, obtain the actual age value of resident i, and divide the resident age into different intervals based on the different impact stages of age, and mark them as ra i =(1,2,3,4); Get the resident's gender information rg i =(0,1), with males being 1 and females being 0; the different age ranges of residents include adolescents 0-18 years old, young people 19-35 years old, middle-aged people 36-55 years old, and elderly people over 56 years old, which are coded with numbers 1, 2, 3, and 4 respectively;

[0109] Step 103: For the dietary preference record, establish a 1 List of dietary combinations, where m 1 is the number of dietary combinations, and dietary combinations are classified based on nutritional components, regional characteristics, and taste types, such as high-protein combinations, vegetarian combinations, spicy flavor combinations, etc. For resident i, the fuzzy logic method is used to quantify dietary preferences, as follows:

[0110] Step U1: Define μ i(ι) is the membership of resident i to the ιth diet combination, where ι is the index of the diet combination type, and ι=1,2,…,m 1 , the value range of ι is [0,1];

[0111] Step U2: Integrate to obtain the dietary preference feature vector of resident i:

[0112]

[0113] Where, υ is the dimension of dietary preference feature vector;

[0114] In step 103, the membership of the diet combination is obtained through questionnaire analysis. The membership reflects the residents' preference for a certain diet combination. For example, if the residents often buy ingredients related to a certain diet combination, then μ iι Close to 1, if residents rarely buy ingredients related to a certain diet combination, then μ iι Close to 0;

[0115] Step 104: For household goods consumption records, determine m 2 Types of household items, including m 2 is the number of household items; collects the amount of consumption of household items of resident i on the kth household item in multiple time periods t (such as each month in the past year) Where k is the index of household products. The future consumption trend is predicted by time series analysis method as follows:

[0116] Step V1: Consumption amount Considered as a time series, the prediction equation is obtained using the autoregressive moving average model:

[0117]

[0118] In the formula, rc′ i(k) is the predicted consumption amount of household goods by resident i for the kth household product, p, q are the orders of the autoregressive moving average model, J is the index of the order of the autoregressive moving average model, θ J are the autoregressive moving average model parameters, is a white noise sequence;

[0119] Step V2: Calculate the characteristic value rh of household goods consumption by resident i i :

[0120]

[0121] In the formula, w k is the weight of the kth household item, which can be determined according to its importance to life (e.g., daily necessities have a high weight). is the average consumption amount of all residents on the kth household product;

[0122] Step 105: For the cleaning and maintenance records of home appliances, a preset period (such as one year) is set, and the cleaning and maintenance status of home appliances of resident i in the period is checked. If there is a cleaning and maintenance record, it is recorded as 1, otherwise it is recorded as 0; at the same time, the number of types of home appliances of resident i is counted, and the type distribution of repaired home appliances is analyzed. For example, home appliances are divided into kitchen appliances, cleaning appliances, entertainment appliances, etc., and the cleaning and maintenance frequency feature vector of home appliances is obtained.

[0123]

[0124] In the formula, represents the proportion of the number of repairs of the first type of household appliances of resident i, the proportion of the number of repairs of the second type of household appliances, ..., The proportion of repair times of each type of household appliances, is the dimension of the feature vector of the frequency of cleaning and maintenance of household appliances;

[0125] Step 106: Extract features based on the pre-processed personalized demand information of residents: Use the Gaussian kernel function to map low-dimensional resident features to a high-dimensional feature space to mine potential feature relationships: The formula of the Gaussian kernel function K(rx,ry) is:

[0126]

[0127] In the formula, rx,ry is the combination of elements in the resident's age, gender, dietary preference feature vector, household goods consumption feature value, and household appliance cleaning and maintenance frequency feature vector, and σ is a hyperparameter that determines the width or smoothness of the Gaussian kernel function;

[0128] Step 107: For resident i, its feature after kernel mapping is expressed as:

[0129]

[0130] Step 108: Use a deep neural network structure to construct a resident demand feature vector: use the kernel-mapped features as the input layer, and perform feature extraction and transformation through multiple hidden layers; set n 1 ,n 2 ,…,n h There are neurons in the hidden layer, where n 1 ,n 2 ,…,n h is the number of neurons in the first hidden layer, the number of neurons in the second hidden layer, ..., the number of neurons in the hth hidden layer, and h is the neuron index of the hidden layer. It can be understood that the number of neurons in each hidden layer is adjusted according to the data scale and complexity. By adjusting the number of neurons in each hidden layer, the performance of the neural network is optimized to better extract and represent the characteristics of residents' needs; set the output of the last hidden layer for:

[0131]

[0132] Then the demand characteristic vector of resident i is

[0133]

[0134] In step 108, the resident demand feature vector It is obtained after complex data analysis and feature extraction, and can reflect the personalized demand information of residents more comprehensively and deeply.

[0135] The steps of the provider service resource module to organize the service provider information including service type, service scope and service price, construct the service resource quantification vector, and calculate the service resource quantification evaluation value include:

[0136] Step 201: Obtain information of each service provider j;

[0137] Step 202: For the service type, a list of all possible service types is established to determine a total of pm possible service types, including catering, housekeeping, home appliance repair, etc. For each service provider, if service provider j provides the lth service type pt j(l) , then a vector of service types is formed

[0138]

[0139] Among them, pm represents the number of service types, and l represents the service type index;

[0140] Step 203: For the service scope, in order to determine the service scope ps of service provider j j , expressed in terms of geographic area; based on the radius calculation of service coverage, the coverage radius of the service provider is clarified, and the service range is approximately regarded as a circle with the community as the center; the coverage area of ​​the service provider is calculated using the circle area formula; if the maximum serviceable distance is used, the value of the distance is directly recorded; for example, if the service coverage radius of service provider j is determined to be 5 kilometers, ps is obtained according to the circle area formula j =π×5 2 =25π square kilometers;

[0141] Step 204: For the service price pp j , that is, recording the average price of the main services provided by service provider j;

[0142] Step 205: Combine the service type vector, service scope and service price into a service resource quantization vector Right now

[0143]

[0144] Step 206: Calculate the service resource quantitative evaluation value pe j :

[0145]

[0146] For example, a service provider's service scope ps j =100 square kilometers, service price pp j =50 yuan, providing 10 types of services but It can be understood that the resource value of the service provider is comprehensively evaluated by multiplying the service scope of the unit price by the number of service types provided. The higher the value, the higher the resource value of the service provider.

[0147] The matching degree acquisition and correction module calculates the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtains the location information of the residents and the service provider, and corrects the basic matching degree based on the service scope of the service provider. The steps to complete the preliminary screening include:

[0148] Step 301: Calculate the resident demand feature vector and service provider resource quantization vector The cosine similarity between them is used to obtain the basic matching degree MD between resident i and service provider j ij :

[0149]

[0150] in,

[0151]

[0152] In step 301, the sum of the products of the corresponding elements of the two vectors is first calculated. Then calculate the magnitude of the two vectors separately Finally, the basic matching degree MD is obtained by dividing the sum of the products of the corresponding elements by the product of the two vector moduli. ij ,This match reflects the similarity between residents’ needs and service ,service providers’ resources;

[0153] Step 302: Use coordinates to represent the location information RL of the resident i =(x i ,y i ), and the location coordinates of service provider j are RL j =(x j ,y j ), and combined with the service provider's service scope ps j , set the distance influence factor λ (for example, 2);

[0154] Step 303: Calculate the distance LD between resident i and service provider j ij :

[0155]

[0156] Step 304: If the distance between the resident and the service provider is within the service range, that is, |RL i -RL j |≤ps j , then the basic matching degree remains unchanged, MD' ij =MD ij , if the distance exceeds the service range, that is, |RL i -RL j |>ps j , then the basic matching degree is corrected according to the correction formula based on the exceeded distance: The greater the distance exceeds, the more obvious the decrease in matching degree is;

[0157] For example, if the distance between the resident and the service provider is 10 km beyond the service range and the service range is 20 km, MD ij =0.6, then

[0158] Step 305: Sort the residents and service providers according to the corrected basic matching degree, and select the part with the highest basic matching degree (for example, the first 30%) as the final matching pairs.

[0159] The resident demand dynamic matching monitoring module establishes a dynamic monitoring mechanism based on event-driven and time window combination; collects multimodal information, uses data fusion technology to combine it with the resident demand feature vector, and generates a new resident demand feature vector; obtains the adaptive monitoring factor through processing and analysis, and calculates the deep matching degree between residents and service providers based on the adaptive monitoring factor. The steps include:

[0160] Step 401: Use smart sensor networks and IoT devices to collect real-time data, and combine the interactive information of the community life service platform. If a resident generates a driving behavior (such as abnormal operation of smart home devices, online service consultation, etc.) to trigger a monitoring event, the resident demand capture process is immediately started;

[0161] In step 401, the interactive information of the community life service platform includes the residents' speeches in community forums and social media groups, and explores potential interests, life problems, and expectations and demands for life services;

[0162] Step 402: Set a time window and adjust the dynamic monitoring interval according to the activity of residents and the frequency of change of demand in different time periods: mark the activity of residents at time τ as A(τ), and the frequency of change of residents' demand at time τ as F(τ); define the monitoring interval function I(τ), if the activity of residents A(τ) ≥ α 1 And the frequency of change of residents' demand F(τ)≥β 1, it is the peak time in the morning and evening, the monitoring frequency is increased, otherwise it is relaxed; where α 1 , β 1 They are the pre-set thresholds for residents’ activity activity and the frequency of changes in residents’ needs, respectively;

[0163] In step 402, real-time monitoring is performed to capture the dynamic changes in residents' needs, and the dynamic monitoring interval is adjusted to increase the monitoring frequency to once every 15 minutes during the morning and evening peak hours, and appropriately relax it to once every 45 minutes during other periods, to ensure that subtle changes in residents' needs can be captured in a timely and accurate manner;

[0164] Step 403: Analyze the interactive information (including text, pictures and videos) of the community life service platform through natural language processing technology to obtain multimodal information;

[0165] Step 404: weighted fusion of the collected resident personalized demand information and multimodal information to obtain a new resident demand feature vector: Obtaining resident demand feature vector Assume the multimodal vector is Define the weight vector w respectively rx and w az ; Weighted fusion of residents’ demand feature vector and multimodal vector is used to obtain new residents’ demand feature vector

[0166]

[0167] In the formula, Indicates the corresponding multiplication of elements, that is, the vector is multiplied by the corresponding weight; |||| indicates the norm of the vector, which is used to normalize the fused vector;

[0168] In step 404, the multimodal vector is obtained in the following manner: using the multimodal fusion technology in deep learning to organically fuse different types of residents' interactive information (text, numerical value, image, etc.); for text data, such as messages left by residents on community forums, using word vector models and recurrent neural networks (RNN) to extract semantic features; for numerical data, such as consumption records, health data, etc., after normalization processing, input into a multi-layer perceptron (MLP) for feature extraction; for image data (if there is a relevant scene image captured by a smart camera), using a convolutional neural network (CNN) to extract visual features; these feature vectors from different modalities are spliced ​​or processed using an attention mechanism to construct a multimodal vector;

[0169] Step 405: Set different seasonal factors according to the season: Set the season to ms, and for refrigeration related services, define the seasonal factor function S cool (ms) is:

[0170]

[0171] In the formula, γ 1 , γ 2 is a seasonally defined constant, and γ 1 >γ 2 , if S cool (ms) = γ 1 , it is determined to be summer and refrigeration related services are implemented;

[0172] For heating related services, define the seasonal factor function S heat (ms) is:

[0173]

[0174] In the formula, γ 3 , γ 4 is a seasonally defined constant, and γ 3 >γ 4 , if S heat (ms) = γ 3 , it is determined to be winter, and heating-related services are promoted;

[0175] In step 405, the seasonal factor is increased for cooling-related services in summer and heating-related services in winter, while the seasonal factor is decreased for some services with opposite seasonality;

[0176] Step 406: Incorporating real-time contextual factors, specifically: weather, location, time, environmental data, and social information. Environmental data is collected through environmental monitoring equipment in the community, including air quality and noise level. Social information includes community event announcements and surrounding traffic conditions.

[0177] Step 407: Process the real-time situation factors to obtain the real-time situation factors: Set the real-time situation factors to include weather lw, location ll, time lt, and environmental data le=(le air ,le noise ) and social information ls=(ls event ,ls traffic ), where le air ,le noise are air quality and noise level, respectively, ls event ,ls traffic They are community activity announcements and surrounding traffic conditions respectively; the real-time situational factors are processed by function Φ to obtain the real-time situational factor LC:

[0178] LC=Φ(lw,ll,lt,le,ls);

[0179] In step 407, when the air quality is poor, life services related to air purification are recommended in a timely manner; if the community holds a large-scale event, convenient travel services or special life services around the event are recommended to residents based on the location and time of the event. At the same time, through real-time analysis of residents' behavior data, the current life scene of the residents is inferred, such as whether they are in a busy weekday, a leisurely weekend holiday, etc., so that the recommendation is more in line with the actual needs of the residents at the moment;

[0180] Step 408: Set the seasonal factor S(ms)=S cool / heat (ms) is organically integrated with the real-time situation factor LC to obtain the adaptive monitoring factor AMF:

[0181] amf=ψS(ms)+(1-ψ)LC,

[0182] Where ψ is the weight used to balance the seasonal factor and the real-time situational factor, and ψ∈[0,1];

[0183] Step 409: Obtain new resident i demand feature vector and service provider j resource quantization vector

[0184] Step 410: Calculate the cosine similarity between the resident demand feature vector and the service provider resource quantization vector as the depth matching degree MB ij , and update the service resource quantitative evaluation value pe j ; Among them, the depth matching degree MB ij The calculation formula is:

[0185]

[0186] Step 411: Adaptive monitoring factor amf and depth matching degree MB ij There is a linear adjustment relationship between them:

[0187]

[0188] In the formula, MB ij ' is the adjusted depth matching degree; η 1 ,η 2 ,η 3 ,η 4 It is an adjustable parameter that is calibrated and optimized through a large amount of historical data and actual business scenarios to achieve the best matching calculation effect. These parameters can control the relative importance and influence of different factors in the final matching calculation. δ is a very small positive number used to prevent the denominator from being zero and ensure the numerical stability of the formula.

[0189] In step 411, the adjustment formula is to transform the depth matching degree MB by logarithmic transformation.ij The exponential function combines various factors and maps the results to the interval [0, 1] through the sigmoid function, making the matching degree intuitively interpretable. The higher the value, the better the matching degree between resident i and service provider j.

[0190] Reference Figure 2 , a smart community life service recommendation method based on big data analysis, including:

[0191] Step 1: Obtain residents' personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and perform feature extraction, and construct residents' demand feature vectors based on the extracted features;

[0192] Step 2: Organize the service provider information including service type, service scope and service price, construct the service resource quantification vector, and calculate the service resource quantification evaluation value;

[0193] Step 3: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening;

[0194] Step 4: Establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

[0195] In addition, the formulas involved in the above are all calculated by removing dimensions and taking their numerical values. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The proportional coefficient in the formula and the various preset thresholds in the analysis process are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data; the size of the proportional coefficient is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the proportional coefficient depends on the amount of sample data and the preliminary setting of the corresponding processing coefficient for each group of sample data by technical personnel in this field; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0196] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically based on the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0197] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0198] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0199] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0200] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0201] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0202] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0203] Finally: The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. An intelligent community life service recommendation system based on big data analysis, characterized by: It includes a resident demand analysis module, a provider service resource module, a matching degree acquisition and correction module, and a resident demand dynamic matching monitoring module; Resident demand analysis module: obtains residents' personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and extracts features, and constructs residents' demand feature vectors based on the extracted features; Provider service resource module: organizes service provider information including service type, service scope and service price, constructs service resource quantitative vector, and calculates service resource quantitative evaluation value; Matching degree acquisition and correction module: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening; Resident demand dynamic matching monitoring module: establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

2. The intelligent community life service recommendation system based on big data analysis according to claim 1 is characterized by: The process of obtaining residents' personalized demand information and extracting features by the resident demand analysis module includes: Collect and pre-process each resident's personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records; The resident information record includes age and gender. For the resident information record, the actual age value of resident i is obtained, and the resident age is divided into different intervals and marked as ra. i =(1,2,3,4); Get the resident's gender information rg i =(0,1), with males as 1 and females as 0; the different age ranges of residents include adolescents, young adults, middle-aged people, and the elderly, and are coded with numbers as 1, 2, 3, and 4 respectively; For dietary preference records, a list of m1 dietary combinations is established, where m1 is the number of dietary combinations. For resident i, the dietary preference is quantified using the fuzzy logic method, as follows: Step U1: Define μ i (ι) is the membership of resident i to the ιth diet combination, where ι is the index of the diet combination type, and ι=1,2,…,m1, and the value range of ι is [0,1]; Step U2: Integrate to obtain the dietary preference feature vector of resident i: Where, υ is the dimension of the dietary preference feature vector; For household goods consumption records, determine m2 types of household goods, where m2 is the number of household goods types; collect the consumption amount of resident i for the kth household goods in multiple time periods t Where k is the index of household products. The future consumption trend is predicted by time series analysis method as follows: Step V1: Consumption amount Considered as a time series, the prediction equation is obtained using the autoregressive moving average model: In the formula, rc′ i(k) is the predicted consumption amount of household goods by resident i for the kth household product, p, q are the orders of the autoregressive moving average model, J is the index of the order of the autoregressive moving average model, θ J are the autoregressive moving average model parameters, is a white noise sequence; Step V2: Calculate the characteristic value rh of household goods consumption by resident i i : In the formula, w k is the weight of the kth household item, is the average consumption amount of all residents on the kth household product; For the cleaning and maintenance records of home appliances, a preset period is set to check the cleaning and maintenance status of home appliances of resident i in the period. If there is a cleaning and maintenance record, it is recorded as 1, otherwise it is recorded as 0. At the same time, the number of types of home appliances of resident i is counted, and the type distribution of repaired home appliances is analyzed to obtain the characteristic vector of home appliance cleaning and maintenance frequency In the formula, represents the proportion of the number of repairs of the first type of household appliances of resident i, the proportion of the number of repairs of the second type of household appliances, ..., The proportion of repair times of each type of household appliances, is the dimension of the feature vector of the frequency of cleaning and maintenance of household appliances; Feature extraction is performed based on the preprocessed personalized demand information of residents: the low-dimensional resident features are mapped to the high-dimensional feature space using the Gaussian kernel function, where the Gaussian kernel function K(rx,ry) formula is: In the formula, rx,ry is the combination of elements in the resident's age, gender, dietary preference feature vector, household goods consumption feature value, and household appliance cleaning and maintenance frequency feature vector, and σ is a hyperparameter; For resident i, its characteristics after kernel mapping are expressed as:

3. The intelligent community life service recommendation system based on big data analysis according to claim 2 is characterized by: The process of constructing the residents' demand feature vector based on the extracted features by the residents' demand analysis module includes: Use deep neural network structure to construct the resident demand feature vector: take the kernel mapped features as the input layer, extract and transform the features through multiple hidden layers; set n1, n2, …, n h There are neurons in the hidden layer, where n1, n2, ..., n h is the number of neurons in the first hidden layer, the number of neurons in the second hidden layer, ..., the number of neurons in the hth hidden layer, where h is the neuron index of the hidden layer; set the output of the last hidden layer for: Then the demand characteristic vector of resident i is 4. The intelligent community life service recommendation system based on big data analysis according to claim 1 is characterized by: The process of the provider service resource module collating service provider information, constructing a service resource quantification vector, and calculating the service resource quantification evaluation value includes: Get the information of each service provider j; For the service type, establish a list of all possible service types, determine the list of pm possible service types, and for each service provider, if service provider j provides the lth service type pt j(l) , then a vector of service types is formed Among them, pm represents the number of service types, and l represents the service type index; For the service scope, in order to determine the service scope ps of service provider j j , expressed in terms of geographic area; based on the radius of service coverage, the coverage radius of the service provider is clarified, and the service range is approximately regarded as a circle with the community as the center; the coverage area of ​​the service provider is calculated using the circle area formula; if the maximum serviceable distance is used, the value of the distance is directly recorded; For service price pp j , that is, recording the average price of the main services provided by service provider j; Combine the service type vector, service scope, and service price into a service resource quantification vector Right now Calculate the quantitative evaluation value of service resources pe j :

5. The intelligent community life service recommendation system based on big data analysis according to claim 1 is characterized by: The process of calculating the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector by the matching degree acquisition and correction module includes: By calculating the resident demand characteristic vector and service provider resource quantization vector The cosine similarity between them is used to obtain the basic matching degree MD between resident i and service provider j ij : in, 6. The intelligent community life service recommendation system based on big data analysis according to claim 5 is characterized by: The matching degree acquisition and correction module obtains the location information of residents and service providers, and corrects the basic matching degree based on the service scope of the service provider. The process of completing the preliminary screening includes: The location information of residents is expressed in coordinate form RL i =(x i ,y i ), and the location coordinates of service provider j are RL j =(x j ,y j ), and combined with the service provider's service scope ps j , set the distance influence factor λ; Calculate the distance LD between resident i and service provider j ij : If the distance between residents and service providers is within the service range, that is, |RL i -RL j |≤ps j , then the basic matching degree remains unchanged, MD' ij =MD ij ; If the distance exceeds the service range, |RL i -RL j |>ps j , then the basic matching degree is corrected according to the correction formula based on the exceeded distance: Residents and service providers are ranked according to the corrected basic matching scores, and the ones with the highest basic matching scores are selected as the final matching pairs.

7. The intelligent community life service recommendation system based on big data analysis according to claim 1 is characterized by: The process of establishing a dynamic monitoring mechanism based on event-driven and time window in the resident demand dynamic matching monitoring module includes: Real-time data is collected using smart sensor networks and IoT devices, combined with interactive information from the community life service platform. If a resident generates a driving behavior that triggers a monitoring event, the resident demand capture process is immediately initiated. Set a time window, and adjust the dynamic monitoring interval according to the activity level of residents and the frequency of changes in demand in different time periods: mark the activity level of residents' activities at time τ as A(τ), and mark the frequency of changes in residents' demand at time τ as F(τ); define the monitoring interval function I(τ), if the activity level of residents' activities A(τ)≥α1 and the frequency of changes in residents' demand F(τ)≥β1, then it is the peak period in the morning and evening, and the monitoring frequency should be increased, otherwise it should be relaxed; where α1 and β1 are the pre-set thresholds for the activity level of residents' activities and the frequency of changes in residents' demand, respectively.

8. The intelligent community life service recommendation system based on big data analysis according to claim 3 is characterized by: The resident demand dynamic matching monitoring module collects multimodal information and combines it with the resident demand feature vector using data fusion technology to generate a new resident demand feature vector. The process includes: Through natural language processing technology, the interactive information of the community life service platform is analyzed to obtain multimodal information; The collected residents’ personalized demand information and multimodal information are weighted and integrated to obtain a new resident demand feature vector: Obtaining the resident demand feature vector Assume the multimodal vector is Define the weight vector w respectively rx and w az ; Weighted fusion of residents’ demand feature vector and multimodal vector is used to obtain new residents’ demand feature vector In the formula, It represents the multiplication of corresponding elements, that is, the vector is multiplied by the corresponding weight; || || represents the norm of the vector.

9. The intelligent community life service recommendation system based on big data analysis according to claim 1 is characterized by: The dynamic matching monitoring module of residents' needs obtains the adaptive monitoring factor through processing and analysis, and calculates the deep matching degree between residents and service providers based on the adaptive monitoring factor, including: Set different seasonal factors according to the season: Set the season to ms, and for refrigeration-related services, define the seasonal factor function S cool (ms) is: Where γ1 and γ2 are seasonal constants, and γ1>γ2. If S cool (ms) = γ1, it is judged to be summer, and refrigeration-related services are promoted; For heating related services, define the seasonal factor function S heat (ms) is: Where γ3 and γ4 are seasonal constants, and γ3>γ4. If S heat (ms) = γ3, it is determined to be winter, and heating-related services are promoted; Incorporate real-time contextual factors, including weather, location, time, environmental data, and social information. Environmental data is collected through environmental monitoring devices in the community, including air quality and noise levels. Social information includes community event announcements and surrounding traffic conditions. The real-time situation factors are processed to obtain the real-time situation factors: the real-time situation factors are set to include weather lw, location ll, time lt, environmental data le=(le air ,le noise ) and social information ls=(ls event ,ls traffic ), where le air ,le noise are air quality and noise level, ls event ,ls traffic They are community activity announcements and surrounding traffic conditions respectively; the real-time situational factors are processed by function Φ to obtain the real-time situational factor LC: LC=Φ(lw,ll,lt,le,ls); The seasonal factor S(ms) = S cool / heat (ms) is organically integrated with the real-time situation factor LC to obtain the adaptive monitoring factor AMF: amf=ψS(ms)+(1-ψ)LC, Where ψ is the weight coefficient used to balance the seasonal factor and the real-time situational factor, and ψ∈[0,1]; Get the new demand characteristic vector of resident i and service provider j resource quantization vector Calculate the cosine similarity between the resident demand feature vector and the service provider resource quantization vector as the deep matching degree MB ij , and update the service resource quantitative evaluation value pe j ; Among them, the depth matching degree MB ij The calculation formula is: Adaptive monitoring factor amf and depth matching degree MB ij There is a linear adjustment relationship between them: In the formula, MB ij ' is the adjusted depth matching degree; η1, η2, η3, η4 are adjustable parameters; δ is a positive number used to prevent the denominator from being zero.

10. A smart community life service recommendation method based on big data analysis, characterized in that: Applied to the smart community life service recommendation system based on big data analysis as described in any one of claims 1 to 9, the method includes: Step 1: Obtain residents' personalized demand information, including resident information records, dietary preference records, household goods consumption records, and home appliance cleaning and maintenance records, and perform feature extraction, and construct residents' demand feature vectors based on the extracted features; Step 2: Organize the service provider information including service type, service scope and service price, construct the service resource quantification vector, and calculate the service resource quantification evaluation value; Step 3: Calculate the basic matching degree based on the resident demand feature vector and the service provider resource quantification vector, obtain the location information of the residents and service providers, and correct the basic matching degree based on the service scope of the service provider to complete the preliminary screening; Step 4: Establish a dynamic monitoring mechanism based on event-driven and time window combination; collect multimodal information, use data fusion technology to combine it with the resident demand feature vector, and generate a new resident demand feature vector; obtain adaptive monitoring factors through processing and analysis, and calculate the deep matching degree between residents and service providers based on the adaptive monitoring factors.

Images