A dining recommendation method and system based on user portrait

Abstract

The present invention discloses a dining recommendation method and system based on user portraits, including: obtaining user historical order information, performing attribute analysis based on the user historical order information, and obtaining attribute analysis result information; performing user preference analysis based on the user historical order information and attribute analysis result information, and constructing a user preference portrait; obtaining user historical browsing information, performing user interest analysis based on the user historical browsing information, and constructing a user interest portrait; predicting the next point of interest for the target user based on the user interest portrait and the user preference portrait, and obtaining point of interest prediction information; obtaining real-time weather information, and performing dining recommendation in combination with the user preference portrait, the user interest portrait, and the point of interest prediction information. Through the method based on user portraits, the diversified needs of users are fully and accurately understood and reflected, and the quality and personalization level of dining recommendation services are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of restaurant recommendation, and in particular to a restaurant recommendation method and system based on user portrait. Background Art

[0002] With the booming development of social media and mobile applications, people's expectations for dining choices have gradually increased. However, traditional dining recommendation systems mainly rely on users' past behavior data, which has certain limitations. Users' tastes, interests, and social backgrounds are multi-dimensional and dynamically changing, and traditional methods are difficult to fully grasp these personalized characteristics.

[0003] The current restaurant recommendation systems on the market often face the problems of information overload and lack of personalization. Users may feel troubled by too many choices, and traditional recommendation systems often fail to accurately grasp the diverse needs of users, resulting in recommended restaurant options that do not always meet users' actual expectations.

[0004] Therefore, there is an urgent need for a more intelligent and personalized dining recommendation method to better meet users' diverse taste preferences, social needs and real-time environmental changes. Summary of the invention

[0005] The present invention overcomes the defects of the prior art and provides a dining recommendation method and system based on user portraits, an important purpose of which is to improve the accuracy and personalization of dining recommendations.

[0006] To achieve the above-mentioned purpose, the first aspect of the present invention provides a dining recommendation method based on user portraits, comprising:

[0007] Obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information;

[0008] Performing user preference analysis based on the user's historical order information and attribute analysis result information, and building a user preference profile;

[0009] Obtain user historical browsing information, perform user interest analysis based on the user historical browsing information, and construct a user interest profile;

[0010] Predict the next point of interest for the target user based on the user interest profile and the user preference profile to obtain point of interest prediction information;

[0011] Obtain real-time weather information and make restaurant recommendations based on user preference profiles, user interest profiles, and POI prediction information.

[0012] In this solution, the attribute analysis is performed based on the user's historical order information to obtain attribute analysis result information, specifically:

[0013] Obtaining historical order information of the user, performing feature extraction on the historical orders of the user, extracting the type, frequency, price and time features of the dishes in the historical orders of the user, and obtaining historical order feature information;

[0014] Preset cuisine categories, perform cuisine attribute analysis based on the historical order feature information, calculate the Euclidean distance between each order and the preset cuisine category, and make a judgment with a preset threshold to obtain cuisine attribute analysis information;

[0015] Acquire order feedback information according to the historical order information, build a semantic analysis model based on natural language processing technology, input the order feedback information into the semantic analysis model to perform semantic analysis of the feedback information, and obtain semantic analysis result information;

[0016] Performing a taste attribute pre-analysis in combination with the historical order feature information and the cuisine attribute analysis information, extracting the taste features of the dishes in each order and the taste features of each cuisine, and counting the number of each taste feature of the dishes in the order to obtain pre-analysis result information;

[0017] Performing taste attribute analysis according to the semantic analysis result information and the pre-analysis result information to obtain taste attribute analysis information;

[0018] Performing a user meal time attribute analysis based on the historical order feature information, dividing each order by time, and obtaining time attribute analysis information;

[0019] Perform order price attribute analysis based on the user's historical order feature information, analyze the total price of each order and the unit price of each meal, and count the frequency of each price to obtain price attribute analysis information;

[0020] The attribute analysis result information is obtained by combining the cuisine attribute analysis information, the taste attribute analysis information, the time attribute analysis information and the price attribute analysis information.

[0021] In this solution, the user preference analysis is performed based on the user's historical order information and attribute analysis result information, and a user preference profile is constructed, specifically as follows:

[0022] Obtain user historical order information and attribute analysis result information;

[0023] Obtain the taste category and cuisine category of the target user according to the attribute analysis result information, and perform analysis based on the principal component analysis method combined with the user's historical order information to obtain principal component analysis information;

[0024] Sort each flavor category and each cuisine category according to the principal component information, perform style preference analysis based on the sorting results, and obtain style preference analysis information;

[0025] Obtaining time attribute analysis information according to the attribute analysis result information, and performing time preference analysis based on a statistical algorithm combined with the price attribute analysis information;

[0026] Count the dining frequency of users in each time period as a time preference analysis indicator and conduct time preference analysis to obtain time preference analysis information;

[0027] Obtain price attribute analysis information according to the attribute analysis result information, and draw a total price trend graph and a unit price trend graph according to the price attribute analysis information;

[0028] Conduct price preference analysis on target customers based on total price trend chart and unit price trend chart, analyze total price preference and unit price preference of target customers, and obtain price preference analysis information;

[0029] Combining the style preference analysis information, the time preference analysis information and the price preference analysis information to form user preference analysis information, and performing user preference level evaluation;

[0030] Preset preference level evaluation indicators, analyze user preference analysis information with the preference level evaluation indicators, evaluate user preference style, dining time and price dimensions and assign corresponding preference levels, and obtain preference level evaluation information;

[0031] A user preference profile is constructed by combining the preference level evaluation information and the user preference analysis information.

[0032] In this solution, the user interest analysis is performed based on the user's historical browsing information, and a user interest portrait is constructed, specifically:

[0033] Obtaining historical browsing information of the user, and extracting features of the historical browsing information of the user, including browsed dishes, browsing frequency, and browsing time period, to obtain historical browsing feature information;

[0034] Building an associated interest analysis model according to an association rule mining algorithm, inputting the historical browsing feature information into the associated interest analysis model for analysis, and obtaining associated interest analysis information;

[0035] Setting interest tags according to the associated interest analysis information, wherein the interest tags are divided into time interest tags and food interest tags, and setting corresponding interest weights;

[0036] Analyze the user's interest preference according to the interest tags and interest weights combined with the associated interest analysis information and historical browsing features to obtain preference analysis information;

[0037] Performing user preference feature analysis based on the associated interest analysis information and the preference degree analysis information based on a clustering algorithm, analyzing the preference characteristics and types of target users, and obtaining preference feature analysis information;

[0038] Build user interest profiles based on preference degree analysis information, associated interest points information, and preference feature analysis information.

[0039] In this solution, the next point of interest of the target user is predicted based on the user interest profile and the user preference profile, specifically:

[0040] The preset attribute categories include location attributes and date attributes. The location attributes are divided into residential, leisure and office, and the date attributes are divided into holiday days and working days.

[0041] Obtain the user's historical dining location information and dining time information, perform attribute analysis based on preset attribute categories, and obtain dining attribute analysis information;

[0042] Preset time intervals, build user access sequences including time, location, and food through meal attribute analysis information, analyze the behavior patterns of target users, and obtain behavior pattern analysis information;

[0043] Build a long-term and short-term preference analysis model based on NGCF to obtain user interest profiles and user preference profiles. Combine dining attribute analysis information and behavior pattern analysis information and input them into the long-term and short-term preference sub-model to conduct long-term and short-term preference analysis on the target users to obtain long-term and short-term preference analysis information.

[0044] Constructing a point of interest prediction model, and inputting the long-term and short-term preference analysis information and the behavior rule analysis information as initial information into the point of interest prediction model to predict the next point of interest;

[0045] The attention mechanism is used to calculate the attention score of each point of interest based on the long-term and short-term preference analysis information and the behavior law analysis information. The score is used as the weight of the point of interest and the interest decision vector is formed through weighted fusion.

[0046] Based on the multi-layer perceptron, the interest decision vector and the interest point are transformed into multi-layer nonlinear transformations and the recommendation score of each interest point is calculated to obtain the prediction information of the candidate interest points.

[0047] The candidate points of interest are sorted according to their recommendation scores, and a selection threshold is preset. The final prediction result is selected by selecting the threshold to obtain the point of interest prediction information.

[0048] In this solution, the real-time weather information is obtained, and restaurant recommendations are made in combination with user preference profiles, user interest profiles, and interest point prediction information, specifically:

[0049] Obtain real-time weather information, user preference profiles, user interest profiles, and point of interest prediction information;

[0050] Based on big data retrieval, the preference and interest portraits of various groups of people are obtained to form a comparative data set, and then the categories are divided. The Manhattan distance of various groups of people is calculated and judged by the preset threshold to obtain the group classification information;

[0051] Integrate the preferences and interests of various groups of people, merge similar interests and preferences, combine time characteristics to form feature sequences, and construct feature maps;

[0052] Calculate the similarity between the user preference profile and the user interest profile and the feature map, and make a judgment with a preset threshold, analyze the similar groups of the target user, and obtain similar group information;

[0053] Sequence extraction is performed on the feature map according to the information of similar people, and the feature sequence of similar people is extracted to obtain similar feature sequence information;

[0054] Conduct suitable catering analysis based on real-time meteorological information, analyze suitable catering under current meteorological characteristics, and obtain suitable catering analysis information;

[0055] Construct a restaurant recommendation model to make restaurant recommendations based on similar feature sequence information, suitable restaurant analysis information, user preference profile, user interest profile and interest point prediction information to obtain restaurant recommendation information;

[0056] Make restaurant recommendations to target users based on restaurant recommendation information, including location, time, and type of food.

[0057] A second aspect of the present invention provides a dining recommendation system based on user portraits, the system comprising: a memory, a processor, the memory containing a dining recommendation method program based on user portraits, the dining recommendation method program based on user portraits when executed by the processor implements the following steps:

[0058] Obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information;

[0059] Performing user preference analysis based on the user's historical order information and attribute analysis result information, and building a user preference profile;

[0060] Obtain user historical browsing information, perform user interest analysis based on the user historical browsing information, and construct a user interest profile;

[0061] Predict the next point of interest for the target user based on the user interest profile and the user preference profile to obtain point of interest prediction information;

[0062] Obtain real-time weather information and make restaurant recommendations based on user preference profiles, user interest profiles, and POI prediction information.

[0063] The present invention discloses a dining recommendation method and system based on user portraits, including: obtaining user historical order information, performing attribute analysis based on the user historical order information, and obtaining attribute analysis result information; performing user preference analysis based on the user historical order information and attribute analysis result information, and constructing a user preference portrait; obtaining user historical browsing information, performing user interest analysis based on the user historical browsing information, and constructing a user interest portrait; predicting the next point of interest for the target user based on the user interest portrait and the user preference portrait, and obtaining point of interest prediction information; obtaining real-time weather information, and performing dining recommendation in combination with the user preference portrait, the user interest portrait, and the point of interest prediction information. Through the method based on user portraits, the diversified needs of users are fully and accurately understood and reflected, and the quality and personalization level of dining recommendation services are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments or exemplary embodiments of the present invention, the drawings required for use in the embodiments or exemplary descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained according to the drawings without paying creative work.

[0065] Figure 1 A flow chart of a dining recommendation method based on user portrait provided by an embodiment of the present invention;

[0066] Figure 2 A dining recommendation flow chart of a dining recommendation method based on user portrait provided by an embodiment of the present invention;

[0067] Figure 3 A block diagram of a dining recommendation system based on user portraits provided by an embodiment of the present invention;

[0068] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0069] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0070] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0071] Figure 1 A flow chart of a dining recommendation method based on user portrait provided by an embodiment of the present invention;

[0072] like Figure 1 As shown, the present invention provides a flow chart of a dining recommendation method based on user portraits, including:

[0073] S102, obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information;

[0074] Obtaining historical order information of the user, performing feature extraction on the historical orders of the user, extracting the type, frequency, price and time features of the dishes in the historical orders of the user, and obtaining historical order feature information;

[0075] Preset cuisine categories, perform cuisine attribute analysis based on the historical order feature information, calculate the Euclidean distance between each order and the preset cuisine category, and make a judgment with a preset threshold to obtain cuisine attribute analysis information;

[0076] Acquire order feedback information according to the historical order information, build a semantic analysis model based on natural language processing technology, input the order feedback information into the semantic analysis model to perform semantic analysis of the feedback information, and obtain semantic analysis result information;

[0077] Performing a taste attribute pre-analysis in combination with the historical order feature information and the cuisine attribute analysis information, extracting the taste features of the dishes in each order and the taste features of each cuisine, and counting the number of each taste feature of the dishes in the order to obtain pre-analysis result information;

[0078] Performing taste attribute analysis according to the semantic analysis result information and the pre-analysis result information to obtain taste attribute analysis information;

[0079] Performing a user meal time attribute analysis based on the historical order feature information, dividing each order by time, and obtaining time attribute analysis information;

[0080] Perform order price attribute analysis based on the user's historical order feature information, analyze the total price of each order and the unit price of each meal, and count the frequency of each price to obtain price attribute analysis information;

[0081] The attribute analysis result information is obtained by combining the cuisine attribute analysis information, the taste attribute analysis information, the time attribute analysis information and the price attribute analysis information.

[0082] It should be noted that, firstly, by obtaining the user's historical order information, the user's dining preferences and behavior patterns are obtained, including key data such as the types of dishes ordered, frequency, price, and dining time. Secondly, through feature extraction, the historical orders are analyzed in detail, and various features are extracted, including the types of dishes, frequency, price, and time, etc., to establish the multi-dimensional characteristics of user behavior, providing a basis for subsequent analysis. Subsequently, the orders are classified into cuisines by pre-set cuisine categories, and the cuisine to which the order belongs is determined, thereby obtaining the cuisine attribute analysis information, refining the user's taste preferences, and improving the accurate cognition of the cuisine. Further, by obtaining order feedback information and using natural language processing technology to build a semantic analysis model, a deep understanding of user feedback is achieved, and the user's subjective evaluation and preference for dishes is excavated, enriching the semantic level of the user portrait. Then, combined with the historical order information and cuisine attributes, the taste features are extracted, and the number of taste features is counted, which provides a basis for the subsequent taste attribute analysis, can more deeply show the user's taste preferences, and provide a more personalized taste reference for catering recommendations. At the same time, we conducted an analysis of the user's dining time attributes, dividing the orders by time to gain a deeper understanding of the user's dining time habits within a day, help understand the user's dining time preferences, and provide time considerations for reasonable recommendations. Finally, we conducted an analysis of the order price attributes, analyzing the total price of the order and the unit price of the meal, and counting the frequency of each price, to gain insight into the user's acceptance of different price ranges, and to provide price guidance for recommendations.

[0083] S104, performing user preference analysis based on the user's historical order information and attribute analysis result information, and constructing a user preference profile;

[0084] Obtain user historical order information and attribute analysis result information;

[0085] Obtain the taste category and cuisine category of the target user according to the attribute analysis result information, and perform analysis based on the principal component analysis method combined with the user's historical order information to obtain principal component analysis information;

[0086] Sort each flavor category and each cuisine category according to the principal component information, perform style preference analysis based on the sorting results, and obtain style preference analysis information;

[0087] Obtaining time attribute analysis information according to the attribute analysis result information, and performing time preference analysis based on a statistical algorithm combined with the price attribute analysis information;

[0088] Count the dining frequency of users in each time period as a time preference analysis indicator and conduct time preference analysis to obtain time preference analysis information;

[0089] Obtain price attribute analysis information according to the attribute analysis result information, and draw a total price trend graph and a unit price trend graph according to the price attribute analysis information;

[0090] Conduct price preference analysis on target customers based on total price trend chart and unit price trend chart, analyze total price preference and unit price preference of target customers, and obtain price preference analysis information;

[0091] Combining the style preference analysis information, the time preference analysis information and the price preference analysis information to form user preference analysis information, and performing user preference level evaluation;

[0092] Preset preference level evaluation indicators, analyze user preference analysis information with the preference level evaluation indicators, evaluate user preference style, dining time and price dimensions and assign corresponding preference levels, and obtain preference level evaluation information;

[0093] A user preference profile is constructed by combining the preference level evaluation information and the user preference analysis information.

[0094] It should be noted that, first, the taste category and cuisine category of the target user are determined according to the attribute analysis result information, and the user's historical order information is analyzed by principal component analysis to obtain principal component analysis information. Subsequently, the taste categories and cuisine categories are sorted according to the principal component analysis information, and the style preference analysis is performed through the sorting results to obtain the user's preference for taste and cuisine, and accurately understand the user's taste tendency and preference for different cuisines. Then, the time attribute analysis information is obtained according to the attribute analysis result information, and the time preference analysis is performed in combination with the statistical algorithm and the price attribute analysis information. By counting the dining frequency of users in each time period, the user's dining time tendency is understood. At the same time, the total price trend chart and the unit price trend chart are drawn using the price attribute analysis information to provide a reference for the subsequent price preference analysis. In the price preference analysis, the target customers are analyzed according to the total price trend chart and the unit price trend chart to understand the user's preference for the total price and unit price, which helps to better understand the user's budget, consumption habits and hobbies. Then, the style preference analysis information, the time preference analysis information and the price preference analysis information are combined to form a comprehensive user preference analysis information, and the user preference level is evaluated. Through the preset preference level evaluation indicators, users are given corresponding preference levels in terms of style, dining time and price, and detailed preference level evaluation information is obtained. Finally, the preference level evaluation information and user preference analysis information are combined to construct the user's preference profile to better meet the user's personalized needs.

[0095] S106, obtaining user historical browsing information, performing user interest analysis based on the user historical browsing information, and constructing a user interest profile;

[0096] Obtaining historical browsing information of the user, and extracting features of the historical browsing information of the user, including browsed dishes, browsing frequency, and browsing time period, to obtain historical browsing feature information;

[0097] Building an associated interest analysis model according to an association rule mining algorithm, inputting the historical browsing feature information into the associated interest analysis model for analysis, and obtaining associated interest analysis information;

[0098] Setting interest tags according to the associated interest analysis information, wherein the interest tags are divided into time interest tags and food interest tags, and setting corresponding interest weights;

[0099] Analyze the user's interest preference according to the interest tags and interest weights combined with the associated interest analysis information and historical browsing features to obtain preference analysis information;

[0100] Performing user preference feature analysis based on the associated interest analysis information and the preference degree analysis information based on a clustering algorithm, analyzing the preference characteristics and types of target users, and obtaining preference feature analysis information;

[0101] Build user interest profiles based on preference degree analysis information, associated interest points information, and preference feature analysis information.

[0102] It should be noted that, first, the user's historical browsing information is feature extracted, including key information such as the browsed dishes, browsing frequency and browsing time, and the historical browsing feature information is obtained. Then, the association rule mining algorithm is used to construct the associated interest analysis model, and the historical browsing feature information is input into the model for analysis to obtain the associated interest analysis information, and the potential association rules in the user's historical browsing behavior are mined to understand the user's interest association. According to the associated interest analysis information, interest tags are set, which are divided into time interest tags and food interest tags, and the corresponding interest weights are set, so as to more accurately express the user's interest in time and food. Next, the user's interest preference degree is analyzed by combining the associated interest analysis information and historical browsing features with interest tags and interest weights, and the user's preference degree for different interest tags is obtained, and the user's interest preference for time and food is quantified. Based on the clustering algorithm, the user preference feature analysis is performed based on the associated interest analysis information and the preference degree analysis information. Through clustering, the user's preference characteristics and types are identified. Finally, the system constructs a user interest profile based on preference degree analysis information, related interest analysis information, and preference feature analysis information, integrates the user's interests in time and food, and provides a more comprehensive understanding of the user to better realize personalized recommendation services.

[0103] S108, predicting the next point of interest for the target user based on the user interest profile and the user preference profile, and obtaining point of interest prediction information;

[0104] The preset attribute categories include location attributes and date attributes. The location attributes are divided into residential, leisure and office, and the date attributes are divided into holiday days and working days.

[0105] Obtain the user's historical dining location information and dining time information, perform attribute analysis based on preset attribute categories, and obtain dining attribute analysis information;

[0106] Preset time intervals, build user access sequences including time, location, and food through meal attribute analysis information, analyze the behavior patterns of target users, and obtain behavior pattern analysis information;

[0107] Build a long-term and short-term preference analysis model based on NGCF to obtain user interest profiles and user preference profiles. Combine dining attribute analysis information and behavior pattern analysis information and input them into the long-term and short-term preference sub-model to conduct long-term and short-term preference analysis on the target users to obtain long-term and short-term preference analysis information.

[0108] Constructing a point of interest prediction model, and inputting the long-term and short-term preference analysis information and the behavior rule analysis information as initial information into the point of interest prediction model to predict the next point of interest;

[0109] The attention mechanism is used to calculate the attention score of each point of interest based on the long-term and short-term preference analysis information and the behavior law analysis information. The score is used as the weight of the point of interest and the interest decision vector is formed through weighted fusion.

[0110] Based on the multi-layer perceptron, the interest decision vector and the interest point are transformed into multi-layer nonlinear transformations and the recommendation score of each interest point is calculated to obtain the prediction information of the candidate interest points.

[0111] The candidate points of interest are sorted according to their recommendation scores, and a selection threshold is preset. The final prediction result is selected by selecting the threshold to obtain the point of interest prediction information.

[0112] It should be noted that, first of all, the system presets attribute categories including location attributes and date attributes, among which location attributes are divided into residential, leisure and office, and date attributes are divided into holiday days and working days. Then, the user's historical dining location information and dining time information are obtained, and attribute analysis is performed in combination with the preset attribute categories to obtain dining attribute analysis information, refine the user's dining habits at different locations and dates, and provide a basis for subsequent behavioral law analysis. By presetting the time interval, the user's access sequence is constructed, including information such as time, location and food. By analyzing the behavioral laws of the target user, the user's dining habits at different time periods and locations are deeply understood. Next, a long-term and short-term preference analysis model is constructed based on NGCF to obtain the user's interest profile and user preference profile, understand the user's long-term and short-term interest preferences, and improve the accurate prediction of the user's interests. Further construct an interest point prediction model, and input the long-term and short-term preference analysis information and behavioral law analysis information as initial information into the model to predict the next interest point. The attention score of each interest point is calculated through the attention mechanism as the interest point weight, and the interest decision vector is formed through weighted fusion. Based on the multi-layer perceptron, the interest decision vector and the interest point are multi-layer nonlinearly transformed, and the recommendation score of each interest point is calculated to obtain the prediction information of the candidate interest point. Finally, by sorting and setting selection thresholds, the next point of interest for the target user is predicted to recommend the food and beverage that the user is interested in, providing the user with food and beverage recommendations that better meet their personalized needs.

[0113] It should be noted that the POI prediction model is a framework constructed based on LSTM, on the basis of which attention mechanism and multi-layer perceptron are added to optimize the model, thereby improving the prediction ability of the model to obtain accurate POI prediction results.

[0114] Furthermore, the real-time location information of the user is obtained, and the high-frequency dietary area of ​​the target user is extracted through the user interest portrait of the target user, and the judgment is made with the real-time location information of the user to obtain the user location judgment information; if the user location judgment information is not in the high-frequency dietary area, the spatial distance between the real-time location information of the user and the high-frequency dietary area is calculated, and the judgment is made with the threshold to determine whether it is in a new dietary area, and the dietary area judgment information is obtained; the user state is analyzed according to the dietary area judgment information and the real-time location information of the user to obtain the real-time state analysis information of the user; according to the real-time state analysis information of the user, it is determined whether to make regional specialty catering recommendations to obtain specialty catering recommendation judgment information; if the specialty catering recommendation judgment information is a recommendation, the regional specialty catering information of the current location is obtained according to the real-time location information of the user, and recommendations are made based on the regional specialty catering information to improve the user's dining experience.

[0115] It should be noted that by analyzing the spatial distance between the user's real-time location and the high-frequency dining areas, it is determined whether the location is outside the high-frequency dining areas, and the user status analysis is performed to determine whether the user is traveling or on a business trip. Based on the status judgment results, regional specialty dining recommendations are made, incorporating regional cultural factors, analyzing the specialty dining culture of the user's area, and providing users with dining recommendations that are more in line with local tastes and cultural atmosphere.

[0116] S110, obtaining real-time weather information, and making restaurant recommendations based on user preference profiles, user interest profiles, and interest point prediction information;

[0117] Obtain real-time weather information, user preference profiles, user interest profiles, and point of interest prediction information;

[0118] Based on big data retrieval, the preference and interest portraits of various groups of people are obtained to form a comparative data set, and then the categories are divided. The Manhattan distance of various groups of people is calculated and judged by the preset threshold to obtain the group classification information;

[0119] Integrate the preferences and interests of various groups of people, merge similar interests and preferences, combine time characteristics to form feature sequences, and construct feature maps;

[0120] Calculate the similarity between the user preference profile and the user interest profile and the feature map, and make a judgment with a preset threshold, analyze the similar groups of the target user, and obtain similar group information;

[0121] Sequence extraction is performed on the feature map according to the information of similar people, and the feature sequence of similar people is extracted to obtain similar feature sequence information;

[0122] Conduct suitable catering analysis based on real-time meteorological information, analyze suitable catering under current meteorological characteristics, and obtain suitable catering analysis information;

[0123] Construct a restaurant recommendation model to make restaurant recommendations based on similar feature sequence information, suitable restaurant analysis information, user preference profile, user interest profile and interest point prediction information to obtain restaurant recommendation information;

[0124] Make restaurant recommendations to target users based on restaurant recommendation information, including location, time, and type of food.

[0125] It should be noted that, first, the preference portraits and interest portraits of various groups of people are obtained through big data retrieval to form a comparative data set, and the categories are divided. By calculating the Manhattan distance between various groups of people and judging with the preset threshold, the information of group classification is obtained. Next, the preferences and interests of various groups of people are integrated, similar interests and preferences are merged, and feature sequences are formed in combination with time characteristics, and feature maps are constructed to understand the similarities between different groups of people. Then, the user preference portrait and user interest portrait are similarly calculated with the feature map, and judged with the preset threshold, and the similar groups of target users are analyzed to obtain similar group information. Through the similar group information, the feature map is sequenced, and the feature sequence of similar groups is extracted, and the behavioral laws and dining habits of similar groups are extracted, so as to map the behavioral laws and dining habits of target users, and provide reference for personalized catering recommendations. Next, suitable catering analysis is performed according to real-time meteorological information, and suitable catering under current meteorological characteristics is analyzed, so as to provide catering recommendations that are closer to the current needs of users according to weather conditions, for example, hot pot is suitable for cold weather and light meals are suitable for hot weather. Finally, a catering recommendation model is constructed to make catering recommendations based on similar feature sequence information, suitable catering analysis information, user preference profile, user interest profile and interest point prediction information. The catering recommendation information is obtained to make catering recommendations for target users, including location, time and type of catering, to meet the personalized catering needs of users. Through various considerations, users are provided with catering recommendation services that are more in line with actual needs.

[0126] Furthermore, similar people's information is obtained, and historical food and beverage purchase information of similar people is obtained based on the similar people's information; food and beverage correlation analysis is performed based on the historical food and beverage purchase information, and the attention scores between each food and beverage are calculated based on the attention mechanism as the confidence of the correlation analysis to obtain the correlation analysis information; the occurrence frequency of related food and beverage is extracted based on the correlation analysis information, and a threshold is preset for judgment, and related food and beverage with a value greater than the preset threshold is defined as highly correlated food and beverage to obtain highly correlated food and beverage analysis information; feature extraction is performed on the historical food and beverage purchase information to obtain historical food and beverage purchase feature information; multiple attribute labels are preset, and historically purchased food and beverages are associated and classified with corresponding attribute labels based on a clustering algorithm to obtain food and beverage attribute classification information; a food and beverage knowledge graph is constructed according to the food and beverage purchase time by combining the historical food and beverage purchase information, the highly correlated food and beverage analysis information and the food and beverage attribute classification information; real-time user behavior information is obtained, including real-time user purchase information, real-time user access information and real-time user interaction information; The method comprises the following steps: extracting features from the real-time purchase behavior information, extracting the user's real-time restaurant purchase information and browsing restaurant information, and obtaining the user's real-time behavior feature information; calculating the similarity between the user's real-time behavior feature information and the restaurant knowledge graph, and judging with a preset threshold, obtaining similar restaurants according to the judgment result, and obtaining similar restaurant information; searching according to the similar restaurant information in combination with the restaurant knowledge graph, extracting the restaurant after the restaurant path where the similar restaurant is located as potential interest restaurant, and obtaining potential interest restaurant information; comparing and judging based on the highly correlated restaurant analysis information and the potential interest restaurant information, judging whether there is highly correlated restaurant; if there is highly correlated restaurant, recommending the corresponding highly correlated restaurant; if there is no highly correlated restaurant, extracting the occurrence frequency of the potential interest restaurant in the restaurant knowledge graph as the recommendation weight, and performing weighted calculation on the potential interest restaurant information; making a restaurant recommendation decision based on the weighted calculation result, and recommending potential intended restaurant to the user, so as to be close to the user's preferences and improve the user experience.

[0127] Figure 2 A dining recommendation flow chart of a dining recommendation method based on user portrait provided by an embodiment of the present invention;

[0128] like Figure 2 As shown, the present invention provides a dining recommendation flow chart of a dining recommendation method based on user portrait, including:

[0129] S202, performing attribute analysis on the target user, analyzing the target user's style preference, time preference, and price preference according to the attribute analysis results, and constructing a preference profile;

[0130] S204, analyzing the interests of the target users and constructing interest portraits;

[0131] S206, predicting the next point of interest by combining the preference portrait and the interest portrait;

[0132] S208, analyzing the characteristics of the target user, and analyzing the restaurants that the user may be interested in through similar groups;

[0133] S210, combining real-time weather conditions to analyze catering suitable for the current weather conditions;

[0134] S212, making restaurant recommendations, including location, time, and type of restaurant.

[0135] Furthermore, a preset information acquisition time interval is used to obtain the target user's catering order information according to the information acquisition time interval; catering information is extracted according to the catering order information, and attribute analysis is performed to analyze the catering attributes of the target user in the current time period to obtain catering attribute analysis information; characteristic information of various categories of catering is obtained based on big data retrieval to form a catering data set; similarity is calculated between the catering attribute analysis information and the catering data set, and a judgment is made with a preset threshold, and catering tendencies are classified according to the judgment result to obtain catering tendency analysis information; a dietary tendency analysis rule is preset, and dietary tendency analysis is performed on the target user in the current time period in combination with the catering tendency analysis information and the catering attribute analysis information; the ratio of the number of caterings of each type of catering tendency of the target user to the total number of caterings is calculated, and a comparative analysis is performed with the dietary tendency analysis rule to obtain dietary tendency analysis information; user portrait information of the target user is obtained, including a user preference portrait and a user interest portrait, and dietary preference change analysis is performed according to the dietary tendency analysis information and the user portrait information to obtain dietary preference change analysis information; personalized catering recommendations are made through dietary preference change analysis information, and catering recommendation strategies and directions are adjusted in real time and dynamically to keep up with the user's real-time preferences and interests;

[0136] It should be noted that by analyzing the user's recent catering orders, analyzing the attributes of the dishes in the orders, and determining whether there have been any changes, the catering attribute tendencies of the target user in the recent time period are obtained, and the user's dietary tendencies are analyzed based on the catering attribute tendency mapping of the target user in the recent time period. The user's dietary habits and preferences are analyzed, and compared with the user portrait of the target user to determine whether there has been a change in dietary tendencies, such as a change from a previous preference for high-calorie foods to a preference for low-calorie foods. The user's real-time tendencies can be understood based on the information analyzed based on the dietary tendency change, and then catering that is close to the real-time situation can be recommended. That is, if the user often chooses low-calorie, high-fiber foods in the current time period, then catering options that meet their healthy dietary goals will be recommended to the user, and more suitable catering will be provided to meet the user's physical and health considerations.

[0137] Figure 3 A dining recommendation system framework based on user portraits provided in one embodiment of the present invention Figure 3The system includes: a memory 31 and a processor 32. The memory 31 includes a dining recommendation method program based on user portraits. When the dining recommendation method program based on user portraits is executed by the processor 32, the following steps are implemented:

[0138] Obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information;

[0139] Performing user preference analysis based on the user's historical order information and attribute analysis result information, and building a user preference profile;

[0140] Obtain user historical browsing information, perform user interest analysis based on the user historical browsing information, and construct a user interest profile;

[0141] Predict the next point of interest for the target user based on the user interest profile and the user preference profile to obtain point of interest prediction information;

[0142] Obtain real-time weather information and make restaurant recommendations based on user preference profiles, user interest profiles, and POI prediction information.

[0143] It should be noted that the present invention provides a dining recommendation method and system based on user portraits, which analyzes the characteristics of target users from the perspective of user preferences and user interests, and analyzes their points of interest based on their browsing behavior, thereby predicting points of interest, including their dining locations, dining interests and dining times, and combining the current real-time weather to analyze suitable dining, giving users a more personalized experience. Considering the user's preferences and the impact of the weather on the user, in order to provide dining recommendations that are more in line with the user's expectations and the current environment, by integrating multiple information sources, it can better meet the user's personalized needs.

[0144] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0145] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0146] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0147] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0148] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention can be essentially or partly reflected in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0149] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A dining recommendation method based on user portrait, characterized in that: include: Obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information; Performing user preference analysis based on the user's historical order information and attribute analysis result information, and building a user preference profile; Obtain user historical browsing information, perform user interest analysis based on the user historical browsing information, and construct a user interest profile; Predict the next point of interest for the target user based on the user interest profile and the user preference profile to obtain point of interest prediction information; Obtain real-time weather information and make restaurant recommendations based on user preference profiles, user interest profiles, and POI prediction information; The performing of user interest analysis based on the user's historical browsing information and constructing a user interest portrait specifically includes: Obtaining historical browsing information of the user, and extracting features of the historical browsing information of the user, including browsed dishes, browsing frequency, and browsing time period, to obtain historical browsing feature information; Building an associated interest analysis model according to an association rule mining algorithm, inputting the historical browsing feature information into the associated interest analysis model for analysis, and obtaining associated interest analysis information; Setting interest tags according to the associated interest analysis information, wherein the interest tags are divided into time interest tags and food interest tags, and setting corresponding interest weights; Analyze the user's interest preference according to the interest tags and interest weights combined with the associated interest analysis information and historical browsing features to obtain preference analysis information; Performing user preference feature analysis based on the associated interest analysis information and the preference degree analysis information based on a clustering algorithm, analyzing the preference characteristics and types of target users, and obtaining preference feature analysis information; Build user interest profiles based on preference degree analysis information, related interest analysis information, and preference feature analysis information; The predicting of the next point of interest of the target user according to the user interest profile and the user preference profile specifically includes: The preset attribute categories include location attributes and date attributes. The location attributes are divided into residential, leisure and office, and the date attributes are divided into holiday days and working days. Obtain the user's historical dining location information and dining time information, perform attribute analysis based on preset attribute categories, and obtain dining attribute analysis information; Preset time intervals, build user access sequences including time, location, and food through meal attribute analysis information, analyze the behavior patterns of target users, and obtain behavior pattern analysis information; Build a long-term and short-term preference analysis model based on NGCF to obtain user interest profiles and user preference profiles. Combine dining attribute analysis information and behavior pattern analysis information and input them into the long-term and short-term preference sub-model to conduct long-term and short-term preference analysis on the target users to obtain long-term and short-term preference analysis information. Constructing a point of interest prediction model, and inputting the long-term and short-term preference analysis information and the behavior rule analysis information as initial information into the point of interest prediction model to predict the next point of interest; The attention mechanism is used to calculate the attention score of each point of interest based on the long-term and short-term preference analysis information and the behavior law analysis information. The score is used as the weight of the point of interest and the interest decision vector is formed through weighted fusion. Based on the multi-layer perceptron, the interest decision vector and the interest point are transformed into multi-layer nonlinear transformations and the recommendation score of each interest point is calculated to obtain the prediction information of the candidate interest points. The candidate points of interest are sorted according to their recommendation scores, and a selection threshold is preset. The final prediction result is selected by selecting the threshold to obtain the prediction information of the points of interest; The acquisition of real-time weather information and the combination of user preference profiles, user interest profiles and point of interest prediction information to make restaurant recommendations specifically include: Obtain real-time weather information, user preference profiles, user interest profiles, and point of interest prediction information; Based on big data retrieval, the preference and interest portraits of various groups of people are obtained to form a comparative data set, and then the categories are divided. The Manhattan distance of various groups of people is calculated and judged by the preset threshold to obtain the group classification information; Integrate the preferences and interests of various groups of people, merge similar interests and preferences, combine time characteristics to form feature sequences, and construct feature maps; Calculate the similarity between the user preference profile and the user interest profile and the feature map, and make a judgment with a preset threshold, analyze the similar groups of the target user, and obtain similar group information; Sequence extraction is performed on the feature map according to the information of similar people, and the feature sequence of similar people is extracted to obtain similar feature sequence information; Conduct suitable catering analysis based on real-time meteorological information, analyze suitable catering under current meteorological characteristics, and obtain suitable catering analysis information; Construct a restaurant recommendation model to make restaurant recommendations based on similar feature sequence information, suitable restaurant analysis information, user preference profile, user interest profile and interest point prediction information to obtain restaurant recommendation information; Make restaurant recommendations to target users based on restaurant recommendation information, including location, time, and restaurant types; Obtain information about similar people, and obtain historical catering information of similar people based on the information about similar people; perform catering correlation analysis based on the historical catering information, calculate the attention scores between various caterings based on the attention mechanism, and use them as the confidence of the correlation analysis to obtain correlation analysis information; extract the occurrence frequency of related catering based on the correlation analysis information, and preset a threshold for judgment, define related catering with a frequency greater than the preset threshold as highly correlated catering, and obtain highly correlated catering analysis information; perform feature extraction on the historical catering information to obtain historical catering feature information; preset multiple attribute labels, and associate and classify historically purchased catering with corresponding attribute labels based on a clustering algorithm to obtain catering attribute classification information; construct a catering knowledge graph based on the catering purchase time by combining the historical catering information, highly correlated catering analysis information, and catering attribute classification information; obtain real-time user behavior information, including real-time user purchase information, real-time user access information, and real-time user interaction information; Feature extraction is performed on the user's real-time purchasing behavior information, and the user's real-time catering purchase information and browsing catering information are extracted to obtain the user's real-time behavior feature information; the user's real-time behavior feature information and the catering knowledge graph are calculated for similarity, and judged with a preset threshold, and similar catering is obtained according to the judgment result to obtain similar catering information; according to the similar catering information and the catering knowledge graph, a search is performed to extract the catering after the catering path where the similar catering is located as potential interest catering, and obtain potential interest catering information; according to the high-correlation catering analysis information and the potential interest catering information, a comparative judgment is made to determine whether there is a highly correlated catering; if there is a highly correlated catering, the corresponding highly correlated catering is recommended; if there is no highly correlated catering, the frequency of occurrence of the potential interest catering in the catering knowledge graph is extracted as a recommendation weight, and the potential interest catering information is weightedly calculated; a catering recommendation decision is made according to the weighted calculation result, and potential intended catering is recommended to the user.

2. A dining recommendation method based on user portrait according to claim 1, characterized in that: The attribute analysis is performed according to the user's historical order information to obtain attribute analysis result information, specifically including: Obtaining historical order information of the user, performing feature extraction on the historical orders of the user, extracting the type, frequency, price and time features of the dishes in the historical orders of the user, and obtaining historical order feature information; Preset cuisine categories, perform cuisine attribute analysis based on the historical order feature information, calculate the Euclidean distance between each order and the preset cuisine category, and make a judgment with a preset threshold to obtain cuisine attribute analysis information; Acquire order feedback information according to the historical order information, build a semantic analysis model based on natural language processing technology, input the order feedback information into the semantic analysis model to perform semantic analysis of the feedback information, and obtain semantic analysis result information; Performing a taste attribute pre-analysis in combination with the historical order feature information and the cuisine attribute analysis information, extracting the taste features of the dishes in each order and the taste features of each cuisine, and counting the number of each taste feature of the dishes in the order to obtain pre-analysis result information; Performing taste attribute analysis according to the semantic analysis result information and the pre-analysis result information to obtain taste attribute analysis information; Performing a user meal time attribute analysis based on the historical order feature information, dividing each order by time, and obtaining time attribute analysis information; Perform order price attribute analysis based on the user's historical order feature information, analyze the total price of each order and the unit price of each meal, and count the frequency of each price to obtain price attribute analysis information; The attribute analysis result information is obtained by combining the cuisine attribute analysis information, the taste attribute analysis information, the time attribute analysis information and the price attribute analysis information.

3. The dining recommendation method based on user portrait according to claim 1, characterized in that: The user preference analysis is performed based on the user's historical order information and attribute analysis result information, and a user preference portrait is constructed, specifically including: Obtain user historical order information and attribute analysis result information; Obtain the taste category and cuisine category of the target user according to the attribute analysis result information, and perform analysis based on the principal component analysis method combined with the user's historical order information to obtain principal component analysis information; Sort each flavor category and each cuisine category according to the principal component information, perform style preference analysis based on the sorting results, and obtain style preference analysis information; Obtaining time attribute analysis information according to the attribute analysis result information, and performing time preference analysis based on a statistical algorithm combined with the price attribute analysis information; Count the dining frequency of users in each time period as a time preference analysis indicator and conduct time preference analysis to obtain time preference analysis information; Obtain price attribute analysis information according to the attribute analysis result information, and draw a total price trend graph and a unit price trend graph according to the price attribute analysis information; Conduct price preference analysis on target customers based on total price trend chart and unit price trend chart, analyze total price preference and unit price preference of target customers, and obtain price preference analysis information; Combining the style preference analysis information, the time preference analysis information and the price preference analysis information to form user preference analysis information, and performing user preference level evaluation; Preset preference level evaluation indicators, analyze user preference analysis information with the preference level evaluation indicators, evaluate user preference style, dining time and price dimensions and assign corresponding preference levels, and obtain preference level evaluation information; A user preference profile is constructed by combining the preference level evaluation information and the user preference analysis information.

4. A restaurant recommendation system based on user portrait, characterized in that: The system includes: a memory and a processor, wherein the memory contains a dining recommendation method program based on user portraits, and when the dining recommendation method program based on user portraits is executed by the processor, the following steps are implemented: Obtaining historical order information of the user, performing attribute analysis based on the historical order information of the user, and obtaining attribute analysis result information; Performing user preference analysis based on the user's historical order information and attribute analysis result information, and building a user preference profile; Obtain user historical browsing information, perform user interest analysis based on the user historical browsing information, and construct a user interest profile; Predict the next point of interest for the target user based on the user interest profile and the user preference profile to obtain point of interest prediction information; Obtain real-time weather information and make restaurant recommendations based on user preference profiles, user interest profiles, and POI prediction information; The performing of user interest analysis based on the user's historical browsing information and constructing a user interest portrait specifically includes: Obtaining historical browsing information of the user, and extracting features of the historical browsing information of the user, including browsed dishes, browsing frequency, and browsing time period, to obtain historical browsing feature information; Building an associated interest analysis model according to an association rule mining algorithm, inputting the historical browsing feature information into the associated interest analysis model for analysis, and obtaining associated interest analysis information; Setting interest tags according to the associated interest analysis information, wherein the interest tags are divided into time interest tags and food interest tags, and setting corresponding interest weights; Analyze the user's interest preference according to the interest tags and interest weights combined with the associated interest analysis information and historical browsing features to obtain preference analysis information; Performing user preference feature analysis based on the associated interest analysis information and the preference degree analysis information based on a clustering algorithm, analyzing the preference characteristics and types of target users, and obtaining preference feature analysis information; Build user interest profiles based on preference degree analysis information, related interest analysis information, and preference feature analysis information; The predicting of the next point of interest of the target user according to the user interest profile and the user preference profile specifically includes: The preset attribute categories include location attributes and date attributes. The location attributes are divided into residential, leisure and office, and the date attributes are divided into holiday days and working days. Obtain the user's historical dining location information and dining time information, perform attribute analysis based on preset attribute categories, and obtain dining attribute analysis information; Preset time intervals, build user access sequences including time, location, and food through meal attribute analysis information, analyze the behavior patterns of target users, and obtain behavior pattern analysis information; Build a long-term and short-term preference analysis model based on NGCF to obtain user interest profiles and user preference profiles. Combine dining attribute analysis information and behavior pattern analysis information and input them into the long-term and short-term preference sub-model to conduct long-term and short-term preference analysis on the target users to obtain long-term and short-term preference analysis information. Constructing a point of interest prediction model, and inputting the long-term and short-term preference analysis information and the behavior rule analysis information as initial information into the point of interest prediction model to predict the next point of interest; The attention mechanism is used to calculate the attention score of each point of interest based on the long-term and short-term preference analysis information and the behavior law analysis information. The score is used as the weight of the point of interest and the interest decision vector is formed through weighted fusion. Based on the multi-layer perceptron, the interest decision vector and the interest point are transformed into multi-layer nonlinear transformations and the recommendation score of each interest point is calculated to obtain the prediction information of the candidate interest points. The candidate points of interest are sorted according to their recommendation scores, and a selection threshold is preset. The final prediction result is selected by selecting the threshold to obtain the prediction information of the points of interest; The acquisition of real-time weather information and the combination of user preference profiles, user interest profiles and point of interest prediction information to make restaurant recommendations specifically include: Obtain real-time weather information, user preference profiles, user interest profiles, and point of interest prediction information; Based on big data retrieval, the preference and interest portraits of various groups of people are obtained to form a comparative data set, and then the categories are divided. The Manhattan distance of various groups of people is calculated and judged by the preset threshold to obtain the group classification information; Integrate the preferences and interests of various groups of people, merge similar interests and preferences, combine time characteristics to form feature sequences, and construct feature maps; Calculate the similarity between the user preference profile and the user interest profile and the feature map, and make a judgment with a preset threshold, analyze the similar groups of the target user, and obtain similar group information; Sequence extraction is performed on the feature map according to the information of similar people, and the feature sequence of similar people is extracted to obtain similar feature sequence information; Conduct suitable catering analysis based on real-time meteorological information, analyze suitable catering under current meteorological characteristics, and obtain suitable catering analysis information; Construct a restaurant recommendation model to make restaurant recommendations based on similar feature sequence information, suitable restaurant analysis information, user preference profile, user interest profile and interest point prediction information to obtain restaurant recommendation information; Make restaurant recommendations to target users based on restaurant recommendation information, including location, time, and restaurant types; Obtain information about similar people, and obtain historical catering information of similar people based on the information about similar people; perform catering correlation analysis based on the historical catering information, calculate the attention scores between various caterings based on the attention mechanism, and use them as the confidence of the correlation analysis to obtain correlation analysis information; extract the occurrence frequency of related catering based on the correlation analysis information, and preset a threshold for judgment, define related catering with a frequency greater than the preset threshold as highly correlated catering, and obtain highly correlated catering analysis information; perform feature extraction on the historical catering information to obtain historical catering feature information; preset multiple attribute labels, and associate and classify historically purchased catering with corresponding attribute labels based on a clustering algorithm to obtain catering attribute classification information; construct a catering knowledge graph based on the catering purchase time by combining the historical catering information, highly correlated catering analysis information, and catering attribute classification information; obtain real-time user behavior information, including real-time user purchase information, real-time user access information, and real-time user interaction information; Feature extraction is performed on the user's real-time purchasing behavior information, and the user's real-time catering purchase information and browsing catering information are extracted to obtain the user's real-time behavior feature information; the user's real-time behavior feature information and the catering knowledge graph are calculated for similarity, and judged with a preset threshold, and similar catering is obtained according to the judgment result to obtain similar catering information; according to the similar catering information and the catering knowledge graph, a search is performed to extract the catering after the catering path where the similar catering is located as potential interest catering, and obtain potential interest catering information; according to the high-correlation catering analysis information and the potential interest catering information, a comparative judgment is made to determine whether there is a highly correlated catering; if there is a highly correlated catering, the corresponding highly correlated catering is recommended; if there is no highly correlated catering, the frequency of occurrence of the potential interest catering in the catering knowledge graph is extracted as a recommendation weight, and the potential interest catering information is weightedly calculated; a catering recommendation decision is made according to the weighted calculation result, and potential intended catering is recommended to the user.

5. A dining recommendation system based on user portrait according to claim 4, characterized in that: Perform attribute analysis based on the user's historical order information to obtain attribute analysis result information, specifically including: Obtaining historical order information of the user, performing feature extraction on the historical orders of the user, extracting the type, frequency, price and time features of the dishes in the historical orders of the user, and obtaining historical order feature information; Preset cuisine categories, perform cuisine attribute analysis based on the historical order feature information, calculate the Euclidean distance between each order and the preset cuisine category, and make a judgment with a preset threshold to obtain cuisine attribute analysis information; Acquire order feedback information according to the historical order information, build a semantic analysis model based on natural language processing technology, input the order feedback information into the semantic analysis model to perform semantic analysis of the feedback information, and obtain semantic analysis result information; Performing a taste attribute pre-analysis in combination with the historical order feature information and the cuisine attribute analysis information, extracting the taste features of the dishes in each order and the taste features of each cuisine, and counting the number of each taste feature of the dishes in the order to obtain pre-analysis result information; Performing taste attribute analysis according to the semantic analysis result information and the pre-analysis result information to obtain taste attribute analysis information; Performing a user meal time attribute analysis based on the historical order feature information, dividing each order by time, and obtaining time attribute analysis information; Perform order price attribute analysis based on the user's historical order feature information, analyze the total price of each order and the unit price of each meal, and count the frequency of each price to obtain price attribute analysis information; The attribute analysis result information is obtained by combining the cuisine attribute analysis information, the taste attribute analysis information, the time attribute analysis information and the price attribute analysis information.

6. A dining recommendation system based on user portrait according to claim 4, characterized in that: The user preference analysis is performed based on the user's historical order information and attribute analysis result information, and a user preference portrait is constructed, specifically including: Obtain user historical order information and attribute analysis result information; Obtain the taste category and cuisine category of the target user according to the attribute analysis result information, and perform analysis based on the principal component analysis method combined with the user's historical order information to obtain principal component analysis information; Sort each flavor category and each cuisine category according to the principal component information, perform style preference analysis based on the sorting results, and obtain style preference analysis information; Obtaining time attribute analysis information according to the attribute analysis result information, and performing time preference analysis based on a statistical algorithm combined with the price attribute analysis information; Count the dining frequency of users in each time period as a time preference analysis indicator and conduct time preference analysis to obtain time preference analysis information; Obtain price attribute analysis information according to the attribute analysis result information, and draw a total price trend graph and a unit price trend graph according to the price attribute analysis information; Conduct price preference analysis on target customers based on total price trend chart and unit price trend chart, analyze total price preference and unit price preference of target customers, and obtain price preference analysis information; Combining the style preference analysis information, the time preference analysis information and the price preference analysis information to form user preference analysis information, and performing user preference level evaluation; Preset preference level evaluation indicators, analyze user preference analysis information with the preference level evaluation indicators, evaluate user preference style, dining time and price dimensions and assign corresponding preference levels, and obtain preference level evaluation information; A user preference profile is constructed by combining the preference level evaluation information and the user preference analysis information.

Images