Tourism customization service method, system and product based on bilateral user matching
By collecting and analyzing multi-source heterogeneous data of users, building a dynamic interest graph, and generating personalized travel guides, the problems of existing platforms in user demand mining and information silos are solved, and more efficient personalized services are achieved.
Patent Information
- Application Number
- CN202510783417.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-12
- Publication Date
- 2025-09-09
AI Technical Summary
Existing tourism service platforms find it difficult to deeply explore users' personalized needs, and their recommendation algorithms are single, resulting in service homogeneity. In addition, users are inefficient in obtaining information across platforms, which easily leads to information island problems.
By collecting users' historical multi-source heterogeneous data, performing preprocessing and sentiment analysis, building a user dynamic interest map, obtaining user needs, and generating personalized travel dynamic strategies.
It improves user experience, avoids data silos, solves service homogeneity issues, can more accurately meet users' personalized needs, and improves strategy selection efficiency.
Smart Images
Figure CN120612155A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of tourism customization, and in particular relates to a tourism customization service method, system and product based on bilateral user matching. Background Art
[0002] At present, with the increasing emotional value consumption of users, the current cultural and tourism industry is in a golden period of consumption upgrading. However, as consumption grows, the contradiction between supply and demand is becoming increasingly acute. Due to false products and traffic fraud, users are bombarded with low-quality travel products, and the chaos of experience degradation and forced shopping caused by the crowding of holidays is frequent.
[0003] High-net-worth users have consumption power, but data on traditional platforms is scattered, and users need to obtain information across platforms, which results in low decision-making efficiency and the problem of information islands. Users are limited by time and energy and find it difficult to screen high-quality information. They are forced to accept standardized itineraries and fall into the dilemma of "paying for trouble". Existing tourism service platforms find it difficult to deeply explore users' personalized needs, have a single recommendation algorithm, and lack personalized customized services, resulting in service homogeneity.
[0004] Based on the above-mentioned deficiencies, how to provide an effective technical solution to solve the problems of information islands and service homogeneity existing in the existing technology has become a difficult problem that needs to be solved urgently in the existing technology. Summary of the Invention
[0005] The purpose of the present invention is to provide a travel customization service method, system and product based on bilateral user matching to solve the above-mentioned problems existing in the prior art.
[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a travel customization service method based on bilateral user matching, comprising: Collecting historical multi-source heterogeneous data of the user, the historical multi-source heterogeneous data including initial social text, initial short video data, and initial consumption behavior, and pre-processing the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; Perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; Build user dynamic interest graphs based on social text, short video data, consumer behavior, text sentiment data, behavioral sentiment data, and image sentiment data; Obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest graphs.
[0007] In one possible design, initial social text, initial short video data, and initial consumption behavior are preprocessed to obtain social text, short video data, and consumption behavior, including: Perform data cleaning, delete duplicate values, and fill missing values on the initial social text, initial short video data, and initial consumption behavior to obtain the filled initial social text, initial short video data, and initial consumption behavior; De-noising the filled initial social text, initial short video data, and initial consumption behavior to obtain the social text, the de-noised initial short video data, and the initial consumption behavior; Performing image enhancement operations on the denoised initial short video data to obtain short video data; The denoised initial consumption behavior is aligned in time and space to obtain the consumption behavior.
[0008] In one possible design, sentiment analysis is performed on social text and short video data to obtain text sentiment data and image sentiment data, and behavioral pattern sentiment analysis is performed on consumer behavior to obtain behavioral sentiment data, including: Extracting sentiment words from social texts to obtain sentiment words, and using a text sentiment analysis model to perform sentiment classification on the sentiment words to obtain text sentiment data, wherein the text sentiment data includes a set of texts with different sentiment polarities and intensities; Extracting features from the short video data to obtain image features, and performing sentiment classification on the image features to obtain image sentiment data, wherein the image sentiment data includes a set of images with different sentiment polarities and intensities; The consumer behavior is classified using the trained behavior emotion classifier to obtain behavior emotion data, which includes a set of behaviors with different emotion polarities and intensities.
[0009] In one possible design, the text sentiment analysis model includes a Transformer-XL layer, a Bi-LSTM layer, a sentiment polarity classification layer, a sentiment intensity classification layer, and an output layer; the process of extracting sentiment vocabulary from social text to obtain sentiment vocabulary, and using the text sentiment analysis model to perform sentiment classification on the sentiment vocabulary to obtain text sentiment data includes: Extract emotional vocabulary from social texts according to preset vocabulary rules to obtain emotional vocabulary; The Transformer-XL layer captures the long-range dependencies of sentiment words and obtains the contextual sentiment feature vector; The contextual sentiment feature vector is processed through the Bi-LSTM layer to obtain forward hidden information and reverse hidden information, which are then combined to obtain the final hidden information. The final hidden information is subjected to sentiment polarity analysis through the sentiment polarity classification layer to obtain the probability distribution of sentiment polarity; The final hidden information is subjected to sentiment intensity analysis through the sentiment intensity classification layer to obtain the probability distribution of sentiment intensity; The output layer obtains text sentiment data based on the probability distribution of sentiment polarity and the probability distribution of sentiment intensity.
[0010] In one possible design, feature extraction is performed on short video data to obtain image features, and based on the image features, image emotion data is obtained, including: Split the short video data into frame sequences; Use the Transformer model to extract features from the frame sequence to obtain image features; Use the trained image sentiment analysis model to perform sentiment analysis on image features to obtain image sentiment data.
[0011] In one possible design, a user dynamic interest graph is constructed based on social text, short video data, consumption behavior, text sentiment data, behavioral sentiment data, and image sentiment data, including: Entity recognition is performed based on social text, short video data, and consumer behavior to obtain entity data and entity connection relationships; Adjust the interest weight of entity data based on text sentiment data, behavioral sentiment data, and image sentiment data, and construct an initial user interest graph based on entity data, entity connection relationships, and interest weights; Obtain real-time multi-source heterogeneous data, use a lightweight graph neural network, dynamically update the initial user interest graph based on the real-time multi-source heterogeneous data, and obtain the user dynamic interest graph.
[0012] In one possible design, based on user needs and user dynamic interest graphs, a dynamic travel guide is obtained, including: Build user profiles based on user dynamic interest graphs; Matching preset travel plans with user needs and user profiles to obtain multiple travel guides. Based on the AI big model and multiple travel guides, a travel guide sequence is obtained, sorted by compliance, and the travel guide with the highest compliance is selected. Acquire local real-time data, which includes weather data and scenic spot heat map data, and dynamically optimize the travel strategy with the highest compliance based on the local real-time data to obtain a dynamic travel strategy.
[0013] In a second aspect, the present invention provides a travel customization service system based on bilateral user matching, which is used to implement the travel customization service method based on bilateral user matching described in the first aspect, comprising: An acquisition and preprocessing module is used to acquire historical multi-source heterogeneous data of users, including initial social text, initial short video data, and initial consumption behavior, and preprocess the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; The sentiment analysis module is used to perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and to perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; The interest graph module is used to construct a user dynamic interest graph based on social text, short video data, consumption behavior, text sentiment data, behavior sentiment data, and image sentiment data; The customized strategy module is used to obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest maps.
[0014] In a third aspect, the present invention provides a computer device comprising a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the travel customization service method based on bilateral user matching as described in the first aspect.
[0015] In a fourth aspect, the present invention provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the travel customization service method based on bilateral user matching as described in the first aspect.
[0016] The beneficial effects of the present invention are as follows: The present invention discloses a method, system and product for customized tourism services based on bilateral matching, including collecting historical multi-source heterogeneous data of users, pre-processing the historical multi-source heterogeneous data of users, performing sentiment analysis on the pre-processed historical multi-source heterogeneous data to obtain text sentiment data, image sentiment data and behavioral sentiment data, adjusting the user's interest weight according to the historical multi-source heterogeneous data, text sentiment data, image sentiment data and behavioral sentiment data, constructing a user dynamic interest graph based on the interest weight and the historical multi-source heterogeneous data, obtaining user needs, and generating a customized tourism dynamic strategy based on the user needs and the user dynamic interest graph. The present invention improves data processing efficiency by centrally collecting and pre-processing the user's historical data on the platform, and recommends a dynamic tourism strategy based on user needs and the user dynamic interest graph. Users can directly view the customized tourism dynamic strategy, improve user experience, avoid the data island problem caused by users querying various types of information, use dynamic strategies to solve the problem of service homogeneity, explore user personalized needs, obtain the tourism dynamic strategy that best meets customer needs, and improve strategy selection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 A flow chart of the travel customization service method provided in Example 1; Figure 2 This is a module diagram of the tourism customization service system provided in Example 2. DETAILED DESCRIPTION
[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the present invention will be briefly introduced below in conjunction with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.
[0019] It should be understood that although the terms "first," "second," etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are merely used to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element without departing from the scope of the exemplary embodiments of the present invention.
[0020] Example 1: like Figure 1 As shown, this embodiment provides a travel customization service method based on bilateral user matching, which can be executed by, but is not limited to, a computer device or a virtual machine with certain computing resources, such as a personal computer or a smart phone, or a virtual machine. The travel customization service method based on bilateral user matching includes, but is not limited to, the following steps: S1. Collect historical multi-source heterogeneous data of the user, including initial social text, initial short video data, and initial consumption behavior, and pre-process the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; Specifically, in step S1, the initial social text, initial short video data, and initial consumption behavior are preprocessed to obtain the social text, short video data, and consumption behavior, including: S11. Clean the initial social text, initial short video data, and initial consumption behavior, remove duplicate values, and fill in missing values to obtain the filled initial social text, initial short video data, and initial consumption behavior; S12. De-noising the filled initial social text, initial short video data, and initial consumption behavior to obtain the social text, the de-noised initial short video data, and the initial consumption behavior; S13. Performing an image enhancement operation on the denoised initial short video data to obtain short video data; S14. Perform temporal and spatial alignment processing on the initial consumption behavior after denoising to obtain the consumption behavior.
[0021] The initial consumption behavior includes but is not limited to purchase history and browsing history including timestamps and spatial locations.
[0022] S2. Perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; Specifically, in step S2, sentiment analysis is performed on social text and short video data to obtain text sentiment data and image sentiment data, and behavioral pattern sentiment analysis is performed on consumer behavior to obtain behavioral sentiment data, including: S21. Extract emotional vocabulary from social text to obtain emotional vocabulary, and use the text sentiment analysis model to perform sentiment classification on the emotional vocabulary to obtain text sentiment data, wherein the text sentiment data includes a text collection with different sentiment polarities and intensities; Among them, the text collection with different sentiment polarity and intensity includes text with positive sentiment polarity and low sentiment intensity, text with positive sentiment polarity and medium sentiment intensity, text with positive sentiment polarity and high sentiment intensity, text with neutral sentiment polarity and low sentiment intensity, text with neutral sentiment polarity and medium sentiment intensity, text with neutral sentiment polarity and high sentiment intensity, text with negative sentiment polarity and low sentiment intensity, text with negative sentiment polarity and medium sentiment intensity, and text with negative sentiment polarity and high sentiment intensity; the text sentiment analysis model is constructed based on the Transformer-XL model, and the text sentiment analysis model includes a Transformer-XL layer, a Bi-LSTM layer, a sentiment polarity classification layer, a sentiment intensity classification layer, and an output layer.
[0023] Specifically, in step S21, sentiment vocabulary is extracted from the social text to obtain sentiment vocabulary, and sentiment classification is performed on the sentiment vocabulary using a text sentiment analysis model to obtain text sentiment data, including: S211. Extract emotional vocabulary from social text according to preset vocabulary rules to obtain emotional vocabulary; Among them, the preset vocabulary rules are set emotional vocabulary rules, such as words such as "happy", "sad" or "angry".
[0024] S212. Capture the long-range dependencies of sentiment words through the Transformer-XL layer to obtain the contextual sentiment feature vector; S213. Process the contextual sentiment feature vector through a Bi-LSTM layer to obtain forward hidden information and reverse hidden information, and combine the forward hidden information and the reverse hidden information to obtain the final hidden information; Among them, the Bi-LSTM layer includes forward LSTM and reverse LSTM. The forward LSTM processes the context emotion feature vector from the forward direction to obtain the forward hidden information, and the reverse LSTM processes the context emotion feature vector from the reverse direction to obtain the reverse hidden information.
[0025] S214. Performing sentiment polarity analysis on the final hidden information through the sentiment polarity classification layer to obtain a probability distribution of sentiment polarity; S215. Performing sentiment intensity analysis on the final hidden information through the sentiment intensity classification layer to obtain a probability distribution of sentiment intensity; S216. The output layer obtains text sentiment data based on the probability distribution of sentiment polarity and the probability distribution of sentiment intensity.
[0026] S22. Extract features from the short video data to obtain image features, perform emotion classification on the image features to obtain image emotion data, wherein the image emotion data includes a set of images with different emotion polarities and intensities; Among them, the set of images with different emotional polarity and intensity includes images with positive emotional polarity and low emotional intensity, images with positive emotional polarity and medium emotional intensity, images with positive emotional polarity and high emotional intensity, images with neutral emotional polarity and low emotional intensity, images with neutral emotional polarity and medium emotional intensity, images with neutral emotional polarity and high emotional intensity, images with negative emotional polarity and low emotional intensity, images with negative emotional polarity and medium emotional intensity, and images with negative emotional polarity and high emotional intensity.
[0027] Specifically, in step S22, feature extraction is performed on the short video data to obtain image features, and emotion classification is performed on the image features to obtain image emotion data, including: S221. Segment the short video data to obtain a frame sequence; S222. Use the Transformer model to extract features from the frame sequence to obtain image features; S223. Use the trained image sentiment analysis model to perform sentiment analysis on image features to obtain image sentiment data.
[0028] S23. Use the trained behavioral emotion classifier to classify consumer behaviors to obtain behavioral emotion data, where the behavioral emotion data includes a set of behaviors with different emotion polarities and intensities.
[0029] Among them, the set of behaviors with different emotional polarity and intensity includes behaviors with positive emotional polarity and low emotional intensity, behaviors with positive emotional polarity and medium emotional intensity, behaviors with positive emotional polarity and high emotional intensity, behaviors with neutral emotional polarity and low emotional intensity, behaviors with neutral emotional polarity and medium emotional intensity, behaviors with neutral emotional polarity and high emotional intensity, behaviors with negative emotional polarity and low emotional intensity, behaviors with negative emotional polarity and medium emotional intensity, and behaviors with negative emotional polarity and high emotional intensity.
[0030] Specifically, the behavior emotion classifier is constructed based on the deep learning model. The training process of the behavior emotion classifier includes: collecting a behavior training set, which includes consumer behaviors with emotion labels, inputting the behavior training set into the behavior emotion classifier for classification, and obtaining a predicted classification result; calculating the loss value of the behavior emotion classifier based on the predicted classification result, emotion label and cross entropy loss; updating the parameters of the behavior emotion classifier based on the loss value to obtain an updated behavior emotion classifier; inputting the behavior training set into the updated behavior emotion classifier for re-training until the preset iteration termination condition is reached, and obtaining a trained behavior emotion classifier.
[0031] S3. Build a user dynamic interest graph based on social text, short video data, consumer behavior, text sentiment data, behavioral sentiment data, and image sentiment data; Specifically, in step S3, a user dynamic interest graph is constructed based on social text, short video data, consumption behavior, text emotion data, behavior emotion data, and image emotion data, including: S31. Perform entity recognition based on social text, short video data, and consumer behavior to obtain entity data and entity connection relationships; S32. Adjust the interest weight of entity data based on text sentiment data, behavioral sentiment data, and image sentiment data, and construct an initial user interest graph based on entity data, entity connection relationships, and interest weights; S33. Obtain real-time multi-source heterogeneous data, use a lightweight graph neural network, dynamically update the initial user interest graph based on the real-time multi-source heterogeneous data, and obtain the user dynamic interest graph.
[0032] Specifically, the code snippet for updating the initial user interest graph using a lightweight graph neural network is as follows: class LightGCN(nn.Module): def __init__(self, num_users, num_items, embedding_dim=64, n_layers=3): super().__init__() self.num_users = num_users self.num_items = num_items self.embedding_dim = embedding_dim self.n_layers = n_layers self.user_embed = nn.Embedding(num_users, embedding_dim) self.item_embed = nn.Embedding(num_items, embedding_dim) nn.init.normal_(self.user_embed.weight, std=0.1) nn.init.normal_(self.item_embed.weight, std=0.1) adj_matrix = self._build_adj_matrix(user_item_matrix) self.adj_matrix = adj_matrix.to(device) Specifically, when using a lightweight graph neural network to update the initial user interest graph, only the local subgraph needs to be updated, giving the user interest graph real-time and variability.
[0033] S4. Obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest graph.
[0034] Specifically, in step S4, based on user needs and user dynamic interest graph, a dynamic travel strategy is obtained, including: S41. Build user profiles based on user dynamic interest graphs; S42. Matching preset travel plans with user needs and user profiles to obtain multiple travel guides. Based on the AI model and the multiple travel guides, a sequence of travel guides is obtained, sorted by degree of compliance, and the travel guide with the highest degree of compliance is selected. Specifically, a user portrait is constructed based on the user's dynamic interest graph as a membership benefit for the user. If the user is an ordinary user, a preset travel plan will be matched according to the user's needs and the user's dynamic interest graph; the degree of compliance will also be displayed to the user, so that the user can intuitively understand the degree of match between the travel guide and needs; the deep sorting model is preferred for the AI large model.
[0035] S43. Obtain local real-time data, which includes weather data and scenic spot heat map data, and dynamically optimize the travel guide with the highest compliance based on the local real-time data to obtain a dynamic travel guide.
[0036] Furthermore, after obtaining the dynamic travel guide, the travel guide containing the matching degree will be displayed to the user for easy viewing. At the same time, other travel guides in the travel guide sequence will also be displayed. The user can select the travel guide he likes and then optimize the travel guide using local real-time data after selection.
[0037] Example 2: like Figure 2 As shown, this embodiment provides a travel customization service system based on bilateral user matching, including: An acquisition and preprocessing module is used to acquire historical multi-source heterogeneous data of users, including initial social text, initial short video data, and initial consumption behavior, and preprocess the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; The sentiment analysis module is used to perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and to perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; The interest graph module is used to construct a user dynamic interest graph based on social text, short video data, consumption behavior, text sentiment data, behavior sentiment data, and image sentiment data; The customized strategy module is used to obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest maps.
[0038] Preferably, the collection and preprocessing module collects user data within this platform. If it is necessary to collect data from other platforms, such as Douyin or Xiaohongshu, the consent of the users and other platforms must be obtained. When cleaning data, the cleaned data uses a JSON configuration file, which can define code field mapping rules for different data sources, support the expansion of data types, and generate interest tags.
[0039] In one possible design, a travel manager module is also included. Users can choose a travel manager according to their needs. The travel manager will be removed from the shelves if the number of orders received exceeds the preset order threshold.
[0040] In a possible design, it also includes a travel butler qualification information display module, which is used to display the butler's qualification information, including the butler's historical service data, user evaluation star rating, skill certification certificate and service price strategy.
[0041] Preferably, the travel customization service system provides users with various travel information, such as the current status of various attractions and users' personal travel experiences. The system pushes customized travel dynamic strategies to users. At the same time, the system also includes a social platform module, similar to Xiaohongshu or Taobao and other similar platforms, for buying and selling customized travel dynamic strategies.
[0042] In this embodiment, the two parties are the service provider and the user. The user selects a travel manager based on user needs, including but not limited to destination information, budget range, time period, and personalized preference tags. The user selects a suitable travel manager and places an order. The system generates a customized dynamic travel guide based on the user's needs. After the manager agrees, the service provider and the user can communicate and adjust the customized dynamic travel guide.
[0043] Specifically, the code snippets that users can filter according to their needs are as follows: doc = nlp_zh(' '.join(keywords)) for token in doc: word = token.text if token.ent_type_ == 'LOC': categories['destination'].append(word) elif token.ent_type_ == 'TIME': categories['time period'].append(word) elif token.ent_type_ == 'MONEY': categories['budget range'].append(word) elif word in custom_dict: categories['preference label'].append(word) return categories When the user travels according to a customized dynamic travel guide, the system obtains the user's real-time location and, based on the user's real-time location, makes phone calls or text messages to the service provider or user from time to time to ensure the safety of both parties. After the trip, the user scores the guide and stores the score in the database, which can be used as a reference for future customers to place orders.
[0044] In one possible design, users can purchase annual cards of different values according to their needs and become members. The membership fees paid for purchasing the annual cards can be deposited as travel funds into an independent account supervised by a licensed financial institution, and the user's exclusive membership rights and interests are activated. If the user has no travel plans, the user can set the financial risk preference for the membership fee. If not set, it can be set according to the system's preset ratio. The system recommends or configures financial management portfolios for customers based on risk preferences. The annualized returns of the financial management portfolios comply with national laws and regulations, and the returns obtained will be returned to the users. Specifically, different levels of membership require different membership fees, and users can choose different levels of membership according to their own needs.
[0045] The working process, working details and technical effects of the travel customization service system provided in this embodiment can be found in the travel customization service method described in Example 1, and will not be described in detail here.
[0046] Example 3: This embodiment provides a computer device comprising a memory, a processor, and a transceiver communicatively connected in sequence. The memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the travel customization service method based on bilateral user matching described in Example 1. Specifically, the memory may include, but is not limited to, random-access memory (RAM), read-only memory (ROM), flash memory, first-input-first-output (FIFO), and / or first-input-last-output (FILO) memory. The processor may include, but is not limited to, an STM32F105 series microprocessor. Furthermore, the computer device may include, but is not limited to, a power supply module, a display screen, and other necessary components.
[0047] The working process, working details and technical effects of the aforementioned computer device provided in this embodiment can be found in the travel customization service method based on bilateral user matching described in Example 1, and will not be repeated here.
[0048] Example 4: This embodiment provides a computer program product, including a computer program or instructions. When the computer program or instructions are executed by a computer, they are used to implement the travel customization service method based on bilateral user matching as described in Example 1.
[0049] The working process, working details and technical effects of the aforementioned computer program product provided in this embodiment can be referred to the travel customization service method based on bilateral user matching as described in Example 1, and will not be described in detail here.
[0050] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.
Claims
1. A travel customization service method based on bilateral user matching, characterized in that: include: Collecting historical multi-source heterogeneous data of the user, the historical multi-source heterogeneous data including initial social text, initial short video data, and initial consumption behavior, and pre-processing the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; Perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; Build user dynamic interest graphs based on social text, short video data, consumer behavior, text sentiment data, behavioral sentiment data, and image sentiment data; Obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest graphs.
2. A travel customization service method based on bilateral user matching according to claim 1, characterized in that: Preprocess the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior, including: Perform data cleaning, delete duplicate values, and fill missing values on the initial social text, initial short video data, and initial consumption behavior to obtain the filled initial social text, initial short video data, and initial consumption behavior; De-noising the filled initial social text, initial short video data, and initial consumption behavior to obtain the social text, the de-noised initial short video data, and the initial consumption behavior; Performing image enhancement operations on the denoised initial short video data to obtain short video data; The denoised initial consumption behavior is aligned in time and space to obtain the consumption behavior.
3. A travel customization service method based on bilateral user matching according to claim 1, characterized in that: Perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data. Perform sentiment analysis on consumer behavior patterns to obtain behavioral sentiment data, including: Extracting sentiment words from social texts to obtain sentiment words, and using a text sentiment analysis model to perform sentiment classification on the sentiment words to obtain text sentiment data, wherein the text sentiment data includes a set of texts with different sentiment polarities and intensities; Extracting features from the short video data to obtain image features, and performing sentiment classification on the image features to obtain image sentiment data, wherein the image sentiment data includes a set of images with different sentiment polarities and intensities; The consumer behavior is classified using the trained behavior emotion classifier to obtain behavior emotion data, which includes a set of behaviors with different emotion polarities and intensities.
4. A travel customization service method based on bilateral user matching according to claim 3, characterized in that: The text sentiment analysis model includes a Transformer-XL layer, a Bi-LSTM layer, a sentiment polarity classification layer, a sentiment intensity classification layer, and an output layer; the sentiment vocabulary extraction from the social text to obtain the sentiment vocabulary, and the sentiment classification of the sentiment vocabulary using the text sentiment analysis model to obtain the text sentiment data, including: Extract emotional vocabulary from social texts according to preset vocabulary rules to obtain emotional vocabulary; The Transformer-XL layer captures the long-range dependencies of sentiment words and obtains the contextual sentiment feature vector; The contextual sentiment feature vector is processed through the Bi-LSTM layer to obtain forward hidden information and reverse hidden information, which are then combined to obtain the final hidden information. The final hidden information is subjected to sentiment polarity analysis through the sentiment polarity classification layer to obtain the probability distribution of sentiment polarity; The final hidden information is subjected to sentiment intensity analysis through the sentiment intensity classification layer to obtain the probability distribution of sentiment intensity; The output layer obtains text sentiment data based on the probability distribution of sentiment polarity and the probability distribution of sentiment intensity.
5. The travel customization service method based on bilateral user matching according to claim 3 is characterized in that: Perform feature extraction on short video data to obtain image features. Based on the image features, image emotion data is obtained, including: Split the short video data into frame sequences; Use the Transformer model to extract features from the frame sequence to obtain image features; Use the trained image sentiment analysis model to perform sentiment analysis on image features to obtain image sentiment data.
6. The travel customization service method based on bilateral user matching according to claim 1, characterized in that: Build a user dynamic interest graph based on social text, short video data, consumer behavior, text sentiment data, behavioral sentiment data, and image sentiment data, including: Entity recognition is performed based on social text, short video data, and consumer behavior to obtain entity data and entity connection relationships; Adjust the interest weight of entity data based on text sentiment data, behavioral sentiment data, and image sentiment data, and construct an initial user interest graph based on entity data, entity connection relationships, and interest weights; Obtain real-time multi-source heterogeneous data, use a lightweight graph neural network, dynamically update the initial user interest graph based on the real-time multi-source heterogeneous data, and obtain the user dynamic interest graph.
7. The travel customization service method based on bilateral user matching according to claim 1, characterized in that: Based on user needs and user dynamic interest graphs, we can obtain dynamic travel strategies, including: Build user profiles based on user dynamic interest graphs; Matching preset travel plans with user needs and user profiles to obtain multiple travel guides. Based on the AI big model and multiple travel guides, a travel guide sequence is obtained, sorted by compliance, and the travel guide with the highest compliance is selected. Acquire local real-time data, which includes weather data and scenic spot heat map data, and dynamically optimize the travel strategy with the highest compliance based on the local real-time data to obtain a dynamic travel strategy.
8. A travel customization service system based on bilateral user matching, used to implement a travel customization service method based on bilateral user matching as claimed in any one of claims 1 to 7, characterized in that: include: An acquisition and preprocessing module is used to acquire historical multi-source heterogeneous data of users, including initial social text, initial short video data, and initial consumption behavior, and preprocess the initial social text, initial short video data, and initial consumption behavior to obtain social text, short video data, and consumption behavior; The sentiment analysis module is used to perform sentiment analysis on social text and short video data to obtain text sentiment data and image sentiment data, and to perform behavioral pattern sentiment analysis on consumer behavior to obtain behavioral sentiment data; The interest graph module is used to construct a user dynamic interest graph based on social text, short video data, consumption behavior, text sentiment data, behavior sentiment data, and image sentiment data; The customized strategy module is used to obtain user needs and obtain dynamic travel strategies based on user needs and user dynamic interest maps.
9. A computer device, characterized in that: The system comprises a memory, a processor and a transceiver which are communicatively connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program and execute the travel customization service method based on bilateral user matching as described in any one of claims 1 to 7.
10. A computer program product comprising a computer program or instructions, characterized in that When executed by a computer, the computer program or the instruction implements the travel customization service method based on bilateral user matching as described in any one of claims 1 to 7.