Search-enhanced large model-based combined tourism commodity recommendation scheme generation method
By using a retrieval-enhanced large model approach, combined with semantic recall and feature scoring strategies, to generate cultural and tourism product combinations, the problem of intelligent cross-industry recommendation was solved, achieving efficient and personalized cultural and tourism product combinations and improving automation and user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NANKAI UNIV
- Filing Date
- 2026-01-30
- Publication Date
- 2026-06-23
AI Technical Summary
Traditional cultural and tourism recommendation systems struggle to achieve intelligent combination recommendations of products across different business formats. They lack personalization and flexibility, and large models lack structured representation and filtering mechanisms in the generation of cultural and tourism products, resulting in low feasibility of recommendation results and an inability to meet users' personalized needs.
The method adopts a retrieval-enhanced large model approach, which extracts preference information from user data, filters product information, and generates a combination of cultural and tourism products that meet user needs by combining semantic recall and feature scoring strategies. The path is optimized by using a multi-starting-point nearest neighbor greedy algorithm to generate image and text details.
It has improved the automation and intelligence of cultural and tourism product combinations, increased product adoption rate and user click-through rate, reduced manual operation costs, and generated combination solutions that are intelligent, personalized and feasible.
Smart Images
Figure CN122264882A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of structured cultural and tourism product combination technology, and in particular to a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model. Background Technology
[0002] Traditional tourism recommendation systems primarily rely on algorithms such as collaborative filtering, content recommendation, or rule matching to make single-item recommendations based on user history or product characteristics. However, these methods often only recommend products within a single industry (such as hotels or attractions), lacking the ability to combine products across different industries and failing to meet users' personalized needs for integrated travel solutions that combine "hotels + dining + entertainment."
[0003] Currently, some platforms attempt to offer bundled tourism products through bundled sales or fixed packages. However, these solutions typically rely on manual configuration or rule combinations based on fixed templates, lacking intelligence and flexibility. Manual configuration is not only costly and slow to update, but also difficult to personalize based on real-time changes in user interests. Rule-based solutions fail to fully understand users' complex natural language descriptions and struggle to accurately match potential travel preferences. Attempts based on knowledge graphs and language models typically utilize large models to help generate and populate node attributes (such as attraction descriptions and restaurant reviews) in the knowledge graph, then use the connections within the graph (such as travel time between attractions) for recommendations. However, the inherent limitation of these methods lies in their core task of "route planning," with the optimization goal of minimizing commuting time, similar to the Traveling Salesman Problem. They can plan the fastest route from attraction A to attraction B and then to attraction C, but lack the ability to "logically combine" across different business sectors and cannot leverage the understanding capabilities of Large Language Models (LLMs).
[0004] Existing large-scale model-based cultural and tourism applications mostly focus on text generation, intelligent customer service, or question-and-answer scenarios. They lack structured representation and screening mechanisms for the combined generation of cultural and tourism products, which are tailored to the characteristics of products across multiple business formats. This results in low feasibility of the generated results. Furthermore, they fail to fully integrate user characteristics and historical order data, leading to insufficient personalization of recommendation results. They also lack a complete process design from semantic recall to combined generation, making it impossible to form an integrated solution in the product filtering, candidate generation, product group generation, and graphic display stages.
[0005] Large models themselves lack knowledge of real-time inventory and specific product information. Directly generating combined products presents a serious illusion problem and cannot guarantee the authenticity, availability, and accuracy of prices of the products.
[0006] Traditional Retrieval Augmentation (RAG) frameworks are primarily designed for unstructured text (such as documents and web pages). Their search engines are difficult to apply directly to structured data like cultural and tourism products, which have multiple business formats, tags, and dimensions (such as price, location, and demographic tags). Designing an efficient search engine that can simultaneously understand users' personalized semantic needs (such as "couples" or "quiet") and accurately match structured product features is a key technical challenge for applying RAG in this field. Summary of the Invention
[0007] This invention aims to at least solve one of the technical problems existing in related technologies. To this end, this invention provides a method for generating combined cultural and tourism product recommendation schemes based on a retrieval-enhanced large-scale model. This method enhances the input context of the large-scale model by retrieving candidate products, ultimately generating logically sound and demand-compliant product combinations, thereby improving the automation level and product adoption rate of cultural and tourism platforms.
[0008] This invention provides a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model, comprising: S1: Acquire user data and use large models to extract user preference information and product strategy from the user data; S2: Obtain product information, filter product information based on user preference information and product grouping strategy, and obtain a filtered product set; S3: Integrate service information based on user data and product information, and adaptively select retrieval enhancement strategies based on service information and user data. Retrieval enhancement strategies include semantic recall and feature scoring. Obtain a list of candidate products from the filtered product set through semantic recall or feature scoring. S4: Combine the candidate product list using a large model to generate a combination of cultural and tourism products that meet user needs; S5: Generate a recommended product recommendation scheme with pictures and text based on the combination of cultural and tourism products.
[0009] Furthermore, the calculation expression for the filtered product set is: in, This is the filtered collection of goods. For the collection of all goods, for The first in One product, For a set of user tags, A collection of product tags. For goods For the user's minimum budget, For the user's maximum budget.
[0010] Furthermore, the user preference information includes explicit preferences and implicit needs.
[0011] Furthermore, semantic recall scoring methods include: S311: Encode user needs into user need semantic vectors, and encode product titles and product description texts into product semantic vectors; S312: Calculate the cosine similarity between the user demand semantic vector and the product semantic vector; S313: Sort the products in the filtered product set according to cosine similarity, select the Top-K candidate products, and obtain the candidate product list.
[0012] Furthermore, the feature scoring methods include: S321: Perform multi-dimensional feature statistics based on user characteristics and historical order data of products to obtain multi-dimensional feature scores of products; S322: Weighted fusion of the multidimensional feature scores of the product to obtain the total product score; S323: Sort the products in the filtered product set according to the total product score, select the products with the highest scores, and obtain a candidate product list.
[0013] Furthermore, the calculation expression for the cultural and tourism product mix is as follows: in, For the final combination of cultural and tourism products, For the large product portfolio model, For user preferences, For the hotel's candidate product list, This is a list of candidate food and beverage items. For the list of candidate products for entertainment, Background knowledge for large models, These are the parameters for the large product portfolio model.
[0014] Furthermore, in step S5, if the user has a travel planning need and the products in the cultural and tourism product combination have latitude and longitude information, the shortest path connecting all products in the cultural and tourism product combination is calculated using a multi-starting-point nearest neighbor greedy algorithm, and the shortest path is integrated into the cultural and tourism product combination.
[0015] This invention also provides a system for generating combined cultural and tourism product recommendation schemes based on a retrieval-enhanced big data model, used to execute the above-mentioned method for generating combined cultural and tourism product recommendation schemes based on a retrieval-enhanced big data model, including: The user demand extraction module acquires user data and uses a large model to extract user preference information and product strategy from the user data. The product filtering module acquires product information and filters the product information based on user preference information and product grouping strategy to obtain a filtered product set. The candidate product acquisition module integrates service information based on user data and product information, and adaptively selects retrieval enhancement strategies based on service information and user data. The retrieval enhancement strategies include semantic recall and feature scoring. The candidate product list is obtained from the filtered product set through semantic recall or feature scoring. The large-model product combination module combines candidate product lists using a large model to generate cultural and tourism product combinations that meet user needs. The image and text details generation module generates image and text product recommendation schemes based on combinations of cultural and tourism products.
[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the method for generating a combined cultural and tourism product recommendation scheme based on a retrieval enhancement big model as described above.
[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model as described above.
[0018] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: This invention accurately identifies personalized preferences based on users' natural language descriptions and feature information, and adaptively filters highly relevant candidate products from multiple business formats such as hotels, restaurants, and entertainment by combining semantic recall or feature scoring strategies. It effectively solves the problem of connecting large models with dynamic product databases. Through a search-then-generate model, it uses real product data to enhance and constrain the generation process of large models, so that the generated product combination schemes combine the intelligence and personalization of large models with the authenticity and feasibility of product data. Furthermore, it generates marketing-attractive graphic and textual details. This invention significantly improves the automation and intelligence of cultural and tourism product combinations, reduces manual operation costs, and increases product adoption rate and user click-through rate, demonstrating good commercial promotion value and application prospects.
[0019] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0021] Figure 1 This is a flowchart illustrating a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model, as provided by the present invention.
[0022] Figure 2 This is a schematic diagram of the structure of a combined cultural and tourism product recommendation scheme generation system based on a retrieval enhancement big model provided by the present invention.
[0023] Figure 3 This is a block diagram of the electronic device provided by the present invention.
[0024] Figure label: 101. User Request Extraction Module; 102. Product Filtering Module; 103. Candidate Product Acquisition Module; 104. Large Model Product Combination Module; 105. Image and Text Details Generation Module; 201. Processor; 202. Communication Bus; 203. Communication Interface; 204. Memory. Detailed Implementation
[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention. The following embodiments are used to illustrate this invention but cannot be used to limit the scope of this invention.
[0026] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0027] The following is combined with Figures 1 to 3This invention describes a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model.
[0028] like Figure 1 As shown, a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced big data model includes: S1: Acquire user data and use large models to extract user preference information and product strategy from the user data; S11: Obtain user data, which includes personalized needs information; User data includes personalized needs information, which includes travel keywords, natural language descriptions, travel budgets, and user profile information. Travel keywords such as cost-effectiveness and tranquility; natural language descriptions such as wanting a cost-effective, tranquil weekend trip suitable for couples; travel budget and user profile information such as age, gender, travel type, and origin; S12: Analyze personalized demand information through a large model to obtain user preferences and product strategy; User preferences include explicit preferences and implicit needs; Explicit preferences such as high cost-performance ratio and couples' travel are used to infer users' implicit needs for hotels, such as quietness and romantic atmosphere. The product strategy includes budget allocation among the three major business sectors: hotels, catering, and entertainment.
[0029] For user data cleaning and format processing, the large language model parses user input and profile information based on semantic understanding and contextual reasoning, automatically extracting user preferences and product combination strategies. For example, by modeling semantic relationships, the model can identify user preferences such as keywords like "high cost-performance ratio" and "couples' travel," and infer users' implicit needs such as the quietness of the hotel and the romantic atmosphere. Meanwhile, the product combination strategy includes budget allocation among the three major business sectors: hotels, restaurants, and entertainment.
[0030] S2: Obtain product information, filter product information based on user preference information and product grouping strategy, and obtain a filtered product set; Acquire product data, clean the product data according to its format, and obtain user product information; set up For the collection of all goods, for The first in One product. For a set of user tags, Let this be a set of product tags. For the user's budget range, , For the user's minimum budget, For the user's maximum budget, For goods The price. Then the filtered product set. The calculation expression is: S3: Integrate service information based on user data and product information, and adaptively select retrieval enhancement strategies based on service information and user data. Retrieval enhancement strategies include semantic recall and feature scoring. Obtain a list of candidate products from the filtered product set through semantic recall or feature scoring. The services include three types: packaged product services, pre-sale packaged product services, and pre-assembled product services. When the service is a product service, the user information includes keywords, descriptions, and user characteristics, while the product information includes characteristics and descriptions. When the service is a pre-sale package service, it includes: user information with keywords, description and user characteristics, while product information only has description and user information only has user characteristics, while product information only has description. When the service is a pre-assembled product service, the user information only contains user characteristics, while the product information includes both characteristics and descriptions. When the service is a pre-sale package service, and the user lacks keywords, description information or the product lacks feature information, the product cannot be recalled. In this case, the system uses the completion module to complete the user's keywords, description information and product features based on the user's feature information and the product's description information using a large model. The service type is determined based on user information, and a search enhancement strategy is adaptively selected based on the service type. The search enhancement strategy includes semantic recall and feature scoring. Semantic recall scoring methods include: S311: Encode user needs into user need semantic vectors, and encode product titles and product description texts into product semantic vectors; S312: Calculate the cosine similarity between the user demand semantic vector and the product semantic vector. The calculation expression is as follows: in, Cosine similarity between user needs and products For user demand semantic vectors, For product semantic vectors, For the first A semantic vector of user needs For the first Each product semantic vector, The number of vectors.
[0031] S313: Sort the products in the filtered product set according to cosine similarity, select the Top-K candidate products, and obtain the candidate product list.
[0032] Top-K candidate products are selected based on similarity ranking to meet the needs of users who require natural language descriptions.
[0033] Feature scoring methods include: S321: Perform multi-dimensional feature statistics based on user characteristics and historical order data of products to obtain multi-dimensional feature scores of products; The feature scoring module is suitable for users with clear profile information. It calculates scores based on user characteristics (such as gender, age, travel type, and origin) and historical order data of products. For example, if female users account for 50% of the historical orders of a certain product, then the product scores 0.5 in the gender dimension. S322: Weighted fusion of the multidimensional feature scores of the product is performed to obtain the total product score. The calculation expression is as follows: in, For the total score, For the first The weighted weights of each feature dimension, For the first Single-dimensional score under each feature dimension For historical order data of products, For user characteristics, The number of multidimensional features; S323: Sort the products in the filtered product set according to the total product score, select the products with the highest scores, and obtain a candidate product list.
[0034] S4: Combine the candidate product list using a large model to generate a combination of cultural and tourism products that meet user needs; Based on understanding user intent and candidate product characteristics, the large language model combines cultural tourism product grouping logic and background knowledge to combine products across multiple business formats and present them as product groups. The model can automatically filter out mutually compatible hotels, restaurants, and entertainment projects based on constraints (such as budget, theme consistency, and business format quantity constraints) to generate complete cultural tourism product combinations that meet user needs.
[0035] For example, when users prefer "luxury and romantic couples' trips", the model will tend to select hotels with bathtubs with views, highly rated Western restaurants, and entertainment activities related to night views, achieving semantic consistency in the combination.
[0036] Through constraint checks, the rationality, feasibility, and diversity of the combination results are ensured. The final calculation expression for the cultural and tourism product combination is as follows: in, For the final combination of cultural and tourism products, For the large product portfolio model, For user preferences, For the hotel's candidate product list, This is a list of candidate food and beverage items. For the list of candidate products for entertainment, Background knowledge for large models, These are the parameters for the large product portfolio model.
[0037] S5: Generate detailed images and text based on the combination of cultural and tourism products.
[0038] If a user has travel planning needs and the products in the cultural and tourism product bundle have latitude and longitude information, the shortest path connecting all products in the cultural and tourism product bundle is calculated using a multi-starting-point nearest neighbor greedy algorithm, and the shortest path is integrated into the cultural and tourism product bundle.
[0039] After generating the cultural and tourism product bundles, the system automatically generates a comprehensive text description, including the bundle name and product details. Based on user needs, it can further utilize a text-to-image model to generate corresponding product images, or directly retrieve existing product images from the database. The system's post-processing module integrates and typesets the text and image information, outputting a detailed product page. This page can be directly displayed on web pages or in cultural and tourism applications. The final generated bundled product content features a consistent style and natural expression, significantly enhancing the user's reading experience and purchase intention.
[0040] In practical deployment, this invention has been applied and verified on a cultural tourism e-commerce platform. Results show that this invention significantly improves the efficiency of combining cultural tourism products, with an average product combination time of 2 minutes and 27 seconds. Simultaneously, the system-generated product combinations have an adoption rate of over 12% by business stakeholders, and the average user click-through rate has increased to over 15%, significantly outperforming traditional manual product combination and rule-based recommendation schemes.
[0041] like Figure 2 As shown, a combined cultural and tourism product recommendation scheme generation system based on a retrieval-enhanced big data model is used to execute the aforementioned method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced big data model, including: The user demand extraction module 101 acquires user data and uses a large model to extract user preference information and product strategy from the user data. The product filtering module 102 acquires product information, filters the product information based on user preference information and product grouping strategy, and obtains a filtered product set. The candidate product acquisition module 103 integrates service information based on user data and product information, and adaptively selects retrieval enhancement strategies based on service information and user data. The retrieval enhancement strategies include semantic recall and feature scoring. The candidate product list is obtained from the filtered product set through semantic recall or feature scoring. The large-scale product combination module 104 combines the candidate product list through the large model to generate a combination of cultural and tourism products that meets the user's needs. The image and text details generation module 105 generates image and text product recommendation schemes based on combinations of cultural and tourism products.
[0042] Through the collaborative work of the above modules, personalized preferences are accurately identified based on users' natural language descriptions and feature information. Combined with semantic recall or feature scoring strategies, highly relevant candidate products from multiple business formats such as hotels, restaurants, and entertainment are adaptively selected. This effectively solves the problem of connecting large models with dynamic product databases. By adopting a search-then-generate model, the generation process of large models is enhanced and constrained by real product data. This results in product combination solutions that combine the intelligence and personalization of large models with the authenticity and feasibility of product data. Furthermore, it generates marketing-attractive graphic and textual details. This invention significantly improves the automation and intelligence of cultural and tourism product combinations, reduces manual operation costs, and increases product adoption rate and user click-through rate. It has good commercial promotion value and application prospects.
[0043] Figure 3 An example is a block diagram of an electronic device, such as... Figure 3 As shown, the electronic device may include a processor 201, a communication interface 203, a memory 204, and a communication bus 202. The processor 201, communication interface 203, and memory 204 communicate with each other via the communication bus 202. The processor 201 can call logical instructions from the memory 204 to execute a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large-scale model.
[0044] Furthermore, the logical instructions in the aforementioned memory 204 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0045] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to execute a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval enhancement big model provided by the above methods.
[0046] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model, as provided by the above methods.
[0047] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0048] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0049] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model, characterized in that, include: S1: Acquire user data and use large models to extract user preference information and product strategy from the user data; S2: Obtain product information, filter product information based on user preference information and product grouping strategy, and obtain a filtered product set; S3: Integrate service information based on user data and product information, and adaptively select retrieval enhancement strategies based on service information and user data. Retrieval enhancement strategies include semantic recall and feature scoring. Obtain a list of candidate products from the filtered product set through semantic recall or feature scoring. S4: Combine the candidate product list using a large model to generate a combination of cultural and tourism products that meet user needs; S5: Generate a recommended product recommendation scheme with pictures and text based on the combination of cultural and tourism products.
2. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, The expression for calculating the filtered set of goods is: in, This is the filtered collection of goods. For the collection of all goods, for The first in One product, For a set of user tags, A collection of product tags. For goods For the user's minimum budget, For the user's maximum budget.
3. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, The user preference information includes explicit preferences and implicit needs.
4. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, Semantic recall scoring methods include: S311: Encode user needs into user need semantic vectors, and encode product titles and product description texts into product semantic vectors; S312: Calculate the cosine similarity between the user demand semantic vector and the product semantic vector; S313: Sort the products in the filtered product set according to cosine similarity, select the Top-K candidate products, and obtain the candidate product list.
5. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, Feature scoring methods include: S321: Perform multi-dimensional feature statistics based on user characteristics and historical order data of products to obtain multi-dimensional feature scores of products; S322: Weighted fusion of the multidimensional feature scores of the product to obtain the total product score; S323: Sort the products in the filtered product set according to the total product score, select the products with the highest scores, and obtain a candidate product list.
6. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, The formula for calculating the cultural and tourism product mix is: in, For the final combination of cultural and tourism products, For the large product portfolio model, For user preferences, For the hotel's candidate product list, This is a list of candidate food and beverage items. For the list of candidate products for entertainment, Background knowledge for large models, These are the parameters for the large product portfolio model.
7. The method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model according to claim 1, characterized in that, In step S5, if the user has a travel planning need and the products in the cultural and tourism product combination have latitude and longitude information, the shortest path connecting all products in the cultural and tourism product combination is calculated using a multi-starting point nearest neighbor greedy algorithm, and the shortest path is integrated into the cultural and tourism product combination.
8. A combined cultural and tourism product recommendation solution generation system based on a retrieval-enhanced large model, characterized in that, A method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model as described in any one of claims 1 to 7, comprising: The user demand extraction module acquires user data and uses a large model to extract user preference information and product strategy from the user data. The product filtering module acquires product information and filters the product information based on user preference information and product grouping strategy to obtain a filtered product set. The candidate product acquisition module integrates service information based on user data and product information, and adaptively selects retrieval enhancement strategies based on service information and user data. The retrieval enhancement strategies include semantic recall and feature scoring. The candidate product list is obtained from the filtered product set through semantic recall or feature scoring. The large-model product combination module combines candidate product lists using a large model to generate cultural and tourism product combinations that meet user needs. The image and text details generation module generates image and text product recommendation schemes based on combinations of cultural and tourism products.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced big data model as described in any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for generating a combined cultural and tourism product recommendation scheme based on a retrieval-enhanced large model as described in any one of claims 1 to 7.