Large-model-driven schematic tourist route map automatic generation method

Abstract

The invention relates to the technical field of map making, in particular to a method for automatically generating a schematic tourist route map driven by a large model, which comprises the following steps of: 1, finely adjusting an open-source large model for tourist demand understanding; 2, generating a schematic tourist route map; and 3, carrying out symbol personalized design and sheet finishing. According to the method, the travel route map subjected to sheet finishing can be better and automatically generated through fine tuning of the large model.

Description

Technical Field

[0001] The present invention relates to the technical field of map making, and in particular to a method for automatically generating a schematic tourist route map driven by a large model. Background Art

[0002] Tourist maps are maps that provide navigation and information for tourists, usually including relevant information such as tourist attractions, landmarks, transportation networks, catering and accommodation. Traditional travel based on travel agencies has relatively fixed tour routes and the presence of tour guides. Tourists do not rely heavily on auxiliary information such as tourist maps. However, tourists who travel independently are more pursuing diverse travel routes, and their needs are often different. Although this group of people can collect travel guides from travel websites, social media and other channels to form a travel planning map that meets their needs, this requires strong information collection capabilities and is time-consuming and labor-intensive.

[0003] To achieve personalized multimodal expression of tourist maps, it is required that the mapping system can fully understand the needs of different users and complete multimodal map generation. However, most traditional automatic map making methods are based on pre-set mapping scenarios, not specific needs of a certain user. In recent years, deep learning large models based on Transformer architecture, such as BERT and GPT series, have greatly promoted the progress of natural language generation. These models can understand the context, adapt to the needs of different scenarios, and generate high-quality texts, and have reached the level of quasi-general artificial intelligence. However, the current practical needs for multimodal automatic expression of personalized tourist maps require analysis of the difficulties in intelligent map expression and research on intelligent map expression methods with generative artificial intelligence as the brain; in addition, although large language models have shown excellent performance in natural language processing, due to the lack of tourism-related corpora and the lack of special training models to understand the relationship between tourist needs and mapping information, it is necessary to further optimize the general large model through incremental pre-training and supervised fine-tuning. Summary of the invention

[0004] The present invention provides a large model driven schematic tourist route map automatic generation method, which can overcome certain or some defects of the prior art.

[0005] The large model-driven schematic tourist route map automatic generation method according to the present invention comprises the following steps:

[0006] Step 1: Fine-tune the open source big model for understanding tourist needs;

[0007] 1.1) Incremental pre-training;

[0008] 1.2) supervised fine-tuning;

[0009] 1.3) Prompt engineering template design;

[0010] Step 2: Generate a schematic travel route map;

[0011] 2.1) Expand the congested area;

[0012] 2.2) Generate regular circuits;

[0013] 2.3) Name and annotation placement;

[0014] Step 3: Personalized symbol design and picture finishing;

[0015] 3.1) Use the open source big model to generate corresponding symbols according to the characteristics of different scenic spots;

[0016] 3.2) The generated symbols are combined with the actual travel route to generate a route map including daily itinerary, sequence of attractions, and transportation information;

[0017] 3.3) Renovate the map, including adjusting the position of text annotations, optimizing the layout of scenic spot symbols, and adding detailed route descriptions and annotations;

[0018] 3.4) Combine the above steps to generate a travel route map.

[0019] Preferably, in step 1.1), specifically:

[0020] 1.1.1) Compile Python crawler scripts and use the BeautifulSoup library to efficiently crawl multi-dimensional data related to attractions from tourism websites, including attraction names, geographic locations, ratings, user reviews, ticket prices, and opening hours;

[0021] 1.1.2) Extract key information from the multidimensional data obtained by the crawler script, including city, attraction name, location, address and introduction;

[0022] 1.1.3) Eliminate attractions with a score below 2.5 and no reviews to ensure data quality;

[0023] 1.1.4) Combine key information into descriptive text to present scenic spot information;

[0024] 1.1.5) Output the descriptive text in a structured JSON format for use in the construction of a standardized dataset in subsequent pre-training;

[0025] 1.1.6) Training large models to build high-quality datasets;

[0026] The model predicts the next word based on the existing word sequence, and after each new word is generated, it is included in the word sequence and recursively performs the next prediction; after each prediction, the model's prediction result is compared with the actual next word and the loss function is calculated; through the back-propagation algorithm, the calculated loss is used to update the model's parameters, so that the model can gradually improve its prediction accuracy.

[0027] Preferably, in step 1.2), specifically:

[0028] 1.2.1) Obtain tourism-related question and answer data through crawler technology;

[0029] 1.2.2) Manual data cleaning, including removing questions and answers that do not meet the needs of personalized travel maps and modifying and improving incomplete questions;

[0030] 1.2.3) Improve the question-answering data, add the step of analyzing key information from the question and then thinking about the next step, that is, use the thinking chain CoT method to build a large instruction training model, supplement the extraction of necessary conditions in the question, analyze the tourism demand conditions and use this as a basis to recommend tourist attractions for tourism route planning;

[0031] 1.2.4) Generate more question-answering data by artificial construction and self-generation; construct a question-answering dataset according to the question-analysis-answer framework, and finally establish supervised fine-tuning instruction data and convert it into JSON format;

[0032] 1.2.5) Based on the incremental pre-trained large model, the instruction data constructed is further supervised and fine-tuned to generate a dedicated large model suitable for the tourism field. During the fine-tuning process, the model will use the instruction data as a reference value to calculate the loss function and adjust the parameters;

[0033] 1.2.6) BLEU-4, ROUGE-L and ROUGE-2 are used as evaluation indicators for evaluation.

[0034] Preferably, in step 1.3), specifically:

[0035] 1.3.1) Based on the two core elements of the travel route map: attractions and days, the needs are divided into three types, and corresponding response requirements are provided for each type:

[0036] (1) Less detailed requirements: The user only specifies the destination; the large model’s response needs to generate a list of attractions near the destination for each day. The recommended attractions should come from the attraction data learned by the model, and the default is a 3-day tour.

[0037] (2) Medium-detailed requirements: The user specifies the destination and the number of days or specific attractions; the large model should plan the route based on the date or attractions requested by the user, and recommend attractions that are not explicitly mentioned by the user but are worth visiting;

[0038] (3) Very detailed requirements: Users specify the destinations and attractions they want to visit every day in detail; the large model must strictly follow the user's requirements to ensure that all attractions are included within the specified time; if the user's list of attractions is not enough to fill the itinerary, it can suggest highly recommended attractions to enrich the user experience while maintaining consistency with their overall preferences;

[0039] 1.3.2) The large model collects necessary mapping information based on the planned route and clarifies seven key information: 1) attraction name; 2) attraction coordinates; 3) attraction ticket price; 4) attractions visited every day; 5) visiting time of each attraction; 6) overall route description; 7) transportation method;

[0040] 1.3.3) Two strategies are defined and incorporated into the prompt template; the first is to combine instances, covering specific examples of different types of user needs and big model responses, to better demonstrate the differences between user inputs at different levels of detail; the second is to adopt a reasoning-and-action framework that allows the big model to collaboratively reason and execute through an iterative process.

[0041] Preferably, in step 2.1), specifically:

[0042] The problem of crowded and uneven distribution of scenic spots is solved by adjusting the positions of points to expand the crowded area so as to achieve uniform distribution. Specifically, points are dispersed in dense areas and concentrated in sparse areas. To simulate this process, the Voronoi diagram is used to describe the spatial distribution of points.

[0043] Preferably, in step 2.2), a grid-based method is used to generate regular connection lines, so that the lines are concise and beautiful, specifically:

[0044] 2.2.1) Reposition the point to the nearest grid corner;

[0045] First determine the appropriate mesh side length. The side length must ensure that the generated mesh contains only one point in each cell. The side length is determined by the following calculation:

[0046]

[0047] In the formula, and are the distances between points i and j in the x and y directions, respectively, and b is the number of points;

[0048] Then, the range of the grid is determined based on the coordinate range of the point set; after the grid is drawn, each point is relocated to the nearest grid corner point;

[0049] 2.2.2) Connect corner points along the grid edges;

[0050] Connect all point features along the grid edge one by one. The connection order follows the order of visiting scenic spots. For each connection, define three constraints:

[0051] (a) The connecting edge should maintain good continuity;

[0052] (b) Connecting edges cannot pass through other point features;

[0053] (c) A connected edge cannot overlap with a previously connected edge;

[0054] To satisfy constraint (a), the change in direction should be minimized; this is achieved by identifying the intersections of horizontal and vertical lines of points; these points are connected along horizontal and vertical lines through these intersections to form connecting edges with minimum change in direction and shortest distance;

[0055] For constraints (b) and (c), a greedy algorithm and a backtracking algorithm are used to connect point by point and judge edge by edge until the final route meets the constraints. Specifically, the n-th order adjacent point is used as the turning point of the route between any two points to avoid the current route passing through other attractions or overlapping with previously existing connection edges.

[0056] As a preferred embodiment, in step 2.3), the problem of name annotation overlapping is solved by designing annotation placement rules, and the steps are as follows:

[0057] 2.3.1) The initial annotation is placed in the upper right position;

[0058] 2.3.2) Identify the overprint mark;

[0059] 2.3.3) Move the overlapping annotation according to the priority of the eight candidate annotation positions;

[0060] 2.3.4) Repeat step 2 until there are no more stamp marks.

[0061] The beneficial effects of the present invention are as follows:

[0062] The present invention automatically generates a tourist route map by fine-tuning a large model and then finishing the map. This automated process not only improves the accuracy and personalization of the route map, but also greatly improves the efficiency and aesthetics of mapping, providing users with a better travel planning experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1It is a flowchart of a method for automatically generating a schematic tourist route map driven by a large model in an embodiment;

[0064] Figure 2 is the Voronoi diagram of the points in the embodiment;

[0065] Figure 3 Schematic diagram of the original scenic spot being updated to the centroid of the corresponding Voronoi unit in the embodiment;

[0066] Figure 4 Schematic diagram of the point homogenization process based on geometric information entropy in the embodiment;

[0067] Figure 5 Schematic diagram of a method for determining the side length of a grid and its corner point positioning in an embodiment;

[0068] Figure 6 Schematic diagram of two ways of connecting point 1 and point 2 with the smallest change in direction in the embodiment;

[0069] Figure 7 is a schematic diagram of a connection edge with an adjacent point as an inflection point in an embodiment;

[0070] Figure 8 A schematic diagram illustrating a tourist route map in an embodiment;

[0071] Fig. 9 It is a schematic diagram of a schematic tourist route map with personalized symbol design and map decoration in an embodiment. DETAILED DESCRIPTION

[0072] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0073] Example 1

[0074] like Figure 1 As shown, this embodiment provides a large model driven schematic tourist route map automatic generation method, which includes the following steps:

[0075] Step 1: Fine-tune the open source big model to understand tourist needs.

[0076] The purpose of incremental pre-training is to expand the knowledge reserve of the big model on tourist attraction related information and enhance its basic knowledge and corpus in the tourism field. The core is to build a massive and high-quality training data set. Using crawler technology, about 70,000 scenic spot data were obtained from common travel websites such as Ctrip and Qunar.com. After data cleaning and screening, 62,000 high-quality data were finally obtained for model training.

[0077] 1.1) Incremental pre-training;

[0078] In step 1.1), specifically:

[0079] 1.1.1) By compiling Python crawler scripts, the BeautifulSoup library is used to efficiently crawl multi-dimensional data related to attractions from tourism websites, including attraction names, geographic locations, ratings, user comments, ticket prices and opening hours; the script combines regular expressions and XPath and other technical means to accurately parse the HTML structure of the web page and extract the required information. In addition, the crawler has designed a variety of strategies to automatically identify and avoid anti-crawling mechanisms, such as dynamically adjusting request header information and access frequency to avoid triggering the website's anti-crawling measures. In order to improve the efficiency of data crawling and reduce the risk of IP blocking, the system also integrates an IP rotation mechanism. In order to enhance the robustness of the system, the crawler introduces a fault recovery mechanism to ensure that it can automatically retry when the network is interrupted or the crawling fails, ensuring the integrity and continuity of data collection. Finally, the crawled data is stored as a CSV file, and the system will automatically check the data format, field integrity and value range before writing the file to prevent abnormal data from entering the file.

[0080] 1.1.2) Extract key information from the multidimensional data obtained by the crawler script, including city, attraction name, location, address and introduction.

[0081] 1.1.3) Eliminate attractions with a rating below 2.5 and no reviews to ensure data quality.

[0082] 1.1.4) Combine key information into descriptive text to present attraction information in a concise and clear manner.

[0083] 1.1.5) Output the descriptive text in a structured JSON format for use in the construction of a standardized dataset in subsequent pre-training.

[0084] 1.1.6) Train large models to learn and construct high-quality datasets.

[0085] Based on the Qwen2.5-7B-Instruct model, the tool LLaMA-Factory recommended by the Qwen official document is used, and the LoRA and QLoRA technologies are used to incrementally pre-train the model, that is, the model predicts the next word based on the existing word sequence (context), and after each new word is generated, it is included in the word sequence and recursively predicts the next step; after each prediction, the model's prediction result is compared with the actual next word, and the loss function is usually the cross entropy loss); through the back-propagation algorithm, the calculated loss is used to update the model parameters, so that the model gradually improves the prediction accuracy. The hyperparameter settings of the training model are shown in Table 1:

[0086] Table 1 Incremental training model hyperparameter settings

[0087]

[0088]

[0089] 1.2) Supervised fine-tuning.

[0090] The goal of supervised fine-tuning is to strengthen the connection between tourist needs and understanding of cartographic information, so that the model can better adapt to specific tasks. The key lies in building a high-quality instruction dataset.

[0091] In step 1.2), specifically:

[0092] 1.2.1) We used crawler technology to obtain nearly 2,000 travel-related Q&A data from the Q&A forums of websites such as Ctrip and Qunar.

[0093] 1.2.2) To ensure the reliability and accuracy of the data, the data was manually cleaned, including removing questions and answers that did not meet the needs of personalized travel maps and modifying incomplete questions; more than 800 question and answer data were obtained.

[0094] 1.2.3) Improve the question-answer data, add the step of analyzing key information from the question and then thinking about the next step, that is, use the Chain of Thought (CoT) method to build a large instruction training model, supplement the extraction of necessary conditions in the question, analyze the tourism demand conditions and recommend tourist attractions for travel route planning based on this; examples are shown in Table 2:

[0095] Table 2 Example of “Question-Analysis-Answer” construction

[0096]

[0097] 1.2.4) Generate more question-answering data by artificial construction and self-generation; construct a question-answering dataset according to the "question-analysis-answer" framework, and finally establish more than 1,600 high-quality supervised fine-tuning instruction data and convert them into JSON format.

[0098] 1.2.5) Based on the incremental pre-trained large model, the instruction data constructed by further supervised fine-tuning is used to generate a dedicated large model suitable for the tourism field; during the fine-tuning process, the model uses the instruction data as a reference value to calculate the loss function and adjust the parameters; thus, the model will automatically align the output format and style with the artificial data set, and the performance in mapping information provision or travel route planning tasks will be significantly improved, and ultimately better meet the personalized travel needs of users. The settings of the supervised fine-tuning hyperparameters are shown in Table 3:

[0099] Table 3. Hyperparameter settings of supervised fine-tuning model

[0100] Hyperparameters Settings cutoff_len (cutoff length) 2048 finetuning_type (finetuning type) LoRA learning_rate 5.0e-05 lora_rank (rank of the LoRA matrix) 8 num_train_epochs (number of training rounds) 3 val_size (validation set ratio) 0.1

[0101] 1.2.6) BLEU-4, ROUGE-L and ROUGE-2 are used as evaluation indicators for evaluation. BLEU-4 refers to the ratio of the number of quadruple in the prediction result to the number of quadruple in the reference result; ROUGE-L is an indicator that measures the similarity between the prediction result and the reference result based on the recall rate of the longest common subsequence between the two; ROUGE-2 is the ratio of the number of identical bigrams in the prediction result and the reference result to the total number of bigrams in the reference result. The larger the value of these indicators, the closer the model's answer is to the human answer. Table 4 lists the indicator results of different models, and it is found that the model with 8-bit quantization pre-training 5epoch+SFT3epoch performs best.

[0102] Table 4 BLEU-4, ROUGE-L and ROUGE-2 results of supervised fine-tuning models

[0103] Model bleu-4 rouge-L rouge-2 Pre-training 3 epochs + SFT 3 epochs with BF16 accuracy 5.1487 14.9252 4.5001 Pre-training 5 epochs + SFT 3 epochs with 8-bit quantization 5.7509 16.3332 5.4467 Qwen2.5-7B-Instruct 4.7268 12.1915 5.0662

[0104] Whether the model is aligned with human preferences cannot be evaluated only by similarity indicators, and manual evaluation is also very important. Therefore, the chat reasoning ability of the model was further evaluated through manual analysis. Taking the input "Prompt: I am going to Xi'an to attend a conference this weekend. Please design a one-day tour route for me in Xi'an and its surrounding areas" as an example, the model output results were manually reviewed and it was found that the text generated by the 8-bit quantized pre-trained 5epoch+SFT3epoch model has the following characteristics:

[0105] (A) High density of factual information: The generated text contains a large amount of accurate and relevant factual information, including detailed descriptions of attractions, routes and time planning. Our model recommends five attractions including Da Ci'en Temple, Bell Tower, Beilin Museum, Xi'an City Wall, and Big Wild Goose Pagoda, and gives a detailed and reasonable time arrangement; while the original model recommends four attractions including Huaqing Hot Spring, Qin Shihuang Mausoleum Museum, Shaanxi History Museum, and Big Wild Goose Pagoda without considering their spatial relationship and time arrangement. In fact, the tour route recommended by the original model is difficult to complete in one day.

[0106] (B) Unified output style: The text structure of the model output is highly consistent, with clear paragraph divisions and arranged in timeline; and the language in the dialogue is very concise and capable. For example, our model gave the following answer for the first attraction: "8:00-9:00: Depart from the hotel and head to the Xi'an City Wall (located in the city center), which is the defense facility of the ancient city and one of the landmark buildings of Xi'an. You can take photos at the South Gate Tower Square, then climb up the city wall, walk along the wall, and feel the historical atmosphere of this ancient city."

[0107] (C) Accurate response to user instructions: The model performs well in understanding and processing different user demand tasks, and can understand the user's personalized needs and make reasonable time and route arrangements. The recommended itinerary given by our model is completely within the second ring road of Xi'an, which meets the requirements of business travelers to arrange their time tightly and avoid long-distance travel.

[0108] 1.3) Prompt engineering template design.

[0109] In step 1.3), specifically:

[0110] 1.3.1) Design a special prompt word template for the tourism-specific large model. According to the two core elements of the tourism route map: attractions and days, the needs are divided into three types, and corresponding response requirements are provided for each type:

[0111] (1) Less detailed requirements: The user only specifies the destination; the large model’s response needs to generate a list of attractions near the destination for each day. The recommended attractions should come from the attraction data learned by the model, and the default is a 3-day tour.

[0112] (2) Medium-detailed requirements: The user specifies the destination and the number of days or specific attractions; the large model should plan the route based on the date or attractions requested by the user, and recommend attractions that are not explicitly mentioned by the user but are worth visiting;

[0113] (3) Very detailed requirements: Users specify in detail the destinations and attractions they want to visit each day; the large model must strictly follow the user's requirements to ensure that all attractions are included within the specified time; if the user's list of attractions is not enough to fill the itinerary, highly recommended attractions can be suggested to enrich the user experience while remaining consistent with their overall preferences.

[0114] 1.3.2) The big model collects necessary mapping information based on the planned route and clarifies seven key information: 1) attraction name; 2) attraction coordinates; 3) attraction ticket price; 4) attractions visited every day; 5) visiting time of each attraction; 6) overall route description; 7) transportation method; In order to facilitate the subsequent compilation program to automatically generate the route map, all mapping information will be stored in JSON format.

[0115] 1.3.3) To reduce the illusions and uncertainties that the big model generates when understanding user needs, two strategies are defined and incorporated into the prompt template. The first is to combine examples, covering specific examples of different types of user needs and big model responses, to better demonstrate the differences between user inputs at different levels of detail. The second is to adopt a Reasoning and Acting Framework, which allows the big model to collaboratively reason and execute through an iterative process. The process includes four key stages: (1) Thinking - semantic understanding and reasoning based on user input; (2) Acting - deciding the next action based on previous thinking; (3) Observing - feedback on the results of the action; (4) Thinking again - re-examining the reasoning based on the observation results and the initial goal.

[0116] Step 2: Generate a schematic tourist route map; the route map obtained by directly connecting the scenic spots has the problems of crowded and uneven distribution of scenic spots, the routes are not concise and beautiful, and the name annotations are overlapped. The following methods can be used to solve this problem:

[0117] 2.1) Expand the congested area.

[0118] In step 2.1), specifically:

[0119] The problem of crowded and uneven distribution of scenic spots is solved by adjusting the positions of points to expand the crowded area so as to achieve uniform distribution. Specifically, points are dispersed in dense areas and concentrated in sparse areas. To simulate this process, the Voronoi diagram is used to describe the spatial distribution of points. Figure 2 Shows the Voronoi diagram of a set of points, where points in dense areas correspond to smaller Voronoi cells and points in sparse areas correspond to larger cells.

[0120] In the generated Voronoi diagram, the original scenic spots are updated to the centroids of the corresponding Voronoi cells, such as Figure 3 These updated points are then used as input to generate the next Voronoi diagram. This process is repeated until the distribution of points reaches an acceptable level of uniformity.

[0121] The uniformity level can be measured by geometric information entropy, and the calculation method is as follows:

[0122]

[0123] Among them, S i (i=1,2,…,N) represents map symbol i th The Voronoi region area is ; S represents the area of ​​the entire map space; N is the total number of map symbols. This formula shows that the more uniform the distribution of map symbols, the greater the geometric information entropy.

[0124] In order to compare the geometric information entropy between maps with different numbers of symbols, the maximum entropy should be used for normalization. i When the maximum entropy is reached, the maximum entropy is The normalized geometric information entropy ranges from 0 to 1.

[0125] Based on multiple experimental observations, the threshold of normalized geometric information entropy is set to 0.8. That is, when the geometric information entropy of the Voronoi diagram exceeds 0.8, the iteration process will terminate. Figure 4 shown.

[0126] 2.2) Generate regular circuits.

[0127] In step 2.2), a grid-based method is used to generate regular connection lines to make the lines simple and beautiful, specifically:

[0128] 2.2.1) Reposition the point to the nearest grid corner;

[0129] First determine the appropriate mesh side length. The side length must ensure that the generated mesh contains only one point in each cell. The side length is determined by the following calculation:

[0130]

[0131] In the formula, and are the distances between points i and j in the x and y directions, respectively. Figure 5 As shown, n is the number of points;

[0132] Then, the range of the grid is determined based on the coordinate range of the point set; after the grid is drawn, each point is relocated to the nearest grid corner point.

[0133] 2.2.2) Connect corner points along the mesh edges.

[0134] Connect all point features along the grid edge one by one. The connection order follows the order of visiting scenic spots. For each connection, define three constraints:

[0135] (a) The connecting edge should maintain good continuity;

[0136] (b) Connecting edges cannot pass through other point features;

[0137] (c) A connecting edge cannot overlap with a previously connected edge.

[0138] To satisfy constraint (a), the change in direction should be minimized; this is achieved by identifying the intersection of the horizontal and vertical lines of the points, such as Figure 6As shown; the points are connected along horizontal and vertical lines through these intersections, forming connecting edges with minimal direction changes and shortest distances.

[0139] For constraints (b) and (c), a greedy algorithm and a backtracking algorithm are used to connect point by point and judge edge by edge until the final route meets the constraints. Specifically, the n-order adjacent points are used as the turning points of the route between any two points to avoid the current route passing through other attractions or overlapping with previously existing connecting edges, such as Figure 7 shown.

[0140] 2.3) Placement of name annotations.

[0141] The problem of name annotation overlap can be solved by designing reasonable annotation placement rules. One of the most common rules is "avoid annotation overlap". In tourist route maps, the term "annotation overlap" includes three types of overlap: overlap between annotations, overlap between annotations and attractions, and overlap between annotations and routes.

[0142] Establish the objective function and combine it with the greedy algorithm to optimize the annotation position. The steps are as follows:

[0143] 2.3.1) The initial annotation is placed in the upper right position;

[0144] 2.3.2) Identify the overprint mark;

[0145] 2.3.3) Move the overlapping annotation according to the priority of the eight candidate annotation positions;

[0146] 2.3.4) Repeat step 2 until there are no more stamp marks.

[0147] The generated schematic tourist route map is as follows Figure 8 shown.

[0148] Step 3: Personalized symbol design and map finishing.

[0149] Customizing the symbol style of scenic spots according to their characteristics is one of the key steps in optimizing tourist route maps. At present, with the maturity of large model technology (such as ChatGPT-4o), it has the ability to generate image results, so these large models can be used to automatically generate symbolic and schematic tourist route maps.

[0150] The details are as follows:

[0151] 3.1) Use the open source big model to generate corresponding symbols according to the characteristics of different scenic spots; these symbols can intuitively reflect the type, scale or characteristics of the scenic spots, and enhance the readability and practicality of the map.

[0152] 3.2) The generated symbols are combined with the actual tourist route to generate a route map including daily itinerary, scenic spot sequence, and transportation information; the large model can not only generate scenic spot symbols, but also generate a route map based on the input information. This step needs to be combined with the specific needs of the user. These route diagrams can clearly show the itinerary for each day and provide tourists with an intuitive route reference.

[0153] 3.3) The map is retouched to ensure the overall aesthetics and practicality of the map, including adjusting the position of text annotations, optimizing the layout of scenic spot symbols, adding detailed route descriptions and annotations, and ensuring the integrity and comprehensibility of map information. The generation of a daily itinerary map requires displaying the main itinerary of each day in sequence, including scenic spots to be visited, transportation routes, estimated time, etc., to ensure that the travel plan is clear and orderly.

[0154] 3.4) Combine the above steps to generate a travel route map, such as Fig. 9 shown.

[0155] This automated process of the present embodiment not only improves the accuracy and personalization of the route map, but also greatly improves the efficiency and aesthetics of mapping, providing users with a better travel planning experience.

[0156] The present invention and its embodiments are described schematically above, and the description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by it and designs a structural method and an embodiment similar to the technical solution without creativity without departing from the purpose of the invention, they shall all fall within the protection scope of the present invention.

Claims

1. A method for automatically generating a schematic tourist route map driven by a large model, characterized in that: The following steps are involved: Step 1: Fine-tune the open source big model for understanding tourist needs; 1.1) Incremental pre-training; 1.2) supervised fine-tuning; 1.3) Prompt engineering template design; Step 2: Generate a schematic travel route map; 2.1) Expand the congested area; 2.2) Generate regular circuits; 2.3) Name and annotation placement; Step 3: Personalized symbol design and picture decoration; 3.1) Use the open source big model to generate corresponding symbols according to the characteristics of different scenic spots; 3.2) The generated symbols are combined with the actual travel route to generate a route map including daily itinerary, sequence of attractions, and transportation information; 3.3) Renovate the map, including adjusting the position of text annotations, optimizing the layout of scenic spot symbols, and adding detailed route descriptions and annotations; 3.4) Combine the above steps to generate a travel route map.

2. The method for automatically generating a schematic tourist route map driven by a large model according to claim 1 is characterized in that: In step 1.1), specifically: 1.1.1) Compile Python crawler scripts and use the BeautifulSoup library to efficiently crawl multi-dimensional data related to attractions from tourism websites, including attraction names, geographic locations, ratings, user reviews, ticket prices, and opening hours; 1.1.2) Extract key information from the multi-dimensional data obtained by the crawler script, including city, attraction name, location, address and introduction; 1.1.3) Eliminate attractions with a score below 2.5 and no reviews to ensure data quality; 1.1.4) Combine key information into descriptive text to present scenic spot information; 1.1.5) Output the descriptive text in a structured JSON format for use in the construction of a standardized dataset in subsequent pre-training; 1.1.6) Training large models to build high-quality datasets; The model predicts the next word based on the existing word sequence, and after each new word is generated, it is included in the word sequence and recursively makes the next prediction; After each prediction, the model's prediction result will be compared with the actual next word to calculate the loss function; through the back propagation algorithm, the calculated loss is used to update the model's parameters, so that the model can gradually improve its prediction accuracy.

3. The method for automatically generating a schematic tourist route map driven by a large model according to claim 2 is characterized in that: In step 1.2), specifically: 1.2.1) Obtain tourism-related question and answer data through crawler technology; 1.2.2) Manual data cleaning, including removing questions and answers that do not meet the needs of personalized travel maps and modifying and improving incomplete questions; 1.2.3) Improve the question-answering data, add the step of analyzing key information from the question and then thinking about the next step, that is, use the thinking chain CoT method to build a large instruction training model, supplement the extraction of necessary conditions in the question, analyze the tourism demand conditions and use this as a basis to recommend tourist attractions for tourism route planning; 1.2.4) Generate more question-answering data by artificial construction and self-generation; construct a question-answering dataset according to the question-analysis-answer framework, and finally establish supervised fine-tuning instruction data and convert it into JSON format; 1.2.5) Based on the incremental pre-trained large model, the instruction data constructed is further supervised and fine-tuned to generate a dedicated large model suitable for the tourism field. During the fine-tuning process, the model will use the instruction data as a reference value to calculate the loss function and adjust the parameters; 1.2.6) BLEU-4, ROUGE-L and ROUGE-2 are used as evaluation indicators for evaluation.

4. The method for automatically generating a schematic tourist route map driven by a large model according to claim 3 is characterized in that: In step 1.3), specifically: 1.3.1) Based on the two core elements of the travel route map: attractions and days, the needs are divided into three types, and corresponding response requirements are provided for each type: (1) Less detailed requirements: The user only specifies the destination; the large model’s response needs to generate a list of attractions near the destination for each day. The recommended attractions should come from the attraction data learned by the model, and the default is a 3-day tour. (2) Medium-detailed requirements: The user specifies the destination and the number of days or specific attractions; the large model should plan the route based on the date or attractions requested by the user, and recommend attractions that are not explicitly mentioned by the user but are worth visiting; (3) Very detailed requirements: Users specify the destinations and attractions they want to visit every day in detail; the large model must strictly follow the user's requirements to ensure that all attractions are included within the specified time; if the user's list of attractions is not enough to fill the itinerary, it can suggest highly recommended attractions to enrich the user experience while maintaining consistency with their overall preferences; 1.3.2) The large model collects necessary mapping information based on the planned route and clarifies seven key information: 1) attraction name; 2) attraction coordinates; 3) attraction ticket price; 4) attractions visited every day; 5) visiting time of each attraction; 6) overall route description; 7) transportation method; 1.3.3) Two strategies are defined and incorporated into the prompt template; the first is to combine instances, covering specific examples of different types of user needs and big model responses, to better demonstrate the differences between user inputs at different levels of detail; the second is to adopt a reasoning-and-action framework that allows the big model to collaboratively reason and execute through an iterative process.

5. The method for automatically generating a schematic tourist route map driven by a large model according to claim 4 is characterized in that: In step 2.1), specifically: The problem of crowded and uneven distribution of scenic spots is solved by adjusting the positions of points to expand the crowded area so as to achieve uniform distribution. Specifically, points are dispersed in dense areas and concentrated in sparse areas. To simulate this process, the Voronoi diagram is used to describe the spatial distribution of points.

6. The method for automatically generating a schematic tourist route map driven by a large model according to claim 5 is characterized in that: In step 2.2), a grid-based method is used to generate regular connection lines, specifically: 2.2.1) Reposition the point to the nearest grid corner; First determine the appropriate edge length of the grid. The edge length must ensure that the generated grid contains only one point in each cell. The edge length is determined by the following calculation: In the formula, and are the distances between points i and j in the x and y directions, respectively, and n is the number of points; Then, based on the coordinate range of the point set, the range of the grid is determined; After the grid is drawn, each point is repositioned to the nearest grid corner; 2.2.2) Connect corner points along the grid edges; Connect all point features along the grid edge one by one. The connection order follows the order of visiting scenic spots. For each connection, define three constraints: (a) The connecting edge should maintain good continuity; (b) Connecting edges cannot pass through other point features; (c) A connected edge cannot overlap with a previously connected edge; To satisfy constraint (a), the change in direction should be minimized; this is achieved by identifying the intersection of the horizontal and vertical lines of the point; The points are connected along horizontal and vertical lines through these intersections, forming connecting edges with minimal direction changes and shortest distances; For constraints (b) and (c), a greedy algorithm and a backtracking algorithm are used to connect point by point and judge edge by edge until the final route meets the constraints. Specifically, the n-th order adjacent point is used as the turning point of the route between any two points to avoid the current route passing through other attractions or overlapping with previously existing connection edges.

7. The method for automatically generating a schematic tourist route map driven by a large model according to claim 6 is characterized in that: In step 2.3), the problem of name annotation overlap is solved by designing annotation placement rules. The steps are as follows: 2.3.1) The initial annotation is placed in the upper right position; 2.3.2) Identify the overprint mark; 2.3.3) Move the overlapping annotation according to the priority of the eight candidate annotation positions; 2.3.4) Repeat step 2 until there are no more stamp marks.

Images