Training method and device of tourism intelligent question and answer large model, medium and equipment

Abstract

The invention provides a training method and device for a tourism intelligent question and answer large model, a medium and equipment. The method comprises the following steps: acquiring tourist related data, extracting tourist behavior characteristics from the tourist related data, and constructing a corresponding tourist portrait; collecting tourism resource data, classifying tourism places, mining characteristics of each tourism place, and adding the tourism resource data, classification and characteristics of each tourism place into a tourism resource knowledge base as a piece of tourism resource knowledge; performing environmental factor evaluation on each tourist site to obtain an environmental factor evaluation result of each tourist site; determining an association relationship among the tourist portrait, the tourism resource knowledge and the environmental factor evaluation result; and obtaining historical question and answer pairs, and performing model training according to the historical question and answer pairs and the association relationship to obtain a convergent tourism intelligent question and answer large model. The trained large model can bring more convenient and personalized service experience to travelers, and meanwhile, the operation efficiency and customer satisfaction of tourism enterprises are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent question-answering technology, and in particular to a training method, device, medium and equipment for a large tourism intelligent question-answering model. Background Art

[0002] With the rapid development of the tourism industry, customers' demands for tourism services are becoming increasingly diversified, and the requirements for the real-time, intelligent and personalized customer service systems are becoming higher and higher. The traditional way of receiving tourists mainly relies on manual customer service, which has problems such as slow response speed, low service efficiency and high labor costs. In order to solve these problems, the tourism industry has begun to introduce intelligent customer service systems to improve service quality and efficiency.

[0003] In the tourism industry, some intelligent customer service systems are based on traditional artificial intelligence technologies, such as natural language processing and speech recognition. However, these technologies sometimes have certain limitations when dealing with complex and changeable tourism consultation issues. Other intelligent customer service systems use fixed question and answer templates and lack personalization. Summary of the invention

[0004] In response to at least one of the above technical problems, an embodiment of the present invention provides a training method, device, medium, and equipment for a large tourism intelligent question-answering model.

[0005] According to the first aspect, the training method of the tourism intelligent question-answering model provided by the embodiment of the present invention includes:

[0006] Acquire tourist-related data, extract tourist behavior characteristics from the tourist-related data, and construct corresponding tourist portraits based on the tourist behavior characteristics;

[0007] Collecting tourism resource data, classifying tourist destinations according to the tourism resource data and exploring the characteristics of each tourist destination, and adding the tourism resource data, classification and characteristics of each tourist destination as a piece of tourism resource knowledge to a tourism resource knowledge base;

[0008] Use machine learning models to evaluate the environmental factors of each tourist destination and obtain the environmental factor evaluation results of each tourist destination;

[0009] Determining the correlation between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result;

[0010] Obtain historical question-answer pairs, perform model training based on the historical question-answer pairs and the association relationship, and obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

[0011] In one embodiment, the acquiring of tourist-related data, extracting tourist behavior characteristics from the tourist-related data, and constructing a corresponding tourist portrait according to the tourist behavior characteristics includes:

[0012] Collecting tourist-related data corresponding to multiple tourists from multiple channels; wherein the tourist-related data includes at least one of basic information of tourists, travel activity information and interactive communication records;

[0013] Preprocessing the visitor-related data; wherein the preprocessing method includes at least one of deduplication, filling missing values, data format conversion processing and standardization processing;

[0014] Extracting the tourist behavior characteristics from the pre-processed tourist-related data; wherein the tourist behavior characteristics include at least one of the tourist's stay time, browsing path, consumption level and interaction frequency;

[0015] According to the tourist behavior characteristics, a corresponding tourist portrait is constructed.

[0016] In one embodiment, the collecting of tourism resource data, classifying tourism destinations according to the tourism resource data and mining the characteristics of each tourism destination, and adding the tourism resource data, classification and characteristics of each tourism destination as a piece of tourism resource knowledge to a tourism resource knowledge base include:

[0017] Collect tourism resource data;

[0018] Cleaning the tourism resource data; wherein the cleaning process includes at least one of deduplication, correction of erroneous information and filling of missing values;

[0019] Unify the formats and structures of tourism resource data in different formats and structures;

[0020] Classifying the tourism resource data, and setting classification labels for corresponding tourist destinations according to the classification results;

[0021] Use association rule mining methods, cluster analysis methods or sentiment analysis methods to explore the characteristics of tourist destinations;

[0022] Adding the tourism resource data, classification labels and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base;

[0023] The tourism resource knowledge in the tourism resource knowledge base is updated regularly.

[0024] In one embodiment, the environmental factors include climate conditions, ecological protection status, safety risk level, and tourism settings and environmental carrying capacity.

[0025] In one embodiment, the environmental factor assessment of each tourist destination is performed using a machine learning model to obtain the environmental factor assessment results of each tourist destination, including:

[0026] Collect data on environmental factors at various tourist destinations;

[0027] Preprocessing the environmental factor data;

[0028] Training an environmental factor assessment model based on the preprocessed environmental factor data; wherein the environmental factor assessment model is constructed based on a machine learning algorithm;

[0029] The environmental factor assessment model is used to assess the environmental factors of each tourist destination to obtain the environmental factor assessment results of each tourist destination; wherein the environmental factor assessment results include the environmental quality grade and the suitable travel time.

[0030] In one embodiment, determining the association relationship between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result includes:

[0031] Use big data processing technology to integrate multiple tourist portraits, tourism resource knowledge of each tourist destination, and environmental factor assessment results of each tourist destination;

[0032] Machine learning technology is used to mine the correlation between the integrated processed tourist portrait, the tourism resource knowledge and the environmental factor evaluation results.

[0033] In one embodiment, the method further comprises:

[0034] Performing performance evaluation on the tourism intelligent question-answering model to obtain evaluation indicators; wherein the evaluation indicators include at least one of accuracy, recall rate and F1 value;

[0035] According to the evaluation indicators, the tourism intelligent question and answer model is optimized.

[0036] According to the second aspect, the training device of the tourism intelligent question-answering large model provided by the embodiment of the present invention includes:

[0037] A portrait construction module is used to obtain tourist-related data, extract tourist behavior characteristics from the tourist-related data, and construct a corresponding tourist portrait based on the tourist behavior characteristics;

[0038] The knowledge construction module is used to collect tourism resource data, classify tourist destinations according to the tourism resource data, and explore the characteristics of each tourist destination, and add the tourism resource data, classification and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base;

[0039] The factor evaluation module is used to evaluate the environmental factors of each tourist destination using a machine learning model to obtain the environmental factor evaluation results of each tourist destination;

[0040] An association determination module, used to determine the association relationship between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result;

[0041] The model training module is used to obtain historical question-answer pairs, and perform model training based on the historical question-answer pairs and the association relationship to obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

[0042] According to a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, enables the computer to execute the method provided in the first aspect.

[0043] According to a fourth aspect, an embodiment of the present invention provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method provided in the first aspect is implemented.

[0044] The training method, device, medium, and equipment of the tourism intelligent question-answering big model provided by the embodiment of the present invention first constructs a tourist portrait, a tourism resource knowledge base, and an environmental factor evaluation, and then mines the association between the three, and then uses the historical question-answer pairs and the association relationship to perform model training to obtain a tourism intelligent question-answering big model. Based on the big model trained by the embodiment of the present invention, when a user asks a question, the user question information is input into the tourism intelligent question-answering big model, and the big model extracts the user portrait from the user question information, and then matches it among the large number of tourist portraits that have been constructed. After matching the most similar tourist portrait, the tourist portrait, the tourist resource knowledge of various places, and the environmental factor evaluation results of various places are used to determine the most suitable tourist destination for the user to travel and the relevant knowledge of the tourist destination. Generate an answer based on the determination result. It can be seen that in the above dialogue, the tourism intelligent question-answering big model takes into account the tourist behavior, tourism resource information, and environmental factors, and can provide personalized, accurate and appropriate answers for tourists. The application of the big model can also greatly improve the processing speed and concurrency of the customer service system, ensuring that efficient and stable services can be maintained under large-scale tourist consultation conditions. In addition, the big model can also learn and optimize the professional knowledge of the tourism industry on a large scale, reduce the dependence on manual customer service, reduce labor costs, and improve the overall quality and satisfaction of tourism services. In short, the big model trained by the embodiment of the present invention can bring more convenient and personalized service experience to passengers, while improving the operational efficiency and customer satisfaction of tourism enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flowchart of a training method for a tourism intelligent question-answering model in one embodiment of the present invention;

[0046] Figure 2 The present invention is a structural block diagram of a training device for a large tourism intelligent question-answering model in one embodiment of the present invention. DETAILED DESCRIPTION

[0047] In the first aspect, the present invention provides a method for training a large tourism intelligent question-answering model. Figure 1 The method comprises the following steps S110 to S150:

[0048] S110, obtaining tourist-related data, extracting tourist behavior characteristics from the tourist-related data, and constructing a corresponding tourist portrait based on the tourist behavior characteristics;

[0049] In one embodiment, the step of obtaining tourist-related data in S110, extracting tourist behavior characteristics from the tourist-related data, and constructing a corresponding tourist portrait according to the tourist behavior characteristics may include S111 to S114:

[0050] S111, collecting tourist-related data corresponding to multiple tourists from multiple channels; wherein the tourist-related data includes at least one of basic information of tourists, travel activity information and interactive communication records;

[0051] Specifically, relevant data of tourists are collected from multiple channels, such as reservation systems, scenic spot ticket systems, tourism service platforms, social media, etc. These data include basic information of tourists, such as age, gender, region, etc., and can also include tourism activity information, such as travel methods, tourist attractions, consumption behavior, etc., and can also include interactive communication records.

[0052] S112, preprocessing the visitor-related data; wherein the preprocessing method includes at least one of deduplication, filling missing values, data format conversion processing and standardization processing;

[0053] Among them, the collected tourist-related data undergoes operations such as removing duplicate data, filling missing values, converting data formats, and standardizing data. The goal of data preprocessing is to improve the quality of data and prepare for tourist behavior analysis.

[0054] S113, extracting the tourist behavior characteristics from the pre-processed tourist-related data; wherein the tourist behavior characteristics include at least one of the tourist's stay time, browsing path, consumption level and interaction frequency;

[0055] Specifically, based on the characteristics of the tourism industry and the laws of tourist behavior, key features that are helpful for tourist behavior analysis are extracted from the pre-processed tourist-related data to ensure that the extracted features are representative and discriminative. These features include tourists' stay time, tour routes, consumption levels, interaction frequency, etc.

[0056] S114. Construct a corresponding tourist portrait based on the tourist behavior characteristics.

[0057] Specifically, the tourist behavior characteristics are analyzed according to the tourist dimension, including identifying whether the tourist type is a first-time tourist or a returning tourist, predicting tourist satisfaction, analyzing the reasons for tourist loss, etc., in order to establish a user portrait.

[0058] Understandably, through the above functions, tourism enterprises can better understand tourists’ needs and develop targeted marketing strategies and measures to improve tourists’ satisfaction for different individuals.

[0059] S120, collecting tourism resource data, classifying tourism destinations according to the tourism resource data and exploring the characteristics of each tourism destination, and adding the tourism resource data, classification and characteristics of each tourism destination as a piece of tourism resource knowledge to a tourism resource knowledge base;

[0060] In one embodiment, the collecting of tourism resource data in S120, classifying tourism destinations according to the tourism resource data and mining the characteristics of each tourism destination, and adding the tourism resource data, classification and characteristics of each tourism destination as a piece of tourism resource knowledge to a tourism resource knowledge base may include S121 to S127:

[0061] S121, collecting tourism resource data;

[0062] In order to ensure the accuracy and completeness of tourism resource data, various channels can be used, such as cooperating with cultural and tourism departments to obtain official data, crawling information on tourism websites, and obtaining it through official accounts of tourist attractions.

[0063] S122, cleaning the tourism resource data; wherein the cleaning method includes at least one of deduplication, correction of erroneous information and filling of missing values;

[0064] S123, unifying the formats and structures of tourism resource data in different formats and structures;

[0065] It is understandable that the formats and structures of tourism resource data from different sources may be different, so it is necessary to unify the formats and structures.

[0066] S124, classifying the tourism resource data, and setting classification labels for corresponding tourist destinations according to the classification results;

[0067] It is understandable that classifying tourism resource data and setting corresponding classification tags facilitates rapid retrieval of relevant information according to user needs.

[0068] S125. Using association rule mining methods, cluster analysis methods or sentiment analysis methods to mine the characteristics of tourist destinations;

[0069] Specifically, through association rule mining methods, cluster analysis methods, sentiment analysis methods and even topic models, tourism resource information can be analyzed and mined to discover the characteristics of tourism resources and even the advantages and potential of tourism resources.

[0070] S126, adding the tourism resource data, classification labels and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base;

[0071] It can be seen that the integration of tourism resource data, classification labels, characteristics and other information has formed a unified and comprehensive tourism resource knowledge base.

[0072] S127. Regularly update the tourism resource knowledge in the tourism resource knowledge base.

[0073] It can be seen that tourism resource information needs to be updated regularly to reflect changes in tourism resources. Tourism resource information updates can be achieved by setting up crawlers to regularly crawl tourism website information, monitoring tourist feedback and evaluation, etc. At the same time, tourism resource information is continuously optimized based on tourists' inquiries and feedback.

[0074] S130, using a machine learning model to evaluate the environmental factors of each tourist destination, and obtaining the environmental factor evaluation results of each tourist destination;

[0075] Among them, the environmental factors may include climate conditions, ecological protection status, safety risk level, and tourism settings and environmental carrying capacity.

[0076] In one embodiment, the use of a machine learning model to evaluate the environmental factors of each tourist destination in S130 to obtain the environmental factor evaluation results of each tourist destination may include:

[0077] S131. Collect environmental factor data of various tourist destinations;

[0078] S132, preprocessing the environmental factor data;

[0079] It can be seen that data on the climate conditions, ecological protection status, safety risk level, tourism facilities and environmental carrying capacity of the tourist destination are collected, and then the collected data are pre-processed by sorting and cleaning.

[0080] S133. Training an environmental factor evaluation model according to the preprocessed environmental factor data; wherein the environmental factor evaluation model is constructed based on a machine learning algorithm;

[0081] Specifically, based on the pre-processed environmental factor data, the environmental factor evaluation model is constructed using machine learning algorithms such as regression analysis and support vector machine. When constructing the model, appropriate evaluation indicators can be selected in combination with the characteristics and needs of the tourism industry.

[0082] S134. Evaluate the environmental factors of each tourist destination using the environmental factor evaluation model to obtain an environmental factor evaluation result for each tourist destination; wherein the environmental factor evaluation result includes an environmental quality grade and a suitable travel time.

[0083] It can be seen that the constructed environmental factor assessment model can be used to assess the environmental factors of tourist destinations and obtain environmental factor assessment results, which include the environmental quality level of the tourist destination, the time suitable for tourism, etc.

[0084] Of course, the environmental factor assessment model can be continuously adjusted and optimized according to environmental changes and tourists' feedback to improve the accuracy and practicality of the assessment. A comprehensive assessment of the environmental factors of tourist destinations is conducive to improving tourists' travel experience.

[0085] S140, determining the correlation relationship between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result;

[0086] In one embodiment, determining the association relationship between the tourist portrait, the tourism resource knowledge, and the environmental factor assessment result in S140 may include:

[0087] S141. Use big data processing technology to integrate multiple tourist portraits, tourism resource knowledge of various tourist destinations, and environmental factor assessment results of various tourist destinations;

[0088] S142. Using machine learning technology to mine the correlation between the integrated tourist portrait, the tourism resource knowledge and the environmental factor assessment results.

[0089] In the above S110-S130, tourist portraits, tourism resource knowledge, and environmental factor evaluation results of each tourist destination can be obtained. The above three types of data are accessed through the API interface and integrated. For example, one piece of integrated data is the tourism resource knowledge of a tourist destination, the environmental factor evaluation results of the tourist destination, and the tourist portrait corresponding to the tourists who have visited the tourist destination. In order to improve the efficiency and accuracy of data processing, the integration process can adopt big data processing technology, such as data processing based on distributed computing frameworks such as Hadoop and Spark. Then, use machine learning techniques, such as cluster analysis and association rule mining, to discover the hidden rules and associations between the three types of data, providing strong support for subsequent model training and optimization.

[0090] S150, obtaining historical question-answer pairs, and performing model training based on the historical question-answer pairs and the association relationship to obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

[0091] Among them, a historical question-answer pair refers to a user question and a corresponding answer in a historical conversation. The association between a historical question-answer pair and the tourism resource knowledge of the tourist destination involved in the historical question-answer pair, the environmental factor assessment results of the tourist destination, and the corresponding tourist portrait of the tourists who have visited the tourist destination is taken as a training sample, and multiple training samples are used for model training to obtain a large tourism intelligent question-answering model.

[0092] Among them, a pre-trained large model is selected as the base model, such as Qwen-72B. Then, the base model is fine-tuned according to the business needs and data characteristics of intelligent customer service in the tourism industry. Fine-tuning can be to modify certain parameters of the model or to add some specific functional modules. For example, an attention mechanism module can be added to the model to improve the model's ability to recognize key information.

[0093] In one embodiment, the method may further include:

[0094] S160, performing performance evaluation on the tourism intelligent question-answering model to obtain evaluation indicators; wherein the evaluation indicators include at least one of accuracy, recall rate and F1 value;

[0095] S170: Optimize the tourism intelligent question-answering model according to the evaluation index.

[0096] After training is completed, the performance of the model is evaluated, and the evaluation indicators include accuracy, recall, F1 value, etc. According to the results of the model evaluation, the model can be further optimized. The optimization method can be to adjust the training parameters, such as learning rate, batch size, etc., or to fine-tune the model again to improve the performance of the model. After the optimization is completed, the model performance reaches the expected goal and the model is deployed to the production environment. At the same time, the model needs to be updated and optimized when the data is updated to keep the model synchronized with the data.

[0097] In the application scenario, when a user asks a question, such as where to travel, the atomic capabilities of the big model, such as text classification, entity recognition, and relationship extraction, are used to deeply understand and analyze the question and determine the problem points that the user is concerned about. The understood user question information is then input into the tourism intelligent question-answering big model. The big model extracts the user portrait from the user question information and then matches it with the large number of tourist portraits that have been constructed in S110. For example,

[0098] Use cosine similarity for matching. Of course, you can also use BM25 algorithm, ES search and other methods to match. After matching the most similar tourist portrait, determine the most suitable tourist destination for the user and the relevant knowledge of the tourist destination, such as characteristics, based on the relationship between the tourist portrait, the knowledge of tourist resources in various places, and the evaluation results of environmental factors in various places. Generate an answer based on the determination result. It can be seen that in the above dialogue, the tourism intelligent question and answer model takes into account tourist behavior, tourism resource information, and environmental factors, and can provide tourists with personalized, accurate and appropriate answers.

[0099] In actual scenarios, the entire conversation process can also be monitored and evaluated, and the tourism intelligent question-answering model can be optimized to improve customer service quality. This module mainly focuses on the following aspects:

[0100] (1) Customer service conversation quality assessment. Conduct in-depth analysis of the conversation content between customer service and users, including: using language models to assess the fluency of the conversation; using entity recognition and relationship extraction technology to assess the accuracy of the conversation content; introducing conversation timing analysis to assess the timeliness of customer service responses.

[0101] (2) Customer service satisfaction evaluation. By collecting tourists’ feedback and evaluation in the intelligent customer service system, we can understand the quality and effectiveness of the answers and adjust and improve the model based on this information. For example, we can use some supervised learning techniques, such as classification or regression analysis, to analyze tourists’ feedback and satisfaction.

[0102] (3) Continuous adjustment and optimization. Based on the above evaluation results, the large model is continuously optimized, including: using data mining and machine learning technologies to mine high-quality conversation data for model training; using transfer learning technology to draw on intelligent customer service cases in other fields to improve the generalization ability of the customer service system; regularly evaluating model performance and adjusting model parameters in a targeted manner to continuously improve the effectiveness of the model's answers.

[0103] The embodiments of the present invention have the following beneficial effects:

[0104] (1) Improve the service efficiency and quality of customer service in the tourism industry. Intelligent customer service can assist manual customer service, reduce the workload of manual customer service, and improve service efficiency. By comprehensively considering tourist behavior, tourism resource information, and environmental factors, intelligent customer service can fully understand tourist needs, tourism resource characteristics, and the environmental conditions of tourist destinations, and provide tourists with more accurate and personalized travel advice and information, thereby improving service quality.

[0105] (2) Improve the utilization rate of tourism resources. By evaluating the characteristics, advantages and potential of tourism resources, it is helpful to rationally develop and utilize tourism resources and improve the attractiveness and competitiveness of tourist destinations.

[0106] (3) Reduce the loss rate of tourists. Through the results of customer service satisfaction evaluation, we can analyze the reasons for the loss of tourists, timely discover and solve potential problems, improve tourism products and services in a targeted manner, and reduce the loss rate of tourists.

[0107] (4) Continuously improve service levels. Customer service effectiveness monitoring and optimization can monitor the performance of the intelligent customer service system, identify problems in a timely manner and make adjustments. At the same time, it can also better adapt to the development of the tourism industry and changes in tourist needs, ensuring the stable operation and continuous improvement of the customer service system.

[0108] It can be seen that the embodiment of the present invention can provide more personalized and accurate tourism consulting services by using big model technology. At the same time, the application of big models can also greatly improve the processing speed and concurrency of the customer service system, ensuring that efficient and stable services can be maintained under large-scale tourist consultation conditions. In addition, big models can also learn and optimize the professional knowledge of the tourism industry on a large scale, reduce dependence on manual customer service, reduce labor costs, and improve the overall quality and satisfaction of tourism services. In short, the big model trained by the embodiment of the present invention can bring more convenient and personalized service experience to passengers, while improving the operational efficiency and customer satisfaction of tourism enterprises.

[0109] In a second aspect, an embodiment of the present invention provides a training device for a large tourism intelligent question-answering model, see Figure 2 , the device 100 comprises:

[0110] A portrait construction module 110 is used to obtain tourist-related data, extract tourist behavior characteristics from the tourist-related data, and construct a corresponding tourist portrait based on the tourist behavior characteristics;

[0111] The knowledge construction module 120 is used to collect tourism resource data, classify tourist destinations according to the tourism resource data, and explore the characteristics of each tourist destination, and add the tourism resource data, classification and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base;

[0112] A factor evaluation module 130 is used to evaluate the environmental factors of each tourist destination using a machine learning model to obtain an environmental factor evaluation result of each tourist destination;

[0113] An association determination module 140 is used to determine the association relationship between the tourist portrait, the tourism resource knowledge and the environmental factor evaluation result;

[0114] The model training module 150 is used to obtain historical question-answer pairs, and perform model training based on the historical question-answer pairs and the association relationship to obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

[0115] In one embodiment, the portrait construction module is specifically used to: collect tourist-related data corresponding to multiple tourists from multiple channels; wherein the tourist-related data includes at least one of tourist basic information, travel activity information and interactive communication records; pre-process the tourist-related data; wherein the pre-processing method includes at least one of deduplication, filling missing values, data format conversion processing and standardization processing; extract the tourist behavior characteristics from the pre-processed tourist-related data; wherein the tourist behavior characteristics include at least one of tourist stay time, browsing path, consumption level and interaction frequency; and construct a corresponding tourist portrait based on the tourist behavior characteristics.

[0116] In one embodiment, the knowledge construction module is specifically used to: collect tourism resource data; clean the tourism resource data; wherein the cleaning method includes at least one of deduplication, correction of erroneous information and filling of missing values; unify the format and structure of tourism resource data of different formats and structures; classify the tourism resource data and set classification labels for corresponding tourist destinations according to the classification results; use association rule mining methods, clustering analysis methods or sentiment analysis methods to mine the characteristics of tourist destinations; add the tourism resource data, classification labels and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base; and regularly update the tourism resource knowledge in the tourism resource knowledge base.

[0117] In one embodiment, the environmental factors include climate conditions, ecological protection status, safety risk level, and tourism settings and environmental carrying capacity.

[0118] In one embodiment, the factor evaluation module is specifically used to: collect environmental factor data of each tourist destination; pre-process the environmental factor data; train an environmental factor evaluation model based on the pre-processed environmental factor data; wherein the environmental factor evaluation model is constructed based on a machine learning algorithm; use the environmental factor evaluation model to evaluate the environmental factors of each tourist destination to obtain environmental factor evaluation results for each tourist destination; wherein the environmental factor evaluation results include environmental quality levels and suitable travel times.

[0119] In one embodiment, the association determination module is specifically used to: use big data processing technology to integrate multiple tourist portraits, tourism resource knowledge of each tourist destination, and environmental factor assessment results of each tourist destination; use machine learning technology to mine the association relationship between the integrated tourist portraits, the tourism resource knowledge and the environmental factor assessment results.

[0120] In one embodiment, the apparatus further comprises:

[0121] The model evaluation module is used to perform performance evaluation on the tourism intelligent question-answering model to obtain evaluation indicators; wherein the evaluation indicators include at least one of accuracy, recall rate and F1 value; and the tourism intelligent question-answering model is tuned according to the evaluation indicators.

[0122] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0123] In a third aspect, an embodiment of the present invention provides a computer-readable medium, on which computer instructions are stored. When the computer instructions are executed by a processor, the processor executes the method provided in the first aspect.

[0124] Specifically, a system or device equipped with a storage medium can be provided, on which software program code that implements the functions of any of the above-mentioned embodiments is stored, and a computer (or CPU or MPU) of the system or device can be enabled to read and execute the program code stored in the storage medium.

[0125] In this case, the program code itself read from the storage medium can realize the function of any one of the above-mentioned embodiments, and thus the program code and the storage medium storing the program code constitute a part of the present invention.

[0126] The storage medium embodiments for providing the program code include a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer by a communication network.

[0127] In addition, it should be clear that the functions of any of the above embodiments can be implemented not only by executing the program code read by the computer, but also by enabling an operating system operating on the computer to complete part or all of the actual operations based on instructions from the program code.

[0128] In addition, it can be understood that the program code read from the storage medium is written to a memory provided in an expansion board inserted into the computer or to a memory provided in an expansion module connected to the computer, and then based on the instructions of the program code, a CPU installed on the expansion board or expansion module is enabled to perform part or all of the actual operations, thereby realizing the functions of any of the above-mentioned embodiments.

[0129] It is understandable that the explanation, specific implementation methods, beneficial effects, examples, etc. of the relevant contents in the computer-readable medium provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0130] In a fourth aspect, an embodiment of the present specification provides a computing device, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, it implements the method in any embodiment of the specification.

[0131] It is understandable that the explanation, specific implementation, beneficial effects, examples, etc. of the relevant contents in the computing device provided in the embodiment of the present invention can be found in the corresponding parts of the method provided in the first aspect, and will not be repeated here.

[0132] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0133] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, widgets, or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0134] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method 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 on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A training method for a large tourism intelligent question-answering model, characterized in that: include: Acquire tourist-related data, extract tourist behavior characteristics from the tourist-related data, and construct corresponding tourist portraits based on the tourist behavior characteristics; Collecting tourism resource data, classifying tourist destinations according to the tourism resource data and exploring the characteristics of each tourist destination, and adding the tourism resource data, classification and characteristics of each tourist destination as a piece of tourism resource knowledge to a tourism resource knowledge base; Use machine learning models to evaluate the environmental factors of each tourist destination and obtain the environmental factor evaluation results of each tourist destination; Determining the correlation between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result; Obtain historical question-answer pairs, perform model training based on the historical question-answer pairs and the association relationship, and obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

2. The method according to claim 1, characterized in that: The step of obtaining tourist-related data, extracting tourist behavior characteristics from the tourist-related data, and constructing a corresponding tourist portrait according to the tourist behavior characteristics includes: Collecting tourist-related data corresponding to multiple tourists from multiple channels; wherein the tourist-related data includes at least one of basic information of tourists, travel activity information and interactive communication records; Preprocessing the visitor-related data; wherein the preprocessing method includes at least one of deduplication, filling missing values, data format conversion processing and standardization processing; Extracting the tourist behavior characteristics from the pre-processed tourist-related data; wherein the tourist behavior characteristics include at least one of the tourist's stay time, browsing path, consumption level and interaction frequency; According to the tourist behavior characteristics, a corresponding tourist portrait is constructed.

3. The method according to claim 1, characterized in that The collecting of tourism resource data, classifying the tourism destinations according to the tourism resource data and exploring the characteristics of each tourism destination, and adding the tourism resource data, classification and characteristics of each tourism destination as a piece of tourism resource knowledge to the tourism resource knowledge base, includes: Collect tourism resource data; Cleaning the tourism resource data; wherein the cleaning process includes at least one of deduplication, correction of erroneous information and filling of missing values; Unify the formats and structures of tourism resource data in different formats and structures; Classifying the tourism resource data, and setting classification labels for corresponding tourist destinations according to the classification results; Use association rule mining methods, cluster analysis methods or sentiment analysis methods to explore the characteristics of tourist destinations; Adding the tourism resource data, classification labels and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base; The tourism resource knowledge in the tourism resource knowledge base is updated regularly.

4. The method according to claim 1, characterized in that: The environmental factors include climate conditions, ecological protection status, safety risk level, and tourism settings and environmental carrying capacity.

5. The method according to claim 4, characterized in that The machine learning model is used to evaluate the environmental factors of each tourist destination, and the environmental factor evaluation results of each tourist destination are obtained, including: Collect data on environmental factors at various tourist destinations; Preprocessing the environmental factor data; Training an environmental factor assessment model based on the preprocessed environmental factor data; wherein the environmental factor assessment model is constructed based on a machine learning algorithm; The environmental factor assessment model is used to assess the environmental factors of each tourist destination to obtain the environmental factor assessment results of each tourist destination; wherein the environmental factor assessment results include the environmental quality grade and the suitable travel time.

6. The method according to claim 1, characterized in that The determining of the correlation between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result includes: Use big data processing technology to integrate multiple tourist portraits, tourism resource knowledge of each tourist destination, and environmental factor assessment results of each tourist destination; Machine learning technology is used to mine the correlation between the integrated processed tourist portrait, the tourism resource knowledge and the environmental factor evaluation results.

7. The method according to claim 1, characterized in that Also includes: Performing performance evaluation on the tourism intelligent question-answering model to obtain evaluation indicators; wherein the evaluation indicators include at least one of accuracy, recall rate and F1 value; According to the evaluation indicators, the tourism intelligent question and answer model is optimized.

8. A training device for a large tourism intelligent question-answering model, characterized in that: include: A portrait construction module is used to obtain tourist-related data, extract tourist behavior characteristics from the tourist-related data, and construct a corresponding tourist portrait based on the tourist behavior characteristics; The knowledge construction module is used to collect tourism resource data, classify tourist destinations according to the tourism resource data, and explore the characteristics of each tourist destination, and add the tourism resource data, classification and characteristics of each tourist destination as a piece of tourism resource knowledge to the tourism resource knowledge base; The factor evaluation module is used to evaluate the environmental factors of each tourist destination using a machine learning model to obtain the environmental factor evaluation results of each tourist destination; An association determination module, used to determine the association relationship between the tourist portrait, the tourism resource knowledge and the environmental factor assessment result; The model training module is used to obtain historical question-answer pairs, and perform model training based on the historical question-answer pairs and the association relationship to obtain a converged tourism intelligent question-answering model; wherein the tourism intelligent question-answering model can generate corresponding answers based on the association relationship and the questions input by the user.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 7.

10. A computing device, characterized in that: The method comprises a memory and a processor, wherein the memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 7 is implemented.

Images