Travel platform system

Abstract

Travel platform system 10 comprises a user interface 14 which receives a natural language query 22 relating to a travel goal from an end user 20. A large language model 12 connects to the user interface 14 and processes the natural language query 22 to identify one or more travel query components. A machine learning component 16, e.g. an artificial neural network, connects to the large language model and operates a machine learning model trained on a dataset of queries and responses relating to travel goals. The machine learning component analyses the travel query components identified in the natural language query and to generates an output, e.g. a natural language response 24, for the or each travel query component according to the machine learning model, and may transmit the response to the interface. A digital twins component 18 may be connected to the large language model and may process the query to identify travel destinations and generate virtual replicas of the travel destinations. The response may include a component derived from the virtual replicas.

Description

This invention relates to a travel platform system. The traditional travel industry has primarily relied on travel agents and, in recent years, websites providing information on destinations, facilities, and prices. Additionally, bloggers on social media platforms share their travel experiences, often earning income through their blogs. Traditional travel platforms offer basic matching for travellers based on destination and travel dates. On social media platforms such as Facebook, individuals share traveling experience after traveling but there is no platform available for matching and guiding before and during traveling planning. A lack of advanced algorithms led to imprecise matching, resulting in suboptimal travel companion experiences. The advent of the Internet led to the development of travel websites, offering information on destinations, accommodations, and pricing. Some of the websites and forums provided basic tips, and user-generated content. These platforms have limitations in providing personalized and purpose-driven travel experiences. They often focus on transactions rather than fostering connections among travellers. Some of the websites give fragmented information, lack of personalized insights, and absence of real-time virtual exploration tools. Social media platforms allowed travellers to share experiences and connect. While valuable for subjective experiences, these blogs lack a systematic integration of technologies for pre-travel planning, matching with like-minded travellers, and fostering purpose-driven travel on a larger scale. They are limited to post-travel sharing, and lack advanced matching algorithms, purpose-driven travel focus, and real-time solutions during travel. Travel bloggers on social media platforms can share their experiences, providing insights and recommendations to a broad audience. Some platforms incorporated basic chatbots for user interaction, however there is limited use of language models, resulting in less natural and intuitive communication. Virtual tours and augmented reality can be provided by different application that offer glimpses of destinations however, there is a lack of integration with purpose-driven travel, advanced matching, and real-time issue resolution during travel. Travel analytics platforms exist that use analytics tools that provide insights into travel trends but are not integrated into a comprehensive travel planning and companion matching platform. These platforms are focused on post-travel data analysis without addressing real-time issues or pre-travel planning. Some platforms encouraged purpose-driven travel, but these are often niche and lack a broad and integrated approach. They have limited scope and have a fragmented focus on specific purposes without comprehensive travel planning features. It is therefore an object of the invention to improve upon the known art. According to a first aspect of the present invention, there is provided a travel platform system comprising a user interface arranged to receive a natural language query relating to a travel goal from an end user, a large language model connected to the user interface and arranged to process the received natural language query to identify one or more travel query components within the received natural language query, and a machine learning component connected to the large language model and operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals and arranged to analyse the one or more travel query components identified in the natural language query and to generate an output for the or each travel query component according to the machine learning model operated by the machine learning component. According to a second aspect of the present invention, there is provided a method of operating a travel platform system comprising receiving a natural language query relating to a travel goal from an end user, processing the received natural language query with a large language model to identify one or more travel query components within the received natural language query, operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals, analysing the one or more travel query components identified in the natural language query, and generating an output for the or each travel query component according to the machine learning model. According to a third aspect of the present invention, there is provided a computer program product on a computer readable medium, the computer program product comprising instructions for operating a computing device, the instructions for receiving a natural language query relating to a travel goal from an end user, processing the received natural language query with a large language model to identify one or more travel query components within the received natural language query, operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals, analysing the one or more travel query components identified in the natural language query, and generating an output for the or each travel query component according to the machine learning model. Owing to the invention, it is possible to provide an improved travel platform system, for example for use in a companion matching system. The essence of the invention lies in employing Advanced Machine Learning (AML) algorithms to intelligently match users with compatible travel companions, considering a myriad of factors such as preferences, interests, and travel goals. This approach is novel and inventive, offering a level of personalization and precision not achieved by existing solutions. The invention provides implementation of AML algorithms within a travel platform's user matching system. Specifically, the invention addresses the technical field of intelligent user matching for travel companions, employing AML to enhance the precision and personalization of recommendations. The application of AML not only enhances the precision and personalization of travel companion recommendations but also extends to three distinct phases: pre-travel, during travel, and post-travel. This comprehensive approach provides a nuanced, adaptable, and responsive user experience. The strategic application of cutting-edge AML techniques allows the system to refine and optimize the process of matching users with compatible travel companions throughout their entire journey. This innovation addresses the limitations of conventional methods, introducing a new level of adaptability, nuance, and responsiveness to user preferences and travel goals. The core of the system involves the incorporation of sophisticated AML algorithms designed to process and analyse user data for intelligent travel companion matching. The invention focuses on enhancing the user matching system within travel platform, leveraging AML to provide more accurate, personalized, and context-aware recommendations across three key phases: pre-travel, during travel, and post-travel. The system achieves higher precision in travel companion recommendations by considering a broad spectrum of factors, including user preferences, interests, and real-time changes in travel goals. This goal is extended to cover recommendations at various stages of the travel journey. The invention involves user interaction with the travel platform application, where natural language inputs are processed by AML algorithms. This interaction is integral to the travel companion matching workflow across all travel phases. Unlike traditional methods, the invention emphasizes the dynamic adaptability of the AML-driven system, allowing the travel platform to respond to evolving user preferences and changing travel objectives at each stage of the travel journey. The functional problem addressed by the travel platform system is the inadequacy of conventional travel companion matching systems in providing precise, personalized recommendations based on user preferences, interests, and travel goals. The problem that is solved by this invention is the inefficiency and inadequacy of conventional travel companion matching systems in addressing the diverse and evolving needs of users throughout their entire travel journey. Existing solutions often fall short in providing precise, personalized, and context-aware recommendations, leading to suboptimal user experiences. Additionally, these systems typically focus on pre-travel matching, neglecting the dynamic nature of user preferences and objectives during and after the travel experience. The travel platform system aims to overcome this limitation by leveraging AML to refine and optimize the matching process to solve following problems. Traditional travel companion matching systems lack the precision needed to consider a comprehensive set of factors, including evolving user preferences, real-time changes in travel goals, and varying interests at different stages of the travel journey. The travel platform system introduces advanced matching, ensuring users connect with compatible travel companions based on shared interests, preferences, and travel goals, fostering meaningful connections. Existing solutions struggle to deliver highly personalized travel companion recommendations that align with individual user profiles, preferences, and the nuanced nature of travel experiences. The travel platform system offers tailored insights, ensuring users are well-prepared with comprehensive information before embarking on their trips. The majority of existing matching systems are static and primarily focus on pre-travel planning, overlooking the dynamic and evolving nature of user requirements during travel and after the completion of the journey. The travel platform system expands its scope to cater to purpose-driven travel, enabling users to connect for initiatives related to study, research, social support, cultural engagement, and community development and the travel platform system facilitates human-like conversations, enhancing the user experience by providing a seamless and intuitive communication platform. Users often miss opportunities for meaningful interactions, collaboration, and shared experiences with like-minded travel companions due to the limitations of current matching systems and this is overcome by the present travel platform system. The travel platform system addresses these challenges and limitations. By integrating Advanced Machine Learning (AML) algorithms into the platform, the travel platform system is able provide a holistic and adaptive solution that delivers precise, personalized, and context-aware travel companion recommendations throughout the entire travel journey, encompassing pretravel planning, during travel, and post-travel reflection. This approach seeks to enhance user satisfaction, foster meaningful connections, and elevate the overall travel experience. In the past, travel companion matching systems often relied on travel agents or manual input without harnessing the power of advanced technologies like AML. Traditional methods lacked the depth and adaptability required for personalized recommendations, leading to suboptimal user experiences. Traditional travel platforms allowed users to find travel companions based on basic criteria such as destination and travel dates. However, the travel agent creates the group through advertisement or other social media campaign and did not delve into deeper user preferences or purposes for travel. The travel platform system builds upon this by incorporating advanced matching algorithms powered by generative Al and machine learning. These algorithms consider a myriad of factors, ensuring a more nuanced and personalized matching experience. In relation to pre-travel planning and guidance some existing platforms provided information about destinations, but the depth and customization were limited. Users had to rely on fragmented sources for travel planning. The travel platform system provides sophisticated advance machine learning algorithms models based on user conversations, offering unparalleled insights and preplanning guidance. In relation to purpose-driven travel, while there were travel platforms catering to specific niches, the emphasis on purpose-driven travel for social or environmental impact was not as prominent. The travel platform system introduces a unique mission by encouraging purpose-driven travel experiences, connecting users with similar goals, and facilitating initiatives to contribute positively to visited communities. For post-travel engagement past attempts using social sharing features did exist, but the emphasis on post-travel engagement, reflections, and a dynamic social feed was not as comprehensive. The travel platform system helps to create a vibrant post-travel community by encouraging users to share real-time experiences and insights. The platform's advance machine learning algorithms process this user-generated content for continuous improvement of the user’s experience of their travel even after they have completed their travel. Embodiments of the present invention will now be described, by way of example only, with reference to the accompanying drawings, in which:- Figure 1 is a schematic diagram of a travel platform system, Figure 2 is a flow diagram of a method of operating the travel platform system of Figure 1, and Figure 3 is a schematic diagram of a computer processing system. Figure 1 shows a travel platform system 10, which comprises four main components, a large language model 12, a user interface 14 connected to the large language model 12, a machine learning component 16 connected to the large language model 12, and a digital twins component 18 connected to the large language model 12. Also shown schematically in Figure 1 is an end user 20 who is able to access the user interface 14. The end user 20 is able to transmit a natural language query 22 to the user interface 14 and will receive a response 24 from the user interface 14. The natural language query 22 relates to a travel goal from the end user 20. The user interface 14, the large language model 12, the machine learning component 16 and the digital twins component 18 could all be embodied as a single dedicated processor that executes their respective functions, or could be embodied as distinctive individual processors carrying out the respective functions or could be implemented as software run by a general purpose computer. In a preferred embodiment, the machine learning component 16 comprises an artificial neural network, embodied either as a hardware component or emulated within a software solution. The large language model 12 is arranged to process the received natural language query 22 to identify one or more travel query components within the received natural language query 22. The machine learning component 16 is operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals. The machine learning component 16 is arranged to analyse the one or more travel query components identified in the natural language query 22 and to generate an output for each travel query component according to the machine learning model operated by the machine learning component 16. The system 10 is provided via a dedicated app that a user can access, for example through a smartphone. The app can provide the user interface 14 while the remaining components can be accessed via a remote connection through the app. The system 10 can be used, for example in a scenario involving an enduser 20 who is engaged in a conversation where the user 20 has expressed a desire to travel to Kamilo Beach, situated on the south-eastern tip of Hawaii. The user 20 wishes to travel to this location with the intention of participating in a cleanup initiative to address plastic bags and trash on the beach. The user 20 has specified a preference to travel in July and has sought assistance in organizing a group of individuals to join the cleanup effort. The user 20 also has a number of additional inquiries including the creation of a replica of Kamilo Beach for a virtual visit, determining the number of individuals needed for the cleanup and facilitating connections with them, inquiring about the expected temperature in July and seeking general recommendations for the trip. This example conversation serves as a basis for exploring the data flow and interaction of technologies within the product architecture of the system 10, as illustrated in the flowchart of Figure 2, which illustrates the various steps in the operation of the travel platform system of Figure 1. The first step in the process is step 1 “User Interaction and Query Submission”. The user’s first action is that the user 20 logs into the application via the user interface 14 and starts a conversation with the large language model 12. The data flow within the system 10 is that all user queries, preferences, and travel goals are submitted to the LLM 12 in natural language. The next step in the process is step 2: “Large Language Model (LLM) Processing”. The technology involved is that the LLM 12 is powered by Advanced Machine Learning (AML) algorithms. The next step is step 3, which comprises data processing. The LLM 12 processes the received user inputs, understanding the user's desire to clean up Kamilo Beach in July, create a virtual replica, find travel companions, check weather, and receive recommendations. The next step is step 4 which is the decision point. The LLM 12 determines that the user's request 22 involves various aspects, in this case replica creation, travel companion matching, weather inquiry, and general recommendations. This is followed by step 5, which comprises the conditional transfer to the machine learning component 16. The technology involved in this step is the operation of the sophisticated advance machine learning component 16. The next step in the process is step 6, which comprises the data transfer. The LLM 12 transfers relevant data to the machine learning component 16 for analytics purposes, considering user preferences, goals, and the specific request. The next step is step 7 which is the application of the machine learning analytics. The technology involved is that the machine learning component 16 is equipped with advanced algorithms. At step 8 data analysis is performed. The machine learning component 16 analyses user data, providing insights such as the number of people required for the cleanup effort, considering historical data and travel patterns. The final step in the process is step 9 which is a contribution from the digital twins component 18. The machine learning insights contribute to enhancing the digital twins component 18 to enable this component to create a lifelike virtual replica of Kamilo Beach. The outputs of the machine learning component 16 and the digital twins component 18 can be provided as a response 24 that is supplied back to the user 20 via the user interface 14 of the application that is operating the travel platform system 10 of Figure 1. In summary, the process discussed above can be used to provide a collaborative approach workflow. The process starts with user query processing. In this case, the user expresses a desire to travel to Kamilo Beach for a cleanup initiative and specifies a preference for July. This query is processed by the Large Language Model (LLM) 12. This produces a decision point for further processing. The LLM 12 identifies whether further processing is needed based on the user's queries 22. In this scenario, additional processing is likely required. If further processing is needed, the LLM 12 transfers relevant data to Advanced Machine Learning (AML) component 16. The machine learning component 16 processes the data for analytics, predictions, and insights. If the user 20 requests a virtual replica of Kamilo Beach, the LLM 12 may transfer data to the digital twins component 18 for processing. The digital twins component 18 process the data, creating a lifelike virtual representation of Kamilo Beach. If the user 20 seeks assistance in organising a group for the cleanup effort, the machine learning component 16 can analyse data to determine the number of individuals required and facilitate connections with them. If the user 20 enquires about the expected temperature in July, the machine learning component 16 can provide insights based on historical data and forecasts. If the user 20 seeks general recommendations for the trip, the machine learning component 16 can generate personalized suggestions based on, for example, user preferences, travel goals, and insights. The results from the machine learning component 16 such as analytics, temperature predictions, and replica information from the digital twins component 18 are integrated. The LLM 12 combines the data, deciding if the response is sufficient. A final response 24, including natural language responses, analytics from the machine learning component 16, and replica information from the digital twins component 18, is sent back to the user interface 14 for presentation to the user 20. Figure 3 shows how the travel platform system 10 can be embodied with a user device 26 such as a smartphone 26 connecting to a server 28 that is running the travel platform system 10. The user’s query 22 is transmitted over a network such as the Internet from the user device 26 to the server 28 and the response 24 is transmitted back from the server 28 to the user device 26. The server physically embodies the functions of the travel platform system 10. The server 28 runs a computer program product from a computer readable medium such as a hard disk. The computer program product comprises instructions for operating a computing device such as the server 28. The instructions are for receiving a natural language query 22 relating to a travel goal from an end user 20, processing the received natural language query 22 with a large language model 12 to identify one or more travel query components within the received natural language query 22, operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals, analysing the one or more travel query components identified in the natural language query 22, and generating an output for the or each travel query component according to the machine learning model. The machine learning component 16 embodies functions of Advance Machine Learning (AML). The travel platform system 10 is constructed so that the AML facilitates intelligent matching of users based on diverse criteria, including preferences, interests, and travel goals. The AML processes contextual data, ensuring recommendations adapt to real-time changes in user preferences and travel objectives. Unlike conventional methods, AML demonstrates dynamic adaptability, allowing the system 10 to respond flexibly to evolving user requirements. AML results in higher precision in travel companion recommendations by considering a broad spectrum of factors, including nuanced user preferences. The AML-driven system 10 personalizes travel experiences by tailoring recommendations based on individual user profiles and historical interactions. The AML provides real-time analytics to determine the number of individuals required for collaborative efforts, such as beach cleanups, and facilitates user connections. The AML seamlessly integrates with Natural Language Processing (NLP) for enhanced user interactions, ensuring the system understands and responds to natural language queries. The model embodied by the machine learning component 16 embodies various technical features. For example, the machine learning component 16 uses feature engineering for user profiling. AML incorporates advanced feature engineering techniques to create comprehensive user profiles, capturing intricate details that contribute to effective matching. The machine learning component 16 can also provide algorithmic learning and adaptation. The AML model embodied by the machine learning component 16 leverages algorithmic learning to adapt and refine its recommendations based on user feedback, ensuring continuous improvement. The machine learning component 16 also provides the dynamic weighting of variables. The AML employs dynamic weighting of variables, allowing the model to assign varying importance to different user preferences and factors during the matching process. The machine learning component 16 provides integration with external data sources. The model run by the machine learning component 16 integrates seamlessly with external data sources, enriching user profiles with additional information for more robust analytics and recommendations. The machine learning component 16 can provide predictive analytics for virtual exploration. The AML includes predictive analytics capabilities for virtual exploration, forecasting user preferences and generating lifelike replicas of destinations. The machine learning component 16 also provides explainability and transparency. The AML incorporates features that enhance model explainability and transparency, providing users with insights into how recommendations are generated. Machine learning is a subfield of artificial intelligence that focuses on developing algorithms and models that enable computer systems to learn patterns, make predictions, and improve their performance over time without being explicitly programmed. Machine learning algorithms use data to identify patterns, make decisions, and adapt their behaviour based on experience. The algorithms used by the machine learning component 16 are designed to deliver precise and highly personalized recommendations. By leveraging advanced machine learning techniques, the machine learning component 16 achieves a level of accuracy that goes beyond conventional methods, ensuring that travel companions and recommendations align closely with individual user preferences. Developing machine learning algorithms for the travel industry poses unique challenges due to the diverse and dynamic nature of customer scenarios. Travelers worldwide have diverse preferences, objectives, and interests. Crafting algorithms that understand and adapt to this diversity requires sophisticated models capable of capturing nuances in user behaviour and preferences. User preferences in the travel industry are dynamic and evolve over time. Machine learning models need to continuously adapt and learn from new data to provide up-to-date and relevant recommendations, especially as user preferences change. Personalizing recommendations for each user involves considering a myriad of factors, such as travel history, interests, trip objectives, and even real-time changes in preferences. Creating algorithms that can handle this level of personalization is a complex task. Understanding the context of user queries and interactions is crucial. The machine learning algorithms used by the travel platform system 10 are able to comprehend the context of travel requests, whether it's planning, during the journey, or post-travel, to provide accurate and meaningful recommendations. The travel industry involves various stages such as before, during, and after travel. The machine learning models used within the travel platform system 10 take account of these stages and provide recommendations that cater to each phase, from pre-travel planning to on-the-go assistance and post-travel reflections. Travel plans can change in real-time due to unforeseen circumstances. The machine learning algorithms used by the travel platform system 10 are adaptable, making decisions and recommendations that align with realtime changes in user preferences or travel conditions. Travel preferences often reflect cultural differences. The machine learning algorithms of travel platform system 10 are culturally sensitive, considering variations in travel expectations and requirements across different regions and communities. Travel data often includes sensitive information. The algorithms used by the travel platform system 10 respect the users’ privacy and adhere to data security standards adds an extra layer of complexity. For scenarios such as collaborative activities during travel, for example group cleanups, the machine learning models used by the travel platform system 10 facilitate connections among users with similar interests or objectives. Collaborative learning adds complexity to the recommendation systems. The machine learning algorithms of the travel platform system 10 are not static and they are configured to continuously learn and improve based on user feedback and changing trends. This travel platform system 10 includes mechanisms for ongoing training and adaptation. One of the key advantages of the algorithms provided within the travel platform system 10 lies in their dynamic adaptability. They can respond in realtime to changes in user preferences and travel goals. This adaptability ensures that travel platform system 10 remains agile and relevant, providing users with recommendations that align with their evolving needs. The complexity of the statistical and mathematical tools utilized in the algorithms within the travel platform system 10 is a testament to the sophistication of the approach delivered by the travel platform system 10. The system 10 employs advanced modelling techniques, including deep learning, neural networks, and ensemble methods, to process large and dynamic datasets effectively. The algorithms used within the travel platform system 10 are not static; they continuously learn and refine themselves in real-time. By training on large datasets with vast amounts of real-time information, the algorithms stay abreast of the latest trends, user behaviours, and contextual changes, ensuring that recommendations remain relevant and up to date. Feature engineering can be used within the algorithms. The travel platform system 10 goes beyond traditional variables and incorporate innovative features that capture intricate details of user preferences. This inventive approach contributes to the uniqueness of travel platform system 10. The algorithms of the travel platform system 10 are equipped with predictive analytics capabilities, enabling virtual exploration that goes beyond simple replication. The algorithms are able to forecast user preferences and generate lifelike replicas of destinations, enhancing the overall virtual travel experience. In the era of responsible Al, the algorithms of the travel platform system 10 are able prioritize explainability and transparency. Users can understand how recommendations are generated, fostering trust and allowing for user feedback that contributes to continuous improvement. The collaborative learning aspect of the algorithms facilitates meaningful user connections. By determining the optimal number of individuals required for collaborative activities, such as beach cleanups, our models contribute to community-building and shared experiences. The combination of advanced machine learning, dynamic adaptability, context-aware matching, precision in recommendations, user interaction with natural language processing, personalized recommendations across travel phases, real-time collaboration for group activities, adaptation to global cultural sensitivity, and continuous learning mechanisms within the travel platform system 10 collectively contribute to the successful operation of the travel platform system 10. These features represent a significant departure from existing solutions. The system 10 involves the integration of advanced machine learning (AML) algorithms within the travel platform system 10. While machine learning is a known concept, the application of sophisticated AML techniques specific to travel companion matching represents a substantial advancement. The intricacies of user preferences, interests, and dynamic travel goals require a level of complexity that goes beyond conventional ML approaches. The travel platform system 10 provides dynamic adaptability, allowing the system 10 to respond to evolving user preferences and changing travel objectives. This level of responsiveness to real-time changes in user preferences and travel conditions sets the solution apart from static or less adaptive approaches commonly found in existing systems. Context-aware matching is a central feature of the travel platform system 10, enabling nuanced recommendations based on the context of user queries. Existing solutions may lack the depth and context-awareness required to deliver highly personalized recommendations that align closely with individual user scenarios. The travel platform system 10 delivers precision in travel companion recommendations by considering a broad spectrum of factors, including user preferences, interests, and real-time changes in travel goals. This precision distinguishes the travel platform system 10 from generic recommendation systems that do not account for the subtleties of travel preferences. The Advance Machine Learning integration of natural language processing (NLP) within user interactions is a distinctive feature of the travel platform system 10. While NLP is utilized in various applications, the specific incorporation into the travel companion matching workflow provided by the travel platform system 10 adds a layer of sophistication. This allows users to engage in natural language conversations with the system 10, enhancing the overall user experience. The travel platform system 10 addresses the challenges of personalization throughout different travel phases: before, during, and after travel. Existing systems tend to focus on specific stages or lack the comprehensive approach required for a seamless and personalized end-to-end travel experience. The collaborative learning aspect of the travel platform system 10, facilitating connections among users with similar interests for group activities during travel, is a novel approach. This collaborative feature distinguishes the travel platform system 10 from systems that primarily focus on individualized recommendations. The travel platform system 10 is able to take into account and adapt global cultural variations in travel expectations and requirements is a key aspect of the travel platform system 10. The system 10 incorporates mechanisms for global cultural sensitivity, ensuring that recommendations align with diverse cultural norms. Existing systems do not exhibit the same level of cultural adaptability. The travel platform system 10 incorporates mechanisms for continuous learning and improvement based on user feedback and changing trends. The ability to evolve over time and continuously improve distinguishes the travel platform system 10 from static or less adaptable systems. The travel platform system 10 uses advanced machine learning (AML) algorithms for intelligent user matching in the travel industry. The multifaceted nature of the travel platform system 10 embedded with advance machine learning algorithms and models, addressing the intricacies of user preferences, dynamic adaptability, context-aware matching, comprehensive consideration of travel phases, NLP integration, global cultural sensitivity, real-time collaboration for group activities, and continuous learning mechanisms, collectively contributes to its operation. These aspects represent innovative steps that require a deep understanding of both the travel industry intricacies and advanced machine learning techniques. The travel platform system 10 is able to deal with the complexity of user preferences. The travel industry involves a highly diverse range of user preferences, objectives, and interests. The complexity of individual preferences, coupled with dynamic changes in travel goals, makes it non-trivial to develop algorithms that accurately match users with compatible travel companions. The machine learning model run by the machine learning component 16 is also able to fulfil a dynamic adaptability requirement. The need for dynamic adaptability to respond to real-time changes in user preferences and travel objectives is a crucial aspect of the operation of the travel platform system 10. Developing algorithms that can adapt dynamically to evolving scenarios and provide personalized recommendations accordingly requires a level of sophistication that is not available with conventional travel recommendation systems. The travel platform system 10 is able to solve context-aware matching challenges. Context-aware matching, considering the nuanced context of user queries, adds a layer of complexity. Existing solutions do not adequately address the subtleties of user requests, making the incorporation of context-aware matching an advantageous feature of the platform 10. The travel platform system 10 is also able to execute comprehensive travel phases consideration. The travel platform system 10 emphasis on personalization throughout different travel phases: before, during, and after travel, requires a comprehensive approach. Conventional systems do not integrate personalization seamlessly across all stages of the travel experience. The travel platform system 10, through the operation of the large language model (LLM) 12 delivers the incorporation of natural language processing (NLP). The integration of natural language processing (NLP) within user interactions is a unique feature. While NLP is a known technology, applying it specifically to enhance user conversations and engagement in the travel companion matching workflow improves the operation of the travel platform system 10. As detailed above, the travel platform system 10 is able to deliver a global cultural sensitivity requirement. The travel platform system 10 addresses global cultural variations in travel expectations and requirements and this is a distinctive feature. Incorporating mechanisms for global cultural sensitivity in travel recommendations goes beyond typical matching algorithms and is an improvement over for those systems that are focused on localized or less adaptable systems. The travel platform system 10 is also able to provide real-time collaboration for group activities. The collaborative learning aspect, facilitating connections among users with similar interests for group activities during travel, is a significant feature of the travel platform system 10. Enabling real-time collaboration for group experiences is not a known feature in traditional travel recommendation systems. The travel platform system 10 also delivers continuous learning and improvement mechanisms. The incorporation of continuous learning and improvement mechanisms based on user feedback and changing trends represents an improvement over traditional systems that lack the adaptability to learn and evolve over time, making this a desirable feature to those accustomed to static recommendation approaches.

Claims

1. A travel platform system (10) comprising:• a user interface (14) arranged to receive a natural language query (22) relating to a travel goal from an end user (20),• a large language model (12) connected to the user interface (14) and arranged to process the received natural language query (22) to identify one or more travel query components within the received natural language query (22), and• a machine learning component (16) connected to the large language model (12) and operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals and arranged to analyse the one or more travel query components identified in the natural language query (22) and to generate an output for the or each travel query component according to the machine learning model operated by the machine learning component (16).

2. A travel platform system according to claim 1, and further comprising a digital twins component (18) connected to the large language model (12), wherein the large language model (12) is arranged to process the received natural language query (22) to identify one or more travel destinations within the received natural language query (22) and wherein the digital twins component (18) is arranged to generate one or more virtual replicas of the identified one or more travel destinations within the received natural language query (22).

3. A travel platform system according to claim 1 or 2, wherein the large language model (12) is further arranged to generate a response (24) from the outputs generated by the machine learning component (16) for the or each travel query component and transmit the response (24) to the user interface (14).

4. A travel platform system according to claim 2 and 3, wherein the large language model (12) is further arranged, when generating a response (24), to include within the response (24) a component derived from the one or more virtual replicas generated by digital twins component (18).

5. A travel platform system according to claim 3 or 4, wherein at least a portion of the response (24) is a natural language response.

6. A travel platform system according to any preceding claim,wherein the machine learning component (16) comprises an artificial neuralnetwork.

7. A method of operating a travel platform system (10) comprising:• receiving a natural language query (22) relating to a travel goal from an end user (20),• processing the received natural language query (22) with a large language model (12) to identify one or more travel query components within the received natural language query (22),• operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals,• analysing the one or more travel query components identified in the natural language query (22), and• generating an output for the or each travel query component according to the machine learning model.

8. A method according to claim 7, and further comprisingprocessing the received natural language query (22) to identify one or moretravel destinations within the received natural language query (22) and generating one or more virtual replicas of the identified one or more travel destinations within the received natural language query (22).

9. A method according to claim 7 or 8, and further comprising generating a response (24) from the outputs generated according to the machine learning model for the or each travel query component and transmitting the response (24).

10. A method according to claim 8 and 9, and further comprising, when generating a response (24), including within the response (24) a component derived from the one or more virtual replicas.

11. A method according to claim 9 or 10, wherein at least a portion of the response (24) is a natural language response.

12. A method according to any one of claims 1, wherein the machine learning model comprises an artificial neural network.

13. A computer program product on a computer readable medium, the computer program product comprising instructions for operating a computing device, the instructions for:• receiving a natural language query (22) relating to a travel goal from an end user (20),• processing the received natural language query (22) with a large language model (12) to identify one or more travel query components within the received natural language query (22),• operating a machine learning model that has been trained on a dataset of queries and responses relating to travel goals,• analysing the one or more travel query components identified in the natural language query (22), and• generating an output for the or each travel query component according to the machine learning model.

14. A computer program product according to claim 13, and further comprising instructions for processing the received natural language query(22) to identify one or more travel destinations within the received natural language query (22) and generating one or more virtual replicas of the identified one or more travel destinations within the received natural language query (22).

15. A computer program product according to claim 13 or 14, and further comprising instructions for generating a response (24) from the outputs generated according to the machine learning model for the or each travel query component and transmitting the response (24).

16. A computer program product according to claim 14 and 15, and further comprising instructions for, when generating a response (24), including within the response (24) a component derived from the one or more virtual replicas.

17. A computer program product according to claim 18 or 19, wherein at least a portion of the response (24) is a natural language response.

18. A computer program product according to any one of claims 13 to 17, the machine learning model comprises an artificial neural network.

Images