Travel information processing method and device, storage medium and program product

Abstract

The embodiment of the invention provides a travel information processing method and device, a storage medium and a program product. In the travel information processing method, user feature data of at least one dimension and at least one type of travel context information of a user can be obtained under the triggering of travel query operation of the user, and travel service information is generated by utilizing a target language model according to the obtained user feature data and the travel context information. Wherein the user feature data of at least one dimension is used for enabling the target language model to deeply understand the travel demand of the user, and the context information is used for enabling the target language model to consider the influence of different information related to the travel on the travel; furthermore, the target language model can fuse the above information, actively identify the personalized travel scene of the user, and mine the potential demand of the user in the personalized scene, thereby facilitating providing accurate travel service for the user.

Description

TECHNICAL FIELD

[0001] The present specification relates to the technical field of information processing, and in particular, to a travel information processing method and device, a storage medium, and a program product. BACKGROUND

[0002] Under the background of the deep integration of intelligent transportation and digital life, travel services are rapidly developing towards intelligence, personalization, and full-process companionship. Travel service platforms mainly focus on meeting the needs of users in specific links, such as providing route planning needs, ticket price query needs, or order execution needs (such as taking a taxi, buying a ticket, or making a reservation). Traditional travel service platforms usually provide homogeneous travel services for different users, resulting in insufficient service accuracy. Therefore, a new solution is needed. SUMMARY

[0003] The embodiments of the present specification provide a travel information processing method, device, storage medium, and program product to improve the accuracy of travel services.

[0004] The embodiments of the present specification provide a travel information processing method, which includes: in response to a travel query request sent by a client according to a travel query operation of a user, obtaining user feature data of at least one dimension of the user; according to the type of the travel query operation, calling at least one task tool to obtain at least one travel context information; providing the user feature data of the at least one dimension and the at least one travel context information to a target language model to obtain travel service information output by the target language model, the travel service information including travel guidance information and / or recommendation information of a service component; and sending the travel service information to the client for display.

[0005] Optionally, according to the type of the travel query operation, calling at least one task tool to obtain at least one travel context information includes: if the travel query operation hits a target shortcut instruction, determining the at least one task tool and its execution order from the registered task tools according to a tool arrangement mode corresponding to the target shortcut instruction; and calling the at least one task tool to obtain the at least one travel context information according to the execution order.

[0006] Optionally, it further includes: if the travel query operation does not hit the target shortcut instruction, calling a planning model to perform task planning according to the user feature data of the at least one dimension and the registered task tools to obtain the at least one task tool and its execution order from the registered task tools; and calling the at least one task tool to obtain the at least one travel context information according to the execution order.

[0007] Optionally, the planning model is invoked to plan the task according to the user characteristic data in the at least one dimension to obtain the at least one task tool and the execution sequence thereof, including: invoking a main planning model to plan the task according to the user characteristic data in the at least one dimension and the registered task tools to obtain the at least one task tool and the execution sequence thereof from the registered task tools; or, invoking the main planning model to plan the task according to the user characteristic data in the at least one dimension to select a target sub-planning model from the at least one sub-planning model, different sub-planning models corresponding to different vertical fields; and invoking the target sub-planning model to plan the task according to the user characteristic data in the at least one dimension and the registered task tools to obtain the at least one task tool and the execution sequence thereof from the registered task tools.

[0008] Optionally, the at least one task tool is invoked to obtain the at least one travel context information in the execution sequence, including at least one of: invoking a fulfillment tool to query fulfillment data according to the user identification information to obtain information of a fulfillment itinerary of the user; invoking a member tool to query member benefit data according to member information of the user to obtain member benefit information of the user; invoking a weather query tool to obtain weather information corresponding to a target location according to information of the target location in the fulfillment itinerary of the user; invoking a traffic perception tool to obtain traffic information of the target location according to information of the target location in the fulfillment itinerary of the user; and invoking a service recall tool to recall at least one candidate service component from a plurality of service components according to at least one of the information of the fulfillment itinerary of the user, the member benefit information, the weather information and the traffic information, and determine service candidate information according to the at least one candidate service component.

[0009] Optionally, the service recall tool is invoked to recall at least one candidate service component from a plurality of service components according to the information of the fulfillment itinerary of the user, including: the service recall tool is invoked to determine a travel node of the user according to the information of the fulfillment itinerary of the user; and at least one candidate service component adapted to the travel node is recalled from the plurality of service components according to the travel node of the user.

[0010] Optionally, the service candidate information is determined according to the at least one candidate service component, including: the travel situation of the user is obtained according to the traffic information and / or the weather information; the recommendation priority of at least one candidate service component adapted to the travel node is determined according to the travel situation; and the service candidate information is determined according to the at least one candidate service component and the recommendation priority thereof.

[0011] Optionally, further comprising: in response to a tool configuration operation on the target shortcut instruction, obtaining the at least one task tool configured for the target shortcut instruction and an execution order thereof, to obtain a tool arrangement manner corresponding to the target shortcut instruction.

[0012] Optionally, further comprising: in response to a service policy update operation, determining an added task tool and / or a task tool to be removed; adding the added task tool to the registered task tools, and / or deregistering the task tool to be removed from the registered task tools, to update the registered task tools.

[0013] Optionally, providing the at least one dimension of user feature data and the at least one travel context information to a target language model to obtain travel service information output by the target language model, includes: generating a prompt word according to the at least one dimension of user feature data and the at least one travel context information; providing the prompt word as input information to the target language model; performing word segmentation on the input information in the target language model to obtain an original text unit; encoding the original text unit and its position information to obtain an embedding representation of the original text unit and position encoding; performing attention calculation on the embedding representation of the original text unit and the position encoding based on an attention mechanism to obtain an attention feature; and iteratively predicting a new text unit in a self-recurrent manner according to the attention feature until the travel service information is output when an iteration condition is met.

[0014] Optionally, the at least one dimension of user feature data includes at least one of travel preference information of the user and member information of the user.

[0015] The embodiments of the present specification further provide a travel information processing method, including: in response to a user's itinerary query operation, sending an itinerary query request to an intelligent travel service, the itinerary query request being used to: make the intelligent travel service obtain at least one dimension of user feature data of the user, and according to a type of the itinerary query operation, call at least one task tool to obtain at least one travel context information; and provide the at least one dimension of user feature data and the at least one travel context information to a target language model to obtain travel service information output by the target language model, the travel service information including: travel guide information and / or recommendation information of a service component; obtaining the travel service information returned by the intelligent travel service, and displaying the travel service information on a dialogue page with the intelligent travel service.

[0016] Optionally, in response to the user's trip query operation, a trip query request is sent to the intelligent travel service, including: displaying a fulfillment notification message of the target fulfillment trip of the user on a first page, and in response to a triggering operation on the fulfillment notification message, sending a trip query request corresponding to the target fulfillment trip to the intelligent travel service; or, displaying an icon of the intelligent travel service on a second page, and in response to a triggering operation on the icon of the intelligent travel service, sending the trip query request to the intelligent travel service; or, detecting a trip query message input by the user on a conversation page with the intelligent travel service, and sending the trip query request to the intelligent travel service according to the trip query message.

[0017] Optionally, the first page or the second page is a travel channel page, the travel channel page contains a plurality of sub-channel pages, and different sub-channel pages correspond to different travel modes; displaying the travel service information on the conversation page with the intelligent travel service includes: determining a target sub-channel page in which the user initiates the trip query operation from a plurality of sub-channels contained in the travel channel page; if the travel service information includes travel service information of fulfillment trips in a plurality of travel modes, respectively, determining a priority of the fulfillment trips in the plurality of travel modes according to a travel mode corresponding to the target sub-channel page; and displaying the travel service information of the fulfillment trips in the plurality of travel modes, respectively, according to the priority.

[0018] The embodiments of the present specification also provide an electronic device, comprising: a memory and a processor; the memory is used to store one or more computer instructions; the processor is used to execute the one or more computer instructions to perform the steps in the method provided by the embodiments of the present specification.

[0019] The embodiments of the present specification also provide a computer readable storage medium storing a computer program, which is executed by a processor to implement the steps in the method provided by the embodiments of the present specification.

[0020] The embodiments of the present specification also provide a computer program product, comprising: computer programs / instructions, which are executed by a processor to implement the steps in the method provided by the embodiments of the present specification.

[0021] In the embodiments of the present specification, at least one dimension of user feature data and at least one travel context information of a user can be obtained under the triggering of a travel query operation of the user, and a travel service information is generated according to the obtained user feature data and travel context information by using a target language model. The at least one dimension of user feature data is used to make the target language model more deeply understand the travel needs of the user, and the at least one context information is used to make the target language model consider the influence of different information related to travel on travel; the target language model can fuse the above-mentioned various information, actively identify the personalized travel scene of the user, and mine the potential needs of the user in the personalized scene, so as to facilitate the provision of accurate travel services for the user. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings, which are included to provide a further understanding of the present specification, constitute a part of the present specification, and the illustrative embodiments of the present specification and their description serve to explain the present specification, and do not constitute an improper limitation on the present specification. In the drawings:

[0023] Figure 1 A flowchart of an example embodiment of the present specification is provided, and the flowchart shows the process of the travel information processing method when the method is executed on the side of the intelligent travel service;

[0024] Figure 2 A flowchart of an example embodiment of the present specification is provided, and the flowchart shows the process of the travel information processing method when the method is executed on the side of the intelligent travel service;

[0025] Figure 3 A schematic diagram of the present specification is provided for an example embodiment of the present specification, and the schematic diagram shows the display of the performance notification card on the travel channel page;

[0026] Figure 4 A schematic diagram of the present specification is provided for an example embodiment of the present specification, and the schematic diagram shows the display of the icon of the intelligent travel service on the travel channel page;

[0027] Figure 5 A schematic diagram of the present specification is provided for an example embodiment of the present specification, and the schematic diagram shows the display of the travel service information on the dialogue page of the intelligent travel service;

[0028] Figure 6 A flowchart of an example embodiment of the present specification is provided, and the flowchart shows the process of the travel information processing method when the method is executed on the side of the intelligent travel service;

[0029] Figure 7 A schematic diagram of the structure of an electronic device is provided for an example embodiment of the present specification. DETAILED DESCRIPTION

[0030] For the purposes of the present description, the technical solutions and advantages thereof, specific terms are used in the present description. As "one embodiment", "an embodiment", and / or "some embodiments" mean a certain feature, structure or characteristic in relation to at least one embodiment of the present description. Therefore, it should be emphasized and noted that the "an embodiment" or "one embodiment" or "one alternative embodiment" mentioned in the present description twice or more in different positions do not necessarily mean the same embodiment. In addition, the skilled in the art can combine and combine the different embodiments or examples described in the present description and the features of the different embodiments or examples without contradiction.

[0031] The terms used in the embodiments of the present description are only for the purpose of describing the specific embodiments of the present description, and are not intended to limit the present description. The singular forms "a", "an" and "the" used in the embodiments of the present description and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Multiple" generally includes at least two, but does not exclude the case of including at least one. "Multiple" generally includes at least two, but does not exclude the case of including at least one.

[0032] The terms used in the embodiments of the present description are only for the purpose of describing the specific embodiments of the present description, and are not intended to limit the present description. The singular forms "a", "an" and "the" used in the embodiments of the present description and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Multiple" generally includes at least two, but does not exclude the case of including at least one. "Multiple" generally includes at least two, but does not exclude the case of including at least one.

[0033] It should be understood that the term "and / or" used herein is only to describe the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which means that there are three cases of A alone, A and B together, and B alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are a "or" relationship.

[0034] It should also be noted that the term "include", "contain" or any other variant thereof is intended to cover non-exclusive inclusion, so that the product or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such product or system. Without more limitation, the element defined by the statement "including a" does not exclude the presence of other identical elements in the product or system including the element.

[0035] In a travel scenario, fulfillment refers to the entire process in which a platform or service provider provides services to a user according to an order agreement. For example, a ticket is successfully issued, a hotel reservation is confirmed, a train ticket is successfully purchased, a private car driver accepts an order and arrives at the pickup point, a trip is changed or completed. Fulfillment trip refers to a trip for which an order has been executed or is being executed.

[0036] Some travel service platforms cannot dynamically perceive and coordinate users' fulfillment trips in multiple travel scenarios when providing travel services for users' fulfillment trips. Typically, they rely on single-travel-scenario travel data or static rules to provide travel services for users, such as providing travel reminders for users' train or flight trips at fixed times, recommending general travel guides to users, etc. In addition, such travel service platforms are usually limited to providing travel services for users based on single-dimensional user data (such as users' historical orders) and cannot deeply integrate multi-dimensional user data. This approach results in service platforms providing homogeneous travel services for different users, which is not accurate enough.

[0037] To address the above technical problems, in some embodiments of the present specification, a solution is provided. The technical solutions provided by the embodiments of the present specification are described in detail below with reference to the accompanying drawings.

[0038] Figure 1 is a flowchart of a travel information processing method provided by an exemplary embodiment of the present specification. The method can include the steps as shown in Figure 1 .

[0039] Step 101, in response to a trip query request sent by a client according to a user's trip query operation, obtaining at least one dimension of user feature data of the user.

[0040] Step 102, according to the type of the trip query operation, calling at least one task tool to obtain at least one travel context information.

[0041] Step 103, providing the at least one dimension of user feature data and the at least one travel context information to a target language model to obtain travel service information output by the target language model, the travel service information including: travel guide information and / or service component recommendation information.

[0042] Step 104, sending the travel service information to the client for display.

[0043] The execution subject of the embodiment can be an intelligent travel service. The intelligent travel service can run on a client device or a server device. If running on a server device, the server device can be implemented as a conventional server, a cloud server, a virtual machine, etc. The embodiment is not limited in this regard. The intelligent travel service can interact with the client to provide travel-related services to a user of the client.

[0044] In step 101, the itinerary query operation of the user is used to query the detail information of the fulfillment itinerary of the user and trigger a dialogue with the intelligent travel service. The user feature data of at least one dimension of the user is used to describe the user from one or more dimensions to perceive the personalized features of the user. The user feature data of the at least one dimension can include at least one of travel preference information of the user and member information of the user.

[0045] The travel preference information is used to describe the personalized selection tendency of the user in the long-term travel process, including but not limited to at least one of the traffic preference, the accommodation preference, and the service preference of the user. The traffic preference can include at least one of the flight preference (prefer early flights, do not accept transfers, etc.), the seat preference (economy class, window seat, etc.), and the transportation tool preference (prefer high-speed rail instead of airplane). The accommodation preference can include at least one of the room type preference (non-smoking room, family room), the hotel brand preference, and the facility preference (need a gym, include breakfast). The service preference can include at least one of the meal preference and the pick-up preference.

[0046] The member information is used to describe the member level of the user in the travel service platform or the cooperative platform of the intelligent travel service. For example, the member level of the user in the travel service platform is a silver card member, and the member level of the user in an airline company is a gold card member.

[0047] In step 102, at least one travel context information can be obtained by calling at least one task tool according to the type of the itinerary query operation. Different types of itinerary query operations can call different types of task tools, and different types of task tools are used to obtain different types of travel context information.

[0048] The trip context information refers to various information associated with the trip behavior of the user and capable of affecting or reflecting the trip process state and experience. The trip context information can include, but is not limited to, at least one of information of a fulfillment itinerary, member benefit information, weather information, traffic information, and service candidate information. The information of the fulfillment itinerary includes one or more fulfillment itineraries of the user, and a fulfillment progress of a single fulfillment itinerary, such as information of flight delay, flight arrival, and online car driver order acceptance. The member benefit information is used to describe exclusive benefits that the user can enjoy in the trip service based on the member identity or member level. For example, exclusive benefits of a general member include: air ticket discount / refund, priority check-in / seat selection, luggage capacity upgrade, etc.; exclusive benefits of a senior member include: free change of flight / return ticket, delayed check-out, check-in gift, etc. The weather information is used to describe current weather, precipitation probability, temperature, wind force, and other meteorological data related to trip comfort and route selection of the departure location, the destination, and the route during the trip period. The traffic information is used to describe the road network operating conditions of the current location of the user, and can include traffic congestion degree, accident, construction, traffic restriction, signal light state, etc., for evaluating trip efficiency and path feasibility.

[0049] The service candidate information is used to describe the service components that the user is likely to need, and recommend the service components to the client for use by the user. The service component refers to a reusable, embeddable, and independently interactive functional unit of a specific service capability encapsulated in a front-end interface such as a client, an applet, or a webpage. It helps the user to quickly initiate and complete a specific service operation through simplified entry, pre-set context, or one-key call. In the trip scenario, the client can provide at least one of a traffic service component, an accommodation service component, and a ticket service component. For example, the traffic service component can include: a map navigation component, a parking lot query component, a ride-hailing component, etc.; the accommodation service component can include: a hotel reservation component, a room type upgrade component, an invoice issuing component, a luggage storage component, etc.; the ticket service component can include: a flight ticket change / refund component, a train ticket standby component, an electronic boarding pass quick access component, etc., without listing one by one. For example, when the user's flight is delayed, the user can be predicted to need a flight change quick entry, and the flight change quick entry is recommended to the user; when the user is a senior member, the user can be predicted to need a hotel delayed check-out application entry, and the hotel delayed check-out application entry is recommended to the user.

[0050] The task tool refers to a functional entrance or service interface encapsulated for completing a specific and atomic service task, and the task tool can be quickly invoked based on a user instruction or a system decision. In this embodiment, one or more task tools can be dynamically configured for the intelligent travel service according to actual needs. For example, in some embodiments, at least one of a fulfillment tool, a member tool, a weather query tool, a traffic perception tool, and a service recall tool can be configured for the intelligent travel service. The operations performed by the fulfillment tool include, but are not limited to, at least one of the following: querying a fulfillment center to obtain all fulfillment trips of a user, a fulfillment progress of a single fulfillment trip (for example, check-in has been completed, the airport has been arrived, etc.), supplementing a fulfillment trip of a user (for example, supplementing a boarding gate, a check-in time, a flight delay situation of an airplane trip, etc.), querying various information of a city corresponding to a station of a trip, and the like.

[0051] The all fulfillment trips of the user can include fulfillment trips of the user in multiple travel scenarios. The travel scenario refers to a complete scenario unit in which the user interacts with the intelligent travel service and invokes related services to complete a travel target, for example, an airplane travel scenario, a train travel scenario, a hotel accommodation scenario, a taxi scenario, a subway travel scenario, a scenic spot ticket verification scenario, a take-out scenario, and the like. Combining the fulfillment trips of the user in multiple travel scenarios can solve complex needs of the user. For example, a demand of the user to travel from A to B can include the following multiple-scenario fulfillment trips: a train trip from A to B, a taxi trip from a train station of B to a hotel C, an accommodation trip in the hotel C, a visit trip to a scenic spot D of B, and the like.

[0052] The member tool is used to query member information and member exclusive benefits of the user. The weather query tool is used to query weather conditions corresponding to a given target location information. The traffic perception tool is used to obtain traffic information of a current location of the user. The service recall tool is used to recall service components that can be used by the user from multiple service components based on the obtained travel context information, and the recalled service components can be sorted.

[0053] In this embodiment, the type of the trip query operation can be divided according to whether the trip query operation hits a preset shortcut instruction. For example, the type of the trip query operation can include a first query type that hits a shortcut instruction or a second query type that does not hit a shortcut instruction. In some optional embodiments, the first query type can be divided into a sub-query type with smaller granularity according to a specific shortcut instruction hit by the query operation, which is not limited in this embodiment. For example, the trip query operation of the user hits a weather query shortcut instruction, and the sub-query type is to query weather information corresponding to a specified trip. For another example, the trip query operation of the user hits a rest room shortcut instruction, and the sub-query type is to query member benefits corresponding to a specified trip.

[0054] In this step, different types of trip query operations can correspond to different task tool invocation manners. According to the type of the trip query operation, the task tools that need to be invoked can be determined, and these task tools are invoked to obtain the corresponding trip context information. In some optional embodiments, if the trip query operation of the user is the first query type that hits the shortcut instruction, the task tools configured for the first query type can be invoked to obtain the corresponding trip service. If the trip query operation of the user is the second query type that does not hit the shortcut instruction, the planning model can be invoked to plan the task tools that need to be invoked.

[0055] After obtaining the at least one trip context information, in step 103, the at least one dimension of user feature data and the at least one trip context information can be provided to the target language model, so that the target language model infers and summarizes the trip service information of the user.

[0056] In this embodiment, the target language model can be an artificial intelligence (AI) model with natural language processing (NLP) capability, for example, it can be a large language model, which is a natural language processing model trained on a large scale. The large language model is usually constructed based on deep learning technology and trained on a large training data set, so as to have strong performance in processing natural language tasks. Given a piece of text (i.e., context), the large language model will try to predict the most likely next word. The number of model parameters supported by the model is not limited in this embodiment, and the goal is to meet the actual needs. In some optional embodiments, the parameter amount of the large language model is greater than a set threshold, which is usually in the order of millions or hundreds of millions. During the training process, these parameters are constantly adjusted and optimized according to the difference between the prediction results of the large language model and the actual results, so as to improve the prediction accuracy of the large language model. In some embodiments, the large language model usually adopts an advanced neural network architecture such as the Transformer architecture to build the model structure, which enables the large language model to capture complex patterns in text and handle long-distance dependencies. After sufficient training, the large language model has strong generation capability and can output coherent and contextually consistent text content according to the given prompt.

[0057] In this embodiment, a large language model pre-trained on a large number of general data sets can be used as a base model of the target language model. For the specific application scenario of this specification, i.e., for the field of travel services, the pre-trained large language model can be fine-tuned on a large number of data sets in the field of travel services, so that the fine-tuned large language model is more suitable for performing travel service related tasks. Specifically, the large language model can be fine-tuned to enable it to summarize travel service information from input information. The travel service information can include at least one of travel guide information and recommendation information of service components. The travel guide information can be in the form of text, image or video, and can display various travel guides, strategies and other information. The recommendation information of service components is used to recommend service components that the user can use at the moment to the client, so as to quickly meet the potential needs of the user.

[0058] After obtaining the travel service information of the user by using the target language model, in step 104, the travel service information can be returned to the client for display.

[0059] In this embodiment, the intelligent travel service can obtain at least one dimension of user feature data and at least one travel context information of the user triggered by the user's travel query operation, and use the target language model to generate travel service information according to the obtained user feature data and travel context information. The at least one dimension of user feature data is used to enable the target language model to better understand the travel needs of the user, and the at least one context information is used to enable the target language model to consider the influence of different travel related information on travel; the target language model can integrate the above-mentioned various information, actively identify the personalized travel scenario of the user, and mine the potential needs of the user in the personalized scenario, so as to facilitate the provision of accurate travel services for the user.

[0060] In some optional embodiments, the intelligent travel service provides the at least one dimension of user feature data and the at least one travel context information to the target language model, and one way to obtain the travel service information output by the target language model can include: generating a prompt word according to the at least one dimension of user feature data and the at least one travel context information. Optionally, in the case that there is one or more rounds of historical dialogue between the user and the intelligent travel service, the intelligent travel service can also add dialogue context memory data to the prompt word. Optionally, the intelligent travel service can also retrieve knowledge related to travel services from a related knowledge base, and fill these retrieved knowledge in the prompt word to enhance the summarization ability of the target language model.

[0061] After the prompt word is generated, the target language model can provide the prompt word as input information to the target language model. In the target language model, the following steps can be mainly performed: preprocessing the input information, encoding (understanding) the preprocessed input information, decoding (generating) the encoding result, and outputting the decoding result as the travel service information.

[0062] In the preprocessing stage of the input information, the target language model can perform word segmentation on the input information to obtain original text units (i.e., tokens), which can be words, subwords, or characters. The target language model can encode the original text units and their position information to obtain embedding representations of the original text units and positional encodings. In the encoding process, the target language model can map a single token to a high-dimensional vector, which is the embedding representation of the token, for capturing the semantic information of the token. The positional encoding of each token is used to represent the position of the token in the sequence corresponding to the input information, so that the target language model understands the word order. In the encoding stage, the target language model can perform deep semantic understanding on various input information through a multi-layer neural network of the Transformer architecture. Specifically, the multi-layer neural network in the target language model can perform attention calculation on the embedding representations of the original text units and the positional encodings based on an attention mechanism to obtain attention features. For example, the correlation weights of each token with all other tokens can be calculated based on a self-attention mechanism (Self-Attention) to obtain a weight calculation result. The weight calculation result is input into a feedforward network for nonlinear transformation and feature extraction of the weight calculation result to obtain the attention features. As the information is transmitted between layers, the target language model gradually extracts semantic, logical, and contextual relationships from shallow lexical and grammatical features to deep semantic and logical features. In the decoding stage, the target language model can iteratively predict new text units in a self-recurrent manner based on the attention features until the iteration condition is met to output the travel service information. In the decoding process, the attention features are mapped to the vocabulary dimension through a projection layer, and a probability distribution is generated after normalization, which represents the possibility of each token in the vocabulary as the next word. The target language model can select the next token from this probability distribution based on a certain sampling strategy. After generating a token each time, it can be used as input for the next step. In the decoding process, the target language model can focus on the context information output by the encoder through the cross-attention mechanism to ensure that the generated content is related to the input information. In this embodiment, after the target language model is trained and fine-tuned, it can efficiently integrate and understand scattered and heterogeneous multi-source information, such as user preference information in user feature data, user historical trips, weather information, traffic conditions, opening hours of scenic spots, flight / train information, etc. in travel context information, which improves the problem of fragmented service caused by multi-source information. With the collaborative understanding of multi-source information and the strong ability in complex reasoning, the target language model can intelligently generate travel guides that meet the individual needs of users and conform to the actual travel conditions, rather than the "one-size-fits-all" general information, effectively improving the accuracy of travel services.

[0063] In some optional embodiments, the type of the trip query operation can include a first query type of hitting a shortcut instruction or a second query type of missing the shortcut instruction. When the intelligent travel service invokes at least one task tool to obtain at least one travel context information according to the type of the trip query operation, if the trip query operation hits the target shortcut instruction, the at least one task tool and an execution order thereof are determined from the registered task tools according to a tool arrangement manner corresponding to the target shortcut instruction, and the at least one task tool is invoked to obtain the at least one travel context information in the execution order.

[0064] Optionally, the target shortcut instruction can be any one of a plurality of different shortcut instructions. For example, the plurality of different shortcut instructions can include a first shortcut query instruction triggered by a card of a travel channel page and a second shortcut query instruction triggered by an icon of the intelligent travel service. If the trip query operation of the user is triggered by the card of the travel channel page, the target shortcut instruction hit by the trip query operation is the first shortcut query instruction, and if the trip query operation of the user is triggered by the icon of the intelligent travel service, the target shortcut instruction hit by the trip query operation is the second shortcut query instruction. In some optional embodiments, the same tool arrangement manner can be configured for the plurality of different shortcut instructions, or different tool arrangement manners can be configured for the plurality of different shortcut instructions, which can be determined according to a defined travel service policy. For example, the tool arrangement manner corresponding to the target shortcut instruction is to sequentially invoke a fulfillment tool, a member tool, a weather tool, and a service recall tool.

[0065] The registered task tool refers to a functional module or an application programming interface (API) that has been declared, configured, and incorporated into a runtime callable list of the intelligent travel service. These functional modules or APIs can be dynamically invoked by the intelligent travel service or a model associated with the intelligent travel service according to user demand.

[0066] In this implementation, a complex tool invocation operation is encapsulated as a one-key direct shortcut instruction. When the query operation of the user hits the target shortcut instruction, the task tool can be automatically invoked according to the tool arrangement manner corresponding to the target shortcut instruction, so that the user can trigger a preset tool invocation process through simple interaction, greatly simplifying the user operation.

[0067] In some optional embodiments, the tool arrangement mode corresponding to the target shortcut instruction can be dynamically updated. Optionally, the intelligent travel service can obtain the at least one task tool and the execution order thereof configured for the target shortcut instruction, and obtain the tool arrangement mode corresponding to the target shortcut instruction, in response to a tool configuration operation on the target shortcut instruction. The tool configuration operation can be triggered by an operation and maintenance personnel or a technical personnel. Based on this implementation, the task tool and the execution order thereof corresponding to the target shortcut instruction can be flexibly defined through the tool configuration operation, so as to facilitate the flexible formulation of different travel service schemes.

[0068] In some optional embodiments, the registered task tools can be dynamically updated. Optionally, the intelligent travel service can determine the added task tools and / or the task tools to be removed, in response to a service policy update operation. The service policy update operation can be triggered by an operation and maintenance personnel or a technical personnel. If there are added task tools, the intelligent travel service can add the added task tools to the registered task tools. If there are task tools to be removed, the intelligent travel service can deregister the task tools to be removed from the registered task tools, so as to update the registered task tools. Based on this implementation, the task tools can be "hot-plugged" along with the dynamic update of the service policy, so as to flexibly optimize the travel service.

[0069] In some optional embodiments, if the itinerary query operation does not hit the target shortcut instruction, the intelligent travel service can call a planning model to perform task planning according to the user feature data of the at least one dimension and the registered task tools, so as to obtain the at least one task tool and the execution order thereof from the registered task tools, and call the at least one task tool to obtain the at least one travel context information according to the execution order.

[0070] The planning model can be the same as or different from the target language model, and the present embodiment does not make any limitation.

[0071] In some optional embodiments, the planning model can include a main planning model and at least one sub-planning model, and different sub-planning models correspond to different vertical fields. For example, the at least one sub-planning model can include, but is not limited to, at least one of a traffic sub-planning model, a lodging sub-planning model, a ticket sub-planning model, and a catering sub-planning model.

[0072] Optionally, in the process of calling the planning model to perform task planning, the intelligent travel service can call the main planning model to perform task planning according to the user feature data of the at least one dimension and the registered task tools, so as to obtain the at least one task tool and the execution order thereof from the registered task tools.

[0073] Optionally, the intelligent travel service can call the main planning model to perform task planning according to the user characteristic data of the at least one dimension, so as to select a target sub-planning model from the at least one sub-planning model. After determining the target sub-planning model, the intelligent travel service can call the target sub-planning model to perform task planning according to the user characteristic data of the at least one dimension and the registered task tools, so as to obtain the at least one task tool and the execution order thereof from the registered task tools.

[0074] The main planning model and any sub-planning model are obtained by fine-tuning a pre-trained base model. The fine-tuning operation enables the base model to perform task tool planning according to input information and a task tool list. Details are not described herein.

[0075] In some optional embodiments, the planning model can be a large language model, for example, a natural language processing model using a Transformer architecture. When calling the main planning model or any sub-planning model, the intelligent travel service can generate prompt words according to the user characteristic data of the at least one dimension and the information of the registered task tools, and input the prompt words into the main planning model or any sub-planning model. Optionally, in the case of one or more rounds of historical dialogues between the user and the intelligent travel service, the intelligent travel service can also add dialogue context memory data to the prompt words. Optionally, the intelligent travel service can also retrieve knowledge related to travel services from a related knowledge base, and fill these retrieved knowledge into the prompt words to enhance the task planning capability of the planning model.

[0076] After determining the at least one task tool and the execution order thereof based on the above-described embodiments, the at least one task tool can be called in the execution order to obtain the at least one travel context information.

[0077] In some optional embodiments, the at least one task tool arranged in the execution order is: a compliance tool, a member tool, a weather query tool, a traffic perception tool, and a service recall tool.

[0078] Optionally, the intelligent travel service can call the compliance tool to query compliance data according to the user's identification information to obtain information of the user's compliance itinerary. The compliance tool can query the compliance data in a compliance center to obtain all compliance itineraries of the user and the compliance progress thereof. The compliance tool can also query a platform related to the compliance itinerary to complete the related data of the compliance itinerary, for example, query the website of an airline corresponding to an airplane itinerary to complete information such as a check-in counter and a boarding gate. Optionally, the compliance tool can also obtain corresponding station information according to a station associated with the compliance itinerary, for example, merchant preferential activities of a train station, distribution of consumption stores of an airport, distribution of rest rooms, etc.

[0079] Optionally, the intelligent travel service can call a member tool to query member benefit data according to member information of the user to obtain member benefit information of the user. Optionally, the intelligent travel service can call a weather query tool to obtain weather information corresponding to a target location in the itinerary of the user according to information of the target location. The target location can include at least one of a departure location and an arrival location. Optionally, the intelligent travel service can call a traffic perception tool to obtain traffic information of the target location according to information of the target location in the itinerary of the user.

[0080] Optionally, the intelligent travel service can call a service recall tool to recall at least one candidate service component from a plurality of service components according to at least one of information of the itinerary of the user, the member benefit information, the weather information, and the traffic information, and determine service candidate information according to the at least one candidate service component.

[0081] Based on this implementation, the intelligent travel service can obtain multi-dimensional travel context information from external platforms or internal platforms by using at least one task tool, so that the intelligent travel service can perceive these travel context information, thereby providing more accurate travel services. In addition, the service recall tool can recall one or more candidate service components, so that the intelligent travel service can predict the needs of the user in advance, realizing the change from "passive operation" to "active accompanying".

[0082] In some optional embodiments, when the intelligent travel service calls the service recall tool to recall at least one candidate service component from a plurality of service components according to information of the itinerary of the user, the service recall tool can determine a travel node of the user according to information of the itinerary of the user. The travel node refers to a key stage identifier set based on a time axis in the user's travel process, which is used to divide different service stages of the itinerary to provide appropriate services for the user at the right time. For example, the travel node can be 2 hours before departure, 30 minutes before arrival, etc. The intelligent travel service can call the service recall tool to recall at least one candidate service component adapted to the travel node from the plurality of service components. For example, when the travel node is 2 hours before departure, the service recall tool can recall a taxi service component, a station real scene navigation service component, a VIP lounge benefit reminder component, etc. For another example, when the travel node is 30 minutes before arrival, the service recall tool can recall a taxi service component, a hotel check-in application component, a take-out service component, etc.

[0083] Based on this implementation, the service recall tool can recall appropriate service components at the right time, thereby facilitating the intelligent travel service to push services related to the travel node to the user to do the right thing at the right time.

[0084] In some optional embodiments, optionally, the intelligent travel service, when calling the service recall tool to determine the service candidate information according to the at least one candidate service component, can further obtain the travel situation of the user according to the traffic information and / or the weather information. The travel situation is used to describe dynamic information that affects travel due to traffic congestion, weather factors, and the like. For example, the travel situation is: according to the traffic information, it is estimated that 45 minutes are needed to reach the airport; or the route to be taken is affected by heavy rain, and it is estimated that 2 hours are needed to reach the train station. The service recall tool determines the recommended priority of the at least one candidate service component adapted to the travel node according to the travel situation, and determines the service candidate information according to the at least one candidate service component and the recommended priority thereof. For example, when the time length from the current time to reach the airport is greater than the time length from the current time to the takeoff of the airplane, the priority of the ticket change component can be improved.

[0085] Based on this manner, different demands of the user can be met under different travel situations, so that on the one hand, solutions can be provided in time when the user faces problems, and the click rate is improved; on the other hand, the service components recalled are sorted, and information overload can be effectively reduced, and the user experience is improved.

[0086] The embodiments of the present specification also provide a travel information processing method that can be executed by a terminal device. Figure 2 is a flowchart of the travel information processing method provided by the embodiments of the present specification when executed on the client side, as Figure 2 indicated, the method mainly includes:

[0087] Step 201, in response to a travel query operation of a user, a travel query request is sent to an intelligent travel service, the travel query request is used to: make the intelligent travel service obtain user feature data of at least one dimension of the user, and according to the type of the travel query operation, at least one travel context information is obtained by calling at least one task tool; and the user feature data of the at least one dimension and the at least one travel context information are provided to a target language model, and travel service information output by the target language model is obtained, the travel service information includes: travel guidance information and / or recommended information of a service component.

[0088] Step 202, the travel service information returned by the intelligent travel service is obtained, and the travel service information is displayed on a dialogue page with the intelligent travel service.

[0089] In this embodiment, the client can integrate a front-end module capable of accessing the intelligent travel service, or the client can access the intelligent travel service through a mini-program, a webpage or an HTML5 (HyperText Markup Language 5) page, which is not limited in this embodiment. The client can provide an access portal and a visual dialogue page of the intelligent travel service, so that the intelligent travel service provides travel-related services to the user in the form of dialogue.

[0090] The client can detect the user's itinerary query operation in various ways. In some optional embodiments, the client can display a fulfillment notification message of the target fulfillment itinerary of the user on a first page, and send an itinerary query request corresponding to the target fulfillment itinerary to the intelligent travel service in response to a triggering operation on the fulfillment notification message. The first page can be the home page, the travel channel page, the order page or the message center page of the client, which is not limited in this embodiment. The fulfillment notification message can be displayed in the form of a card, such as the fulfillment notification card shown in Figure 3 In the travel scenario, the fulfillment notification card is an interactive interface component in the form of a card for delivering service execution status and key information to the user. The fulfillment notification card can structure the key information (such as time, location, status, operation button, jump link, etc.) in the fulfillment process in a visually independent card component to efficiently display itinerary-related information. If the client detects a selection operation on the fulfillment itinerary card, it is determined that the user's itinerary query operation is detected.

[0091] It is worth noting that in some embodiments, when the first page is a travel channel page, the travel channel page can include multiple sub-channel pages, and different sub-channel pages correspond to different travel modes. The travel channel page can display entry icons of sub-channel pages corresponding to multiple different travel modes, such as the entry icons of sub-channel pages corresponding to multiple travel modes of taxi, hitchhiking, train, flight, hotel, cycling and bus ticket shown in Figure 3 If the user has a fulfillment itinerary of a certain travel mode, the client can display a fulfillment notification card corresponding to the travel mode in the sub-channel page corresponding to the travel mode. For example, the client can display a fulfillment notification message of the target fulfillment itinerary of the user in the sub-channel page corresponding to the target fulfillment itinerary. As shown in Figure 3 The client can display a fulfillment notification card corresponding to the airplane itinerary in the sub-channel page corresponding to the flight. Of course, if the user has a train itinerary, the client can display a fulfillment notification card corresponding to the train itinerary in the sub-channel page corresponding to the train, which is not illustrated.

[0092] In some alternative embodiments, the client may display an icon of the smart travel service on a second page and, in response to a triggering operation on the icon, send a trip query request to the smart travel service. This second page can be the client's homepage, travel channel page, order page, or message center page; this embodiment is not limited to this. For example, such as... Figure 4 As shown, the second page can be a travel channel page, with a bubble-style "AI Travel Assistant" entry icon displayed at the bottom for accessing smart travel services. If the client detects a trigger action on the smart travel service icon, it confirms that a user's trip query operation has been detected.

[0093] Optionally, when the second page is the travel channel page, the travel channel page can contain multiple sub-channel pages, each corresponding to a different travel mode. The travel channel page can include entry icons for multiple sub-channel pages corresponding to various travel modes, such as... Figure 4 The image shows entry icons for various travel modes, including taxis, ride-sharing, trains, air tickets, hotels, cycling, and bus tickets, each corresponding to a sub-channel page. The client can display the icon of this smart travel service on any target sub-channel page on the second page and respond to user actions triggered by the icon on any target sub-channel page by sending a trip query request to the smart travel service. This trip query request can include a query request for fulfilled trips under one or more travel modes, and these one or more travel modes must at least include the travel mode corresponding to the target sub-channel page. For example... Figure 4 As shown, when the client displays the entry icon of the AI ​​travel assistant in the sub-channel page corresponding to the air ticket, it can respond to the user's trigger operation on the icon of the smart travel service in the sub-channel page corresponding to the air ticket, and send a trip query request associated with the air travel mode to the smart travel service, or send a trip query request associated with the air travel mode and a trip query request associated with at least one other travel mode besides air.

[0094] In some alternative embodiments, the client may display a dialogue page with the smart mobility service. If a user-inputted trip query message is detected on the dialogue page, it is determined that a user's trip query operation has been detected, and the client sends the trip query request to the smart mobility service based on the trip query message. The client may display the trip query message on the dialogue page, and the trip query message and the travel service information returned by the smart mobility service form a round of dialogue. The dialogue between the user and the smart mobility service can continue for multiple rounds, and the processing method for each round is the same or similar, which will not be described further.

[0095] The alternative implementation method for intelligent travel services to return travel service information based on the itinerary query request can be referred to the description in the foregoing embodiments, and will not be described again here.

[0096] The travel service information returned by the intelligent travel service can be displayed on the conversation page with the intelligent travel service. As shown in Figure 5 , the travel service information related to the user's flight from city A to city B can be displayed on the conversation page with the intelligent travel service for the user to view.

[0097] It is worth noting that in some embodiments, the user can initiate the trip query operation through any sub-channel page in the travel channel page. In some embodiments, when the user has a plurality of fulfillment trips under different travel modes, the intelligent travel service can return travel service information of the fulfillment trips under different travel modes. Alternatively, the client determines the target sub-channel page in which the user initiates the trip query operation from a plurality of sub-channels included in the travel channel page. If the travel service information includes travel service information of the fulfillment trips under different travel modes, the client can determine the priority of the fulfillment trips under different travel modes according to the travel mode corresponding to the target sub-channel page, and display the travel service information of the fulfillment trips under different travel modes according to the priority.

[0098] Alternatively, if the user initiates the trip query operation on the target sub-channel page in the travel channel page, the client can take the fulfillment trip under the travel mode corresponding to the target sub-channel as the highest priority fulfillment trip, and preferentially display the travel service information of the fulfillment trip. As shown in Figure 3 and Figure 4 , assuming that the user initiates the trip query operation by clicking on the fulfillment notification card or the entry icon of the AI travel assistant on the sub-channel page corresponding to the ticket, the client can preferentially display the travel service information of the user's flight after obtaining the travel service information of the fulfillment trips under different travel modes.

[0099] Alternatively, if the travel service information includes travel service information of the fulfillment trips under different travel modes, the client can superimpose information cards of the fulfillment trips under different travel modes on the conversation page, and the first-layer information card is the information card of the fulfillment trip under the travel mode corresponding to the target sub-channel page, and the client can switch the display of the information cards of the plurality of fulfillment trips according to the user's left swipe and right swipe operations.

[0100] Based on this implementation, the main query intention of the user can be determined by detecting the sub-channel page in which the user initiates the trip query operation, so that the travel service information strongly associated with the query intention can be preferentially displayed for the user, and the query demand of the user can be timely met.

[0101] In this embodiment, a client's trip query request can trigger the intelligent travel service to acquire user feature data of at least one dimension and at least one travel context information for the user. The service then uses a target language model to generate travel service information based on the acquired user feature data and travel context information. Specifically, the user feature data of at least one dimension enables the target language model to gain a deeper understanding of the user's travel needs, while the context information enables the target language model to consider the impact of different travel-related information on the trip. The target language model can integrate these multiple pieces of information to proactively identify the user's personalized travel scenarios and uncover the user's potential needs within those personalized scenarios, thereby facilitating the provision of accurate travel services to the user.

[0102] The following will combine Figure 6 The flowchart shown further illustrates the travel information processing method provided in the embodiments of this specification. For example, users can initiate a trip query operation through various methods. Figure 6 As shown, users can access the travel service by clicking the fulfillment notification card on the travel channel homepage to initiate a trip query, or by clicking the smart travel service icon, or by initiating a query conversation on the smart assistant dialogue page to query existing fulfilled trips. After entering the travel service, users can initiate a conversation or not. If a user initiates a conversation, the smart travel service needs to perform a security check on the information entered by the user. Optionally, if the user has historical conversations with the smart travel service, the conversation context memory data can also be loaded. Figure 6 As shown, the intelligent mobility service can perform information collection operations, primarily to collect user characteristic data in at least one dimension, such as user travel preferences and membership information. When a user initiates a conversation, the intelligent mobility service can also recall task tools related to the content of that conversation.

[0103] like Figure 6 As shown, the intelligent travel service can determine whether a user's trip query operation has triggered a shortcut. If a user initiates a trip query by clicking the fulfillment notification card on the travel channel homepage, or by clicking the intelligent travel service icon, then the user's trip query operation has triggered a shortcut. If the trip query operation has triggered a shortcut, the relevant task tool is invoked according to the tool arrangement corresponding to the shortcut to obtain the relevant travel context information. The shortcut supports pluggable configuration of one or more task tools, and their execution order can be arranged.

[0104] like Figure 6As shown, the travel service related task tools can include: a compliance tool, a member tool, a weather query tool, a service recall tool, a traffic perception tool, etc. The above task tools support dynamic loading or unloading to ensure timely service policy updates.

[0105] Among them, the service recall tool can recall scenario-based service components (such as components such as taking a taxi, station real scene navigation, and VIP hall benefit reminder) based on the user's current travel node (such as 2 hours before departure, 30 minutes before arrival, etc.), and can dynamically adjust the priority of the recalled service components in combination with the current travel situation, for example, preferentially pushing the rescheduling entry in the event of sudden congestion. Thus, the user can be dynamically recommended a more appropriate service at the current time, improving the end-to-end experience.

[0106] As shown, Figure 6 If the user's travel query operation does not hit the shortcut instruction, the intelligent travel service can call the planning model to perform overall task planning. In this implementation, the user can specify a decision model or a deep reasoning model as the planning model. The intelligent travel service can construct a prompt word according to the user feature data of at least one dimension and the information of the registered task tools, and input the prompt word into the planning model. Optionally, the prompt word can also include at least one of the dialogue context memory data, the historical planning tasks and the task tools planned thereby, the execution process of the task tools, and the execution results of the task tools, for the planning model to use for reference, and the present embodiment does not limit. As shown, Figure 6 As shown, the task planning is a process of circular processing, and if the maximum number of cycles is reached or the task state is an end state, the task planning operation is stopped.

[0107] As shown, Figure 6 The planning model includes a general main planning model and a vertical field sub-planning model. The main planning model can be used to perform task planning first, and the task planning process of the main planning model includes: thinking, action, execution, etc. In addition, the task planning process of the main planning model also includes: asking the user a counter-question to further clarify the user's intention, and after performing the action, it can also be observed whether the execution result meets the requirements. In some simple task processing scenarios, the action determined by the main planning model through reasoning thinking is a call action to the task tool, and the main planning model can directly execute the call action to the task tool and observe the execution result. In some complex task processing scenarios, as shown, Figure 6 The action determined by the main planning model through reasoning thinking is a call action to the vertical field sub-planning model. In this case, the main planning model can call the sub-planning model to perform task planning until the maximum number of cycles is reached or the task state is an end state. Among them, the execution stage of the main planning model and the sub-planning model can concurrently call the task tool as needed to improve planning efficiency.

[0108] After the task tool is called to obtain one or more travel context information, the intelligent travel service can convert the travel context information into natural language, and call a target language model to summarize the travel context information converted into natural language. The target language model can be a quick decision model or a deep reasoning model selected by the user. The intelligent travel service can generate a prompt word according to the user feature data of at least one dimension of the user and at least one travel context information, and provide the prompt word to the target language model. The prompt word can also include dialogue context memory data of the user and the intelligent travel service, travel knowledge data, etc. The target language model can generate travel service information according to the input prompt word, and the travel service information can include travel guides and dynamic service candidate information. The intelligent travel service can output the travel service information as the dialogue content with the user, record the current dialogue into the dialogue context memory data, and facilitate subsequent dialogue consumption and perception.

[0109] In the above embodiments, the intermediate processes and results of each link, such as the information collected by the intelligent travel service, the thinking content of the planning model, the counter-question content of the target language model, the intelligent service card, etc., can be assembled into corresponding protocol cards for streaming output and dynamic update. Before output, the intelligent travel service can perform security checks on the output content, and can display the output content according to the card style customized by the user.

[0110] It should be noted that the execution subject of each step of the method provided in the above embodiments can be the same device, or the method can also be executed by different devices as the execution subject. For example, the execution subject of steps 101 to 104 can be device A; for another example, the execution subject of steps 101 and 102 can be device A, and the execution subject of step 103 can be device B; and so on.

[0111] In addition, in some of the processes described in the above embodiments and the accompanying drawings, a plurality of operations appearing in a specific order are included, but it should be clearly understood that these operations can be executed or in parallel without the order in which they appear in this text. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the "first", "second", etc. described herein are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence. Also, "first" and "second" are not of different types.

[0112] In this specification, unless explicitly stated otherwise, "receiving and sending data" does not necessarily mean direct receiving and sending; it can also mean indirect receiving and sending. For example, A receiving data sent by B can be understood as A directly receiving the data sent by B, or it can be understood as A indirectly receiving the data sent by B through other entities such as C. Similarly, B sending data to A can be understood as B sending the data directly to A, or it can be understood as B indirectly sending the data to A through other entities such as C. Here, C can be one entity, or it can be two or more entities.

[0113] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this manual are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation portals are provided for users to choose to authorize or refuse.

[0114] Figure 7 This specification illustrates a schematic diagram of the structure of an electronic device provided in an exemplary embodiment, which is applicable to the travel information processing method provided in the foregoing embodiments. Figure 7 As shown, the electronic device 700 mainly consists of a communication interface 702, a user interface 704, a processor 706, and a memory 708. These components are interconnected and communicate with each other through a system bus, network, or other connection mechanism 710. The communication interface 702 enables the device 700 to communicate with other devices, access networks, and transmission networks via analog or digital modulation. For example, the communication interface 702 may include a chipset and antenna for wireless communication with a radio access network or access point. Furthermore, the communication interface 702 can be a wired interface such as Ethernet, Token Ring, or a USB port, or a wireless interface such as Wi-Fi (Wireless Fidelity), Bluetooth, Global Positioning System (GPS), or wide-area wireless interface such as WiMAX (Wireless Maximum Interoperability) or LTE (Long Term Evolution). Of course, the communication interface 702 can also support other forms of physical layer interfaces and standard or proprietary communication protocols. The communication interface 702 may also include multiple physical communication interfaces, such as a Wi-Fi interface, a Bluetooth interface, and a wide-area wireless interface.

[0115] The user interface 704 includes input components such as a keypad, keyboard, touch- sensitive or presence-sensitive panel, computer mouse, trackball, joystick, microphone, still camera, and video camera, among other possibilities. The user interface 704 also includes output components such as a display screen (which can be combined with a touch-sensitive panel), a CRT, LCD, LED, display, a printer, other suitable output devices, and the like. The user interface 704 can also generate audible outputs through a loudspeaker, a loudspeaker cone, a audio output port, audio output device, earphones, and the like. In some embodiments, the user interface 704 can include software, circuits, or other forms of logic that enable the external user input / output devices to transmit data to and receive data from the user interface 704. Additionally or alternatively, the electronic device 700 can support remote access from other devices through the communication interface 702 or another physical interface (not shown). The user interface 704 can be configured to receive user input, the location and movement of which can be indicated by a pointer or cursor as described herein. The user interface 704 can also be configured as a display device for rendering or displaying a text segment.

[0116] The processor 706 can include one or more general-purpose processors and / or special- purpose processors. The memory 708 can include one or more volatile and / or non-volatile storage components, and can be integral to, or separate from, the processor(s) 706. The memory 708 can include removable and / or non-removable components.

[0117] The processor 706 is capable of executing program instructions stored in the memory 708 (e.g., compiled or interpreted

[0118] Memory 708 can include a non-transitory computer-readable medium, such as static random-access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), a magnetic storage, a flash memory, a magnetic or optical disk, or another similar storage medium. Stored on memory 708 are program instructions 718. The program instructions 718 are designed to cause processor 706 to perform any of the methodologies, processes, or functions disclosed in the specification and / or the accompanying figures. The program instructions 718 can cause processor 706 to utilize data 712.

[0119] For example, program instructions 718 can include an operating system 722 (e.g., an operating system kernel, device drivers, and / or other modules) installed on device 700, as well as one or more applications 720 (e.g., a browser, a social application, or a gaming application). Similarly, data 712 can include operating system data 716 and application data 714. Operating system data 716 is primarily accessible to operating system 722, while application data 714 is primarily accessible to one or more applications 720. Application data 714 can be located in a file system that is visible or hidden to a user of device 700.

[0120] Applications 720 can communicate with operating system 722 through one or more application programming interfaces (APIs). These APIs facilitate applications 720 reading and / or writing application data 714, communicating or receiving information via communication interface 702, receiving or displaying information on user interface 704, and the like.

[0121] In some terminology, applications 720 can be referred to simply as “apps.” Furthermore, applications 720 can be downloaded to device 700 through one or more online application stores or application markets. However, applications can also be installed on device 700 through other means, such as through a web browser or a physical interface on electronic device 700 (e.g., a USB port).

[0122] Accordingly, the embodiments of the present specification also provide a computer readable storage medium storing a computer program, which causes the processor to be able to implement each step in the above method embodiments when the computer program is executed by the processor. The computer readable storage medium includes volatile or non-volatile or a combination thereof, and can be removable or non-removable. Examples of the computer readable storage medium include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital video disc (DVD) or other optical storage, magnetic cassette, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium

[0123] Accordingly, the embodiments of the present specification also provide a computer program product, which includes a computer program or instructions, which causes the processor to be able to implement each step in the above method embodiments when the computer program or instructions are executed by the processor. It should be understood that each process or a combination of multiple processes in the above method flow can be implemented by the computer program or instructions. In addition, these computer programs or instructions can be applied to the processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing devices, so that the processor of the general-purpose computer, the special-purpose computer, the embedded processor or other programmable data processing devices can be implemented as a device for implementing the corresponding functions in the above method embodiments.

[0124] It should also be noted that the terms "comprising", "including", or any other variant thereof are intended to cover non-exclusive inclusion, so that processes, methods, products or devices including a series of elements not only include those elements, but also include other elements not explicitly listed, or further include elements inherent in such processes, methods, products or devices. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, product or device including the element.

[0125] The above merely provides the example of the present application, but does not serve to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application should be included in the scope of claims of the present application.

Claims

1. A method for processing travel information, characterized in that, include: In response to a trip query request sent by the client based on the user's trip query operation, obtain user feature data of at least one dimension of the user; Based on the type of the trip query operation, at least one task tool is invoked to obtain at least one type of travel context information; The user feature data of at least one dimension and the travel context information of at least one dimension are provided to the target language model to obtain the travel service information output by the target language model. The travel service information includes: travel guidance information and / or service component recommendation information. The travel service information is sent to the client for display.

2. The method according to claim 1, characterized in that, Depending on the type of the trip query operation, at least one task tool is invoked to obtain at least one type of travel context information, including: If the trip query operation matches the target shortcut, then according to the tool arrangement method corresponding to the target shortcut, at least one task tool and its execution order are determined from the registered task tools; According to the execution order, the at least one task tool is invoked to obtain the at least one travel context information.

3. The method according to claim 2, characterized in that, Also includes: If the trip query operation does not match the target shortcut, the planning model is invoked to perform task planning based on the user feature data of at least one dimension and the registered task tools, so as to obtain the at least one task tool and its execution order from the registered task tools; According to the execution order, the at least one task tool is invoked to obtain the at least one travel context information.

4. The method according to claim 3, characterized in that, Invoking a planning model to perform task planning based on user characteristic data of at least one dimension, resulting in at least one task tool and its execution order, including: Invoke the master planning model to perform task planning based on the user feature data of at least one dimension and the registered task tools, so as to obtain the at least one task tool and its execution order from the registered task tools; or, The main planning model is invoked to perform task planning based on user feature data of at least one dimension, so as to select a target sub-planning model from the at least one sub-planning model, with different sub-planning models corresponding to different vertical domains; and the target sub-planning model is invoked to perform task planning based on user feature data of at least one dimension and the registered task tools, so as to obtain the at least one task tool and its execution order from the registered task tools.

5. The method according to any one of claims 2-4, characterized in that, According to the execution order, the at least one task tool is invoked to obtain the at least one travel context information, including at least one of the following: The fulfillment tool is invoked to query the fulfillment data based on the user's identification information, thereby obtaining the user's fulfillment itinerary information; The membership tool is invoked to query the membership benefits data based on the user's membership information, thereby obtaining the user's membership benefits information; The weather query tool is invoked to obtain the weather information corresponding to the target location based on the information of the target location in the user's fulfillment itinerary; The traffic sensing tool is invoked to obtain traffic information at the target location based on the information of the target location in the user's fulfillment journey; The service recall tool is invoked to recall at least one candidate service component from multiple service components based on at least one of the user's fulfillment itinerary information, membership benefits information, weather information, and traffic information, and service candidate information is determined based on the at least one candidate service component.

6. The method according to claim 5, characterized in that, The service recall tool is invoked to recall at least one candidate service component from a variety of service components based on the user's fulfillment schedule information, including: The service recall tool is invoked to determine the user's travel nodes based on the user's fulfillment travel information; Based on the user's itinerary node, at least one candidate service component that is compatible with the itinerary node is recalled from the multiple service components.

7. The method according to claim 6, characterized in that, Based on the at least one candidate service component, service candidate information is determined, including: Based on the traffic information and / or weather information, the user's travel status is obtained; Based on the travel pattern, determine the recommendation priority of at least one candidate service component that is compatible with the travel node; The service candidate information is determined based on the at least one candidate service component and its recommendation priority.

8. The method according to any one of claims 2-4, characterized in that, Also includes: In response to the tool configuration operation of the target shortcut, the system obtains the at least one task tool configured for the target shortcut and its execution order, thereby obtaining the tool arrangement method corresponding to the target shortcut.

9. The method according to any one of claims 2-4, characterized in that, Also includes: In response to service policy update operations, identify newly added task tools and / or task tools to be removed; Add the newly added task tool to the registered task tools, and / or unregister the task tool to be removed from the registered task tools to update the registered task tools.

10. The method according to any one of claims 1-4, characterized in that, The user feature data of at least one dimension and the travel context information of at least one dimension are provided to the target language model to obtain travel service information output by the target language model, including: Based on the user feature data of at least one dimension and the travel context information of at least one dimension, generate prompt words; The prompt words are provided as input information to the target language model; In the target language model, the input information is segmented into words to obtain original text units; The original text unit and its position information are encoded to obtain the embedded representation and position encoding of the original text unit; Based on the attention mechanism, attention calculation is performed on the embedding representation and position encoding of the original text unit to obtain attention features; Based on the attention features, new text units are predicted iteratively in an autoregressive manner until the iteration conditions are met, at which point the travel service information is output.

11. The method according to any one of claims 1-4, characterized in that, The user feature data of at least one dimension includes at least one of the user's travel preference information and the user's membership information.

12. A method for processing travel information, characterized in that, include: In response to a user's trip query operation, a trip query request is sent to the intelligent travel service. The trip query request is used to: enable the intelligent travel service to obtain user feature data of at least one dimension of the user, and, according to the type of the trip query operation, call at least one task tool to obtain at least one travel context information. In addition, the user feature data of the at least one dimension and the at least one travel context information are provided to the target language model to obtain travel service information output by the target language model, wherein the travel service information includes: travel guidance information and / or service component recommendation information; Obtain the travel service information returned by the intelligent travel service, and display the travel service information on the dialogue page with the intelligent travel service.

13. The method according to claim 12, characterized in that, In response to a user's trip query request, a trip query request is sent to the intelligent travel service, including: The first page displays the fulfillment notification message for the user's target fulfillment trip, and in response to the triggering operation of the fulfillment notification message, sends a trip query request corresponding to the target fulfillment trip to the smart travel service; or, The second page displays the icon of the smart travel service and, in response to a triggering operation on the icon, sends the trip query request to the smart travel service; or, The system detects the user's input trip query message on the dialogue page with the smart travel service and sends the trip query request to the smart travel service based on the trip query message.

14. The method according to claim 13, characterized in that, The first page or the second page is a travel channel page, which contains multiple sub-channel pages, and different sub-channel pages correspond to different travel methods; The travel service information is displayed on the dialogue page with the intelligent travel service, including: From the multiple sub-channels included in the travel channel page, determine the target sub-channel page from which the user initiated the trip query operation; If the travel service information includes travel service information for each of the various travel modes, then the priority of the various travel modes' fulfillment trips is determined based on the travel mode corresponding to the target sub-channel page. Based on the stated priority, the travel service information for each of the various travel modes is displayed.

15. An electronic device, characterized in that, include: Memory and processor; The memory is used to store one or more computer instructions; The processor is configured to execute one or more computer instructions for performing the steps of the method according to any one of claims 1-14.

16. A computer-readable storage medium storing a computer program, characterized in that, When a computer program is executed by a processor, it is able to perform the steps of the method described in any one of claims 1-14.

17. A computer program product, characterized in that, include: A computer program / instruction that, when executed by a processor, enables the implementation of the steps in the method described in any one of claims 1-14.

Images