Conversational express ordering method, device, equipment, medium and product

Abstract

The application provides a conversational express ordering method, device, equipment, medium and product. It relates to the fields of artificial intelligence and logistics technology. The method comprises the following steps: determining a user target intention based on user input information of the current round of conversation, wherein the user input information comprises picture content and / or natural language text content; extracting corresponding field content from the user input information of the current round of conversation according to a preset field extraction rule and storing the field content into order state data in combination with a conversation context; performing integrity verification on the current order state data to generate a field to be filled in prompt message; in response to the order state data passing the integrity verification, generating an order information completeness prompt message, and storing the receiver information and the sender information into a user historical information library. Through the maintenance of the order state data, the application realizes progressive information collection, completes the multi-round conversation ordering process, and improves the overall ordering efficiency and user experience.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence and logistics technology, and in particular to a conversational express delivery ordering method, apparatus, equipment, medium and product. Background Technology

[0002] Against the backdrop of the continued rapid development of cross-border e-commerce, the demand for international express delivery is showing a rapid growth trend. When placing an order on an international express delivery platform, users need to fill in complete information such as the names, addresses, phone numbers, postal codes of both the sender and recipient, as well as the name, quantity, weight, and value of the goods. This process is cumbersome and time-consuming. Due to differences in international address standards, the colloquial nature of user expressions, and the segmented provision of information, manual entry is prone to errors, omissions, or non-standard information. At the same time, international express delivery involves complex rules such as acceptance standards, shipping costs, delivery times, and customs policies in various countries, resulting in high costs for querying and verification. Traditional interaction methods are difficult to meet users' needs for convenient and intelligent order placement.

[0003] Current international express ordering systems primarily employ solutions such as question-and-answer FAQ interaction, fixed form filling, keyword and regular expression matching for information extraction, static rule base validation, and single-modal text interaction. The FAQ system responds to user inquiries based on preset rules, the form system guides users to enter information item by item according to fixed fields, keywords and regular expressions are used to identify key information such as address and item, the rule base carries acceptance and delivery restrictions and shipping cost calculation logic, and the system generally only supports plain text input interaction.

[0004] However, existing solutions struggle to extract multidimensional structured information from conversational and fragmented input, and do not support mixed text and image multimodal parsing. Multi-round, progressive information cannot be effectively integrated, leading to information overlay or loss, resulting in cumbersome order placement processes and poor user experience. Summary of the Invention

[0005] The dialog-based express ordering method, apparatus, equipment, medium, and product provided in this application are used to achieve rapid and accurate determination of dual-purpose items.

[0006] In a first aspect, embodiments of this application provide a conversational express delivery ordering method, the method comprising:

[0007] The user's target intent is determined based on the user input information in this round of dialogue, including image content and / or natural language text content;

[0008] If the user's target intent is to place an order, the corresponding field content is extracted from the user's input information in this round of dialogue according to the preset field extraction rules and stored in the order status data in combination with the dialogue context;

[0009] Completeness verification is performed based on the current order status data to generate a message prompting users to fill in fields that need to be filled in. This message prompts users to supplement their order information.

[0010] In response to the order status data passing the integrity verification, a complete order information prompt message is generated. The complete order information prompt message is used to prompt the user to confirm the order information and complete the order, and the recipient information and sender information are stored in the user's historical information database.

[0011] In one possible implementation, the user input information includes image content and natural language text content, and the step of determining the user's target intent based on the user input information in the current round of dialogue further includes:

[0012] A preset extraction algorithm is used to extract text content from the image, and then concatenated with the natural language text content to generate the target query content;

[0013] The intent classification result of the target query content is determined based on a preset prompt word template. The intent classification result includes the target intent identifier and the corresponding confidence score.

[0014] The target query content is identified as a pending node based on the target intent identifier and the corresponding confidence score. The pending node is used to execute the processing flow for different intents.

[0015] In one possible implementation, the step of determining the node to be processed for the target query content based on the target intent identifier and the corresponding confidence score includes:

[0016] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the first preset identifier, then the target query content is rewritten based on the historical dialogue.

[0017] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the second preset identifier, then the corresponding query interface or preset knowledge base will be called to perform a query based on the target query content, and a response will be given based on the query results.

[0018] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the third preset identifier, then modify the field content in the order status data based on the historical dialogue.

[0019] If the confidence score is less than the preset confidence threshold, a list of query questions is generated based on the target query content. The list of query questions is used to determine the target intent.

[0020] In one possible implementation, the step of extracting corresponding field content from the user input information of the current round of dialogue according to preset field extraction rules and storing it in the order status data in conjunction with the dialogue context includes:

[0021] Determine the type of information contained in the user input;

[0022] If the information type is recipient information or sender information, the corresponding field content is extracted from the user input information according to the preset fill fields. The field content includes the recipient's or sender's personal information and address information, and the personal information and address information are mapped to standard fields.

[0023] If the information type is item information, the corresponding field content is extracted from the user input information. The field content includes the item name and item attributes, and the acceptance and delivery standards are queried based on the item name.

[0024] The extracted field content is compared with the corresponding field content already stored in the current order status data;

[0025] If the extracted field content is inconsistent with the stored content or the corresponding field is empty, the extracted field content will be stored in the order status data.

[0026] In one possible implementation, the information type is recipient information, and the method further includes:

[0027] Retrieve default sender information from the user's historical information database and store it in the order status data;

[0028] The system matches the recipient information included in the user's input with the information from the user's historical information database.

[0029] If the matching similarity is greater than the first similarity threshold, the information content included in the matching result will be stored in the order status data;

[0030] If the matching similarity is greater than or equal to the second similarity threshold and less than or equal to the first similarity threshold, then a candidate recipient list is output, which is used to confirm the target recipient information.

[0031] If the matching similarity is less than the second similarity threshold, a prompt message for the fields to be filled in will be output.

[0032] In one possible implementation, it also includes:

[0033] The order status data is stored using a document-based database with checkpoints.

[0034] In response to a user reconnecting after interrupting the conversation, the order status data is restored based on the session identifier.

[0035] In one possible implementation, it also includes:

[0036] Perform error detection on order status data;

[0037] If an error in the field format or a logical mismatch is detected in the order status data, the content of the erroneous field should be cleared.

[0038] Generate a message indicating the fields to be filled in, including error messages.

[0039] Secondly, embodiments of this application provide a conversational express delivery ordering device, comprising:

[0040] The determination module is used to determine the user's target intent based on the user input information in the current round of dialogue, wherein the user input information includes image content and / or natural language text content;

[0041] The extraction module is used to extract the corresponding field content from the user input information in the current round of dialogue according to the preset field extraction rules if the user's target intent is to place an order.

[0042] The storage module is used to store the extracted field content into the order status data in conjunction with the dialogue context;

[0043] The generation module is used to perform integrity verification based on the current order status data to generate a prompt message for fields to be filled in. The prompt message for fields to be filled in is used to prompt the user to supplement the order information.

[0044] The generation module is further configured to generate a complete order information prompt message in response to the order status data passing the integrity check. This complete order information prompt message is used to prompt the user to confirm the order information and complete the order.

[0045] The storage module is also used to store the recipient information and sender information into the user's historical information database.

[0046] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0047] The memory stores computer-executed instructions;

[0048] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0049] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0050] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0051] The conversational express ordering method, apparatus, device, medium, and product provided in this application support mixed text and image input, and are compatible with real-world usage scenarios such as user screenshots, chat logs, and photographed addresses, thus broadening the interaction methods. It proactively identifies the user's true intent, distinguishing between different needs such as ordering, querying, modifying, and canceling, avoiding invalid processes and improving system response efficiency. If the intent is to place an order, it automatically extracts receiving and shipping information and item information from colloquial and fragmented information in a structured manner, eliminating the need for manual form filling by the user. Order status data is uniformly stored, achieving centralized management of information throughout the process, ensuring no loss or confusion. It supports multi-round progressive input, allowing users to provide information in stages while storing it segment by segment, avoiding information overwriting. Based on the current order status data, i.e., the already filled information, it dynamically calculates missing fields, only prompting the user for unfilled content, eliminating the need for repeated confirmation. It achieves intelligent guided ordering, reducing user understanding costs and operation steps. It improves information completeness, avoiding order failures due to missing fields. After the order data passes integrity verification, the receiving and shipping information is stored in the user's historical information database. This completes the automatic persistent storage of information, forming a user-specific address memory database. Subsequent use allows for fuzzy matching and autofill, significantly reducing repetitive input. This improves order efficiency for existing users, providing a smarter experience the more you use it. Attached Figure Description

[0052] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0053] Figure 1 An application scenario diagram of a conversational express delivery ordering method provided in this application;

[0054] Figure 2 A flowchart illustrating a conversational express delivery ordering method provided in an embodiment of this application;

[0055] Figure 3 A signaling interaction flowchart for a conversational express delivery ordering method provided in another embodiment of this application;

[0056] Figure 4 A schematic diagram of the structure of a conversational express delivery ordering device provided in an embodiment of this application;

[0057] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0058] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0059] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0060] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0061] To clearly understand the technical solution of this application, the solutions of the prior art will be described in detail first.

[0062] With the rapid development of cross-border e-commerce, the volume of international express delivery has continued to climb, and users' demands for the convenience of ordering parcels are constantly increasing. Ordering international express parcels requires filling in multiple details, including sender and recipient information, address, contact information, postal code, and item name, quantity, weight, and declared value, making the process complex. Due to factors such as varying address standards across different countries, users' habitual colloquial descriptions, and the provision of information in stages, manual entry is prone to errors, omissions, or non-standard formats. Furthermore, international express delivery involves complex rules regarding acceptance standards, shipping costs, delivery times, and customs supervision in multiple countries, resulting in high query and verification costs. Traditional interaction methods are no longer sufficient to meet users' needs for intelligent, one-stop ordering. Currently, international express ordering systems mostly employ question-and-answer intelligent customer service, fixed form filling, keyword and regular expression information extraction, and static rule validation, and generally only support text-based single-modal interaction. FAQ customer service replies to inquiries according to preset rules, form systems guide users to fill in fixed fields, keywords and regular expressions are used to extract address and item information, and rule bases are used to implement acceptance and delivery restriction judgments and cost calculations. However, the existing solutions still have obvious shortcomings: they are difficult to accurately extract structured information from users' colloquial and fragmented descriptions, and cannot handle multimodal input that mixes images and text; in the process of multi-round progressive information collection, they cannot effectively merge historical information with new information, which is prone to information overwriting or loss, resulting in cumbersome order placement steps, repeated verification, and a poor overall user experience.

[0063] Therefore, to address the technical challenges of existing technologies and reduce the complexity of order placement for users, the system supports user input via spoken language or images. It first performs multimodal intent recognition on user input, uniformly processing images and natural language text to accurately determine if the user intends to place an order, thus guiding the user through the appropriate process. To solve the problems of unstructured spoken and fragmented information and the easy overwriting and loss of information in multiple rounds, after recognizing an order intention, information is extracted according to standard field rules and centrally managed in the order status dataset, ensuring orderly storage of multi-round information without loss or overwriting. To further enhance the user experience and avoid repetitive input, the system dynamically generates prompts for fields to be filled in based on stored order status data, only reminding users of missing fields and guiding them to gradually complete the order, achieving progressive intelligent order placement. After the order information is complete, the shipping and receiving information is automatically stored in the user's historical information database for subsequent fuzzy matching and automatic filling, improving reuse efficiency and user experience.

[0064] Figure 1 This is a diagram illustrating an application scenario for implementing the conversational express delivery ordering method provided in this application, such as... Figure 1 As shown in the diagram, the scenario corresponding to the conversational express delivery ordering method provided in this application includes: a user terminal 101, an intelligent dialogue server 102, and a user historical information database 103. It is understood that the conversational express delivery ordering device is integrated into the intelligent dialogue server 102.

[0065] Specifically, the user inputs interactive information, including images and / or natural language text, into the intelligent dialogue server 102 via user terminal 101. The intelligent dialogue server 102 processes the user input in this round of dialogue to determine the user's target intent. If it is determined to be an intention to place an order, the intelligent dialogue server 102 extracts the corresponding field content from the user input in this round of dialogue according to preset field extraction rules and stores it in the maintained order status data. Then, the intelligent dialogue server 102 performs integrity verification based on the current order status data to generate a prompt message for fields to be filled, which is fed back to user terminal 101 to guide the user to supplement the information. When the order status data passes the integrity verification, the intelligent dialogue server 102 generates a complete order information prompt message to prompt the user to confirm the order information and complete the order, and synchronously stores the verified recipient and sender information in the user's historical information database 105. The relevant order status and prompt information are synchronized to user terminal 101 in real time for display to the user.

[0066] For example, the screen displayed on user terminal 101 could be as shown in screen 104, where the user inputs: "Send a package to Xiao Wang in Tokyo." The intelligent dialogue server 102, after extracting information from preset fields, sends the message to the user: "We have already filled in the recipient Xiao Wang's information for you. Please complete the item information." Upon seeing the reply, the user inputs: "A piece of clothing." This process continues until the order information is complete.

[0067] Optionally, the conversational express ordering method provided in this application can also be used for domestic express ordering scenarios or other express ordering scenarios. The international express ordering scenario provided in this embodiment is only an example and is not limited to such scenarios.

[0068] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0069] Figure 2 This is a flowchart illustrating a conversational express delivery ordering method provided in an embodiment of this application, as shown below. Figure 2 As shown, the execution subject of this embodiment is a conversational express delivery ordering device. This device can be implemented through a computer program, or through a medium storing the relevant computer program, such as a USB flash drive and / or optical disc, or through a physical device integrating or installing the relevant computer program, such as a chip, server, or server cluster. The conversational express delivery ordering method provided in this embodiment includes the following steps:

[0070] S201. Determine the user's target intent based on the user input information in this round of dialogue. The user input information includes image content and / or natural language text content.

[0071] User input information refers to all interactive content submitted by the user during the order placement process, including one or both of natural language text content and image content.

[0072] Natural language text content refers to spoken or written descriptions input by users in the form of text, speech-to-text, etc., such as "I want to send a mobile phone to Xiao Wang in Tokyo".

[0073] Among them, image content refers to images or screenshots uploaded by users that contain information such as addresses, items, waybills, and chat logs.

[0074] User target intent refers to the user's true purpose identified by analyzing user input, including the intent to place an order, the intent to query, the intent to modify, the intent to cancel, the intent to inquire, etc.

[0075] Specifically, the system first receives all content submitted by the user in the current dialogue round and determines whether the input type is plain text, plain image, or a combination of text and image. If image content is present, the system first extracts and cleanses the image, converting any recognizable text information into text format. Then, the extracted image text is concatenated with the natural language text directly input by the user to form a complete and consistent piece of content to be analyzed.

[0076] Furthermore, the assembled content is fed into the intent determination module, which analyzes the user's core needs by combining the contextual dialogue history. Key actions, objects, and scene information within the content are identified, and this information is then matched with preset intent types to determine the user's current desired action. Based on the analysis results, a clear target intent is provided, and the intent confidence level is marked.

[0077] Optionally, if the user's core purpose is determined to be to create a courier order, then it is determined to be an order placement intent; if it is to inquire about shipping costs, delivery time, and collection and delivery rules, then it is determined to be a query intent; if it is to modify the address or item information, then it is determined to be a modification intent.

[0078] S202. If the user's target intent is to place an order, extract the corresponding field content from the user's input information in this round of dialogue according to the preset field extraction rules and store it in the order status data in combination with the dialogue context.

[0079] Among them, the order placement intent refers to the user's current core purpose of initiating international / domestic express delivery and creating new orders. This is a target intent that requires collecting delivery and collection information, item information, and completing the order placement process.

[0080] Among them, the preset field extraction rules refer to the pre-defined specifications for identifying and extracting the information required for express delivery orders from users' natural language or image text, including the types of fields that can be extracted, extraction logic, and verification methods.

[0081] The field content refers to the specific information extracted from user input according to preset rules that can be directly used to generate an order, including sender and recipient information, address information, item information, etc.

[0082] Among them, order status data refers to the dataset used to uniformly store, manage, and maintain all information of the current order throughout the entire dialogue process, recording all collected fields, missing fields, information modification records, and order progress status.

[0083] The dialogue context refers to all relevant information that the user has previously expressed, that the system has identified and temporarily stored in the current dialogue history, including the address, time, and needs mentioned in previous rounds.

[0084] Specifically, the system reads pre-configured rules for extracting express delivery order information, identifying the standard fields to be extracted, including recipient information, sender information, country, province, city, address, telephone number, postal code, item name, quantity, weight, and value. Then, it performs semantic parsing on the user input that has undergone multimodal fusion, identifying descriptive statements related to the express delivery order, filtering out irrelevant content such as chatter and interjections, and retaining valid information usable for field extraction. Based on preset extraction rules, it identifies and separates the specific content corresponding to each standard field from the valid information. For user-speaking, vague, or fragmented descriptions, it transforms them into standardized, usable field content through semantic understanding and contextual association.

[0085] Furthermore, the extracted information is formatted, with standardized wording, corrected obvious errors, and missing elements added to ensure the field content conforms to the entry standards of the express delivery order system, guaranteeing information usability and standardization. Then, the field content extracted in this round is combined and supplemented with historical information already stored in the dialogue context, integrating, updating, and writing complete and valid information into the uniformly maintained order status data. If a field with the same name already exists in the order status data, it is updated using an incremental merging method. If it does not exist, the corresponding field content is directly added, ensuring that information collected from multiple rounds of dialogue is accumulated, not lost, and not overwritten.

[0086] S203. Perform integrity verification based on the current order status data to generate a message prompting users to fill in fields to complete their orders.

[0087] The current order status data refers to the set of all order-related information that has been extracted, merged, and uniformly maintained from user input during multiple rounds of dialogue.

[0088] Among them, the fields to be filled refer to standard fields that are necessary to complete the express delivery order, but have not yet been collected or are incomplete in the current order status data.

[0089] Among them, the "fields to be filled" prompt message refers to the natural language prompt content generated by the system for the required fields and already filled information in the order, which is used to guide users to fill in the missing content and improve the order information.

[0090] Integrity verification refers to the process of checking each item of order status data according to the preset mandatory order field specifications to determine whether there are missing, empty, or non-compliant data items.

[0091] Specifically, based on all the standard required fields for placing an express delivery order, the stored field content in the current order status data is compared one by one with the complete list of required fields. Fields that have not been collected, are empty, or do not meet the specifications are filtered out, forming a set of missing fields. According to the information logic of international express delivery orders, the missing fields are prioritized, with core required fields such as address, phone number, and item name being prompted first, followed by supplementary fields such as postal code and weight.

[0092] Furthermore, based on the sorted missing fields, the information is organized into user-friendly natural language statements, clearly informing users of the required supplementary information, avoiding the use of technical jargon, and keeping the prompts clear and concise. The generated prompts for fields to be filled are used as the core response content of this round of dialogue, integrated with other business information, and sent to the user's end to guide them to complete the information in the next round of input until the order information is complete.

[0093] S204. In response to the order status data passing the integrity check, generate a complete order information prompt message. The complete order information prompt message is used to prompt the user to confirm the order information and complete the order, and stores the recipient information and sender information in the user's historical information database.

[0094] The user history information database refers to a collection of data used to store a user's frequently used shipping information, historical receiving information, and frequently used addresses for long-term storage, which can be automatically reused and intelligently filled when placing subsequent orders.

[0095] Specifically, all required fields in the current order status data are checked item by item to confirm that all key fields have been collected, the content is legal and valid, and the format conforms to international express delivery standards, thus determining that the order information meets the completeness requirements. The system generates a message indicating that the order information is complete and pushes this message to the user's interface to clearly inform the user that the order information has been filled in completely and that they can confirm the information and execute the final order placement operation. Simultaneously with pushing the message, complete and standardized recipient and sender information is separated and extracted from the validated order status data, removing irrelevant content such as temporary process data and interaction records. The extracted recipient and sender information undergoes format standardization, field regularization, and legality verification to ensure that information such as address, telephone number, and postal code meets the requirements for long-term storage and reuse.

[0096] Furthermore, the standardized sender and recipient information is persistently saved in the corresponding user's historical information database as archives of the user's frequently used sending information and historical receiving information.

[0097] Optionally, the stored shipping information can be marked as valid and the latest update time can be recorded to facilitate priority matching, automatic identification and quick filling when placing subsequent orders, thus avoiding repeated input by the user.

[0098] The conversational express ordering method provided in this application supports mixed text and image input, and is compatible with real-world usage scenarios such as user screenshots, chat logs, and photo addresses, thus broadening the interaction methods. It proactively identifies the user's true intent, distinguishing between different needs such as ordering, querying, modifying, and canceling, avoiding invalid processes and improving system response efficiency. If the intent is to place an order, it automatically extracts receiving and shipping information and item information from conversational and fragmented information in a structured manner, eliminating the need for manual form filling. Order status data is uniformly stored, achieving centralized management of information throughout the process, ensuring no loss or confusion. It supports multi-round progressive input, allowing users to provide information in stages while storing it segment by segment, avoiding information overwriting. Based on the current order status data, i.e., the already filled information, missing fields are dynamically calculated, only prompting the user for unfilled content, eliminating the need for repeated confirmation. It achieves intelligent guided ordering, reducing user understanding costs and operation steps. It improves information completeness, avoiding order failures due to missing fields. After the order data passes integrity verification, the receiving and shipping information is stored in the user's historical information database. This completes the automatic persistent storage of information, forming a user-specific address memory database. Subsequent use allows for fuzzy matching and autofill, significantly reducing repetitive input. This improves order efficiency for existing users, providing a smarter experience the more you use it.

[0099] As an optional implementation, based on the above embodiments, the user input information includes image content and natural language text content. The step of determining the user's target intent based on the user input information in this round of dialogue further includes:

[0100] The text content is extracted from the image using a preset extraction algorithm and then concatenated with natural language text content to generate the target query content.

[0101] The intent classification results of the target query content are determined based on the preset prompt word template. The intent classification results include the target intent identifier and the corresponding confidence score.

[0102] The target query content is identified as a pending node based on the target intent identifier and the corresponding confidence score. The pending node is used to execute the processing flow for different intents.

[0103] Among them, the preset extraction algorithm refers to the pre-configured algorithm used to identify and extract text information from the image, such as optical character parsing rules.

[0104] The target query content refers to the complete user input information used for intent recognition after multimodal fusion.

[0105] The preset prompt word template refers to a pre-built set of prompt words used to guide the large model in determining intent. This includes multiple intent categories and corresponding descriptive information for each intent. Intents include placing an order, querying shipping standards, querying shipping costs and delivery time, modifying an order, modifying pick-up time, querying self-delivery points, querying routes, historical orders, casual conversation, confirming / rejecting, and transferring to human assistance.

[0106] Among them, the target intent identifier refers to the classification mark used to distinguish different intents such as placing an order, querying, modifying, and canceling.

[0107] Among them, the pending node refers to the subsequent business processing unit allocated according to the intent type, which is used to perform operations such as information extraction, rule query, and order modification.

[0108] Specifically, the system first checks if the user input contains image content. If so, a preset extraction algorithm is used to perform optical character recognition (OCR) on the image, extracting valid text information and filtering out irrelevant content. The extracted image text is then semantically fused with the user's directly input natural language text to form a complete and continuous target query. Based on a preset prompt template, the content is semantically parsed, and the user's core needs are determined by considering the historical dialogue context. The corresponding intent classification result is output, including a target intent identifier and a confidence score indicating the reliability of the recognition. The business processing direction is determined based on the target intent identifier, and the reliability of the recognition result is assessed using the confidence score.

[0109] Optionally, when the confidence level meets the preset threshold, the user directly enters the pending node corresponding to the intent; when the confidence level is insufficient, the user enters the clarification or guessing node to guide the user to further clarify their needs and finally determine the pending node of the target query content.

[0110] The conversational express delivery ordering method provided in this application supports mixed text and image input, and is compatible with real-world usage scenarios such as screenshots, photos, and chat logs. It converts image information into parsable text, expanding the system's input types and improving user interaction convenience. Standardized templates enhance the stability and accuracy of intent recognition, enabling precise judgment of users' true needs from colloquial and fragmented descriptions; outputting confidence scores provides a basis for subsequent decisions, reducing the probability of misjudgment. Based on intent and confidence, it determines the nodes to be processed, achieving intelligent routing of the process, allowing the system to automatically enter the corresponding processing stage according to user needs, avoiding invalid execution; combining confidence scores for branch judgment, it proactively clarifies when there is uncertainty in recognition, improving the overall robustness of the process and user experience.

[0111] As an optional implementation, based on the above embodiments, the process node for determining the target query content is determined according to the target intent identifier and the corresponding confidence score, including:

[0112] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the first preset identifier, then the target query content is rewritten based on the historical dialogue.

[0113] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the second preset identifier, then the corresponding query interface or preset knowledge base will be called to perform a query based on the target query content, and a response will be given based on the query results.

[0114] If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the third preset identifier, then modify the field content in the order status data based on the historical dialogue.

[0115] If the confidence score is less than the preset confidence threshold, a list of query questions is generated based on the target query content. The list of query questions is used to determine the target intent.

[0116] Among them, the pre-set reliability threshold refers to the pre-set score threshold used to judge whether the intent recognition result is credible and whether it can directly enter the corresponding business process.

[0117] The first preset identifier corresponds to an intent identifier that indicates the user's intent to confirm, refuse, or make a referential statement, and the complete information needs to be supplemented in combination with the context.

[0118] The second preset identifier corresponds to the user's intent to inquire about services such as freight rates, delivery times, collection and delivery standards, and service outlets.

[0119] Among them, the third preset identifier corresponds to the user's intent to modify order editing information such as receiving information, sending information, or item information.

[0120] Among them, historical dialogues refer to all interaction records before the current round of input in the current session, including extracted information, order status, and user statements.

[0121] Rewriting the target query content refers to supplementing vague, abbreviated, or referential user input into complete, explicit, and directly processable query content based on the context of the dialogue. For example, resolving the previous address to a specific address.

[0122] The pre-built knowledge base refers to a collection of questions and answers about common issues, rules, and operational guidelines in the express delivery business that are built and stored in advance.

[0123] The query question list refers to the optional questions generated by the system to confirm and clarify the user's true needs when the confidence level of intent recognition is insufficient.

[0124] Specifically, when the confidence score is greater than or equal to the preset confidence threshold and the target intent is the first preset identifier, the historical dialogue information is read, and the user's input in this round is resolved and the content is completed in combination with the context. The omitted, vague or referential expressions are rewritten into complete and clear target query content so as to facilitate accurate processing in the future.

[0125] Optionally, during the rewriting process, it is determined whether to continue rewriting based on a preset upper limit for the number of rewrites. If the current number of rewrites is greater than or equal to the upper limit for the number of rewrites, it is determined to be a chatty intention and is processed according to the processing method for chatty intentions.

[0126] Optionally, when the confidence score is greater than or equal to a preset confidence threshold, and the target intent is a second preset identifier, the corresponding business query interface is called according to the intent type, or a matching search is performed in a preset knowledge base to obtain relevant results such as shipping costs, delivery time, and acceptance / delivery rules, and a response is generated based on the query results. This can include integrating multi-source information, reminder notes, QA knowledge, API results, etc., to generate a natural and accurate response through a prompt project. Different prompt templates can be selected based on the information source: pure reminder, pure knowledge, or a combination of reminder and knowledge.

[0127] For example, the acceptance and delivery standard query extracts the item name and country of origin, calls the ISOP API to query acceptance and delivery rules, such as whether delivery is possible and any restrictions, and caches the results in isop_records to avoid duplicate calls. The shipping cost and delivery time query uses the sender's and recipient's city codes to call the recommendation system API to query shipping costs and delivery times, parses the returned results to generate a natural language response, such as an estimated delivery time of 3 days and a shipping cost starting from 128 yuan. The QA knowledge base retrieval uses vector similarity to search the SF Express International Business FAQ knowledge base, providing a reference for response generation.

[0128] Optionally, when the confidence score is greater than or equal to the preset confidence threshold and the target intent is the third preset identifier, the fields and target content that the user needs to modify are determined based on the historical dialogue, the corresponding field content in the order status data is updated, and other filled information remains unchanged.

[0129] Optionally, when the confidence score is less than the preset confidence threshold, it is impossible to accurately determine the user's true intention. Based on the user's input, a list of multiple-choice query questions is generated, and the user is asked to clarify and confirm. After clarifying the target intention through user feedback, the user can proceed to the corresponding pending node.

[0130] The conversational express ordering method provided in this application effectively handles user omissions, references, and confirmations by rewriting query content, and completes ambiguous information based on context, improving dialogue coherence and comprehension accuracy, and avoiding process interruptions due to incomplete information. The intent of the second preset identifier is directly routed to the corresponding business capability, automatically completing real-time data queries and knowledge Q&A without manual intervention, improving consultation response speed and reply accuracy. Users can flexibly modify order information in multi-turn dialogues, updating only the target field without affecting other content, achieving progressive and editable intelligent ordering, and improving usage flexibility. When intent is uncertain, the method proactively clarifies to the user, avoiding erroneous operations due to misjudgment, improving system robustness, reducing error rate, and enhancing overall interaction reliability.

[0131] As an optional implementation, based on the above embodiments, the corresponding field content is extracted from the user input information of the current round of dialogue according to preset field extraction rules and stored in the order status data in combination with the dialogue context, including:

[0132] Determine the type of information contained in the user input;

[0133] If the information type is recipient information or sender information, the corresponding field content is extracted from the user input information according to the preset fields. The field content includes the recipient's or sender's personal information and address information, and the personal information and address information are mapped to standard fields.

[0134] If the information type is item information, the corresponding field content is extracted from the user input information. The field content includes the item name and item attributes, and the acceptance and delivery standards are queried based on the item name.

[0135] The extracted field content is compared with the corresponding field content already stored in the current order status data;

[0136] If the extracted field content is inconsistent with the stored content or the corresponding field is empty, the extracted field content will be stored in the order status data.

[0137] Among them, information type refers to the order information category corresponding to the user input content, which mainly includes three categories: recipient information, sender information, and item information.

[0138] The preset fields are standard information items required to complete a courier order, including personal and address information of the sender and recipient, and item-related information, such as recipient's name, country, province, city, street, postal code, and telephone number.

[0139] Standard fields refer to fixed field names and data formats that are used internally in accordance with business specifications, ensuring that information can be stored and processed uniformly.

[0140] Among them, item attributes refer to relevant parameters used to describe the characteristics of an item, including quantity, weight, size, value, currency, brand, etc.

[0141] Among them, the standards for accepting and sending items refer to the rules that determine whether items can be sent, the restrictions, and the required qualifications, based on national policies, transportation restrictions, aviation safety, and other requirements.

[0142] Specifically, semantic parsing is performed on user input to determine whether it primarily involves recipient, sender, or item information, thus classifying the information type. If the information type is recipient or sender information, the corresponding personal and address information is extracted from the user input according to preset fields. The extracted results are mapped to standard fields using `process_map_fields`, and the address is standardized by calling the ICMS API to query city codes. If the information type is item information, the item name and attributes are extracted from the user input. Based on the item name and the sending / receiving country information, the corresponding acceptance and delivery standards are queried to determine whether delivery is permitted and any related restrictions. For example, the ISOP API is called to query acceptance and delivery standards, and the results are cached to avoid duplicate queries.

[0143] Furthermore, after information extraction is complete, the newly extracted field content is compared item by item with the corresponding field content already stored in the order status data to determine if there are any discrepancies or if the original field is empty. If the newly extracted content is inconsistent with the stored content, or if the corresponding field was originally empty, the latest content extracted in this round is used to update and overwrite the corresponding field in the order status data, completing the dynamic update and supplementation of order information.

[0144] The conversational express ordering method provided in this application classifies user input, enabling the system to execute different extraction logics based on information categories, improving parsing accuracy and avoiding interference between different types of information. It automatically extracts structured information from colloquial descriptions, unifies field formats, and ensures that addresses and personal information are standardized and usable. It stores unified order status data, achieving multi-round information accumulation and saving without loss or overwriting. While extracting item information, it automatically performs compliance checks and identifies prohibited and restricted items in advance. It stores acceptance standards and item information together, providing a basis for subsequent order verification, prompts, and responses, improving order security and process smoothness.

[0145] As an optional implementation, based on the above embodiments, the information type is recipient information, and the method further includes:

[0146] Retrieve default sender information from the user's historical information database and store it in the order status data;

[0147] The system matches the recipient information included in the user's input with the information from the user's historical information database.

[0148] If the matching similarity is greater than the first similarity threshold, the information content included in the matching result will be stored in the order status data;

[0149] If the matching similarity is greater than or equal to the second similarity threshold and less than or equal to the first similarity threshold, then a candidate recipient list is output. The candidate recipient list is used to confirm the target recipient information.

[0150] If the matching similarity is less than the second similarity threshold, a prompt message for the fields to be filled in will be output.

[0151] The default sender information refers to the frequently used sender information that the user has saved in advance or prioritizes, including complete information such as name, phone number, address, and postal code.

[0152] The first similarity threshold refers to the preset similarity threshold used to determine a complete match that can be automatically filled in.

[0153] The second similarity threshold refers to a preset similarity threshold used to determine partial matching and to display a candidate list for user selection.

[0154] The recipient candidate list refers to a list of multiple similar historical recipient information displayed when the matching results are not unique or completely certain.

[0155] Among them, the target recipient information refers to the correct recipient information that the user ultimately confirms is used to generate the order.

[0156] Specifically, the system reads pre-saved default sender information from the user's historical information database and directly stores it into the order status data, eliminating the need for the user to fill it in repeatedly. It then extracts recipient-related content from the user's current input, searches and compares it in the user's historical information database, and calculates the similarity between each historical record and the current input.

[0157] Optionally, if the calculated matching similarity is greater than the first similarity threshold, it is determined to be an exact match, and the complete information in the historical record is automatically extracted and stored in the order status data.

[0158] Optionally, if the matching similarity is between the second similarity threshold and the first similarity threshold, it is determined to be a fuzzy match. Multiple similar records are selected from historical information, and a candidate list of recipients is generated and displayed to the user, who then confirms the target recipient information.

[0159] Optionally, if the matching similarity is less than the second similarity threshold, it is determined that there is no valid matching record, and a prompt message for fields to be filled in is generated to guide the user to enter the recipient information completely.

[0160] The conversational express ordering method provided in this application automatically loads default sender information, eliminating the need for users to repeatedly enter sender information, simplifying the operation process, and significantly improving ordering efficiency for existing users. Historical matching of recipient information enables intelligent reuse of users' historical addresses, reducing manual input and errors, and lowering the cost of filling out forms. When matching is uncertain, an optional list is provided, balancing intelligence and accuracy, and avoiding incorrect entry. When there is no historical record, a smooth switch to guided mode ensures the process is complete and usable, improving system robustness.

[0161] As an optional implementation, based on the above embodiments, it further includes:

[0162] A document-based database is used to store checkpoints for order status data.

[0163] In response to a user reconnecting after interrupting the conversation, order status data is restored based on the session identifier.

[0164] Checkpoint storage refers to saving the complete order status data at a key node in the dialogue process, forming a recoverable data snapshot.

[0165] Interrupted conversation refers to the state in which the current order session is paused or terminated due to reasons such as the user exiting the page, closing the program, or disconnecting from the network.

[0166] Reconnection refers to the process where, after an interruption, a user re-enters the order interface, reconnects, and continues to interact with the system.

[0167] The session identifier is a unique number used to identify a complete order placement conversation, serving as a basis for distinguishing different users and different sessions.

[0168] Among them, restoring order status data refers to restoring order information, dialogue progress, and filled content to the state before the interruption based on the saved checkpoint data.

[0169] Specifically, at key nodes in the dialogue process, the current complete order status data is structured and written to a document-based database in document format, thus saving the checkpoints. Each time information is updated or the process jumps, the system persistently stores the latest status to ensure no data loss. In response to a user re-entering the system after an interruption, order status data is restored based on the session identifier. When a user re-enters the system after an interruption, the system obtains the session identifier corresponding to the current dialogue and retrieves matching checkpoint data from the document-based database based on this identifier. Once valid data is retrieved, the order information, filled fields, dialogue progress, and other content saved in the checkpoint are completely restored, allowing the user to directly return to the order placement state before the interruption and continue subsequent operations.

[0170] The conversational express ordering method provided in this application employs checkpoint storage to persistently save order status data, unaffected by page refreshes, program exits, or service restarts, ensuring the stability and reliability of multi-turn conversation information. Based on session identifiers, it supports breakpoint-resume ordering, allowing users to exit and return at any time to continue without re-entering information, significantly improving ease of use and user experience. This makes multi-turn conversational ordering highly stable and fault-tolerant, solving the problems of data loss and inability to recover from interruptions in traditional conversational ordering, thus improving system reliability and usability.

[0171] As an optional implementation, based on the above embodiments, it further includes:

[0172] Perform error detection on order status data;

[0173] If an error in the field format or a logical mismatch is detected in the order status data, the content of the erroneous field should be cleared.

[0174] Generate a message indicating the fields to be filled in, including error messages.

[0175] Error detection refers to the process of checking the legality of various items in the order status data in accordance with the format specifications and business logic of international express delivery information.

[0176] Among them, field format errors refer to field content that does not meet the system's format requirements, such as incorrect postal code digits, telephone numbers containing illegal characters, or empty addresses.

[0177] Among them, logical mismatch refers to business logic conflicts between fields, such as inconsistencies between country and city, mismatches between province and city, and violations of restriction rules by the sending and receiving areas.

[0178] Error messages are explanatory texts used to inform users of specific errors and their causes, guiding them to re-enter the correct information.

[0179] Specifically, according to preset validation rules, all fields in the order status data are checked one by one to determine if there are any invalid formats or content conflicts. When a field with a format error or logical mismatch is detected, the original content of the field is automatically cleared, retaining the field position but not the erroneous data, to prevent abnormal data from affecting order generation. Based on the specific error detected, a corresponding natural language error message is generated and integrated into the prompt message for the field to be filled in, and sent to the user, clearly informing them of the content and requirements that need to be corrected, guiding the user to provide the correct information again.

[0180] The conversational express ordering method provided in this application improves order success rate by detecting data anomalies before order submission through error detection, thus avoiding problems such as order failure, customs clearance failure, and transit returns due to incorrect information. It proactively cleans up invalid or erroneous data to prevent erroneous information from being carried over into subsequent business processes, ensuring the accuracy and standardization of order data. It generates error messages that clearly inform users of the reasons for errors, eliminating the need for guesswork or repeated attempts, reducing comprehension costs, improving information correction efficiency, and enhancing the interactive experience.

[0181] Figure 3 This is a signaling interaction flowchart of a conversational express delivery ordering method provided in another embodiment of this application, taking a user wanting to send an express delivery to someone else as an example. Figure 3 As shown, the conversational express ordering method provided in this embodiment includes the following steps:

[0182] S301. The user enters the information "sent to Xiao Wang in Tokyo" on the visual interface of the terminal device.

[0183] S302, The conversational express ordering device determines the user's target intent as the intention to place an order based on the user's input information.

[0184] S303, The conversational express ordering device determines the type of information contained in the user input information.

[0185] S304. The conversational express order placement device determines that the information type is recipient information, retrieves the default sender information from the user's historical information database, and stores it in the order status data.

[0186] S305, the conversational express ordering device matches the recipient information included in the user's input information with the user's historical information database.

[0187] S306. The conversational express ordering device stores the information content of the matching results with a similarity greater than the first similarity threshold into the order status data.

[0188] S307, The conversational express ordering device generates a prompt message for fields to be filled in based on the current order status data.

[0189] S308. Output a message to the terminal device's visual interface that says "The recipient Xiao Wang's information has been automatically filled in. Please fill in the item information."

[0190] S309. The user enters item information "a mobile phone" on the visual interface of the terminal device.

[0191] The S310 conversational express ordering device extracts the item name and item attributes from the user's input information and stores them in the order status data.

[0192] S311, The conversational express ordering device queries the acceptance and delivery standards based on the item name through a preset query interface.

[0193] S312. Based on the query results and current order status data, the conversational express order placement device generates a prompt message for fields to be filled in.

[0194] S313. Output a prompt message to the terminal device's visual interface: "Mobile phone is allowed to be mailed. Please fill in the weight and value."

[0195] S314. The user enters the remaining information "200 grams, 800 yuan" on the visual interface of the terminal device.

[0196] S315, The conversational express order placement device performs integrity verification on order status data.

[0197] S316. In response to the order status data passing the integrity verification, the conversational express ordering device stores the recipient information and sender information in the user's historical information database.

[0198] S317. The conversational express ordering device allows users to check shipping costs and delivery times through a preset query interface.

[0199] S318, The conversational express ordering device generates a response message based on the query results.

[0200] S319. Output the message "Order information is complete, shipping fee is 128, estimated arrival time is 3 days, please reply yes to confirm the order" on the visual interface of the terminal device.

[0201] In this embodiment, the implementation method and technical effect of S301-S319 are similar to those of the corresponding solutions in the above embodiments, and will not be repeated here.

[0202] Figure 4 A schematic diagram of the conversational express ordering device provided in this application is shown below. Figure 4 As shown, the conversational express ordering device 40 provided in this embodiment includes: a determination module 41, an extraction module 42, a generation module 43, and a storage module 44.

[0203] The system comprises the following modules: Determination module 41, which determines the user's target intent based on the user input information in the current dialogue, including image content and / or natural language text content; Extraction module 42, which, if the user's target intent is to place an order, extracts the corresponding field content from the user input information in the current dialogue according to preset field extraction rules; Storage module 44, which stores the extracted field content in the order status data in conjunction with the dialogue context; Generation module 43, which performs integrity verification based on the current order status data to generate a field to be filled prompt message, which prompts the user to supplement order information; Generation module 43, which also generates a complete order information prompt message in response to the order status data passing integrity verification, which prompts the user to confirm the order information and complete the order; and Storage module 44, which stores the recipient and sender information in the user's historical information database.

[0204] The conversational express delivery ordering device provided in this embodiment can execute... Figure 2 The implementation principles and technical effects of the methods shown are similar, and will not be repeated here.

[0205] Optionally, the user input information includes image content and natural language text content. The determining module 41, when determining the user's target intent based on the user input information in this round of dialogue, specifically performs the following: extracting text content from the image using a preset extraction algorithm and concatenating it with the natural language text content to generate the target query content; determining the intent classification result of the target query content based on a preset prompt word template, the intent classification result including the target intent identifier and the corresponding confidence score; and determining the pending processing node of the target query content based on the target intent identifier and the corresponding confidence score, the pending processing node being used to execute the processing flow for different intents.

[0206] Optionally, the determining module 41, when determining the pending node of the target query content based on the target intent identifier and the corresponding confidence score, specifically performs the following: if the confidence score is greater than or equal to a preset confidence threshold and the target intent identifier is a first preset identifier, then rewrite the target query content based on historical dialogue; if the confidence score is greater than or equal to a preset confidence threshold and the target intent identifier is a second preset identifier, then call the corresponding query interface or preset knowledge base to perform a query based on the target query content, and reply based on the query results; if the confidence score is greater than or equal to a preset confidence threshold and the target intent identifier is a third preset identifier, then modify the field content in the order status data based on historical dialogue; if the confidence score is less than a preset confidence threshold, then generate a query question list based on the target query content, and the query question list is used to determine the target intent.

[0207] Optionally, the extraction module 42, when extracting corresponding field content from the user input information in the current round of dialogue according to preset field extraction rules and storing it in the order status data in conjunction with the dialogue context, specifically performs the following: determining the information type contained in the user input information; if the information type is recipient information or sender information, then extracting the corresponding field content from the user input information according to preset fields, the field content including the recipient's or sender's personal information and address information, and mapping the personal information and address information to standard fields; if the information type is item information, then extracting the corresponding field content from the user input information, the field content including the item name and item attributes, and querying the acceptance and delivery standards based on the item name; comparing the extracted field content with the corresponding field content already stored in the current order status data; if the extracted field content is inconsistent with the stored content or the corresponding field is empty, then storing the extracted field content in the order status data.

[0208] Optionally, the conversational express ordering device provided in this embodiment further includes an acquisition module, a matching module, and an output module.

[0209] Correspondingly, the acquisition module is used to retrieve default sender information from the user's historical information database, and the storage module 44 is also used to store the default sender information in the order status data; the matching module is used to match the recipient information included in the user's input information in the user's historical information database; the storage module 44 is also used to store the information content included in the matching result in the order status data if the matching similarity is greater than the first similarity threshold; the output module is used to output a candidate recipient list if the matching similarity is greater than or equal to the second similarity threshold and less than or equal to the first similarity threshold, and the candidate recipient list is used to confirm the target recipient information; if the matching similarity is less than the second similarity threshold, a prompt message for fields to be filled is output.

[0210] Optionally, the conversational express ordering device provided in this embodiment also includes a recovery module.

[0211] Correspondingly, the storage module 44 is also used to perform checkpoint storage of order status data using a document-oriented database; the recovery module is used to recover order status data based on the session identifier in response to a user reconnecting after interrupting the conversation.

[0212] Optionally, the conversational express ordering device provided in this embodiment also includes a detection module and a clearing module.

[0213] Correspondingly, the detection module is used to detect errors in the order status data; the clearing module is used to clear the content of the erroneous fields if an error in the field format or a logical mismatch is detected in the order status data; and the generation module 43 is also used to generate a prompt message for the fields to be filled in, which includes error messages.

[0214] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Figure 5 As shown, the electronic device 50 provided in this embodiment includes a processor 51 and a memory 52. ​​The processor 51 and the memory 52 are connected via a bus and communicate with each other.

[0215] In the specific implementation process, the processor 51 executes the computer execution instructions stored in the memory 52, causing the processor 51 to perform the above-described method.

[0216] The specific implementation process of processor 51 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0217] In the above embodiments, it should be understood that the processor 51 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0218] The memory 52 may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0219] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0220] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0221] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0222] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, 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), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0223] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0224] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0225] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0226] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0227] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0228] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0229] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A conversational express delivery ordering method, characterized in that, The method includes: The user's target intent is determined based on the user input information in this round of dialogue, including image content and / or natural language text content; If the user's target intent is to place an order, the corresponding field content is extracted from the user's input information in this round of dialogue according to the preset field extraction rules and stored in the order status data in combination with the dialogue context; Completeness verification is performed based on the current order status data to generate a message prompting users to fill in fields that need to be filled in. This message prompts users to supplement their order information. In response to the order status data passing the integrity verification, a complete order information prompt message is generated. The complete order information prompt message is used to prompt the user to confirm the order information and complete the order, and the recipient information and sender information are stored in the user's historical information database.

2. The method according to claim 1, characterized in that, The user input information includes image content and natural language text content. The step of determining the user's target intent based on the user input information in this round of dialogue further includes: A preset extraction algorithm is used to extract text content from the image, and then concatenated with the natural language text content to generate the target query content; The intent classification result of the target query content is determined based on a preset prompt word template. The intent classification result includes the target intent identifier and the corresponding confidence score. The target query content is identified as a pending node based on the target intent identifier and the corresponding confidence score. The pending node is used to execute the processing flow for different intents.

3. The method according to claim 2, characterized in that, The step of determining the pending nodes of the target query content based on the target intent identifier and the corresponding confidence score includes: If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the first preset identifier, then the target query content is rewritten based on the historical dialogue. If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the second preset identifier, then the corresponding query interface or preset knowledge base will be called to perform a query based on the target query content, and a response will be given based on the query results. If the confidence score is greater than or equal to the preset confidence threshold, and the target intent identifier is the third preset identifier, then modify the field content in the order status data based on the historical dialogue. If the confidence score is less than the preset confidence threshold, a list of query questions is generated based on the target query content. The list of query questions is used to determine the target intent.

4. The method according to claim 1, characterized in that, The step of extracting corresponding field content from the user input information in the current round of dialogue according to preset field extraction rules and storing it in the order status data in combination with the dialogue context includes: Determine the type of information contained in the user input; If the information type is recipient information or sender information, the corresponding field content is extracted from the user input information according to the preset fill fields. The field content includes the recipient's or sender's personal information and address information, and the personal information and address information are mapped to standard fields. If the information type is item information, the corresponding field content is extracted from the user input information. The field content includes the item name and item attributes, and the acceptance and delivery standards are queried based on the item name. The extracted field content is compared with the corresponding field content already stored in the current order status data; If the extracted field content is inconsistent with the stored content or the corresponding field is empty, the extracted field content will be stored in the order status data.

5. The method according to claim 4, characterized in that, The information type is recipient information, and the method further includes: Retrieve default sender information from the user's historical information database and store it in the order status data; The system matches the recipient information included in the user's input with the information from the user's historical information database. If the matching similarity is greater than the first similarity threshold, the information content included in the matching result will be stored in the order status data; If the matching similarity is greater than or equal to the second similarity threshold and less than or equal to the first similarity threshold, then a candidate recipient list is output, which is used to confirm the target recipient information. If the matching similarity is less than the second similarity threshold, a prompt message for the fields to be filled in will be output.

6. The method according to claim 1, characterized in that, Also includes: The order status data is stored using a document-based database with checkpoints. In response to a user reconnecting after interrupting the conversation, the order status data is restored based on the session identifier.

7. The method according to claim 1, characterized in that, Also includes: Perform error detection on order status data; If an error in the field format or a logical mismatch is detected in the order status data, the content of the erroneous field should be cleared. Generate a message indicating the fields to be filled in, including error messages.

8. A conversational express delivery ordering device, characterized in that, include: The determination module is used to determine the user's target intent based on the user input information in the current round of dialogue, wherein the user input information includes image content and / or natural language text content; The extraction module is used to extract the corresponding field content from the user input information in the current round of dialogue according to the preset field extraction rules if the user's target intent is to place an order. The storage module is used to store the extracted field content into the order status data in conjunction with the dialogue context; The generation module is used to perform integrity verification based on the current order status data to generate a prompt message for fields to be filled in. The prompt message for fields to be filled in is used to prompt the user to supplement the order information. The generation module is further configured to generate a complete order information prompt message in response to the order status data passing the integrity check. This complete order information prompt message is used to prompt the user to confirm the order information and complete the order. The storage module is also used to store the recipient information and sender information into the user's historical information database.

9. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

11. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method according to any one of claims 1-7.

Images