Method and device for realizing list operation based on heterogeneous cyclic dialogue and medium

Abstract

The invention provides a method and a device for realizing list operation based on heterogeneous cyclic dialogue and a medium, and relates to the technical field of computers. The method comprises the following steps: acquiring an original list, a plurality of predefined operation dimensions and a plurality of predefined operations; the method for executing multiple rounds of dialogues based on heterogeneous circulation with a user comprises the steps that an Nth round of user dialogue instruction is received, a first operation dimension, a first operation and a pointed object list in the Nth round of user dialogue instruction are recognized, the first operation is executed on the first operation dimension in the object list to obtain an Nth round of execution list, the object list is an original list or an execution list of a certain round in previous N-1 rounds. The heterogeneous cyclic dialogue with the user is realized based on list interaction, some lists obtained in the multiple rounds are taken as the basis of a new round of dialogue according to the user dialogue instruction, and the method can adapt to a complex dialogue scene of thinking jump of the user.

Description

Technical Field

[0001] The present application relates at least to the field of computer technology, and in particular to a method, device, and medium for implementing list operations based on heterogeneous loop dialogues. Background Art

[0002] Existing voice-interaction interfaces on smartphones have very limited support for retrieving list information through voice. Users can only browse lists linearly or switch back to the graphical user interface (GUI) interface, making the entire process inefficient and unnatural. Lists are a common and important component of smartphone GUIs, containing a large number of items with rich semantic information that must be retrieved during GUI usage. The inability of voice interfaces to provide this information significantly limits their interactive capabilities, making it impossible to efficiently complete user interaction tasks that require GUI list information, thus affecting the user experience. Summary of the Invention

[0003] In response to the above-mentioned shortcomings, the present application provides a method, device and medium for implementing list operations based on heterogeneous loop dialogues to solve the following technical problem: how to improve the efficiency of interacting with list content during dialogues with users.

[0004] In a first aspect, the present application provides a method for implementing list operations based on heterogeneous loop dialogues, the method comprising:

[0005] Get the original list and predefined multiple operation dimensions and multiple operations;

[0006] Conduct heterogeneous, loop-based multi-turn conversations with users, including:

[0007] Receive the Nth round of user dialogue instructions,

[0008] Identify the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions.

[0009] Execute the first operation on the first operation dimension in the object list to obtain the Nth round execution list.

[0010] The object list is the original list or a round execution list in the first N-1 rounds.

[0011] Furthermore, the original list and multiple predefined operation dimensions and multiple operations are obtained, including:

[0012] Get the list of original tariff packages preset by the operator;

[0013] Obtain the consideration dimensions of the user's package subscription, including package identification, package price, several package service contents, and package attributes. Define an operation dimension based on each consideration dimension.

[0014] Get the user's habitual operations on the package list when selecting a package. Habitual operations include selecting list items, adding list items to favorites, and sorting list items. Define an operation based on each habitual operation.

[0015] Furthermore, identifying the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions specifically includes:

[0016] Identifying, based on keywords set for each operation dimension and each operation, a plurality of first operation dimensions and corresponding first operations included in the Nth round of user dialogue instructions;

[0017] Obtain the previous N-1 rounds of user dialogue instructions, and identify the contextual relationship between the Nth round of user dialogue instructions and the previous N-1 rounds of user dialogue instructions;

[0018] According to the context and several first operations, the original list or a list of execution rounds in the previous N-1 rounds is obtained as the object list pointed to by the current round.

[0019] Further, wherein:

[0020] The keywords set for each operation dimension correspond to several list tags in the original tariff package list, including:

[0021] The package identifier corresponds to the rate code, rate name, and rate description.

[0022] The package price corresponds to the monthly rental fee and the price of traffic, voice, SMS and MMS outside the package.

[0023] Several package service contents correspond to the amount of data traffic, voice duration, number of SMS messages, and number of MMS messages within the package.

[0024] Package attributes include the network type, location, and whether the package is shareable, transferable, contracted, broadband, and TV.

[0025] Each list tag has several pre-set operations that it supports.

[0026] Furthermore, a first operation is performed on the first operation dimension in the object list to obtain an N-th round execution list, specifically including:

[0027] Sequentially selecting a first operation dimension and a corresponding first operation from the multiple first operation dimensions and the corresponding multiple first operations of this round;

[0028] Execute a corresponding first operation on a first operation dimension in the object list of this round to obtain a first execution list of this round of the first operation dimension and the first operation;

[0029] Identify the intersection / union relationship between the multiple first operation dimensions corresponding to a first operation in this round, and calculate the intersection / union of the first execution lists of the corresponding multiple first operation dimensions in this round according to the intersection / union relationship to obtain the second execution list of the current round for the first operation;

[0030] Identify the replacement relationship between the first operations of the current round, and replace the second execution list of the current round of the first operations according to the replacement relationship to obtain a third execution list of the current round.

[0031] Furthermore, based on the context relationship and the plurality of first operations, obtaining the object list pointed to by the original list or a certain round execution list in the previous N-1 rounds, and performing a corresponding first operation on a certain first operation dimension in the object list of the current round to obtain the first execution list of the current round for the certain first operation dimension and the certain first operation, specifically includes:

[0032] If the user dialogue instruction of this round includes an operation of selecting a list item / sorting list items, identifying, based on the context, that the object list specified in the user dialogue instruction of this round is the original list or an execution list of a round in the previous N-1 rounds, and performing the operation of selecting a list item / sorting list items on a first operation dimension corresponding to the operation of selecting a list item / sorting list items in the object list of this round, to obtain a first execution list of this round for the operation of selecting a list item / sorting list items in the first operation dimension;

[0033] If the user dialogue instruction of this round includes a favorite list item operation, the first object list specified in the user dialogue instruction of this round is identified as the original list or a round of execution list in the previous N-1 rounds according to the context relationship, and the second object list is a round of first execution list obtained by the previous execution of the favorite list item operation on the same first operation dimension, and the favorite list item operation is performed on a first operation dimension corresponding to the favorite list item operation in the first object list of this round, and the favorite list items are added to the second object list of this round to obtain the first execution list of this round of the favorite list item operation of a certain first operation dimension.

[0034] Furthermore, identifying the intersection / union relationship between the multiple first operation dimensions corresponding to a first operation in this round, and finding the intersection / union of the first execution lists of the corresponding multiple first operation dimensions in this round according to the intersection / union relationship to obtain the second execution list of the current round for the first operation, specifically includes:

[0035] If the user dialogue instruction in this round includes executing selection list item operations on multiple first operation dimensions, identifying intersection / union relationships between the corresponding multiple first operation dimensions, and performing intersection / union of the first execution lists of the current round of selection list item operations on the multiple first operation dimensions to obtain a second execution list of the current round of selection list item operations;

[0036] If the user dialogue instructions in this round include executing favorite list item operations on multiple first operation dimensions, identify that the corresponding multiple first operation dimensions are in a union relationship, and calculate the union of the first execution lists of this round of favorite list item operations of the multiple first operation dimensions to obtain the second execution list of this round of favorite list item operations.

[0037] Furthermore, identifying a replacement relationship between the first operations of the current round, and performing replacement on the second execution list of the current round of the first operations according to the replacement relationship, specifically includes:

[0038] If the user dialogue instructions in this round include sorting the list items after selecting a list item / collecting a list item, the second execution list of this round of selecting a list item / collecting a list item is replaced with the sorted list.

[0039] In a second aspect, the present application provides a device for implementing list operations based on heterogeneous loop dialogues, the device comprising:

[0040] The acquisition module is used to obtain the original list and multiple predefined operation dimensions and multiple operations;

[0041] The dialogue module, connected to the acquisition module, is used to perform heterogeneous cycle-based multi-round dialogues with the user, including:

[0042] The receiving unit is used to receive the Nth round of user dialogue instructions,

[0043] The recognition unit is connected to the receiving unit and is used to recognize the first operation dimension, the first operation and the list of objects pointed to in the Nth round of user dialogue instructions.

[0044] an operation unit connected to the identification unit, configured to perform a first operation on a first operation dimension in the object list to obtain an Nth round execution list,

[0045] The object list is the original list or a round execution list in the first N-1 rounds.

[0046] In a third aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the method for implementing list operations based on heterogeneous loop dialogue as described above is implemented.

[0047] The present application provides a method, device, and medium for implementing list operations based on heterogeneous loop dialogues. A heterogeneous loop dialogue is implemented with a user based on list interaction. Certain lists obtained in the previous multiple rounds are used as the basis for a new round of dialogue according to user dialogue instructions. This can adapt to complex dialogue scenarios where the user's thinking jumps. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flowchart of a method for implementing list operations based on heterogeneous loop dialogues according to an embodiment of the present application;

[0049] Figure 2 This is a schematic diagram of the structure of a device for implementing list operations based on heterogeneous loop dialogues according to an embodiment of the present application;

[0050] Figure 3 This is a structural diagram of a dialogue module according to an embodiment of the present application;

[0051] Figure 4 Schematic diagram comparing various dialogue structures in the embodiments of the present application, where a is a single-round dialogue structure, b is a heterogeneous linear dialogue structure, c is a homogeneous loop dialogue structure, and d is a heterogeneous loop dialogue structure;

[0052] Figure 5 This is a schematic diagram illustrating a heterogeneous loop dialogue according to an embodiment of the present application;

[0053] Figure 6 This is an information flow diagram of a method for implementing list operations based on heterogeneous loop dialogues in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the technical solution of the present application, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0055] It should be understood that the specific embodiments and drawings described herein are only used to explain the present application, rather than to limit the present application.

[0056] It can be understood that, in the absence of conflict, the various embodiments and features in the embodiments of the present application can be combined with each other.

[0057] It will be understood that, for the sake of ease of description, the drawings of this application only show the parts related to this application, while the parts not related to this application are not shown in the drawings.

[0058] It can be understood that each module and unit involved in the embodiments of the present application may correspond to only one physical structure, or may be composed of multiple physical structures, or multiple modules and units may be integrated into one physical structure.

[0059] It is understood that, in the absence of conflict, the functions and steps marked in the flowcharts and block diagrams of the present application may occur in an order different from that marked in the drawings.

[0060] It is understood that the flowcharts and block diagrams of the present application illustrate the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the various embodiments of the present application. Each box in the flowchart or block diagram may represent a module, unit, program segment, or code, which contains executable instructions for implementing the specified functions. Moreover, each box or combination of boxes in the block diagram and flowchart may be implemented by a hardware-based device that implements the specified functions, or by a combination of hardware and computer instructions.

[0061] It can be understood that the modules and units involved in the embodiments of the present application can be implemented by software or hardware, for example, the modules and units can be located in a processor.

[0062] Example 1:

[0063] like Figure 1 As shown, the present application provides a method for implementing list operations based on heterogeneous loop dialogues, the method comprising:

[0064] S1, obtaining an original list and multiple predefined operation dimensions and multiple operations;

[0065] S2. Conduct a multi-round conversation with the user based on heterogeneous cycles, including:

[0066] Receive the Nth round of user dialogue instructions,

[0067] Identify the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions.

[0068] Execute the first operation on the first operation dimension in the object list to obtain the Nth round execution list.

[0069] The object list is the original list or a round execution list in the first N-1 rounds.

[0070] In this embodiment, the method realizes heterogeneous cyclic dialogue with the user based on list interaction, and uses certain lists obtained in the previous rounds as the basis for a new round of dialogue according to the user's dialogue instructions, which can adapt to complex dialogue scenarios where the user's thinking jumps. Figure 1 The method shown is applied to Figure 2 and 3 The device shown.

[0071] More specifically, intelligent customer service has been widely adopted across various industries in modern society, allowing users to trigger functions within the graphical user interface (GUI) through text or voice commands. However, existing carriers' intelligent customer service systems have very limited support for accessing list information through text or voice. Users can only browse through the list linearly or switch back to the GUI interaction mode. This entire process is inefficient and user-unfriendly, and many users still choose to contact human customer service. This significantly wastes labor costs and fails to meet users' personalized needs, severely reducing the user experience. Lists are a common and important component of intelligent customer service GUIs, containing a large number of items with rich semantic information that must be retrieved during GUI usage. The voice interface's difficulty in providing list information significantly limits its interactive capabilities, preventing users from completing interactive tasks that require GUI list information, such as selecting and ordering their desired packages. Voice assistants are widely used in smartphones, allowing users to filter tariff plans through voice commands, making the entire package selection process convenient and efficient. However, the aforementioned operational issues still exist.

[0072] In light of this, this embodiment provides an intelligent tariff package selection system and method based on heterogeneous cyclical dialogue and list interaction. This system aims to help users select the most suitable package through natural and intelligent voice interaction. The system supports multiple rounds of heterogeneous cyclical dialogue and can dynamically obtain, filter, and sort list information to optimize the user's interactive experience. Based on the conversation between the user and the system, the system helps users select packages. This system leverages voice human-computer interaction to fully consider user needs across multiple dimensions (price, bandwidth, network type, location, etc.), helping users accurately and quickly select the appropriate package. It also supports a heterogeneous cyclical dialogue structure, allowing users to freely switch between different dialogue states, thus optimizing the interactive experience. By introducing a "heterogeneous cyclical dialogue" approach, the system gradually identifies user needs over multiple rounds of dialogue and dynamically adjusts the recommendation list in real time, ensuring that each recommendation accurately matches the user's personalized needs. This system fully considers the user's immediate needs and provides multiple historical records, potentially including dimensions such as price, bandwidth, network type, and location. It also fully considers price, SMS, MMS, and other indicators, taking into account user preferences, providing more highly matched packages and helping users accurately and quickly select the appropriate package. The adopted "heterogeneous cyclic dialogue structure" supports multiple parallel dialogue states and free switching, and can provide a more flexible and natural interactive experience according to the changes in different user needs and context; it can dynamically adjust the recommendation list (Nth round execution list) in multiple rounds of dialogue, and intelligently optimize the recommendation results based on user input and historical behavior to ensure that each recommendation is more accurate and more in line with user needs; it supports pure voice input and feedback.

[0073] In one embodiment, S1, obtaining an original list and a plurality of predefined operation dimensions and a plurality of operations, specifically includes:

[0074] Get the list of original tariff packages preset by the operator;

[0075] Obtain the consideration dimensions of the user's package subscription, including package identification, package price, several package service contents, and package attributes. Define an operation dimension based on each consideration dimension.

[0076] Get the user's habitual operations on the package list when selecting a package. Habitual operations include selecting list items, adding list items to favorites, and sorting list items. Define an operation based on each habitual operation.

[0077] In this embodiment, there is currently no efficient solution for selecting appropriate tariffs through human-computer interaction, and existing human-computer interaction methods have the following problems: existing methods that support smartphone interaction correspond one-to-one with the graphical user interface (GUI). The system can only access a very limited amount of interface content and cannot access list data, forcing users to complete interaction tasks without obtaining complete information. Existing voice assistants only support single-round conversations with preset information for dimensions such as price and time, and cannot provide optimal solutions in multiple dimensions (package price, in-package calls, in-package data usage, etc.). In existing heterogeneous linear conversation systems, users interact with the system through multiple rounds of conversations, and the system filters and sorts based on the activated business scenarios in each round. However, this method is overly dependent on previous business scenarios, cannot flexibly respond to changes in user needs, and cannot update recommendation results in real time to adapt to users' dynamic needs. Because the trigger conditions for each round of conversation are fixed, the system lacks flexible multi-state switching and dynamic adaptation capabilities, resulting in unpersonalized and unresponsive recommendations. To address the above issues, this embodiment proposes a human-computer interaction system based on list access to select tariffs, optimizes data with strong list attributes in the tariff selection process, supports pure voice input and output, and ensures the importance of pure voice information access. By modeling and interpreting user behavior in package selection, this system uses a heterogeneous loop dialogue system to support multiple dialogue states, so that users can obtain the required package tariff type naturally, completely and accurately.

[0078] In one embodiment, identifying the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions specifically includes:

[0079] Identifying, based on keywords set for each operation dimension and each operation, a plurality of first operation dimensions and corresponding first operations included in the Nth round of user dialogue instructions;

[0080] Obtain the previous N-1 rounds of user dialogue instructions, and identify the contextual relationship between the Nth round of user dialogue instructions and the previous N-1 rounds of user dialogue instructions;

[0081] According to the context and several first operations, the original list or a list of execution rounds in the previous N-1 rounds is obtained as the object list pointed to by the current round.

[0082] In this embodiment, if Figure 4 The figure shows a comparison of several dialogue structures. The dialogue structure describes the transition relationship between dialogue and state. Based on the analysis of dialogue structures supported by the dialogue information acquisition system, the main types are as follows:

[0083] a) Single-turn dialogue: In this structure, the user issues a command, and the system responds and returns a result. There is no continuous interaction or context between the user and the system. Some systems only support single-turn dialogue, with no connection between different user commands. Although some systems introduce multi-turn dialogue to handle ambiguity in user commands, users do not express further interactive intent in these conversations, so they are still essentially single-turn dialogue structures.

[0084] b) Heterogeneous linear dialogue: This structure supports multiple rounds of dialogue, but user commands and system responses follow a predefined path. The system filters or navigates by parsing input keywords, but this lacks flexibility. This type of system includes multiple dialogue states and supports multiple rounds of user dialogue. However, users can only transition between dialogue states along a given path. For example, users acquire information through voice through three stages: keyword input, result browsing, and content browsing.

[0085] c) Homogeneous Loop Dialogue: In this structure, switching between and managing different dialogue states is relatively simple. Homogeneous loop dialogue systems combine historical interactions to understand the semantics of commands. However, these systems only support a single dialogue state, and the intent of a user's command is the sum of the intents of all previous commands. For example, if list information is organized into a decision tree, the user can only traverse the tree by continuously adding filter conditions to ultimately locate the target item.

[0086] d) Heterogeneous Circular Dialogue: This system supports different types of dialogue (e.g., call charges, data usage, and call time), and allows users to switch freely between these states. This system is suitable for complex scenarios, such as first checking the package, then deciding on the data package, and finally ordering the voice package. The heterogeneous cyclical dialogue system supports multiple dialogue states (call time, package data usage, package price, etc.), and users can switch freely between these dialogue states. Users can filter list information by jumping through historical commands. The system understands and integrates the user's command intent, supporting multiple rounds of interleaved interaction.

[0087] The list-based interactive system designed in this embodiment supports a heterogeneous loop structure. The system can support multiple parallel conversation states and switch freely between these states according to the user's instructions. The system can maintain multiple different conversation states, such as one state for processing call duration screening and another state for processing traffic package recommendations. Users can switch at any time, which is more in line with the user's natural interactive behavior. The same is true for the different operations specified in the user instructions. There are many types of instructions, which mainly record the content of the table items filtered out by the user's intention each time. Through different states, you can customize the call time, traffic, package price, etc. The user interaction system of this embodiment supports the user's natural interactive behavior based on a heterogeneous loop structure. The difference between heterogeneous loop dialogues and the other three is shown in Table 1 below:

[0088] Table 1 Comparison of various dialogue structures

[0089]

[0090] In one embodiment, wherein:

[0091] The keywords set for each operation dimension correspond to several list tags in the original tariff package list, including:

[0092] The package identifier corresponds to the rate code, rate name, and rate description.

[0093] The package price corresponds to the monthly rental fee and the price of traffic, voice, SMS and MMS outside the package. The service contents of several packages correspond to the amount of traffic, voice duration, number of SMS and MMS within the package.

[0094] Package attributes include the network type, location, and whether the package is shareable, transferable, contracted, broadband, and TV.

[0095] Each list tag has several pre-set operations that it supports.

[0096] In this embodiment, through keyword judgment, such as storing keywords in the form of a dictionary in the system, for example, call time includes keywords defined in the form of call time, chat time, call duration, etc., when the system captures keywords in different states, it will automatically switch between different states. In the implementation of the package selection task, this embodiment summarizes the main dimensions that affect user choices in the rate selection process, and sorts out each dimension label, label data type, and table operations supported by each label. Different labels can support different operations. An example of displaying a rate list label is shown in Table 2 below.

[0097] Table 2 Tariff dimension examples

[0098] Label Tag Type Support Operations Tariff Code char Ask, refer, collect, browse Rate Name char Ask, refer, collect, browse Rate Description char Ask, refer, collect, browse Monthly rental fee int Filter, sort, query, statistics, collection, browse Flow rate within the set int Filter, sort, query, statistics, collection, browse In-suite voice int Filter, sort, query, statistics, collection, browse In-suite text messages int Filter, sort, query, statistics, collection, browse In-package MMS int Filter, sort, query, statistics, collection, browse Out-of-pocket traffic price float Filter, sort, query, statistics, collection, browse Out-of-package voice price float Filter, sort, query, statistics, collection, browse Out-of-package SMS price float Filter, sort, query, statistics, collection, browse Package MMS price float Filter, sort, query, statistics, collection, browse Share bool Filter, query, save, browse Whether to carry forward bool Filter, query, save, browse Is there a contract bool Filter, query, save, browse Network Type char Ask, collect, browse Place of origin char Filter, query, save, browse Is there broadband? bool Filter, query, save, browse Is there a TV? bool Filter, query, save, browse

[0099] In one embodiment, performing a first operation on a first operation dimension in the object list to obtain an Nth round execution list specifically includes:

[0100] Sequentially selecting a first operation dimension and a corresponding first operation from the multiple first operation dimensions and the corresponding multiple first operations of this round;

[0101] Execute a corresponding first operation on a first operation dimension in the object list of this round to obtain a first execution list of this round of the first operation dimension and the first operation;

[0102] Identify the intersection / union relationship between the multiple first operation dimensions corresponding to a first operation in this round, and calculate the intersection / union of the first execution lists of the corresponding multiple first operation dimensions in this round according to the intersection / union relationship to obtain the second execution list of the current round for the first operation;

[0103] Identify the replacement relationship between the first operations of the current round, and replace the second execution list of the current round of the first operations according to the replacement relationship to obtain a third execution list of the current round.

[0104] In this embodiment, based on user research, a list-based package interaction system was designed, implementing a heterogeneous, cyclical dialogue approach. Lists are a key information storage method. Through heterogeneous, cyclical dialogues, the list content is continuously updated. After the dialogue ends, a final list that meets the user's requirements is obtained. List operations can be broadly divided into three categories, totaling seven operations: 1) Modification operations: These result in changes to the list content, such as filtering, sorting, and adding to favorites; 2) Indication operations: These indicate a specific list item or items, such as referencing or browsing; and 3) Information operations: These are used to obtain specific attributes (inquiries) or overall information (statistics) about a particular attribute in the list. These classifications are based on analysis of conversation data. Modification operations are associated with the user's selection intent. This embodiment primarily records tables for these operations. This allows for further recording of valid operation information, making operations more explicit and facilitating state transitions. Different conversations trigger different operations, each resulting in a different table (satisfying the output of the previous round of conversation). Users can initiate a new round of conversation based on the existing table output. The specific descriptions of each of the above list operations are as follows: Filtering, users use filtering operations to delete items that do not meet user needs; Sorting, users use sorting operations to adjust the list order and promote the position of user-preferred items in the list; Favorites, users use favorites operations to add one or more list items to favorites to facilitate subsequent in-depth comparison. The favorite operation needs to be combined with the reference operation to determine the items to be collected; Reference, users use reference operations to indicate the items they have visited; Browse, users use browse operations to access items in the current list based on the list index; Query, users use query operations to obtain the attribute values ​​of items; Statistics, users use statistics operations to obtain the overall situation of attributes, such as average, most, least, most expensive, cheapest, etc.

[0105] In one embodiment, based on the context relationship and the plurality of first operations, obtaining an object list pointed to by an original list or a certain execution list in the previous N-1 rounds as the current round, and performing a first operation corresponding to a first operation dimension in the object list of the current round to obtain a first execution list of the current round for the first operation dimension and the first operation, specifically includes:

[0106] If the user dialogue instruction of this round includes an operation of selecting a list item / sorting list items, identifying, based on the context, that the object list specified in the user dialogue instruction of this round is the original list or an execution list of a round in the previous N-1 rounds, and performing the operation of selecting a list item / sorting list items on a first operation dimension corresponding to the operation of selecting a list item / sorting list items in the object list of this round, to obtain a first execution list of this round for the operation of selecting a list item / sorting list items in the first operation dimension;

[0107] If the user dialogue instruction of this round includes a favorite list item operation, the first object list specified in the user dialogue instruction of this round is identified as the original list or a round of execution list in the previous N-1 rounds according to the context relationship, and the second object list is a round of first execution list obtained by the previous execution of the favorite list item operation on the same first operation dimension, and the favorite list item operation is performed on a first operation dimension corresponding to the favorite list item operation in the first object list of this round, and the favorite list items are added to the second object list of this round to obtain the first execution list of this round of the favorite list item operation of a certain first operation dimension.

[0108] In this embodiment, by investigating the logic of package selection, learning user selection habits, designing a user cognitive model, and designing a package selection human-computer interaction system based on list operation. This system constructs the current interaction context by learning user interaction behavior, system voice feedback, and list abstract results to infer user cognition, and based on this, supplements the implicit operation of user instructions and accurately understands user intentions. The interaction diagram is as follows Figure 5 As shown. Through the interaction between the user and the system, the system learns the user's requirements for call duration, traffic volume, and package price, and provides a package that meets the user's requirements. The content of the interaction is the user's requirements for each indicator in the package. The system voice feedback is the system's feedback to the user. The feedback content is the refreshed table information. For example, if the user wants to use more traffic and the package price is not higher than 30 yuan, the table information will be updated, and the packages with a price lower than 30 yuan will be sorted in descending order according to the traffic volume. This table is the final result of the package content category. For example, Figure 5 In the interaction diagram, "there are 3 items with a price less than 80 and meeting all the previous conditions", "meeting all the previous conditions" is an implicit operation and needs to be obtained based on context information. The final result is a rearranged table. For example, the original list may be a list arranged by package price. When a conversation occurs, the user needs a list that meets the requirement of at least 100 minutes of calls per month. The list modification part will delete the calls with a time of less than 100 minutes to form a new list, which will be modified in sequence. Figure 6 As shown, historical access data includes previous operation content, table content, etc., and current access content is the current conversation content between the user and the system. Different conversation contents can produce different operation methods (filtering, sorting, collecting, browsing, etc.). In the process of switching between different tags, a heterogeneous circular conversation is formed.

[0109] In one embodiment, identifying an intersection / union relationship between a plurality of first operation dimensions corresponding to a first operation in a current round, and performing intersection / union of the corresponding first execution lists of the current round of the plurality of first operation dimensions according to the intersection / union relationship to obtain a second execution list of the current round of the first operation specifically includes:

[0110] If the user dialogue instruction in this round includes executing selection list item operations on multiple first operation dimensions, identifying intersection / union relationships between the corresponding multiple first operation dimensions, and performing intersection / union of the first execution lists of the current round of selection list item operations on the multiple first operation dimensions to obtain a second execution list of the current round of selection list item operations;

[0111] If the user dialogue instructions in this round include executing favorite list item operations on multiple first operation dimensions, identify that the corresponding multiple first operation dimensions are in a union relationship, and calculate the union of the first execution lists of this round of favorite list item operations of the multiple first operation dimensions to obtain the second execution list of this round of favorite list item operations.

[0112] In this embodiment, the package selection logic is based on the comparison and screening of price, traffic, and number of voices; operations on these packages (sorting, collecting, etc.) are the user's selection habits; the entire conversation process is based on the user's cognitive model to complete operations such as sorting and collecting some tag content; in one conversation, the user can simultaneously filter the package price, traffic, and voice. When identifying the user conversation, it can jump to the list obtained after a certain round of execution. Lists can be recorded separately for multiple operation dimensions and multiple operations. In view of the human characteristic of jumping thinking, it can look forward to find a suitable list for this round of processing, which reflects the characteristics of heterogeneous cyclic conversation. There are seven list operation options that users can perform in package selection. The user selection of the context can be retained, and the table interaction system can be operated to finally select a package that meets the customer's expectations. The information processing and table operation flow diagram is shown as follows. Figure 6 shown.

[0113] In one embodiment, identifying a replacement relationship between a plurality of first operations in a current round, and performing replacement on a second execution list of the current round of the plurality of first operations according to the replacement relationship, specifically includes:

[0114] If the user dialogue instructions in this round include sorting the list items after selecting a list item / collecting a list item, the second execution list of this round of selecting a list item / collecting a list item is replaced with the sorted list.

[0115] In this embodiment, new lists are recorded mainly for three of the seven list operations, including filtering, sorting, and collection. Different tables are formed according to the actual needs corresponding to the three operations. The filtering operation records the filtered tables for each operation dimension separately, such as items with traffic greater than 30 and items with prices less than 80, and then the intersection of the two is calculated to obtain the items selected by the user in this round. If the user modifies the instruction to an item with a price less than 100 in a subsequent round of dialogue, the item with traffic greater than 30 is still retained in combination with the context to better identify the user's intention. For the remaining operations, the design of which lists to record and how to operate between lists is based on the actual human habits of using these instructions.

[0116] In summary, this embodiment proposes a list-based interactive rate selection system based on a heterogeneous cyclic dialogue structure, which can flexibly support users to switch freely between multiple dialogue states and provide a natural and smooth interactive experience. The designed list-level interaction method can intelligently match and filter between multiple dimensions (such as package price, call duration, data traffic, etc.), ensuring that users receive package recommendations that best suit their needs based on multi-dimensional information. It supports pure voice interaction, and users can directly perform complex list operations with the system through voice input and feedback without relying on a graphical user interface (GUI). In a fully voice-based environment, users can still efficiently select packages that meet their needs. It is intelligent and efficient, and can solve the problems of low accuracy and low efficiency caused by manual comparison and recommendation in the current package selection process, thereby improving customer perception and enhancing corporate image.

[0117] Example 2:

[0118] like Figure 2 and 3 As shown, the present application provides a device for implementing list operations based on heterogeneous loop dialogues, the device comprising:

[0119] Acquisition module 1, used to acquire the original list and multiple predefined operation dimensions and multiple operations;

[0120] The dialogue module 2 is connected to the acquisition module 1 and is used to perform a multi-round dialogue with the user based on a heterogeneous cycle, including:

[0121] The receiving unit 21 is used to receive the Nth round of user dialogue instructions,

[0122] The identification unit 22 is connected to the receiving unit 21 and is used to identify the first operation dimension, the first operation and the list of objects pointed to in the Nth round of user dialogue instructions.

[0123] The operation unit 23 is connected to the identification unit 22 and is used to perform a first operation on the first operation dimension in the object list to obtain an N-th round execution list.

[0124] The object list is the original list or a round execution list in the first N-1 rounds.

[0125] In one embodiment, the acquisition module 1 specifically includes:

[0126] Get the original list unit, used to obtain the original tariff package list preset by the operator;

[0127] The operation dimension acquisition unit is used to obtain the consideration dimensions of the user's package subscription. The consideration dimensions include package identifier, package price, several package service contents, and package attributes. An operation dimension is defined based on each consideration dimension.

[0128] The operation acquisition unit is used to obtain the user's habitual operations on the package list when selecting a package. The habitual operations include selecting list items, adding list items to favorites, and sorting list items. An operation is defined based on each habitual operation.

[0129] In one embodiment, the identification unit 22 specifically includes:

[0130] A keyword recognition unit, configured to recognize a plurality of first operation dimensions and corresponding first operations contained in the Nth round of user dialogue instructions based on keywords set for each operation dimension and each operation;

[0131] A context recognition unit is used to obtain the previous N-1 rounds of user dialogue instructions and identify the contextual relationship between the Nth round of user dialogue instructions and the previous N-1 rounds of user dialogue instructions;

[0132] The object recognition unit is connected to the keyword recognition unit and the context recognition unit, and is used to obtain the original list or a round execution list in the previous N-1 rounds as the object list pointed to by the current round based on the context relationship and a plurality of first operations.

[0133] In one embodiment, wherein:

[0134] The keywords set for each operation dimension correspond to several list tags in the original tariff package list, including:

[0135] The package identifier corresponds to the rate code, rate name, and rate description.

[0136] The package price corresponds to the monthly rental fee and the price of traffic, voice, SMS and MMS outside the package.

[0137] Several package service contents correspond to the amount of data traffic, voice duration, number of SMS messages, and number of MMS messages within the package.

[0138] Package attributes include the network type, location, and whether the package is shareable, transferable, contracted, broadband, and TV.

[0139] Each list tag has several pre-set operations that it supports.

[0140] In one embodiment, the operating unit 23 specifically includes:

[0141] A sequential selection unit, configured to sequentially select a first operation dimension and a corresponding first operation from a plurality of first operation dimensions and corresponding first operations of this round;

[0142] A first execution unit, connected to the sequential selection unit, is configured to execute a corresponding first operation on a first operation dimension in the object list of this round, so as to obtain a first execution list of this round for the first operation dimension and the first operation;

[0143] A second execution unit, connected to the first execution unit, is configured to identify an intersection / union relationship between multiple first operation dimensions corresponding to a first operation in this round, and calculate an intersection / union of the first execution lists of this round of the corresponding multiple first operation dimensions based on the intersection / union relationship to obtain a second execution list of this round of the first operation;

[0144] The third execution unit is connected to the second execution unit and is used to identify the replacement relationship between the multiple first operations in this round, and replace the multiple first operations in the second execution list of this round according to the replacement relationship to obtain the third execution list of this round.

[0145] In one embodiment, the context identification unit and the first execution unit specifically include:

[0146] a selection / sorting identification unit configured to, if the current round of user dialogue instructions includes a select list item operation / sort list item operation, identify, based on context, the object list specified in the current round of user dialogue instructions as the original list or an execution list in a previous N-1 rounds; and a selection and sorting execution unit configured to execute the select list item operation / sort list item operation on a first operation dimension corresponding to the select list item operation / sort list item operation in the current round of object lists, to obtain a first execution list of the current round for the select list item operation / sort list item operation in the first operation dimension;

[0147] A collection identification unit is used to identify, if the current round of user dialogue instructions includes a collection list item operation, whether the first object list specified in the current round of user dialogue instructions is the original list or a certain round of execution list in the previous N-1 rounds, and the second object list is a certain round of first execution list obtained by the previous execution of the collection list item operation on the same first operation dimension; and a collection execution unit is used to execute the collection list item operation on a certain first operation dimension corresponding to the collection list item operation in the current round of the first object list, add the collection list item to the current round of the second object list, and obtain the current round of the first execution list of the collection list item operation of the certain first operation dimension.

[0148] In one embodiment, the second execution unit specifically includes:

[0149] A selection intersection and union unit is configured to, if the current round of user dialogue instructions includes executing selection list item operations on multiple first operation dimensions, identify the intersection / union relationship between the corresponding multiple first operation dimensions, and calculate the intersection / union of the current round of first execution lists of the selection list item operations on the multiple first operation dimensions to obtain a current round of second execution list of the selection list item operations;

[0150] The collection merging unit is used to identify that the corresponding multiple first operation dimensions are in a union relationship if the current round of user dialogue instructions includes executing collection list item operations on multiple first operation dimensions, and to calculate the union of the current round of first execution lists of the collection list item operations of the multiple first operation dimensions to obtain the current round of second execution list of the collection list item operations.

[0151] In one embodiment, the third execution unit specifically includes:

[0152] The sorting replacement unit is used to replace the second execution list of the current round of the selection list item operation / the collection list item operation with the sorted list if the current round of user dialogue instructions includes performing a sorting operation on the list items after performing the selection list item operation / the collection list item operation.

[0153] Example 3:

[0154] Embodiment 3 of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the method for implementing list operations based on heterogeneous loop dialogues as described in Embodiment 1 or the device for implementing list operations based on heterogeneous loop dialogues as described in Embodiment 2 is implemented.

[0155] The computer-readable storage medium includes volatile or non-volatile, removable or non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, computer program elements or other data). Computer-readable storage media include, but are not limited to, RAM (Random Access Memory), ROM (Read-Only Memory), EEPROM (Electrically Erasable Programmable read only memory), flash memory or other memory technology, CD-ROM (Compact Disc Read-Only Memory), digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer.

[0156] In addition, the present application may also provide a computer device comprising a memory and a processor, wherein the memory stores a computer program. When the processor executes the computer program stored in the memory, the processor executes the method for implementing list operations based on heterogeneous loop dialogues as described in Example 1. The computer device may be the device for implementing list operations based on heterogeneous loop dialogues as described in Example 2.

[0157] The memory is connected to the processor, the memory may be a flash memory, a read-only memory or other memory, and the processor may be a central processing unit or a single-chip microcomputer.

[0158] Embodiments 1-3 of the present application provide a method, device, and medium for implementing list operations based on heterogeneous loop dialogues, implementing heterogeneous loop dialogues with users based on list interactions, and using certain lists obtained in the previous multiple rounds as the basis for a new round of dialogue according to user dialogue instructions, which can adapt to complex dialogue scenarios where users' thoughts jump.

[0159] It is understood that the above embodiments are merely exemplary embodiments for illustrating the principles of the present application, and the present application is not limited thereto. Those skilled in the art may make various modifications and improvements without departing from the spirit and substance of the present application, and such modifications and improvements are also considered to be within the scope of protection of the present application.

Claims

1. A method for implementing list operations based on heterogeneous loop dialogue, characterized in that: The method comprises: Get the original list and predefined multiple operation dimensions and multiple operations; Conduct heterogeneous, loop-based multi-turn conversations with users, including: Receive the Nth round of user dialogue instructions, Identify the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions. Execute the first operation on the first operation dimension in the object list to obtain the Nth round execution list. The object list is the original list or a round execution list in the first N-1 rounds.

2. The method according to claim 1, characterized in that Get the original list and multiple predefined operation dimensions and multiple operations, including: Get the list of original tariff packages preset by the operator; Obtain the consideration dimensions of the user's package subscription, including package ID, package price, several package service contents, and package attributes. Define an operation dimension based on each consideration dimension. Get the user's habitual operations on the package list when selecting a package. Habitual operations include selecting list items, adding list items to favorites, and sorting list items. Define an operation based on each habitual operation.

3. The method according to any one of claims 1-2, characterized in that Identifying the first operation dimension, the first operation, and the list of objects pointed to in the Nth round of user dialogue instructions, specifically including: Identifying, based on keywords set for each operation dimension and each operation, a plurality of first operation dimensions and corresponding first operations included in the Nth round of user dialogue instructions; Obtain the previous N-1 rounds of user dialogue instructions, and identify the contextual relationship between the Nth round of user dialogue instructions and the previous N-1 rounds of user dialogue instructions; According to the context and several first operations, the original list or a list of execution rounds in the previous N-1 rounds is obtained as the object list pointed to by the current round.

4. The method according to claim 3, characterized in that in: The keywords set for each operation dimension correspond to several list tags in the original tariff package list, including: The package identifier corresponds to the rate code, rate name, and rate description. The package price corresponds to the monthly rental fee and the price of traffic, voice, SMS and MMS outside the package. The service contents of several packages correspond to the amount of traffic, voice duration, number of SMS and MMS within the package. Package attributes include the network type, location, and whether the package is shareable, transferable, contracted, broadband, and TV. Each list tag has several pre-set operations that it supports.

5. The method according to claim 3, characterized in that Perform the first operation on the first operation dimension in the object list to obtain the Nth round execution list, specifically including: Sequentially selecting a first operation dimension and a corresponding first operation from the multiple first operation dimensions and the corresponding multiple first operations of this round; Execute a corresponding first operation on a first operation dimension in the object list of this round to obtain a first execution list of this round of the first operation dimension and the first operation; Identify the intersection / union relationship between the multiple first operation dimensions corresponding to a first operation in this round, and calculate the intersection / union of the first execution lists of the corresponding multiple first operation dimensions in this round according to the intersection / union relationship to obtain the second execution list of the current round for the first operation; Identify the replacement relationship between the first operations of the current round, and replace the second execution list of the current round of the first operations according to the replacement relationship to obtain a third execution list of the current round.

6. The method according to claim 5, characterized in that According to the context relationship and several first operations, the original list or a round execution list in the previous N-1 rounds is obtained as the object list pointed to in the current round, and a first operation corresponding to a first operation dimension in the object list of the current round is performed to obtain the first execution list of the current round for the first operation dimension and the first operation, specifically including: If the user dialogue instruction of this round includes an operation of selecting a list item / sorting list items, identifying, based on the context, that the object list specified in the user dialogue instruction of this round is the original list or an execution list of a round in the previous N-1 rounds, and performing the operation of selecting a list item / sorting list items on a first operation dimension corresponding to the operation of selecting a list item / sorting list items in the object list of this round, to obtain a first execution list of this round for the operation of selecting a list item / sorting list items in the first operation dimension; If the user dialogue instruction of this round includes a favorite list item operation, the first object list specified in the user dialogue instruction of this round is identified as the original list or a round of execution list in the previous N-1 rounds according to the context relationship, and the second object list is a round of first execution list obtained by the previous execution of the favorite list item operation on the same first operation dimension, and the favorite list item operation is performed on a first operation dimension corresponding to the favorite list item operation in the first object list of this round, and the favorite list items are added to the second object list of this round to obtain the first execution list of this round of the favorite list item operation of a certain first operation dimension.

7. The method according to claim 5, characterized in that Identify the intersection / union relationship between the multiple first operation dimensions corresponding to a first operation in this round, and calculate the intersection / union of the first execution lists of the corresponding multiple first operation dimensions in this round according to the intersection / union relationship to obtain the second execution list of the first operation in this round, specifically including: If the user dialogue instruction in this round includes executing selection list item operations on multiple first operation dimensions, identifying intersection / union relationships between the corresponding multiple first operation dimensions, and performing intersection / union of the first execution lists of the current round of selection list item operations on the multiple first operation dimensions to obtain a second execution list of the current round of selection list item operations; If the user dialogue instructions in this round include executing favorite list item operations on multiple first operation dimensions, identify that the corresponding multiple first operation dimensions are in a union relationship, and calculate the union of the first execution lists of this round of favorite list item operations of the multiple first operation dimensions to obtain the second execution list of this round of favorite list item operations.

8. The method according to claim 5, characterized in that Identifying a replacement relationship between multiple first operations in this round, and performing replacement on the second execution list of the multiple first operations in this round according to the replacement relationship, specifically including: If the user dialogue instructions in this round include sorting the list items after selecting a list item / collecting a list item, the second execution list of this round of selecting a list item / collecting a list item is replaced with the sorted list.

9. A device for implementing list operations based on heterogeneous loop dialogue, characterized in that: The device comprises: The acquisition module is used to obtain the original list and multiple predefined operation dimensions and multiple operations; The dialogue module, connected to the acquisition module, is used to perform heterogeneous cycle-based multi-round dialogues with the user, including: The receiving unit is used to receive the Nth round of user dialogue instructions, The recognition unit is connected to the receiving unit and is used to recognize the first operation dimension, the first operation and the list of objects pointed to in the Nth round of user dialogue instructions. an operation unit connected to the identification unit, configured to perform a first operation on a first operation dimension in the object list to obtain an Nth round execution list, The object list is the original list or a round execution list in the first N-1 rounds.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for implementing list operations based on heterogeneous loop dialogues according to any one of claims 1 to 8 is implemented.

Images