WO2022012234A1

WO2022012234A1 - Associated recommendation method, smart device and service device

Info

Publication number: WO2022012234A1
Application number: PCT/CN2021/099448
Authority: WO
Inventors: 邵星阳; 杨善松; 刘永霞; 殷腾龙; 修媛媛; 岳文浩
Original assignee: 海信视像科技股份有限公司
Priority date: 2020-07-17
Filing date: 2021-06-10
Publication date: 2022-01-20

Abstract

Provided in the present application are an associated recommendation method, a smart device and a service device, which are used to make associated recommendations on the basis of a query phrase. The smart device comprises: an input interface and a controller. The controller is configured for use in: sending, to the service device, a query phrase which is inputted by means of the input interface so as to allow the service device to search, within an established knowledge graph, for an associated entity that corresponds to a target entity, wherein the knowledge graph is used to represent the semantic relationship between entities, and the target entity is an entity that is determined on the basis of the query phrase; and acquiring and outputting the associated entity from the service device.

Description

A kind of association recommendation method, intelligent device and service device

This application claims the priority of the Chinese patent application filed on July 17, 2020 with the application number 202010692647.0, and the application title is "A method of association recommendation, intelligent equipment and service equipment", and filed in August 2020 The priority of the Chinese patent application filed with the Chinese Patent Office on the 7th with the application number of 202010790355.0 and the application title of "An Information Recommending Method and Service Equipment", the entire contents of which are incorporated in this application by reference.

technical field

The present application relates to the technical field of smart devices, and in particular, to a method for association recommendation, a smart device and a service device.

Background technique

At present, users can query and obtain information through smart devices such as smart TVs and smart speakers. In order to better meet users' needs for information acquisition, recommendation systems emerge as the times require. The recommendation system can perform associated recommendation on the information queried by the user, so as to provide the user with relevant information of the queried information.

Relevant recommendation has many advantages. For example, in this era of information overload, in the face of massive data, it is difficult for users to query all the information that may be of interest one by one. Help users to quickly find more information that may be of interest; for another example, when a user queries for some information, such as querying a specified person, if the specified person is not found, it is recommended to recommend related people of the specified person rather than simply returning "query failure". The result is a better user experience.

In order to further improve the user experience, how to provide a more comprehensive and rich association recommendation is a problem that still needs to be solved.

SUMMARY OF THE INVENTION

Embodiments of the present application provide an association recommendation method, an intelligent device, and a service device, which are used for association recommendation according to a query sentence.

In a first aspect, an intelligent device is provided, including: an input interface and a controller;

The above controller is configured to:

Send the query statement input through the above input interface to the service device, so that the service device searches for the associated entity corresponding to the target entity in the established knowledge graph, and the above knowledge graph is used to represent the semantic relationship between entities, and the above target entity is The entity determined according to the above query statement;

Obtain the above-mentioned associated entity from the above-mentioned service device and output it.

In a second aspect, a service device is provided, including:

The receiving module is used to receive the current query statement from the smart device;

The association recommendation module is used to find the associated entity corresponding to the target entity in the established knowledge graph, the knowledge graph is used to represent the semantic relationship between the entities, and the target entity is the entity determined according to the current query sentence.

In a third aspect, a method for association recommendation is provided, applied to a smart device, including:

Send the query statement input through the input interface to the service device, so that the service device searches for the associated entity corresponding to the target entity in the established knowledge graph. The above-mentioned knowledge graph is used to represent the semantic relationship between entities, and the above-mentioned target entity is based on The entity identified by the above query statement;

In a fourth aspect, an association recommendation method is provided, applied to a service device, including:

Receive the current query statement from the smart device;

The associated entity corresponding to the target entity is searched in the established knowledge graph, the above knowledge graph is used to represent the semantic relationship between entities, and the above target entity is the entity determined according to the above current query sentence.

In a fifth aspect, a service device is provided, including: an input interface and a processor;

The processor is configured to:

When a query statement is received through the input interface, the query statement is input into the configured question answering system to obtain a reply statement corresponding to the query statement;

Find the answer entity associated with the reply sentence in the established knowledge graph;

A recommended sentence is generated according to the corresponding relationship between attributes and attribute values configured by the answer entity; the recommended sentence includes a preset number of N attributes and corresponding attribute values, where N is greater than or equal to 1.

A sixth aspect provides an information recommendation method, including:

When receiving a query statement, inputting the query statement into the configured question answering system to obtain a reply statement corresponding to the query statement;

Description of drawings

In order to illustrate the embodiments of the present application more clearly, the accompanying drawings required in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, without creative efforts Additional drawings can be obtained from these drawings.

FIG. 1 exemplarily shows a schematic diagram of an operation scene between a display device and a control apparatus according to some embodiments;

FIG. 2 exemplarily shows a hardware configuration block diagram of a display device 200 according to some embodiments;

FIG. 3 exemplarily shows a hardware configuration block diagram of the control device 100 according to some embodiments;

FIG. 4 exemplarily shows a schematic diagram of a network architecture provided by an embodiment of the present application;

FIG. 5 exemplarily shows a flowchart of the association recommendation method provided by the embodiment of the present application;

6A-6G exemplarily show the schematic diagrams of the display screen when the associated entity is output;

FIG. 7 exemplarily shows a flowchart of an association recommendation method applied to a service device;

FIG. 8 illustrates a schematic diagram of a knowledge graph according to some embodiments;

FIG. 9 exemplarily shows a flowchart of an information recommendation method according to some embodiments;

FIG. 10 exemplarily shows a schematic diagram of a GUI provided by a service device when displaying a recommendation sentence according to some embodiments;

FIG. 11 exemplarily shows a flow chart of the implementation of step 103 according to some embodiments;

FIG. 12 exemplarily shows a flowchart of the implementation of step 333 according to some embodiments.

detailed description

In order to make the purpose and implementation of the present application clearer, the exemplary embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the exemplary embodiments of the present application. Obviously, the described exemplary embodiments are only the Some embodiments are claimed, but not all embodiments.

It should be noted that the brief description of the terms in the present application is only for the convenience of understanding the embodiments described below, rather than intended to limit the embodiments of the present application. Unless otherwise specified, these terms are to be understood according to their ordinary and ordinary meanings.

The terms "first", "second", "third", etc. in the description and claims of this application and the above drawings are used to distinguish similar or similar objects or entities, and are not necessarily meant to limit specific Sequential or sequential, unless otherwise noted. It is to be understood that the terms so used are interchangeable under appropriate circumstances.

The terms "comprising" and "having", and any variations thereof, are intended to cover but not exclusively include, for example, a product or device that incorporates a series of components is not necessarily limited to all components explicitly listed, but may include no explicit other components listed or inherent to these products or devices.

The term "module" refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic, or combination of hardware or/and software code capable of performing the functions associated with that element.

FIG. 1 is a schematic diagram of an operation scenario between a display device and a control apparatus according to an embodiment. As shown in FIG. 1 , the user can operate the display device 200 through the smart terminal 300 or the control device 100 .

The control apparatus 100 may be a remote controller, and the communication between the remote controller and the display device includes infrared protocol communication or Bluetooth protocol communication, and other short-distance communication methods, and the display device 200 is controlled wirelessly or wiredly. The user can control the display device 200 by inputting user instructions through keys on the remote control, voice input, control panel input, and the like.

In some embodiments, a smart terminal 300 (eg, a mobile terminal, a tablet computer, a computer, a notebook computer, etc.) can also be used to control the display device 200 . For example, the display device 200 is controlled using an application running on the smart device.

In some embodiments, the display device 200 may also be controlled in a manner other than the control apparatus 100 and the smart device 300. For example, the module for acquiring voice commands configured inside the display device 200 may directly receive the user's voice command for control. , the user's voice command control can also be received through a voice control device provided outside the display device 200 device.

In some embodiments, the display device 200 is also in data communication with the server 400 . The display device 200 may be allowed to communicate via local area network (LAN), wireless local area network (WLAN), and other networks. The server 400 may provide various contents and interactions to the display device 200 .

FIG. 2 exemplarily shows a configuration block diagram of the control apparatus 100 according to an exemplary embodiment. As shown in FIG. 2 , the control device 100 includes a controller 110 , a communication interface 130 , a user input/output interface 140 , a memory, and a power supply. The control device 100 can receive the user's input operation instruction, and convert the operation instruction into an instruction that the display device 200 can recognize and respond to, and play an intermediary role between the user and the display device 200 .

FIG. 3 is a block diagram showing a hardware configuration of the display apparatus 200 according to an exemplary embodiment.

Display apparatus 200 includes at least some of tuner 210, communicator 220, detector 230, external device interface 240, controller 250, display 275, audio output interface 285, memory, power supply, and user interface.

The display 275 includes a display screen component for presenting pictures, and a driving component for driving image display, for receiving image signals output from the controller, components for displaying video content, image content, and menu manipulation interfaces, and user manipulation UI interfaces .

The display 275 can be a liquid crystal display, an OLED display, and a projection display, as well as some projection devices and projection screens.

The communicator 220 is a component for communicating with external devices or servers according to various communication protocol types. For example, the communicator may include a Wifi module, a Bluetooth module, a wired Ethernet module and other network communication protocol chips or near field communication protocol chips, and at least some of the infrared receivers. The display device 200 may establish transmission and reception of control signals and data signals with the external control device 100 or the server 400 through the communicator 220 .

The user interface can be used to receive control signals from the control device 100 (eg, an infrared remote control, etc.).

The detector 230 is used to collect external environment or external interaction signals. For example, the detector 230 includes a light receiver, a sensor for collecting ambient light intensity; alternatively, the detector 230 includes an image collector, such as a camera, which can be used to collect external environmental scenes, user attributes or user interaction gestures, or , the detector 230 includes a sound collector, such as a microphone, for receiving external sound.

The external device interface 240 may include but is not limited to the following: any one of a high-definition multimedia interface interface (HDMI), an analog or data high-definition component input interface (component), a composite video input interface (CVBS), a USB input interface (USB), an RGB port, etc. or multiple interfaces. It may also be a composite input/output interface formed by a plurality of the above-mentioned interfaces.

The controller 250 and the tuner 210 may be located in different separate devices, that is, the tuner 210 may also be located in an external device of the main device where the controller 250 is located, such as an external set-top box.

The controller 250 controls the operation of the display device and responds to the user's operation through various software control programs stored in the memory. The controller 250 controls the overall operation of the display apparatus 200 . For example, in response to receiving a user command for selecting a UI object to be displayed on the display 275, the controller 250 may perform an operation related to the object selected by the user command.

Objects can be any of the optional objects, such as hyperlinks, icons, or other actionable controls. The operations related to the selected object include: displaying operations connected to hyperlinked pages, documents, images, etc., or executing operations of programs corresponding to the icons.

In some embodiments, the user may input user commands on a graphical user interface (GUI) displayed on the display 275, and the user input interface receives the user input commands through the graphical user interface (GUI). Alternatively, the user may input a user command by inputting a specific sound or gesture, and the user input interface recognizes the sound or gesture through a sensor to receive the user input command.

"User interface" can refer to the medium interface for interaction and information exchange between application programs or operating systems and users, which realizes the conversion between the internal form of information and the form acceptable to users. The commonly used form of user interface is Graphical User Interface (GUI), which refers to a user interface related to computer operations displayed in a graphical manner. It can be an icon, window, control and other interface elements displayed on the display screen of the electronic device, wherein the control can include icons, buttons, menus, tabs, text boxes, dialog boxes, status bars, navigation bars, Widgets, etc. visual interface elements. In some embodiments, a system may include a kernel (Kernel), a command parser (shell), a file system, and applications. Together, the kernel, shell, and file system make up the basic operating system structures that allow users to manage files, run programs, and use the system. After power-on, the kernel starts, activates the kernel space, abstracts hardware, initializes hardware parameters, etc., runs and maintains virtual memory, scheduler, signals and inter-process communication (IPC). After the kernel starts, the shell and user applications are loaded. An application is compiled into machine code after startup, forming a process.

In order to provide a more comprehensive and rich association recommendation, the embodiment of the present application provides an association recommendation method, which can determine the corresponding target entity according to the knowledge graph when it is determined that the target entity needs to be associated recommendation according to the query entered by the user. , and output the associated entity. When the user asks the associated object of the target entity, the associated entity may include the answer entity.

In order to clearly illustrate the embodiments of the present application, explanations of some related terms are given below.

Knowledge graph: Also known as scientific knowledge graph, it uses visualization technology to describe knowledge resources and their carriers, and mines, analyzes, constructs, draws and displays knowledge and their interconnections. A knowledge graph is essentially a semantic network that can represent semantic relationships between entities. In the knowledge graph, entities are used as vertices or nodes, and relationships are used as edges. The knowledge graph can be constructed in various ways, and the embodiment of the present application does not focus on how to construct the knowledge graph, so it will not be described in detail.

Entity: In the knowledge graph, a node is called an entity, and an entity can be replaced by things that exist objectively and can be distinguished from each other, such as specific people, things, things, institutions, abstract concepts, etc.

To clearly illustrate the embodiments of the present application, a network architecture provided by the embodiments of the present application is described below with reference to FIG. 4 .

Referring to FIG. 4 , FIG. 4 is a schematic diagram of a network architecture provided by an embodiment of the present application. In Figure 4, the smart device is used to receive the input information and output the processing result of the information; the speech recognition service device is an electronic device deployed with a speech recognition service, the semantic service device is an electronic device deployed with a semantic service, and the business service device Electronic devices deployed with business services. Of course, one or more services can be integrated into the same electronic device. For example, speech recognition services and semantic services can be integrated on one electronic device, and semantic services and business services can also be integrated on one electronic device. Services and business services are integrated on electronic devices. The number of electronic devices is not limited here, as long as the corresponding functions can be accomplished.

The electronic device here may include a server, a computer, etc. Here, the speech recognition service, the semantic service (also referred to as a semantic engine) and the business service are web services that can be deployed on the electronic device, wherein the speech recognition service is used to convert audio Recognized as text, the semantic service is used to semantically parse the text, and the business service is used to provide specific services such as weather query service, music query service, etc. In one embodiment, there may be multiple business service devices deployed with different business services in the architecture shown in FIG. 4 .

The following describes the process of processing the information input to the display device based on the architecture shown in FIG. 4 as an example. Taking the information input to the display device as a query sentence input by voice as an example, the above process may include the following three stages:

1. Speech recognition stage

After receiving the query sentence input by voice, the smart device can upload the audio of the query sentence to the voice recognition service device, so that the voice recognition service device can recognize the audio as text through the voice recognition service and return it to the display device.

In one embodiment, before uploading the audio of the query sentence to the speech recognition service device, the smart device may perform denoising processing on the audio of the query sentence, where the denoising processing may include steps such as removing echoes and ambient noise.

2. Semantic understanding stage

The display device uploads the text of the query sentence recognized by the speech recognition service to the semantic service device, so that the semantic service device performs semantic analysis on the text through the semantic service to obtain the business field and intent of the text.

3. Response stage

According to the semantic analysis result of the text of the query sentence, the semantic service device sends a query instruction to the corresponding business service device to obtain the query result given by the business service. The display device can obtain the query result from the semantic service device and output it. As an embodiment, the semantic service device may also send the semantic parsing result of the query statement to the display device, so that the display device outputs the feedback statement in the semantic parsing result.

So far, the description of the process of processing the information input to the display device based on the architecture shown in FIG. 4 is completed.

It should be noted that the architecture shown in FIG. 4 is only an example, and does not limit the protection scope of the present application. In the embodiments of the present application, other architectures may also be used to implement similar functions, which will not be repeated here.

The association recommendation method provided by the embodiment of the present application is described below. In one embodiment, the method can be applied to smart devices such as display devices, smart TVs, smart speakers, smart phones, and the like. When describing the various embodiments below, the description is mainly based on the application of the association recommendation method to a display device in a smart device as an example. The specific implementation of the association recommendation method applied to other smart devices is similar to that of the display device, and the following description will not be described one by one. .

From the perspective of a display device, the association recommendation method provided by the embodiment of the present application is described below with reference to FIG. 5 :

FIG. 5 exemplarily shows a flowchart of the association recommendation method provided by the embodiment of the present application. With reference to the display device shown in FIG. 2 and FIG. 5 , the process may include the following steps:

Step S01, sending the received query statement to the service device, so that the service device searches the established knowledge graph for an associated entity corresponding to a target entity, where the target entity is an entity determined according to the query statement.

In one example, the above query statement may include: a query statement input through voice, a query statement input through a control device, and the like. The user can obtain the required information by entering a query statement into the smart device. The query statement can include various contents, for example, it can be a single entity name, such as "actor A", "movie B", etc.; another example, it can be a Sentences, such as "who is the star of movie B"; another example, can be words without actual meaning, such as "that", "I think"; and so on. In one embodiment, the above query statement is a query statement input through an input interface.

In an example, the service device in this step S01 is deployed with at least a semantic service. As an embodiment, the display device may invoke a voice recognition service, and send the query sentence inputted by voice to the service device after recognizing it as text.

In this step S01, how the service device specifically determines the target entity according to the query statement and how to search for the associated entity corresponding to the target entity is described below with an example, and will not be repeated here.

Step S02, the above-mentioned associated entity is acquired from the service device and output.

In some embodiments, the search for the associated entity from the target entity is to broaden the recommended resources related to the user's voice input command. For example, when the target entity is A, the service device will return the recommended resources related to A to the terminal device, and these recommended resources are all related to A. Based on the above purpose of expanding the recommended resources, when the target entity input by the user is A, the service device also returns another entity B associated with A, which can also be selected by the user. When the user still has B based on the recommended content When B is selected, the service device will also deliver B-related recommended resources to the terminal device. Of course, after the user selects entity B, the service device can also deliver another entity C and D associated with entity B. In this way, although the user only inputs entity A once, the terminal device can present recommended resources of entity B, entity C, and entity D other than entity A, which enriches the diversity of recommended resources.

In one example, the display device outputting the associated entity may include: outputting the associated entity through a voice, displaying the associated entity through a display, and the like. The following description is given by way of example with reference to FIGS. 6A-6G , and details are not repeated here.

Through the process shown in FIG. 5 , the display device can send the received query statement to the service device, so as to obtain and output the associated entity found by the service device according to the query statement. In this way, during the interaction between the display device and the user, a relatively comprehensive association recommendation can be implemented for the query sentence input to the display device.

Figures 6A-6G exemplarily show schematic diagrams of the display screen when the associated entity is output. The following describes how the display device outputs the associated entity in the above step S02 with reference to Figures 6A-6G:

In an example, the service device returns the associated entity in the form of entity information. As an embodiment, the displaying device outputting the associated entity may include: displaying the entity information of the associated entity through a display. The above entity information may include: entity name, entity-related pictures, and the like. The entity-related pictures here, for example, when the entity is a person, the entity-related picture can be a person's avatar, a recent photo of the person, etc.; when the entity is a film and television, the entity-related picture can be a film and television cover, etc.; when the entity is a game, the entity-related pictures The image can be a game icon; etc.

Each entity corresponds to a recommended resource entry, and the entity can be selected. After receiving the user's selection of a specific entity, the entry into the resource is triggered, and the user's selection is sent to the service device, and the service device returns the corresponding recommendation according to the selected entity. resource.

FIG. 6A exemplarily shows a schematic diagram of a GUI provided by the display device when outputting the associated entity.

As shown in FIG. 6A, the display device may provide a GUI 600 to the display, the GUI 600 includes an associated entity display area 61 for displaying associated entities, and the display area includes items 611-618, which correspond to associated entities 1-8, respectively. Items 611-618 may be specifically displayed as names and/or related pictures of associated entities.

As an embodiment, the entity information of the associated entity obtained from the service device may include a business domain to which the associated entity belongs, and the business domain here may also be referred to as an entity category, such as characters, movies, music, and so on. The display device may display associated entities of different business fields in different areas according to the business fields to which the associated entities belong. As shown in FIG. 6B , the associated entity display area is further divided into two areas, where associated entities belonging to characters and associated entities belonging to movies are arranged respectively, wherein entities 1-4 belong to characters, and entities 5-8 belong to movies.

In one embodiment, the display device may use at least one associated entity among the associated entities acquired from the service device as a focused associated entity to display. Compared with other associated entities, the display device can display more entity information, occupy more display area for display, and so on, for the highlighted associated entities. As shown in FIG. 6C , entity 1 is an associated entity that is highlighted. As an embodiment, the first associated entity among the associated entities returned by the service device may be used as the highlighted associated entity.

In one example, the displaying device outputting the associated entity may include: outputting entity information of the associated entity through voice. As an embodiment, the display device may generate the text to be outputted by voice according to a preset text template and entity information of the associated entity, and output the text by voice. Taking the text template as "recommend [entity name] for you" and the entity name of the associated entity as "person A", the text to be outputted by voice generated according to the text template and the entity name of the associated entity is "recommended for you" Person A". As another embodiment, the display device may acquire and output the text to be outputted by voice from the service device. As an embodiment, when the display device voice outputs the associated entity, the text corresponding to the voice output may be displayed on the display.

In one example, outputting the associated entity by the display device may include: outputting the associated entity through a voice and displaying the associated entity through a display, which will be described below with reference to FIGS. 6D-6E as examples.

As an embodiment, the associated entity output through the voice may be one of the associated entities displayed on the display. As shown in FIG. 6D , the GUI 600 provided by the display device to the display further includes a voice text display area 62 for displaying the text corresponding to the voice output. In FIG. 6D , the text displayed in the voice text display area 62 is the text corresponding to the voice output entity 1 , and the entity 1 is an associated entity that is highlighted in the associated entity display area 61 .

As an embodiment, the related entities output through voice may be at least two related entities among the related entities displayed on the display. As shown in FIG. 6E , the text displayed in the voice text display area 62 is the text corresponding to the voice output entities 1 - 4 , and the entities 1 - 4 are a plurality of related entities displayed in the related entity display area 61 .

In one example, when the display device acquires the associated entity from the service device, it also acquires the target entity corresponding to the associated entity. As an embodiment, the display device may output the target entity by voice, and display the target entity and the associated entity through the display.

As shown in FIG. 6F , the GUI 600 provided by the display device to the display further includes a target entity display area 63 for displaying the target entity. Specifically, the target entity display area 63 can be used to display entity information of the target entity. In FIG. 6F , the target entity display area 63 displays the entity A as the target entity, the text displayed in the voice text display area 62 is the text corresponding to the voice output entity A, and the association entity display area 61 displays the association corresponding to the entity A entity.

In one example, when outputting the associated entity, the display device may further display the text of the query statement received in step S01 through the display. As shown in FIG. 6G , the GUI 600 provided by the display device to the display further includes a query statement display area 64 for displaying the text of the received query statement. As an embodiment, the text of the query statement may be the text obtained by calling the speech recognition service on the display device to recognize the query statement.

In an example, when the display device acquires the associated entity from the service device, it can also acquire the query result of the query statement. Here, for the acquisition of the query result, reference may be made to the above description based on the architecture shown in FIG. 6 . In one embodiment, the display device may respectively display the query result and the associated entity in different areas on the display, which will not be described in detail here.

It should be noted that, when outputting an associated entity through the above embodiments, each embodiment may be used separately, or each embodiment may be used in combination, which is not specifically limited in the present application. Moreover, in practical application, there are multiple implementation manners for the display device to output the associated entity, and the above embodiments are only examples, and are not intended to be limiting.

FIG. 7 exemplarily shows a flowchart of an association recommendation method applied to a service device. As shown in Figure 7, the process includes the following steps:

Step S11, the service device receives the query statement from the display device, and determines whether the query statement satisfies the preset association recommendation condition; if so, determines the target entity according to the query statement.

As an embodiment, after receiving the query statement from the display device, the service device may perform semantic analysis on the query statement, so as to subsequently determine whether the query statement satisfies the associated recommendation condition. In one embodiment, the service device may perform semantic parsing on the query statement through the deployed semantic service. This application does not limit the specific semantic parsing process.

In practical application, there are many ways to determine the target entity according to the query statement. In one example, for a query statement that satisfies the preset association recommendation conditions, if the query statement contains an entity name, the entity corresponding to the entity name contained in the query statement can be determined as the target entity; if the query statement does not contain an entity name If the entity name is included, the query statement is used as the current query statement, and the target entity is determined according to the previous query statement of the current query statement. The previous query statement and the current query statement here come from the same display device, and the process of determining the target entity according to the previous query statement can refer to the above-mentioned process of determining the target entity according to the current query statement.

As an example, the query statement received within a preset time before the current query statement and received most recently from the current query statement may be determined as the previous query statement, and if there is no previous query statement, the query statement received in advance may be determined as the previous query statement The default entity set is determined as the target entity.

For ease of understanding, the following describes how to determine the target entity based on the query statement, in conjunction with how to determine whether the query statement satisfies the preset association recommendation conditions:

In one example, when the query statement is a single entity name, it is determined that a preset association recommendation condition is satisfied. The entity name here, such as: character name, movie name, music name, game name, food name, etc. As an embodiment, after the query statement is determined to be a single entity name, the entity corresponding to the single entity name may be determined as the target entity.

In another example, when the information to be acquired by the query statement cannot be found, it is determined that the query statement satisfies the preset association recommendation condition. Here, the information to be obtained by the query statement cannot be found. Taking the query statement as "who is the daughter of person A?" as an example, if the service device does not find the information of the daughter of person A, it can be determined that the query cannot be found. The information to be obtained by the statement. As an embodiment, when the information to be acquired by the query statement cannot be found, the service device may determine the entity included in the query statement as the target entity. When the information to be obtained by the query statement cannot be found, returning the relevant information of the query can achieve a better user experience than simply returning the query result.

In another example, when the query statement does not have clear semantics, it is determined that a preset association recommendation condition is satisfied. For example, query statements that do not have explicit semantics here can be "that", "I think", and so on. As an embodiment, when the query statement does not have explicit semantics, the service device may determine the entity corresponding to the entity name contained in the previous query statement of the received query statement as the target entity. Generally speaking, when the requirements are not clear and the requirements cannot be accurately described, users will input query statements without clear semantics. At this time, relevant recommendations are made for the query statements previously output by the user, which can help users find the required information. As well as guide the user to the next query.

In another example, when the number of query statements with the same semantics received from the same display device within a certain period of time reaches a threshold, it is determined that a preset association recommendation condition is satisfied. As an embodiment, when the semantic similarity of the two query statements reaches a threshold, it may be determined that the semantics of the two query statements are the same. As an embodiment, the service device may determine the entity contained in the query statement with the same semantics as the target entity. The number of query statements with the same semantics reaches the threshold, which means that the user has input query statements with the same semantics to the display device multiple times. Generally speaking, this happens because the query that the user has performed does not get the desired query result. In this case, the association recommendation can achieve a better user experience than simply returning the same query results.

It should be noted that, when determining whether the query statement satisfies the preset association recommendation condition through the above embodiments, each embodiment may be used separately, or at least two embodiments may be used in combination. When used in combination, as an embodiment, when the query statement does not satisfy all the association recommendation conditions in the embodiments used in combination, it may be determined that it does not satisfy the association recommendation conditions. The above embodiments are only examples, and other implementation manners that are easy to think of in practical applications will not be described in detail here.

Step S12, according to the preset association recommendation rule, search for the associated entity corresponding to the target entity in the established knowledge graph.

This step S12 is executed when it is determined in the above step S11 that the query statement satisfies the association recommendation condition, and the target entity has been determined according to the query statement.

In an example, the association recommendation rules in this step S12 include: association recommendation rules corresponding to the business domain to which the target entity belongs. As an embodiment, the service device is provided with a corresponding relationship between a business domain and an association recommendation rule. Before searching for the associated entity corresponding to the target entity according to the associated recommendation rule, the service device may determine the business domain to which the target entity belongs, and search for the associated recommendation rule corresponding to the business domain in the correspondence between the business domain and the associated recommendation rule. As an embodiment, the business domain to which the target entity belongs can be determined in various ways, for example, determining the business domain of the target entity according to the attribute value of the business domain attribute of the target entity, determining the business domain of the target entity according to the tag of the target entity, and so on. .

In one embodiment, the association recommendation rule corresponding to the business domain may be determined according to the analysis of query sentences input by a large number of users. For example, a large number of adjacent query statements input by the user can be obtained to analyze the relationship existing among the entities queried by the user, and the like. As an embodiment, the association recommendation rule may be appropriately adjusted according to the query statement input by the user within a certain period of time.

Specifically, how to find the associated entity corresponding to the target entity in the knowledge graph according to the association recommendation rule is described below with an example, and will not be repeated here.

In one example, the business domain to which the target entity belongs may correspond to one association recommendation rule, or may correspond to at least two association recommendation rules.

When the business domain to which the target entity belongs corresponds to an association recommendation rule, as an embodiment, the service device may search the knowledge graph for the associated entity corresponding to the target entity according to the association recommendation rule.

When the business domain to which the target entity belongs corresponds to at least two association recommendation rules, as an embodiment, the service device may search the knowledge graph for the associated entity corresponding to the target entity according to the at least two association recommendation rules. As an embodiment, the associated entities returned by the service device to the display device are in a certain order. For the same associated entities found according to the above at least two association recommendation rules, they can be placed in the respective associated entities during sorting. front end.

When the business domain to which the target entity belongs corresponds to at least two association recommendation rules, as another embodiment, the service device may search for the associated entity corresponding to the target entity according to one of the above at least two association recommendation rules, and, in the found association, When the entity does not meet the preset conditions, the search is further performed according to other association recommendation rules in the above at least two association recommendation rules. The preset conditions here may include: the number of found associated entities is less than a threshold, and the like.

Through the process shown in FIG. 7 , the association recommendation method provided by the embodiment of the present application is described from the perspective of a service device. Through the process shown in FIG. 7 , target entities in different business fields can be recommended according to different association recommendation rules.

In one example, the service device may execute the process shown in FIG. 7 through multiple modules. As an embodiment, the service device may include a receiving module and an association recommendation module. Among them, the receiving module is used to receive the query statement from the display device; the association recommendation module is used to determine whether the query statement satisfies the preset association recommendation condition; The association recommendation rule of , finds the associated entity corresponding to the target entity in the knowledge graph.

In the process shown in Figure 7, how to find the associated entity corresponding to the target entity in the knowledge graph according to the association recommendation rule is described as an example:

In one example, the business domain may include at least one of characters, movies, music, games, and food. The following is an example description of the association recommendation rules corresponding to each business field:

As an embodiment, the association recommendation rule corresponding to a character may include: recommending a character entity that has a specified character relationship with the target entity, where the specified character relationship may include: spouse, child, parent, teammate, mentor, etc.; recommending a character entity with the target entity Characters belonging to the same occupation, the same occupation here may include: doctors, film and television actors, singers, athletes, etc. For some occupations with multiple subclasses, different subclasses can be determined as different occupations respectively. For example, an athlete has multiple sub-categories according to different sports, such as basketball players, football players, table tennis players, etc., and different sub-categories can be regarded as different occupations respectively.

As an embodiment, the association recommendation rules corresponding to movies may include: recommend movies of the same movie type as the target entity; recommend the star of the target entity; recommend the director of the target entity; recommend the same movie as the star of the target entity; recommend the target entity other films directed by the director;

As an embodiment, the associated recommendation rule corresponding to music may include: recommending music belonging to the same music style as the target entity, and recommending music with a high similarity in name to the target entity when recommending. Here, the music style may include: national style , pop, classical, lyrical, etc.; recommends the same music as the original singer of the target entity; recommends the same music as the target entity's lyricist; recommends the same music as the target entity's composer; recommends singing by a non-original singer the target entity; etc.

As an embodiment, the association recommendation rule corresponding to the game may include: recommending games that belong to the same game type as the target entity; recommending games that belong to the same theme as the target entity. The above-mentioned game types may include: first-person shooter, puzzle, strategy, role-playing, etc., and the above-mentioned themes may include: war, strategy, fashion, etc.

As an embodiment, the associated recommendation rule corresponding to the food may include: recommending the food belonging to the same cuisine as the target entity; recommending the food belonging to the same taste as the target entity. The above-mentioned cuisines may include the eight traditional Chinese cuisines, and may also include other cuisines such as Beijing cuisine and Northeastern cuisine. The above-mentioned tastes may include: one or a combination of sour, sweet, bitter, spicy, salty, and fresh.

The target entity in the association recommendation rule in the above embodiments is the entity determined according to the query statement.

In one example, searching for the associated entity corresponding to the target entity in the knowledge graph according to the association recommendation rule may include: according to the business domain to which the target entity belongs, the association relationship between the target entity and other entities and/or the attributes possessed by the target entity, in the Determine the associated entity corresponding to the target entity in the knowledge graph.

As an embodiment, according to the business domain to which the target entity belongs, an entity with a specified association relationship with the target entity can be searched in the knowledge graph, and the found entity can be determined as the associated entity corresponding to the target entity; The business domain to which it belongs, searches the knowledge graph for entities with the same specified attribute value as the target entity, and determines the found entity as the associated entity corresponding to the target entity.

The above-mentioned specified relationship and specified attribute value may be determined according to the association recommendation rule corresponding to the business domain to which the target entity belongs. For example, if the business field to which the target entity belongs is a person, the specified relationship can be determined to be a specified person relationship according to the association recommendation rules corresponding to the person, and the entity whose relationship with the target entity is a specified person relationship can be searched in the knowledge graph. The received entity is determined to be the associated entity of the target entity; it is also possible to determine the “occupational attribute value” of the target entity with the specified attribute value according to the association recommendation rules corresponding to the character, and find the entity with the specified attribute value in the knowledge graph, The found entity is determined to be the associated entity of the target entity.

In one embodiment, the knowledge graph may be a knowledge graph constructed based on tools such as Neo4j, TigerGraph, and OrientDB. The following takes the knowledge graph constructed based on Neo4j as an example to describe how to find the associated entity corresponding to the target entity in the knowledge graph:

In one example, for a knowledge graph constructed based on Neo4j, a specified entity can be searched in the knowledge graph through a rule expression relationship chain. The rule expresses the relationship chain and can search for the specified entity by defining the attributes of the entity, the relationship between the entity and the entity, etc. As an embodiment, the service device may generate a rule expression relationship chain according to the association recommendation rule, so as to find the associated entity corresponding to the target entity in the knowledge graph.

When generating a rule expression relationship chain according to an association recommendation rule, as an embodiment, the service device may determine the corresponding relationship between the association recommendation rule corresponding to the business domain to which the target entity belongs, according to the preset association recommendation rule and the corresponding relationship between the rule expression relationship chain template. A rule expression relationship chain template, which includes a part that needs to input keywords; the service device can obtain the specified attribute value of the target entity as a keyword, input it into the determined rule expression relationship chain template, and generate a rule expression relationship chain.

As an embodiment, by expressing a relationship chain template by a rule, the number and order of the found associated entities may also be limited.

In order to facilitate the understanding of the above content, the following description is combined with specific examples:

For example: the target entity is sweet and sour pork loin, the business field of the target entity is gourmet, and the associated recommendation rule corresponding to the gourmet is: recommend the gourmet that belongs to the same cuisine as the target entity, then the service device searches for the gourmet that is the same as the "recommended and the target entity belongs to the same cuisine" "The corresponding rule expression relationship chain template.

Suppose the found rule expression relationship chain template is as follows:

"match(n)-[r:relate]-(m)-[t:relate]-(s:'food')where id(n)=ENTITY and r.name='cuisine'and m.name=' [cuisine attribute value]'and t.name='cuisine'and n.name='[name attribute value]'and s.describeNum>50 RETURN s order by s.describeNum LIMIT 10".

The rule expression relationship chain is used to find entities whose business domain is cuisine and whose cuisine is [cuisine attribute value]. Among them, [cuisine attribute value], [name attribute value] are the parts that need to input keywords.

First, briefly describe the meaning of some content in the rule expression relationship chain template:

"(n)-[r:relate]-(m)-[t:relate]-(s:'food')" means that entity n has relationship r with entity m, and entity m has relationship t with entity s;

"r.name='Cuisine'" means that the name of the relation r is a cuisine, "t.name='Cuisine'" means that the name of the relation t is a cuisine, and "n.name='[name attribute value]'" means that the entity n The name of m is [name attribute value], "m.name='[cuisine attribute value]'" means the name of m is [cuisine attribute value];

"RETURN s" indicates that the output entity is the entity represented by s in the rule expression relationship chain template; "order by s.describeNum" indicates that the output entities are sorted according to the describeNum attribute; "LIMIT 10" indicates that the number of output entities is limited to 10.

According to the above rule expression relationship chain template, the service device can obtain the attribute value of the "cuisine" attribute and the attribute value of the "name" attribute of the target entity sweet and sour pork loin as a keyword, and input the keyword into the rule expression relationship chain template to generate Rules express chains of relationships.

Taking the attribute value of the "cuisine" attribute of sweet and sour pork loin as "Zhejiang cuisine" as an example, the generated rule expression relationship chain is as follows:

"match(n)-[r:relate]-(m)-[t:relate]-(s:'food')where id(n)=ENTITY and r.name='cuisine'and m.name=' Zhejiang cuisine'and t.name='cuisine'and n.name='sweet and sour pork loin'and s.describeNum>50 RETURN s order by s.describeNum LIMIT 10".

It should be noted that the above embodiment is only an example, and other association recommendation rules may exist in practical application, as well as other ways of finding the associated entity corresponding to the target entity in the knowledge graph, which will not be described one by one here.

As described in the above embodiment, the display device can send the received query statement to the service device, so that the service device can determine the target entity that needs to be recommended for association according to the query statement, and search the knowledge graph for the corresponding target entity. The associated entity is obtained by the display device and output, so that during the interaction between the display device and the user, a relatively rich and comprehensive association recommendation can be made for the query statement input by the user, thereby improving the user experience.

Taking a character as an example, FIG. 8 exemplarily shows a schematic diagram of a knowledge graph according to some embodiments. As shown in Figure 8, in the knowledge graph, a node is called an entity, and the entity can be replaced by things that exist objectively and can be distinguished from each other, such as specific people, things, things, institutions, abstract concepts, etc.

For entities in the knowledge graph, the corresponding relationship between attributes and attribute values can be configured according to actual requirements during specific implementation. Taking the entity "Liu Moumou" in Fig. 8 as an example, the pre-configured attributes of this entity may be: occupation, height, gender, etc. Correspondingly, the attribute values may be: specific occupation type (taking the driver as an example), specific height value (take 180cm as an example), specific gender (take male as an example).

In addition, different entities in the knowledge graph can be connected by directional lines. This wire can be used to represent semantic relationships between different entities. Taking the entity "Liu Moumou" and the entity "Zhao Moumou" in Figure 8 as an example, the connection between the entity "Liu Moumou" and the entity "Zhao Moumou" is that the entity "Liu Moumou" points to the entity "Zhao Moumou". XX", if the semantic relationship corresponding to the link is a wife relationship, then the link can indicate that the wife of the entity "Ryu Moumou" is the entity "Zhao Moumou".

In some embodiments, the service device includes: an input interface and a processor. The processor is configured to: when a query statement is received through the input interface, input the query statement into the configured question answering system to obtain a reply statement corresponding to the query statement; find the corresponding reply statement in the established knowledge graph The associated answer entity; according to the corresponding relationship between the configured attributes and attribute values of the answer entity, a recommended sentence is generated; the recommended sentence includes a preset number of N attributes and corresponding attribute values, where N is greater than or equal to 1.

Correspondingly, this embodiment also provides some methods for implementing corresponding operations of the foregoing processor. Please refer to FIG. 9 , which is a schematic flowchart of an information recommendation method provided by an embodiment of the present application.

As shown in Figure 9, the process may include the following steps:

Step 31: When the query sentence is received, the query sentence is input into the configured question answering system to obtain a reply sentence corresponding to the query sentence.

In one example, the query statement may be input to the service device by the smart device. That is, the service device and the smart device are two independent devices. In this case, the service device can be a server. In another example, the query statement may be directly input to the service device by the user, or input to the service device by the user operating the control device of the service device. That is, the service device and the aforementioned smart device can be understood as one device.

It should be noted that the query statement received by the service device may be a query statement in a voice format or a query statement in a text format.

It should also be noted that the above-mentioned smart devices may be smart speakers, smart TVs, mobile phones, computers, or the aforementioned display devices and other devices, and the application does not limit the specific types of smart devices.

As an example, the above query statement may be input to the service device through an input interface on the service device.

As an example, since the specific content of the query statement is determined by the user's will, the specific content of the query statement is varied. However, in practical applications, the service device cannot generate a corresponding recommendation statement for any query statement. Therefore, it is necessary to set recommendation conditions in advance to determine whether a recommendation statement can be generated for the received query statement, and when the received query statement is received If the query statement satisfies the above recommendation conditions, a recommendation statement is generated for the query statement.

Optionally, the preset recommendation condition may include: the query statement includes at least a main entity and a semantic relationship between the main entity and the answer entity to be queried, and the semantic relationship is the same as the preset semantic relationship.

The following is an example of the main entity in the query statement and the semantic relationship between the main entity and the answer entity to be queried:

For example, when a query statement includes a person and a person relationship, the person in the query statement can be used as the main entity, and the person relationship can be used as a semantic relationship. For example, if the query sentence is "Who is Yao's wife?", the main entity in the query sentence is "Yao", and the semantic relationship between the main entity "Yao" and the answer entity to be queried is is "wife".

Of course, the main entity and semantic relationship in the query statement can be, in addition to the above examples of characters and character relationships, regions and positions (such as governors, mayors, etc.), film and television works and work information (such as actors, directors, screenwriters, etc.) , producer, host, etc.), songs and song information (e.g. singer, lyricist, MV director, producer, arranger, etc.), countries and rulers/national leaders (e.g. president, prime minister, prime minister, deputy President, Emperor, etc.), books and book information (author, translator, editor-in-chief, etc.), school and school information (principal, famous alumni, etc.), etc. This application does not limit the main entity and semantic relationship specifically included in the query statement .

It should be noted that, the semantic relationship in the above-mentioned preset recommendation conditions may be set according to the actual situation, and the present application does not specifically limit the semantic relationship in the recommendation conditions.

When the service device receives the query statement and determines that the query statement satisfies the above-mentioned preset recommendation conditions, the service device will input the query statement into the configured question answering system, so that the question answering system output corresponds to the query The statement's reply statement. In one example, the reply statement may include the answer entity queried by the query statement. In another example, the reply statement may carry the answer entity ID of the answer entity queried by the query statement (the entity ID will be described below). Of course, the reply statement can also include other information according to the actual situation, such as the main entity in the query statement, and the semantic relationship between the main entity in the query statement and the answer entity to be queried. The content is not specifically limited.

It should be noted that the above question and answer system can be deployed in a service device, so that the service device can locally input query sentences into the question and answer system, and locally obtain the reply sentences output by the question and answer system. Of course, the above question answering system can also be deployed in other devices (denoted as the target device) other than the service device. In this way, the service device needs to input the query sentence into the question answering system through the communication connection with the target device, and obtain the question answer Reply statement output by the system.

It should also be noted that the above question answering system can be implemented in various implementation manners. For example, the question answering system is implemented through a search engine, or the question answering system is implemented through a knowledge graph, etc. This application does not limit the specific implementation of the question answering system.

Step 32: Find the answer entity associated with the reply sentence in the established knowledge graph.

Before performing this step 32, a knowledge graph needs to be pre-established in the service device, and the knowledge graph can be established by referring to the establishment methods well known to those skilled in the art, and details are not repeated here. Since there may be different entities with the same entity name in the knowledge graph, in order to realize the distinction between the aforementioned different entities with the same entity name, each entity in the knowledge graph can be assigned a unique entity ID.

In the embodiments of the present application, there are multiple implementations for finding the answer entity associated with the reply sentence in the established knowledge graph. As one of the implementation manners, as mentioned above, in the case where the reply sentence carries the answer entity ID of the answer entity, an entity matching the answer entity ID can be searched in the established knowledge graph, and then the found This entity is identified as the answer entity. In this implementation manner, since the answer entity ID can uniquely identify the answer entity, even if there are multiple other entities with the same entity name as the answer entity, the answer entity can be accurately searched.

Step 33 , generating a recommended sentence according to the corresponding relationship between the configured attributes and attribute values of the answer entity; the recommended sentence includes a preset number of N attributes and corresponding attribute values, where N is greater than or equal to 1.

Before performing this step 33, the corresponding relationship between attributes and attribute values is configured for the answer entity in the knowledge graph in advance, and the corresponding relationship between the attributes and attribute values can be used to describe the relevant information of the answer entity. The following table exemplarily lists some attributes used to configure entities, see Table 1:

Table 1

第一级属性类别first-level attribute category	第二级属性类别second-level attribute category	第三级属性类别third-level attribute category	属性Attributes
人物figure	通用Universal	人物关系Character Relationships	朋友friend
人物figure	通用Universal	人物关系Character Relationships	前任predecessor
人物figure	通用Universal	人物关系Character Relationships	儿子son
人物figure	通用Universal	人物关系Character Relationships	母亲Mother
人物figure	通用Universal	人物关系Character Relationships	好友friend
人物figure	通用Universal	人物关系Character Relationships	女儿daughter
人物figure	通用Universal	人物关系Character Relationships	妻子wife
人物figure	通用Universal	人物关系Character Relationships	丈夫husband
人物figure	通用Universal	人物关系Character Relationships	父亲Father
人物figure	通用Universal	人物关系Character Relationships	初恋first love
人物figure	通用Universal	人物属性character attributes	星座Constellation
人物figure	通用Universal	人物属性character attributes	籍贯Hometown
人物figure	通用Universal	人物属性character attributes	身高height
人物figure	通用Universal	人物属性character attributes	出生地place of birth
人物figure	通用Universal	人物属性Character attributes	职业Profession

人物figure	通用Universal	人物属性character attributes	出生日期date of birth
人物figure	通用Universal	人物属性character attributes	代表作品representative work
人物figure	通用Universal	人物属性character attributes	主要成就Major achievements
人物figure	娱乐人物entertainment figures		艺名stage name
人物figure	娱乐人物entertainment figures	演员actor	饰演play
人物figure	娱乐人物entertainment figures		粉丝名fan name
人物figure	娱乐人物entertainment figures		昵称Nick name
人物figure	娱乐人物entertainment figures		经纪公司agency
人物figure	娱乐人物entertainment figures		三围measurements
人物figure	娱乐人物entertainment figures		出道时间Debut time
人物figure	体育人物sports figures		国家nation
人物figure	体育人物sports figures		运动项目sports
人物figure	体育人物sports figures	运动员athlete	教练coach
人物figure	体育人物sports figures	运动员athlete	所属运动队sports team
人物figure	体育人物sports figures	运动员athlete	场上位置Field position
人物figure	体育人物sports figures	运动员athlete	场上号码Field number
人物figure	体育人物sports figures	运动员athlete	惯用脚dominant foot

Here, the third row in Table 1 above is used as an example for illustration, where "predecessor" is an attribute, and the attribute "predecessor" belongs to the third-level attribute category "person relationship", and the third-level attribute category "Character relationship" is under the second-level attribute category "General", and the second-level attribute category "General" is under the second-level attribute category "Person". As for the specific content of the attribute value corresponding to the attribute "predecessor", it can be set according to the actual situation, which is not specifically limited in this application.

It should be noted that the first-level, second-level and third-level attribute categories in the above Table 1 are used to classify different attributes, so as to facilitate the management and planning of different attributes, and are not necessary in the process of configuring attributes for entities. .

It should also be noted that the attributes listed in Table 1 above are only examples, and other attributes may also be configured for entities according to actual situations in practical applications, which are not specifically limited in this application.

In the embodiment of the present application, the N attributes and corresponding attribute values included in the recommendation sentence are attributes and attribute values included in the corresponding relationship between attributes and attribute values preconfigured by the answer entity. As for how to generate a recommendation sentence including N attributes and corresponding attribute values, the following will be described in detail with reference to the process shown in FIG. 11 , and details will not be described here for the time being.

It can be seen from the above step 33 that since the attributes and the corresponding attribute values record the relevant information of the answer entity, the generated recommendation sentence also includes a preset number of N attributes and corresponding attribute values, so through the recommendation The statement can learn the relevant information of the answer entity corresponding to the query statement, which can expand the user's vision, help the user to understand more knowledge related to the answer entity, let the user have a specific understanding of the answer entity, and stimulate the user's Cognitive interest, triggering further searches for answer entities.

In the embodiment of the present application, there are multiple implementations for the service device to output the recommended sentence.

As one of the implementation manners, as mentioned above, when the service device and the smart device are not the same device, the service device can send the recommended sentence and the reply sentence to the smart device, and the smart device sends the received recommended sentence and reply sentence For output, for example, output is performed by means of voice playback, or, if a display is provided, output is performed by means of display display.

As another implementation manner, as mentioned above, when the service device and the smart device are the same device, the service device can output through voice playback, or, in the case of a display, output through display display .

In the following, in conjunction with the GUI on the display device shown in FIG. 10 , the service device sends some related information to the display device, and the display device presents the related interface. How to display the above recommended sentence on the display will be described in detail, which will not be repeated here.

It can be seen from the above technical solutions that in this application, since the N attributes included in the recommendation sentence and the corresponding attribute values are used to describe the relevant information of the answer entity, the user can learn the relevant information of the answer entity through the recommendation sentence. information, which can expand the user's vision, help the user to understand more knowledge related to the answer entity, let the user have a specific understanding of the answer entity, and at the same time stimulate the user's cognitive interest and trigger further search for the answer entity.

Hereinafter, in conjunction with the GUI shown in FIG. 10 , how the display device displays the recommended sentence is exemplarily described:

FIG. 10 exemplarily shows a schematic diagram of the GUI provided by the display device when displaying the recommended sentence. As shown in FIG. 10 , the display provides a GUI 800 for presenting a query sentence and a reply sentence and a recommended sentence returned by the service device. The GUI 800 includes a first display area 81 for displaying the query sentence, and the first display area 81 includes an item 81a. Item 81a is used to display a query statement. The GUI 800 also includes a second display area 82 for displaying reply sentences and a third display area 83 for displaying recommended sentences, wherein the second display area 82 includes an item 82a, and the item 82a is used for displaying reply sentences; The display area 83 includes an item 83a, and the item 83a is used to display a recommended sentence.

It should be noted that the above-mentioned FIG. 10 is only a schematic diagram illustrating a GUI for displaying recommended sentences. In practical applications, other contents other than the above query sentence, reply sentence and recommendation sentence may also be displayed according to the actual situation, which is not specifically limited in this application.

The following is a detailed description of how to generate a recommendation sentence according to the corresponding relationship between the configured attributes and attribute values of the answer entity in the above step 33:

Referring to FIG. 11 , it is a flowchart of the implementation of step 103 provided in this embodiment of the present application.

As shown in Figure 11, the process may include the following steps:

Step 331: Select N attributes and corresponding attribute values in the corresponding relationship between attributes and attribute values that have been configured in the answer entity.

As an example, in this step 331, among all the attributes-attribute value correspondences that have been configured in the answer entity in the knowledge graph, there are multiple implementations for selecting N attributes and corresponding attribute values.

As one of the implementation manners, a selection ratio may be preset, such as 10%, 20%, etc. This application does not specifically limit the preset selection ratio. Based on this, when selecting N attributes and corresponding attribute values, the corresponding relationship between the configured attributes and attribute values of the answer entity can be selected according to the preset selection ratio, so as to select the attributes with the preset selection ratio. , and the corresponding attribute value.

As another implementation manner, a selection quantity may be preset, such as 2, 4, etc., and this application does not specifically limit the preset selection quantity. Based on this, when selecting N attributes and corresponding attribute values, the corresponding relationship between the configured attributes and attribute values of the answer entity can be selected according to the preset selection number, so as to select the preset selection number of attributes. , and the corresponding attribute value.

Step 332 , determine whether the selected attribute values have the same attribute value as the main entity in the query statement or the reply statement; if so, go back to step 331 ; if not, go to step 333 .

In the embodiment of the present application, the reason why each selected attribute value is judged to determine whether the attribute value is consistent with the main entity in the query sentence or the reply sentence is because: the attribute configured for the answer entity in advance Among the values, there may be an attribute with the same value as the main entity in the query or reply statement. For example, the query sentence is "who is the singer of song A?", the reply sentence corresponding to the query sentence is "the singer of song A is Zhao", and the attribute in the knowledge graph corresponding to Zhao's answer entity is configured The attribute value corresponding to "Masterpiece" is "Song A". In this way, if the attribute value of "song A" and its corresponding attribute "representative work" are selected to generate a recommended sentence, the recommended sentence can include "Zhao's representative work is song A", and the relationship between the recommended sentence and the reply sentence The meanings are similar, so the recommendation effect that the recommendation sentence should have cannot be achieved. Therefore, by judging each of the selected attribute values, it is possible to avoid recommending useless information for the user by the subsequently generated recommendation sentences.

Step 333: Input each attribute and the corresponding attribute value into the set recommended sentence generation model to obtain a recommended sentence.

This step 333 is executed on the premise that each of the selected attribute values is different from the main entity in the reply sentence or the query sentence.

In the case where each selected attribute value is different from the main entity in the reply sentence or query sentence, it means that each attribute value in the subsequently generated recommendation sentence is different from the main entity, and neither the generation nor the reply will be generated. Statements with similar meanings are recommended statements. Based on this, each selected attribute and the corresponding attribute value can be input into the preset recommended sentence generation model to obtain the recommended sentence. The following will describe in detail how the recommended sentence generation model in this step 333 obtains the recommended sentence with reference to the process shown in FIG. 12 , which will not be repeated here. The following describes in detail how the recommended sentence generation model in the above step 333 obtains the recommended sentence:

Referring to FIG. 12 , FIG. 12 is a flowchart of implementing step 333 provided in this embodiment of the present application.

As shown in Figure 12, the process may include:

Step 3331, for each attribute of the input and the corresponding attribute value, find the corresponding target sentence template in the set sentence template, fill the attribute and the attribute value correspondingly to the filling position of the target sentence template, according to the target sentence The populated attributes and attribute values in the template generate corresponding recommended sub-statements.

As an example, in this step 3331, at least one statement template may be preset in the statement generation model, and these statement templates may be statement templates of the same category, or may be statement templates of different categories respectively.

When the statement generation model includes only one type of statement template (referred to as the first type of statement template), the first type of statement template includes at least: a filling position (referred to as the first filling position) for filling attributes, for The filling position of the filling attribute value (referred to as the second filling position), and the connective connecting the first filling position and the second filling position. For example, the statement template can be "[attribute] is [attribute value]", where "[attribute]" is the first fill position for filling the attribute and "[attribute value]" is the first fill position for filling the attribute value The second stuffing position, "is" is a connective that connects the first stuffing position and the second stuffing position.

Optionally, when the statement generation model includes only one type of statement template, when the statement generation model generates a recommended sub-statement for each input attribute and the corresponding attribute value, it may include:

For each attribute and the corresponding attribute value, a target sentence template is selected from the set first type of sentence templates. Then, the attribute is correspondingly filled into the first filling position in the target sentence template, and the attribute value is correspondingly filled into the second filling position in the target sentence template. Finally, set corresponding punctuation marks for the above target sentence template filled with attributes and attribute values to generate corresponding recommended sub-sentences. For example, if the attribute is height, the corresponding attribute value is 180cm. Here, taking the target sentence template as "[attribute] is [attribute value]" as an example, after filling the aforementioned assumed height and 180cm corresponding to the height into the target sentence template, a filling such as "height is 180cm" can be obtained. A target statement template with attributes and corresponding attribute values. After that, add corresponding punctuation marks such as ";" to the target sentence template filled with attributes and corresponding attribute values, and then a recommended sub-sentence such as "height is 180 cm;" can be obtained.

When the statement generation model includes at least two types of statement templates, the first type of statement template and the second type of statement template, the first type of statement template is the same as the first type of statement template in the foregoing implementation manner, which is not repeated here. Repeat. The second type of statement template is different from the first type of statement template. The second type of statement template includes at least: a filling position for filling the answer entity (referred to as the fifth filling position), a filling position for filling the attribute (referred to as the first filling position) Three padding bits), the padding position (referred to as the fourth padding position) for padding the attribute value, the first connective connecting the third padding position and the fourth padding position, and the fifth padding position and the third padding position. second conjunction. For example, the second type of statement template can be "[attribute] of [answer entity] is [attribute value]", where "[answer entity]" is the fifth padding position used to populate the answer entity name, "[ attribute]" is the third fill position for filling the attribute, "[attribute value]" is the fourth fill position for filling the attribute value, "yes" is the first connection connecting the third fill position and the fourth fill position word, "of" is the second connective that connects the fifth stuffing position and the third stuffing position.

Optionally, when the statement generation model includes two types of statement templates, when the statement generation model generates a recommended sub-statement for each attribute of the input and the corresponding attribute value, it may include: each attribute of the input and the corresponding An attribute and a corresponding attribute value are selected from the attribute values as the first attribute and attribute value. Select a target sentence template (marked as the first target sentence template) from the second type of sentence templates, fill the first attribute and attribute value into the above-mentioned first target sentence template and add punctuation to obtain the corresponding first target sentence template. Recommended sub-statements for attributes and attribute values. For each of the remaining attributes and corresponding attribute values, select a target sentence template (referred to as the second target sentence template) in the first type of sentence template, and fill the attribute and corresponding attribute value into the above-mentioned second sentence template. The target sentence template is added and punctuation marks are added to obtain the recommended sub-sentence corresponding to the attribute and the corresponding attribute value.

It should be noted that the above statement template may also include other information according to the actual situation, and the connectives in the above statement template and the punctuation marks set for the statement template can also be selected according to the actual situation, which are not limited in this application.

Step 3332, splicing each attribute and the recommended sub-sentences corresponding to the corresponding attribute values to obtain a recommended sentence.

In this step 3332, there are multiple implementations for splicing each recommended sub-sentence into a recommended sentence.

As one of the implementations, if the recommended sub-sentences corresponding to the input attributes and corresponding attribute values are all generated according to the aforementioned first-type sentence template, then the recommended sub-sentences can be directly spliced . For example, splicing can be performed in a random manner, in the order in which each attribute and corresponding attribute value are input into the recommended sentence generation model, and the like. The present application does not specifically limit the splicing method of each recommended sub-sentence.

As another implementation manner, if there is a recommended sub-sentence (referred to as the first recommended sub-sentence) generated according to the foregoing second type of sentence template in the recommended sub-sentences corresponding to the input attributes and corresponding attribute values, it is necessary to The first recommended sub-sentence is used as the first recommended sub-sentence in the recommended sentence. Afterwards, the other recommended sub-sentences except the first recommended sub-sentence are spliced after the first recommended sub-sentence to generate a recommended sentence. In the process of splicing the remaining recommended sub-sentences after the first recommended sub-sentence, the splicing can be performed in a random manner, in the order in which each attribute and the corresponding attribute value are input into the recommended sentence generation model, etc. It does not specifically limit the splicing method of the remaining recommended sub-sentences.

The above describes in detail how the recommended sentence generation model in the above step 333 obtains the recommended sentence. It should be noted that the above is only an exemplary description, and is not intended to limit the present application.

In a specific implementation, the present invention also provides some non-volatile computer storage media, wherein the computer storage medium can store a program, and when the program is executed, it can include various embodiments of the screen saver display method and the screen saver jump method provided by the present invention some or all of the steps in . The storage medium may be a magnetic disk, an optical disk, a read-only memory (English: read-only memory, abbreviated as: ROM) or a random access memory (English: random access memory, abbreviated as: RAM) and the like.

Those skilled in the art can clearly understand that the technology in the embodiments of the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such understanding, the technical solutions in the embodiments of the present invention may be embodied in the form of software products in essence or the parts that make contributions to the prior art, and the computer software products may be stored in a storage medium, such as ROM/RAM , magnetic disk, optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of the present invention.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: The technical solutions described in the foregoing embodiments can still be modified, or some or all of the technical features thereof can be equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the technical solutions of the embodiments of the present application. Scope.

For the convenience of explanation, the above description has been made in conjunction with specific embodiments. However, the above exemplary discussions are not intended to be exhaustive or to limit implementations to the specific forms disclosed above. Numerous modifications and variations are possible in light of the above teachings. The above embodiments are chosen and described to better explain the principles and practical applications, so as to enable those skilled in the art to better utilize the described embodiments and various modified embodiments suitable for specific use considerations.

Claims

An intelligent device, comprising: an input interface and a controller;

The controller is configured to:

Send the query statement input through the input interface to the service device, so that the service device searches for the associated entity corresponding to the target entity in the established knowledge graph, where the knowledge graph is used to represent the semantic relationship between entities, and the The target entity is the entity determined according to the query statement;

Acquire the associated entity from the service device and output.
The smart device according to claim 1, which is a display device;

The controller is configured to output the associated entity by voice, and/or output the associated entity to a display to display the associated entity through the display.
A service device comprising:

The receiving module is used to receive the current query statement from the smart device;

The association recommendation module is used to find the associated entity corresponding to the target entity in the established knowledge graph, the knowledge graph is used to represent the semantic relationship between the entities, and the target entity is the entity determined according to the current query sentence.
The service device according to claim 3, wherein the association recommendation module is specifically used for:

According to the business domain to which the target entity belongs, the knowledge graph is searched for an entity having a specified relationship with the target entity, and the found entity is determined as the associated entity corresponding to the target entity.
The service device according to claim 3, wherein the association recommendation module is specifically used for:

According to the business domain to which the target entity belongs, an entity having the same specified attribute value as the target entity is searched in the knowledge graph, and the found entity is determined as an associated entity corresponding to the target entity.
The service device according to any one of claims 4 or 5, wherein the business field includes: at least one of characters, movies, music, games and food.
The service device according to claim 3, wherein the association recommendation module is specifically used for:

When the current query statement satisfies the preset association recommendation condition, if the current query statement contains an entity name, the entity corresponding to the entity name is determined as the target entity;

If the current query statement does not contain an entity name, the entity corresponding to the entity name in the received previous query statement is determined as the target entity.
An association recommendation method applied to smart devices, including:

Send the query statement input through the input interface to the service device, so that the service device searches for the associated entity corresponding to the target entity in the established knowledge graph, the knowledge graph is used to represent the semantic relationship between entities, and the target entity is the entity determined according to the query statement;

Acquire the associated entity from the service device and output.
An association recommendation method, applied to service equipment, includes:

Receive the current query statement from the smart device;

The associated entity corresponding to the target entity is searched in the established knowledge graph, where the knowledge graph is used to represent the semantic relationship between entities, and the target entity is an entity determined according to the current query sentence.
The method according to claim 9, wherein searching the associated entity corresponding to the target entity in the established knowledge graph, comprising:

According to the business domain to which the target entity belongs, search for an entity with a specified association relationship with the target entity in the knowledge graph, and determine the found entity as an associated entity corresponding to the target entity; and/or,

According to the business domain to which the target entity belongs, determine the specified attribute value of the target entity, search for the entity with the specified attribute value in the knowledge graph, and determine the found entity as the corresponding entity of the target entity. Associated entity.
A service device, comprising: an input interface and a processor;

The processor is configured to:

When a query statement is received through the input interface, the query statement is input into the configured question answering system to obtain a reply statement corresponding to the query statement;

Find the answer entity associated with the reply sentence in the established knowledge graph;

A recommended sentence is generated according to the corresponding relationship between attributes and attribute values configured by the answer entity; the recommended sentence includes a preset number of N attributes and corresponding attribute values, where N is greater than or equal to 1.
The service device according to claim 11, wherein the corresponding relationship between attributes and attribute values configured by the answer entity comprises: the N attributes and the corresponding relationship between corresponding attribute values.
The service device of claim 11,

The processor is configured to output the recommended sentence by voice; or,

The processor is configured to display the recommended sentence through a display.
The service device according to claim 11, wherein the reply sentence carries an answer entity ID;

The answer entity associated with the reply sentence includes: an answer entity corresponding to the answer entity ID in the established knowledge graph.
The service device according to claim 11, wherein the generating a recommended sentence comprises:

The N attributes and corresponding attribute values are input into the set recommended sentence generation model to obtain the recommended sentence.
The service device according to claim 15, wherein the recommended sentence generation model obtains the recommended sentence in the following manner:

For each attribute of the input and the corresponding attribute value, find the corresponding target sentence template in the set sentence template, fill the attribute and attribute value into the filling position of the target sentence template, according to the target sentence template The filled attributes and attribute values in the statement template generate corresponding recommended sub-statements;

The recommended sentences are obtained by splicing each attribute and the recommended sub-sentences corresponding to the corresponding attribute values.
The service device according to claim 11, 15 or 16, wherein the reply statement at least comprises: a main entity, a semantic relationship between the main entity and the answer entity, and the answer entity;

The attribute values corresponding to the N attributes in the recommendation sentence are different from the main entity.
An information recommendation method including:

When receiving a query statement, inputting the query statement into the configured question answering system to obtain a reply statement corresponding to the query statement;

Find the answer entity associated with the reply sentence in the established knowledge graph;

A recommended sentence is generated according to the corresponding relationship between attributes and attribute values configured by the answer entity; the recommended sentence includes a preset number of N attributes and corresponding attribute values, where N is greater than or equal to 1.
The method according to claim 18, wherein the corresponding relationship between attributes and attribute values configured by the answer entity comprises: the N attributes and the corresponding relationship between corresponding attribute values.
The method of claim 18,

Output the recommended sentence by voice; or,

The recommended sentence is displayed on the display.
The method of claim 18, wherein the reply statement carries an answer entity ID;

The answer entity associated with the reply sentence includes: an answer entity corresponding to the answer entity ID in the established knowledge graph.
The method according to claim 18, wherein the generating a recommendation statement comprises:

The N attributes and corresponding attribute values are input into the set recommended sentence generation model to obtain the recommended sentence.
The method according to claim 22, wherein the recommended sentence generation model obtains the recommended sentence in the following manner:

For each attribute of the input and the corresponding attribute value, find the corresponding target sentence template in the set sentence template, fill the attribute and attribute value into the filling position of the target sentence template, according to the target sentence template The filled attributes and attribute values in the statement template generate corresponding recommended sub-statements;

The recommended sentences are obtained by splicing each attribute and the recommended sub-sentences corresponding to the corresponding attribute values.
The method according to claim 18, 22 or 23, wherein the query statement at least comprises: a main entity, a semantic relationship between the main entity and the answer entity;

The attribute values corresponding to the N attributes in the recommendation sentence are different from the main entity.