CN114925189A

CN114925189A - Content recommendation method and device based on vehicle-mounted conversation and storage medium

Info

Publication number: CN114925189A
Application number: CN202210593274.0A
Authority: CN
Inventors: 易紫阳
Original assignee: FAW Group Corp
Current assignee: FAW Group Corp
Priority date: 2022-05-27
Filing date: 2022-05-27
Publication date: 2022-08-19

Abstract

The invention provides a content recommendation method based on vehicle-mounted conversation, which comprises the following steps: and S1, acquiring the voice information sent by the user. S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, go to the dialogue strategy module. S3, the conversation strategy module decides whether to enter the next round of interaction, and when the conversation strategy module judges that a next-round interaction is not needed, the conversation strategy module enters a recommendation module to perform recommendation operation; otherwise go to S1 for the next round of interaction. The invention makes user's preference clear step by step through dialogue interaction without affecting user experience.

Description

Content recommendation method and device based on vehicle-mounted conversation and storage medium

Technical Field

The invention relates to a content recommendation method, a content recommendation device and a storage medium, in particular to a content recommendation method, a content recommendation device and a storage medium based on vehicle-mounted conversation.

Background

Speech recognition is a high technology by which machines translate speech signals into corresponding text files or commands by recognition and understanding. As a special research field, speech recognition is a cross discipline, which is closely linked to many disciplines such as acoustics, phonetics, linguistics, digital signal processing theory, information theory, computer disciplines, and the like.

Speech recognition is an important part of man-machine interface and is also a very important application technology in the processing of world speech signals. The purpose of speech recognition is that machines "understand" human speech, which is also an important aspect of machine intelligence.

The difficulty of recognizing speech by a machine is somewhat similar to how a person with a bad foreign language listens to a foreign person, which is associated with different speakers, different speaking speeds, different speaking contents, and different environmental conditions. The characteristics of the speech signal itself cause difficulties in speech recognition. These characteristics include variability, dynamics, transients, continuity, etc.

At present, the most widely used voice control technology is the application in household appliances. For example, sound-controlled car audio of Sony corporation, a KD-LXSO type sound-controlled box (for car) of JVC corporation, America InVoca omnibearing sound remote controller, Italy Delonghi microwave oven, etc. can all execute various functions by voice commands. The panasonic electric appliance industry started to market 36-inch televisions with voice recognition function in 2001 and 12 months, "Bs digital high-definition televisions with built-in AV hard disks". The product is equipped with a microphone remote control for inputting instructions in language. The user can easily select a specific television station of BS digital broadcasting or terrestrial wave analog broadcasting by reporting the name or channel number of the television station by voice.

When a conventional remote controller is used, it is necessary to perform operations step by step on a menu screen when searching for program information, making a reservation for recording, and the like, and it is possible to omit operations of the middle layer when performing voice operations. The voice control is not only applied to household appliances, but also has more and more extensive application in the aspects of communication, automatic control, home automation and the like.

In the vehicle-mounted scene, the voice assistant is often passive to play various roles, such as: navigation, music playing, video playing, etc.

For music playback, the user's preferences are not constant, such as a user who previously liked to listen to bach, and recently tried to listen to zhou jeron. Based on the contradiction between the user portrait and the recent requirement of the user, multiple rounds of conversations can be generated, and the user often generates aversive psychology due to too many interactions, so that the user experience is influenced.

The problem of current on-vehicle recommendation:

1. the active conversation capability of the voice assistant is not fully utilized in the current vehicle-mounted scene to explore the preference of the user. The push can be performed only according to the user portrait or the keyword is extracted according to the voice input of the user.

2. When a certain scene is faced with no user data, only the preferences of other users can be referred to, and the actual requirements of the users cannot be represented.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a content recommendation method based on vehicle-mounted conversation, which comprises the following steps:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, go to the dialogue strategy module.

S3, the conversation strategy module decides whether to enter the next round of interaction, and when the conversation strategy module judges that a next-round interaction is not needed, the conversation strategy module enters a recommendation module to perform recommendation operation; otherwise go to S1 for the next round of interaction.

By the method, when the emotional tendency and the user preference are positive or other, the conversation strategy module judges whether the next round of interaction is needed, if so, the user is guided to continuously input the voice information, and if not, the user is switched to the recommending module to recommend commodities or services, such as songs, to the user.

The conversation content is based on a specific scene, and the preference of the user in the specific scene is determined by analyzing the conversation content, constructing a knowledge graph of the corresponding scene, carrying out conversation guidance based on prior knowledge and the like, so that the conversation recommendation process is more consistent with the service scene.

The conversation strategy is characterized in that in the conversation strategy, the exploration process of user preference is focused more by improving a model structure in conversation management, introducing more prior knowledge bases, perfecting a structure of conversation management and the like, so that the user preference is obtained in as few conversation turns as possible.

The user's preferences are gradually clarified through dialog interactions without affecting the user experience.

And when the emotional tendency or the preference of the user is negative, directly switching to a recommending module to recommend the goods or the services to the user. When the user expresses negative emotional tendency, the user preference can be timely stopped being explored, and the goods and the services can be timely recommended.

Furthermore, the recommending module takes the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix, and constructs a two-stage recommending system based on recall and sequencing.

Furthermore, the recommendation module further constructs a feature matrix by taking the current user input and the short-term user log as short-term preferences according to the preferences in the features of the user and taking the preferences in the user portrait as long-term preferences.

Furthermore, the recommendation module takes the current user input and the short-term user log as short-term preferences according to the preferences in the characteristics of the user, takes the preferences in the user portrait as long-term preferences, and performs weighted calculation on the short-term preferences and the long-term preferences to obtain the user preferences.

Further, the short-term user log comprises types of songs listened to by the user within the short-term time, song listening time and song types not listened to by the user.

Further, the short-term duration does not exceed 7 days.

Further, the recommendation module uses an attention weighting calculation.

Further, the dialogue strategy module judges the next action based on the DQN network of each step time state.

There is also provided a vehicle dialog based content recommendation device comprising a processor and a memory having executable code stored thereon, which when executed by the processor causes the processor to perform the above method.

There is also provided a non-transitory machine-readable storage medium having stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform the above-described method.

The invention makes the user's preference clear step by step through dialogue interaction without affecting the user experience. And when the emotional tendency or the preference of the user is negative, directly switching to a recommending module to recommend the goods or the services to the user. When the user expresses negative emotional tendency, the user preference can be timely stopped being explored, and the goods and the services can be timely recommended.

When facing a new scene, the invention can integrate the long-term preference and the short-term preference of the user and determine the real preference of the user in time. Meanwhile, the active conversation capability of the voice assistant is fully utilized to explore the preference of the user.

Drawings

FIG. 1 is a schematic view of the present invention.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.

Example 1:

the content recommendation method based on the vehicle-mounted dialogue comprises the following steps:

and S1, acquiring the voice information sent by the user.

The text content expressed by the user is judged whether the emotional state is proved, negative or other through a text classification model by using a natural language processing technology.

S3, the conversation strategy module determines whether to enter the next round of interaction, and when the conversation strategy module judges that a next round of interaction is not needed, the conversation strategy module enters a recommendation module to perform recommendation operation; otherwise go to S1 to enter the next round of interaction.

The user's preferences are progressively clarified through dialog interactions without affecting the user experience.

And when the emotional tendency or the preference of the user is negative, directly switching to a recommending module to recommend the goods or the services to the user. When the user expresses negative emotional tendency, the user preference can be timely stopped being explored, and the goods and services can be timely recommended.

Example 2:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, the conversation strategy module is switched to. And the recommending module takes the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix to construct a two-stage recommending system based on recall and sequencing.

Example 3:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, go to the dialogue strategy module. And the recommending module is used for constructing a two-stage recommending system based on recall and sequencing by taking the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix. The recommendation module also constructs a feature matrix according to the preference in the features of the user by taking the current user input and the short-term user log as short-term preferences and taking the preference in the user portrait as long-term preferences.

S3, the conversation strategy module determines whether to enter the next round of interaction, and when the conversation strategy module judges that a next round of interaction is not needed, the conversation strategy module enters a recommendation module to perform recommendation operation; otherwise go to S1 for the next round of interaction.

Example 4:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, go to the dialogue strategy module. And the recommending module is used for constructing a two-stage recommending system based on recall and sequencing by taking the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix. And the recommendation module is used for taking the current user input and the short-term user log as short-term preferences according to the preferences in the characteristics of the user, taking the preferences in the user portrait as long-term preferences, and performing weighted calculation on the short-term preferences and the long-term preferences to obtain the user preferences.

Example 5:

and S1, acquiring the voice information sent by the user.

The recommendation module is used for taking the current user input and the short-term user log as short-term preferences according to the preferences in the characteristics of the user, taking the preferences in the user portrait as long-term preferences, and performing weighted calculation on the short-term preferences and the long-term preferences to obtain the user preferences.

The short-term user log contains the types of songs listened to by the user in the short-term duration, the duration of listening to the songs and the types of songs not listened to by the user. Preferably, the short term does not exceed 7 days, for example 3 days. Through the short-term user log, the recent preference of the user can be known, and the goods or services can be conveniently and pertinently recommended.

Example 6:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, go to the dialogue strategy module. And the recommending module is used for constructing a two-stage recommending system based on recall and sequencing by taking the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix.

The short-term user log contains the types of songs listened to by the user in the short-term duration, the duration of listening to the songs and the types of songs not listened to by the user. Preferably, the short term does not exceed 7 days, for example 3 days. By the short-term user log, the recent preference of the user can be known, and the goods or services can be conveniently and pertinently recommended.

Preferably, the recommendation module uses an attention weighting calculation.

Example 7:

referring to fig. 1, the content recommendation method based on the vehicle-mounted dialogue according to the embodiment includes the following steps:

and S1, acquiring the voice information sent by the user.

S2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, the conversation strategy module is switched to.

The dialogue content is based on a specific scene, and the preference of the user in the specific scene is determined by analyzing the dialogue content, constructing a knowledge graph of the corresponding scene, carrying out dialogue guidance based on prior knowledge and the like, so that the dialogue recommendation process is more consistent with the business scene.

And the conversation strategy module judges the action of the next step based on the DQN network of the time state of each step.

Example 8:

the embodiment provides a content recommendation device based on an in-vehicle dialogue, which comprises a processor and a memory.

The processor may be a multi-core processor or may include a plurality of processors. In some embodiments, the processor may comprise a general-purpose host processor and one or more special coprocessors such as a Graphics Processor (GPU), a Digital Signal Processor (DSP), or the like. In some embodiments, the processor may be implemented using custom circuitry, such as an Application Specific Integrated Circuit (ASIC) or a Field Programmable Gate Array (FPGA).

The memory may include various types of storage units, such as system memory, Read Only Memory (ROM), and permanent storage. Wherein the ROM may store static data or instructions for the processor or other modules of the computer. The persistent storage device may be a read-write storage device. The persistent storage may be a non-volatile storage device that does not lose stored instructions and data even after the computer is powered down. In some embodiments, the persistent storage device employs a mass storage device (e.g., magnetic or optical disk, flash memory) as the persistent storage device. In other embodiments, the permanent storage may be a removable storage device (e.g., floppy disk, optical drive). The system memory may be a read-write memory device or a volatile read-write memory device, such as a dynamic random access memory. The system memory may store instructions and data that some or all of the processors require at runtime. Further, the memory may comprise any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), magnetic and/or optical disks, may also be employed. In some embodiments, the memory may include a removable storage device that is readable and/or writable, such as a Compact Disc (CD), digital versatile disc read only (e.g., DVD-ROM, dual layer DVD-ROM), Blu-ray disc read only, ultra-compact disc, flash memory card (e.g., SD card, min SD card, Micro-SD card, etc.), magnetic floppy disk, and the like. Computer-readable storage media do not contain carrier waves or transitory electronic signals transmitted by wireless or wired means.

The memory has stored thereon executable code which, when executed by the processor, causes the processor to perform the methods described in the above embodiments.

Example 9:

the present embodiments provide a non-transitory machine-readable storage medium having stored thereon executable code, which when executed by a processor of an electronic device, causes the processor to perform the above-described method.

A non-transitory machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) having stored thereon executable code (or a computer program, or computer instruction code) which, when executed by a processor of an electronic device (or a computing device, a server, etc.), causes the processor to perform the various steps of the above-described method according to the invention.

Those of skill would further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or combinations of both.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

It will be apparent to those skilled in the art that various modifications and variations can be made in the above embodiments of the present invention without departing from the spirit of the invention.

Claims

1. A content recommendation method based on vehicle-mounted conversation is characterized by comprising the following steps:

s1, acquiring voice information sent by a user;

s2, analyzing the user emotional tendency and/or user preference contained in the voice information through a language understanding module; when the emotional tendency or the user preference is analyzed to be negative, the recommending module carries out recommending operation; otherwise, switching to a conversation strategy module;

2. The content recommendation method based on vehicle-mounted dialogue according to claim 1, characterized in that: and the recommending module takes the user portrait constructed according to the user history log as a user characteristic matrix and an article matrix to construct a two-stage recommending system based on recall and sequencing.

3. The content recommendation method based on vehicle dialog according to claim 2, characterized in that: the recommendation module also constructs a feature matrix according to the preference in the features of the user by taking the current user input and the short-term user log as short-term preferences and taking the preference in the user portrait as long-term preferences.

4. The content recommendation method based on vehicle-mounted dialogue according to claim 2, characterized in that: and the recommendation module is used for taking the current user input and the short-term user log as short-term preferences according to the preferences in the characteristics of the user, taking the preferences in the user portrait as long-term preferences, and performing weighted calculation on the short-term preferences and the long-term preferences to obtain the user preferences.

5. The content recommendation method based on vehicle-mounted dialogue according to claim 4, wherein: the short-term user log comprises types of songs listened to by the user within the short-term time length, the song listening time length and the song types not listened to by the user.

6. The content recommendation method based on vehicle-mounted dialogue according to claim 5, characterized in that: the short term duration does not exceed 7 days.

7. The content recommendation method based on vehicle dialog according to claim 6, wherein: the recommendation module uses the attention weighting calculation.

8. The content recommendation method based on vehicle-mounted dialogue according to claim 1, characterized in that: and the conversation strategy module judges the action of the next step based on the DQN network of the time state of each step.

9. A content recommendation device based on vehicle-mounted conversation is characterized in that: comprising a processor and a memory having executable code stored thereon, which when executed by the processor causes the processor to perform the method of any of claims 1 to 8.

10. A non-transitory machine-readable storage medium, characterized in that: stored thereon executable code which, when executed by a processor of an electronic device, causes the processor to perform the method of any of claims 1 to 8.