Dynamic fact-based ambiguous sentence disambiguation method and device, equipment and storage medium

Abstract

The invention provides an ambiguous sentence disambiguation method and device based on a dynamic fact, equipment and a storage medium, the method discriminates ambiguous sentences by introducing a pre-configured ambiguous sentence library, and performs text data intention judgment by combining dynamic fact data, so that disambiguation processing of the ambiguous sentences is well realized, and the disambiguation efficiency is improved. According to the technical scheme, the system can more accurately judge the real intention of the user according to the dynamic fact data, so that targeted replies or questions are provided, an excessively simple and rough processing mode is avoided, and more accurate and personalized voice assistant interaction experience is realized.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and storage medium for disambiguating ambiguous sentences based on dynamic facts. Background Art

[0002] Voice assistant is a relatively common artificial intelligence application. Many personal mobile phones, computers and various smart home devices are equipped with such applications. Voice assistants answer users' questions mainly by identifying the user's speaking intention and slot. The so-called intention is what the user wants the voice assistant to do, execute a command (such as adding a schedule) or answer a question (such as what is the capital of the United States). The slot is some keywords that the user tells when executing a type of intention, such as the location and time when checking the weather.

[0003] However, during the interaction, there may be ambiguous sentences, which hinders AI from providing accurate services to users. Limited by the current level of technology, disambiguation (i.e. eliminating ambiguity and locking in the user's true intention from multiple understandings) has always been a difficult point in the natural language understanding task. At present, there are mainly the following methods to deal with such problems:

[0004] 1) Treat them uniformly according to one intent, for example, according to a more conservative intent classification;

[0005] 2) Treat it as a vague intention and use a unified language to further question the user, allowing the user to express his or her intention in a more specific way.

[0006] 3) Treat it as fuzzy intent, use different wordings according to the number of possible intentions to further question the user, and let the user express his or her intention in a more specific way.

[0007] Existing methods are too simplistic, crude and mechanical, and are unable to provide intelligent and personalized targeted responses, resulting in a poor user experience. Summary of the invention

[0008] The object of the present invention is to provide a method, device, equipment and storage medium for disambiguating ambiguous sentences based on dynamic facts.

[0009] The present invention provides a method for disambiguating ambiguous sentences based on dynamic facts, which comprises the steps of:

[0010] Configuring an ambiguous sentence library, configuring common ambiguous sentences and intention branches of ambiguous sentences related to the current device to the ambiguous sentence library;

[0011] Get the user's input voice and transcribe it into corresponding text data;

[0012] Comparing the text data with the content of the ambiguous sentence library, and obtaining the intention when determining that the text data is an ambiguous sentence;

[0013] Based on the intention of the text data, dynamic fact data is acquired, where the dynamic fact data includes interaction data between the current user and the device and / or current state information of the device;

[0014] Based on the text data and the dynamic fact data, respond to the voice input by the user.

[0015] As a further improvement of the present invention, the acquiring of dynamic fact data based on the intention of the text data specifically includes:

[0016] Based on the intention of the text data, determining the type of the dynamic fact data to be acquired;

[0017] According to the required type of dynamic fact data, corresponding dynamic fact data is acquired.

[0018] As a further improvement of the present invention, the obtaining of the corresponding fact dynamic fact data specifically includes:

[0019] According to the required type of dynamic fact data, current user and device interaction information data and / or current device inventory status information and / or current inventory item information are obtained.

[0020] As a further improvement of the present invention, the configuration ambiguous sentence library further includes:

[0021] The response discourse term sentence corresponding to the ambiguous sentence is configured into the ambiguous sentence library.

[0022] As a further improvement of the present invention, the responding to the voice input by the user specifically includes:

[0023] Based on the text data and the dynamic fact data, the voice input by the user is answered and replied, or further questions are asked based on the response words.

[0024] As a further improvement of the present invention, it also includes:

[0025] When it is determined that the text data is not an ambiguous sentence, the natural language understanding module parses the semantics of the text data and outputs corresponding reply content.

[0026] The present invention also provides an ambiguous sentence disambiguation device based on dynamic facts, which comprises:

[0027] An ambiguous sentence library configuration module, configured to configure common ambiguous sentences and intention branches of ambiguous sentences related to the current device into the ambiguous sentence library;

[0028] A text acquisition module, which is configured to acquire the voice input by the user and transcribe it into corresponding text data;

[0029] An ambiguity identification and ambiguity information acquisition module, which is configured to compare the text data with the content of the ambiguous sentence library, and when the text data is determined to be an ambiguous sentence, obtain its intention;

[0030] A dynamic fact data acquisition module, configured to acquire dynamic fact data based on the intention of the text data, wherein the dynamic fact data includes interaction data between the current user and the device and / or current state information of the device;

[0031] The semantic clarification module is configured to respond to the voice input by the user based on the text data and the dynamic fact data.

[0032] As a further improvement of the present invention, the dynamic fact data acquisition module is configured to determine the type of the dynamic fact data to be acquired based on the intention of the text data; and acquire the corresponding fact dynamic fact data according to the type of the dynamic fact data required.

[0033] As a further improvement of the present invention, the dynamic fact data acquisition module is configured to acquire current user and device interaction information data and / or current device inventory status information and / or current inventory item information according to the required dynamic fact data type.

[0034] As a further improvement of the present invention, the ambiguous sentence library configuration module is further configured to configure the response discourse term sentence corresponding to the ambiguous sentence into the ambiguous sentence library.

[0035] As a further improvement of the present invention, the semantic clarification module is further configured to answer and reply to the voice input by the user based on the text data and the dynamic fact data, or to further inquire based on the response words.

[0036] The present invention also provides a refrigeration device, comprising:

[0037] A memory for storing executable instructions;

[0038] The processor is used to implement the above-mentioned method for disambiguating ambiguous sentences based on dynamic facts when running the executable instructions stored in the memory.

[0039] The present invention also provides a computer storage medium storing a computer program for executing the above-mentioned method for disambiguating ambiguous sentences based on dynamic facts.

[0040] The beneficial effects of the present invention are as follows: the present invention distinguishes ambiguous sentences by introducing a pre-configured ambiguous sentence library, and judges the intention of text data by combining dynamic fact data, thereby better realizing the disambiguation processing of ambiguous sentences. The system can more accurately judge the user's true intention based on dynamic fact data, thereby providing targeted replies or follow-up questions, avoiding overly simple and crude processing methods, and realizing a more accurate and personalized voice assistant interaction experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a schematic diagram of the steps of an ambiguous sentence disambiguation method based on dynamic facts in one embodiment of the present invention.

[0042] Figure 2 It is a structural block diagram of an ambiguous sentence disambiguation device based on dynamic facts in one embodiment of the present invention. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in combination with the specific implementation methods of the present invention and the corresponding drawings. Obviously, the described implementation methods are only part of the implementation methods of the present invention, not all implementation methods. Based on the implementation methods of the present invention, all other implementation methods obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0044] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0045] This implementation compares the user's input voice data with a pre-configured ambiguous sentence library, which can effectively determine whether the user's voice input belongs to ambiguous content, and when determining that it belongs to ambiguous content, combines the current dynamic fact data to perform semantic discrimination of ambiguous sentences, thereby improving the recognition ability of sentences with ambiguous content and the accuracy of content responses, enabling the device's voice interaction system to more accurately understand the user's intentions, thereby increasing the intelligence of the voice assistant and user satisfaction.

[0046] An ambiguous sentence is one that can be understood or interpreted in multiple ways. The same sentence may have different meanings due to different contexts, voice tones, intonations, and other factors, making it difficult to accurately determine the user's true intention in natural language processing and dialogue systems. For example, if a user says "there are apples in the refrigerator", this sentence may mean that the user has stored apples in the refrigerator, or it may be a statement about the current storage situation in the refrigerator, that is, the apples are already stored in the refrigerator.

[0047] like Figure 1 As shown, this embodiment provides an ambiguous sentence disambiguation method based on dynamic facts, which includes the steps of:

[0048] S1: Configure an ambiguous sentence library, and configure common ambiguous sentences and intention branches of ambiguous sentences related to the current device into the ambiguous sentence library.

[0049] S2: Obtain the voice input by the user and transcribe it into corresponding text data.

[0050] S3: Compare the text data with the contents of the ambiguous sentence library, and when the text data is determined to be an ambiguous sentence, obtain its intent.

[0051] S4: Based on the intent of the text data, dynamic fact data is obtained, where the dynamic fact data includes interaction data between the current user and the device and / or current status information of the device.

[0052] S5: Respond to the user's voice input based on text data and dynamic fact data.

[0053] In this embodiment, the method is explained by taking a smart refrigerator as an example, that is, the common ambiguous sentences in the ambiguous sentence library are common ambiguous sentences in the refrigerator field. In other embodiments of the present invention, if applied to different devices, the content in the ambiguous sentence library can be adjusted according to the functions and scenarios of the specific devices. For example, if applied to a smart speaker, the ambiguous sentences in the ambiguous sentence library can be associated with the functions and operations of the speaker. Different devices may face different types of ambiguous problems, so corresponding adjustments can be made according to the characteristics of the device to achieve better ambiguity elimination and intelligent reply effects.

[0054] In step S1, it also includes: configuring the response discourse sentence corresponding to the ambiguous sentence into the ambiguous sentence library.

[0055] Response scripts are response strategies designed by humans or intelligent AI to ensure that responses are reasonable and accurate, thus meeting user needs. When the voice assistant encounters ambiguous sentences in actual applications, choosing appropriate response scripts can provide a smarter and more personalized interactive experience. This configuration method further enhances the intelligence of the voice assistant, enabling it to better adapt to different scenarios and user needs.

[0056] Intent branches refer to the semantic interpretation of different user intentions in natural language processing tasks. In voice assistants or dialogue systems, user input usually expresses a specific intention, and the system needs to understand the user's needs and provide corresponding responses or perform corresponding operations based on the intention. By distinguishing intent branches, the user's true intention can be better captured and processed accordingly. For intent branches, judgments can be made based on keywords or contextual information in text data, and classification can be performed. In a smart refrigerator, intent branches may include "query ingredients", "add ingredients", "delete ingredients", etc.

[0057] By configuring the ambiguous sentence library, you can pre-store the relevant content of ambiguous sentences to provide support for subsequent ambiguous sentence recognition. Since the ambiguous sentence library has stored the information of common ambiguous sentences, the system can directly match the ambiguous sentence library when recognizing user input in real time, avoiding complex natural language understanding calculations every time, thereby saving computing resources and improving the system's response speed. In addition, during use, the content of the ambiguous sentence library can be dynamically updated and expanded according to the actual application scenario.

[0058] In step S2, the voice data may be transcribed into text data by referring to the existing voice transcription technology, which will not be described in detail here.

[0059] In step S3, the acquired text data is compared with the contents of the ambiguous sentence library. When it is determined to be an ambiguous sentence, its intention is obtained based on the contents stored in the ambiguous sentence library for subsequent disambiguation processing and intelligent reply.

[0060] When making comparisons, the text data can be preprocessed, such as word segmentation, stop word removal, etc., so as to match the sentences in the ambiguous sentence library. The preprocessed text data is compared with each ambiguous sentence in the ambiguous sentence library. The similarity between the text data and the ambiguous sentences can be determined by using, for example, a string matching algorithm or a similarity calculation method based on word vectors.

[0061] When it is determined that it is not an ambiguous sentence, a normal voice interaction process is performed, and the natural language understanding module analyzes the semantics of the text data and outputs the corresponding reply content. The natural language understanding module is a prior art and will not be described in detail here.

[0062] In step S4, it specifically includes:

[0063] S41: Based on the intent of the text data, determine the type of dynamic fact data that needs to be obtained.

[0064] S42: Acquire corresponding dynamic fact data according to the required dynamic fact data type.

[0065] Specifically, the dynamic fact data includes user-device interaction information data and / or current device inventory status information and / or current inventory item information, etc.

[0066] In this embodiment, the user-device interaction information data is the user's current actions of taking and placing food, adjusting the refrigerator temperature, adjusting the placement of items in the refrigerator inventory, etc. The current device inventory status information is the food storage status of the refrigerator refrigerator, freezer, ice making room, etc. The current inventory item information is the detailed information of specific food in the refrigerator, such as the name of the food, storage time, shelf life, etc. In other embodiments of the present invention, the dynamic fact data content can also be adjusted according to the specific device and its application scenario.

[0067] In step S41, the system will determine the type of dynamic fact data that needs to be obtained based on the acquired ambiguous sentence intention branch. The intention information can tell the system the user's possible specific intention, for example: if the user's intention is to query ingredients, the dynamic fact data is the current device inventory status information and / or the current device inventory status information; if the user's intention is to query the food situation, etc., the dynamic fact data is the current device inventory status information; if the user's intention is to add ingredients, delete ingredients, etc., the dynamic fact data is the user and device interaction information data. Based on different intentions, the system will determine what type of dynamic fact data is needed to better understand the user's needs, thereby improving the accuracy of understanding ambiguous sentences. By obtaining these dynamic fact data, the voice assistant can perceive the user's environment and behavior changes in real time, combined with the subsequent interaction process, so as to provide personalized, real-time services and improve the user experience.

[0068] In step S5, it specifically includes:

[0069] Based on text data and dynamic factual data, the user's voice input is answered and responded to, or further questions are asked based on the response language.

[0070] Combining the pre-configured ambiguous sentence intention branches and the detected dynamic fact data, the true meaning of the ambiguous sentence can be understood more accurately, so that the response words in the pre-configured ambiguous sentence library can be used to generate answers to users, or the natural language understanding module can be used to parse the text data semantics and output the corresponding reply content. When it is found that there is still ambiguity or uncertainty, further questions can be asked based on the pre-configured words. These questions can help clarify the user's intentions so as to more accurately meet the user's needs.

[0071] Based on the content of the output reply, the voice interaction system will convert the generated answer into speech form, convert it into natural speech through the speech synthesis module, and then output it to the user.

[0072] like Figure 2As shown, based on the same inventive concept, the present invention also provides an ambiguous sentence disambiguation device based on dynamic facts, which includes:

[0073] The ambiguous sentence library configuration module 1 is configured to configure common ambiguous sentences and intention branches of ambiguous sentences related to the current device into the ambiguous sentence library.

[0074] The text acquisition module 2 is configured to acquire the voice input by the user and transcribe it into corresponding text data.

[0075] The ambiguity identification and ambiguity information acquisition module 3 is configured to compare the text data with the content of the ambiguous sentence library, and when the text data is determined to be an ambiguous sentence, obtain its intention.

[0076] The dynamic fact data acquisition module 4 is configured to acquire dynamic fact data based on the intention of the text data, and the dynamic fact data includes the interaction data between the current user and the device and / or the current state information of the device.

[0077] The semantic clarification module 5 is configured to respond to the voice input by the user based on the text data and the dynamic fact data.

[0078] Furthermore, the dynamic fact data acquisition module 4 is configured to determine the type of dynamic fact data to be acquired based on the intention of the text data; and acquire the corresponding dynamic fact data according to the type of dynamic fact data required.

[0079] Furthermore, the dynamic fact data acquisition module 4 is configured to acquire current user-device interaction information data and / or current device inventory status information and / or current inventory item information according to the required dynamic fact data type.

[0080] Furthermore, the ambiguous sentence library configuration module 3 is also configured to configure the response discourse term sentence corresponding to the ambiguous sentence into the ambiguous sentence library.

[0081] Furthermore, the semantic clarification module 5 is also configured to answer and reply to the voice input by the user based on the text data and dynamic fact data, or to ask further questions based on the response words.

[0082] Based on the same inventive concept, this embodiment also provides a refrigeration device, which includes:

[0083] A memory for storing executable instructions;

[0084] The processor is configured to implement the above-mentioned method for disambiguating ambiguous sentences based on dynamic facts when running the executable instructions stored in the memory.

[0085] This embodiment also provides a computer storage medium, in which a computer program is stored. The computer program is used to execute the above-mentioned method for disambiguating ambiguous sentences based on dynamic facts.

[0086] In summary, this embodiment introduces a pre-configured ambiguous sentence library to distinguish ambiguous sentences, and judges the intention of text data by combining dynamic fact data, thereby better achieving the disambiguation of ambiguous sentences. The system can more accurately judge the user's true intention based on dynamic fact data, thereby providing targeted replies or follow-up questions, avoiding overly simple and crude processing methods, and achieving a more accurate and personalized voice assistant interaction experience.

[0087] It should be understood that although this specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each implementation mode may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

[0088] The series of detailed descriptions listed above are only specific descriptions of feasible implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent implementation methods or changes that do not deviate from the technical spirit of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for disambiguating ambiguous sentences based on dynamic facts, characterized in that: Includes steps: Configuring an ambiguous sentence library, configuring common ambiguous sentences and intention branches of ambiguous sentences related to the current device to the ambiguous sentence library; Get the user's input voice and transcribe it into corresponding text data; Comparing the text data with the content of the ambiguous sentence library, and obtaining the intention when determining that the text data is an ambiguous sentence; Based on the intention of the text data, dynamic fact data is acquired, where the dynamic fact data includes interaction data between the current user and the device and / or current state information of the device; Based on the text data and the dynamic fact data, respond to the voice input by the user.

2. The method for disambiguating ambiguous sentences based on dynamic facts according to claim 1, characterized in that: The acquiring of dynamic fact data based on the intention of the text data specifically includes: Based on the intention of the text data, determining the type of the dynamic fact data to be acquired; According to the required type of dynamic fact data, corresponding dynamic fact data is acquired.

3. The method for disambiguating ambiguous sentences based on dynamic facts according to claim 2, characterized in that: The obtaining of the corresponding dynamic fact data specifically includes: According to the required type of dynamic fact data, current user and device interaction information data and / or current device inventory status information and / or current inventory item information are obtained.

4. The method for disambiguating ambiguous sentences based on dynamic facts according to claim 1, characterized in that: The configuration ambiguous sentence library also includes: The response discourse term sentence corresponding to the ambiguous sentence is configured into the ambiguous sentence library.

5. The method for disambiguating ambiguous sentences based on dynamic facts according to claim 4, characterized in that: The responding to the voice input by the user specifically includes: Based on the text data and the dynamic fact data, the voice input by the user is answered and replied, or further questions are asked based on the response words.

6. The method for disambiguating ambiguous sentences based on dynamic facts according to claim 1, characterized in that: Also includes: When it is determined that the text data is not an ambiguous sentence, the natural language understanding module parses the semantics of the text data and outputs corresponding reply content.

7. An ambiguous sentence disambiguation device based on dynamic facts, characterized in that: include: An ambiguous sentence library configuration module, configured to configure common ambiguous sentences and intention branches of ambiguous sentences related to the current device into the ambiguous sentence library; A text acquisition module, which is configured to acquire the voice input by the user and transcribe it into corresponding text data; An ambiguity identification and ambiguity information acquisition module, which is configured to compare the text data with the content of the ambiguous sentence library, and when the text data is determined to be an ambiguous sentence, obtain its intention; A dynamic fact data acquisition module, configured to acquire dynamic fact data based on the intention of the text data, wherein the dynamic fact data includes interaction data between the current user and the device and / or current state information of the device; The semantic clarification module is configured to respond to the voice input by the user based on the text data and the dynamic fact data.

8. The device for disambiguating ambiguous sentences based on dynamic facts according to claim 7, characterized in that: The dynamic fact data acquisition module is configured to determine the type of the dynamic fact data to be acquired based on the intention of the text data; and acquire the corresponding dynamic fact data according to the type of the dynamic fact data to be acquired.

9. The device for disambiguating ambiguous sentences based on dynamic facts according to claim 7, characterized in that: The dynamic fact data acquisition module is configured to acquire current user-device interaction information data and / or current device inventory status information and / or current inventory item information according to the required dynamic fact data type.

10. The device for disambiguating ambiguous sentences based on dynamic facts according to claim 7, characterized in that: The ambiguous sentence library configuration module is also configured to configure the response discourse term sentence corresponding to the ambiguous sentence into the ambiguous sentence library.

11. The device for disambiguating ambiguous sentences based on dynamic facts according to claim 7, characterized in that: The semantic clarification module is further configured to answer and reply to the voice input by the user based on the text data and the dynamic fact data, or to conduct further inquiries based on the response words.

12. A refrigeration device, characterized in that: include: A memory for storing executable instructions; A processor, configured to implement the method for disambiguating ambiguous sentences based on dynamic facts as described in any one of claims 1 to 5 when running the executable instructions stored in the memory.

13. A computer storage medium, characterized in that: A computer program is stored therein, which is used to execute the method for disambiguating ambiguous sentences based on dynamic facts as described in any one of claims 1 to 5.

Images