Man-machine conversation-based fault processing method, device, equipment, storage and program

Abstract

The invention relates to a man-machine conversation-based fault processing method, device and equipment, a storage medium and a computer program, and the method comprises the steps: obtaining input data of a user through preset input equipment, carrying out the semantic recognition of the input data through a pre-trained intention recognition model, and obtaining a semantic recognition result, if the semantic recognition result is a common question, performing similarity matching on the semantic recognition result by using a pre-established common question answering library to obtain a matching result, and outputting an answering result according to the matching result, otherwise, querying the semantic recognition result in a pre-established knowledge graph database to obtain a query result, and outputting an answering result according to the query result, displaying the answering result to the user, obtaining a feedback result of the user on the answering result, if the feedback result is unsatisfactory, prompting the user to perform question input again and obtaining input data of the user again, and otherwise, confirming that the process is ended.

Description

Technical Field

[0001] The present disclosure relates to the technical field of railway intelligent operation and maintenance, and in particular to a fault handling method, device, equipment, medium and program based on human-computer dialogue. Background Art

[0002] In the process of railway operation and maintenance, timely detection and handling of faults is the key to ensuring the safe operation of trains. Traditional fault handling methods rely on manual inspections and telephone reports, which have problems such as low efficiency, prone to errors, and slow response. With the development of artificial intelligence and Internet of Things technologies, the rapid identification and handling of faults through human-computer dialogue and cross-device interaction has become a research hotspot.

[0003] However, the existing fault handling methods based on human-computer dialogue are mostly limited to the interaction of a single device or a single mode, making it difficult to achieve seamless collaboration among multiple devices and comprehensive integration of multimodal information, which affects the efficiency and accuracy of fault handling. Summary of the invention

[0004] The present invention provides a human-computer dialogue-based fault handling method, device, equipment, medium and program to solve the problems of low efficiency and low accuracy of the existing human-computer dialogue-based fault handling method.

[0005] In a first aspect, the present disclosure provides a fault handling method based on human-computer dialogue, comprising:

[0006] Using a preset input device to obtain user input data;

[0007] Using a pre-trained intent recognition model to perform semantic recognition on the input data to obtain a semantic recognition result;

[0008] Determining whether the semantic recognition result is a common problem;

[0009] If the semantic recognition result is a common question, similarity matching is performed on the semantic recognition result using a pre-established common question answering library to obtain a matching result, and an answer result is output according to the matching result;

[0010] If the semantic recognition result is not a common problem, query the semantic recognition result in a pre-built knowledge graph database to obtain a query result, and output a solution result according to the query result;

[0011] Displaying the answer result to the user and obtaining the user's feedback on the answer result;

[0012] If the feedback result is unsatisfactory, the user is prompted to re-input the question, and the process returns to the step of obtaining the user's input data by using the preset input device;

[0013] If the feedback result is satisfactory, the confirmation process ends.

[0014] In some embodiments, the step of obtaining user input data using a preset input device includes:

[0015] Obtaining an input instruction from a user, and determining whether the input instruction is voice data or text data;

[0016] If the input instruction is voice data, the input instruction is converted into text data through voice recognition.

[0017] If the input instruction is text data, segmenting the text data to obtain segmented text;

[0018] The segmented text is subjected to part-of-speech tagging to obtain a tagged text, and named entity recognition is performed on the tagged text to obtain the input data.

[0019] In some embodiments, after obtaining the user's input data using a preset input device, the method further includes:

[0020] Obtain multimodal fault data of users;

[0021] Performing semantic analysis on text information contained in the multimodal fault data to obtain semantic features;

[0022] Extracting features of images contained in the multimodal fault data to obtain image features;

[0023] Using a pre-trained fault recognition model to perform fault recognition according to the semantic features and the image, and obtaining a fault recognition result;

[0024] Generate a fault handling method according to the fault identification result using a pre-established fault database;

[0025] The troubleshooting method is presented to the user.

[0026] In some embodiments, after obtaining the user's input data using a preset input device, the method further includes:

[0027] Obtain multimodal fault data of users;

[0028] Performing semantic analysis on text information contained in the multimodal fault data to obtain semantic features;

[0029] Extracting features of images contained in the multimodal fault data to obtain image features;

[0030] Using a pre-trained fault recognition model to perform fault recognition according to the semantic features and the image, and obtaining a fault recognition result;

[0031] Generate a fault handling method according to the fault identification result using a pre-established fault database;

[0032] The troubleshooting method is presented to the user.

[0033] In some embodiments, determining whether the semantic recognition result is a common problem includes:

[0034] Obtain the question type annotation contained in the semantic recognition result;

[0035] Matching is performed in a preset frequently asked questions database according to the question type annotation to obtain a matching result;

[0036] If the matching result is empty, it is determined that the semantic recognition result is not a common problem;

[0037] If the matching result is not empty, the semantic recognition result is determined to be a common problem.

[0038] In some embodiments, the using of a pre-established FAQ library to perform similarity matching on the semantic recognition results to obtain matching results includes:

[0039] Performing word segmentation processing on the semantic recognition result to obtain a word segmentation recognition result;

[0040] Obtaining a vector representation of each word in the word segmentation recognition result to obtain a word vector set;

[0041] Calculate an average word vector based on the word vector set;

[0042] Calculate the cosine similarity between the vector of each question text in the FAQ library and the average word vector;

[0043] The question text with the largest cosine similarity in the FAQ database is selected as the matching result.

[0044] In a second aspect, the present disclosure provides a fault handling device based on human-computer dialogue, comprising:

[0045] A data acquisition module, used to acquire user input data using a preset input device;

[0046] A semantic recognition module, used to perform semantic recognition on the input data using a pre-trained intent recognition model to obtain a semantic recognition result;

[0047] A question determination module is used to determine whether the semantic recognition result is a common question. If the semantic recognition result is a common question, a pre-established common question answering library is used to perform similarity matching on the semantic recognition result to obtain a matching result, and an answer result is output according to the matching result. If the semantic recognition result is not a common question, a pre-established knowledge graph database is used to query the semantic recognition result to obtain a query result, and an answer result is output according to the query result.

[0048] The data feedback module is used to display the answer result to the user and obtain the user's feedback on the answer result. If the feedback result is unsatisfactory, the user is prompted to re-enter the question and return to the step of obtaining the user's input data using the preset input device. If the feedback result is satisfactory, the confirmation process ends.

[0049] In a third aspect, the present disclosure provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method described in the above aspects.

[0050] In a fourth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in the above aspects when executed by a processor.

[0051] In a fifth aspect, the present disclosure provides a computer program product, including a computer program, which implements the steps of the method described in the above aspects when executed by a processor.

[0052] The present disclosure provides a fault handling method, device, equipment, storage medium and computer program based on human-computer dialogue, which obtains user input data by using a preset input device, performs semantic recognition on the input data by using a pre-trained intention recognition model, obtains a semantic recognition result, determines whether the semantic recognition result is a common problem, and if the semantic recognition result is a common problem, performs similarity matching on the semantic recognition result by using a pre-established common problem answering library to obtain a matching result, outputs an answer result according to the matching result, and if the semantic recognition result is not a common problem, queries the semantic recognition result in a pre-built knowledge graph database to obtain a query result, outputs an answer result according to the query result, displays the answer result to the user, obtains the user's feedback result on the answer result, and if the feedback result is unsatisfactory, prompts the user to re-input the question, returns to the step of obtaining the user's input data by using the preset input device, and if the feedback result is satisfactory, confirms that the process ends. Thus, the problems of low efficiency and low accuracy of the existing fault handling method based on human-computer dialogue are solved, and the efficiency and accuracy of the fault handling method based on human-computer dialogue are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The present disclosure will be described in more detail below based on embodiments and with reference to the accompanying drawings:

[0054] Figure 1 A flowchart of a fault handling method based on human-computer dialogue provided by an embodiment of the present disclosure;

[0055] Figure 2 A functional module diagram of a fault handling device based on human-computer dialogue provided in an embodiment of the present disclosure.

[0056] In the drawings, the same reference numerals are used for the same components, and the drawings are not drawn to scale. DETAILED DESCRIPTION

[0057] In order to enable those skilled in the art to better understand the technical solution of the present disclosure, and to fully understand and implement how the present disclosure applies technical means to solve technical problems and achieve the corresponding technical effects, the technical solution in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only embodiments of a part of the present disclosure, not all of the embodiments. The embodiments of the present disclosure and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of the present disclosure. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.

[0058] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0059] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0060] Embodiment 1

[0061] Figure 1 The following is a flow chart of a fault handling method based on human-computer dialogue provided by an embodiment of the present disclosure. Figure 1 As shown, a fault handling method based on human-computer dialogue includes:

[0062] S1. Obtain user input data using a preset input device.

[0063] In the embodiment of the present invention, the obtaining of user input data by using a preset input device refers to obtaining voice data or text data input by the user and preprocessing the received data for subsequent processing.

[0064] In the embodiment of the present invention, the step of obtaining the user's input data by using a preset input device includes:

[0065] Obtaining an input instruction from a user, and determining whether the input instruction is voice data or text data;

[0066] If the input instruction is voice data, the input instruction is converted into text data through voice recognition.

[0067] If the input instruction is text data, segmenting the text data to obtain segmented text;

[0068] The segmented text is subjected to part-of-speech tagging to obtain a tagged text, and named entity recognition is performed on the tagged text to obtain the input data.

[0069] In detail, the conversion of the input command into text data through speech recognition may be the conversion of the input command into text data using existing speech recognition technology. Speech recognition technology is a computer technology that converts human speech into text or commands. It usually uses acoustic models and language models to analyze speech input and predict the most likely text or command. It includes four steps: sampling, parsing, matching and recognition: sampling converts speech signals into digital signals; parsing extracts feature information; matching processes feature information and compares it with the model; and recognizing and determining speech content.

[0070] In detail, the word segmentation process is a basic step in natural language processing (NLP), which involves breaking down a text string into smaller units, which may be words, phrases or other meaningful symbols.

[0071] In the embodiment of the present invention, after obtaining the user's input data by using the preset input device, the method further includes:

[0072] Obtain multimodal fault data of users;

[0073] Performing semantic analysis on text information contained in the multimodal fault data to obtain semantic features;

[0074] Extracting features of images contained in the multimodal fault data to obtain image features;

[0075] Using a pre-trained fault recognition model to perform fault recognition according to the semantic features and the image, and obtaining a fault recognition result;

[0076] Generate a fault handling method according to the fault identification result using a pre-established fault database;

[0077] The troubleshooting method is presented to the user.

[0078] In detail, the multimodal fault data includes user's voice data, text data and fault image data.

[0079] In detail, the fault identification model is a model built based on machine learning or deep learning algorithms, which can integrate information from multiple modes to make fault judgments.

[0080] In detail, the fault database is a database containing a large number of common fault problems and solutions.

[0081] In detail, displaying the fault handling method to the user includes but is not limited to displaying relevant text to the user using a display, and playing relevant voice to the user using a speaker.

[0082] In the embodiment of the present invention, by using a preset input device to obtain the input data of the user, the efficiency of semantic recognition of the input data is improved.

[0083] S2. Use a pre-trained intent recognition model to perform semantic recognition on the input data to obtain a semantic recognition result.

[0084] In the embodiment of the present invention, the intention recognition model is a semantic recognition model, which is a technology used in the field of natural language processing (NLP) to understand and interpret the meaning of natural language text. These models involve machine learning technology, which aims to enable computers to recognize and understand the intention and meaning in language.

[0085] In the embodiment of the present invention, the semantic recognition of the input data using the pre-trained intent recognition model to obtain the semantic recognition result includes:

[0086] Using a bag-of-words model to extract features from the input data to obtain a text vector;

[0087] Using the intention recognition model to encode the text vector to obtain semantic information;

[0088] Using the attention mechanism module in the intention recognition model to identify the importance of the semantic information, and obtain an importance recognition result;

[0089] The intention recognition model is used to output a semantic recognition result according to the semantic information and the importance recognition result.

[0090] In detail, the bag-of-words model is a very basic and important concept in the field of natural language processing and text mining. It is a simplified method for text representation, and its core idea is to convert text into a vector form so that computers can process it. The bag-of-words model regards the text as a "bag of words", in which each word is disordered and independent, while ignoring the order and grammatical structure of the words, and it constructs a feature vector of the text by counting the number of times each word appears in the text.

[0091] In detail, using the attention mechanism module in the intent recognition model to identify the importance of the semantic information refers to identifying the importance according to the importance weights of different parts of speech in the semantic information.

[0092] In the embodiment of the present invention, semantic recognition is performed on the input data using a pre-trained intent recognition model to obtain a semantic recognition result, thereby improving the efficiency of subsequent problem solving.

[0093] S3. Determine whether the semantic recognition result is a common problem.

[0094] In the embodiment of the present invention, the semantic recognition result includes labeling information of the question type.

[0095] In an embodiment of the present invention, the determination of whether the semantic recognition result is a common question is performed by matching the question type annotations contained in the semantic recognition result in a preset common question database, and determining whether it is a common question based on the matching results.

[0096] In the embodiment of the present invention, determining whether the semantic recognition result is a common problem includes:

[0097] Obtain the question type annotation contained in the semantic recognition result;

[0098] Matching is performed in a preset frequently asked questions database according to the question type annotation to obtain a matching result;

[0099] If the matching result is empty, it is determined that the semantic recognition result is not a common problem;

[0100] If the matching result is not empty, the semantic recognition result is determined to be a common problem.

[0101] In the embodiment of the present invention, by judging whether the semantic recognition result is a common problem, the efficiency of subsequent different processing according to the judgment result is improved.

[0102] If the semantic recognition result is a frequently asked question, S4 is executed to perform similarity matching on the semantic recognition result using a pre-established frequently asked question answering library to obtain a matching result, and an answer result is output according to the matching result.

[0103] In the embodiment of the present invention, when the semantic recognition result is a frequently asked question, the question can be matched with a pre-established frequently asked question answering library for similarity to confirm the specific question and then find the corresponding answer.

[0104] In an embodiment of the present invention, the semantic recognition result is matched for similarity using a pre-established FAQ library to obtain a matching result, by converting the semantic recognition result into a vector form, and then calculating the cosine similarity between the converted vector and the text in the FAQ library, and selecting the text with the largest pre-similarity as the matching result.

[0105] In the embodiment of the present invention, the similarity matching of the semantic recognition results using the pre-established FAQ library to obtain the matching results includes:

[0106] Performing word segmentation processing on the semantic recognition result to obtain a word segmentation recognition result;

[0107] Obtaining a vector representation of each word in the word segmentation recognition result to obtain a word vector set;

[0108] Calculate an average word vector based on the word vector set;

[0109] Calculate the cosine similarity between the vector of each question text in the FAQ library and the average word vector;

[0110] The question text with the largest cosine similarity in the FAQ database is selected as the matching result.

[0111] In detail, the bag-of-words model may be used to obtain the vector representation of each word in the word segmentation recognition result.

[0112] In detail, the cosine similarity is a measurement method for measuring the similarity between two vectors, and their similarity is measured by calculating the cosine value of the angle between the two vectors.

[0113] In an embodiment of the present invention, when the semantic recognition result is a frequently asked question, a matching result is obtained by performing similarity matching on the semantic recognition result using a pre-established frequently asked question answering library, and an answer result is output based on the matching result, thereby improving the efficiency and accuracy of obtaining the answer result.

[0114] If the semantic recognition result is not a common problem, execute S5, query the semantic recognition result in a pre-built knowledge graph database to obtain a query result, and output a solution result based on the query result.

[0115] In an embodiment of the present invention, when the semantic recognition result is not a frequently asked question, relevant matching cannot be performed in the frequently asked questions library, and therefore it is necessary to use a pre-built knowledge graph database for query.

[0116] In an embodiment of the present invention, querying the semantic recognition result in a pre-constructed knowledge graph database to obtain a query result refers to performing a data query in the knowledge graph database based on the keywords in the semantic recognition result to obtain multiple query results, and selecting the most relevant data as the final query result.

[0117] In an embodiment of the present invention, when the semantic recognition result is not a common question, the query result is obtained by querying the semantic recognition result in a pre-constructed knowledge graph database, and the answer result is output according to the query result, thereby improving the efficiency of outputting the answer result.

[0118] S6. Display the answer result to the user and obtain the user's feedback on the answer result.

[0119] In the embodiment of the present invention, presenting the answer result to the user includes but is not limited to presenting the text of the answer result to the user using a display screen, and playing the audio of the answer result to the user using a speaker.

[0120] In the embodiment of the present invention, the obtaining of the user's feedback result on the answer result is achieved by obtaining the user's touch operation on a preset display screen, wherein the touch operation includes the user's selection of satisfaction or dissatisfaction.

[0121] In the embodiment of the present invention, the efficiency of the fault handling process is improved by displaying the answer result to the user and obtaining the user's feedback result on the answer result.

[0122] S7. Determine whether the feedback result is satisfactory.

[0123] If the feedback result is unsatisfactory, S8 is executed to prompt the user to re-input the question, and the process returns to the step of obtaining the user's input data by using the preset input device.

[0124] In the embodiment of the present invention, when the feedback result is unsatisfactory, it means that the user's input instruction may be wrong, or there is an error in the analysis of the user's input instruction. Therefore, it is necessary to re-execute the step of obtaining the user's input data using the preset input device to give a correct answer result.

[0125] If the feedback result is satisfactory, S9 is executed to end the confirmation process.

[0126] In the embodiment of the present invention, when the feedback result is satisfactory, it means that the solution result meets the user's expectations, and therefore the fault handling process is confirmed to be finished.

[0127] Embodiment 2

[0128] Based on the above embodiments, this embodiment provides an application example.

[0129] The specific implementation includes but is not limited to the following steps:

[0130] 6.1. System architecture design: The system includes cross-device communication module, multimodal information processing module, fault identification and location module, treatment measure recommendation module and user interaction interface.

[0131] 6.2. Cross-device communication module: uses multiple communication technologies such as Bluetooth, Wi-Fi, 4G / 5G, etc. to achieve data transmission and synchronization between different devices.

[0132] 6.3. Multimodal information processing module: Utilizes speech recognition, natural language processing, image processing and other technologies to analyze and integrate multimodal information from different devices.

[0133] 6.4. Fault identification and location module: Based on the preset fault knowledge base and machine learning model, intelligent analysis of multimodal information is performed to achieve automatic identification and location of faults.

[0134] 6.5. Treatment measures recommendation module: automatically recommend corresponding treatment measures based on fault type and severity, and generate standardized treatment processes.

[0135] 6.6. User interaction interface: Provides an intuitive and easy-to-use user interaction interface that supports multiple interaction methods such as voice and touch, to realize the reporting of fault information, tracking of processing progress, and feedback of processing results.

[0136] 7.1 Several key steps of the human-computer dialogue processing system:

[0137] 7.1.1. Question reception and analysis:

[0138] The system first receives questions submitted by users via voice, text or other means.

[0139] Preprocess the received questions, including word segmentation, part-of-speech tagging, named entity recognition, etc., to facilitate subsequent processing.

[0140] 7.1.2. Intent Recognition:

[0141] Use a pre-trained intent recognition model (such as a binary classification or multi-classification model) to determine the type of user questions, whether they are domain-related or non-domain-related questions, or further identify which specific domain and subcategory they belong to.

[0142] Intent recognition models are usually trained based on a large amount of labeled data to improve recognition accuracy and generalization ability.

[0143] 7.1.3. Question classification and query:

[0144] For domain-related questions, the system needs to further determine whether it is a graph question and answer (query based on the knowledge graph) or a FAQ question and answer (matching based on a frequently asked questions library).

[0145] For graph question and answer, the system will query the corresponding knowledge graph database based on the content of the question and the vertical field it belongs to to obtain accurate answers.

[0146] For FAQ questions and answers, the system will match the user's questions with the pre-configured FAQ library to find the most relevant answers.

[0147] 7.1.4. Answer generation and feedback:

[0148] When the system successfully finds the answer, it will return the answer to the user in an appropriate manner (such as voice, text, etc.).

[0149] If the query is unsuccessful or the match is low, the system may take some remedial actions, such as asking the user for more information or forwarding the question to manual processing.

[0150] 7.1.5. Continuous optimization and iteration:

[0151] The system will be continuously optimized and improved based on user feedback and usage, including updating the intent recognition model, expanding the knowledge graph database, optimizing the FAQ library, etc., to improve the overall human-computer dialogue effect.

[0152] The cross-device multimodal interaction system aims to provide users with a seamless and natural interaction experience between different devices.

[0153] The system has the following features:

[0154] 7.2.1. Multimodal Input:

[0155] The system supports multiple input methods, including voice, text, image, touch, etc. Users can choose the most appropriate input method according to their habits and scenarios.

[0156] 7.2.2. Multimodal Fusion:

[0157] The system can integrate input information from different modalities to obtain a more comprehensive and accurate user intent, which usually requires the use of advanced algorithms and technologies, such as deep learning and natural language processing.

[0158] 7.2.3. Cross-device compatibility:

[0159] The system can run on different devices (such as smartphones, tablets, smart terminal devices, etc.) and achieve seamless connection and interaction between devices. This usually requires the use of unified communication protocols and interface standards.

[0160] 7.2.4. Intelligent decision-making and control:

[0161] The system can make intelligent decisions and controls based on the fused multimodal information and generate corresponding output instructions in response to user operations or requests. This usually requires the help of powerful cognitive models and decision-making algorithms.

[0162] 7.2.5. Personalization and Adaptation:

[0163] The system can provide a personalized interactive experience based on the user's habits, preferences and contextual information, and continuously learn and optimize as the user uses it. This helps to improve user satisfaction and loyalty.

[0164] 7.2.6. Security and Privacy Protection:

[0165] The system takes security and privacy protection into full consideration during design and implementation to ensure that user data is not leaked or abused. Encryption technology, access control and other measures are used to ensure user data security.

[0166] Embodiment 3

[0167] like Figure 2 , which is a functional module diagram of a fault handling device based on human-computer dialogue provided in this embodiment.

[0168] The human-computer dialogue-based fault handling device 100 of the present invention can be installed in an electronic device. According to the functions to be implemented, the human-computer dialogue-based fault handling device 100 can include a data acquisition module 101, a semantic recognition module 102, a problem judgment module 103, and a data feedback module 104. The module of the present invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0169] In this embodiment, the functions of each module / unit are as follows:

[0170] The data acquisition module 101 is used to acquire the user's input data using a preset input device;

[0171] The semantic recognition module 102 is used to perform semantic recognition on the input data using a pre-trained intention recognition model to obtain a semantic recognition result;

[0172] The question determination module 103 is used to determine whether the semantic recognition result is a common question. If the semantic recognition result is a common question, the semantic recognition result is matched by similarity using a pre-established common question answering library to obtain a matching result, and an answer result is output according to the matching result. If the semantic recognition result is not a common question, the semantic recognition result is queried in a pre-established knowledge graph database to obtain a query result, and an answer result is output according to the query result.

[0173] The data feedback module 104 is used to display the answer result to the user and obtain the user's feedback on the answer result. If the feedback result is unsatisfactory, the user is prompted to re-enter the question and return to the step of obtaining the user's input data using the preset input device. If the feedback result is satisfactory, the confirmation process ends.

[0174] Embodiment 4

[0175] Based on the above embodiments, this embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the following steps:

[0176] Using a preset input device to obtain user input data;

[0177] Using a pre-trained intent recognition model to perform semantic recognition on the input data to obtain a semantic recognition result;

[0178] Determining whether the semantic recognition result is a common problem;

[0179] If the semantic recognition result is a common question, similarity matching is performed on the semantic recognition result using a pre-established common question answering library to obtain a matching result, and an answer result is output according to the matching result;

[0180] If the semantic recognition result is not a common problem, query the semantic recognition result in a pre-built knowledge graph database to obtain a query result, and output a solution result according to the query result;

[0181] Displaying the answer result to the user and obtaining the user's feedback on the answer result;

[0182] If the feedback result is unsatisfactory, the user is prompted to re-input the question, and the process returns to the step of obtaining the user's input data by using the preset input device;

[0183] If the feedback result is satisfactory, the confirmation process ends.

[0184] In some implementations of this embodiment, a computer-readable storage medium is provided, on which a computer program is stored, characterized in that when the computer program is executed by a processor, the following steps are implemented:

[0185] Using a preset input device to obtain user input data;

[0186] Using a pre-trained intent recognition model to perform semantic recognition on the input data to obtain a semantic recognition result;

[0187] Determining whether the semantic recognition result is a common problem;

[0188] If the semantic recognition result is a common question, similarity matching is performed on the semantic recognition result using a pre-established common question answering library to obtain a matching result, and an answer result is output according to the matching result;

[0189] If the semantic recognition result is not a common problem, query the semantic recognition result in a pre-built knowledge graph database to obtain a query result, and output a solution result according to the query result;

[0190] Displaying the answer result to the user and obtaining the user's feedback on the answer result;

[0191] If the feedback result is unsatisfactory, the user is prompted to re-input the question, and the process returns to the step of obtaining the user's input data by using the preset input device;

[0192] If the feedback result is satisfactory, the confirmation process ends.

[0193] In some implementations of this embodiment, a computer program product is provided, including a computer program, characterized in that when the computer program is executed by a processor, the steps of the method described in the above embodiment are implemented.

[0194] The processor may include, but is not limited to, one or more processors or microprocessors, etc. Each processor may be an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field programmable gate array (FPGA), a controller, a microcontroller, a microprocessor or other electronic components to execute the methods in the above embodiments.

[0195] Computer-readable storage media can be implemented by any type of volatile or non-volatile storage device or a combination thereof. Computer-readable storage media may include, but are not limited to, for example, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, computer storage media (such as hard disks, floppy disks, solid-state drives, removable disks, Blu-ray disks, etc.).

[0196] The computer-readable storage medium may also store at least one computer executable program, such as a computer-readable instruction. The computer-readable storage medium includes, but is not limited to, for example, a volatile memory and / or a non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The computer-readable storage medium may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. For example, a non-transitory computer-readable storage medium may be connected to a computing device such as a computer, and then, when the computing device runs the computer-readable instructions stored on the computer-readable storage medium, the various methods described above may be performed.

[0197] In addition, the computer device may also include (but not limited to) a data bus, an input / output (I / O) bus, a display, and input / output devices (eg, keyboard, mouse, speaker, etc.), etc.

[0198] The processor may communicate with external devices via an I / O bus via a wired or wireless network.

[0199] In one embodiment, the at least one computer executable instruction may also be compiled into or constitute a software product / computer program product, wherein one or more computer executable instructions are executed by a processor to perform the various functions and / or method steps in the embodiments described in the present technology.

[0200] In the embodiments provided in the present disclosure, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0201] It should be noted that in the present disclosure, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element limited by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0202] Although the embodiments disclosed in the present disclosure are as above, the above contents are only embodiments adopted for facilitating the understanding of the present disclosure and are not intended to limit the present disclosure. Any technician in the technical field to which the present disclosure belongs can make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed in the present disclosure, but the scope of patent protection of the present disclosure shall still be subject to the scope defined in the attached claims.

Claims

1. A fault handling method based on human-computer dialogue, characterized in that: include: Using a preset input device to obtain user input data; Using a pre-trained intent recognition model to perform semantic recognition on the input data to obtain a semantic recognition result; Determining whether the semantic recognition result is a common problem; If the semantic recognition result is a common question, similarity matching is performed on the semantic recognition result using a pre-established common question answering library to obtain a matching result, and an answer result is output according to the matching result; If the semantic recognition result is not a common problem, query the semantic recognition result in a pre-built knowledge graph database to obtain a query result, and output a solution result according to the query result; Displaying the answer result to the user and obtaining the user's feedback on the answer result; If the feedback result is unsatisfactory, the user is prompted to re-input the question, and the process returns to the step of obtaining the user's input data by using the preset input device; If the feedback result is satisfactory, the confirmation process ends.

2. The method according to claim 1, characterized in that The method of obtaining the user's input data by using a preset input device includes: Obtaining an input instruction from a user, and determining whether the input instruction is voice data or text data; If the input instruction is voice data, the input instruction is converted into text data through voice recognition. If the input instruction is text data, segmenting the text data to obtain segmented text; The segmented text is subjected to part-of-speech tagging to obtain a tagged text, and named entity recognition is performed on the tagged text to obtain the input data.

3. The method according to claim 1, characterized in that After obtaining the user's input data by using the preset input device, the method further includes: Obtain multimodal fault data of users; Performing semantic analysis on text information contained in the multimodal fault data to obtain semantic features; Extracting features of images contained in the multimodal fault data to obtain image features; Using a pre-trained fault recognition model to perform fault recognition according to the semantic features and the image, and obtaining a fault recognition result; Generate a fault handling method according to the fault identification result using a pre-established fault database; The troubleshooting method is presented to the user.

4. The method according to claim 1, characterized in that: The using of the pre-trained intention recognition model to perform semantic recognition on the input data to obtain a semantic recognition result includes: Using a bag-of-words model to extract features from the input data to obtain a text vector; Using the intention recognition model to encode the text vector to obtain semantic information; Using the attention mechanism module in the intention recognition model to identify the importance of the semantic information, and obtain an importance recognition result; The intention recognition model is used to output a semantic recognition result according to the semantic information and the importance recognition result.

5. The method according to claim 1, characterized in that The determining whether the semantic recognition result is a common problem includes: Obtain the question type annotation contained in the semantic recognition result; Matching is performed in a preset frequently asked questions database according to the question type annotation to obtain a matching result; If the matching result is empty, it is determined that the semantic recognition result is not a common problem; If the matching result is not empty, the semantic recognition result is determined to be a common problem.

6. The method according to claim 1, characterized in that The method of performing similarity matching on the semantic recognition results using the pre-established FAQ library to obtain matching results includes: Performing word segmentation processing on the semantic recognition result to obtain a word segmentation recognition result; Obtaining a vector representation of each word in the word segmentation recognition result to obtain a word vector set; Calculate an average word vector based on the word vector set; Calculate the cosine similarity between the vector of each question text in the FAQ library and the average word vector; The question text with the largest cosine similarity in the FAQ database is selected as the matching result.

7. A fault handling device based on human-computer dialogue, characterized in that: include: A data acquisition module, used to acquire user input data using a preset input device; A semantic recognition module, used to perform semantic recognition on the input data using a pre-trained intent recognition model to obtain a semantic recognition result; A question determination module is used to determine whether the semantic recognition result is a common question. If the semantic recognition result is a common question, a pre-established common question answering library is used to perform similarity matching on the semantic recognition result to obtain a matching result, and an answer result is output according to the matching result. If the semantic recognition result is not a common question, a pre-established knowledge graph database is used to query the semantic recognition result to obtain a query result, and an answer result is output according to the query result. The data feedback module is used to display the answer result to the user and obtain the user's feedback on the answer result. If the feedback result is unsatisfactory, the user is prompted to re-enter the question and return to the step of obtaining the user's input data using the preset input device. If the feedback result is satisfactory, the confirmation process ends.

8. A computer device comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images