Intelligent methods, devices, equipment, and storage media for finding missing persons based on intent recognition

Abstract

This invention relates to artificial intelligence technology and discloses an intelligent person-finding method, apparatus, device, and storage medium based on intent recognition. The method includes: acquiring user input voice and converting it to text to obtain a text instruction; determining whether the string length of the text instruction is greater than a preset threshold; if it is greater, segmenting the text instruction using a word segmentation tool to obtain the target person's name; if it is less than or equal to the threshold, using a pre-trained binary classification network of a person-finding intent recognition model to perform person-finding intent recognition on the text instruction, obtaining an intent recognition result; and locating the target person's name based on the intent recognition result and the name confidence recognition network in the person-finding intent recognition model; and querying public information about the target person's name. This invention can improve the accuracy of intelligent person-finding through address book by accurately identifying user intent and precise person-finding instructions.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to an intelligent method, apparatus, device, and computer-readable storage medium for finding people based on intent recognition. Background Technology

[0002] With the development of speech recognition technology, voice commands have become a common control technology in artificial intelligence, such as controlling smart homes and interacting with robots through voice.

[0003] However, in the field of intelligent contact search, due to the possibility of deviations in individual words when translating user input voice, searching for keywords such as "find" or "find" can easily lead to situations where the search program is accidentally triggered when searching for files, or incorrect search results are obtained under the search program, such as "find **" becoming "Zhao **", failing to accurately obtain the result the user wants to find. Summary of the Invention

[0004] This invention provides an intelligent person-finding method, device, equipment, and storage medium based on intent recognition. Its main purpose is to improve the accuracy of intelligent person-finding in the address book by accurately identifying user intent and precise person-finding instructions.

[0005] To achieve the above objectives, the present invention provides an intelligent person-finding method based on intent recognition, comprising:

[0006] The system acquires the user's voice input and performs a voice-to-text conversion operation on the voice input to obtain text instructions.

[0007] Based on a preset intent recognition strategy, determine whether the string length of the text instruction is greater than a preset threshold;

[0008] When the length of the text instruction string is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the segmentation results.

[0009] When the length of the text instruction string is less than the preset threshold, the binary classification network of the pre-trained missing person intention recognition model is used to recognize the missing person intention of the text instruction, and the intention recognition result is obtained. Then, based on the intention recognition result and the name confidence recognition network in the missing person intention recognition model, the name of the text instruction is located to obtain the target name.

[0010] A contact list search is performed based on the target person's name to obtain the public information corresponding to that target person's name.

[0011] Optionally, before using the binary classification network of the pre-trained missing person intent recognition model to recognize the missing person intent in the text instruction, the method further includes:

[0012] Obtain a pre-built set of sentence samples and a pre-built model for recognizing the intent to find people;

[0013] One statement sample is extracted from the statement sample set in sequence, and the statement sample is subjected to binary semantic recognition using the missing person intent recognition model to obtain the intent type;

[0014] When the intent type is a preset non-person-finding intent, the output layer of the person-finding intent recognition model is used to output a preset blank name character as the predicted name;

[0015] When the intent type is a preset missing person intent, the pinyin characters corresponding to the sentence sample are obtained, and the missing person intent recognition model is used to evaluate the name confidence of the pinyin characters. Based on the confidence evaluation results, the word with the highest confidence score is selected as the predicted name.

[0016] Using a pre-constructed intention-name combination cross-entropy loss function, the loss value of the intention category and the predicted name is calculated to obtain the combination loss value. The model combination parameters are obtained by minimizing the combination loss value.

[0017] The model combination parameters are updated by using a forward feedback neural network to obtain an updated missing person intention recognition model.

[0018] Determine whether the combined loss value is less than a preset qualified threshold;

[0019] When the combined loss value is greater than the qualified threshold, return to the above steps of extracting one sentence sample from the sentence sample set in sequence, and iteratively update the updated missing person intention recognition model;

[0020] When the combined loss value is less than or equal to the qualified threshold, the finally optimized updated missing person intention recognition model is used as the trained missing person intention recognition model.

[0021] Optionally, the step of using a word segmentation tool to segment the text instruction and filtering the target person's name from the segmentation results includes:

[0022] The text instruction is segmented using a word segmentation tool, and the segmentation results are then tagged with part-of-speech tags and arranged in order to obtain a set of segmentation results.

[0023] The target names are obtained by filtering the word segmentation results set.

[0024] Optionally, the step of querying the address book based on the target person's name to obtain the public information corresponding to the target person's name includes:

[0025] Determine the source of the target person's name;

[0026] When the target person's name comes from the missing person intention recognition model, the target person's name is searched in the address book according to the preset query strategy to obtain the public information corresponding to the target person's name;

[0027] When the target person's name comes from the word segmentation tool, determine whether the target person's name is an empty set;

[0028] When the target person name obtained by the word segmentation tool is an empty set, a prompt message indicating that the intention is not to find a person is generated.

[0029] When the target person name obtained by the word segmentation tool is not an empty set, the target person name is subjected to a precise address book matching operation.

[0030] Determine whether the target name achieves an exact match;

[0031] When the target name achieves an exact match, the public information corresponding to the target name is output.

[0032] When the target name is not accurately matched, the text command of the missing person intention recognition model is used to identify the name and retrieve the public information corresponding to the target name.

[0033] Optionally, the step of performing a contact list query on the target name according to a preset query strategy to obtain the public information corresponding to the target name includes:

[0034] According to the preset query strategy, the target name is converted into pinyin to obtain the original pinyin, and the original pinyin is converted into front and back nasal sounds to obtain the converted pinyin.

[0035] Based on the original pinyin and the converted pinyin, a string-based precise query is performed on the pre-built address book to obtain a set of query persons, and the public information of each person's name in the set of query persons is displayed.

[0036] To address the aforementioned problems, the present invention also provides an intelligent person-finding device based on intent recognition, the device comprising:

[0037] The text acquisition module is used to acquire the user's voice input and perform voice-to-text conversion on the voice input to obtain text instructions;

[0038] The character length filtering module is used to determine whether the string length of the text instruction is greater than a preset threshold according to a preset intent recognition strategy. When the string length of the text instruction is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the word segmentation results.

[0039] The intent recognition name module is used to identify the intent of the text instruction when the string length of the text instruction is less than the preset threshold, by using the binary classification network of the pre-trained missing person intent recognition model to identify the intent of the text instruction, obtain the intent recognition result, and locate the name of the text instruction based on the intent recognition result and the name confidence recognition network in the missing person intent recognition model to obtain the target name.

[0040] The data query module is used to perform a contact list query based on the target person's name to obtain the public information corresponding to the target person's name.

[0041] Optionally, before using the binary classification network of the pre-trained missing person intent recognition model to recognize the missing person intent in the text instruction, the method further includes:

[0042] Obtain a pre-built set of sentence samples and a pre-built model for recognizing the intent to find people;

[0043] One statement sample is extracted from the statement sample set in sequence, and the statement sample is subjected to binary semantic recognition using the missing person intent recognition model to obtain the intent type;

[0044] When the intent type is a preset non-person-finding intent, the output layer of the person-finding intent recognition model is used to output a preset blank name character as the predicted name;

[0045] When the intent type is a preset missing person intent, the pinyin characters corresponding to the sentence sample are obtained, and the missing person intent recognition model is used to evaluate the name confidence of the pinyin characters. Based on the confidence evaluation results, the word with the highest confidence score is selected as the predicted name.

[0046] Using a pre-constructed intention-name combination cross-entropy loss function, the loss value of the intention category and the predicted name is calculated to obtain the combination loss value. The model combination parameters are obtained by minimizing the combination loss value.

[0047] The model combination parameters are updated by using a forward feedback neural network to obtain an updated missing person intention recognition model.

[0048] Determine whether the combined loss value is less than a preset qualified threshold;

[0049] When the combined loss value is greater than the qualified threshold, return to the above steps of extracting one sentence sample from the sentence sample set in sequence, and iteratively update the updated missing person intention recognition model;

[0050] When the combined loss value is less than or equal to the qualified threshold, the finally optimized updated missing person intention recognition model is used as the trained missing person intention recognition model.

[0051] Optionally, the step of using a word segmentation tool to segment the text instruction and filtering the target person's name from the segmentation results includes:

[0052] The text instruction is segmented using a word segmentation tool, and the segmentation results are then tagged with part-of-speech tags and arranged in order to obtain a set of segmentation results.

[0053] The target names are obtained by filtering the word segmentation results set.

[0054] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0055] At least one processor; and,

[0056] A memory communicatively connected to the at least one processor; wherein,

[0057] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the aforementioned intelligent person-finding method based on intent recognition.

[0058] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the aforementioned intelligent person-finding method based on intent recognition.

[0059] This invention acquires user input voice and converts it into text commands. Then, by limiting the string length, long text commands are segmented and queried to obtain the target name. Short text commands are identified using a pre-trained missing person intent recognition model. This missing person intent recognition model is a neural network model combining a binary classification network and a name confidence recognition network. The binary classification network determines whether the user's intent has entered the missing person search process, while the name confidence recognition network segments the statement by calculating confidence levels, increasing the accuracy and completeness of name recognition and thus improving the accuracy of intelligent missing person search within the address book. Furthermore, by querying long and short strings separately, the missing person intent recognition model avoids parsing long text commands, increasing the speed of the intelligent missing person search process. Therefore, this invention provides an intelligent missing person search method, device, equipment, and storage medium based on intent recognition, which can accurately identify missing person commands and improve the accuracy of intelligent missing person search within the address book by recognizing user intent. Attached Figure Description

[0060] Figure 1 This is a flowchart illustrating an intelligent person-finding method based on intent recognition provided in an embodiment of the present invention.

[0061] Figure 2 This is a detailed flowchart illustrating one step of an intelligent missing person search method based on intent recognition, provided in an embodiment of the present invention.

[0062] Figure 3 This is a detailed flowchart illustrating one step of an intelligent missing person search method based on intent recognition, provided in an embodiment of the present invention.

[0063] Figure 4 This is a detailed flowchart illustrating one step of an intelligent missing person search method based on intent recognition, provided in an embodiment of the present invention.

[0064] Figure 5 A functional block diagram of an intelligent person-finding device based on intent recognition provided in an embodiment of the present invention;

[0065] Figure 6 This is a schematic diagram of the structure of an electronic device that implements the intelligent person-finding method based on intent recognition, according to an embodiment of the present invention.

[0066] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0067] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0068] This application provides an intelligent missing person search method based on intent recognition. In this application, the executing entity of the intelligent missing person search method based on intent recognition includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the intelligent missing person search method based on intent recognition can be executed by software or hardware installed on a terminal device or a server device, and the software can be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0069] Reference Figure 1 The diagram shown is a flowchart illustrating an intelligent person-finding method based on intent recognition according to an embodiment of the present invention. In this embodiment, the intelligent person-finding method based on intent recognition includes:

[0070] S1. Obtain the user's voice input and perform a voice-to-text operation on the voice input to obtain text instructions.

[0071] In this embodiment of the invention, when a user uses the frequently-asked questions (FAQ) function, the user's voice is obtained as the input voice.

[0072] Then, the input speech is converted to text using voice-to-text technology to obtain text instructions.

[0073] S2. Based on the preset intent recognition strategy, determine whether the string length of the text instruction is greater than a preset threshold.

[0074] It should be understood that the longer the voice input, the higher the accuracy of the translation result. Therefore, the longer the text command string, the easier it is to identify the user's intent and the corresponding target name. Conversely, the shorter the string length, the more difficult it is. For example, "Find Zhao A at **Bank** Branch" can be identified with a high degree of confidence as "Zhao A" as the target name. However, if it is "Find b", it may become "Find Zhao b" during the speech-to-text process. Furthermore, it is impossible to confirm whether "b" is a person's name or a file name, nor can it be confirmed whether "Zhao b" or "b" is the accurate target name. Therefore, a more accurate identification method is needed to make the judgment.

[0075] Therefore, in this embodiment of the invention, the intent recognition strategy is as follows: query the string length of the text instruction, perform word segmentation query on long text instructions, and perform model query on short text instructions.

[0076] When the length of the text instruction string is greater than or equal to the preset threshold, then S3, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the segmentation results.

[0077] For details, please refer to the following: Figure 2 As shown in this embodiment of the invention, the step of using a word segmentation tool to segment the text instruction and filtering the target person's name from the segmentation results includes:

[0078] S31. Use a word segmentation tool to segment the text instruction, and perform part-of-speech tagging and sorting operations on the segmentation results to obtain a set of segmentation results;

[0079] S32. Filter the word segmentation result set to obtain the target person's name.

[0080] In this embodiment of the invention, the word segmentation tool or Chinese word segmentation system can be used to perform word segmentation, part-of-speech tagging and sorting operations on the word segmentation tool to obtain a word segmentation result set. Then, by filtering each noun in the word segmentation result set, and then filtering person names from each noun, the target person name is obtained. The target person name obtained by word segmentation filtering is a dataset, which may be multiple or may be empty.

[0081] When the length of the text instruction string is less than the preset threshold, S4, the text instruction is used to identify the intention of finding a person by using the binary classification network of the pre-trained missing person intention recognition model to obtain the intention recognition result, and the text instruction is used to locate the name based on the intention recognition result and the name confidence recognition network in the missing person intention recognition model to obtain the target name.

[0082] In this embodiment of the invention, the missing person intent recognition model is a neural network model based on a binary classification network and a name confidence recognition network. The binary classification network is used to identify whether a user's action intent is to find a missing person through semantic recognition, while the name confidence recognition network extracts two to four characters to obtain words, examines the confidence of each word as a name, and selects the name with the highest confidence as the target name.

[0083] In this embodiment of the invention, due to the special nature of personal names, such as "National Day" which may be a festival or a personal name, it is necessary to first identify the user's intent and determine whether the user wants to search for a person. Specifically, this embodiment of the invention quantizes and encodes the text instruction to obtain quantized text. Then, a feature extraction network performs convolution, pooling, and flattening operations on the quantized text to obtain a set of feature sequences. Each feature sequence is then imported into the binary classification network for a fully connected operation to obtain combined features. The category of the combined features is identified to determine whether it is a preset type of intent such as searching for a person or address. Intent classification is performed to determine the user's intent type. When the intent type is not searching for a person, a preset personal name character is output as the predicted personal name, such as "Null". When the intent type is searching for a person, the text instruction is then converted into pinyin. The personal name confidence recognition network extracts the pinyin result and performs personal name confidence analysis to identify the target personal name. The personal name confidence includes analysis of user behavior habits, tone of voice, and pauses.

[0084] In detail, during the name confidence analysis process, the user's input text commands can be traversed and extracted using preset 2-byte, 3-byte, and 4-byte frames. For example, "I want to find the National Day Manager of ** Bank", we get [2 bytes: ... "of Guo", "National Day", "Qingjing" ...; 3 bytes: "of the bank", "of National Day" ...; 4 bytes ...]. Then, we query the preset list of surnames by name confidence and infer the part of speech of each segment by the part of speech before and after each segment. We then perform a weighted calculation on each segment to obtain the confidence of each segment belonging to a person's name.

[0085] For details, please refer to the following: Figure 3 As shown in this embodiment of the invention, before using the binary classification network of the pre-trained missing person intention recognition model to recognize the missing person intention in the text instruction, the method further includes:

[0086] S401. Obtain a pre-built set of sentence samples and a pre-built model for recognizing the intent to find people;

[0087] S402. Extract one statement sample from the statement sample set in sequence, and use the missing person intention recognition model to perform binary classification semantic recognition on the statement sample to obtain the intention type;

[0088] S403. Determine whether the intent type is a search for a person;

[0089] When the intent type is a preset non-person-finding intent, S404, the output layer of the person-finding intent recognition model is used to output a preset blank name character as the predicted name;

[0090] When the intent type is a preset missing person intent, S405, obtain the pinyin characters corresponding to the sentence sample, and use the missing person intent recognition model to evaluate the name confidence of the pinyin characters, and select the word with the highest confidence score as the predicted name based on the confidence evaluation result;

[0091] S406. Using the pre-constructed intention-name combination cross-entropy loss function, calculate the loss value for the intention category and the predicted name to obtain the combination loss value, and minimize the combination loss value to obtain the model combination parameters when the combination loss value is minimized.

[0092] S407. The model combination parameters are updated in reverse using a forward feedback neural network to obtain an updated missing person intention recognition model.

[0093] S408. Determine whether the combined loss value is less than a preset qualified threshold.

[0094] When the combined loss value is greater than the qualified threshold, return to step S402 above and iteratively update the updated missing person intention recognition model;

[0095] When the combined loss value is less than or equal to the qualified threshold, S409, the finally optimized updated missing person intention recognition model is used as the trained missing person intention recognition model.

[0096] In this embodiment of the invention, the set of sentence samples includes multiple types of phrase texts, such as text expressing the intention to find a person, text expressing the intention to search for a file, and single names or filenames. This embodiment of the invention performs forward propagation calculations on the sentence samples using a pre-built person-finding intention recognition model. Then, using a feedback neural network, through the cross-entropy algorithm and gradient descent method, it calculates a combined loss value by combining the model's predicted name and intention type with the actual results corresponding to the sentence samples. The model combination parameters corresponding to minimizing the combined loss value are then fed back in reverse to update the person-finding intention recognition model, resulting in an updated person-finding intention recognition model. The cross-entropy algorithm is obtained by weighting the cross-entropy loss function of the binary classification network and the cross-entropy loss function of the name confidence recognition network. The cross-entropy loss functions of the binary classification network and the name confidence recognition network are common loss functions in softmax, which will not be elaborated here.

[0097] In this invention, the progress of training is judged by monitoring the combined loss value during the training process. When the combined loss value is less than a preset qualified threshold, such as 0.1, it indicates that the error of the prediction result of the updated missing person intention recognition model is small, and the training can be ended, resulting in a completed missing person intention recognition model.

[0098] S5. Perform a contact list query based on the target person's name to obtain the public information corresponding to the target person's name.

[0099] For details, please refer to the following: Figure 4 As shown in the embodiment of the present invention, step S5 includes:

[0100] S51. Determine the source of the target person's name;

[0101] When the target person's name comes from the missing person intention recognition model, S52, the target person's name is searched in the address book according to the preset query strategy to obtain the public information corresponding to the target person's name;

[0102] When the target person name comes from the word segmentation tool, S53: Determine whether the target person name is an empty set;

[0103] When the target person name obtained by the word segmentation tool is an empty set, S54, generate a prompt message that is not intended to find a person;

[0104] When the target person name obtained by the word segmentation tool is not an empty set, S55, perform a precise address book matching operation on the target person name;

[0105] S56. Determine whether the target name achieves an exact match;

[0106] When the target name achieves an exact match, S57, output the public information corresponding to the target name;

[0107] When the target name is not accurately matched, S58, the text command of the missing person intention recognition model is used to identify the name and query the public information corresponding to the target name.

[0108] In this embodiment of the invention, the process of querying the address book based on the target name can be divided into two categories: one is the target name obtained by the word segmentation model, and the other is the target name obtained by the missing person intent recognition model. The word segmentation model yields a large number of target names with poor name completeness, while the missing person intent recognition model provides more accurate target names. Therefore, different methods are needed for querying the target name. In this embodiment of the invention, the address book query is performed using exact matching, meaning that every character in the string must correspond exactly to the target name for a match to be considered complete; no extra characters or missing characters are allowed.

[0109] Regarding the target name from which the model originates, in this embodiment of the invention, the step of performing a contact list query on the target name according to a preset query strategy to obtain the public information corresponding to the target name includes:

[0110] According to the preset query strategy, the target name is converted into pinyin to obtain the original pinyin, and the original pinyin is converted into front and back nasal sounds to obtain the converted pinyin.

[0111] Based on the original pinyin and the converted pinyin, a string-based precise query is performed on the pre-built address book to obtain a set of query persons, and the public information of each person's name in the set of query persons is displayed.

[0112] In this embodiment of the invention, when the pinyin of the target person's name is "jinchen", the target person can be extracted from the address book by means of string precision, such as "jinchen", "zhaojinchen", "wangjinchen", etc., which contain the string "jinchen", and the public information of the target person can be displayed.

[0113] In addition, since searching for names is similar to searching for organizations or departments, it can be applied to various query services in user voice interaction scenarios, in addition to searching for names.

[0114] This invention acquires user input voice and converts it into text commands. Then, by limiting the string length, long text commands are segmented and queried to obtain the target name. Short text commands are identified using a pre-trained missing person intent recognition model. This missing person intent recognition model is a neural network model combining a binary classification network and a name confidence recognition network. The binary classification network determines whether the user's intent has entered the missing person search process, while the name confidence recognition network segments the sentence by calculating confidence, increasing the accuracy and completeness of name recognition and thus improving the accuracy of intelligent missing person search in the address book. Furthermore, by querying long and short strings separately, the missing person intent recognition model avoids parsing long text commands, increasing the speed of the intelligent missing person search process. Therefore, this invention provides an intelligent missing person search method based on intent recognition, which can accurately identify missing person commands by recognizing user intent, thereby improving the accuracy of intelligent missing person search in the address book.

[0115] like Figure 5 The diagram shown is a functional block diagram of an intelligent person-finding device based on intent recognition provided in an embodiment of the present invention.

[0116] The intelligent person-finding device 100 based on intent recognition described in this invention can be installed in an electronic device. Depending on the functions implemented, the intelligent person-finding device 100 based on intent recognition may include a text acquisition module 101, a character length filtering module 102, an intent-based name recognition module 103, and a data query module 104. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0117] In this embodiment, the functions of each module / unit are as follows:

[0118] The text acquisition module 101 is used to acquire the user's input voice and perform a voice-to-text operation on the input voice to obtain text instructions;

[0119] The character length filtering module 102 is used to determine whether the string length of the text instruction is greater than a preset threshold according to a preset intent recognition strategy, and when the string length of the text instruction is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the word segmentation results.

[0120] The intent recognition name module 103 is used to perform search intent recognition on the text instruction using the binary classification network of the pre-trained search intent recognition model when the string length of the text instruction is less than the preset threshold, to obtain the intent recognition result, and to locate the name of the text instruction based on the intent recognition result and the name confidence recognition network in the search intent recognition model, so as to obtain the target name.

[0121] The data query module 104 is used to perform a contact list query based on the target person's name to obtain the public information corresponding to the target person's name.

[0122] In detail, the modules described in the intent-based intelligent missing person search device 100 in this application embodiment employ the same methods as described above during use. Figures 1 to 4 The method uses the same technical means as the intelligent missing person search method based on intent recognition described in the article, and can produce the same technical effect, so it will not be repeated here.

[0123] like Figure 6 The diagram shown is a structural schematic of an electronic device 1 that implements an intelligent person-finding method based on intent recognition, according to an embodiment of the present invention.

[0124] The electronic device 1 may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as an intelligent person-finding program based on intent recognition.

[0125] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device 1, connecting various components of the electronic device via various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing an intelligent missing person search program based on intent recognition) and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0126] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of an intent-based intelligent missing person search program, but also to temporarily store data that has been output or will be output.

[0127] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0128] The communication interface 13 is used for communication between the electronic device 1 and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), or, optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0129] Figure 6 Only electronic devices with components are shown; those skilled in the art will understand that... Figure 6 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0130] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0131] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0132] The intelligent missing person search program based on intent recognition stored in the memory 11 of the electronic device 1 is a combination of multiple instructions. When run in the processor 10, it can achieve the following:

[0133] The system acquires the user's voice input and performs a voice-to-text conversion operation on the voice input to obtain text instructions.

[0134] Based on a preset intent recognition strategy, determine whether the string length of the text instruction is greater than a preset threshold;

[0135] When the length of the text instruction string is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the segmentation results.

[0136] When the length of the text instruction string is less than the preset threshold, the binary classification network of the pre-trained missing person intention recognition model is used to recognize the missing person intention of the text instruction, and the intention recognition result is obtained. Then, based on the intention recognition result and the name confidence recognition network in the missing person intention recognition model, the name of the text instruction is located to obtain the target name.

[0137] A contact list search is performed based on the target person's name to obtain the public information corresponding to that target person's name.

[0138] Specifically, the specific implementation method of the processor 10 for the above instructions can be referred to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, and will not be repeated here.

[0139] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0140] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following:

[0141] The system acquires the user's voice input and performs a voice-to-text conversion operation on the voice input to obtain text instructions.

[0142] Based on a preset intent recognition strategy, determine whether the string length of the text instruction is greater than a preset threshold;

[0143] When the length of the text instruction string is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the segmentation results.

[0144] When the length of the text instruction string is less than the preset threshold, the binary classification network of the pre-trained missing person intention recognition model is used to recognize the missing person intention of the text instruction, and the intention recognition result is obtained. Then, based on the intention recognition result and the name confidence recognition network in the missing person intention recognition model, the name of the text instruction is located to obtain the target name.

[0145] A contact list search is performed based on the target person's name to obtain the public information corresponding to that target person's name.

[0146] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0147] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0148] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0149] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0150] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0151] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0152] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0153] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0154] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for intelligent missing person search based on intent recognition, characterized in that, The method includes: The system acquires the user's voice input and performs a voice-to-text conversion operation on the voice input to obtain text instructions. Based on a preset intent recognition strategy, determine whether the string length of the text instruction is greater than a preset threshold; When the length of the text instruction string is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the segmentation results. When the length of the text instruction string is less than the preset threshold, the binary classification network of the pre-trained missing person intention recognition model is used to recognize the missing person intention of the text instruction, and the intention recognition result is obtained. Then, based on the intention recognition result and the name confidence recognition network in the missing person intention recognition model, the name of the text instruction is located to obtain the target name. A contact list search is performed based on the target person's name to obtain the public information corresponding to the target person's name; The training of the missing person intention recognition model includes: Obtain a pre-built set of sentence samples and a pre-built model for recognizing the intent to find people; One statement sample is extracted from the statement sample set in sequence, and the pre-built missing person intent recognition model is used to perform binary semantic recognition on the statement sample to obtain the intent type; When the intent type is a preset non-person-finding intent, the output layer of the person-finding intent recognition model is used to output a preset blank name character as the predicted name; when the intent type is a preset person-finding intent, the pinyin characters corresponding to the sentence sample are obtained, and the person-finding intent recognition model is used to score the name confidence of the pinyin characters, and the word with the highest confidence score is selected as the predicted name. The combined loss value is obtained by calculating the loss value of the intent type and the predicted person name using a pre-constructed intent-name combination cross-entropy loss function. The combined loss value is minimized to obtain the model combination parameters when the combined loss value is minimized. The model combination parameters are then updated by using a forward feedback neural network to perform network inverse update to obtain an updated person search intent recognition model. Determine whether the combined loss value is less than a preset qualified threshold. If the combined loss value is greater than the qualified threshold, return to the step of extracting one sentence sample from the sentence sample set in sequence, and iteratively update the updated missing person intention recognition model. When the combined loss value is less than or equal to the qualified threshold, the updated missing person intention recognition model is taken as the completed missing person intention recognition model.

2. The intelligent person-finding method based on intent recognition as described in claim 1, characterized in that, The step of using a word segmentation tool to segment the text instruction and filtering the target person's name from the segmentation results includes: The text instruction is segmented using a word segmentation tool, and the segmentation results are then tagged with part-of-speech tags and arranged in order to obtain a set of segmentation results. The target names are obtained by filtering the word segmentation results set.

3. The intelligent person-finding method based on intent recognition as described in claim 1, characterized in that, The step of querying the address book based on the target person's name to obtain the public information corresponding to the target person's name includes: Determine the source of the target person's name; When the target person's name comes from the missing person intention recognition model, the target person's name is searched in the address book according to the preset query strategy to obtain the public information corresponding to the target person's name; When the target person's name comes from the word segmentation tool, determine whether the target person's name is an empty set; When the target person name obtained by the word segmentation tool is an empty set, a prompt message indicating that the intention is not to find a person is generated. When the target person name obtained by the word segmentation tool is not an empty set, the target person name is subjected to a precise address book matching operation. Determine whether the target name achieves an exact match; When the target name achieves an exact match, the public information corresponding to the target name is output. When the target name is not accurately matched, the text command of the missing person intention recognition model is used to identify the name and retrieve the public information corresponding to the target name.

4. The intelligent person-finding method based on intent recognition as described in claim 3, characterized in that, The step of performing a contact list query on the target name according to a preset query strategy to obtain the public information corresponding to the target name includes: According to the preset query strategy, the target name is converted into pinyin to obtain the original pinyin, and the original pinyin is converted into front and back nasal sounds to obtain the converted pinyin. Based on the original pinyin and the converted pinyin, a string-based precise query is performed on the pre-built address book to obtain a set of query persons, and the public information of each person's name in the set of query persons is displayed.

5. An intelligent person-finding device based on intent recognition, characterized in that, The device includes: The text acquisition module is used to acquire the user's voice input and perform voice-to-text conversion on the voice input to obtain text instructions; The character length filtering module is used to determine whether the string length of the text instruction is greater than a preset threshold according to a preset intent recognition strategy. When the string length of the text instruction is greater than or equal to the preset threshold, the text instruction is segmented using a word segmentation tool, and the target person's name is obtained from the word segmentation results. The intent recognition name module is used to identify the intent of the text instruction when the string length of the text instruction is less than the preset threshold, by using the binary classification network of the pre-trained missing person intent recognition model to identify the intent of the text instruction, obtain the intent recognition result, and locate the name of the text instruction based on the intent recognition result and the name confidence recognition network in the missing person intent recognition model to obtain the target name. The data query module is used to perform a contact list query based on the target person's name to obtain the public information corresponding to the target person's name; The training of the missing person intention recognition model includes: Obtain a pre-built set of sentence samples and a pre-built model for recognizing the intent to find people; One statement sample is extracted from the statement sample set in sequence, and the pre-built missing person intent recognition model is used to perform binary semantic recognition on the statement sample to obtain the intent type; When the intent type is a preset non-person-finding intent, the output layer of the person-finding intent recognition model is used to output a preset blank name character as the predicted name; when the intent type is a preset person-finding intent, the pinyin characters corresponding to the sentence sample are obtained, and the person-finding intent recognition model is used to score the name confidence of the pinyin characters, and the word with the highest confidence score is selected as the predicted name. The combined loss value is obtained by calculating the loss value of the intent type and the predicted person name using a pre-constructed intent-name combination cross-entropy loss function. The combined loss value is minimized to obtain the model combination parameters when the combined loss value is minimized. The model combination parameters are then updated by using a forward feedback neural network to perform network inverse update to obtain an updated person search intent recognition model. Determine whether the combined loss value is less than a preset qualified threshold. If the combined loss value is greater than the qualified threshold, return to the step of extracting one sentence sample from the sentence sample set in sequence, and iteratively update the updated missing person intention recognition model. When the combined loss value is less than or equal to the qualified threshold, the updated missing person intention recognition model is taken as the completed missing person intention recognition model.

6. The intelligent person-finding device based on intent recognition as described in claim 5, characterized in that, The step of using a word segmentation tool to segment the text instruction and filtering the target person's name from the segmentation results includes: The text instruction is segmented using a word segmentation tool, and the segmentation results are then tagged with part-of-speech tags and arranged in order to obtain a set of segmentation results. The target names are obtained by filtering the word segmentation results set.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the intelligent person-finding method based on intent recognition as described in any one of claims 1 to 4.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the intelligent person-finding method based on intent recognition as described in any one of claims 1 to 4.

Images