Text processing method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a text processing method and device, electronic equipment, a storage medium and a program product. The method comprises the steps that a first text of a first object is extracted from a dialogue text, the first text comprises a keyword, a second text of a second object is extracted from the dialogue text, dialogue time of the second text is earlier than dialogue time of the first text, and if the second text does not comprise a text with the same meaning as the keyword, the first text is determined to be an abnormal text. Through the method, the accuracy of determining the abnormal text can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a text processing method, device, electronic device, storage medium and program product. Background Art

[0002] Dialogue text refers to the text generated when two objects use language to talk. In the related art, in order to identify whether the dialogue text includes abnormal text, a text library corresponding to the abnormal text can be established. Then, after obtaining the dialogue text, it can be identified whether there is text belonging to the text library in the dialogue text. If there is text belonging to the text library in the dialogue text, the text belonging to the text library is determined to be abnormal text. However, due to the diversity of dialogue texts, different vocabulary usage habits of different objects, etc., the text library cannot contain all abnormal texts, which leads to low accuracy in identifying whether there is abnormal text in the dialogue text. Summary of the invention

[0003] The embodiments of the present application provide a text processing method, device, electronic device, storage medium and program product, which can improve the accuracy of determining abnormal text.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] The present application provides a text processing method, the method comprising:

[0006] Extracting a first text of a first object from the conversation text, wherein the first text contains a keyword;

[0007] Extracting a second text of a second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text;

[0008] If the second text does not include text having the same meaning as the keyword, the first text is determined to be an abnormal text.

[0009] The present application provides a text processing device, the device comprising:

[0010] An extraction module, configured to extract a first text of a first object from the conversation text, wherein the first text contains keywords;

[0011] The extraction module is further used to extract a second text of a second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text;

[0012] A determination module is configured to determine that the first text is an abnormal text if the second text does not include any text having the same meaning as the keyword.

[0013] An embodiment of the present application provides an electronic device, the electronic device comprising:

[0014] A memory for storing computer executable instructions or computer programs;

[0015] The processor is used to implement the text processing method provided in the embodiment of the present application when executing the computer executable instructions or computer programs stored in the memory.

[0016] An embodiment of the present application provides a computer-readable storage medium storing a computer program or computer-executable instructions for implementing the text processing method provided in the embodiment of the present application when executed by a processor.

[0017] An embodiment of the present application provides a computer program product, including a computer program or computer executable instructions. When the computer program or computer executable instructions are executed by a processor, the text processing method provided by the embodiment of the present application is implemented.

[0018] The embodiments of the present application have the following beneficial effects:

[0019] In the text processing method provided in the embodiment of the present application, a first text of a first object can be extracted from a conversation text, the first text contains keywords, and a second text of a second object can be extracted from the conversation text, the conversation time of the second text is earlier than the conversation time of the first text. If the second text does not include text with the same meaning as the keyword, the first text is determined to be an abnormal text.

[0020] In the process of determining abnormal text, the present application does not need to rely on a text library for various dialogue texts, and can determine abnormal text based only on the dialogue text. When the second text of the second object does not include text with the same meaning as the keyword, the first text can be determined as an abnormal text. Even if different objects have different vocabulary usage habits, as long as the meaning is the same as the keyword, the abnormal text can be determined. In other words, even if there is diversity in the dialogue texts and different objects have different vocabulary usage habits, the abnormal text can be determined, thereby improving the accuracy of determining the abnormal text. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a structural diagram of a text processing system provided in an embodiment of the present application;

[0022] Figure 2 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0023] Figure 3 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 1 ;

[0024] Figure 4 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 2 ;

[0025] Figure 5 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 3 . DETAILED DESCRIPTION

[0026] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0027] In the following description, reference is made to “some embodiments”, which describe a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0028] In the following description, the terms "first\second\third" involved are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0029] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0030] Unless otherwise defined, all technical and scientific terms used in the embodiments of the present application have the same meanings as those commonly understood by those skilled in the art. The terms used in the embodiments of the present application are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0031] The relevant data collection and processing in the embodiments of this application should be strictly in accordance with the requirements of relevant laws and regulations when applied in examples, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing within the scope of authorization of laws and regulations and the personal information subject.

[0032] During the research process, the inventor discovered that the related technology has the following technical problems, which are described in detail below.

[0033] 1. In the related art, in order to identify whether there is abnormal text in the dialogue text, a text library corresponding to the abnormal text can be established. Then, after obtaining the dialogue text, it can be identified whether there is text belonging to the text library in the dialogue text. If there is text belonging to the text library in the dialogue text, the text belonging to the text library is determined to be abnormal text. However, the period of building the text library is long, and due to the diversity of dialogue texts and the vocabulary usage habits of different objects, the text library cannot contain all abnormal texts. In other words, the text library cannot cover all abnormal texts, resulting in low accuracy in determining abnormal texts.

[0034] Embodiments of the present application provide a text processing method, a text processing device, an electronic device, a computer-readable storage medium, and a computer program product, which can improve the accuracy of determining abnormal text.

[0035] See also Figure 1 , Figure 1 is a structural diagram of a text processing system provided in an embodiment of the present application, Figure 1 The text processing system 100 shown is used to support a text processing application. The terminal 400 is connected to the server 200 via the network 300. The network 300 can be a wide area network or a local area network, or a combination of the two.

[0036] In response to the text processing instruction triggered by the graphical interface 410 - 1 of the terminal 400 , the terminal 400 may send a text processing request to the server 200 . After receiving the text processing request, the server 200 may obtain the conversation text of the conversation between the first object and the second object.

[0037] After acquiring the dialogue text of the first object and the second object, the server 200 can extract the first text of the first object from the dialogue text, and the first text contains the keyword. The server 200 can also extract the second text of the second object from the dialogue text, and the dialogue time of the second text is earlier than the dialogue time of the first text. If the second text does not include text with the same meaning as the keyword, the first text is determined to be an abnormal text.

[0038] In some embodiments, in order to facilitate the user to know the abnormal text, the server 200 can send the abnormal text to the terminal 400, so that the terminal 400 can display the abnormal text on the graphical interface 410-1. In this way, the accuracy of determining the abnormal text can be improved.

[0039] In some embodiments, the server 200 may also determine the number of abnormal texts, and generate an alarm message when the number is greater than an abnormal number threshold, and then send the alarm message to the terminal 400, so that the terminal 400 can display the alarm message on the graphical interface 410-1. In this way, it is possible to implement an alarm for abnormal texts, so that the first object can reduce the number of abnormal texts during the process of communicating with other objects (including the second object), thereby improving the service quality of the first object providing services to other objects.

[0040] The electronic device provided in the embodiment of the present application is introduced below. The electronic device that implements the text processing method provided in the embodiment of the present application can be a terminal, a server, or a combination of the two. The terminal can be various types of terminals such as a laptop computer, a tablet computer, a desktop computer, a set-top box, a smart phone, a smart speaker, a smart watch, a smart TV, and a car terminal.

[0041] The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, 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 communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. The terminal and the server can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiments of the present application.

[0042] See also Figure 2 , Figure 2 is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application, Figure 2 The electronic device shown includes: at least one processor 410, a memory 450, at least one network interface 420 and a user interface 430. The various components in the electronic device are coupled together via a bus system 440. It is understood that the bus system 440 is used to achieve connection and communication between these components. In addition to the data bus, the bus system 440 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, the bus system 440 is not shown in FIG. Figure 2 Various buses are labeled as bus system 440 .

[0043] The processor 410 can be an integrated circuit chip with signal processing capabilities, such as a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc., where the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0044] The user interface 430 includes one or more output devices 431 that enable presentation of media content, including one or more speakers and / or one or more visual display screens. The user interface 430 also includes one or more input devices 432, including user interface components that facilitate user input, such as a keyboard, mouse, microphone, touch screen display, camera, other input buttons and controls.

[0045] The memory 450 may be removable, non-removable, or a combination thereof. Exemplary hardware devices include solid-state memory, hard disk drives, optical disk drives, etc. The memory 450 may optionally include one or more storage devices that are physically remote from the processor 410.

[0046] The memory 450 includes a volatile memory or a nonvolatile memory, and may also include both volatile and nonvolatile memories. The nonvolatile memory may be a read-only memory (ROM), and the volatile memory may be a random access memory (RAM). The memory 450 described in the embodiments of the present application is intended to include any suitable type of memory.

[0047] In some embodiments, memory 450 can store data to support various operations, examples of which include programs, modules, and data structures, or a subset or superset thereof, as exemplarily described below.

[0048] Operating system 451, including system programs for processing various basic system services and performing hardware-related tasks, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks;

[0049] A network communication module 452, used to reach other electronic devices via one or more (wired or wireless) network interfaces 420, exemplary network interfaces 420 include: Bluetooth, Wireless Compatibility Certification (WiFi), and Universal Serial Bus (USB), etc.;

[0050] a presentation module 453 for enabling presentation of information via one or more output devices 431 (e.g., display screen, speaker, etc.) associated with the user interface 430 (e.g., a user interface for operating peripherals and displaying content and information);

[0051] The input processing module 454 is used to detect one or more user inputs or interactions from one of the one or more input devices 432 and translate the detected inputs or interactions.

[0052] In some embodiments, the text processing device provided in the embodiments of the present application can be implemented in software. Figure 2 The text processing device 455 stored in the memory 450 is shown, which can be software in the form of a program and a plug-in, etc., including the following software modules: an extraction module 4551 and a determination module 4552. These modules are logical, so they can be arbitrarily combined or further split according to the functions implemented. The functions of each module will be explained below.

[0053] In other embodiments, the text processing device provided in the embodiments of the present application can be implemented in hardware. As an example, the text processing device provided in the embodiments of the present application can be a processor in the form of a hardware decoding processor, which is programmed to execute the text processing method provided in the embodiments of the present application. For example, the processor in the form of a hardware decoding processor can adopt one or more application-specific integrated circuits (Application Specific Integrated Circuit, ASIC), digital signal processors (Digital Signal Processor, DSP), programmable logic devices (Programmable Logic Device, PLD), complex programmable logic devices (Complex Programmable Logic Device, CPLD), field programmable gate arrays (Field-Programmable Gate Array, FPGA) or other electronic components.

[0054] In some embodiments, the terminal or server can implement the text processing method provided by the embodiment of the present application by running various computer executable instructions or computer programs. For example, computer executable instructions can be commands, machine instructions or software instructions at the microprogram level. The computer program can be a native program or software module in the operating system; it can be a local (Native) application (APPlication, APP), that is, a program that needs to be installed in the operating system to run, such as an APP or instant messaging APP in the financial field; it can also be a small program that can be embedded in any APP, that is, a program that can be run only by downloading it to a browser environment. In short, the above-mentioned computer executable instructions can be instructions in any form, and the above-mentioned computer program can be an application, module or plug-in in any form.

[0055] The text processing method provided by the embodiment of the present application is described below. As mentioned above, the electronic device implementing the text processing method of the embodiment of the present application can be a terminal, a server, or a combination of the two. Figure 3 , Figure 3 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 1 , the following combination Figure 3 The steps shown, taking the execution subject as a server as an example, illustrate the text processing method provided in the embodiment of the present application.

[0056] In step 101, a first text of a first object is extracted from a conversation text.

[0057] In actual application, when text processing is required, in response to a text processing instruction triggered on a display interface of a server, or in response to a text processing request received by a server, a conversation text of a first object and a second object can be obtained, and then the first text of the first object can be extracted from the conversation text of the first object and the second object. The first text contains keywords.

[0058] In some embodiments, the first object is an agent object, which can also be called a customer service object. The agent is a customer service object that provides services to customers. The second object is a customer object, which can also be called a user object. Customers are used to receive services. The conversation text is the text of the conversation between the first object and the second object during the process of the first object providing services to the second object.

[0059] To facilitate understanding of the dialogue text, the following is an example of the dialogue text. For example, for a telephone communication scenario involving the financial field, the dialogue text may be the text of the dialogue between the first and second objects during the process of the first object recommending financial products in the financial field to the second object. The dialogue text may also be the text of the dialogue between the first and second objects during the process of the first object returning a visit or reminding the second object of the financial products held by the second object. The dialogue text may also be the text of the dialogue between the first and second objects during the process of the second object asking the first object for information about financial products, etc. The specific text may be determined in combination with the actual situation.

[0060] In some embodiments, the first object may be the object conducting the interview (ie, the interviewee), the second object may be the object being interviewed (ie, the interviewee), and the conversation text may be the corresponding text during the process of the first object interviewing the second object.

[0061] The following describes how to obtain the text of the conversation between the first object and the second object. In some embodiments, the video and / or audio of the conversation between the first object and the second object may be obtained. For the video of the conversation between the first object and the second object, the audio may be extracted from the video of the conversation between the first object and the second object, that is, the audio of the conversation between the first object and the second object may be obtained.

[0062] For the audio of the conversation between the first object and the second object, the automatic speech recognition (ASR) technology can be used to convert the audio of the conversation between the first object and the second object into text, so as to obtain the conversation text of the conversation between the first object and the second object. In this case, the text of the first object and the text of the second object in the conversation text are sorted according to the conversation time sequence, and the conversation time is the start time when the first object or the second object says the words corresponding to the text.

[0063] There are multiple texts of the first object, and one text of the first object is a sentence spoken by the first object. Different texts of the first objects correspond to different conversation times. Correspondingly, there are multiple texts of the second object, and one text of the second object is a sentence spoken by the second object. Different texts of the first objects correspond to different conversation times. The conversation times corresponding to the texts of the first object and the second object are different.

[0064] For example, the conversation text includes the text corresponding to the words spoken by the first object at conversation time 1, the text corresponding to the words spoken by the second object at conversation time 2, the text corresponding to the words spoken by the first object at conversation time 3, the text corresponding to the words spoken by the second object at conversation time 4, etc., wherein conversation time 1 is earlier than conversation time 2, conversation time 2 is earlier than conversation time 3, conversation time 3 is earlier than conversation time 4, and conversation time 1-conversation time 4 are all the starting times of the words corresponding to the texts.

[0065] In some embodiments, the first object and the second object can have a conversation in a conversation window, and the text of the conversation between the first object and the second object can be used as the conversation text. The text of the first object and the text of the second object in the conversation text are arranged in order of conversation time, and the conversation time is the time when the first object or the second object sends the corresponding text.

[0066] The following introduces a method for extracting the first text of the first object. The first method for extracting the first text is to identify keywords from the conversation text of the first object and the second object after obtaining the conversation text of the first object and the second object to obtain a text including the keywords.

[0067] The keywords may be set according to actual usage requirements. For example, in order to improve the security of personal information of the second object or other objects, the keywords may be words corresponding to user identity information, words corresponding to time, words corresponding to financial product information, etc.

[0068] For another example, in a scenario of a conversation in a conversation window, that is, a scenario of an online conversation between a first object and a second object, in order to improve the network environment, the keywords may be words that cause discomfort to the objects.

[0069] After obtaining the text including the keyword, it can be determined whether the text including the keyword is the text of the first object. If the text including the keyword is not the text of the first object, that is, the text including the keyword is the text of the second object, the text including the keyword may not be extracted.

[0070] If the text including the keyword is the text of the first object, the text including the keyword is determined as the first text, so that the first text of the first object can be extracted from the dialogue text of the first object and the second object, and the first text includes the keyword.

[0071] A second extraction method for the first text can be to split the object text of the conversation between the first object and the second object according to the first object and the second object, thereby obtaining the text of the first object and the text of the second object, wherein the text of the first object and the text of the second object are sorted according to the conversation time.

[0072] The text of the first object includes texts of different conversation times. For each text of the conversation time, it is identified whether the text of the conversation time includes a keyword. If the text of the conversation time includes the keyword, the text of the conversation time can be extracted from the conversation text, and the text of the conversation time is used as the first text, and the first text includes the keyword. If the text of the conversation time does not include the keyword, the text of the conversation time may not be processed.

[0073] In the present application, the number of first texts may be one or more, and one first text includes one or more keywords. The multiple keywords corresponding to one first text may be the same or different, which may be determined according to actual conditions. In this way, the first text of the first object can be accurately obtained, and the first text is used to subsequently determine the abnormal text, so as to improve the accuracy of determining the abnormal text.

[0074] In step 102, a second text of a second object is extracted from the conversation text.

[0075] In some embodiments, step 102 can be implemented in the following manner: for the first extraction method of the first text and the second extraction method of the first text, after determining the first text, based on the conversation time of the first text, the text of the second object whose conversation time is before the conversation time of the first text and whose conversation time has a difference with the conversation time of the first text less than a first threshold value can be taken as the second text, thereby achieving the extraction of the second text of the second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text.

[0076] The first threshold is the time difference threshold, which can be set according to actual use requirements, and the second text can be the text of the second object corresponding to one conversation time, or the second text can be the text of the second object corresponding to multiple conversation times spliced ​​together in the order of the conversation time. In this way, an accurate second text can be obtained.

[0077] For the first extraction method of the first text, since the text of the first object and the text of the second object in the conversation text are sorted in chronological order according to the conversation time, after determining the first text, the first position of the first text in the conversation text can be determined, and the text of the second object whose position is before the first position and the amount of text between the position and the first position is less than a second threshold is taken as the second text, thereby realizing the extraction of the second text of the second object from the conversation text.

[0078] Among them, the second threshold is the position interval threshold, which can be set according to actual usage requirements. In this way, the second text of the second object can be accurately determined, and the second text is used for subsequent determination of abnormal text, so as to improve the accuracy of determining abnormal text later.

[0079] In step 103, if the second text does not include text having the same meaning as the keyword, the first text is determined to be an abnormal text.

[0080] In some embodiments, step 103 can be implemented in the following manner: in order to determine whether the first text is an abnormal text, it can be first determined whether the second text includes text with the same meaning as the keyword. The following describes various methods for determining whether the second text includes text with the same meaning as the keyword.

[0081] The first determination method is to determine whether the second text includes the keyword for each keyword in each first text. If the second text includes the keyword, it is determined that the second text includes text with the same meaning as the keyword. Therefore, the first text can be determined to be a normal text (i.e., non-abnormal text), and the normal text is associated with the keyword. If the second text does not include the keyword, other determination methods can be used for determination.

[0082] The second determination method is to determine the third text corresponding to the keyword from pre-set candidate keywords and candidate third texts when the number of characters of the keyword is greater than or equal to a preset threshold value, wherein the keyword and the third text have the same meaning, for example, the keyword is fourteen days and the third text is two weeks.

[0083] After determining the third text with the same meaning as the keyword, it can be determined whether the second text includes the third text. If the second text includes the third text, it is determined that the second text includes text with the same meaning as the keyword. Therefore, it can be determined that the first text is normal text. If the second text does not include the third text, it can be determined that the second text does not include text with the same meaning as the keyword, or other determination methods can be used for determination.

[0084] In a third determination method, the second text is split based on a third threshold to obtain multiple second subtexts, the number of characters corresponding to each second subtext is less than or equal to the third threshold, and for each second subtext, it can be determined whether the keyword includes the second subtext. The number of characters of the keyword can be greater than, equal to, or less than the third threshold.

[0085] If the keyword includes the second subtext, it is determined that the second text includes text with the same meaning as the keyword, so the first text can be determined to be a normal text (ie, non-abnormal text). If the keyword does not include any of the multiple second subtexts, other determination methods can be used for determination.

[0086] The fourth determination method is that when the number of characters of the keyword is less than a preset threshold, the second text is split based on the fourth threshold to obtain multiple first sub-texts, that is, the second text includes multiple first sub-texts, and the number of characters corresponding to each first sub-text is less than or equal to the fourth threshold, and the fourth threshold is less than the third threshold.

[0087] For the fourth determination method, see Figure 4 , Figure 4 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 2 , the following combination Figure 4 The steps shown in the figure are used to illustrate the fourth determination method. Figure 3 Before step 103 shown, you can execute Figure 4 Steps shown.

[0088] In step 401 , when the number of characters of a keyword is less than a preset number threshold, for each first subtext, the similarity between the keyword and the first subtext is determined.

[0089] In some embodiments, step 401 may be implemented by determining the first pinyin corresponding to the keyword and determining the second pinyin corresponding to each first subtext, and then calculating the similarity between the first pinyin and the second pinyin for each second pinyin.

[0090] In some embodiments, the Jaccard similarity coefficient between the first pinyin and the second pinyin can be calculated. The Jaccard similarity coefficient is an indicator for measuring the similarity between characters, and the Jaccard similarity coefficient between the first pinyin and the second pinyin can be used as the similarity between the first pinyin and the second pinyin. Of course, the similarity between the first pinyin and the second pinyin can also be calculated by the pinyin editing example method, the cosine similarity method, etc., which is not limited here.

[0091] In some embodiments, the number of first subtexts is multiple, the number of similarities between the first pinyin and the second pinyin is multiple, and the number of first subtexts is consistent with the number of similarities. After calculating the similarity between the first pinyin and the second pinyin, for each similarity, it can be determined whether the similarity is greater than a similarity threshold.

[0092] If there is a similarity greater than the similarity threshold, it can be determined that the second text includes text with the same meaning as the keyword. If there is no similarity greater than the similarity threshold, that is, the similarities are all less than or equal to the similarity threshold, step 402 can be performed.

[0093] The following describes a method for determining the similarity threshold. In some embodiments, the first pinyin corresponding to the keyword can be determined, and the second pinyin corresponding to each first subtext can be determined. For each second pinyin, the first number of different elements in the first pinyin and the second pinyin can be determined.

[0094] The following describes a method for determining the first number of different elements in a first pinyin and a second pinyin. In some embodiments, it can be determined whether the elements at the same position in the first pinyin and the second pinyin are the same. If the elements at the same position in the first pinyin and the second pinyin are the same, no processing is required. If the elements at the same position in the first pinyin and the second pinyin are different, the element can be marked as a different element and the first number of different elements can be determined.

[0095] In some embodiments, there is an arrangement order for the elements in the first pinyin and the second pinyin. According to the arrangement order, if the elements at the same position in the first pinyin and the second pinyin are different, the element can be marked as the first element and the number of first elements can be determined.

[0096] According to the reverse order of the arrangement order, if the elements at the same position in the first pinyin arranged in reverse and the second pinyin arranged in reverse are the same, the element can be marked as the second element, and the number of second elements can be determined, and the difference between the number of first elements and the number of second elements can be determined as the first number of different elements.

[0097] For example, if the first pinyin is "yinian" and the second pinyin is "yinan", according to the arrangement order, for "yinian" and "yinan", "ian" can be marked as the first element, and the number of first elements is determined to be 3. According to the reverse order of the arrangement order, that is, for "nainiy" and "naniy", "na" can be marked as the second element, and the number of second elements is determined to be 2, and then the first number of different elements can be determined to be 1. In this way, a more accurate first number of different elements can be determined, and the first number is used to determine the similarity threshold, so that a more accurate similarity threshold can be determined.

[0098] After determining the first quantity, a similarity threshold can be determined based on the second quantity of elements in the first pinyin, the third quantity of elements in the second pinyin, and the first quantity. In some embodiments, the sum of the second quantity and the third quantity can be determined, and then the ratio between the first quantity and the sum is determined as the similarity threshold. In this way, an accurate similarity threshold can be determined.

[0099] In step 402, if the similarity is less than or equal to the similarity threshold, it is determined that the second text does not include text with the same meaning as the keyword.

[0100] When there is no similarity greater than the similarity threshold, that is, the similarities are all less than or equal to the similarity threshold, it can be determined that the second text does not include text with the same meaning as the keyword, and further the first text can be determined to be an abnormal text.

[0101] The fourth determination method is described below in conjunction with formula (1).

[0102]

[0103] Among them, S a is the second pinyin corresponding to the first subtext, S info-text is the first pinyin corresponding to the keyword, Jaccard (S a ,S info-text ) is the similarity between the keyword and the first subtext, is the similarity threshold, L a is the third number of elements in the second spelling, L info-text is the second number of elements in the first pinyin, La +L info-text is the sum of the second number and the third number, and L is the first number of different elements.

[0104] If each first subtext and the keyword satisfy formula (1), it means that there is no similarity greater than the similarity threshold, that is, the similarities are less than or equal to the similarity threshold. It can be determined that the second text does not include text with the same meaning as the keyword, and further it can be determined that the first text is an abnormal text.

[0105] If any of the first subtexts and the keyword does not satisfy formula (1), it means that there is a similarity greater than the similarity threshold, and it can be determined that the second text includes text with the same meaning as the keyword, and further the first text can be determined to be a normal text.

[0106] There may be text with text input errors or audio-to-text errors in the conversation text. For the erroneous text (text input errors or audio-to-text errors), the first determination method to the third determination method cannot determine whether the second text includes text with the same meaning as the keyword. However, the fourth determination method can accurately determine whether the second text includes text with the same meaning as the keyword based on the similarity between the first pinyin corresponding to the keyword and the second pinyin corresponding to the first sub-text, as well as the similarity threshold, thereby determining whether the first text is an abnormal text, thereby further improving the accuracy of determining abnormal text.

[0107] That is to say, the fourth determination method can avoid the problem of reduced accuracy in determining abnormal text due to text input errors or audio-to-text errors, thereby achieving accurate determination of abnormal text even in the presence of text input errors or audio-to-text errors, thereby further improving the accuracy of determining abnormal text.

[0108] A fifth determination method is that when the number of characters of the keyword is greater than or equal to the preset number threshold, and the number of characters of the keyword is less than the third threshold, that is, when the number of characters of the keyword is less than the number of characters of the second subtext, for each second subtext, it can be determined whether the second subtext includes the keyword. If the second subtext includes the keyword, it is determined that the second text includes text with the same meaning as the keyword, and therefore, the first text can be determined to be a normal text. If the second subtext does not include the keyword, other determination methods can be used for determination.

[0109] The relationship between the first determination method and the fifth determination method is described below. For the first determination method, if the second text does not include the keyword, it can be determined whether the number of characters of the keyword is less than a preset number threshold.

[0110] When the number of characters of the keyword is greater than or equal to the preset number threshold, the second determination method can be used first, that is, determining the third text corresponding to the keyword, and determining whether the second text includes the third text. If the second text includes the third text, it is determined that the second text includes text with the same meaning as the keyword, and therefore, the first text can be determined to be a normal text.

[0111] In some embodiments, if the second text does not include the third text, a third determination method can also be used to determine whether the keyword includes the second sub-text. If the keyword includes the second sub-text, it is determined that the second text includes text with the same meaning as the keyword. Therefore, the first text can be determined to be a normal text.

[0112] If the keyword does not include the second subtext, it can be determined whether the number of characters of the keyword is less than the third threshold. If the number of characters of the keyword is greater than or equal to the third threshold, it can be determined that the second text does not include text with the same meaning as the keyword, and then the first text can be determined to be an abnormal text.

[0113] In some embodiments, if the second text does not include the third text, it can be determined whether the number of characters of the keyword is less than the third threshold. In the case where the number of characters of the keyword is less than the third threshold, the fifth determination method can be used, that is, determining whether the second subtext includes the keyword. If the second subtext includes the keyword, it is determined that the second text includes text with the same meaning as the keyword, that is, the first text is a normal text. If the second subtext does not include the keyword, it is determined that the second text does not include text with the same meaning as the keyword, and the first text can be determined to be an abnormal text.

[0114] When the number of characters in the keyword is less than the preset number threshold, in some embodiments, a fourth determination method may be used. For details, see Figure 4 After determining the similarity, if the similarities are less than or equal to the similarity threshold, it is determined that the second text does not include text with the same meaning as the keyword, and the first text can be determined to be an abnormal text. If the similarity is greater than the similarity threshold, it is determined that the second text includes text with the same meaning as the keyword, and the first text can be determined to be a normal text.

[0115] In some embodiments, the third determination method can be used to determine whether the keyword includes the second subtext. If the keyword includes the second subtext, it is determined that the second text includes text with the same meaning as the keyword, so the first text can be determined to be a normal text. If the keyword does not include the second subtext, the fourth determination method can be used, which will not be described in detail here.

[0116] In combination with the above content, it can be known that after executing the first determination method, the second determination method can be executed, after executing the second determination method, the third determination method can be executed, or the third determination method and the fifth determination method can be executed. After executing the first determination method, the fourth determination method can also be executed. After the first determination method, the third determination method and the fourth determination method can also be executed.

[0117] In some embodiments, after executing the first determination method, the third determination method may be executed, and then the fifth determination method may be executed, and then the second determination method may be executed. In some embodiments, after executing the first determination method, the third determination method may be executed, and then the fourth determination method may be executed. Specifically, it may be set according to actual use requirements, that is, the first determination method-the fifth determination method provided in the embodiment of the present application may be used in combination, and there is a progressive relationship between the determination methods used in combination, which can more accurately determine the abnormal text, and further improve the accuracy of determining the abnormal text, and can solve the above-mentioned technical problem 1.

[0118] In some embodiments, the keyword is composed of a word representing a quantity and a word representing a unit, for example, the keyword is "fifteen days", where "fifteen" is a word representing a quantity and the word representing a unit is "day". In view of this situation, one or more of the first to fifth determination methods described above may be used to determine whether the first text is an abnormal text.

[0119] Among them, the first determination method, the second determination method and the third determination method are consistent with the above content. For details, please refer to the above description and will not be repeated here. When the number of characters corresponding to the word representing the quantity is less than 2, the probability of audio-to-text error or text input error is high. In this case, the fourth determination method can be used to determine whether the first text is an abnormal text.

[0120] For the fourth determination method, when the number of characters corresponding to the word representing the quantity is less than 2, the similarity between the word representing the quantity and the first subtext can be determined for each first subtext. If the similarities are all less than or equal to the similarity threshold, it is determined that the second text does not include text with the same meaning as the keyword, and then the first text can be determined to be an abnormal text. If there is a similarity greater than the similarity threshold, it is determined that the second text includes text with the same meaning as the keyword, and then the first text can be determined to be a normal text.

[0121] When the number of characters corresponding to the word representing quantity is greater than or equal to 2, the probability of audio-to-text error or text input error is small. Therefore, the fourth determination method is not used to determine whether the first text is an abnormal text, but the fifth determination method is used to determine whether the first text is an abnormal text.

[0122] For the fifth determination method, when the number of characters corresponding to the word representing quantity is greater than or equal to 2, for each second subtext, it can be determined whether the second subtext includes the word representing quantity. If the second subtext includes the word representing quantity, it is determined that the second text includes text with the same meaning as the keyword, and therefore, the first text can be determined to be a normal text. If the second subtext does not include the word representing quantity, it is determined that the second text does not include text with the same meaning as the keyword, and therefore, the first text can be determined to be an abnormal text. In this way, the abnormal text can be accurately determined for the case where the keyword is composed of a word representing quantity and a word representing a unit.

[0123] In some embodiments, Figure 3 After step 103 shown, that is, after determining that the first text is an abnormal text, the number of abnormal texts may be determined, and when the number is greater than an abnormal number threshold, an alarm message may be generated.

[0124] In some embodiments, the number of first objects may be multiple. For each first object, the first number of abnormal texts corresponding to the first object may be determined. When the first number is greater than a first abnormal number threshold, an alarm message for the first object is generated.

[0125] In some embodiments, the alarm level of the keyword can be preset, and there is an associated first abnormality quantity threshold for the alarm level. The higher the alarm level, the smaller the first abnormality quantity threshold, that is, the alarm level and the first abnormality quantity threshold are inversely correlated.

[0126] After determining that the first text is an abnormal text, a first number of abnormal texts corresponding to the first object can be determined, and based on the alarm level of the keyword in the abnormal text, a first abnormal number threshold associated with the alarm level is determined. If the abnormal text includes multiple keywords, the first abnormal number threshold associated with the highest alarm level is determined.

[0127] When the first number is less than or equal to the first abnormal number threshold, no processing is required. When the first number is greater than the first abnormal number threshold, an alarm message is generated for the first object. For example, in the scenario of telephone communication involving the financial field, the first object cannot actively tell the time information to the second object. When the first text is determined to be an abnormal text, it means that the first object has violated the rules during the telephone communication, that is, the first object actively told the time information to the second object. When the first number of abnormal texts is greater than 2 (the first abnormal number threshold), an alarm message is generated for the first object. The alarm message can be "there is a violation in the conversation with the second object."

[0128] In some embodiments, there is a corresponding relationship between the warning information and the warning level. For different warning levels, the warning information can be different. When the warning level is greater than the warning level threshold, the conversation between the first object and any second object can be suspended. In this way, after determining the abnormal text, the first object corresponding to the abnormal text can be warned so that the first object can make improvements to the abnormal text.

[0129] In some embodiments, for each keyword, a second number of abnormal texts corresponding to the keyword can be determined, and a second abnormal number threshold can be determined based on the keyword. If the second number is not greater than the second abnormal number threshold, the abnormal text may not be processed.

[0130] When the second number is greater than the second abnormal number threshold, an alarm message for the keyword can be generated. In this way, after the abnormal text is determined, an alarm can be issued for the object corresponding to the abnormal text so that improvements can be made for the keyword.

[0131] In order to facilitate understanding of the text processing method provided in the embodiment of the present application, Figure 5 The text processing method provided in the embodiment of the present application is described in detail. Figure 5 , Figure 5 This is a flowchart of the text processing method provided in the embodiment of the present application. Figure 3 .

[0132] See also Figure 5 In step 501, a first text of the first object is extracted from a conversation text of the first object and the second object.

[0133] The first text includes keywords. Step 501 and Figure 3 The step 101 shown is the same, and for details, please refer to the description of step 101. As an example of step 501, for a scenario of telephone communication involved in the financial field, the conversation text can be obtained first.

[0134] The dialogue text may be: Customer (i.e. the second object): Hello, who is this? Agent (i.e. the first object): Hello, are you A's friend? Customer: Well, what's your matter? You tell me. Agent: Oh, this is B. Has A contacted you? Customer: He said it will only take fifteen days. Has it not been processed yet? Agent: No, please tell him. Customer: It will probably be processed in a few days. I'll talk to him. Agent: It has been overdue for fifteen days. I'm worried that it will affect A. Customer: OK, I understand.

[0135] The keyword is fifteen days. After obtaining the conversation text, the conversation text between the first object and the second object can be split to obtain the text of the first object: "Agent: Hello, are you A's friend? Agent: Oh, this is B. Has A contacted you? Agent: No, please tell him. Agent: It has been overdue for fifteen days. I am worried that it will affect A." The number of texts of the first object is 4, and each text in the first object is sorted according to the conversation time.

[0136] By splitting the conversation text between the first object and the second object, we can also get the text of the second object: "Customer: Hello, who is this? Customer: Well, what's your business? Tell me. Customer: He said it would only take fifteen days. Hasn't it been processed yet? Customer: It will probably be processed in a few days. I'll talk to him. Customer: OK, I got it." The number of texts of the second object is 5, and each text in the second object is sorted according to the conversation time.

[0137] For the text of the first object, it can be identified whether the text of each conversation time includes keywords. "Agent: It has been overdue for fifteen days. I am worried that it will affect A." includes the keyword "fifteen days", so "Agent: It has been overdue for fifteen days. I am worried that it will affect A." can be used as the first text.

[0138] In step 502, a second text of a second object is extracted from the conversation text.

[0139] Step 502 and Figure 3 The step 102 shown is consistent. For details, please refer to the description of step 102, which will not be repeated here. As an example of step 502, following the example of step 501, the difference between the conversation time of "Customer: It will probably be processed in a few days. I will talk to him." and the conversation time of "Agent: It has been overdue for fifteen days. I am worried that it will affect A." in the text of the second object is less than the first threshold.

[0140] In the text of the second object, the difference between the conversation time of "Customer: He said it would only be fifteen days, hasn't it been processed yet?" and the conversation time of "Agent: It has been overdue for fifteen days, I am worried that it will have an impact on A." is less than the first threshold.

[0141] Then, we can concatenate “Customer: He said it would take only fifteen days. Hasn’t it been processed yet?” and “Customer: It will probably be processed in a few days. Let me talk to him about it.” in the order of conversation time from morning to night to get the second text “Customer: He said it would take only fifteen days. Hasn’t it been processed yet? Customer: It will probably be processed in a few days. Let me talk to him about it.”

[0142] In step 503, it is determined whether the second text includes a keyword.

[0143] If the second text includes the keyword, step 504 is executed, and if the second text does not include the keyword, step 505 is executed. As an example of step 503, following the example of step 502, the second text includes the keyword, so step 504 can be executed. Among them, step 503 corresponds to the first determination method.

[0144] In step 504, it is determined that the second text includes text having the same meaning as the keyword, and the first text is determined to be a normal text.

[0145] In step 505, the second text is split based on the third threshold to obtain a plurality of second subtexts, and for each second subtext, it is determined whether the keyword includes the second subtext.

[0146] The third threshold may be 3, and step 504 corresponds to the third determination method. When the second subtext is not included in the keyword, step 506 may be performed. When the second subtext is included in the keyword, step 504 may be performed.

[0147] In step 506, a third text having the same meaning as the keyword is obtained, and it is determined whether the second text includes the third text.

[0148] Step 506 corresponds to the second determination method. When the second text includes the third text, step 504 is executed. When the second text does not include the third text, step 507 can be executed, or step 508 can be executed.

[0149] In step 507, the second text is split based on the second threshold to obtain a plurality of first subtexts, and for each first subtext, the similarity between the keyword and the first subtext is determined.

[0150] Step 507 corresponds to the fourth determination method. When the similarities are both less than or equal to the similarity threshold (corresponding to Figure 5 The similarity in is not greater than the threshold), step 509 may be performed, and if there is a similarity greater than the similarity threshold (corresponding to Figure 5 is greater than a threshold), step 504 may be executed.

[0151] In step 508 , for each second sub-text, it is determined whether the second sub-text includes a keyword.

[0152] Step 508 corresponds to the fifth determination method. When the second subtext includes the keyword, step 504 may be executed. When the second subtext does not include the keyword, step 509 may be executed.

[0153] In step 509, it is determined that the second text does not include any text having the same meaning as the keyword, and the first text is determined to be an abnormal text.

[0154] In the process of determining abnormal text, the present application does not need to rely on a text library for various dialogue texts, and can determine abnormal text based only on the dialogue text. When the second text of the second object does not include text with the same meaning as the keyword, the first text can be determined as an abnormal text. Even if different objects have different vocabulary usage habits, as long as the meaning is the same as the keyword, the abnormal text can be determined. In other words, even if there is diversity in the dialogue texts and different objects have different vocabulary usage habits, the abnormal text can be determined, thereby improving the accuracy of determining the abnormal text.

[0155] Moreover, in case of text input errors or audio-to-text conversion errors, the fourth determination method is adopted, which can accurately determine whether the second text includes text with the same meaning as the keyword based on the similarity between the first pinyin corresponding to the keyword and the second pinyin corresponding to the first sub-text, and the similarity threshold, so as to determine whether the first text is an abnormal text, thereby improving the accuracy of determining abnormal text.

[0156] The following is a description of an exemplary structure of a text processing device 455 provided in an embodiment of the present application implemented as a software module. In some embodiments, Figure 2 As shown, the software modules stored in the text processing device 455 of the memory 450 may include:

[0157] The extraction module 4551 is used to extract a first text of a first object from the conversation text, wherein the first text contains keywords.

[0158] The extraction module 4551 is further configured to extract a second text of a second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text.

[0159] The determination module 4552 is configured to determine that the first text is an abnormal text if the second text does not include any text having the same meaning as the keyword.

[0160] In some embodiments, the second text includes a first sub-text; the determination module 4552 is also used to determine the similarity between the keyword and the first sub-text for each of the first sub-texts before determining that the first text is an abnormal text if the second text does not include text with the same meaning as the keyword, when the number of characters of the keyword is less than a preset number threshold; if the similarity is less than or equal to the similarity threshold, determine that the second text does not include text with the same meaning as the keyword.

[0161] In some embodiments, the determination module 4552 is also used to determine the first pinyin corresponding to the keyword and the second pinyin corresponding to the first subtext before determining the similarity between the keyword and the first sub-text; for each of the second pinyins, determine the first number of different elements in the first pinyin and the second pinyin; and determine the similarity threshold based on the second number of elements in the first pinyin, the third number of elements in the second pinyin, and the first number.

[0162] In some embodiments, the determination module 4552 is further used to determine the sum of the second number and the third number; and determine the ratio between the first number and the sum as the similarity threshold.

[0163] In some embodiments, the determination module 4552 is further used to mark the elements at the same position in the first pinyin and the second pinyin as different elements if the elements are different, and determine a first number of the different elements.

[0164] In some embodiments, the determination module 4552 is also used to determine that before the first text is an abnormal text if the second text does not include text with the same meaning as the keyword, if the number of characters of the keyword is greater than or equal to a preset threshold, if the second text does not include the keyword, determine a third text with the same meaning as the keyword; if the second text does not include the third text, determine that the second text does not include text with the same meaning as the keyword.

[0165] In some embodiments, the determination module 4552 is further used to determine the number of abnormal texts after determining that the first text is abnormal text; and generate an alarm message when the number is greater than an abnormal number threshold.

[0166] The embodiment of the present application provides a computer program product, which includes a computer program or a computer executable instruction, and the computer program or the computer executable instruction is stored in a computer-readable storage medium. The processor of the electronic device reads the computer program or the computer executable instruction from the computer-readable storage medium, and the processor executes the computer program or the computer executable instruction, so that the electronic device executes the text processing method described in the embodiment of the present application.

[0167] The present application embodiment provides a computer-readable storage medium in which computer executable instructions or computer programs are stored. When the computer executable instructions or computer programs are executed by a processor, the processor will be caused to execute the text processing method provided by the present application embodiment, for example, Figure 3 The text processing method shown.

[0168] In some embodiments, the computer-readable storage medium may be a memory such as RAM, ROM, flash memory, magnetic surface memory, optical disk, or CD-ROM; or may be various devices including one or any combination of the above memories.

[0169] In some embodiments, computer executable instructions may be in the form of a program, software, software module, script or code, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment.

[0170] As an example, computer-executable instructions may, but need not, correspond to a file in a file system, may be stored as part of a file that stores other programs or data, for example, in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files storing one or more modules, subroutines, or code portions).

[0171] As an example, computer executable instructions or computer programs may be deployed to be executed on one electronic device, or on multiple electronic devices located at one site, or on multiple electronic devices distributed at multiple sites and interconnected by a communication network.

[0172] In the process of determining abnormal text, the present application does not need to rely on a text library for various dialogue texts, and can determine abnormal text based only on dialogue text. If the second text of the second object includes text with the same meaning as the keyword, the first text can be determined as abnormal text. Even if different objects have different vocabulary usage habits, as long as the meaning is the same as the keyword, the abnormal text can be determined. In other words, even if there is diversity in dialogue texts and different objects have different vocabulary usage habits, the abnormal text can be determined, thereby improving the accuracy of determining abnormal text and solving technical problem 1.

[0173] Moreover, in case of text input errors or audio-to-text conversion errors, the fourth determination method is adopted, which can accurately determine whether the second text includes text with the same meaning as the keyword based on the similarity between the first pinyin corresponding to the keyword and the second pinyin corresponding to the first sub-text, and the similarity threshold, so as to determine whether the first text is an abnormal text, thereby improving the accuracy of determining abnormal text.

[0174] The above is only an embodiment of the present application and is not intended to limit the protection scope of the present application. Any modifications, equivalent substitutions and improvements made within the spirit and scope of the present application are included in the protection scope of the present application.

Claims

1. A text processing method, characterized in that: The method comprises: Extracting a first text of a first object from the conversation text, wherein the first text contains a keyword; Extracting a second text of a second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text; If the second text does not include text having the same meaning as the keyword, the first text is determined to be an abnormal text.

2. The method according to claim 1, characterized in that The second text includes a first subtext; If the second text does not include text with the same meaning as the keyword, before determining that the first text is an abnormal text, the method further includes: When the number of characters of the keyword is less than a preset number threshold, determining, for each of the first subtexts, a similarity between the keyword and the first subtext; If the similarity is less than or equal to the similarity threshold, it is determined that the second text does not include text with the same meaning as the keyword.

3. The method according to claim 2, characterized in that Before determining the similarity between the keyword and the first subtext, the method further includes: Determine a first pinyin corresponding to the keyword, and determine a second pinyin corresponding to the first subtext; For each of the second pinyins, determining a first number of elements that are different between the first pinyin and the second pinyin; The similarity threshold is determined based on a second number of elements in the first pinyin, a third number of elements in the second pinyin, and the first number.

4. The method according to claim 3, characterized in that The determining the similarity threshold based on the second number of elements in the first pinyin, the third number of elements in the second pinyin, and the first number includes: determining a sum of the second number and the third number; The ratio between the first number and the sum is determined as the similarity threshold.

5. The method according to claim 3, characterized in that: The determining a first number of different elements in the first pinyin and the second pinyin includes: If the elements at the same position in the first pinyin and the second pinyin are different, the elements are marked as different elements, and a first number of the different elements is determined.

6. The method according to claim 1, characterized in that If the second text does not include text with the same meaning as the keyword, before determining that the first text is an abnormal text, the method further includes: In the case where the number of characters of the keyword is greater than or equal to a preset number threshold, if the second text does not include the keyword, determining a third text having the same meaning as the keyword; If the second text does not include the third text, it is determined that the second text does not include text having the same meaning as the keyword.

7. A text processing device, characterized in that: The device comprises: An extraction module, configured to extract a first text of a first object from the conversation text, wherein the first text contains keywords; The extraction module is further used to extract a second text of a second object from the conversation text, wherein the conversation time of the second text is earlier than the conversation time of the first text; A determination module is configured to determine that the first text is an abnormal text if the second text does not include any text having the same meaning as the keyword.

8. An electronic device, characterized in that: The electronic device comprises: A memory for storing computer executable instructions or computer programs; The processor is used to implement the text processing method according to any one of claims 1 to 6 when executing the computer executable instructions or computer programs stored in the memory.

9. A computer-readable storage medium storing computer-executable instructions or a computer program, characterized in that: When the computer executable instructions or computer program are executed by a processor, the text processing method according to any one of claims 1 to 6 is implemented.

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

Images