Text processing method and device

Abstract

The embodiment of the invention provides a text processing method and device, and the method comprises the steps: inputting an obtained to-be-processed text into a detection model, and obtaining a text type; inputting the to-be-processed text and the text type into a keyword extraction model to obtain a keyword of the text type; and determining an initial recommendation value in the to-be-processed text based on the keyword, comparing the initial recommendation value with a recommendation value which is in the same type as the initial recommendation value in a database, and determining a target recommendation value, so that the efficiency of recording manual work can be improved, and the face visit duration is greatly shortened.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and in particular to a text processing method. One or more embodiments of this specification also relate to a text processing apparatus, a computing device, and a computer-readable storage medium. Background Technology

[0002] With the advancement of data processing technology and the rapid popularization of mobile Internet, computer technology has been widely applied to various fields of society. In the claims process, when recorders communicate with users through face-to-face interviews, the record-taking and inquiry process takes up a lot of time. The recorders need to repeatedly confirm the information collected with the users before manually filling it into the forms of the claims processing system. This not only makes manual record-taking costly but also takes a long time to process the project, which greatly affects the efficiency of project processing. Summary of the Invention

[0003] In view of this, embodiments of this specification provide a text processing method. One or more embodiments of this specification also relate to a text processing apparatus, a computing device, and a computer-readable storage medium to address the technical deficiencies existing in the prior art.

[0004] According to a first aspect of the embodiments of this specification, a text processing method is provided, comprising: Input the acquired text to be processed into the detection model to obtain the text type; Input the text to be processed and the text type into the keyword extraction model to obtain the keywords of the text type; Based on the keywords, an initial recommendation value is determined in the text to be processed, and the initial recommendation value is compared with recommendation values ​​of the same type in the database to determine the target recommendation value.

[0005] According to a second aspect of the embodiments of this specification, a text processing apparatus is provided, comprising: The text acquisition module is configured to input the acquired text to be processed into the detection model to obtain the text type; The keyword extraction module is configured to input the text to be processed and the text type into the keyword extraction model to obtain keywords of the text type; The recommendation value determination module is configured to determine an initial recommendation value in the text to be processed based on the keywords, and compare the initial recommendation value with recommendation values ​​of the same type in the database to determine a target recommendation value.

[0006] According to a third aspect of the embodiments of this specification, a computing device is provided, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, wherein the processor executes the computer-executable instructions to implement the steps of the text processing method.

[0007] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores computer-executable instructions, which, when executed by a processor, implement the steps of any of the text processing methods described herein.

[0008] The video processing method provided in this specification involves inputting the acquired text to be processed into a detection model to obtain the text type; inputting the text to be processed and the text type into a keyword extraction model to obtain keywords of the text type; determining an initial recommendation value in the text to be processed based on the keywords; and comparing the initial recommendation value with recommendation values ​​of the same type in the database to determine a target recommendation value.

[0009] One embodiment of this specification obtains the initial recommendation value of keywords in the text to be processed by a detection model and a keyword extraction model, and then obtains the target recommendation value. The target recommendation value can be displayed in the input box recorded by the recorder. This realizes the automatic identification of keyword information from the real-time dialogue text of the video interview and displays it in the form floating window of the inquiry record. It obtains intelligent recommendations of key content, which not only improves the efficiency of manual recording work, but also greatly reduces the interview time. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a text processing method provided in one embodiment of this specification; Figure 2 This is a schematic diagram illustrating the effect of a hospital recommendation value in a text processing method provided in one embodiment of this specification; Figure 3 This is a schematic diagram illustrating the effect of a time recommendation value for a text processing method provided in one embodiment of this specification; Figure 4 This is a flowchart illustrating the processing procedure of a text processing method provided in one embodiment of this specification. Figure 5 This is a schematic diagram of the structure of a text processing device provided in one embodiment of this specification; Figure 6 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0011] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0012] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0013] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0014] First, the terms and concepts used in one or more embodiments of this specification will be explained.

[0015] Video interview: The process of recording personnel conducting an interview with a user who has reported a case via remote video conversation. Interview transcript: During the claims process, this transcript records the information obtained by staff when asking users about their medical records and personal details. ASR: Automatic Speech Recognition, is a technology that converts audio signals into text signals. NER: Named Entity Recognition, is a technology that can identify entity names (address, time, disease, hospital, examination, etc.) in a text. Entity Linking refers to associating entities identified by NER with corresponding entities in an existing knowledge base. Topic Detection identifies and categorizes important topics mentioned in a conversation.

[0016] SDK: Software Development Kit, is a collection of development tools that software engineers use to create application software for specific software packages, software frameworks, hardware platforms, operating systems, etc.

[0017] This specification provides a text processing method, and also relates to a text processing apparatus, a computing device, and a computer-readable storage medium, which will be described in detail in the following embodiments.

[0018] The text processing methods provided in the embodiments of this specification can be applied to insurance claims scenarios, security agency inquiry record scenarios, or hospital consultation scenarios. The embodiments of this specification do not limit the application scenarios. Taking the text processing of inquiry records in the insurance claims scenario as an example, the text processing methods are described in detail.

[0019] In policyholder claim scenarios, recorders currently primarily use video interviews to review users' medical information, aiming to reduce the cost of offline assessments and improve overall claims efficiency. However, due to the complexity of claims, a single remote video interview is time-consuming. Through communication with users and information collection, recorders have found that the written record-taking and questioning phase consumes a significant amount of time. Recorders frequently need to confirm details such as time, location, hospital, illness, and examination methods with users before manually filling them into the claims system form. Therefore, the text processing method provided in this specification can intelligently recommend key content from the recorded record based on ASR results. It automatically identifies keyword information from the real-time dialogue text of the video interview and displays it in the form's floating window, reducing manual filling costs and improving the overall efficiency of video interviews.

[0020] Most insurance companies currently lack dedicated tools for conducting video interviews. Typically, during the interview, recorders manually fill out forms one by one. This method is inefficient, requiring repeated confirmation of information or searching for details within the context. Therefore, the lack of a system for keyword recognition, extraction, and intelligent recommendation to assist recorders in filling out the forms also fails to effectively address the time-consuming process.

[0021] It should be noted that the text processing method provided in the embodiments of this specification is applied to a server. The text to be processed is the text after processing the audio in the video received by the client. In practical applications, users can conduct video interviews with the recording personnel based on the client. The client uploads the obtained audio data to the server to obtain keywords in the audio, which facilitates the recording personnel to quickly fill in the form later.

[0022] Figure 1 A flowchart of a text processing method according to an embodiment of this specification is shown, which specifically includes the following steps.

[0023] Step 102: Input the acquired text to be processed into the detection model to obtain the text type.

[0024] The text to be processed can be understood as the text resulting from processing the audio in the video.

[0025] The detection model can be understood as a topic detection model for speech text. This topic detection model can be pre-trained. Inputting speech text into the topic detection model will output the topic type corresponding to the speech text.

[0026] In practical applications, the server inputs the acquired text to be processed into the detection model to perform topic detection. Then, it can output the text type corresponding to the text to be processed. For example, after inputting the text to be processed into the detection model, the output text type of the text to be processed is "work trajectory" or "medical treatment process".

[0027] After obtaining the audio data from the video, the server can process the audio data to obtain the text to be processed; specifically, before inputting the obtained text to be processed into the detection model, the process further includes: The receiving client processes the acquired voice data based on a preset voice processing method to determine the voice text, and then extracts the voice text according to preset text extraction conditions to obtain the text to be processed; or The system receives initial text edited by the user through the client and extracts the initial text according to preset text extraction conditions to obtain the text to be processed.

[0028] The preset speech processing method can be understood as the system's preset processing method for converting speech into text, such as ASR technology.

[0029] Among them, the preset text extraction conditions can be understood as the preset number or content of text based on the application scenario.

[0030] The initial text can be understood as the text entered by the user through the client, and this input text may be incorrect or abbreviated, which does not meet the needs of the recorder to fill out the form.

[0031] Specifically, the server can receive the voice data obtained by the client after processing it according to a preset voice processing method, and then extract the voice text according to preset text extraction conditions to obtain the text to be processed; or the server can receive the initial text edited by the user through the client, and then extract the initial text according to preset text extraction conditions to obtain the text to be processed.

[0032] In practical applications, the client can convert the voice recording of a conversation between a person recording a conversation with a user into text using ASR technology. It can also include the role information of the person recording the conversation and the user, which can be uploaded to the server via the SDK. After receiving the voice-to-text recording, the server can extract the context information of the conversation over the past 20 rounds, thus obtaining the text to be processed. It should be noted that the preset voice processing method is not limited to ASR technology; it can also include any technology that converts voice to text. The number of conversations extracted by the preset text extraction conditions is not limited to this and can be set according to different application scenarios. Alternatively, the user can manually input initial text through the client, and this initial text can also be extracted according to the preset text extraction conditions, with the extracted portion serving as the text to be processed.

[0033] The text processing method provided in the embodiments of this specification obtains the text to be processed by processing voice data or input text, so as to facilitate subsequent keyword extraction from the text to be processed and obtain the target recommendation value of the keyword.

[0034] Step 104: Input the text to be processed and the text type into the keyword extraction model to obtain the keywords of the text type.

[0035] Keyword extraction models can be understood as models that extract keywords from text, such as NER or entity chain indexing models.

[0036] Specifically, the server inputs the text to be processed and the text type into the keyword extraction model, thereby obtaining the keywords for that text type.

[0037] In practical applications, the server can perform entity recognition on the text to be processed. Entity recognition uses NER and entity chaining models to identify and extract keywords from the current and context text. The keyword extraction model can be pre-trained, and this specification does not limit this in any way. After the keyword extraction model processes the text to be processed and the text type, it can obtain keywords of the text type. For example, if the text to be processed is the dialogue text between the recorder and the patient, and the determined text type is "medical treatment process", then the output keywords of the medical treatment process can include five types: hospital, disease, time, address, and examination items.

[0038] Step 106: Determine an initial recommendation value in the text to be processed based on the keywords, and compare the initial recommendation value with recommendation values ​​of the same type in the database to determine the target recommendation value.

[0039] The initial recommendation value can be understood as the content of the keyword appearing in the text to be processed.

[0040] The target recommendation value can be understood as the recommendation value that matches the initial recommendation value in the preset database.

[0041] Specifically, the server can determine an initial recommendation value in the text to be processed based on certain keywords, and compare the initial recommendation value with recommendation values ​​of the same type in the database to determine the target recommendation value.

[0042] In practical applications, the server compares the initial recommended value determined by the keyword in the text to be processed with the recommended value in the database to determine whether the database has a recommended value that is the same as the initial recommended value, thus determining the accuracy of the target recommended value under that keyword. Continuing with the previous example, hospitals, diseases, addresses, and examination items will be associated and compared with existing databases to provide accurate entity results. For example, when the keyword is determined to be a hospital, if the text to be processed only contains "People's Hospital," the server will extract the most recently mentioned province, city, and district name from the context of the text to be processed, such as "**Province People's Hospital" or "**Province **City People's Hospital." When the keyword is determined to be an address, if the text to be processed only contains "**Road No. 18," the server will extract the most recently mentioned address from the context of the text to be processed, such as "**Province **City **District **Road No. 18." The server will then compare these input values ​​with the database to determine if there are correct results. If multiple valid results exist, they will all be returned as the target recommended value.

[0043] To obtain recommendation values ​​corresponding to keywords mentioned in the text to be processed, target recommendation values ​​with high similarity can be achieved through similarity calculation. Specifically, comparing the initial recommendation value with recommendation values ​​of the same type in the database to determine the target recommendation value includes: Identify recommendation values ​​in the database that are of the same type as the initial recommendation values. The initial recommendation value and the recommendation value are compared to calculate their similarity, and the recommendation value that meets the preset similarity condition is determined as the target recommendation value.

[0044] Specifically, the server can determine the recommended value in the data that is the same type as the initial recommended value. For example, if the initial recommended value is the name of a hospital, then it needs to determine the hospital name stored in the database, calculate the similarity between the initial recommended value and the recommended value in the database that is the same type as the initial recommended value, and based on the calculation result, determine the recommended value that meets the preset similarity conditions as the target recommended value.

[0045] In practical applications, taking the initial recommendation value as the name of a hospital as an example, the server can determine the recommendation values ​​of multiple hospital names in the database, calculate the text similarity between the initial recommendation value and the multiple hospital name recommendation values, and if the preset similarity condition is that the similarity threshold is 80%, then the similarity calculation result that meets the 80% similarity recommendation value can be used as the final target recommendation value.

[0046] For example, if the initial recommendation value is "People's Hospital", then there will be a large number of hospital names in the database. For instance, there might be five hospital names: "First People's Hospital of Province A", "Second People's Hospital of Province A", "People's Hospital of Province A", "Children's Hospital of Province A", and "People's Hospital of City C, Province B". The initial recommendation value "People's Hospital" is then compared with the names of these five hospitals. If the similarity threshold is 80%, then "People's Hospital of Province A" is determined to have a similarity of 90%, and "First People's Hospital of Province A" and "Second People's Hospital of Province A" have similarities of 80% each. Finally, the hospital names with a similarity of 80% or higher can be determined as the target recommendation values. That is, there are three target recommendation values: "People's Hospital of Province A", "First People's Hospital of Province A", and "Second People's Hospital of Province A".

[0047] The text processing method provided in this specification calculates the similarity between the initial recommendation value and the recommendation value in the database to obtain the target recommendation value, thereby achieving accuracy in determining the recommendation value. This facilitates quick manual form filling and effectively solves the problem of time-consuming manual processing.

[0048] To further improve the efficiency of manual form filling, the obtained target recommendation values ​​can be sorted to facilitate subsequent record-keeping personnel in quickly selecting keyword input values ​​from the target recommendation values. Specifically, determining the recommendation values ​​that meet the preset similarity conditions as target recommendation values ​​includes: If it is determined that there are at least two recommended values ​​that meet the preset similarity conditions, the at least one recommended value is sorted according to a preset sorting method, and the sorted at least two recommended values ​​are determined as the target recommended values.

[0049] Specifically, if the server determines that at least two of the target recommendation values ​​meet the preset similarity conditions, it will sort at least one recommendation value according to a preset sorting method, and determine the at least two sorted recommendation values ​​as the final target recommendation values, which will be displayed on the client interface used by the recorder.

[0050] In practical applications, the target recommendation value selected based on similarity calculation can be one, two, or more. When there are more than two target recommendation values, the specific sorting order displayed on the client's interface directly affects the speed at which the recorder selects the corresponding target recommendation value. Therefore, at least two target recommendation values ​​can be sorted according to the similarity ranking method, such as from high to low or from low to high. This manual does not impose any restrictions on this method and can be set according to different application scenarios and actual needs. Finally, the sorted target recommendation values ​​are displayed on the client's interface.

[0051] Furthermore, when there are a large number of target recommended values, the client can also filter them according to a preset number threshold. For example, if the server returns 7 target recommended values ​​to the client, then the client's recommended value display box will only have five positions. The client can then filter out the top five with high similarity based on the sorting results and display them on the client. The specific display method is not limited in this embodiment of the specification.

[0052] The text processing method provided in this specification sorts the target recommended values ​​and then displays them on the client for the recorder to quickly select, eliminating the need to repeatedly search among multiple target recommended values. This not only speeds up the server's processing speed but also reduces the selection time for the recorder, further improving the recorder's work efficiency.

[0053] After returning the target recommendation value to the client, the server can then receive the selection results of the recorder regarding the target recommendation value, and then store the target recommendation value; specifically, after determining at least two sorted recommendation values ​​as the target recommendation value, the process further includes: The system returns the at least two target recommendation values ​​to the client's selection box for the keyword, receives the target recommendation value selected by the user through the selection box on the client, and stores the target recommendation value and the keyword.

[0054] The keyword selection box can be understood as the selection box that the recorder clicks when entering the keyword input value.

[0055] Specifically, the server returns at least two target recommendation values ​​to the client's selection box for the keyword. After the recorder selects the input value through the client, the server receives the target recommendation value selected by the user through the client in the selection box. This target recommendation value is the final input value to be selected for the keyword. At the same time, the target recommendation value and the keyword can be stored in the background for subsequent data analysis.

[0056] Continuing with the previous example, if there are three target recommended values ​​under the keywords displayed on the client's screen, namely "People's Hospital of Province A", "First People's Hospital of Province A", and "Second People's Hospital of Province A", then these three target recommended values ​​can be displayed in the candidate box of the keyword hospital. The recorder can click "People's Hospital of Province A" according to the content of the text to be processed, and then "People's Hospital of Province A" will be entered into the candidate box and sent to the server. At this time, the target recommended value that the server can receive is "People's Hospital of Province A", and then further store "People's Hospital of Province A" and the keyword hospital in the backend system.

[0057] The text processing method provided in this specification, by returning the target recommended value to the client's selection box, simultaneously receiving the target recommended value selected by the recorder, and storing the target recommended value, achieves automatic filling of keyword input values ​​and speeds up the processing speed of the recorder.

[0058] Furthermore, in one embodiment of the text processing method provided in this specification, there is another situation where the data content in the database on the server is not updated in a timely manner, and the selected target recommended value has been modified. In this case, the recorder can adjust the input value accordingly after selecting the target recommended value. Specifically, one embodiment of this specification also includes: The system receives the modified input value from the user through the client, after the user adjusts the target recommended value in the selection box, and stores the modified input value and the keyword.

[0059] Modifying the input value can be understood as the input value after the recorder has made corresponding adjustments to the target recommended value.

[0060] In practical applications, when there is a difference between the target recommended value displayed on the client and the actual input value of the keyword, the recorder can directly modify the input value of the keyword in the client's input box. The client will then return the modified input value of the keyword to the server. After receiving the modified input value of the target recommended value from the client's selection box, the server can store the modified input value and the corresponding keyword.

[0061] Furthermore, after the server stores the modified input value and its corresponding keywords, it not only facilitates subsequent data analysis but also allows for database updates. This ensures that when the modified input value is applied again, the target recommended value can be output accurately without requiring manual modification, thus reducing subsequent manual operations.

[0062] For example, if the target recommendation value selected by the user on the client side is "People's Hospital of Province A", but the user knows that the hospital has recently changed its official name and should fill in "Central People's Hospital of Province A", then the user can modify the target recommendation value and obtain the modification input. The server can then store the modification input and the keyword hospital as data points in the background system.

[0063] The text processing method provided in this specification allows modification of the selected target recommendation value according to the actual situation, and returns the modified target recommendation value to the server. The server stores the modified target recommendation value and keywords, which not only updates the database but also facilitates subsequent recorders to quickly determine the final target recommendation value, reduces the subsequent modification work of the target recommendation value, and speeds up the work efficiency of recorders.

[0064] Furthermore, another embodiment of the text processing method provided in this specification further includes: If it is determined that the at least one target recommendation value is empty, the system receives the target input value of the user through the client in the keyword selection box, and stores the target input value and the keyword.

[0065] The target input value can be understood as the target content that the recorder enters into the keyword selection box based on the content of the conversation with the user.

[0066] Specifically, if the server determines that the target recommendation value is empty, it means that the target recommendation value corresponding to the keyword has not been determined in real time from the audio of the video conversation. In this case, the recorder can enter the target input value in the candidate box of the keyword through the client. After receiving the target input value, the server can store the target input value and the keyword. It should be noted that the situation where the target recommendation value is empty may be that the target recommendation value cannot be matched in the text to be processed in real time, or that the content of the keyword is too complex and the speech recognition system cannot accurately convert it into text information. This specification does not impose too many limitations on this embodiment.

[0067] In practical applications, for the content entered by the recorder, the target input value and keywords can be stored in the database later to update the database.

[0068] The text processing method provided in this specification, when it is determined that the target recommended value displayed in the selection box is empty, can update the database by having the recorder manually input the target input value, so that the target recommended value will not be empty in the future, thus speeding up the work efficiency of the recorder.

[0069] Below is a diagram illustrating the client-side interface where personnel fill in keywords during an insurance claims interview. (See also...) Figure 2 , Figure 2 The diagram shows the effect of displaying hospital recommendation values ​​on the client side of the text processing method provided in the embodiments of this specification.

[0070] Figure 2 This is a diagram illustrating the record-keeping process during face-to-face interviews between personnel and users. Figure 2 The data displayed shows the records of medical visits by the personnel. Specifically, in the selection box A with the keyword "first-time hospital," the target recommended value determined by the server is displayed. Figure 2 The right side is dialog box B, which records the dialogue between the person and the user. For example, the dialogue might contain "**Provincial Affiliated Hospital" and "**Provincial People's Hospital". The client converts the dialogue into speech-to-text and extracts the text according to preset text extraction conditions to obtain the text to be processed. The server then inputs the text to be processed into the detection model and keyword extraction model to determine the initial recommendation value. Once a recommendation value with high similarity to the initial recommendation value is found in the database, the determined target recommendation value can be displayed in the client's selection box, for example... Figure 2 The server will determine the target recommendation values ​​as "**Provincial People's Hospital" and "**Provincial **University Affiliated People's Hospital", and display them in the output. Figure 2 In the selection box A.

[0071] Based on this, the method for selecting target recommendation values ​​for keywords provided in the embodiments of this specification processes the text converted from the dialog box, performs topic detection and keyword extraction on the text to be processed, and then determines the target recommendation value and displays it in the client for the recorder to select. This realizes intelligent recommendation of key content, which not only improves the efficiency of manual record-keeping work, but also greatly reduces the time of face-to-face interviews.

[0072] See Figure 3 , Figure 3 The diagram illustrates the effect of the recommended time value displayed on the client side of the text processing method provided in the embodiments of this specification.

[0073] Figure 3 This is a diagram illustrating the record-keeping process during face-to-face interviews between personnel and users. Figure 3The data displayed shows the records of medical visits by the personnel. Specifically, in the selection box A with the keyword "first medical visit time," the target recommended value determined by the server is displayed. Figure 3 The right side is dialog box B, which records the conversation between the person and the user. For example, the conversation might contain "May 26, 2020". The client converts the conversation into speech-to-text and extracts the text according to preset text extraction criteria to obtain the text to be processed. The server then inputs the text to be processed into the detection model and keyword extraction model to determine the initial recommendation value. Once a recommendation value with high similarity to the initial recommendation value is found in the database, the determined target recommendation value can be displayed in the client's selection box, for example... Figure 3 The server determined the target recommended values ​​as "2020-05-26", "2020-01-01", and "2020-05-01", and displayed them in the output. Figure 3 In the selection box A.

[0074] Based on this, the method for selecting target recommendation values ​​for keywords provided in the embodiments of this specification processes the text converted from the dialog box, performs topic detection and keyword extraction on the text to be processed, and then determines the target recommendation value and displays it in the client for the recorder to select. This realizes intelligent recommendation of key content, which not only improves the efficiency of manual record-keeping work, but also greatly reduces the time of face-to-face interviews.

[0075] The following is in conjunction with the appendix Figure 4 Taking the text processing method provided in this specification in the application of video face-to-face interviews as an example, the text processing method will be further explained. Among other things, Figure 4 The present specification shows a flowchart of a text processing method according to an embodiment, which specifically includes the following steps.

[0076] It should be noted that traditional methods cannot effectively improve the efficiency of record-keeping in video interviews. Information such as time, province, city, district, and detailed address often requires significant time to fill in and confirm. The text processing method provided in this specification uses an intelligent keyword recommendation function to identify the topic of the conversation and information such as time, address, hospital, disease, and examination items mentioned in the conversation text. It then identifies the keyword information and displays it on the corresponding record form, thereby improving the overall efficiency of the record-keeping and inquiry process for record-keepers.

[0077] In addition, for the standalone keyword extraction task, this solution uses NER and entity chaining methods; however, depending on the actual project needs, keyword extraction can be achieved using regular expression rules or a keyword dictionary.

[0078] This specification describes a three-terminal interaction method involving a user, a client, and a server. The client includes a user client and a recorder client; the server includes a reading module, a detection module, an extraction module, a storage module, and a database.

[0079] Step 402: The user makes a video call with the recording personnel through the user client.

[0080] Step 404: The user client converts the acquired speech into text.

[0081] Specifically, the voice recordings of the conversation between the recorder and the user are first converted into text using ASR technology, which also includes the role information of the user and the recorder. This text is then sent to the backend server system via the SDK.

[0082] Step 406: The user client calls the server algorithm interface to send the converted text and user identity information to the server's reading module.

[0083] Step 408: The server's reading module reads the converted speech text, obtains the text to be processed, and calls the intelligent minutes algorithm interface to send the text to be processed to the server's detection module.

[0084] Specifically, the detection model can be a topic detection model, which can determine which stage the current conversation has reached, such as asking about life trajectory or medical treatment process, and give the topic type result.

[0085] Step 410: The server's reading module stores the read voice text into the server's storage module.

[0086] Step 412: The text type output by the server's detection module and the text to be processed are sent to the server's extraction module.

[0087] Step 414: The server extraction module extracts keywords to obtain text-type keywords.

[0088] Specifically, the keyword extraction model can achieve entity recognition. It uses NER and entity chaining model to identify and extract keywords from the current and context text. Currently, the keywords include five types: hospital, disease, time, address, and examination items.

[0089] Step 416: The server extraction module returns text-type keywords and target recommendation values ​​to the server's reading module.

[0090] Step 418: The server's reading module returns the algorithm result to the user client, and the user client stores the latest algorithm result.

[0091] Step 420: The recorder in the user's record clicks the keyword selection box in the interview record.

[0092] Step 422: Record the target recommended value corresponding to the selection box displayed on the personnel's client.

[0093] Step 424: The database on the server filters valid entities and returns them to the user client.

[0094] Specifically, during the screening and comparison process, hospitals, diseases, addresses, and examination items are linked and compared with existing knowledge bases to provide accurate entity results. For hospitals and addresses, contextual information is added for association. For example, if the dialogue only contains "People's Hospital" or "No. 18, ** Road," the most recently mentioned province, city, or district will be extracted from the context to complete the name, such as "**Province People's Hospital," "**Province **City People's Hospital," or "**Province **City **District **Road No. 18," before comparing it with the hospital / address knowledge base to provide the correct result. If multiple valid results exist, they will all be returned as recommendations. For time-related terms like "last year," "the year before last," "yesterday," and "May," the specific year and date will be identified based on the current date.

[0095] Furthermore, within a single topic, such as a medical experience, there may be multiple medical records, each with corresponding entities such as hospital and disease. Therefore, the recommendation value displays a list of entities of the same type within the same topic, showing the first five results.

[0096] Step 426: The recorder client displays the target recommended value to the recorder among the users.

[0097] Specifically, the recommended value list (including topic information) is returned to the client. The topics detected by the algorithm are associated with the client's form structure, and the recommended values ​​are displayed in the form floating window corresponding to the hospital or time.

[0098] Step 428: The recorder in the user's client selects the target recommended value, or manually enters the keyword input value, or modifies the target recommended value.

[0099] Specifically, when the recorder moves the cursor over a form, a list of recommended values ​​will be displayed in a floating window. Selecting a recommended value will automatically fill it into the form.

[0100] Step 430: The server's database stores records the user ID, topic, input value, and keywords selected by the user.

[0101] Specifically, the recorder moves the cursor away from the floating window in the client and the selected information is stored as embedded data on the server for subsequent data analysis.

[0102] The text processing method provided in this specification achieves the acquisition of target recommendation values ​​through interaction between the user, client, and server. During video interviews, it identifies topic categories and extracts keyword information from real-time dialogue text, including time, hospital, disease, detailed address, and examination items. Time entities can automatically identify the corresponding year and specific date for last year, the year before last, yesterday, and the day after tomorrow; hospitals and addresses can be identified as "People's Hospital" or "No. 18, ** Road," automatically identified as "**Province People's Hospital" or "**Province **City **District **Road No. 18" based on the contextual province, city, and district address information; multiple different models (topic detection, NER, entity chaining) are integrated during the intelligent recommendation process; the algorithm recognition results are automatically displayed on the corresponding record forms, and record forms containing different topics only display the keywords identified under that topic.

[0103] Corresponding to the above method embodiments, this specification also provides embodiments of a text processing device. Figure 5 A schematic diagram of the structure of a text processing apparatus according to one embodiment of this specification is shown. Figure 5 As shown, the device includes: The text acquisition module 502 is configured to input the acquired text to be processed into the detection model to obtain the text type; The keyword extraction module 504 is configured to input the text to be processed and the text type into the keyword extraction model to obtain the keywords of the text type; The recommendation value determination module 506 is configured to determine an initial recommendation value in the text to be processed based on the keywords, and compare the initial recommendation value with recommendation values ​​of the same type in the database to determine a target recommendation value.

[0104] Optionally, the recommended value determination module 506 is further configured to: Identify recommendation values ​​in the database that are of the same type as the initial recommendation values. The initial recommendation value and the recommendation value are compared to calculate their similarity, and the recommendation value that meets the preset similarity condition is determined as the target recommendation value.

[0105] Optionally, the device further includes: The receiving client processes the acquired voice data based on a preset voice processing method to determine the voice text, and then extracts the voice text according to preset text extraction conditions to obtain the text to be processed; or The system receives initial text edited by the user through the client and extracts the initial text according to preset text extraction conditions to obtain the text to be processed.

[0106] Optionally, the recommended value determination module 506 is further configured to: If it is determined that there are at least two recommended values ​​that meet the preset similarity conditions, the at least one recommended value is sorted according to a preset sorting method, and the sorted at least two recommended values ​​are determined as the target recommended values.

[0107] Optionally, the device further includes: The system returns the at least two target recommendation values ​​to the client's selection box for the keyword, receives the target recommendation value selected by the user through the selection box on the client, and stores the target recommendation value and the keyword.

[0108] Optionally, the device further includes: The system receives the modified input value from the user through the client, after the user adjusts the target recommended value in the selection box, and stores the modified input value and the keyword.

[0109] Optionally, the device further includes: If it is determined that the at least one target recommendation value is empty, the system receives the target input value of the user through the client in the keyword selection box, and stores the target input value and the keyword.

[0110] The text processing device provided in this specification obtains the initial recommendation value of keywords in the text to be processed by passing the acquired text through a detection model and a keyword extraction model, and then obtains the target recommendation value. The target recommendation value can be displayed in the input box recorded by the recorder. This realizes the automatic identification of keyword information from the real-time dialogue text of the video interview and displays it in the form floating window of the inquiry record, thus obtaining intelligent recommendations of key content. This not only improves the efficiency of manual recording work, but also greatly reduces the interview time.

[0111] The above is an illustrative scheme of a text processing device according to this embodiment. It should be noted that the technical solution of this text processing device and the technical solution of the above-described text processing method belong to the same concept. For details not described in detail in the technical solution of the text processing device, please refer to the description of the technical solution of the above-described text processing method.

[0112] Figure 6 A structural block diagram of a computing device 600 according to one embodiment of this specification is shown. The components of the computing device 600 include, but are not limited to, a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and a database 650 is used to store data.

[0113] The computing device 600 also includes an access device 640, which enables the computing device 600 to communicate via one or more networks 660. Examples of these networks include a Public Switched Telephone Network (PSTN), a Local Area Network (LAN), a Wide Area Network (WAN), a Personal Area Network (PAN), or a combination of communication networks such as the Internet. The access device 640 may include one or more of any type of wired or wireless network interface (e.g., a Network Interface Card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) interface, a Wi-MAX interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0114] In one embodiment of this specification, the above-described components of the computing device 600 and Figure 6 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 6 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0115] The computing device 600 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or PCs. The computing device 600 can also be a mobile or stationary server.

[0116] The processor 620 is configured to execute the following computer-executable instructions, wherein the processor executes the computer-executable instructions to implement the steps of the text processing method.

[0117] The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the above-described text processing method belong to the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the above-described text processing method.

[0118] An embodiment of this specification also provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the steps of the text processing method.

[0119] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium and the technical solution of the above-described text processing method belong to the same concept, and all details not described in detail in the technical solution of the storage medium can be found in the description of the technical solution of the above-described text processing method.

[0120] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0121] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0122] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0123] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0124] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A text processing method, comprising: inputting an obtained to-be-processed text into a detection model to obtain a text type; inputting the to-be-processed text and the text type into a keyword extraction model to obtain a keyword of the text type, wherein the keyword extraction model processes the to-be-processed text and the text type to obtain the keyword of the text type; determining an initial recommended value based on the keyword in the to-be-processed text, and comparing the initial recommended value with recommended values of the same type as the initial recommended value in a database to determine a target recommended value.

2. The text processing method of claim 1, before the inputting an obtained to-be-processed text into a detection model, further comprising: receiving a voice text determined by a client based on a preset voice processing manner after processing obtained voice data, and extracting the voice text according to a preset text extraction condition to obtain a to-be-processed text; or receiving an initial text edited by a user through a client, and extracting the initial text according to a preset text extraction condition to obtain a to-be-processed text.

3. The text processing method of claim 2, after the determining at least two recommended values in sequence as target recommended values, further comprising: returning the at least two target recommended values to a to-be-selected box of the client for the keyword, and receiving a target recommended value selected by a user through the to-be-selected box of the client, and storing the target recommended value and the keyword.

4. The text processing method of claim 3, further comprising: receiving a modified input value adjusted by the user through the client for a target recommended value in the to-be-selected box, and storing the modified input value and the keyword.

5. The text processing method of claim 3, further comprising: in a case where the at least one target recommended value is empty, receiving a target input value of the user through the client in the to-be-selected box of the keyword, and storing the target input value and the keyword.

6. A text processing apparatus, comprising: a text obtaining module configured to input an obtained to-be-processed text into a detection model to obtain a text type; a keyword extraction module configured to input the to-be-processed text and the text type into a keyword extraction model to obtain a keyword of the text type, wherein the keyword extraction model processes the to-be-processed text and the text type to obtain the keyword of the text type; a recommended value determining module configured to determine an initial recommended value based on the keyword in the to-be-processed text, and compare the initial recommended value with recommended values of the same type as the initial recommended value in a database to determine a target recommended value.

7. A computing device, comprising: a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, wherein the processor executes the computer executable instructions to implement the steps of the text processing method of any one of claims 1-5.

8. A computer readable storage medium storing computer instructions, which, when executed by a processor, implement the steps of the text processing method of any one of claims 1-5.

Images