Prompter Method and Device, Electronic Device, and Storage Medium

Abstract

An embodiment of the present application provides a prompting method, a device, an electronic device, and a storage medium, belonging to the field of artificial intelligence technology. The method includes: receiving a question search request; wherein the question search request includes a trigger request and a target question; extracting keywords from the target question according to the trigger request to obtain target keywords; performing label matching search on the target keywords according to a preset target label hierarchy to obtain candidate question-and-answer data; wherein the candidate question-and-answer data includes candidate questions and candidate answers; extracting keywords from the candidate questions to obtain candidate keywords; calculating a matching degree according to the candidate keywords and the target keywords to obtain a target matching score; performing answer screening processing on the candidate answers according to the target matching score to obtain a target answer; and recommending and displaying the target answer. The embodiment of the present application can improve the accuracy of question answering during prompting.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to a prompter method and device, an electronic device, and a storage medium. Background Art

[0002] Currently, in order to meet the Q&A needs in on-site conversation scenarios, the person being questioned often needs to consult a large amount of relevant materials in advance to answer different questions raised by the questioner. However, this method is often limited by the memory and on-the-spot reaction ability of the person being questioned, which will affect the accuracy of answering questions. Therefore, how to improve the accuracy of answering questions has become a technical problem to be solved urgently. Summary of the Invention

[0003] The main purpose of the embodiments of this application is to propose a prompter method and device, an electronic device, and a storage medium, which can improve the accuracy of answering questions when prompting.

[0004] To achieve the above object, a first aspect of the embodiments of this application proposes a prompter method, and the method includes:

[0005] Receiving a question search request; wherein, the question search request includes a trigger request and a target question;

[0006] Extracting target keywords from the target question according to the trigger request;

[0007] Performing label matching search on the target keywords according to a preset label hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers;

[0008] Extracting candidate keywords from the candidate questions;

[0009] Calculating a target matching score according to the candidate keywords and the target keywords;

[0010] Performing answer screening processing on the candidate answers according to the target matching score to obtain a target answer;

[0011] Recommending and displaying the target answer.

[0012] In some embodiments, the step of extracting target keywords from the target question according to the trigger request includes:

[0013] Invoking a preset keyword extraction model according to the trigger request;

[0014] Extracting initial keywords from the target question through the keyword extraction model;

[0015] Perform keyword screening on the initial keyword according to preset screening conditions to obtain the target keyword.

[0016] In some embodiments, the preset tag hierarchy includes multiple target tags. The step of performing tag matching search on the target keyword according to the preset tag hierarchy to obtain candidate Q&A data includes:

[0017] Perform tag matching search on the target keyword according to the target tag to obtain the matching probability value for each preset tag category;

[0018] Perform Q&A screening on the reference Q&A data in the target tag according to the matching probability value to obtain the candidate Q&A data.

[0019] In some embodiments, the step of calculating the matching degree between the candidate keyword and the target keyword to obtain the target matching score includes:

[0020] Calculate the similarity between the target keyword and the candidate keyword to obtain the similarity matching score;

[0021] Calculate the search matching degree between the target keyword and the candidate keyword to obtain the search matching score;

[0022] Perform weighted calculation on the similarity matching score and the search matching score through a preset adjustment factor to obtain the target matching score.

[0023] In some embodiments, the step of performing answer screening on the candidate answer according to the target matching score to obtain the target answer includes:

[0024] Compare the target matching score with a preset matching score threshold;

[0025] Use the candidate answer with the target matching score greater than the matching score threshold as the target answer.

[0026] In some embodiments, before the step of performing tag matching search on the target keyword according to the preset tag hierarchy to obtain candidate Q&A data, the method further includes pre-constructing the preset tag hierarchy, specifically including:

[0027] Obtain reference Q&A text;

[0028] Extract keywords from the reference Q&A text to obtain reference keywords;

[0029] Input the reference keywords into a pre-constructed initial tag hierarchy;

[0030] Perform label matching on the reference keyword through the initial label hierarchy to obtain label data corresponding to the reference keyword;

[0031] Optimize the initial label hierarchy according to the label data to obtain the preset label hierarchy.

[0032] In some embodiments, before the step of pre-constructing the preset label hierarchy, the method further includes pre-constructing the initial label hierarchy, specifically including:

[0033] Obtain historical Q&A texts;

[0034] Extract keywords from the historical Q&A texts to obtain historical keywords;

[0035] Construct the initial label hierarchy according to the historical keywords and preset label categories.

[0036] To achieve the above object, a second aspect of the embodiments of the present application proposes a prompter device, the device includes:

[0037] A search request acquisition module, configured to receive a question search request; wherein, the question search request includes a trigger request and a target question;

[0038] A target keyword extraction module, configured to extract keywords from the target question according to the trigger request to obtain target keywords;

[0039] A candidate Q&A data acquisition module, configured to perform label matching search on the target keyword according to the preset label hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers;

[0040] A candidate keyword extraction module, configured to extract keywords from the candidate questions to obtain candidate keywords;

[0041] A matching score calculation module, configured to calculate a matching degree according to the candidate keywords and the target keywords to obtain a target matching score;

[0042] A screening module, configured to perform answer screening processing on the candidate answers according to the target matching score to obtain a target answer;

[0043] A prompter display module, configured to recommend and display the target answer.

[0044] To achieve the above object, a third aspect of the embodiments of the present application provides an electronic device, which includes a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing the connection and communication between the processor and the memory. When the program is executed by the processor, the method described in the first aspect above is realized.

[0045] To achieve the above object, a fourth aspect of the embodiments of the present application provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to realize the method described in the first aspect above.

[0046] The prompting method and device, electronic device and storage medium proposed by the present application receive a question search request; wherein, the question search request includes a trigger request and a target question; then, keyword extraction is performed on the target question according to the trigger request to obtain target keywords. This method can identify key information in the target question and reduce the difficulty of question search. Furthermore, label matching search is performed on the target keywords according to a preset label hierarchy, and it is relatively convenient to match candidate Q&A data corresponding to the target keywords. The candidate Q&A data includes candidate questions and candidate answers. After obtaining the candidate Q&A data, keyword extraction is performed on the candidate questions to obtain candidate keywords, the matching degree is calculated according to the candidate keywords and the target keywords to obtain a target matching score, and finally, answer screening processing is performed on the candidate answers according to the target matching score to obtain a target answer, and the target answer is recommended and displayed. Through the matching score calculation, the relevance between different candidate keywords and the target keywords can be clearly determined, and the candidate keyword most relevant to the target keyword can be obtained. Therefore, the candidate answer corresponding to the candidate keyword is used as the target answer, and the target answer is recommended and displayed to the person being questioned, which can make the generated target answer more in line with the current question search requirements, enabling the person being questioned to answer the corresponding question in a timely and accurate manner according to the target answer, and improving the accuracy of question answering during prompting. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a flowchart of the prompting method provided by the embodiments of the present application;

[0048] Figure 2 is Figure 1 a flowchart of step S102 in

[0049] Figure 3 is another flowchart of the prompting method provided by the embodiments of the present application;

[0050] Figure 4It is another flowchart of the teleprompter method provided by the embodiments of the present application;

[0051] Figure 5 It is Figure 1 the flowchart of step S103 in

[0052] Figure 6 It is Figure 1 the flowchart of step S105 in

[0053] Figure 7 It is Figure 1 the flowchart of step S106 in

[0054] Figure 8 It is the structural schematic diagram of the teleprompter device provided by the embodiments of the present application;

[0055] Figure 9 It is the hardware structural schematic diagram of the electronic device provided by the embodiments of the present application. Detailed implementation manners

[0056] In order to make the objectives, technical solutions and advantages of the present application more clear and understandable, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] It should be noted that although functional module division is performed in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order from the module division in the device or the order in the flowchart. Terms such as "first" and "second" in the description, claims and the above-mentioned drawings are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence.

[0058] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0059] First, several nouns involved in the present application are analyzed:

[0060] Artificial Intelligence (AI): It is a new technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence. Artificial intelligence is a branch of computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. The research in this field includes robots, speech recognition, image recognition, natural language processing, and expert systems, etc. Artificial intelligence can simulate the information process of human consciousness and thinking. It also refers to the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results.

[0061] Natural Language Processing (NLP): NLP uses computers to process, understand, and apply human languages (such as Chinese, English, etc.). NLP is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, and is often referred to as computational linguistics. Natural language processing includes syntactic analysis, semantic analysis, discourse understanding, etc. Natural language processing is commonly used in technical fields such as machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, information intention recognition, information extraction and filtering, text classification and clustering, public opinion analysis, and opinion mining. It involves data mining related to language processing, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computing.

[0062] Information Extraction (NER): It is a text processing technology that extracts factual information such as specified types of entities, relationships, and events from natural language texts and forms structured data output. Information extraction is a technology for extracting specific information from text data. Text data is composed of some specific units, such as sentences, paragraphs, and texts. Text information is exactly composed of some small specific units, such as characters, words, phrases, sentences, paragraphs, or combinations of these specific units. Extracting noun phrases, personal names, place names, etc. from text data are all text information extraction. Of course, the information extracted by text information extraction technology can be various types of information.

[0063] Web crawler (also known as web spider, web robot, and more commonly called web chaser in the FOAF community): A web crawler is a program or script that automatically fetches information from the World Wide Web according to certain rules.

[0064] Automatic Speech Recognition (ASR): Automatic speech recognition technology is a technology that converts human speech into text. The input of speech recognition is generally a time-domain speech signal, which is mathematically represented by a series of vectors indicating the signal length (length T) and dimension (dimension d). The output of this automatic semantic recognition technology is text, represented by a series of tokens indicating the field length (length N) and different tokens.

[0065] Trie tree: Also known as a word search tree or a key tree, it is a tree structure and a variant of a hash tree. A typical application is for counting and sorting a large number of strings (but not limited to strings), so it is often used by search engine systems for text word frequency statistics. Its advantages are: minimizing unnecessary string comparisons. The core idea of a Trie is to trade space for time. By using the common prefixes of strings to reduce the overhead of query time to achieve the purpose of improving efficiency. Three basic properties of a prefix tree: (1) The root node does not contain a character, and each node except the root node contains only one character. (2) From the root node to a certain node, the characters passed through on the path are concatenated to form the string corresponding to that node. (3) All child nodes of each node contain different characters.

[0066] Currently, in order to meet the Q&A needs of on-site conversation scenarios, the person being questioned often needs to consult a large number of relevant materials in advance and directly answer or refer to them according to memory on-site to deal with different questions raised by the questioner. This method often requires the person being questioned to have a basic understanding and strong memory in various professional fields, and also requires those who prepare the reference materials to be as refined and accurate as possible. In fact, there is a contradiction that the more reference materials there are, the higher the probability of "hitting" the question, but the more materials the person being questioned needs to read on-site, the lower the possibility of remembering all the content, which affects the accuracy of answering questions. Therefore, how to improve the accuracy of answering questions has become a technical problem to be solved urgently.

[0067] Based on this, the embodiments of the present application provide a prompter method and device, a prompter, an electronic device, and a storage medium, which can improve the accuracy of answering questions when prompting.

[0068] The prompter method and device, prompter, electronic device, and storage medium provided by the embodiments of the present application are specifically described through the following embodiments. First, the prompter method in the embodiments of the present application is described.

[0069] Embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is a theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results.

[0070] Artificial intelligence basic technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation / interaction systems, mechatronics, etc. Artificial intelligence software technologies mainly include several major directions such as computer vision technology, robotics, biometric technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0071] The teleprompter method provided by embodiments of the present application relates to the field of artificial intelligence technology. The teleprompter method provided by embodiments of the present application can be applied to a terminal, or to a server side, or can also be software running on a terminal or a server side. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc.; the server side can be configured as an independent physical server, or can be configured as a server cluster or distributed system composed of multiple physical servers, or can also be configured as a cloud server providing 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, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the teleprompter method, etc., but is not limited to the above forms.

[0072] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and so on. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in a distributed computing environment, where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0073] Figure 1 is an optional flowchart of the teleprompter method provided by embodiments of the present application. Figure 1The method in [it] may include but is not limited to steps S101 to S107.

[0074] Step S101, receiving a question search request; wherein, the question search request includes a trigger request and a target question;

[0075] Step S102, extracting keywords from the target question according to the trigger request to obtain target keywords;

[0076] Step S103, performing label matching search on the target keywords according to a preset label hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers;

[0077] Step S104, extracting keywords from the candidate questions to obtain candidate keywords;

[0078] Step S105, calculating a matching degree according to the candidate keywords and the target keywords to obtain a target matching score;

[0079] Step S106, performing answer screening processing on the candidate answers according to the target matching score to obtain a target answer;

[0080] Step S107, recommending and displaying the target answer.

[0081] Steps S101 to S107 of the embodiments of the present application extract keywords from the target question according to the trigger request to obtain target keywords, which can identify key information in the target question and reduce the difficulty of question search. Performing label matching search on the target keywords according to the preset label hierarchy can more conveniently match candidate Q&A data corresponding to the target keywords. Among them, the candidate Q&A data includes candidate questions and candidate answers. After obtaining the candidate Q&A data, keywords are extracted from the candidate questions to obtain candidate keywords, and the matching degree is calculated according to the candidate keywords and the target keywords. Through the matching score calculation, the relevance between different candidate keywords and the target keywords can be clearly determined, and the candidate keyword most relevant to the target keyword can be obtained. Thus, the candidate answer corresponding to the candidate keyword is used as the target answer, and the target answer is recommended and displayed to the person being questioned, which can make the generated target answer more in line with the current question search requirements, enabling the person being questioned to answer the corresponding question in a timely and accurate manner according to the target answer, and can improve the accuracy of question answering when prompting.

[0082] In step S101 of some embodiments, according to the question given by the questioner, the user can input the corresponding question on the teleprompter, so that the teleprompter can receive the corresponding question search request; wherein, the question search request includes a trigger request and a target question. It should be noted that when the user inputs the corresponding question, the user can input the corresponding question by editing text, or by voice, or in other ways to input the question into the teleprompter for searching, without limitation.

[0083] For example, when the user inputs the question of the questioner into the teleprompter by voice input, the teleprompter can use ASR technology to parse and process the voice question. Specifically, the ASR technology adopted by the teleprompter uses the basic elements of sound as units. Since different words are composed of different phonemes, by identifying which phonemes exist in the input voice, and then combining these phonemes into the recognized text, the corresponding voice text can be obtained. Therefore, the teleprompter is provided with a phoneme mapping table according to the basic elements of sound, and this phoneme mapping table can reflect the corresponding relationship between the voice signal and the phoneme. According to this phoneme mapping table, the voice signal of the target question input by voice is subjected to voice recognition processing, the phonemes existing in the voice signal of the target question are recognized, and then these phonemes are combined into the recognized text to obtain the voice feature word segment corresponding to each voice signal, and the voice question is converted into a text question, so as to search for answers to this text question, making the teleprompter process multimodal and improving the diversity and universality of the answer search method.

[0084] Please refer to Figure 2 , in some embodiments, step S102 may include but is not limited to steps S201 to S203:

[0085] Step S201, calling a preset keyword extraction model according to the trigger request;

[0086] Step S202, extracting keywords from the target question through the keyword extraction model to obtain initial keywords;

[0087] Step S203, performing keyword screening processing on the initial keywords according to preset screening conditions to obtain target keywords.

[0088] In steps S201 and S202 of some embodiments, the teleprompter calls a preset keyword extraction model according to a trigger request instruction. Among them, the preset keyword extraction model is an ESIM model based on a trie tree. The training process of this keyword extraction model can specifically include: obtaining sample question data, which is labeled with question part-of-speech category labels, question intent category labels, etc., performing segmentation processing on the sample question data to obtain sample question fields, traversing all sample question fields through the preset trie tree to obtain sample question keywords. Matching the sample question keywords with the question part-of-speech category labels and question intent category labels to construct a keyword comparison table. According to the matching relationship between the sample question keywords and the sample questions in the keyword comparison table, updating the loss function of the keyword extraction model until the number of iterations meets the preset iteration condition, stopping updating the loss function of the keyword extraction model, and obtaining the final keyword extraction model.

[0089] In step S202 of some embodiments, first, fields lengths, statement categories, etc. that meet the requirements can be preset in the Jieba tokenizer in the keyword extraction model, so as to perform segmentation processing on the target question through the Jieba tokenizer to obtain multiple target question fields. Since the corresponding question keywords are pre-stored in the preset trie tree in the keyword extraction model, and the question keywords are used as the child nodes of the trie tree, input the target question fields into the keyword extraction model to extract the initial keywords. Starting from the root node in the trie tree, gradually search downward, traverse all leaf nodes, calculate the similarity between the question keywords of each leaf node and the current target question field, and extract the question keywords of the leaf nodes with a similarity greater than the preset threshold as the initial keywords.

[0090] In step S203 of some embodiments, keyword screening processing is performed on the initial keywords according to the preset screening conditions. Among them, the preset screening conditions can be that the word segment length of the initial keyword meets the preset length, or the similarity of the initial keyword is the highest, etc., or others. For example, using techniques such as TF-IDF and TextRank, screening and sorting the initial keywords according to the screening conditions to obtain keywords that can better reflect the current corpus and are related to the theme and scenario, and taking the initial keywords that meet the screening conditions as the target keywords.

[0091] Please refer to Figure 3 , in some embodiments, before step S103, the teleprompter method includes pre-constructing a preset label hierarchy, specifically including steps S301 to S305:

[0092] Step S301, obtaining reference Q&A text;

[0093] Step S302, performing keyword extraction on the reference Q&A text to obtain reference keywords;

[0094] Step S303: Input the reference keywords into the initially constructed tag hierarchy.

[0095] Step S304: Perform tag matching on the reference keywords through the initial tag hierarchy to obtain the tag data corresponding to the reference keywords.

[0096] Step S305: Optimize the initial tag hierarchy according to the tag data to obtain the preset tag hierarchy.

[0097] In step S301 of some embodiments, reference Q&A texts can be obtained by writing a web crawler to perform targeted data crawling after setting the data source. For example, the reference Q&A texts can be Q&A materials pre-organized according to the requirements of the current press conference, that is, the reference Q&A texts are some newly added Q&A materials targeted, in addition to the questions raised by the questioners in previous press conferences and the professional answer materials corresponding to the questions. It should be noted that the reference Q&A texts are natural language texts, and the reference Q&A texts include reference questions and reference answers.

[0098] In step S302 of some embodiments, first obtain multiple reference Q&A paragraphs in the reference Q&A text based on the optical character recognition technology (OCR), and then use natural language processing technology (NLP) to perform text parsing on each reference Q&A paragraph to obtain multiple reference keywords. Specifically, the above-mentioned pre-trained keyword extraction model can be used to extract keywords from the reference Q&A text. First, the required field length, sentence category, etc. can be preset in the Jieba tokenizer in the keyword extraction model, so as to perform segmentation processing on the reference Q&A text through the Jieba tokenizer to obtain multiple reference Q&A fields. Since the corresponding Q&A keywords are pre-stored in the trie tree preset in the keyword extraction model, and the Q&A keywords are used as the child nodes of the trie tree, the reference Q&A fields are input into the keyword extraction model to extract the reference keywords. Starting from the root node in the trie tree, search downward step by step, traverse all the leaf nodes, calculate the similarity between the Q&A keywords of each leaf node and the current reference Q&A field, and extract the Q&A keywords of the leaf nodes with similarity greater than the preset threshold as the reference keywords.

[0099] In step S303 of some embodiments, the reference keywords are input into the initial tag hierarchy, where the initial tag hierarchy can be constructed according to historical Q&A texts and preset tag categories, and the initial tag hierarchy includes the questions raised by the questioners in previous press conferences, the professional answer materials corresponding to the questions, and the Q&A tags of these Q&A materials, etc.

[0100] In step S304 of some embodiments, label matching is performed on the reference keywords through the initial label hierarchy to obtain the label data corresponding to the reference keywords, where the label data includes the overall three-level labels. Specifically, the reference keywords are matched to the labels under the corresponding initial label hierarchy, thereby forming a structured display. The reference keywords are matched with the third-level label names in the initial label hierarchy. In order to quickly locate the labels corresponding to the reference Q&A text, a triple is used to define the overall three-level labels of the initial label hierarchy. The representation form of the overall three-level label is "First-level label - Second-level label - Third-level label".

[0101] For example, in some specific application scenarios, the first-level label is "Life & Health", the second-level label is "Life Insurance" under the first-level label, and the third-level label is "Operating Profit & Remaining Margin". It is specifically divided into the following three cases:

[0102] The first case: When the number of label matching results is 0, first, the first-level and second-level labels under the initial label hierarchy are matched using the newly added reference Q&A text; then, the answer keywords of the corresponding answer text in the newly added reference Q&A text are extracted, and the extracted answer keywords are matched with the third-level label names in the initial label hierarchy to obtain the overall three-level labels of the newly added reference Q&A text.

[0103] The second case: When the number of label matching results is 1, the corresponding overall three-level labels are directly obtained according to the Q&A label classification under the initial label hierarchy.

[0104] The third case: When the number of label matching results is greater than 1, the reference questions of the newly added reference Q&A text are cleaned and filtered according to the first-level and second-level labels to obtain the overall three-level labels. Specifically, for the reference keywords that may appear under multiple labels, when the keywords in the corresponding first-level and second-level labels appear in the reference questions, it is considered to belong to that label. For example: When the reference keyword extracted from the newly added reference question is "Operating Profit", this "Operating Profit" may be involved in multiple different first-level labels. If the reference keyword in the newly added reference question contains the first-level label "Technology", the overall three-level label of the newly added reference question is "Technology - Overall Technology - Technology Performance"; if the second-level label "Life Insurance" appears in the newly added reference question, the overall three-level label should be "Life & Health - Life Insurance - Operating Profit & Remaining Margin".

[0105] In step S305 of some embodiments, when optimizing the initial tag hierarchy according to the tag data, it is necessary to review and confirm the initial tag hierarchy and the Q&A tag classification. Specifically, when the reference keyword obtained by parsing is a keyword under the existing tag hierarchy, the reference Q&A pair corresponding to the reference keyword is automatically classified under the corresponding Q&A tag classification, and the current Q&A tag classification is reviewed; when the reference keyword obtained by parsing is not a keyword under the existing tag hierarchy, first, add the reference keyword to the initial tag hierarchy, update and review the current initial tag hierarchy; then automatically classify the Q&A pair corresponding to the reference keyword under the corresponding Q&A tag classification; after that, review the updated Q&A tag classification again to obtain the preset tag hierarchy.

[0106] Please refer to Figure 4 , in some embodiments, before pre-constructing the preset tag hierarchy, the prompter method further includes pre-constructing the initial tag hierarchy, specifically including steps S401 to S403:

[0107] Step S401, obtain historical Q&A texts;

[0108] Step S402, extract keywords from the historical Q&A texts to obtain historical keywords;

[0109] Step S403, construct the initial tag hierarchy according to the historical keywords and the preset tag categories.

[0110] In step S401 of some embodiments, historical Q&A texts can be obtained by writing a web crawler and performing targeted data crawling after setting the data source. For example, the data source can be the materials of press conferences held in the industry, or the industry information on the forum, etc. The historical Q&A texts can be the Q&A materials saved from past press conferences, that is, the questions raised by the questioners at past press conferences, and the corresponding professional answers, etc. It should be noted that the historical Q&A texts are natural language texts.

[0111] In step S402 of some embodiments, the process of extracting keywords from the historical Q&A texts to obtain historical keywords is basically the same as the process of extracting keywords from the reference Q&A texts to obtain reference keywords, which will not be elaborated here.

[0112] In step S403 of some embodiments, the parsed initial keywords are added to the initial tag hierarchy to construct the tag hierarchy corresponding to the historical Q&A texts. Specifically, the parsed initial keywords are added under the corresponding tag hierarchy, and the historical Q&A pairs corresponding to the initial keywords are automatically classified under the corresponding Q&A tag classification. Among them, the Q&A tag classification under the initial tag hierarchy includes: first-level tags, second-level tags, and third-level tags.

[0113] In some specific application scenarios, the first-level tags have the highest generality and are generally the refinement of business segments; the second-level tags are the business themes under the corresponding business segments in the first-level tags; the third-level tags are more specific business contents in the corresponding business themes in the second-level tags, that is, the first-level tags contain the second-level tags, and the second-level tags contain the third-level tags. The historical Q&A pairs include the historical questions corresponding to the historical keywords and the corresponding historical answers.

[0114] Please refer to Figure 5 , in some embodiments, the preset tag hierarchy includes multiple target tags, and step S103 may further include but is not limited to steps S501 to S502:

[0115] Step S501, perform tag matching search on the target keywords according to the target tags to obtain the matching probability values of each preset tag category;

[0116] Step S502, perform Q&A screening processing on the reference Q&A data in the target tags according to the matching probability values to obtain candidate Q&A data.

[0117] In step S501 of some embodiments, perform tag matching search on the target keywords according to the preset function and the target tags, where the preset function may be the softmax function. For example, create a probability distribution on each target tag through the softmax function to obtain the matching probability values of the target keywords belonging to each target tag.

[0118] In step S502 of some embodiments, extract the reference Q&A data in the target tags where the matching probability values are greater than or equal to the matching probability threshold, and use the reference Q&A data as the candidate Q&A data.

[0119] In step S104 of some embodiments, the process of extracting candidate keywords from the candidate questions is basically the same as the process of extracting the target keywords. Specifically, first, the required field lengths, statement categories, etc. can be preset in the Jieba tokenizer in the keyword extraction model, so as to perform segmentation processing on the candidate questions through the Jieba tokenizer to obtain multiple candidate question fields. Since the corresponding question keywords are pre-stored in the trie tree preset in the keyword extraction model, and the question keywords are used as the child nodes of the trie tree, input the candidate question fields into the keyword extraction model to extract the candidate keywords. Starting from the root node in the trie tree, gradually search downward, traverse all the leaf nodes, calculate the similarity between the question keywords of each leaf node and the current candidate question, and extract the question keywords of the leaf nodes with similarity greater than the preset threshold as the candidate keywords.

[0120] Please refer to Figure 6, in some embodiments, step S105 may further include, but is not limited to, steps S601 to S603:

[0121] Step S601, calculate the similarity between the target keyword and the candidate keyword to obtain a similarity matching score;

[0122] Step S602, calculate the search matching degree between the target keyword and the candidate keyword to obtain a search matching score;

[0123] Step S603, perform weighted calculation on the similarity matching score and the search matching score through a preset adjustment factor to obtain a target matching score.

[0124] In step S601 of some embodiments, when calculating the similarity between the target keyword and the candidate keyword, collaborative filtering algorithms such as the cosine similarity algorithm can be used to calculate the similarity between the target keyword and the candidate keyword. For example, first vectorize the target keyword and the candidate keyword segment to obtain the target keyword vector u and the candidate keyword vector v, and calculate the similarity matching score y1 of the target keyword and the candidate keyword according to the formula of the cosine similarity algorithm (as shown in formula 1).

[0125]

[0126] In step S602 of some embodiments, when calculating the search matching degree between the target keyword and the candidate keyword, ES search matching can be performed on the target keyword and the candidate keyword, that is, text recall is performed on the candidate keyword based on matching algorithms such as BM25, and further rough and fine sorting is performed to obtain the search matching score at this time. Taking the BM25 matching algorithm as an example, the input target keyword is segmented, and then the relevance of each field in the target keyword to the candidate keyword is calculated, and then weighted summation is performed. The search matching score y2 of the target keyword and the candidate keyword is obtained. The calculation formula is as shown in formula (2):

[0127]

[0128] Among them, Wi is a preset weight, and R(qi, d) is the relevance score of each field to the candidate keyword. For the relevance score of each field to the candidate keyword, collaborative filtering algorithms such as the cosine similarity algorithm can also be used to calculate.

[0129] In step S603 of some embodiments, dynamic weight combination can be performed on the similarity matching score and the search matching score according to the preset adjustment factor. Specifically, the adjustment factors can be α and β, and the similarity matching score and the search matching score are weighted calculated according to formula (3) to obtain the target matching score y. Formula (3) is expressed as:

[0130] Y = α * y1 + β * y2, Formula (3)

[0131] Please refer to Figure 7 , in some embodiments, step S106 may further include but is not limited to steps S701 to S702:

[0132] Step S701, compare the target matching score with a preset matching score threshold;

[0133] Step S702, use the candidate answer with a target matching score greater than the matching score threshold as the target answer.

[0134] In step S701 of some embodiments, compare the magnitude relationship between the target matching score and the preset matching score threshold, where the preset matching score threshold can be set according to the actual situation, and the matching score threshold is any value between 0 and 100, without limitation.

[0135] In step S702 of some embodiments, by comparing the magnitude relationship between the target matching score and the preset matching score threshold, use the candidate answer with a target matching score greater than the matching score threshold as the target answer.

[0136] Furthermore, in order to avoid an excessive number of target answers affecting the review efficiency, in some specific scenarios, the candidate answers with a target matching score greater than the matching score threshold can be sorted, and the candidate answers are sorted in descending order according to the target matching score, and the top five candidate answers are selected as the target answers.

[0137] In step S107 of some embodiments, the target answer can be recommended and displayed through various terminals such as a PC or a PAD, and the target answer information is updated online in real time. This method can improve the display form of traditional paper materials, meet the requirements of paperless, and also meet the prompting needs in different scenarios, improving the applicability of this prompting method.

[0138] The prompting method according to the embodiment of the present application receives a question search request; wherein, the question search request includes a trigger request and a target question; furthermore, keyword extraction is performed on the target question according to the trigger request to obtain target keywords. This method can identify the key information in the target question and reduce the difficulty of question search. Furthermore, according to the preset tag hierarchy, tag matching search is performed on the target keywords, and it is more convenient to match the candidate Q&A data corresponding to the target keywords. The candidate Q&A data includes candidate questions and candidate answers. After obtaining the candidate Q&A data, keyword extraction is performed on the candidate questions to obtain candidate keywords. The matching degree is calculated based on the candidate keywords and the target keywords to obtain a target matching score. Finally, the candidate answers are screened according to the target matching score to obtain a target answer, and the target answer is recommended and displayed. Through the matching score calculation, the relevance between different candidate keywords and the target keywords can be clearly determined, and the candidate keyword most relevant to the target keywords can be obtained. Therefore, the candidate answer corresponding to the candidate keyword is used as the target answer, and the target answer is recommended and displayed to the person being questioned, which can make the generated target answer more in line with the current question search requirements. This method can quickly and accurately achieve high-precision search for the target question, enabling the person being questioned to answer the corresponding question in a timely and accurate manner according to the target answer, and improving the accuracy of question answering during prompting.

[0139] Please refer to Figure 8 , the embodiment of the present application further provides a prompting device, which can implement the above-mentioned prompting method. The device includes:

[0140] A search request acquisition module 801, configured to receive a question search request; wherein, the question search request includes a trigger request and a target question;

[0141] A target keyword extraction module 802, configured to perform keyword extraction on the target question according to the trigger request to obtain target keywords;

[0142] A candidate Q&A data acquisition module 803, configured to perform tag matching search on the target keywords according to the preset tag hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers;

[0143] A candidate keyword extraction module 804, configured to perform keyword extraction on the candidate questions to obtain candidate keywords;

[0144] A matching score calculation module 805, configured to calculate the matching degree based on the candidate keywords and the target keywords to obtain a target matching score;

[0145] An answer screening module 806, configured to screen the candidate answers according to the target matching score to obtain a target answer;

[0146] The teleprompter display module 807 is used to recommend and display the target answer.

[0147] In some embodiments, the target keyword extraction module 802 includes:

[0148] A call unit, configured to call a preset keyword extraction model according to a trigger request;

[0149] A first keyword extraction unit, configured to extract keywords from the target question through the keyword extraction model to obtain initial keywords;

[0150] A keyword screening unit, configured to perform keyword screening processing on the initial keywords according to preset screening conditions to obtain target keywords.

[0151] In some embodiments, the teleprompter device further includes a preset tag hierarchy construction module, and the preset tag hierarchy construction module includes:

[0152] A reference Q&A text acquisition unit, configured to acquire reference Q&A text;

[0153] A second keyword extraction unit, configured to extract keywords from the reference Q&A text to obtain reference keywords;

[0154] An input unit, configured to input the reference keywords into a pre-constructed initial tag hierarchy;

[0155] A first tag matching unit, configured to perform tag matching on the reference keywords through the initial tag system to obtain tag data corresponding to the reference keywords;

[0156] An optimization unit, configured to optimize the initial tag hierarchy according to the tag data to obtain a preset tag hierarchy.

[0157] In some embodiments, the preset tag hierarchy includes multiple target tags, and the candidate Q&A data acquisition module 803 includes:

[0158] A second tag matching unit, configured to perform tag matching search on the target keywords according to the target tags to obtain a matching probability value for each preset tag category;

[0159] A Q&A screening unit, configured to perform Q&A screening processing on the reference Q&A data in the target tags according to the matching probability value to obtain candidate Q&A data.

[0160] In some embodiments, the matching score calculation module 805 includes:

[0161] A similarity calculation unit, configured to calculate the similarity between the target keyword and the candidate keyword to obtain a similarity matching score;

[0162] A matching degree calculation unit for calculating the search matching degree between a target keyword and a candidate keyword to obtain a search matching score;

[0163] A weighted calculation unit for performing weighted calculation on the similarity matching score and the search matching score through a preset adjustment factor to obtain a target matching score.

[0164] In some embodiments, the answer screening module 806 includes:

[0165] A comparison unit for comparing the target matching score with a preset matching score threshold;

[0166] A target answer determination unit for using the candidate answer with a target matching score greater than the matching score threshold as the target answer.

[0167] It should be noted that the prompter device in the embodiments of the present application is used to implement the above-mentioned prompter method. The prompter device in the embodiments of the present application corresponds to the foregoing prompter method. For the specific processing process, please refer to the foregoing prompter method and will not be elaborated herein.

[0168] The embodiments of the present application also provide a prompter, and the prompter includes the prompter device in the above embodiments. The prompter in the embodiments of the present application is used to implement the above-mentioned prompter method. The prompter in the embodiments of the present application corresponds to the foregoing prompter method. For the specific processing process, please refer to the foregoing prompter method and will not be elaborated herein.

[0169] The embodiments of the present application also provide an electronic device. The electronic device includes: a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing the connection and communication between the processor and the memory. When the program is executed by the processor, the above-mentioned prompter method is implemented. The electronic device can be any intelligent terminal including a tablet computer, an in-vehicle computer, etc.

[0170] Please refer to Figure 9 , Figure 9 which shows the hardware structure of an electronic device in another embodiment. The electronic device includes:

[0171] A processor 901, which can be implemented in ways such as a general-purpose CPU (Central Processing Unit, central processor), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0172] The memory 902 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM), etc. The memory 902 can store an operating system and other application programs. When implementing the technical solutions provided in the embodiments of this specification through software or firmware, the relevant program codes are stored in the memory 902, and the processor 901 is called to execute the prompting method of the embodiments of this application;

[0173] The input / output interface 903 is used to implement information input and output;

[0174] The communication interface 904 is used to implement communication and interaction between this device and other devices. It can communicate through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.);

[0175] The bus 905 transmits information between the various components of the device (such as the processor 901, the memory 902, the input / output interface 903, and the communication interface 904);

[0176] Among them, the processor 901, the memory 902, the input / output interface 903, and the communication interface 904 are communicatively connected to each other inside the device through the bus 905.

[0177] The embodiments of this application also provide a storage medium. The storage medium is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the above-mentioned prompting method.

[0178] As a non-transitory computer-readable storage medium, the memory can be used to store non-transitory software programs and non-transitory computer-executable programs. In addition, the memory can include high-speed random access memory, and can also include non-transitory memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transitory solid-state storage devices. In some embodiments, the memory optionally includes a memory remotely provided with respect to the processor, and these remote memories can be connected to the processor through a network. Examples of the above-mentioned network include but are not limited to the Internet, an enterprise intranet, a local area network, a mobile communication network, and combinations thereof.

[0179] The teleprompter method, teleprompter device, teleprompter, electronic device and storage medium according to the embodiments of the present application receive a question search request; wherein the question search request includes a trigger request and a target question; and then extract keywords from the target question according to the trigger request to obtain target keywords. This method can identify the key information in the target question and reduce the difficulty of question search. Furthermore, by performing label matching search on the target keywords according to the preset label hierarchy, it is possible to more conveniently match the candidate Q&A data corresponding to the target keywords. The candidate Q&A data includes candidate questions and candidate answers. After obtaining the candidate Q&A data, extract keywords from the candidate questions to obtain candidate keywords, calculate the matching degree between the candidate keywords and the target keywords to obtain a target matching score, and finally perform answer screening processing on the candidate answers according to the target matching score to obtain a target answer, and recommend and display the target answer. Through the calculation of the matching score, it is possible to clearly determine the relevance between different candidate keywords and the target keywords, and obtain the candidate keyword most relevant to the target keyword, so as to use the candidate answer corresponding to the candidate keyword as the target answer and recommend and display the target answer to the person being questioned, which can make the generated target answer more in line with the current question search requirements, enabling the person being questioned to answer the corresponding question in a timely and accurate manner according to the target answer, and improving the accuracy of question answering during teleprompter use.

[0180] The embodiments described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation to the technical solutions provided by the embodiments of the present application. Those skilled in the art will know that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0181] Those skilled in the art can understand that Figure 1-7 the technical solutions shown do not constitute a limitation to the embodiments of the present application, and may include more or fewer steps than those shown, or combine some steps, or different steps.

[0182] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0183] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices can be implemented as software, firmware, hardware and their appropriate combinations.

[0184] In the description of this application and the above-mentioned accompanying drawings, terms such as "first", "second", "third", "fourth", etc. (if any) are used to distinguish similar objects and do not necessarily describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that comprises a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0185] It should be understood that in this application, "at least one (item)" means one or more, and "a plurality" means two or more. "And / or" is used to describe the association relationship of associated objects and indicates that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and both A and B exist at the same time. Among them, A and B can be singular or plural. The character " / " generally means that the associated objects before and after are in an "or" relationship. "At least one (one) of the following" or its similar expressions refer to any combination of these items, including any combination of single items (ones) or plural items (ones). For example, at least one (one) of a, b, or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0186] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the above-mentioned division of units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling, direct coupling, or communication connection to each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical, mechanical, or other forms.

[0187] The units described above as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0188] In addition, in each embodiment of the present application, each functional unit can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0189] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes multiple instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in each embodiment of the present application. The foregoing storage medium includes: various media that can store programs such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0190] The preferred embodiments of the embodiments of the present application have been described above with reference to the accompanying drawings. This does not limit the scope of the rights of the embodiments of the present application. Any modifications, equivalent replacements, and improvements made by those skilled in the art without departing from the scope and essence of the embodiments of the present application shall be within the scope of the rights of the embodiments of the present application.

Claims

1. A teleprompter method, characterized in that, The method includes: Receiving a question search request; wherein, the question search request includes a trigger request and a target question; Performing keyword extraction on the target question according to the trigger request to obtain target keywords; performing tag matching search on the target keywords according to a preset tag hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers; Performing keyword extraction on the candidate questions to obtain candidate keywords; calculating a matching degree between the candidate keywords and the target keywords to obtain a target matching score; performing answer screening processing on the candidate answers according to the target matching score to obtain a target answer; recommending and displaying the target answer; The performing keyword extraction on the target question according to the trigger request to obtain target keywords includes: Invoking a preset keyword extraction model according to the trigger request; performing keyword extraction on the target question through the keyword extraction model to obtain initial keywords; performing keyword screening processing on the initial keywords according to preset screening conditions to obtain target keywords; The calculating a matching degree between the candidate keywords and the target keywords to obtain a target matching score includes: Calculating a similarity matching degree score between the target keywords and the candidate keywords; calculating a search matching degree score between the target keywords and the candidate keywords; performing weighted calculation on the similarity matching degree score and the search matching degree score through a preset adjustment factor to obtain the target matching score; Before performing tag matching search on the target keywords according to the preset tag hierarchy to obtain candidate Q&A data, the method further includes pre-constructing the preset tag hierarchy, specifically including: Obtaining reference Q&A texts; obtaining a plurality of reference Q&A paragraphs in the reference Q&A texts based on optical character recognition technology, and performing text parsing on each reference Q&A paragraph by using natural language processing technology to obtain a plurality of reference keywords; Inputting the reference keywords into a pre-constructed initial tag hierarchy, wherein the initial tag hierarchy is constructed according to historical Q&A texts and preset tag categories; Matching the reference keywords with the third-level tag names in the initial tag hierarchy to obtain tag data corresponding to the reference keywords including overall third-level tags; optimizing the initial tag hierarchy according to the tag data to obtain the preset tag hierarchy.

2. The teleprompter method according to claim 1, wherein The preset tag hierarchy includes a plurality of target tags, and the step of performing tag matching search on the target keywords according to the preset tag hierarchy to obtain candidate Q&A data includes: Performing tag matching search on the target keywords according to the target tags to obtain a matching probability value for each preset tag category; Performing Q&A screening processing on the reference Q&A data in the target tags according to the matching probability value to obtain the candidate Q&A data.

3. The teleprompter method according to claim 1, wherein The step of performing answer screening processing on the candidate answers according to the target matching score to obtain a target answer includes: Comparing the target matching score with a preset matching score threshold; Use the candidate answer with the target matching score greater than the matching score threshold as the target answer.

4. The teleprompter method according to claim 1, characterized in that, Before the step of pre - constructing the preset tag hierarchy, the method further includes pre - constructing the initial tag hierarchy, specifically including: Obtain historical Q&A texts; Extract keywords from the historical Q&A texts to obtain historical keywords; Construct the initial tag hierarchy according to the historical keywords and preset tag categories.

5. A teleprompter device, characterized in that, The device includes: A search request acquisition module, configured to receive a question search request; wherein, the question search request includes a trigger request and a target question; A target keyword extraction module, configured to extract keywords from the target question according to the trigger request to obtain target keywords; A candidate Q&A data acquisition module, configured to perform tag - matching search on the target keywords according to a preset tag hierarchy to obtain candidate Q&A data; wherein, the candidate Q&A data includes candidate questions and candidate answers; A candidate keyword extraction module, configured to extract keywords from the candidate questions to obtain candidate keywords; A matching score calculation module, configured to calculate a matching degree between the candidate keywords and the target keywords to obtain a target matching score; A screening module, configured to perform answer screening processing on the candidate answers according to the target matching score to obtain a target answer; A prompt display module, configured to recommend and display the target answer; The extracting keywords from the target question according to the trigger request to obtain target keywords includes: Call a preset keyword extraction model according to the trigger request; extract initial keywords from the target question through the keyword extraction model; perform keyword screening processing on the initial keywords according to preset screening conditions to obtain target keywords; The calculating the matching degree between the candidate keywords and the target keywords to obtain a target matching score includes: Calculate a similarity matching degree score between the target keywords and the candidate keywords; calculate a search matching degree score between the target keywords and the candidate keywords; perform weighted calculation on the similarity matching degree score and the search matching degree score through a preset adjustment factor to obtain the target matching score; Before performing the tag - matching search on the target keywords according to the preset tag hierarchy to obtain candidate Q&A data, the device further includes pre - constructing the preset tag hierarchy, specifically including: Obtain reference Q&A texts; obtain multiple reference Q&A paragraphs in the reference Q&A texts based on optical character recognition technology, and perform text parsing on each reference Q&A paragraph using natural language processing technology to obtain multiple reference keywords; Input the reference keywords into the pre - constructed initial tag hierarchy, where the initial tag hierarchy is constructed according to historical Q&A texts and preset tag categories; Match the reference keywords with the third - level tag names in the initial tag hierarchy to obtain tag data corresponding to the reference keywords including the overall third - level tags; optimize the initial tag hierarchy according to the tag data to obtain the preset tag hierarchy.

6. An electronic device, characterized in that, The electronic device includes a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for implementing connection communication between the processor and the memory. When the program is executed by the processor, the steps of the teleprompter method according to any one of claims 1 to 4 are implemented.

7. A storage medium, which is a computer-readable storage medium for computer-readable storage, and is characterized in that The storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of the teleprompter method according to any one of claims 1 to 4.

Images