Conference visualization display method and system and storage medium

Abstract

The invention relates to the field of conference visual display, in particular to a conference visual display method and system and a storage medium. Comprising data acquisition, processing, intelligent analysis and recommendation modules; the data processing module is used for cleaning and denoising data, performing word segmentation, part-of-speech tagging and named entity recognition by utilizing a natural language processing technology, and extracting key information; the intelligent analysis module constructs an emotion analysis model through a recurrent neural network, and calculates an emotion tendency value of each candidate; the recommendation module compares the emotional tendency value with a standard threshold value, marks different candidates and screens matched recommendation content for display, for example, positive candidates display successful cases, neutral candidates display basic information, and negative candidates are not displayed on a current screen; the conference data can be efficiently processed, the emotional tendency is accurately analyzed, personalized recommendation display is realized, and the conference efficiency and the decision-making quality are improved.

Description

Technical Field

[0001] The present invention relates to the field of conference visual display, and specifically provides a conference visual display method, system and storage medium. Background Art

[0002] In the current conference scenario, with the continuous development of information technology, the amount of data generated by conferences is increasing exponentially and is of various types. Traditional conference display and analysis methods have gradually revealed many problems, which have become the research background and innovation driving force of the present invention. There are various data sources involved in conferences, such as voice speeches, voting data, text information in chat windows and discussion areas, etc. These data have different formats and uneven quality, and there are problems such as speech recognition errors, invalid voting data, and a large amount of interference information in the text content. When converting speech to text, it is easily affected by background noise, resulting in errors and garbled characters in the text. During the voting process, there may be situations such as duplicate voting and format errors, making it difficult to directly use the data for effective analysis and increasing the difficulty of information extraction and utilization.

[0003] It is difficult to grasp the emotional tendencies of participants. During the conference decision-making process, it is crucial to understand the emotional tendencies of participants towards different topics and candidates. However, traditional methods lack effective means for accurate analysis. The opinions and emotions of participants are often scattered in various speeches and text exchanges. Manual analysis is not only time-consuming and laborious, but also difficult to accurately capture and quantify emotional tendencies. This results in the inability of conference organizers and decision-makers to timely and accurately understand the attitudes of participants, affecting the scientificity and rationality of conference decisions.

[0004] To address the above deficiencies, a technical solution is provided. Summary of the Invention

[0005] To solve the technical problems raised in the above background art, the present invention provides a conference visual display method, system and storage medium.

[0006] The object of the present invention can be achieved through the following technical solutions: In the first aspect of the present invention, a conference visual display system is provided, including a data acquisition module, a data processing module, an intelligent analysis module, a recommendation module and a database.

[0007] The data acquisition module obtains data content according to various data sources in the conference and transmits it to the data processing module in real time. The specific process is as follows:

[0008] Real-time collect the voice input in the meeting and convert it into voice text content; real-time obtain the number and information of voter accounts, label the meeting initiator as candidate A, and label the accounts of each voter as candidate B and candidate C in the order of entering the meeting until labeled as candidate Z. If the number of voters exceeds 24, they will be labeled as candidate A2, candidate B2, and candidate C2 in turn until all voters are labeled. Real-time obtain the voting data in the meeting; then real-time capture the text information in the chat window and discussion area; send the collected voice text content, voting data, and text information to the data processing module.

[0009] The data processing module cleans and denoises based on the collected voice text content, voting data, and text information, and performs word segmentation, part-of-speech tagging, and named entity recognition on the text type data through natural language processing technology. The specific process is as follows:

[0010] The data processing module scans the voice text content, obtains the recognition error data in the voice text content, and corrects it. For example, convert the wrongly formed words into correct expressions, and then remove the extra spaces, line breaks, and special characters in the text; identify the interfering words generated by background noise in the voice text content, extract the preset stop word list in the database, and the stop word list includes modal particles and emojis, and remove the noise words according to the stop word list;

[0011] Identify the invalid voting records in the voting data, including duplicate votes and voting content with incorrect formats, delete the invalid voting records, and at the same time obtain the candidate accounts corresponding to the invalid voting records, and send the content of "Voting error, please vote again" to the display screen of the candidate accounts;

[0012] Similarly, clean and denoise the content in the text information; perform word segmentation on the processed voice text content and text information, divide them into several word sequences according to the word boundaries, and then obtain the grammatical properties of the word sequences through a part-of-speech tagging tool and preset tagging rules. The grammatical properties include nouns, verbs, and adjectives. Finally, extract the emotional expression words, evaluation objects, and emotional intensity words of each candidate through named entity recognition technology, including words such as hate, satisfaction, disappointment, very, extremely, and a little. Integrate the words of each candidate extracted in this meeting into each meeting vocabulary library and send it to the intelligent analysis module.

[0013] The intelligent analysis module obtains different emotional probability values for each candidate according to the emotion analysis model and calculates the emotional tendency value. The specific process is as follows:

[0014] Select a recurrent neural network as the architecture of the sentiment analysis model; extract the positive sentiment cases, neutral sentiment cases, and negative sentiment cases of the text content in the database and integrate them into a sentiment dataset. Use the sentiment dataset to train the sentiment analysis model. Divide the sentiment dataset into a 70% training set, a 15% validation set, and a 15% test set. Use each conference vocabulary library as an input feature quantity, and use the training set to train the selected model. Continuously adjust the model parameters during the training process, select a loss function for parameter update until obtaining each sentiment probability value. Each sentiment probability value includes a positive probability value, a neutral probability value, and a negative probability value, which are respectively marked as TY, TI, and TF, and substitute them into a preset formula for calculation Obtain the sentiment tendency values of each candidate and send the sentiment tendency values of each candidate to the recommendation module.

[0015] The recommendation module compares the sentiment tendency values of each candidate with the standard sentiment tendency threshold, marks different candidates, and screens and displays the recommended content that matches them. The specific process is as follows:

[0016] Extract the standard sentiment tendency threshold in the database. If the sentiment tendency value of any candidate is greater than the standard sentiment tendency threshold, it indicates that the corresponding candidate has a high enthusiasm for the evaluation of this meeting, and mark it as a positive candidate. If the sentiment tendency value of any candidate is less than the standard sentiment tendency threshold, it indicates that the corresponding candidate has a high negative emotion, and mark it as a negative candidate. If the sentiment tendency value of any candidate is equal to the standard sentiment tendency threshold, it indicates that the corresponding candidate has neither obvious positive evaluation nor negative view, and mark it as a neutral candidate; thus, count each positive candidate, each negative candidate, and each neutral candidate; if it is determined that the current is a positive candidate speaking, use orange as the main color for the speech box and voting data to be displayed on the display screen, obtain the successful cases in the account of this positive candidate in the database, and place them in the center position of the display screen in chronological order; if it is determined that the current is a negative candidate speaking, use blue as the main color for the speech box and voting data, and the screen does not display. If it is determined that the current is a neutral candidate speaking, use gray as the main color for the speech box and voting data, and obtain the basic information and professional skills of the candidate in the database and place them on the display screen.

[0017] Please refer to Figure 2 As shown, the second aspect of the present invention provides a conference visualization display method, and its specific steps are as follows:

[0018] Step 1: Data collection: Obtain data content according to various data sources in the conference and transmit it to data processing in real time;

[0019] Step 2: Data processing: Clean the collected voice text content, voting data and text information, and use natural language processing technology to perform word segmentation, part-of-speech tagging and named entity recognition on the text type data to extract key information;

[0020] Step 3: Intelligent analysis: Obtain different emotional probability values ​​of each candidate based on the sentiment analysis model, and calculate the sentiment tendency value;

[0021] Step 4: Recommendation: Compare the sentiment tendency value of each candidate with the standard sentiment tendency threshold, mark them as different candidates, and screen and display the matching recommended content.

[0022] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0023] Compared with the prior art, the present invention has the following beneficial effects: the intelligent analysis module uses a recurrent neural network to construct a sentiment analysis model, and through a large number of sentiment case training, it can accurately obtain the different sentiment probability values ​​of each candidate, and then calculate the sentiment tendency value. This process quantifies the participants' emotional attitudes towards the candidates, allowing the conference organizers and participants to intuitively understand everyone's views on different candidates, providing a strong reference for conference decision-making;

[0024] The data processing module can accurately extract key information from complex data through cleaning, denoising and natural language processing technologies; for example, it can correct errors in voice text, remove invalid records in voting data, and identify and extract emotion-related words to ensure that the data input into subsequent modules is clean and accurate, providing a reliable basis for analysis and decision-making;

[0025] The recommendation module classifies and labels candidates based on their sentiment values ​​and filters matching recommendations. For positive candidates, successful cases are displayed to reinforce their strengths. For neutral candidates, basic information and skills are displayed to facilitate a comprehensive understanding. Although negative candidates are not displayed on the current screen, targeted improvement directions can be provided. This personalized recommendation display allows participants to quickly focus on key points and improve the pertinence and efficiency of meeting discussions and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. The following drawings are not intentionally scaled to the actual size, and the focus is on illustrating the main purpose of the present invention.

[0027] Figure 1This is the principle block diagram of the present invention.

[0028] Figure 2 This is the method step diagram of the present invention. Detailed implementation manners

[0029] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only partial embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts also belong to the scope of protection of the present invention.

[0030] Please refer to Figure 1 As shown, the first aspect of the present invention provides a conference visualization display system, including a data acquisition module, a data processing module, an intelligent analysis module, a recommendation module, and a database.

[0031] The data acquisition module obtains data content according to various data sources in the conference and transmits it to the data processing module in real time. The specific process is as follows:

[0032] Real-time collect the voice input in the conference and convert it into voice text content; real-time obtain the number and information of voter accounts, mark the conference initiator as candidate A, and mark the accounts of each voter in the order of entering the conference as candidate B and candidate C until marked as candidate Z. If the number of voters exceeds 24, they will be marked as candidate A2, candidate B2, and candidate C2 in turn until all voters are marked. Real-time obtain the voting data in the conference; then real-time capture the text information in the chat window and discussion area; send the collected voice text content, voting data, and text information to the data processing module.

[0033] The data processing module performs cleaning and denoising on the collected voice text content, voting data, and text information, and performs word segmentation, part-of-speech tagging, and named entity recognition on text type data through natural language processing technology. The specific process is as follows:

[0034] The data processing module scans the voice text content, obtains the recognition error data in the voice text content, and corrects it. For example, convert the wrongly formed words into correct expressions, and then remove the extra spaces, line breaks, and special characters in the text; identify the interference words generated by background noise in the voice text content, extract the preset stop word list in the database, and the stop word list includes modal particles and emoticons, and remove the noise words according to the stop word list;

[0035] Identify invalid voting records in the voting data, including duplicate votes and incorrectly formatted voting content, delete the invalid voting records, and at the same time obtain the candidate accounts corresponding to the invalid voting records, and send the content "Voting error, please vote again" to the display screens of these candidate accounts;

[0036] Similarly, clean and denoise the content in the text information; perform word segmentation on the processed speech text content and text information, divide them into several word sequences according to word boundaries, and then use a part-of-speech tagging tool and preset tagging rules to obtain the grammatical properties of the word sequences. The grammatical properties include nouns, verbs, and adjectives. Finally, use named entity recognition technology to extract the emotional expression words, evaluation objects, and emotional intensity words of each candidate, including words such as dislike, satisfaction, disappointment, very, extremely, and a little. Integrate the words of each candidate extracted in this meeting into each meeting vocabulary library and send it to the intelligent analysis module.

[0037] The intelligent analysis module obtains different emotional probability values for each candidate according to the sentiment analysis model and calculates the emotional tendency value. The specific process is as follows:

[0038] Select a recurrent neural network as the architecture of the sentiment analysis model; extract the positive sentiment cases, neutral sentiment cases, and negative sentiment cases in the text content of the database and integrate them into a sentiment dataset. Use the sentiment dataset to train the sentiment analysis model. Divide the sentiment dataset into a 70% training set, a 15% validation set, and a 15% test set. Use each meeting vocabulary library as the input feature quantity, and use the training set to train the selected model. Continuously adjust the model parameters during the training process, select a loss function for parameter update until the emotional probability values are obtained. The emotional probability values include positive probability values, neutral probability values, and negative probability values, which are respectively marked as TY, TI, and TF, and substitute the values of the three into a preset formula for calculation Obtain the emotional tendency values of each candidate, and send the emotional tendency values of each candidate to the recommendation module.

[0039] The recommendation module compares the emotional tendency values of each candidate with the standard emotional tendency threshold, marks different candidates, and screens and displays the recommended content that matches them. The specific process is as follows:

[0040] Extract the standard sentiment tendency threshold in the database. If the sentiment tendency value of any candidate is greater than the standard sentiment tendency threshold, it indicates that the corresponding candidate has a high enthusiasm for the evaluation of this meeting, and mark it as a positive candidate. If the sentiment tendency value of any candidate is less than the standard sentiment tendency threshold, it indicates that the corresponding candidate has a high negative emotion, and mark it as a negative candidate. If the sentiment tendency value of any candidate is equal to the standard sentiment tendency threshold, it indicates that the corresponding candidate has neither obvious positive evaluation nor negative view, and mark it as a neutral candidate. Thus, count the positive candidates, negative candidates, and neutral candidates respectively. If it is determined that the current speaker is a positive candidate, use orange as the main color for the speech box and voting data and display them on the display screen. Obtain the successful cases in the account of this positive candidate in the database and display them in the center of the display screen in chronological order. If it is determined that the current speaker is a negative candidate, use blue as the main color for the speech box and voting data, and the screen does not display. If it is determined that the current speaker is a neutral candidate, use gray as the main color for the speech box and voting data, and obtain the basic information and professional skills of the candidate in the database and display them on the display screen.

[0041] Please refer to Figure 2 As shown, the second aspect of the present invention provides a method for visual display of a meeting, and its specific steps are as follows:

[0042] Step 1: Data collection: Obtain data content according to various data sources in the meeting and transmit it to data processing in real time;

[0043] Step 2: Data processing: Clean the collected speech text content, voting data, and text information, and perform word segmentation, part-of-speech tagging, and named entity recognition on the text type data through natural language processing technology to extract key information;

[0044] Step 3: Intelligent analysis: Obtain different sentiment probability values of each candidate according to the sentiment analysis model and calculate the sentiment tendency value;

[0045] Step 4: Recommendation: Compare the sentiment tendency values of each candidate with the standard sentiment tendency threshold, mark them as different candidates, and screen and display the recommended content that matches them.

[0046] An embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0047] The foregoing is a description of the present invention and should not be construed as limiting thereof. Although several exemplary embodiments of the present invention have been described, those skilled in the art will readily appreciate that many modifications can be made to the exemplary embodiments without departing from the novel teachings and advantages of the present invention. Accordingly, all such modifications are intended to be included within the scope of the present invention as defined by the claims. It should be understood that the foregoing is a description of the present invention and should not be considered limited to the specific embodiments disclosed, and modifications to the disclosed embodiments as well as other embodiments are intended to be included within the scope of the appended claims. The present invention is defined by the claims and their equivalents.

Claims

1. A conference visualization display system, comprising a data processing module, an intelligent analysis module, a recommendation module and a database, characterized in that: The data processing module cleans and denoises the collected voice text content, voting data and text information, and uses natural language processing technology to perform word segmentation, part-of-speech tagging and named entity recognition on text type data; the recommendation module compares the sentiment tendency value of each candidate with the standard sentiment tendency threshold, marks them as different candidates, and screens and displays the recommended content that matches them; The intelligent analysis module obtains the different emotional probability values ​​of each candidate according to the sentiment analysis model, calculates the sentiment tendency value, and selects the recurrent neural network as the architecture of the sentiment analysis model; Extract positive emotion cases, neutral emotion cases and negative emotion cases from the text content in the database and integrate them into an emotion data set. Use the emotion data set to train the emotion analysis model. Divide the emotion data set into 70% training set, 15% validation set and 15% test set. Use each conference vocabulary library as input feature quantity. Use the training set to train the selected model. During the training process, continuously adjust the model parameters, select the loss function to update the parameters until the emotion probability values ​​are obtained. The emotion probability values ​​include positive probability values, neutral probability values ​​and negative probability values, which are marked as TY, TI and TF respectively, and substitute their values ​​into the preset formula The sentiment tendency value of each candidate is calculated and sent to the recommendation module.

2. A conference visualization display system according to claim 1, characterized in that: The data processing module performs cleaning and denoising based on the collected voice text content, voting data and text information. The specific process is as follows: The data processing module scans the speech text content, obtains the recognition error data in the speech text content, and corrects it, such as converting the correct but incorrect words into the correct expression, and then removing the extra spaces, line breaks and special characters in the text; identifies the interference words generated by background noise in the speech text content, extracts the stop word list preset in the database, the stop word list includes modal particles and emoticons, and removes the noise words according to the stop word list; Identify invalid voting records in the voting data, including duplicate voting and voting content with incorrect format, delete the invalid voting records, and obtain the candidate account corresponding to the invalid voting record, and send the text "voting error, please vote again" to the display screen of the candidate account; Clean and denoise the content of text information.

3. A conference visualization display system according to claim 2, characterized in that: The data processing module performs word segmentation, part-of-speech tagging and named entity recognition on text type data through natural language processing technology, which is specifically: The processed speech text content and text information are segmented and divided into several word sequences according to word boundaries. The grammatical properties of the word sequences are obtained through part-of-speech tagging tools and preset tagging rules. The grammatical properties include nouns, verbs and adjectives. Finally, the named entity recognition technology is used to extract the emotion expression vocabulary, evaluation object and emotion intensity vocabulary of each candidate, including the words hate, satisfy, disappoint, very, and a little bit. The vocabulary of each candidate after the extraction of this meeting is integrated into the vocabulary library of each meeting and sent to the intelligent analysis module.

4. A conference visualization display system according to claim 1, characterized in that: The recommendation module compares the sentiment value of each candidate with the standard sentiment threshold, marks them as different candidates, and screens the matching recommended content for display. The specific process is as follows: Extract the standard emotional tendency threshold in the database. If the emotional tendency value of any candidate is greater than the standard emotional tendency threshold, mark it as a positive candidate. If the emotional tendency value of any candidate is less than the standard emotional tendency threshold, it means that the corresponding candidate has high negative emotions and is marked as a negative candidate. If the emotional tendency value of any candidate is equal to the standard emotional tendency threshold, mark it as a neutral candidate. In this way, statistics are obtained for each positive candidate, each negative candidate and each neutral candidate. If it is determined that the current speaker is a positive candidate, the speech box and voting data are displayed on the display screen with orange as the main color, and the successful cases in the positive candidate account in the database are obtained and placed in the center of the display screen in chronological order. If it is determined that the current speaker is a negative candidate, the speech box and voting data are blue as the main color, and the screen is not displayed. If it is determined that the current speaker is a neutral candidate, the speech box and voting data are gray as the main color, and the basic information and professional skills of the candidates in the database are obtained and placed on the display screen.

5. A conference visualization display system according to claim 1, characterized in that: It also includes a data acquisition module, which obtains data content based on various data sources in the meeting and transmits it to the data processing module in real time. The specific process is as follows: Collect voice input in the meeting in real time and convert it into voice text content; obtain the number and information of voters' accounts in real time, mark the initiator of the meeting as candidate A, and mark the accounts of the remaining voters as candidate B and candidate C in the order of entering the meeting until they are marked as candidate Z. If the number of voters exceeds 24, they will be marked as candidate A2, candidate B2 and candidate C2 in turn until all voters are marked, and obtain the voting data in the meeting in real time; then capture the text information of the chat window and discussion area in real time; send the collected voice text content, voting data and text information to the data processing module.

6. A conference visualization display method, characterized in that: A conference visualization display system applied to any one of claims 1 to 5, wherein the specific steps are as follows: Step 1: Obtain data content based on multiple data sources within the meeting and transmit it to data processing in real time; Step 2: Clean the collected voice text content, voting data and text information, and use natural language processing technology to perform word segmentation, part-of-speech tagging and named entity recognition on the text type data to extract key information; Step 3: Obtain different sentiment probability values ​​of each candidate according to the sentiment analysis model, and calculate the sentiment tendency value; Step 4: Compare the sentiment tendency value of each candidate with the standard sentiment tendency threshold, mark them as different candidates, and screen and display the recommended content that matches them.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the conference visualization display method described in claim 6 is implemented.

Images