CN116306587A - Internet negative public opinion early warning method - Google Patents

Internet negative public opinion early warning method Download PDF

Info

Publication number
CN116306587A
CN116306587A CN202310307147.4A CN202310307147A CN116306587A CN 116306587 A CN116306587 A CN 116306587A CN 202310307147 A CN202310307147 A CN 202310307147A CN 116306587 A CN116306587 A CN 116306587A
Authority
CN
China
Prior art keywords
public opinion
data information
opinion data
negative
extracted
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310307147.4A
Other languages
Chinese (zh)
Inventor
韩益亮
崔骕
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Engineering University of Chinese Peoples Armed Police Force
Original Assignee
Engineering University of Chinese Peoples Armed Police Force
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Engineering University of Chinese Peoples Armed Police Force filed Critical Engineering University of Chinese Peoples Armed Police Force
Priority to CN202310307147.4A priority Critical patent/CN116306587A/en
Publication of CN116306587A publication Critical patent/CN116306587A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/205—Parsing
    • G06F40/216—Parsing using statistical methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/951—Indexing; Web crawling techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/20—Natural language analysis
    • G06F40/279—Recognition of textual entities
    • G06F40/284—Lexical analysis, e.g. tokenisation or collocates
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/01—Social networking

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Economics (AREA)
  • Computing Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明公开一种互联网负面舆情预警方法，包括：采集舆情数据信息，并舆情数据信息进行预处理；对预处理后的舆情数据信息进行情感极性识别，以识别标记正面舆情数据信息、负面舆情数据信息和中立舆情数据信息；对负面舆情数据信息进行主题提取；计算提取的主题和与该提取的主题相关联的舆情数据信息的数量的爆发指数；计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息的情绪指数；计算提取的主题和与该提取的主题相关联的负面舆情数据信息的传播指数；计算提取的同一主题的爆发指数、情绪指数和传播指数的总和以获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警。

Figure 202310307147

The invention discloses a method for early warning of negative public opinion on the Internet, comprising: collecting public opinion data information, and preprocessing the public opinion data information; performing emotional polarity identification on the preprocessed public opinion data information, so as to identify and mark positive public opinion data information and negative public opinion information Data information and neutral public opinion data information; theme extraction of negative public opinion data information; calculation of the extracted theme and the outbreak index of the number of public opinion data information associated with the extracted theme; calculation of the extracted theme and the relationship between the extracted theme Calculate the sentiment index of the associated positive public opinion data information and negative public opinion data information; calculate the dissemination index of the extracted topic and the negative public opinion data information associated with the extracted topic; calculate the outbreak index, sentiment index and dissemination index of the same topic extracted The sum of the negative public opinion index is obtained to obtain a comprehensive negative public opinion index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, an early warning will be given to the topic and the corresponding public opinion event.

Figure 202310307147

Description

一种互联网负面舆情预警方法A method for early warning of negative Internet public opinion

技术领域technical field

本发明涉及信息处理技术领域，更具体地涉及一种互联网负面舆情预警方法。The present invention relates to the technical field of information processing, and more specifically relates to a method for early warning of negative Internet public opinion.

背景技术Background technique

网络舆情是指在互联网上流行的对社会问题不同看法的网络舆论，是社会舆论的一种表现形式，是通过互联网传播的公众对现实生活中某些热点、焦点问题所持的有较强影响力、倾向性的言论和观点。Internet public opinion refers to the Internet public opinion with different views on social issues popular on the Internet. , Predisposed remarks and opinions.

目前，由于网络信息和受众之间存在一定的信息差，导致各种网络负面舆情事件突发、频发，进而使得相关部门正常的舆论工作受到巨大的冲击，如果不能及时疏解和响应，将严重削弱政府话语权的影响力。然而，目前对负面网络舆情的早期预警很少，不能单纯通过几条社交网络负面情绪信息进行预警，使得相关部门的舆论工作开展大大受阻。At present, due to a certain information gap between network information and audiences, various negative network public opinion incidents occur suddenly and frequently, and the normal public opinion work of relevant departments has been greatly impacted. If they cannot be resolved and responded in time, serious Weaken the influence of the government's voice. However, there are very few early warnings of negative online public opinion at present, and early warnings cannot be made simply through a few negative emotional information on social networks, which greatly hinders the development of public opinion work by relevant departments.

鉴于此，有必要提供一种互联网负面舆情预警方法，以根据负面舆情主题鉴别负面舆情事件，并通过对负面舆情主题的爆发指数、情绪指数和传播指数进行计算、阈值判定等，实现对负面舆情事件的早期预警。In view of this, it is necessary to provide an early warning method for negative public opinion on the Internet to identify negative public opinion events according to the theme of negative public opinion, and to realize the detection of negative public opinion through the calculation of the outbreak index, sentiment index and transmission index of the theme of negative public opinion, threshold determination, etc. Early warning of incidents.

发明内容Contents of the invention

本发明所要解决的技术问题是提供一种互联网负面舆情预警方法，以实现对各类网络负面舆情的早期预警提示。The technical problem to be solved by the present invention is to provide an early warning method for negative Internet public opinion, so as to realize early warning prompts for various types of negative Internet public opinion.

为解决上述技术问题，提供一种互联网负面舆情预警方法，包括：In order to solve the above-mentioned technical problems, a method for early warning of negative Internet public opinion is provided, including:

从社交网络平台中采集舆情数据信息，并对所采集的舆情数据信息进行预处理；Collect public opinion data information from social network platforms, and preprocess the collected public opinion data information;

利用ernie3.0预训练模型对预处理后的舆情数据信息进行情感极性识别，以识别标记正面舆情数据信息、负面舆情数据信息和中立舆情数据信息；Use the ernie3.0 pre-training model to identify the emotional polarity of the preprocessed public opinion data information to identify and mark positive public opinion data information, negative public opinion data information and neutral public opinion data information;

对负面舆情数据信息进行主题提取；Subject extraction of negative public opinion data information;

计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，并在爆发指数超过第一预设阈值时进行负面舆情爆发过快预警；Calculate the outbreak index of the extracted topic and the amount of public opinion data information associated with the extracted topic within a first preset time period, and give an early warning of the outbreak of negative public opinion when the outbreak index exceeds the first preset threshold;

计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，并在情绪指数超过第二预设阈值时进行负面舆情波动过大预警；Calculate the sentiment index of the extracted theme and the positive public opinion data information and negative public opinion data information associated with the extracted theme within a second preset time period, and perform a negative public opinion fluctuation process when the sentiment index exceeds the second preset threshold. big warning;

计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，并在传播指数超过第三预设阈值时进行负面舆情主题过多预警；Calculating the spread index of the extracted topic and the negative public opinion data information associated with the extracted topic within a third preset time period, and giving an early warning of excessive negative public opinion topics when the spread index exceeds the third preset threshold;

计算提取的同一主题的爆发指数、情绪指数和传播指数的总和，以获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警通知。Calculate the sum of the extracted outbreak index, sentiment index, and communication index of the same topic to obtain a comprehensive negative public opinion index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, an early warning notification will be issued for the topic and the corresponding public opinion event .

其进一步技术方案为：所述计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，具体包括：根据公式

Figure BDA0004147213730000021
计算提取的主题和与该提取的主题相关联的舆情数据信息的数量的爆发指数EI；其中，/>
Figure BDA0004147213730000022
为第i天出现提取的主题t的负面舆情数据信息的数量，
Figure BDA0004147213730000031
为第n天出现提取的主题t的负面舆情数据信息的数量。Its further technical solution is: said calculation of the extracted theme and the outbreak index of the quantity of public opinion data information associated with the extracted theme within the first preset time period, specifically includes: according to the formula
Figure BDA0004147213730000021
Calculate the topic of extraction and the outbreak index EI of the quantity of public opinion data information associated with the topic of this extraction; Wherein, />
Figure BDA0004147213730000022
is the number of negative public opinion data information of the topic t extracted on the i-th day,
Figure BDA0004147213730000031
The number of negative public opinion data information extracted for topic t that appeared on the nth day.

其进一步技术方案为：所述计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，具体包括：根据公式

Figure BDA0004147213730000032
计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息的情绪指数SI；其中，/>
Figure BDA0004147213730000033
为第i天出现提取的主题t的正面舆情数据信息的数量，/>
Figure BDA0004147213730000034
为第n天出现提取的主题t的负面舆情数据信息的数量。Its further technical solution is: said calculation of the extracted topic and the sentiment index of the positive public opinion data information and negative public opinion data information associated with the extracted topic within the second preset time period, specifically including: according to the formula
Figure BDA0004147213730000032
Calculate the sentiment index SI of the extracted theme and the positive public opinion data information and negative public opinion data information associated with the extracted theme; where, />
Figure BDA0004147213730000033
is the number of positive public opinion data information of the topic t extracted on the i-th day, />
Figure BDA0004147213730000034
The number of negative public opinion data information extracted for topic t that appeared on the nth day.

其进一步技术方案为：所述计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，具体包括：根据公式

Figure BDA0004147213730000035
计算提取的主题和与该提取的主题相关联的负面舆情数据信息的传播指数；其中，/>
Figure BDA0004147213730000036
为第n天出现提取的主题t的负面舆情数据信息的数量，/>
Figure BDA0004147213730000037
为第i天出现提取的主题t的负面舆情数据信息的数量。Its further technical solution is: said calculation of the extracted theme and the spread index of the negative public opinion data information associated with the extracted theme within the third preset time period, specifically including: according to the formula
Figure BDA0004147213730000035
Calculate the dissemination index of the extracted topic and the negative public opinion data information associated with the extracted topic; where, />
Figure BDA0004147213730000036
The number of negative public opinion data information extracted for topic t that appeared on the nth day, />
Figure BDA0004147213730000037
The number of negative public opinion data information of topic t extracted for the i-th day.

其进一步技术方案为：所述对该主题和对应的舆情事件进行早期预警通知，具体包括：通过发送弹窗提醒/向手机发送短信/手机微信公众号发送通知/来电通话的方式对该主题和对应的舆情事件进行早期预警通知。Its further technical solution is: the early warning notification of the topic and corresponding public opinion events, specifically including: sending a pop-up window reminder/sending a text message to a mobile phone/sending a notification from a mobile phone WeChat official account/calling the topic and the corresponding public opinion event Early warning notifications for corresponding public opinion events.

其进一步技术方案为：所述对负面舆情数据信息进行主题提取，具体包括：采用TF-IDF算法对负面舆情数据信息的关键词进行提取，每一关键词标记为一主题。Its further technical solution is: the topic extraction of the negative public opinion data information specifically includes: using the TF-IDF algorithm to extract keywords of the negative public opinion data information, and each keyword is marked as a topic.

其进一步技术方案为：所述采用TF-IDF算法对负面舆情数据信息的关键词进行提取，之后还包括：显示每个关键词对应的权重信息。Its further technical solution is: the TF-IDF algorithm is used to extract the keywords of the negative public opinion data information, and then it also includes: displaying the weight information corresponding to each keyword.

其进一步技术方案为：所述第一预设阈值、第二预设阈值和第三预设阈值的取值均为1，所述第四预设阈值的取值为2。Its further technical solution is: the values of the first preset threshold, the second preset threshold and the third preset threshold are all 1, and the value of the fourth preset threshold is 2.

其进一步技术方案为：所述对所采集的舆情数据信息进行预处理，具体包括：Its further technical solution is: the preprocessing of the collected public opinion data information includes:

删除所采集的舆情数据信息中不必要的空格和换行符；Delete unnecessary spaces and line breaks in the collected public opinion data;

和/或，删除所采集的舆情数据信息中的@+用户名、表情符和邮箱；And/or, delete @+username, emoji and email in the collected public opinion data information;

和/或，取消所采集的舆情数据信息中的转义HTML标记；And/or, cancel the escaped HTML tags in the collected public opinion data information;

和/或，用URL替换所采集的舆情数据信息中提到的超链接；And/or, replace the hyperlinks mentioned in the collected public opinion data information with URLs;

和/或，将所采集的舆情数据信息中的繁体字转换为简体字。And/or, convert traditional Chinese characters in the collected public opinion data information into simplified Chinese characters.

本发明的有益技术效果在于：本发明通过从社交网络平台中采集舆情数据信息并进行预处理，并对预处理后的舆情数据信息进行情感极性识别以识别标记舆情数据信息中的正面舆情数据信息、负面舆情数据信息和中立舆情数据信息，且对所有负面舆情数据信息进行主题提取，进而计算所提取的主题的爆发指数、情绪指数和传播指数，并在爆发指数、情绪指数和传播指数分别超过其设定的阈值时进行预警，还可结合爆发指数、情绪指数和传播指数获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警通知，可知，本发明实现了对社交网络平台海量数据的采集、分析和预警，即可根据负面舆情主题鉴别负面舆情事件，并通过对负面舆情主题的爆发指数、情绪指数和传播指数进行计算、阈值判定等，实现对与该主题对应的舆情事件的早期预警，从而能够在舆情发生的第一时间从源头发现并掌握舆情动态，实现预警，为企业、公司和政府等机构提供高效准确的针对网络舆情的预警信息，在热点舆情事件爆发早期做到提前感知和预防处理，从而把舆情隐患尽量制止在摇篮中。The beneficial technical effects of the present invention are: the present invention collects public opinion data information from the social network platform and performs preprocessing, and performs emotional polarity recognition on the preprocessed public opinion data information to identify and mark positive public opinion data in the public opinion data information Information, negative public opinion data information and neutral public opinion data information, and subject extraction is performed on all negative public opinion data information, and then the outbreak index, sentiment index and communication index of the extracted topics are calculated, and the outbreak index, sentiment index and communication index are calculated respectively. When the threshold set by it is exceeded, an early warning will be issued, and a comprehensive negative public opinion index can be obtained by combining the outbreak index, sentiment index and communication index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, the topic and corresponding public opinion events will be analyzed Early warning notification, it can be seen that the present invention realizes the collection, analysis and early warning of massive data on social network platforms, and can identify negative public opinion events according to the theme of negative public opinion, and carry out the outbreak index, sentiment index and transmission index of the theme of negative public opinion. Calculation, threshold determination, etc., to realize early warning of public opinion events corresponding to the topic, so as to discover and grasp public opinion dynamics from the source as soon as public opinion occurs, realize early warning, and provide efficient and accurate information for enterprises, companies, governments and other institutions Early warning information for Internet public opinion, early perception and preventive treatment of hot public opinion events, so as to prevent hidden dangers of public opinion in the cradle as much as possible.

附图说明Description of drawings

图1是本发明一种互联网负面舆情预警方法一具体实施例的流程示意图。FIG. 1 is a schematic flow chart of a specific embodiment of a method for early warning of negative Internet public opinion in the present invention.

具体实施方式Detailed ways

为使本领域的普通技术人员更加清楚地理解本发明的目的、技术方案和优点，以下结合附图和实施例对本发明做进一步的阐述。In order to make those skilled in the art more clearly understand the purpose, technical solutions and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

参照图1，图1为本发明一种互联网负面舆情预警方法一具体实施例的流程示意图。在附图所示的实施例中，所述互联网负面舆情预警方法包括：Referring to FIG. 1 , FIG. 1 is a schematic flowchart of a specific embodiment of a method for early warning of negative Internet public opinion in the present invention. In the embodiment shown in the accompanying drawings, the Internet negative public opinion early warning method includes:

S101、从社交网络平台中采集舆情数据信息，并对所采集的舆情数据信息进行预处理。S101. Collect public opinion data information from a social networking platform, and preprocess the collected public opinion data information.

该步骤中，所述社交网络平台包括微博、贴吧、新闻、论坛、微信以及数字报等，本实施例中，可采集在预设社交网络平台在预设时间段内产生的舆情数据信息，其中，所述舆情数据信息包括一个或多个文本，例如，预设社交网络平台为微博时，所述舆情数据信息可以包括帖子、评论信息以及转发帖子信息中的至少一个。In this step, the social network platform includes Weibo, Tieba, news, forums, WeChat and digital newspapers, etc. In this embodiment, the public opinion data information generated on the preset social network platform within a preset time period can be collected, Wherein, the public opinion data information includes one or more texts, for example, when the preset social network platform is Weibo, the public opinion data information may include at least one of posts, comment information, and forwarded post information.

具体地，所述对所采集的舆情数据信息进行预处理具体包括：删除所采集的舆情数据信息中不必要的空格和换行符；和/或，删除所采集的舆情数据信息中的@+用户名、表情符和邮箱；和/或，取消所采集的舆情数据信息中的转义HTML标记；和/或，用URL替换所采集的舆情数据信息中提到的超链接；和/或，将所采集的舆情数据信息中的繁体字转换为简体字。Specifically, the preprocessing of the collected public opinion data information specifically includes: deleting unnecessary spaces and line breaks in the collected public opinion data information; and/or deleting @+user in the collected public opinion data information name, emoji, and email address; and/or, cancel the escaped HTML tags in the collected public opinion data information; and/or, replace the hyperlinks mentioned in the collected public opinion data information with URLs; and/or, replace The traditional characters in the collected public opinion data information are converted into simplified characters.

S102、利用ernie3.0预训练模型对预处理后的舆情数据信息进行情感极性识别，以识别标记正面舆情数据信息、负面舆情数据信息和中立舆情数据信息。S102. Use the ernie3.0 pre-training model to identify the emotional polarity of the preprocessed public opinion data information, so as to identify and mark positive public opinion data information, negative public opinion data information, and neutral public opinion data information.

该步骤中，使用ernie3.0模型对预处理后的舆情数据信息进行情感极性识别，识别标记舆情数据信息中哪些舆情数据信息为正面舆情数据信息，哪些为负面舆情数据信息，哪些为中立舆情数据信息，并按照时间顺序保存。In this step, use the ernie3.0 model to identify the emotional polarity of the preprocessed public opinion data information, and identify which public opinion data information in the marked public opinion data information is positive public opinion data information, which is negative public opinion data information, and which is neutral public opinion data information The data information is stored in chronological order.

S103、对负面舆情数据信息进行主题提取。S103. Extracting topics from the negative public opinion data information.

该步骤中，采用TF-IDF算法(Term Frequency-Inverse Document Frequency,词频-逆文档频率算法)对所有的负面舆情数据信息的关键词进行提取，每一关键词标记为一主题，且还可显示每个关键词对应的权重信息，还可根据关键词对应的权重大小进行排名，以供用户查看，且用户还可以查看与该关键词相关的舆情数据信息，例如当社交网络平台为微博时，还可以查看与该关键词相关的总帖子数量、负面的帖子数量、负面的帖子百分比、某天某个时间的负面帖子、根据关键词搜索的负面帖子等。In this step, the TF-IDF algorithm (Term Frequency-Inverse Document Frequency, term frequency-inverse document frequency algorithm) is used to extract the keywords of all negative public opinion data information, each keyword is marked as a topic, and can also be displayed The weight information corresponding to each keyword can also be ranked according to the weight corresponding to the keyword for users to view, and the user can also view public opinion data information related to the keyword. For example, when the social network platform is Weibo, You can also view the total number of posts related to the keyword, the number of negative posts, the percentage of negative posts, negative posts at a certain time of day, negative posts searched by keyword, etc.

S104、计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，并在爆发指数超过第一预设阈值时进行负面舆情爆发过快预警。S104. Calculate the outbreak index of the extracted topic and the amount of public opinion data information associated with the extracted topic within the first preset time period, and give an early warning of the rapid outbreak of negative public opinion when the outbreak index exceeds the first preset threshold .

优选地，该步骤中，所述第一预设阈值为1。Preferably, in this step, the first preset threshold is 1.

本发明中，所述计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，具体包括：根据公式

Figure BDA0004147213730000061
计算提取的主题和与该提取的主题相关联的舆情数据信息的数量的爆发指数EI；其中，/>
Figure BDA0004147213730000062
为第i天出现提取的主题t的负面舆情数据信息的数量，/>
Figure BDA0004147213730000063
为第n天出现提取的主题t的负面舆情数据信息的数量。In the present invention, the calculation of the extracted subject and the outbreak index of the quantity of public opinion data information associated with the extracted subject within the first preset time period specifically includes: according to the formula
Figure BDA0004147213730000061
Calculate the topic of extraction and the outbreak index EI of the quantity of public opinion data information associated with the topic of this extraction; Wherein, />
Figure BDA0004147213730000062
is the number of negative public opinion data information of topic t extracted on the i-th day, />
Figure BDA0004147213730000063
The number of negative public opinion data information extracted for topic t that appeared on the nth day.

可理解地，第一预设时间段可根据用户选择进行设定，而Relu为激活函数，又叫线性整流函数，当Relu函数括号内的数值小于0时激活函数的结果变为0，当大于零时保持原数值，当EI≥1时，第n天出现提取的主题t的负面舆情数据信息的数量已经达到前n-1天平均值的2倍，增长程度达到100％，触发爆发指数预警，可通过发送弹窗提醒/向手机发送短信/手机微信公众号发送通知/来电通话的方式进行负面舆情爆发过快预警；而当

Figure BDA0004147213730000071
小于之前n-1天均值时，负面情绪舆论呈现负增长，触发Relu激活函数，EI值为零，不再为爆发指数提供预警信息。Understandably, the first preset time period can be set according to the user's choice, and Relu is an activation function, also known as a linear rectification function. When the value in the parentheses of the Relu function is less than 0, the result of the activation function becomes 0, and when it is greater than Keep the original value at zero time. When EI ≥ 1, the number of negative public opinion data information extracted on the topic t on the nth day has reached twice the average value of the previous n-1 days, and the growth rate has reached 100%, triggering an outbreak index warning , you can send a pop-up window reminder/send a text message to a mobile phone/send a notification from a mobile phone WeChat official account/call a phone to give an early warning of the rapid outbreak of negative public opinion; and when
Figure BDA0004147213730000071
When it is less than the average value of the previous n-1 days, negative sentiment and public opinion show negative growth, triggering the Relu activation function, and the EI value is zero, which no longer provides early warning information for the outbreak index.

S105、计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，并在情绪指数超过第二预设阈值时进行负面舆情波动过大预警。S105. Calculate the sentiment index of the extracted topic and the positive public opinion data information and negative public opinion data information associated with the extracted topic within a second preset time period, and perform negative public opinion when the sentiment index exceeds the second preset threshold Warning of excessive volatility.

该步骤中，所述第二预设阈值为1。In this step, the second preset threshold is 1.

本发明中，所述计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，具体包括：根据公式

Figure BDA0004147213730000072
计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息的情绪指数SI；其中，/>
Figure BDA0004147213730000073
为第i天出现提取的主题t的正面舆情数据信息的数量，/>
Figure BDA0004147213730000074
为第n天出现提取的主题t的负面舆情数据信息的数量。In the present invention, the calculation of the extracted topic and the sentiment index of the positive public opinion data information and negative public opinion data information associated with the extracted topic within the second preset time period specifically includes: according to the formula
Figure BDA0004147213730000072
Calculate the sentiment index SI of the extracted theme and the positive public opinion data information and negative public opinion data information associated with the extracted theme; where, />
Figure BDA0004147213730000073
is the number of positive public opinion data information of the topic t extracted on the i-th day, />
Figure BDA0004147213730000074
The number of negative public opinion data information extracted for topic t that appeared on the nth day.

可理解地，当SI≥1时，第n天出现提取的主题t的负面舆情数据信息和正面舆情数据信息的比值已经达到前n-1天均值的2倍，增长程度达到100％，触发情绪指数预警，而当第n天比值小于之前n-1天比值的均值时，积极情绪舆论占据主导地位，触发Relu激活函数，SI值为零，不再为情绪指数提供预警信息。Understandably, when SI≥1, the ratio of negative public opinion data information and positive public opinion data information of topic t extracted on the nth day has reached twice the average value of the previous n-1 days, and the growth rate has reached 100%, triggering emotions Index early warning, and when the ratio on the nth day is less than the average value of the previous n-1 day ratio, positive sentiment and public opinion dominate, triggering the Relu activation function, the SI value is zero, and no longer provides early warning information for the sentiment index.

S106、计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，并在传播指数超过第三预设阈值时进行负面舆情主题过多预警。S106. Calculate the spread index of the extracted topic and the negative public opinion data information associated with the extracted topic within a third preset time period, and give an early warning of too many negative public opinion topics when the spread index exceeds a third preset threshold.

该步骤中，所述第三预设阈值为1。In this step, the third preset threshold is 1.

本发明中，所述计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，具体包括：根据公式

Figure BDA0004147213730000081
计算提取的主题和与该提取的主题相关联的负面舆情数据信息的传播指数；其中，/>
Figure BDA0004147213730000082
为第n天出现提取的主题t的负面舆情数据信息的数量，/>
Figure BDA0004147213730000083
为第i天出现提取的主题t的负面舆情数据信息的数量。In the present invention, the calculation of the extracted theme and the spread index of the negative public opinion data information associated with the extracted theme within the third preset time period specifically includes: according to the formula
Figure BDA0004147213730000081
Calculate the dissemination index of the extracted topic and the negative public opinion data information associated with the extracted topic; where, />
Figure BDA0004147213730000082
The number of negative public opinion data information extracted for topic t that appeared on the nth day, />
Figure BDA0004147213730000083
The number of negative public opinion data information of topic t extracted for the i-th day.

在本实施例中，使用传播指数时底数log可采用10为底数，而针对更大级别数据时可以对底数进行适当更改。In this embodiment, the base log can use 10 as the base number when using the propagation index, and the base number can be appropriately changed for larger-level data.

S107、计算提取的同一主题的爆发指数、情绪指数和传播指数的总和，以获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警通知。S107. Calculate the sum of the extracted outbreak index, sentiment index, and communication index of the same topic to obtain a comprehensive negative public opinion index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, conduct an early analysis of the topic and corresponding public opinion events Early warning notification.

该步骤中，根据公式POI＝SI+EI+DI计算获得综合负面舆情指数POI，所述第四预设阈值优选为2。In this step, the comprehensive negative public opinion index POI is calculated according to the formula POI=SI+EI+DI, and the fourth preset threshold is preferably 2.

可理解地，本发明互联网负面舆情预警方法可运行于搭载在终端设备(例如手机、pad等)的app中，通过该app可以实现用户和系统之间的交互。Understandably, the method for early warning of negative Internet public opinion of the present invention can be run in an app carried on a terminal device (such as a mobile phone, a pad, etc.), and the interaction between the user and the system can be realized through the app.

综上可知，本发明通过从社交网络平台中采集舆情数据信息并进行预处理，并对预处理后的舆情数据信息进行情感极性识别以识别标记舆情数据信息中的正面舆情数据信息、负面舆情数据信息和中立舆情数据信息，且对所有负面舆情数据信息进行主题提取，进而计算所提取的主题的爆发指数、情绪指数和传播指数，并在爆发指数、情绪指数和传播指数分别超过其设定的阈值时进行预警，还可结合爆发指数、情绪指数和传播指数获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警通知，可知，本发明以负面情绪主题为切入口，根据负面舆情主题鉴别负面舆情事件，通过对舆情数据信息进行情感极性识别、主题提取，随后从提取的主题的爆发指数、情绪指数、传播指数三个维度以及综合负面舆情指数实现对与该主题对应的舆情事件的早期预警，从而能够在舆情发生的第一时间从源头发现并掌握舆情动态，实现预警，为企业、公司和政府等机构提供高效准确的针对网络舆情的预警信息，在热点舆情事件爆发早期做到提前感知和预防处理，从而把舆情隐患尽量制止在摇篮中。In summary, the present invention collects and preprocesses public opinion data information from a social network platform, and performs emotional polarity recognition on the preprocessed public opinion data information to identify and mark positive public opinion data information and negative public opinion data information in the public opinion data information. Data information and neutral public opinion data information, and subject extraction of all negative public opinion data information, and then calculate the outbreak index, sentiment index and transmission index of the extracted topics, and when the outbreak index, sentiment index and transmission index exceed their settings respectively Early warning is given when the threshold is reached, and the comprehensive negative public opinion index can be obtained by combining the outbreak index, sentiment index and communication index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, an early warning notification will be given for the topic and the corresponding public opinion event, It can be seen that the present invention takes negative emotional topics as the entry point, identifies negative public opinion events according to negative public opinion topics, and carries out emotional polarity identification and topic extraction on public opinion data information, and then extracts from the outbreak index, emotional index, and spread index of the extracted topics. The three dimensions and the comprehensive negative public opinion index realize the early warning of the public opinion events corresponding to the topic, so that the public opinion dynamics can be discovered and grasped from the source as soon as the public opinion occurs, and the early warning can be realized, providing enterprises, companies, governments and other institutions with Efficient and accurate early warning information for Internet public opinion, early perception and preventive treatment of hot public opinion events in the early stage, so as to prevent hidden dangers of public opinion in the cradle as much as possible.

以上所述仅为本发明的优选实施例，而非对本发明做任何形式上的限制。本领域的技术人员可在上述实施例的基础上施以各种等同的更改和改进，凡在权利要求范围内所做的等同变化或修饰，均应落入本发明的保护范围之内。The above descriptions are only preferred embodiments of the present invention, and do not limit the present invention in any form. Those skilled in the art can make various equivalent changes and improvements on the basis of the above-mentioned embodiments, and all equivalent changes or modifications made within the scope of the claims shall fall within the protection scope of the present invention.

Claims (9)

1.一种互联网负面舆情预警方法，其特征在于，包括：1. A method for early warning of negative Internet public opinion, characterized in that it comprises: 从社交网络平台中采集舆情数据信息，并对所采集的舆情数据信息进行预处理；Collect public opinion data information from social network platforms, and preprocess the collected public opinion data information; 利用ernie3.0预训练模型对预处理后的舆情数据信息进行情感极性识别，以识别标记正面舆情数据信息、负面舆情数据信息和中立舆情数据信息；Use the ernie3.0 pre-training model to identify the emotional polarity of the preprocessed public opinion data information to identify and mark positive public opinion data information, negative public opinion data information and neutral public opinion data information; 对负面舆情数据信息进行主题提取；Subject extraction of negative public opinion data information; 计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，并在爆发指数超过第一预设阈值时进行负面舆情爆发过快预警；Calculate the outbreak index of the extracted topic and the amount of public opinion data information associated with the extracted topic within a first preset time period, and give an early warning of the outbreak of negative public opinion when the outbreak index exceeds the first preset threshold; 计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，并在情绪指数超过第二预设阈值时进行负面舆情波动过大预警；Calculate the sentiment index of the extracted theme and the positive public opinion data information and negative public opinion data information associated with the extracted theme within a second preset time period, and perform a negative public opinion fluctuation process when the sentiment index exceeds the second preset threshold. big warning; 计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，并在传播指数超过第三预设阈值时进行负面舆情主题过多预警；Calculating the spread index of the extracted topic and the negative public opinion data information associated with the extracted topic within a third preset time period, and giving an early warning of excessive negative public opinion topics when the spread index exceeds the third preset threshold; 计算提取的同一主题的爆发指数、情绪指数和传播指数的总和，以获得综合负面舆情指数，并在综合负面舆情指数超过第四预设阈值时，对该主题和对应的舆情事件进行早期预警通知。Calculate the sum of the extracted outbreak index, sentiment index, and communication index of the same topic to obtain a comprehensive negative public opinion index, and when the comprehensive negative public opinion index exceeds the fourth preset threshold, an early warning notification will be issued for the topic and the corresponding public opinion event . 2.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述计算提取的主题和与该提取的主题相关联的舆情数据信息的数量在第一预设时间段内的爆发指数，具体包括：根据公式
Figure FDA0004147213720000021
计算提取的主题和与该提取的主题相关联的舆情数据信息的数量的爆发指数EI；其中，/>
Figure FDA0004147213720000022
为第i天出现提取的主题t的负面舆情数据信息的数量，/>
Figure FDA0004147213720000023
为第n天出现提取的主题t的负面舆情数据信息的数量。2. Internet negative public opinion early warning method as claimed in claim 1, is characterized in that, the topic that described calculation extracts and the quantity of the public opinion data information associated with the topic of this extraction is the outbreak index in the first preset time period , including: according to the formula
Figure FDA0004147213720000021
Calculate the topic of extraction and the outbreak index EI of the quantity of public opinion data information associated with the topic of this extraction; Wherein, />
Figure FDA0004147213720000022
is the number of negative public opinion data information of topic t extracted on the i-th day, />
Figure FDA0004147213720000023
The number of negative public opinion data information extracted for topic t that appeared on the nth day. 3.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息在第二预设时间段内的情绪指数，具体包括：根据公式
Figure FDA0004147213720000024
计算提取的主题和与该提取的主题相关联的正面舆情数据信息和负面舆情数据信息的情绪指数SI；其中，/>
Figure FDA0004147213720000025
为第i天出现提取的主题t的正面舆情数据信息的数量，/>
Figure FDA0004147213720000026
为第n天出现提取的主题t的负面舆情数据信息的数量。3. Internet negative public opinion early warning method as claimed in claim 1, is characterized in that, the positive public opinion data information and the negative public opinion data information associated with the theme extracted by the calculation and the extracted theme are within the second preset time period. The sentiment index within, specifically includes: According to the formula
Figure FDA0004147213720000024
Calculate the sentiment index SI of the extracted theme and the positive public opinion data information and negative public opinion data information associated with the extracted theme; where, />
Figure FDA0004147213720000025
is the number of positive public opinion data information of the topic t extracted on the i-th day, />
Figure FDA0004147213720000026
The number of negative public opinion data information extracted for topic t that appeared on the nth day. 4.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述计算提取的主题和与该提取的主题相关联的负面舆情数据信息在第三预设时间段内的传播指数，具体包括：根据公式
Figure FDA0004147213720000027
计算提取的主题和与该提取的主题相关联的负面舆情数据信息的传播指数；其中，/>
Figure FDA0004147213720000028
为第n天出现提取的主题t的负面舆情数据信息的数量，/>
Figure FDA0004147213720000029
为第i天出现提取的主题t的负面舆情数据信息的数量。4. Internet negative public opinion early-warning method as claimed in claim 1, is characterized in that, the dissemination index of the theme extracted by said calculation and the negative public opinion data information associated with the theme extracted in the third preset time period, Specifically include: According to the formula
Figure FDA0004147213720000027
Calculate the dissemination index of the extracted topic and the negative public opinion data information associated with the extracted topic; where, />
Figure FDA0004147213720000028
The number of negative public opinion data information extracted for topic t that appeared on the nth day, />
Figure FDA0004147213720000029
The number of negative public opinion data information of topic t extracted for the i-th day. 5.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述对该主题和对应的舆情事件进行早期预警通知，具体包括：通过发送弹窗提醒/向手机发送短信/手机微信公众号发送通知/来电通话的方式对该主题和对应的舆情事件进行早期预警通知。5. The Internet negative public opinion early warning method as claimed in claim 1, wherein the early warning notification of the topic and corresponding public opinion events specifically includes: sending pop-up reminders/sending text messages/mobile phone WeChat to mobile phones The official account sends notifications/incoming calls to provide early warning notifications for the topic and corresponding public opinion events. 6.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述对负面舆情数据信息进行主题提取，具体包括：采用TF-IDF算法对负面舆情数据信息的关键词进行提取，每一关键词标记为一主题。6. Internet negative public opinion early warning method as claimed in claim 1, is characterized in that, described subject extraction is carried out to negative public opinion data information, specifically comprises: adopting TF-IDF algorithm to extract the keyword of negative public opinion data information, every A keyword is marked as a topic. 7.如权利要求6所述的互联网负面舆情预警方法，其特征在于，所述采用TF-IDF算法对负面舆情数据信息的关键词进行提取，之后还包括：显示每个关键词对应的权重信息。7. Internet negative public opinion early warning method as claimed in claim 6, is characterized in that, described employing TF-IDF algorithm extracts the keyword of negative public opinion data information, also comprises afterwards: display the weight information corresponding to each keyword . 8.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述第一预设阈值、第二预设阈值和第三预设阈值的取值均为1，所述第四预设阈值的取值为2。8. The Internet negative public opinion early warning method as claimed in claim 1, wherein the values of the first preset threshold, the second preset threshold and the third preset threshold are all 1, and the fourth preset threshold is 1. Set the value of the threshold to 2. 9.如权利要求1所述的互联网负面舆情预警方法，其特征在于，所述对所采集的舆情数据信息进行预处理，具体包括：9. Internet negative public opinion early warning method as claimed in claim 1, is characterized in that, described public opinion data information collected is carried out preprocessing, specifically comprises: 删除所采集的舆情数据信息中不必要的空格和换行符；Delete unnecessary spaces and line breaks in the collected public opinion data; 和/或，删除所采集的舆情数据信息中的@+用户名、表情符和邮箱；And/or, delete @+username, emoji and email in the collected public opinion data information; 和/或，取消所采集的舆情数据信息中的转义HTML标记；And/or, cancel the escaped HTML tags in the collected public opinion data information; 和/或，用URL替换所采集的舆情数据信息中提到的超链接；And/or, replace the hyperlinks mentioned in the collected public opinion data information with URLs; 和/或，将所采集的舆情数据信息中的繁体字转换为简体字。And/or, convert traditional Chinese characters in the collected public opinion data information into simplified Chinese characters.
CN202310307147.4A 2023-03-27 2023-03-27 Internet negative public opinion early warning method Pending CN116306587A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310307147.4A CN116306587A (en) 2023-03-27 2023-03-27 Internet negative public opinion early warning method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310307147.4A CN116306587A (en) 2023-03-27 2023-03-27 Internet negative public opinion early warning method

Publications (1)

Publication Number Publication Date
CN116306587A true CN116306587A (en) 2023-06-23

Family

ID=86803044

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310307147.4A Pending CN116306587A (en) 2023-03-27 2023-03-27 Internet negative public opinion early warning method

Country Status (1)

Country Link
CN (1) CN116306587A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117743581A (en) * 2023-12-26 2024-03-22 中国农业科学院农业信息研究所 Intervention method for agricultural product quality safety network rumors

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117743581A (en) * 2023-12-26 2024-03-22 中国农业科学院农业信息研究所 Intervention method for agricultural product quality safety network rumors
CN117743581B (en) * 2023-12-26 2024-06-11 中国农业科学院农业信息研究所 A method for intervening in online rumors about agricultural product quality and safety

Similar Documents

Publication Publication Date Title
US10911383B2 (en) Spam filtering and person profiles
CN107343077B (en) Method, device and equipment for identifying malicious phone and establishing identification model
US10778624B2 (en) Systems and methods for spam filtering
US10120859B2 (en) Message sentiment analyzer and message preclusion
Petersen et al. # Covid-19: An exploratory investigation of hashtag usage on Twitter
CN101784022A (en) Method and system for filtering and classifying short messages
CN104580758A (en) Incoming call comment information acquisition method and device
CN104219672B (en) Incoming call or short message recognition methods and device
CN106936971B (en) A kind of incoming person's information presentation system and reminding method
CN116306587A (en) Internet negative public opinion early warning method
Kabha et al. Comparison study between the UAE, the UK, and India in dealing with WhatsApp fake news
CN115017289A (en) A kind of customer service method and device based on knowledge base
CN105530386B (en) The determination method and its application process and system of a kind of communication identifier type of number
KR20130083286A (en) System for processing spam for mobile phone
CN105468768A (en) System monitoring method of WeChat public sentiment
CN110245352A (en) A kind of public sentiment hot word analysis method and device
JP5516264B2 (en) Questionnaire survey device and questionnaire survey method
JP3227155U (en) Message filtering system for filtering and managing user-generated text content on social networks
Cao et al. Ontology-based spam detection filtering system
CN107895284A (en) Exchange page management-control method, device, terminal and computer-readable recording medium
US20210144112A1 (en) Method, system and apparatus for providing a contextual keyword collective for communication events in a multicommunication platform environment
US20210312510A1 (en) Customer-sentiment driven workflow based on social media data
CN103107936B (en) A kind of method and device sending information
Sung et al. Development of humming call system for blocking spam on a smartphone
JP2020052595A (en) Extractor and program

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination