WO2023082698A1 - 公众满意度的分析方法、存储介质及电子设备 - Google Patents

公众满意度的分析方法、存储介质及电子设备 Download PDF

Info

Publication number
WO2023082698A1
WO2023082698A1 PCT/CN2022/107244 CN2022107244W WO2023082698A1 WO 2023082698 A1 WO2023082698 A1 WO 2023082698A1 CN 2022107244 W CN2022107244 W CN 2022107244W WO 2023082698 A1 WO2023082698 A1 WO 2023082698A1
Authority
WO
WIPO (PCT)
Prior art keywords
public
data
satisfaction
negative
positive
Prior art date
Application number
PCT/CN2022/107244
Other languages
English (en)
French (fr)
Inventor
季婧
刘益东
王君
Original Assignee
上海蜜度信息技术有限公司
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 上海蜜度信息技术有限公司 filed Critical 上海蜜度信息技术有限公司
Publication of WO2023082698A1 publication Critical patent/WO2023082698A1/zh

Links

Images

Classifications

    • 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
    • G06F16/353—Clustering; Classification into predefined classes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/048—Activation functions

Definitions

  • the invention belongs to the technical field of data analysis and relates to an analysis method, in particular to an analysis method of public satisfaction, a storage medium and electronic equipment.
  • the existing technology analyzes the degree of public attention or public dissemination of a certain event data based more on social media likes, comments and other communication attributes, and the data processing method is relatively simple.
  • this processing and analysis method cannot reflect the trend formed by public attention, for example, whether the public holds more positive or negative attitudes.
  • the purpose of the present invention is to provide an analysis method, storage medium and electronic equipment for public satisfaction, which is used to solve the problem that the prior art cannot be further analyzed on the basis of public attention and dissemination.
  • the issue of public satisfaction is to provide an analysis method, storage medium and electronic equipment for public satisfaction, which is used to solve the problem that the prior art cannot be further analyzed on the basis of public attention and dissemination.
  • the present invention provides an analysis method of public satisfaction on the one hand.
  • the analysis method of public satisfaction includes: collecting public feedback data; classifying the public feedback data, and determining the public feedback data.
  • the category of the public feedback data determine the attributes of each category of public feedback data according to the sensitivity of the public feedback data; the attributes include positive attributes and negative attributes; based on the attributes of all categories of public feedback data, determine public satisfaction Spend.
  • the step of collecting public feedback data includes: setting a collection time period; within the collection time period, acquiring public feedback data generated under all channels.
  • the category of public feedback data includes: at least one of public opinion data, media data or network report data.
  • the step of determining the attributes of each category of public feedback data according to the sensitivity of the public feedback data includes: analyzing the sensitivity of the public opinion data, if the public opinion data is not sensitive data, then determine the public opinion data as positive data; if the public opinion data is sensitive data, then determine the public opinion data as negative data; perform event clustering on the media data, determine the event information and data information of the media data; performing event clustering on the network report data, and determining the event information of the network report data.
  • the event information of the media data includes the number of positive media events and the number of negative media events; the data information of the media data includes the amount of positive information about positive media events and the amount of negative information about negative media events ;
  • the event information of the network report data refers to the number of network report negative events.
  • the step of determining public satisfaction based on the attributes of all types of public feedback data includes: determining the positive score or negative score of the public feedback data based on the attributes of all types of public feedback data. score; analyze the negative sentiment distribution of the public feedback data; calculate the positive total score or negative total score of all categories of public feedback data according to the positive score, the negative score and the negative sentiment distribution; combine the positive total score score and the negative total score to determine a total satisfaction score; modify the total satisfaction score, and evaluate the public satisfaction according to the revised total satisfaction score.
  • the positive score includes positive points for public opinion and/or positive points for media;
  • the negative score includes negative points for public opinion, negative points for media, and/or negative points for network reports.
  • the step of modifying the total satisfaction score includes: judging whether the total satisfaction score is greater than 0, and if the total satisfaction score is greater than 0, using the first function to calculate the The total score of satisfaction is corrected; if the total score of satisfaction is not greater than 0, the total score of satisfaction is corrected by using the second function.
  • another aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for analyzing public satisfaction is realized.
  • the last aspect of the present invention provides an electronic device, including: a processor and a memory; the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory, To make the electronic device execute the analysis method of public satisfaction.
  • the public satisfaction analysis method, storage medium and electronic equipment of the present invention have the following beneficial effects:
  • the present invention classifies the public feedback data into categories and determines the positive or negative attributes of each category of public feedback data to analyze public satisfaction. Compared with the prior art, the present invention can conduct two-way consideration from the positive and negative information of data and the positive and negative information of events, and can further analyze public satisfaction on the basis of public attention and dissemination.
  • FIG. 1 shows a schematic flow chart of an embodiment of the public satisfaction analysis method of the present invention.
  • FIG. 2 is a flow chart of data collection in an embodiment of the public satisfaction analysis method of the present invention.
  • FIG. 3 is a flow chart of attribute determination in an embodiment of the public satisfaction analysis method of the present invention.
  • FIG. 4 is a flow chart of public satisfaction analysis in an embodiment of the public satisfaction analysis method of the present invention.
  • FIG. 5 is a schematic diagram showing the structural connection of the electronic device of the present invention in an embodiment.
  • the public satisfaction analysis method, storage medium and electronic equipment described in the present invention can be bidirectionally considered from the positive and negative information of data and the positive and negative information of events, and can further analyze public satisfaction on the basis of public attention and dissemination.
  • the present invention is not the satisfaction of a single event, but the satisfaction of a certain type of event. For example, it can be a city or region's ecological environment, a city or region's public health safety satisfaction. Another example is that the public has a high degree of attention to the Olympic Games in a certain period of time. The high degree of attention analyzed in the prior art cannot reflect everyone's attitude towards the Olympic Games, that is, whether they are satisfied.
  • the analysis method of the public satisfaction degree described in the present invention can further analyze and obtain information such as whether everyone is satisfied with the Olympic Games, and how much the degree of satisfaction is.
  • FIG. 1 is a schematic flow chart of an embodiment of the public satisfaction analysis method of the present invention.
  • the analysis method of public satisfaction includes the following steps:
  • FIG. 2 shows a flow chart of data collection in an embodiment of the public satisfaction analysis method of the present invention.
  • S11 specifically includes the following steps:
  • the collection time period is, for example, 1 hour. In addition, other reasonable time periods may also be set according to collection requirements.
  • public feedback data generated under all channels that is, public feedback data generated on the entire network or online such as public opinion methods, media methods, and online reporting platforms.
  • a piece of public feedback data refers to a piece of text data.
  • the type of public feedback data includes: at least one of public opinion data, media data, or network report data.
  • category classification can be implemented according to the way of data collection. In practical applications, judge according to the source of data collection. For example, if the collected data comes from the government’s official reporting platform, it is determined that the data is network reporting data; if the collected data comes from official media, it is determined that the data is media data; the public opinion data can be the data obtained by screening keywords from the entire network data .
  • the BiLSTM model or other models that can realize binary classification can be used to judge the positive attributes and negative attributes.
  • the BiLSTM model used in the present invention is generated through a large amount of positive data and negative data training.
  • FIG. 3 shows a flow chart of attribute determination in an embodiment of the public satisfaction analysis method of the present invention.
  • S13 specifically includes the following steps:
  • sensitivity classification is performed on public opinion data, non-sensitive public opinion data is classified as positive data, and sensitive public opinion data is classified as negative data.
  • the number of positive public opinion data is taken as the amount of positive public opinion information, and the number of negative public opinion data is taken as the amount of negative public opinion information.
  • sentiment classification is carried out on the public opinion data to obtain the proportion of negative information.
  • S132 Perform event clustering on the media data, and determine event information of the media data and data information of the media data.
  • the event information of the media data includes the number of positive media events and the number of negative media events; the data information of the media data includes positive information about positive media events and negative information about negative media events.
  • event clustering is performed on the media data, positive media events and negative media events are determined, and the number of positive media events and the number of negative media events are counted respectively. And classify the sensitivity of a single piece of data. According to the amount of sensitive information, it is judged that the media data is positive media data or negative media data. The number of articles is regarded as the amount of negative information of media negative events.
  • the data reported on the Internet are all considered as negative information, and the number of negative events reported on the Internet is obtained through event clustering.
  • S133 Perform event clustering on the network report data, and determine event information of the network report data.
  • the event information of the network report data refers to the number of negative events reported by the network.
  • the principle is as follows: segment the text of each piece of media data or network report data, use the word2vec model, and calculate the average to obtain the sentence vector of the sentence.
  • the similarity calculation is performed on the sentence vectors of all texts, and the texts with high similarity are clustered as a class of events.
  • Word2vec is a group of related models used to generate word vectors. These models are shallow, two-layer neural networks trained to reconstruct linguistic word texts.
  • FIG. 4 shows a flow chart of public satisfaction analysis in an embodiment of the public satisfaction analysis method of the present invention.
  • S14 specifically includes the following steps:
  • the positive score includes positive public opinion plus points and/or positive media plus points;
  • the negative score includes negative public opinion minus points, negative media minus points and/or negative net report negative points.
  • the positive and negative scores of various types of data are obtained through independent logarithmization of the variables of information volume and number of events, or logarithmization of the number of information volume events .
  • positive points for online reports are set to 0.
  • s ln(a).
  • a represents the amount of information and the number of events or the number of information events
  • s represents the positive and negative scores.
  • the distribution of negative emotions in public opinion proportion of negative information * f.
  • f represents the scaling factor used to set the negative sentiment distribution of public opinion to a converted value of the same magnitude as the other fractions.
  • the range of positive public opinion points is 0-0.1
  • S143 Calculate positive total scores or negative total scores of all categories of public feedback data according to the positive scores, the negative scores and the negative emotion distribution.
  • the positive total score is the total positive plus points
  • the negative total score is the total negative minus points.
  • the total negative minus points and the total positive plus points are respectively calculated by weighting:
  • Total negative points a* negative negative points for public opinion + b* distribution of negative emotions for public opinion + c* negative points for media negative points + d* negative points for network reports;
  • Total positive points o*positive points of public opinion+p*(1-distribution of negative emotions of public opinion)+q*positive points of media+r*positive points of network reports.
  • the weight values a, b, c, and d are set to be 0.15, 0.25, 0.40, and 0.20 respectively, and the weight values o, p, q, and r are set to be 0.15, 0.25, 0.40, and 0.20, respectively.
  • other reasonable weight values set according to business requirements and which type of data to focus on are within the protection scope of the present invention.
  • total score total positive plus points - total negative minus points.
  • the step of modifying the total satisfaction score includes:
  • the sigmoid score is corrected according to the set threshold.
  • the Sigmoid function is a common S-type function, also known as the S-type growth curve. In information science, due to its single-increase and inverse function single-increase properties, the Sigmoid function is often used as the activation function of the neural network to map variables between 0 and 1.
  • x represents the original total score
  • y represents the revised score
  • i, j, m, n are formula constants.
  • the formula constant is not a specific value, but is obtained according to different event types. That is, different event types take different values. In the actual setting process, a certain event direction is first selected, and then adjusted according to the data and results of different cities, and repeatedly based on the feedback from the business side.
  • the business side will give us feedback such as whether the overall score is high or low, or whether it is sensitive to different inputs, etc.
  • i, m can help control the sensitivity to the input
  • j, n can help control the overall score.
  • the characteristics of the Sigmoid function include: (1) It is monotonically increasing. (2) The function value range is between (0, 1). (3) Can accept positive and negative input. (4) The change is slower when the input is larger or smaller. In addition, the use of other functions with the above characteristics (1)-(4) for correction is also within the protection scope of the present invention.
  • the present invention before analyzing and processing the public feedback data, it is also possible to divide different regions, analyze public satisfaction for different regions, and then quantitatively display and intuitively present the analysis results of different regions. Compare the analysis results of different regions, and analyze the differences in public satisfaction for the same type of events (such as ecological environment or public health security) in different regions. Specifically, assuming that the collection time period is 1 hour, a y value will be generated every 1 hour for city A and city B. Thus, multiple y values of city A and city B can be obtained through statistical charts and Excel tables In other forms, it can show users the change of public satisfaction in city A or city B over time. It can also show the difference between public satisfaction in city A and city B in the same time period. Available public satisfaction information.
  • the analysis of public satisfaction can be performed only according to the positive and negative attributes of the data, or only based on the positive and negative attributes of the clustered events, and can also be combined with the positive and negative attributes of the data 1, positive and negative attributes of the event are analyzed, combined analysis is the preferred method, but other independent analysis methods are also within the protection scope of the present invention.
  • the present invention is based on the public satisfaction calculation method of events and information statistics after positive and negative sensitive classification.
  • the data is firstly classified by sensitivity , considering positive and negative plus and minus points, can further reflect the satisfaction of the public.
  • the present invention uses a modified sigmoid function adjusted by a specific threshold to calculate the score, which makes the score calculation more flexible and changeable, and the parameters of the function can be adjusted according to a large amount of actual data, so as to achieve a score that is more consistent with human subjective feelings.
  • This embodiment provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the method for analyzing public satisfaction is implemented.
  • the aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method embodiments; and the aforementioned computer-readable storage medium includes: ROM, RAM, magnetic disk or optical disk and other computer storage media that can store program codes.
  • FIG. 5 is a schematic diagram showing the structural connection of the electronic device of the present invention in an embodiment.
  • this embodiment provides an electronic device 5, which specifically includes: a processor 51 and a memory 52; the memory 52 is used to store computer programs, and the processor 51 is used to execute the programs stored in the memory 52. A computer program, so that the electronic device 5 executes each step of the public satisfaction analysis method.
  • processor 51 can be general-purpose processor, comprises central processing unit (Central Processing Unit, be called for short CPU), network processor (Network Processor, be called for short NP) etc.; Can also be Digital Signal Processing (Digital Signal Processing, be called for short DSP) ), Application Specific Integrated Circuit (ASIC for short), Field Programmable Gate Array (Field Programmable GateArray, FPGA for short) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.
  • Central Processing Unit be called for short CPU
  • Network Processor Network Processor
  • NP Network Processor
  • CPU central processing unit
  • Network Processor be called for short NP
  • ASIC Application Specific Integrated Circuit
  • FPGA Field Programmable GateArray
  • the above-mentioned memory 52 may include a random access memory (Random Access Memory, RAM for short), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.
  • RAM Random Access Memory
  • non-volatile memory such as at least one disk memory.
  • the electronic device may include memory, storage controller, one or more processing units (CPU), peripheral interface, RF circuit, audio circuit, speaker, microphone, input/output (I/O) Subsystems, display screens, other output or control devices, and components such as external ports; said computers include, but are not limited to, devices such as desktop computers, laptop computers, tablet computers, smartphones, smart TVs, personal digital assistants (Personal Digital Assistant , referred to as PDA) and other personal computers, in other embodiments, the electronic device can also be a server, and the server can be arranged on one or more physical servers according to various factors such as functions and loads, or can be arranged by The cloud server formed by a distributed or centralized server cluster is not limited in this embodiment.
  • the public satisfaction analysis method, storage medium and electronic equipment of the present invention classify the public feedback data and determine the positive or negative attributes of each category of public feedback data, so as to analyze the public satisfaction. satisfaction.
  • the present invention can conduct two-way consideration from the positive and negative information of data and the positive and negative information of events, and can further analyze public satisfaction on the basis of public attention and dissemination. The invention effectively overcomes various shortcomings in the prior art and has high industrial application value.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Evolutionary Computation (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本发明提供一种公众满意度的分析方法、存储介质及电子设备，所述公众满意度的分析方法包括：采集公众反馈数据；对所述公众反馈数据进行类别划分，确定所述公众反馈数据的类别；根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性；所述属性包括正面属性和负面属性；基于所有类别的公众反馈数据的属性，确定公众满意度。本发明可以从数据正负面信息与事件正负面信息进行双向考虑，可以在公众关注度、传播度的基础上进一步分析公众满意度。

Description

公众满意度的分析方法、存储介质及电子设备 技术领域
本发明属于数据分析的技术领域，涉及一种分析方法，特别是涉及一种公众满意度的分析方法、存储介质及电子设备。
背景技术
针对各种渠道下产生的公众数据，现有技术更多地根据社交媒体的点赞、评论等传播属性，分析某一事件数据的公众关注度或公众传播度，数据处理方式较简单。然而通过这样的处理分析方式并不能体现公众关注度所形成的趋势，比如，公众更多地是持有正面态度还是负面态度。
在目前的分析方式中，基于文本特征的关注度计算方式以及基于原始媒体数据量的直接计算方式等技术手段都是只能获取到公众对事件的关注程度。例如，以基于文本特征的关注度计算方式为例，在源数据中提取关键词，并构建关注度指数模型；实时获取网络数据，根据关注度指数模型计算获得事件关注度指数，通过关注度指数体现出公众对事件的关注程度。
因此，如何提供一种公众满意度的分析方法、存储介质及电子设备，以解决现有技术无法在公众关注度、传播度的基础上进一步分析公众满意度等缺陷，成为本领域技术人员亟待解决的技术问题。
发明内容
鉴于以上所述现有技术的缺点，本发明的目的在于提供一种公众满意度的分析方法、存储介质及电子设备，用于解决现有技术无法在公众关注度、传播度的基础上进一步分析公众满意度的问题。
为实现上述目的及其他相关目的，本发明一方面提供一种公众满意度的分析方法，所述公众满意度的分析方法包括：采集公众反馈数据；对所述公众反馈数据进行类别划分，确定所述公众反馈数据的类别；根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性；所述属性包括正面属性和负面属性；基于所有类别的公众反馈数据的属性，确定公众满意度。
于本发明的一实施例中，所述采集公众反馈数据的步骤包括：设置一采集时间段；在所述采集时间段内，获取所有渠道下产生的公众反馈数据。
于本发明的一实施例中，所述公众反馈数据的类别包括：舆论数据、媒体数据或网络举 报数据中的至少一种。
于本发明的一实施例中，所述根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性的步骤包括：分析所述舆论数据的敏感性，若所述舆论数据为非敏感数据，则将所述舆论数据确定为正面数据；若所述舆论数据为敏感数据，则将所述舆论数据确定为负面数据；对所述媒体数据进行事件聚类，确定所述媒体数据的事件信息和所述媒体数据的数据信息；对所述网络举报数据进行事件聚类，确定所述网络举报数据的事件信息。
于本发明的一实施例中，所述媒体数据的事件信息包括媒体正面事件个数和媒体负面事件个数；所述媒体数据的数据信息包括媒体正面事件正面信息量和媒体负面事件负面信息量；所述网络举报数据的事件信息是指网络举报负面事件个数。
于本发明的一实施例中，所述基于所有类别的公众反馈数据的属性，确定公众满意度的步骤包括：基于所有类别的公众反馈数据的属性，确定所述公众反馈数据的正面得分或负面得分；分析所述公众反馈数据的负面情绪分布；根据所述正面得分、所述负面得分和所述负面情绪分布计算所有类别的公众反馈数据的正面总得分或负面总得分；结合所述正面总得分和所述负面总得分确定满意度总分；修正所述满意度总分，根据修正后的满意度总分评估所述公众满意度。
于本发明的一实施例中，所述正面得分包括舆论正面加分和/或媒体正面加分；所述负面得分包括舆论负面减分、媒体负面减分和/或网络举报负面减分。
于本发明的一实施例中，所述修正所述满意度总分的步骤包括：判断所述满意度总分是否大于0，若所述满意度总分大于0，则利用第一函数对所述满意度总分进行修正；若所述满意度总分不大于0，则利用第二函数对所述满意度总分进行修正。
为实现上述目的及其他相关目的，本发明另一方面提供一种计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现所述的公众满意度的分析方法。
为实现上述目的及其他相关目的，本发明最后一方面提供一种电子设备，包括：处理器及存储器；所述存储器用于存储计算机程序，所述处理器用于执行所述存储器存储的计算机程序，以使所述电子设备执行所述的公众满意度的分析方法。
如上所述，本发明所述的公众满意度的分析方法、存储介质及电子设备，具有以下有益效果：
本发明通过对公众反馈数据进行类别划分，确定每一类别的公众反馈数据的正面属性或负面属性，以此来分析公众满意度。本发明与现有技术相比，可以从数据正负面信息与事件正负面信息进行双向考虑，可以在公众关注度、传播度的基础上进一步分析公众满意度。
附图说明
图1显示为本发明的公众满意度的分析方法于一实施例中的原理流程图。
图2显示为本发明的公众满意度的分析方法于一实施例中的数据采集流程图。
图3显示为本发明的公众满意度的分析方法于一实施例中的属性确定流程图。
图4显示为本发明的公众满意度的分析方法于一实施例中的公众满意度分析流程图。
图5显示为本发明的电子设备于一实施例中的结构连接示意图。
元件标号说明
1               电子设备
11              处理器
12              存储器
S11～S14        步骤
S111～S112      步骤
S131～S133      步骤
S141～S145      步骤
具体实施方式
以下通过特定的具体实例说明本发明的实施方式，本领域技术人员可由本说明书所揭露的内容轻易地了解本发明的其他优点与功效。本发明还可以通过另外不同的具体实施方式加以实施或应用，本说明书中的各项细节也可以基于不同观点与应用，在没有背离本发明的精神下进行各种修饰或改变。需说明的是，在不冲突的情况下，以下实施例及实施例中的特征可以相互组合。
需要说明的是，以下实施例中所提供的图示仅以示意方式说明本发明的基本构想，遂图示中仅显示与本发明中有关的组件而非按照实际实施时的组件数目、形状及尺寸绘制，其实际实施时各组件的型态、数量及比例可为一种随意的改变，且其组件布局型态也可能更为复杂。
本发明所述的公众满意度的分析方法、存储介质及电子设备可以从数据正负面信息与事件正负面信息进行双向考虑，可以在公众关注度、传播度的基础上进一步分析公众满意度。本发明并不是对单一事件的满意度，而是对某一种事件类型的满意度。比如可以是一个城市或地区的生态环境方面、一个城市或地区的公共卫生安全方面的满意度。又比如在一段时间 内公众某段时间对奥运会的关注度高，现有技术中分析出的关注度高并不能反映大家对奥运会的态度，即是否满意。本发明所述的公众满意度的分析方法可以进一步分析得出大家对奥运会是否满意，满意度是多少等信息。
以下将结合图1至图5详细阐述本实施例的一种公众满意度的分析方法、存储介质及电子设备的原理及实施方式，使本领域技术人员不需要创造性劳动即可理解本实施例的公众满意度的分析方法、存储介质及电子设备。
请参阅图1，显示为本发明的公众满意度的分析方法于一实施例中的原理流程图。如图1所示，所述公众满意度的分析方法具体包括以下几个步骤：
S11，采集公众反馈数据。
请参阅图2，显示为本发明的公众满意度的分析方法于一实施例中的数据采集流程图。如图2所示，S11具体包括以下步骤：
S111，设置一采集时间段。
具体地，所述采集时间段例如为1小时，除此之外，也可以根据采集需求设置其他合理的时间段。
S112，在所述采集时间段内，获取所有渠道下产生的公众反馈数据。
具体地，在1小时内，获取所有渠道下产生的公众反馈数据，即舆论方式、媒体方式、网络举报平台等全网或线上产生的公众反馈数据。于实际应用中，一条公众反馈数据是指一条文本数据。
S12，对所述公众反馈数据进行类别划分，确定所述公众反馈数据的类别。
于一实施例中，所述公众反馈数据的类别包括：舆论数据、媒体数据或网络举报数据中的至少一种。
具体地，类别划分可以依据数据采集的方式实现。于实际应用中，根据数据采集的来源判断。例如，采集数据源于政府官方举报平台，则判定该数据为网络举报数据；采集数据源于官方媒体，则判定该数据为媒体数据；舆论数据可以是全网数据中根据关键词筛选得到的数据。
S13，根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性；所述属性包括正面属性和负面属性。
具体地，关于正面属性和负面属性可用BiLSTM模型或其他可以实现二分类的模型进行判断。本发明所用的BiLSTM模型通过大量正面数据和负面数据训练生成。
请参阅图3，显示为本发明的公众满意度的分析方法于一实施例中的属性确定流程图。 如图3所示，S13具体包括以下步骤：
S131，分析所述舆论数据的敏感性，若所述舆论数据为非敏感数据，则将所述舆论数据确定为正面数据；若所述舆论数据为敏感数据，则将所述舆论数据确定为负面数据。
具体地，对舆论数据进行敏感性分类，将非敏感的舆论数据划分为正面数据，将敏感的舆论数据划分为负面数据。将舆论正面数据的条数作为舆论正面信息量，将舆论负面数据的条数作为舆论负面信息量。
进一步地，对舆论数据进行情绪分类，得到负面信息占比。
具体地，情绪分类可以使用一个情绪分类模型，例如可以是多分类Bert模型。进行喜、怒、悲伤、恐惧、惊讶、中性的多分类。假设一共N条数据，经过情绪分类模型后，得到如下4中负面情绪：情绪怒的舆论数据为a条、情绪悲伤的舆论数据为b条、情绪恐惧的舆论数据为c条、情绪惊讶的舆论数据为d条，则负面信息占比＝(a+b+c+d)/N。
S132，对所述媒体数据进行事件聚类，确定所述媒体数据的事件信息和所述媒体数据的数据信息。
于一实施例中，所述媒体数据的事件信息包括媒体正面事件个数和媒体负面事件个数；所述媒体数据的数据信息包括媒体正面事件正面信息量和媒体负面事件负面信息量。
具体地，对媒体数据进行事件聚类，确定媒体正面事件和媒体负面事件，分别统计媒体正面事件个数和媒体负面事件个数。并对单条数据进行敏感性分类，根据敏感性信息数量，判断媒体数据为媒体正面数据或媒体负面数据，将统计的媒体正面数据的条数作为媒体正面事件正面信息量，将统计的媒体负面数据的条数作为媒体负面事件负面信息量。
于实际应用中，例如有100条媒体数据，经过事件聚类后确定正面事件个数为1个，负面事件个数为1个，经过单条媒体数据的敏感性分类后确定媒体正面数据为60条，媒体负面数据为40条，则媒体正面事件正面信息量为60，媒体负面事件负面信息量为40。
具体地，网络举报数据均认为是负面信息，通过事件聚类，得到网络举报负面事件个数。
S133，对所述网络举报数据进行事件聚类，确定所述网络举报数据的事件信息。
于一实施例中，所述网络举报数据的事件信息是指网络举报负面事件个数。
具体地，针对媒体数据和网络举报数据的聚类，原理如下：对每条媒体数据或网络举报数据的文本进行分词，使用word2vec模型，并求平均值得到句子的句向量。对所有文本的句向量进行相似度计算，相似度高的文本作为一类事件来聚类。其中，Word2vec，是一群用来产生词向量的相关模型。这些模型为浅而双层的神经网络，用来训练以重新建构语言学之词文本。
S14，基于所有类别的公众反馈数据的属性，确定公众满意度。
请参阅图4，显示为本发明的公众满意度的分析方法于一实施例中的公众满意度分析流程图。如图4所示，S14具体包括以下步骤：
S141，基于所有类别的公众反馈数据的属性，确定所述公众反馈数据的正面得分或负面得分。
于一实施例中，所述正面得分包括舆论正面加分和/或媒体正面加分；所述负面得分包括舆论负面减分、媒体负面减分和/或网络举报负面减分。
具体地，根据整理数据，通过对信息量和事件个数类变量进行独立对数化，或对信息量 事 件个数进行对数化，得到各类数据的正负面得分。包括：舆论负面减分、舆论正面加分、媒体负面减分、媒体正面加分、网络举报负面减分。特别的，网络举报正面加分设置为0。
其中，对数化所用到的函数为：s＝ln(a)。a表示信息量和事件个数或信息量 事件个数，s表示正负面得分。
S142，分析所述公众反馈数据的负面情绪分布。
具体地，舆论负面情绪分布＝负面信息占比*f。其中f表示缩放比例因子，用于将舆论负面情绪分布设置为与其他分数量级相同的转换数值。例如，舆论正面加分范围在0-0.1，负面信息占比为0-0.01，则令f＝10，舆论负面情绪分布也变成0-0.1之间。
S143，根据所述正面得分、所述负面得分和所述负面情绪分布计算所有类别的公众反馈数据的正面总得分或负面总得分。
具体地，正面总得分为总正面加分，负面总得分为总负面减分，通过加权分别计算出总负面减分和总正面加分：
总负面减分＝a*舆论负面减分+b*舆论负面情绪分布+c*媒体负面减分+d*网络举报负面减分；
总正面加分＝o*舆论正面加分+p*(1-舆论负面情绪分布)+q*媒体正面加分+r*网络举报正面加分。
其中，权重值a+b+c+d＝1，权重值o+p+q+r＝1。
于实际应用中，设权重值a、b、c、d分别为0.15、0.25、0.40、0.20，设权重值o、p、q、r分别为0.15、0.25、0.40、0.20。除此之外，其他根据业务需求以及重点侧重哪类数据所设置的合理权重值均在本发明保护的范围内。
S144，结合所述正面总得分和所述负面总得分确定满意度总分。
具体地，总分＝总正面加分-总负面减分。
S145，修正所述满意度总分，根据修正后的满意度总分评估所述公众满意度。
于一实施例中，所述修正所述满意度总分的步骤包括：
判断所述满意度总分是否大于0，若所述满意度总分大于0，则利用第一函数对所述满意度总分进行修正；若所述满意度总分不大于0，则利用第二函数对所述满意度总分进行修正。
具体地，根据设定的阈值进行sigmoid分数修正。Sigmoid函数是一个常见的S型函数，也称为S型生长曲线。在信息科学中，由于其单增以及反函数单增等性质，Sigmoid函数常被用作神经网络的激活函数，将变量映射到0，1之间。
于实际应用中，当总分大于0时，使用第一函数计算：
Figure PCTCN2022107244-appb-000001
当总分小于或等于0时，使用第二函数计算：
Figure PCTCN2022107244-appb-000002
其中，x表示原始总分，y表示修正后的分数，i，j，m，n为公式常数。所述公式常数不是特定的某一个值，是根据不同事件类型分别得到的。即不同的事件类型取值不同。实际设置过程中是先选定某一事件方向，根据不同城市的数据和结果，反复根据业务方的反馈来进行调整的。
于实际应用中，主要是对sigmoid中的i，j，m，n进行调整，以达到与人为主观感受更加相符的得分。比如一个城市出现重大卫生安全事故，那么如果现在关心的是城市卫生安全方面的满意度，那么满意度在这段时间内会出现下降。再比如，对于某些方面有比如主流媒体的排名，会作为城市间对比的依据。
在i，j，m，n参数的具体调整过程中，比如得到一个城市一个月的数据，业务方会反馈给我们比如分数整体偏高或偏低，或对不同输入是否敏感等。i，m可以帮助控制对输入的敏感性，j，n可以帮助控制整体分数高低。
需要说明的是，Sigmoid函数的特点包括：(1)是单调递增的。(2)函数值范围在(0，1)之间。(3)可以接受正数和负数的输入。(4)当输入越大或越小的时候变化越缓慢。除此之外，其他的具备上述特点(1)-(4)的函数用于修正也在本发明保护的范围内。
进一步地，本发明中还可以在对公众反馈数据进行分析处理之前进行不同地区的划分，针对不同地区进行公众满意度的分析，进而定量显示、比较直观地呈现不同地区的分析结果，且对不同地区的分析结果进行对比，分析不同地区针对同一类事件(例如生态环境方面或公共卫生安全方面)公众满意度的差异。具体地，设采集时间段为1小时，则针对A城市和B 城市，均会每隔1小时生成一个y值，由此，A城市和B城市的多个y值可以通过统计图、Excel表等形式向用户呈现A城市或B城市随着时间变化公众满意度的变化，也可以呈现A城市与B城市在同一时间段内公众满意度的区别，还可以呈现任何可以通过y值直接或间接能获知的公众满意度信息。
于本发明不同的实施例中，可以只根据数据的正面与负面属性进行公众满意度的分析，也可以只根据聚类后事件的正面与负面属性进行分析，还可以结合数据的正面与负面属性、事件的正面与负面属性进行分析，结合分析为优选方式，但其他单独的分析方式也在本发明保护的范围内。
由此，本发明基于正负面敏感分类后事件、信息统计的公众满意度计算方式，相比于传统各类指数计算方式针对原始数据量的直接缩放、加权计算方式，先对数据进行敏感性分类，考虑正负面加减分，可以进一步体现出公众的满意度。本发明根据特定阈值调整的变体sigmoid函数进行分数修正的分数计算方式，使得分数计算更加灵活多变，可以根据大量实际数据进行函数的参数调整，以达到与人为主观感受更加相符的得分。
本发明所述的公众满意度的分析方法的保护范围不限于本实施例列举的步骤执行顺序，凡是根据本发明的原理所做的现有技术的步骤增减、步骤替换所实现的方案都包括在本发明的保护范围内。
本实施例提供一种计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现所述公众满意度的分析方法。
本领域普通技术人员可以理解：实现上述各方法实施例的全部或部分步骤可以通过计算机程序相关的硬件来完成。前述的计算机程序可以存储于一计算机可读存储介质中。该程序在执行时，执行包括上述各方法实施例的步骤；而前述的计算机可读存储介质包括：ROM、RAM、磁碟或者光盘等各种可以存储程序代码的计算机存储介质。
请参阅图5，显示为本发明的电子设备于一实施例中的结构连接示意图。如图5所示，本实施例提供一种电子设备5，具体包括：处理器51及存储器52；所述存储器52用于存储计算机程序，所述处理器51用于执行所述存储器52存储的计算机程序，以使所述电子设备5执行所述公众满意度的分析方法的各个步骤。
上述的处理器51可以是通用处理器，包括中央处理器(Central Processing Unit，简称CPU)、网络处理器(Network Processor，简称NP)等；还可以是数字信号处理器(Digital Signal Processing，简称DSP)、专用集成电路(Alication Specific Integrated Circuit，简称ASIC)、现场可编程门阵列(Field Programmable GateArray，简称FPGA)或者其他可编程逻辑器件、分 立门或者晶体管逻辑器件、分立硬件组件。
上述的存储器52可能包含随机存取存储器(Random Access Memory，简称RAM)，也可能还包括非易失性存储器(non-volatilememory)，例如至少一个磁盘存储器。
于实际应用中，所述电子设备可以是包括存储器、存储控制器、一个或多个处理单元(CPU)、外设接口、RF电路、音频电路、扬声器、麦克风、输入/输出(I/O)子系统、显示屏、其他输出或控制设备，以及外部端口等组件的计算机；所述计算机包括但不限于如台式电脑、笔记本电脑、平板电脑、智能手机、智能电视、个人数字助理(Personal Digital Assistant，简称PDA)等个人电脑，在另一些实施方式中，所述电子设备还可以是服务器，所述服务器可以根据功能、负载等多种因素布置在一个或多个实体服务器上，也可以是由分布的或集中的服务器集群构成的云服务器，本实施例不作限定。
综上所述，本发明所述公众满意度的分析方法、存储介质及电子设备通过对公众反馈数据进行类别划分，确定每一类别的公众反馈数据的正面属性或负面属性，以此来分析公众满意度。本发明与现有技术相比，可以从数据正负面信息与事件正负面信息进行双向考虑，可以在公众关注度、传播度的基础上进一步分析公众满意度。本发明有效克服了现有技术中的种种缺点而具有高度产业利用价值。
上述实施例仅例示性说明本发明的原理及其功效，而非用于限制本发明。任何熟悉此技术的人士皆可在不违背本发明的精神及范畴下，对上述实施例进行修饰或改变。因此，举凡所属技术领域中具有通常知识者在未脱离本发明所揭示的精神与技术思想下所完成的一切等效修饰或改变，仍应由本发明的权利要求所涵盖。

Claims (10)

  1. 一种公众满意度的分析方法，其特征在于，所述公众满意度的分析方法包括：
    采集公众反馈数据；
    对所述公众反馈数据进行类别划分，确定所述公众反馈数据的类别；
    根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性；所述属性包括正面属性和负面属性；
    基于所有类别的公众反馈数据的属性，确定公众满意度。
  2. 根据权利要求1所述的公众满意度的分析方法，其特征在于，所述采集公众反馈数据的步骤包括：
    设置一采集时间段；
    在所述采集时间段内，获取所有渠道下产生的公众反馈数据。
  3. 根据权利要求1所述的公众满意度的分析方法，其特征在于，所述公众反馈数据的类别包括：舆论数据、媒体数据或网络举报数据中的至少一种。
  4. 根据权利要求3所述的公众满意度的分析方法，其特征在于，所述根据所述公众反馈数据的敏感性确定每一类别的公众反馈数据的属性的步骤包括：
    分析所述舆论数据的敏感性，若所述舆论数据为非敏感数据，则将所述舆论数据确定为正面数据；若所述舆论数据为敏感数据，则将所述舆论数据确定为负面数据；
    对所述媒体数据进行事件聚类，确定所述媒体数据的事件信息和所述媒体数据的数据信息；
    对所述网络举报数据进行事件聚类，确定所述网络举报数据的事件信息。
  5. 根据权利要求4所述的公众满意度的分析方法，其特征在于，
    所述媒体数据的事件信息包括媒体正面事件个数和媒体负面事件个数；
    所述媒体数据的数据信息包括媒体正面事件正面信息量和媒体负面事件负面信息量；
    所述网络举报数据的事件信息是指网络举报负面事件个数。
  6. 根据权利要求1所述的公众满意度的分析方法，其特征在于，所述基于所有类别的公众反馈数据的属性，确定公众满意度的步骤包括：
    基于所有类别的公众反馈数据的属性，确定所述公众反馈数据的正面得分或负面得分；
    分析所述公众反馈数据的负面情绪分布；
    根据所述正面得分、所述负面得分和所述负面情绪分布计算所有类别的公众反馈数据的正面总得分或负面总得分；
    结合所述正面总得分和所述负面总得分确定满意度总分；
    修正所述满意度总分，根据修正后的满意度总分评估所述公众满意度。
  7. 根据权利要求6所述的公众满意度的分析方法，其特征在于，
    所述正面得分包括舆论正面加分和/或媒体正面加分；
    所述负面得分包括舆论负面减分、媒体负面减分和/或网络举报负面减分。
  8. 根据权利要求6所述的公众满意度的分析方法，其特征在于，所述修正所述满意度总分的步骤包括：
    判断所述满意度总分是否大于0，若所述满意度总分大于0，则利用第一函数对所述满意度总分进行修正；若所述满意度总分不大于0，则利用第二函数对所述满意度总分进行修正。
  9. 一种计算机可读存储介质，其上存储有计算机程序，其特征在于，该计算机程序被处理器执行时实现权利要求1至8中任一项所述的公众满意度的分析方法。
  10. 一种电子设备，其特征在于，包括：处理器及存储器；
    所述存储器用于存储计算机程序，所述处理器用于执行所述存储器存储的计算机程序，以使所述电子设备执行如权利要求1至8中任一项所述的公众满意度的分析方法。
PCT/CN2022/107244 2021-11-10 2022-07-22 公众满意度的分析方法、存储介质及电子设备 WO2023082698A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111323562.6 2021-11-10
CN202111323562.6A CN113779258B (zh) 2021-11-10 2021-11-10 公众满意度的分析方法、存储介质及电子设备

Publications (1)

Publication Number Publication Date
WO2023082698A1 true WO2023082698A1 (zh) 2023-05-19

Family

ID=78873755

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/107244 WO2023082698A1 (zh) 2021-11-10 2022-07-22 公众满意度的分析方法、存储介质及电子设备

Country Status (2)

Country Link
CN (1) CN113779258B (zh)
WO (1) WO2023082698A1 (zh)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113779258B (zh) * 2021-11-10 2022-03-11 上海蜜度信息技术有限公司 公众满意度的分析方法、存储介质及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104137128A (zh) * 2011-12-27 2014-11-05 汤姆森路透社全球资源公司 用于使用源于社交媒体的数据和情绪分析来生成企业绿色分数的方法及系统
CN110134849A (zh) * 2019-05-20 2019-08-16 瑞森网安(福建)信息科技有限公司 一种网络舆情监控方法及系统
CN111914096A (zh) * 2020-07-06 2020-11-10 同济大学 基于舆情知识图谱的公共交通乘客满意度评价方法及系统
CN112612895A (zh) * 2020-12-29 2021-04-06 中科院计算技术研究所大数据研究院 一种主体话题态度指数的计算方法
TWM617933U (zh) * 2021-07-02 2021-10-01 大數軟體有限公司 新聞及公眾意見分析系統
CN113779258A (zh) * 2021-11-10 2021-12-10 上海蜜度信息技术有限公司 公众满意度的分析方法、存储介质及电子设备

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8554701B1 (en) * 2011-03-18 2013-10-08 Amazon Technologies, Inc. Determining sentiment of sentences from customer reviews
US10474752B2 (en) * 2011-04-07 2019-11-12 Infosys Technologies, Ltd. System and method for slang sentiment classification for opinion mining
CN102929861B (zh) * 2012-10-22 2015-07-22 杭州东信北邮信息技术有限公司 一种文本情感指数计算方法和系统
CN110020159B (zh) * 2017-12-11 2021-05-07 网智天元科技集团股份有限公司 基于数据特征的舆情分析方法和系统
CN109271512B (zh) * 2018-08-29 2023-11-24 中国平安保险(集团)股份有限公司 舆情评论信息的情感分析方法、装置及存储介质
CN109684481A (zh) * 2019-01-04 2019-04-26 深圳壹账通智能科技有限公司 舆情分析方法、装置、计算机设备和存储介质
CN110232149B (zh) * 2019-05-09 2022-03-01 北京邮电大学 一种热点事件检测方法和系统
CN111026868B (zh) * 2019-12-05 2022-07-15 厦门市美亚柏科信息股份有限公司 一种多维度舆情危机预测方法、终端设备及存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104137128A (zh) * 2011-12-27 2014-11-05 汤姆森路透社全球资源公司 用于使用源于社交媒体的数据和情绪分析来生成企业绿色分数的方法及系统
CN110134849A (zh) * 2019-05-20 2019-08-16 瑞森网安(福建)信息科技有限公司 一种网络舆情监控方法及系统
CN111914096A (zh) * 2020-07-06 2020-11-10 同济大学 基于舆情知识图谱的公共交通乘客满意度评价方法及系统
CN112612895A (zh) * 2020-12-29 2021-04-06 中科院计算技术研究所大数据研究院 一种主体话题态度指数的计算方法
TWM617933U (zh) * 2021-07-02 2021-10-01 大數軟體有限公司 新聞及公眾意見分析系統
CN113779258A (zh) * 2021-11-10 2021-12-10 上海蜜度信息技术有限公司 公众满意度的分析方法、存储介质及电子设备

Also Published As

Publication number Publication date
CN113779258B (zh) 2022-03-11
CN113779258A (zh) 2021-12-10

Similar Documents

Publication Publication Date Title
CN111079639B (zh) 垃圾图像分类模型构建的方法、装置、设备及存储介质
CN110674850A (zh) 一种基于注意力机制的图像描述生成方法
US20120095943A1 (en) System for training classifiers in multiple categories through active learning
CN111814842B (zh) 基于多通路图卷积神经网络的对象分类方法及装置
CN111538828B (zh) 文本情感分析方法、装置、计算机装置及可读存储介质
CN108073988B (zh) 一种基于强化学习的法律认知方法、装置和介质
CN112348629A (zh) 一种商品信息推送方法和装置
CN107688870A (zh) 一种基于文本流输入的深度神经网络的分层因素可视化分析方法及装置
WO2023082698A1 (zh) 公众满意度的分析方法、存储介质及电子设备
Khashman et al. Anticipation of political party voting using artificial intelligence
CN113627151B (zh) 跨模态数据的匹配方法、装置、设备及介质
WO2019242453A1 (zh) 信息处理方法及装置、存储介质、电子装置
CN110019556B (zh) 一种话题新闻获取方法、装置及其设备
Khan et al. News popularity prediction with ensemble methods of classification
CN116522013B (zh) 基于社交网络平台的舆情分析方法及系统
CN115905648B (zh) 基于高斯混合模型的用户群和金融用户群分析方法及装置
Hantke et al. Trustability-based dynamic active learning for crowdsourced labelling of emotional audio data
KR102053858B1 (ko) 콘텐츠 평점 산출 방법
CN112987940B (zh) 一种基于样本概率量化的输入方法、装置和电子设备
Rezaeenour et al. Developing a new hybrid intelligent approach for prediction online news popularity
US20230260044A1 (en) Generation method and information processing apparatus
Sebastiani Sentiment Quantification of User-Generated Content.
Shandilya et al. Fairness for whom? Understanding the reader’s perception of fairness in text summarization
KR102624636B1 (ko) 자체 피드백 기반의 설문지 생성 시스템
KR102624634B1 (ko) 자가학습 기반 인공지능을 이용한 비대면 설문조사에서 중도이탈 응답 대체 시스템

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22891507

Country of ref document: EP

Kind code of ref document: A1