WO2018184518A1

WO2018184518A1 - Microblog data processing method and device, computer device and storage medium

Info

Publication number: WO2018184518A1
Application number: PCT/CN2018/081697
Authority: WO
Inventors: 王健宗; 黄章成; 吴天博; 肖京
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-04-07
Filing date: 2018-04-03
Publication date: 2018-10-11
Also published as: CN107797983A

Abstract

A microblog data processing method comprises: detecting a hot event in real-time by monitoring microblog streaming data (402); analyzing emotional tendencies shown in microblogs containing the hot event (404); determining corresponding emotional index values according to the analysis on the emotional tendencies (406); and generating a trend chart of time series of emotional indexes according to the emotional index values (408).

Description

Microblog data processing method, device, computer device and storage medium

This application claims the priority of the Chinese patent application filed on April 7, 2017, the Chinese Patent Office, the application number is 2017102256815, and the invention is entitled "microblog data processing method, device, computer equipment and storage medium". The citations are incorporated herein by reference.

Technical field

The present application relates to the field of computer processing, and in particular, to a microblog data processing method, apparatus, computer device, and storage medium.

Background technique

With the development of social media, social networking sites, online communities, and Weibo have gradually become an indispensable part of people's lives, and also the main channel for information dissemination in the current era. The tremendous changes in the form of information dissemination have impacted investors' original information utilization patterns and investment beliefs, and have profoundly affected the information transmission and financial ecology of capital markets. Behavioral finance theory believes that investors' investment decision-making behavior is affected by factors such as investor's attention and emotion. However, although the traditional stock market forecasting model also involves emotional sentiment analysis, most sentiment analysis is based on news pages. Can not truly reflect the attention and emotions of investors, resulting in large deviations in the stock market forecast.

Summary of the invention

According to various embodiments of the present application, a microblog data processing method, apparatus, computer device, and storage medium are provided.

A microblog data processing method, comprising:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

Determining a corresponding emotional index value according to the sentiment orientation analysis; and

A time series of emotional index charts is generated based on the emotional index values.

A microblog data processing device includes:

a discovery module, configured to discover hotspot events in real time by monitoring microblogging stream data;

An analysis module, configured to perform an emotional tendency analysis on the microblog containing the hot event;

a determining module, configured to determine a corresponding emotional index value according to the sentiment orientation analysis; and

And a generating module, configured to generate a time series of emotional index charts according to the emotional index value.

A computer device comprising a memory and one or more processors having stored therein computer readable instructions, the computer readable instructions being executed by the one or more processors such that the one or more The processors perform the following steps:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

One or more non-transitory computer readable storage mediums storing computer readable instructions, when executed by one or more processors, cause the one or more processors to perform the following steps:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

Details of one or more embodiments of the present application are set forth in the accompanying drawings and description below. Other features, objects, and advantages of the invention will be apparent from the description and appended claims.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings used in the description of the embodiments will be briefly described below. It is obvious that the drawings in the following description are only some embodiments of the present application. Other drawings may also be obtained from those of ordinary skill in the art in light of the inventive work.

FIG. 1 is an application scenario diagram of a microblog data processing method according to one or more embodiments.

2 is a block diagram of the internal structure of a terminal in accordance with one or more embodiments.

3 is a block diagram of a server in accordance with one or more embodiments.

4 is a flow diagram of a method of microblog data processing in accordance with one or more embodiments.

5 is a flow diagram of a method for discovering hotspot events in real time by monitoring microblogging stream data in accordance with one or more embodiments.

6 is a flow diagram of a method for performing sentiment orientation analysis of a microblog containing a hotspot event, in accordance with one or more embodiments.

7 is a flow diagram of a method for determining a corresponding sentiment index value based on sentiment orientation analysis in accordance with one or more embodiments.

FIG. 8 is a block diagram of a microblog data processing apparatus in accordance with one or more embodiments.

9 is a block diagram of a discovery module in accordance with one or more embodiments.

Figure 10 is a block diagram of an analysis module in accordance with one or more embodiments.

11 is a block diagram of a determination module in accordance with one or more embodiments.

detailed description

In order to make the objects, technical solutions, and advantages of the present application more comprehensible, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the application and are not intended to be limiting.

The microblog data processing method provided by the present application can be applied to the application scenario shown in FIG. 1 . The terminal 102 communicates with the server 104 over a network. The terminal 102 can be, but is not limited to, various personal computers, notebook computers, smart phones, tablets, and portable wearable devices, and the server 104 can be implemented with a stand-alone server or a server cluster composed of a plurality of servers. First, the terminal 102 uploads the microblog stream data to the server 104. The server 104 discovers the hotspot event in real time by monitoring the microblog stream data, performs sentiment orientation analysis on the microblog containing the hot event, and determines corresponding corresponding according to the sentiment tendency analysis. An emotional index value; and an emotional index chart that generates a time series based on the emotional index value.

As shown in FIG. 2, in one embodiment, the internal structure of the terminal 102 is as shown in FIG. 2, including a processor, a memory, a network interface, a display screen, and an input device connected through a system bus. The processor of the terminal is used to provide calculation and control capabilities to support the operation of the entire terminal. The memory of the terminal includes a non-transitory computer readable storage medium, an internal memory. The non-transitory computer readable storage medium of terminal 102 stores an operating system and computer readable instructions that, when executed, cause the processor to perform a method of processing microblog data. The internal memory provides an environment for the operation of an operating system and computer readable instructions in a non-transitory computer readable storage medium. The network interface is used to connect to the network for communication. The display screen of the terminal 102 may be a liquid crystal display or an electronic ink display screen. The input device may be a touch layer covered on the display screen, or may be a button, a trackball or a touchpad provided on the outer casing of the electronic device, or may be An external keyboard, trackpad, or mouse. The terminal can be a tablet, a laptop, a desktop computer, or the like. It will be understood by those skilled in the art that the structure shown in FIG. 2 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the terminal to which the solution of the present application is applied. The specific terminal may include a ratio. More or fewer components are shown in the figures, or some components are combined, or have different component arrangements.

As shown in FIG. 3, in one of the embodiments, the internal structure of the server 104 is as shown in FIG. 3, including a processor, a memory, and a network interface connected through a system bus. Among them, the server's processor is used to provide computing and control capabilities, supporting the operation of the entire server. The memory of the server includes a non-transitory computer readable storage medium, an internal memory. The non-transitory computer readable storage medium can store operating systems and computer readable instructions. When the computer instructions are executed, the processor can be caused to perform a microblog data processing method. The server's network interface is used to communicate with external servers and terminals over a network connection. Those skilled in the art can understand that the structure shown in FIG. 3 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the server to which the solution of the present application is applied. The specific server may include a ratio. More or fewer components are shown in the figures, or some components are combined, or have different component arrangements.

As shown in FIG. 4, in one embodiment, a microblog data processing method is proposed, which can be applied to a terminal or a server, and specifically includes the following steps:

Step 402: Discover hotspot events in real time by monitoring the microblog stream data.

In one of the embodiments, the hotspot event refers to an event with a current degree of interest or a relatively large influence. By monitoring the data flow data of the Weibo platform, the current hot topic can be found in real time, and the event corresponding to the hot topic is a hot event. Hot events can reflect investors' attention and emotional tendencies, so timely detection of hot events is conducive to discovering investors' attention and emotional tendencies.

In one of the embodiments, the real-time hot topic of the microblog platform is discovered through a social media real-time hotspot discovery algorithm. Specifically, firstly, the real-time main microblog flow of the microblog platform is obtained by dynamically updating the microblog flow algorithm. Among them, the main microblog stream refers to the most representative part of the microblog data stream sampled from a large number of real-time microblog data streams. The dynamic update microblog flow algorithm actually refers to an update rule for the monitored microblog account, and the update rule is used to ensure that the microblog data stream obtained through the monitoring account can fully reflect the overall microblog flow. For example, the Weibo data stream was originally built by collecting Weibo from the selected 100,000 Weibo accounts, but the 100,000 accounts may have different operating conditions over time, such as , stop updating, hacking, or the like, or the real-time microblogging stream of the 100,000 Weibo accounts can not represent the entire real-time microblog data stream, then you need to dynamically update the selected Weibo account, for example, delete more than Add a new active account to the account that is not updated for 5 days. Secondly, the main microblog stream is vectorized to calculate the acceleration of the first-order word frequency and the second-order word frequency. Among them, vectorization means that the obtained main microblog stream is expressed in the form of a word vector. Specifically, the word2vec method can be used to represent the main microblog stream as a vector form, and word2vec is a tool for converting words into a vector form. You can simplify the processing of text content into vector operations in vector space. Because the problem of natural language understanding is to be converted to machine learning, you first need to mathematicalize the text, which is to convert the text into a mathematical representation. For example, the microphone is expressed as [0.792, -0.177, -0.17, 0.109.-0.542, ...]. The text is represented by a mathematical vector. The semantic similarity of the text can be expressed by calculating the similarity in the vector space. The first-order word frequency refers to the frequency at which a single word appears, and the second-order word frequency refers to the frequency at which two words appear at the same time. Finally, according to the preset acceleration thresholds, real-time monitoring and discovery of microblog hot topics. For example, an acceleration threshold corresponding to the first-order word frequency and the second-order word frequency may be set. When the acceleration threshold is reached, the corresponding topic is determined as a hot topic, and the corresponding event is a hot event. In addition, it is also possible to obtain the migration of attention within the hot topic by paying attention to the change of the keyword, and by monitoring the migration of the focus of the hot event, the affected stock may be acquired in real time.

In another embodiment, first, the microblog stream data is acquired in real time, and the microblog stream data is vectorized; secondly, the vectorized microblog data stream is monitored, and the frequency and number of occurrences of each word are recorded; finally, according to each The frequency and frequency of occurrences of the word determine the current hot event. In this embodiment, in order to perform real-time monitoring on the microblog stream, the microblog stream data needs to be converted into a vector representation for monitoring, so after acquiring the microblog stream data, each microblog is converted into a vector. Form to represent. Then, the vectorized microblog data is monitored, and the frequency and number of occurrences of each word are recorded. When the frequency and frequency of occurrence of a word are relatively high, the corresponding event is determined as a hot event.

Step 404: Perform an emotional tendency analysis on the microblog containing the hot event.

In one embodiment, in order to obtain the investor's emotional tendency, after obtaining the microblog containing the hot event, an emotional tendency analysis is performed for each microblog. In terms of emotional distraction, according to the latest research results of psychology, human emotions are divided into four categories, namely happiness, sadness, anger and surprise. Specifically, in order to analyze the sentiment orientation of each microblog, it is necessary to establish an emotional tendency analysis model. Before the establishment of the sentiment orientation analysis model, it is first necessary to determine the training set for training the sentiment orientation analysis model. The training set is a four-category microblog data set after emotion annotation. The emotion annotation is a hybrid annotation method based on dynamic annotation dictionary and manual annotation. The advantage of the hybrid labeling method is that, on the one hand, the error of simply using one type of labeling method can be avoided, and on the other hand, the time of manual labeling can be saved. Secondly, features are extracted from the microblog texts after emotional annotation for training the sentiment orientation analysis model. In one embodiment, feature extraction is performed using an LDA model and a word2vec model, wherein the LDA model is used to predict the topic distribution of each microblog; the word2vec model is used to obtain a word vector representation for each microblog. In another embodiment, principal component analysis may also be employed to extract features. Finally, the sentiment orientation analysis model is used to analyze the sentiment orientation. If the sentiment orientation analysis model is trained based on the extracted LDA topic features and the word2vec word vector feature, the sentiment orientation analysis model will be acquired after the training is completed. The LDA topic feature corresponding to the new Weibo and the word2vec word vector feature are used together as an input vector of the sentiment orientation analysis model, and the sentiment orientation analysis model outputs the sentiment tendency corresponding to each Weibo. Emotional tendencies are also divided into four categories: happiness, sadness, anger and surprise.

Step 406: Determine a corresponding emotional index value according to the sentiment orientation analysis.

In one embodiment, after analyzing the sentiment orientation of the microblog containing the hot event according to the sentiment orientation analysis model, the number of microblogs corresponding to each type of emotion is counted, and according to the emotion category, the emotions can be divided into four categories, respectively. It is happiness, sadness, anger and surprise. The emotional index value corresponding to the hot event is calculated according to the number of microblogs corresponding to each type of emotion. Among them, the emotional index includes one or more of a happy emotional proportion, a sad emotional proportion, an angry emotional proportion, a surprise emotional proportion, and an emotional titer. Among them, the emotional valence refers to the reflection value of the overall microblogging sentiment, and the emotional valence calculation formula is:

SentmentValence=log{(1+P)/(1+N)}, where P is the total number of positive emotional microblogs, and N is the total number of negative emotional microblogs. The total number of positive emotional microblogs is the sum of the number of happy and amazed emotions, and the total number of negative emotions is the sum of the number of sad and angry emotions. Specifically, for example, the number of total microblogs related to hot events of the day is 100,000, of which, the number of microblogs of happy emotions is 30,000, the number of microblogs of sad emotions is 10,000, and the number of microblogs of angry emotions For 10,000, the number of microblogs with amazed emotions is 50,000, then the corresponding proportion of happy emotions is 0.3, the proportion of sadness and anger is 0.1, and the proportion of surprise emotions is 0.5. The emotional titer is log3.

Step 408: Generate a time series of emotional index charts according to the emotional index value.

In one embodiment, after determining the corresponding emotion index value by the sentiment orientation analysis, the emotion index value is correspondingly stored with the corresponding statistical time point or time period, and the emotion corresponding to each time point in the period is recorded. After the index value, a time series of sentiment index charts are generated in time order, and the sentiment index charts can fully and intuitively reflect the fluctuation of investor sentiment within a certain period of time. In one of the embodiments, by analyzing the correlation between the trend of the sentiment index and the trend of the stock market, the sentiment factor affecting the stock market trend can be determined. The sentiment factor can be used together with other stock market data as a predictor of the stock market forecasting model to make corresponding stock market forecasts. Due to the full consideration of the bettor's attention and emotion, it is beneficial to improve the accuracy of stock market forecasting.

In one embodiment, the hot spot event is detected in real time by monitoring the microblog stream data, and the sentiment tendency analysis is performed on the microblog containing the hot event, and then the corresponding sentiment index value is determined according to the sentiment orientation analysis, and the time is generated according to the sentiment index value. The emotional index chart of the sequence. The microblog data processing method selects a microblog hotspot event that can reflect the investor's attention and emotion, analyzes the emotional tendency of the microblog hot event, and generates a time series emotional index chart according to the emotional index value, The emotional index chart can fully and intuitively reflect the investor's attention and emotions. If applied to the stock market forecast, it will help improve the accuracy of the stock market forecast.

As shown in FIG. 5, in one embodiment, the step 402 of discovering hotspot events in real time by monitoring the microblog stream data includes:

Step 402A: Acquire microblog stream data in real time, and vectorize the microblog stream data.

In one embodiment, in order to obtain the attention and emotion of the investor in time, the microblog stream data is acquired in real time, and the initially obtained microblog data stream is text data, text data and ordinary numerical data or generic data. Differently, text data is a kind of semi-structured data. Before text data analysis, text data needs to be preprocessed, and vectorized values are used to express these semi-structured text data. In one embodiment, before the vectorization, the text data is subjected to word segmentation processing. After the word segmentation process, a text data can be represented as a multi-dimensional vector represented by a plurality of keywords, which facilitates subsequent monitoring of the feature words. .

Step 402B, monitoring the vectorized microblog stream data, and recording the frequency and frequency of occurrence of each feature word.

In one embodiment, the feature word refers to a vocabulary that can represent an event. Not all words in a weibo text need to be recorded, because some common words may appear in any event, so recording these commonly used words Words have no meaning, for example, the auxiliary words "", "land" and so on. After vectorizing the microblog stream data, the vectorized microblog stream data is monitored in real time, and the frequency and number of occurrences of each feature word are monitored. The frequency refers to the number of occurrences of the word in a unit time. The number of times refers to the total number of occurrences of the word. By monitoring the frequency of occurrence of each feature word, you can find hot topics that appear in a short period of time. By recording the number of occurrences of each feature word, you can examine the degree of attention of a topic over a long period of time.

Step 402C: Determine a current hot event according to the frequency and number of occurrences of each feature word.

In one embodiment, the frequency and the number of occurrences of each feature word are monitored and recorded, and when the frequency or number of one or more feature words reaches a preset threshold, the event corresponding to the one or more feature words is determined. For hot events. The frequency threshold and the number of times thresholds may be respectively set. In one embodiment, when the frequency of occurrence of the feature word reaches the frequency threshold or when the number of occurrences of the feature word reaches the threshold of the number of times, the event corresponding to the feature word is determined as the hot event. In another embodiment, when the frequency of occurrence of the feature word reaches the frequency threshold, and the number of times also reaches the threshold of the number of times, the event corresponding to the feature word is determined as the hot event. According to the frequency of the monitored one or more characteristic words, a hot event can be found in a short time, and by monitoring the number of occurrences of one or more characteristic words, the hot topic in a period of time can be inspected, thereby determining the popularity of the topic. And duration. This method facilitates the timely discovery and recording of hotspot events.

As shown in FIG. 6, in one embodiment, the step of performing an sentiment orientation analysis on the microblog containing the hot event includes:

Step 404A: Extract the acquired LDA topic feature and the word2vec word vector feature of each microblog containing the hot event.

In one embodiment, after the hotspot event is found by monitoring the microblog stream data, all the microblogs containing the hotspot event are obtained, and each of the obtained microblogs containing the hotspot event is respectively extracted with the corresponding LDA subject feature and word2vec. Word vector feature. The LDA theme model and the word2vec word vector model are pre-trained, wherein the topic distribution of the microblog can be predicted by inputting the vector text of each microblog into the LDA topic model, for example, the probability distribution of the first 250 dimensions is taken as the strip. The 250-dimensional feature of Weibo; you can obtain the vector representation of any Weibo word by inputting the vector text of each Weibo into the word2vec model. You can also select the first 250 dimensions as the feature, which can be obtained by simple simultaneous dimension addition. The 500-dimensional word vector representation of the microblog. Among them, word2vec is an efficient tool for characterizing words as real-valued vectors. It can use the idea of deep learning to simplify the processing of text content into vector operations in k-dimensional vector space.

In step 404B, the LDA topic feature and the word2vec word vector feature are substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output.

In one embodiment, after acquiring the LDA topic feature and the word2vec word vector feature corresponding to each microblog, the acquired feature is substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output. Among them, the sentiment orientation analysis model can be established by Boosted tree algorithm, among which Boosted tree algorithm is mainly used for prediction of multi-classification problems. In one of the embodiments, the emotional propensity is divided into four categories, namely happiness, sadness, anger, and surprise. Before adopting the sentiment orientation analysis model, we first need to establish an sentiment orientation analysis model. Specifically, the four-category annotation method is used to emotionally label the microblog dataset, and the emotion-labeled microblog dataset is used as the training set. In each centralized microblog, the LDA topic feature and the word2vec word vector feature are substituted into the initialized sentiment orientation analysis model to obtain the corresponding model parameters, and the final sentiment orientation analysis model is obtained. Among them, the initial sentiment orientation analysis model is established by using the Boosted tree method. Each feature needs to be standardized before training. The formula for index normalization is: ZXt=(Xt-X)/σ, where Xt is the specific value of each dimension, and ZXt is the corresponding value of each dimension of the data. X is the mean of all the values of the column features, and σ is the standard deviation of all the values of the column features. The formula is used to standardize the features of each dimension to facilitate the subsequent machine learning to obtain the sentiment orientation analysis model. Therefore, when analyzing the sentiment orientation of each microblog with hot events, it is necessary to extract the LDA theme feature and word2vec word vector in each microblog as the input vector of the sentiment orientation analysis model, and then output the sentiment tendency corresponding to each microblog. For example, whether it is happy or sad, angry or amazed.

As shown in FIG. 7, in one of the embodiments, the step of determining the corresponding emotional index value according to the sentiment orientation analysis comprises:

Step 406A: According to the sentiment orientation analysis, the microblogs containing the hot event are classified according to the emotion category, and the number of microblogs corresponding to each type of emotion is separately counted.

In one embodiment, after each sentiment containing the hot event is sentimentally analyzed, the microblog containing the hot event is classified according to the sentiment category, and the number of microblogs corresponding to each type of emotion is separately counted. Among them, the emotional categories are divided into four categories, namely happiness, sadness, anger and surprise. In order to obtain the investor's emotions, the number of microblogs corresponding to each type of emotion is separately counted, which facilitates the subsequent calculation of the sentiment index.

Step 406B: Determine an emotional index value corresponding to the hot event according to the counted number of microblogs corresponding to each type of emotion.

In one embodiment, the sentiment index includes one or more of a happy emotional proportion, a sad emotional proportion, an angry emotional proportion, a surprise emotional proportion, and an emotional titer. Among them, the emotional valence refers to the reflection value of the overall microblogging emotion, and the emotional valence calculation formula is: SentmentValence=log{(1+P)/(1+N)}, where P is the total number of positive emotional microblogs, N The total number of microblogs for negative emotions. The total number of positive emotional microblogs is the sum of the number of happy and amazed emotions, and the total number of negative emotions is the sum of the number of sad and angry emotions. The emotional index value is the value of the corresponding emotional index. For example, the number of total microblogs related to hot events of the day is 100,000, of which the number of microblogs with happy emotions is 30,000, the number of microblogs with sad emotions is 10,000, and the number of microblogs with angry emotions is 1. 10,000, the number of microblogs of surprise emotion is 50,000, then the emotional index values include: happy emotional proportion 0.3, sadness and anger emotional ratio are 0.1, surprise emotional proportion 0.5, emotional valence log3. Subsequent calculations of the stock market based on the calculated emotional index values. It should be noted that the hot event here does not refer to a hot event, but all hot events in a period of time.

In one embodiment, after the step of generating a time series of the emotional index chart according to the emotional index value, the following comprises: determining an affective factor affecting the stock market data according to the sentiment index chart, and using the sentiment factor together with other stock market data as the stock market The predictors of the predictive model are correlated with the stock market forecast.

In one of the embodiments, after obtaining the time series of the emotional index charts, the emotional factors affecting the stock market data are determined by the emotional index charts. By analyzing the time series of sentiment index charts and the daily returns, prices and transaction amounts of the market, the correlation sentiment index is extracted as the forecast factor of the stock market, that is, the sentiment factors affecting the stock market data are determined. Among them, the emotional index includes emotional valence, happy emotional proportion, sad emotional proportion, anger emotional proportion, surprise emotional proportion and so on. Therefore, in the analysis, each factor needs to be analyzed to screen out one or more factors that have an impact on the stock market data, and together as an emotional factor of the stock market data. Then the determined sentiment factor and other traditional stock market factors are used as predictors of the stock market forecasting model to make corresponding stock market forecasts. Other stock market data include opening price, closing price, highest price, lowest price, turnover, and yield. Specifically, firstly, through the correlation analysis between the sentiment index and the daily yield, price and transaction amount of the market, the correlation sentiment index is extracted as the forecasting feature of the stock market, and the machine learning algorithm is used to train the emotional characteristics. The stock market forecasting model, according to the calculated sentiment index value (ie, the sentiment index value corresponding to the sentiment factor) and other stock market data obtained, use the stock market forecasting model to perform corresponding stock market forecasting. The stock market forecast here can predict the profit status of the market, and can also predict the profit status of a single stock. The different forecasting purposes are different. By considering the investor's attention and emotions and making corresponding stock market forecasts, it is conducive to improving the accuracy and reliability of stock market forecasts and providing a reliable basis for decision makers.

In addition, in one of the embodiments, by monitoring the microblog stream data, it is also possible to find a hot event that has a great impact on the society and lasts for a long time, and it is beneficial to the stock selection by analyzing the hot event with great influence. Because the hot events that affect the big cause will cause excessive attention of the investing public, causing irrational fluctuations in current prices, and returning to basic values in the future, there will be some buying opportunities in the middle. Specifically, first, determining an event influence of a hot event, when the event influence is greater than a preset threshold, selecting a stock or industry in the stock market name or a stock industry classification name that includes an event keyword to form a candidate stock pool; The information selects the fundamental stocks from the candidate stock pool; determines the influence on the relevant stocks according to the event period and the market trend, and then feeds back to the stock selection timing model according to the corresponding influence degree for stock selection. For example, in the short-term irrational fluctuations caused by prices such as gold and oil, the US general election can choose the right time to buy at the bottom.

It should be understood that although the various steps in the flowcharts of FIGS. 4 through 7 are sequentially displayed as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Except as explicitly stated herein, the execution of these steps is not strictly limited, and the steps may be performed in other orders. Moreover, at least some of the steps in Figures 4 to 7 may comprise a plurality of sub-steps or stages, which are not necessarily performed at the same time, but may be performed at different times, these sub-steps or stages The order of execution is not necessarily performed sequentially, but may be performed alternately or alternately with at least a portion of other steps or sub-steps or stages of other steps.

As shown in FIG. 8, in one embodiment, a microblog data processing apparatus is proposed, the apparatus comprising:

The discovery module 802 is configured to discover hotspot events in real time by monitoring the microblog stream data.

The analysis module 804 is configured to perform sentiment orientation analysis on the microblog containing the hot event.

The determining module 806 is configured to determine a corresponding emotional index value according to the sentiment orientation analysis.

The generating module 808 is configured to generate a time series of sentiment index charts according to the emotion index values.

As shown in FIG. 9, in one embodiment, the discovery module 702 includes:

The obtaining module 802A is configured to acquire microblog stream data in real time, and vectorize the microblog stream data.

The recording module 802B is configured to monitor the vectorized microblog stream data, and record the frequency and frequency of occurrence of each feature word.

The hotspot event determining module 802C is configured to determine a current hotspot event according to the frequency and number of occurrences of each feature word.

As shown in FIG. 10, in one embodiment, the analysis module 704 includes:

The extracting module 804A is configured to extract the acquired LDA topic feature and the word2vec word vector feature of each microblog containing the hot event.

The output module 804B is configured to substitute the LDA topic feature and the word2vec word vector feature into the sentiment orientation analysis model, and output the sentiment tendency of each microblog.

As shown in FIG. 11, in one of the embodiments, the determining module 706 includes:

The statistics module 806A is configured to classify the microblogs containing the hot event according to the sentiment analysis according to the sentiment analysis, and separately count the number of microblogs corresponding to each type of emotion.

The sentiment index value determining module 806B is configured to determine an emotional index value corresponding to the hotspot event according to the counted number of microblogs corresponding to each type of sentiment.

In one embodiment, the microblog data processing apparatus further includes: a stock market prediction module, configured to determine an sentiment factor affecting the stock market data according to the sentiment index chart, and use the sentiment factor together with other stock market data as a stock market forecast The predictive factors of the model are corresponding to the stock market forecast.

Each of the above-described microblog data processing devices may be implemented in whole or in part by software, hardware, and combinations thereof. The network interface may be an Ethernet card or a wireless network card. The above modules may be embedded in the hardware in the processor or in the memory in the server, or may be stored in the memory in the server, so that the processor calls the corresponding operations of the above modules. The processor can be a central processing unit (CPU), a microprocessor, a microcontroller, or the like.

The apparatus for controlling the underwriting process described above can be implemented in the form of a computer readable instruction that can be run on a terminal or server as shown in FIG. 2 or 3.

In one embodiment, a computer device is proposed. The internal structure of the computer device may correspond to the structure as shown in FIG. 2 or 3, that is, the computer device may be either a server or a terminal, and includes a series of storage. The computer readable instructions on the memory, when the computer readable instructions are executed by the processor, may implement the microblog data processing method proposed by the embodiments of the present application. The embodiment of the present application proposes a computer device. The computer device includes a series of computer readable instructions stored on a memory, and when the computer readable instructions are executed by the processor, the microblog data processing method proposed by the embodiments of the present application can be implemented. The computer device includes a memory, a processor, and computer readable instructions stored on the memory and operative on the processor, the processor executing the computer readable instructions to implement the step of monitoring a microblog stream The data is used to discover hot events in real time; the sentiment orientation analysis is performed on the microblogs containing the hot event; the corresponding emotional index values are determined according to the sentiment orientation analysis; and the emotional index charts of the time series are generated according to the emotional index values.

In one embodiment, the performing, by the processor, the real-time hotspot event by monitoring the microblog stream data includes: acquiring microblog stream data in real time, vectorizing the microblog stream data; and monitoring the vectorization The microblog stream data records the frequency and number of occurrences of each feature word; the current hot event is determined according to the frequency and frequency of occurrence of each feature word.

In one embodiment, the performing, by the processor, performing the sentiment orientation analysis on the microblog that will include the hotspot event comprises: extracting the acquired LDA topic feature of each microblog containing the hot event and word2vec The word vector feature; the LDA theme feature and the word2vec word vector feature are substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output.

In one embodiment, the determining, by the processor, the determining the corresponding sentiment index value according to the sentiment orientation analysis comprises: separately counting the number of microblogs corresponding to each type of emotion according to the sentiment category; The number of microblogs corresponding to each type of emotion determines the emotional index value corresponding to the hot event.

In one embodiment, the processor is further configured to: determine an affective factor affecting the stock market data according to the sentiment index chart, and use the sentiment factor together with other stock market data as a predictor of the stock market forecasting model Carry out the corresponding stock market forecast.

In one of the embodiments, one or more non-transitory computer readable storage media storing computer-executable instructions are provided, the computer-executable instructions being executed by one or more processors such that the one Or the plurality of processors perform the following steps: discovering hotspot events in real time by monitoring the microblogging stream data; performing sentiment orientation analysis on the microblogs containing the hotspot events; determining corresponding emotional index values according to the sentiment orientation analysis; The value generates a time series of emotional index charts.

A person skilled in the art can understand that all or part of the process of implementing the above embodiment method can be completed by a computer program to instruct related hardware, and the computer program can be stored in a computer readable storage medium. When executed, the flow of an embodiment of the methods as described above may be included. The storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

The technical features of the above-described embodiments may be arbitrarily combined. For the sake of brevity of description, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction between the combinations of these technical features, All should be considered as the scope of this manual.

The above-mentioned embodiments are merely illustrative of several embodiments of the present application, and the description thereof is more specific and detailed, but is not to be construed as limiting the scope of the invention. It should be noted that a number of variations and modifications may be made by those skilled in the art without departing from the spirit and scope of the present application. Therefore, the scope of the invention should be determined by the appended claims.

Claims

A microblog data processing method, comprising:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

Determining a corresponding emotional index value according to the sentiment orientation analysis; and

A time series of emotional index charts is generated based on the emotional index values.
The method according to claim 1, wherein the detecting hotspot events in real time by monitoring the microblog stream data comprises:

Acquiring the microblog stream data in real time, and vectorizing the microblog stream data;

Monitoring the vectorized microblog stream data, recording the frequency and number of occurrences of each feature word; and

The current hot event is determined based on the frequency and number of occurrences of each feature word.
The method according to claim 1, wherein the analyzing the sentiment of the microblog containing the hot event comprises:

Extracting the LDA topic feature and the word2vec word vector feature of each microblog containing the hot event; and

The LDA topic feature and the word2vec word vector feature are substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output.
The method according to claim 1, wherein said determining a corresponding emotional index value based on said sentiment orientation analysis comprises:

According to the sentiment orientation analysis, the microblogs containing the hot event are classified according to the emotion category, and the number of microblogs corresponding to each type of emotion is separately counted;

The emotion index value corresponding to the hot event is determined according to the counted number of microblogs corresponding to each type of emotion.
The method according to claim 1, wherein after the generating the time series of the emotional index chart according to the emotional index value, the method further comprises:

Determining an affective factor affecting stock market data based on the sentiment index chart; and

The sentiment factor is used together with other stock market data as a predictor of the stock market forecasting model to make corresponding stock market forecasts.
A microblog data processing device includes:

a discovery module, configured to discover hotspot events in real time by monitoring microblogging stream data;

An analysis module, configured to perform an emotional tendency analysis on the microblog containing the hot event;

a determining module, configured to determine a corresponding emotional index value according to the sentiment orientation analysis; and

And a generating module, configured to generate a time series of emotional index charts according to the emotional index value.
The apparatus of claim 6, the discovery module comprising:

An acquiring module, configured to acquire microblog stream data in real time, and vectorize the microblog stream data;

a recording module, configured to monitor the vectorized microblog stream data, and record the frequency and frequency of occurrence of each feature word;

The hotspot event determining module is configured to determine a current hotspot event according to the frequency and the number of times each feature word appears.
The apparatus according to claim 6, wherein the analysis module comprises:

An extracting module, configured to extract an LDA topic feature and a word2vec word vector feature of each microblog containing the hot event; and

The output module is configured to substitute the LDA topic feature and the word2vec word vector feature into the sentiment orientation analysis model, and output the sentiment tendency of each microblog.
The apparatus according to claim 6, wherein the determining module comprises:

a statistical module, configured to classify the microblogs containing the hot event according to the sentiment orientation according to the sentiment analysis, and separately count the number of microblogs corresponding to each type of emotion; and

The sentiment index value determining module is configured to determine an emotional index value corresponding to the hot event according to the counted number of microblogs corresponding to each type of sentiment.
The device according to claim 6, wherein the device further comprises:

The stock market forecasting module is configured to determine an sentiment factor affecting the stock market data according to the sentiment index chart, and use the sentiment factor together with other stock market data as a predictor of the stock market forecasting model to perform a corresponding stock market forecast.
A computer device comprising a memory and one or more processors, the memory storing computer readable instructions, the computer readable instructions being executed by the one or more processors such that the one or more The processors perform the following steps:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

Determining a corresponding emotional index value according to the sentiment orientation analysis; and

A time series of emotional index charts is generated based on the emotional index values.
The computer device according to claim 11, wherein the discovering hotspot events in real time by monitoring the microblogging stream data comprises:

Acquiring the microblog stream data in real time, and vectorizing the microblog stream data;

Monitoring the vectorized microblog stream data, recording the frequency and number of occurrences of each feature word; and

The current hot event is determined based on the frequency and number of occurrences of each feature word.
The computer device according to claim 11, wherein the analyzing the sentiment of the microblog containing the hot event comprises:

Extracting the LDA topic feature and the word2vec word vector feature of each microblog containing the hot event; and

The LDA topic feature and the word2vec word vector feature are substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output.
The computer device according to claim 11, wherein the determining the corresponding emotional index value according to the sentiment orientation analysis comprises:

According to the sentiment orientation analysis, the microblogs containing the hot event are classified according to the emotion category, and the number of microblogs corresponding to each type of emotion is separately counted;

The emotion index value corresponding to the hot event is determined according to the counted number of microblogs corresponding to each type of emotion.
The computer device according to claim 11, wherein after the generating an emotional index chart of the time series according to the emotional index value, the processor is further configured to perform the following steps:

Determining an affective factor affecting stock market data based on the sentiment index chart; and

The sentiment factor is used together with other stock market data as a predictor of the stock market forecasting model to make corresponding stock market forecasts.
One or more non-transitory computer readable storage mediums storing computer readable instructions, when executed by one or more processors, cause the one or more processors to perform the following steps:

Discover hot events in real time by monitoring Weibo stream data;

Conducting emotional sentiment analysis on Weibo containing the hot event;

Determining a corresponding emotional index value according to the sentiment orientation analysis; and

A time series of emotional index charts is generated based on the emotional index values.
The storage medium according to claim 16, wherein the discovering hotspot events in real time by monitoring the microblogging stream data comprises:

Acquiring the microblog stream data in real time, and vectorizing the microblog stream data;

Monitoring the vectorized microblog stream data, recording the frequency and number of occurrences of each feature word; and

The current hot event is determined based on the frequency and number of occurrences of each feature word.
The storage medium according to claim 16, wherein the analyzing the sentiment of the microblog containing the hot event comprises:

Extracting the LDA topic feature and the word2vec word vector feature of each microblog containing the hot event; and

The LDA topic feature and the word2vec word vector feature are substituted into the sentiment orientation analysis model, and the sentiment tendency of each microblog is output.
The storage medium according to claim 16, wherein the determining the corresponding emotional index value according to the sentiment orientation analysis comprises:

According to the sentiment orientation analysis, the microblogs containing the hot event are classified according to the emotion category, and the number of microblogs corresponding to each type of emotion is separately counted;

The emotion index value corresponding to the hot event is determined according to the counted number of microblogs corresponding to each type of emotion.
The storage medium according to claim 16, wherein after the generating an emotional index chart according to the emotional index value, the processor is further configured to perform the following steps:

Determining an affective factor affecting stock market data based on the sentiment index chart; and

The sentiment factor is used together with other stock market data as a predictor of the stock market forecasting model to make corresponding stock market forecasts.