Information processing method, device, equipment and medium based on sentiment analysis

Abstract

The present invention relates to the fields of artificial intelligence technology and financial technology, and discloses an information processing method based on sentiment analysis. The method comprises the following steps: collecting data and information, including text data, user behavior data, and event information; performing natural language processing and sentiment analysis on the text data to identify the sentiment characteristics of the text data; analyzing user behavior data to determine behavior patterns; analyzing event information to determine event change trends; generating sentiment state indicators based on the sentiment characteristics, behavior patterns, and event change trends; monitoring online platforms to obtain user sentiment feedback; comprehensively analyzing the sentiment state indicators and user sentiment feedback to generate sentiment analysis results; and generating target output content based on the sentiment analysis results. By combining sentiment analysis with behavior data and event change trends, the present invention can capture the user's emotional state in real time, dynamically adjust decision recommendations, and effectively respond to market or environmental changes.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence technology and financial technology, and in particular to an information processing method, device, equipment and storage medium based on sentiment analysis. Background Art

[0002] With the rapid development of financial markets, robo-advisory systems have become a crucial tool in investment management. Traditional robo-advisory systems rely on historical data, algorithmic models, and big data analytics to provide investors with asset allocation, risk assessment, and other services through analysis of market fundamentals and technical data. However, these systems still have limitations in terms of the accuracy and personalization of investment advice.

[0003] First, the core of traditional robo-advisory systems relies on historical data and fixed algorithmic models, ignoring the dynamics of market and investor sentiment. In actual investment decision-making, market sentiment often significantly influences asset price fluctuations. Especially during major financial market events or periods of significant market volatility, investor emotions can lead to irrational decisions. Traditional robo-advisory systems fail to fully account for this factor, resulting in a lack of timeliness and flexibility in responding to rapid market changes.

[0004] Secondly, traditional robo-advisory systems pay little attention to managing investor sentiment. Investor sentiment is a crucial factor influencing decision-making, but traditional systems typically rely solely on historical data and provide standardized investment advice based on fixed models, lacking dynamic analysis of individual investor sentiment fluctuations. This standardized advice model ignores the individual needs of investors and fails to effectively help them make sound investment decisions amidst fluctuating emotions.

[0005] Furthermore, traditional robo-advisory systems lack the ability to respond to market changes in real time. Because these systems rely primarily on historical data and static algorithmic models, they often struggle to capture the immediate impact of market trends or unexpected events on market sentiment. Real-time information such as market sentiment, news reports, and social media comments significantly influences market sentiment, but traditional advisory systems are limited in their ability to analyze this unstructured data, resulting in delayed responses to market changes and an inability to provide sufficiently flexible and dynamic investment advice. Summary of the Invention

[0006] The main purpose of the present invention is to provide an information processing method, device, equipment and storage medium based on sentiment analysis, aiming to solve the technical problems that the existing technology fails to combine market sentiment and user sentiment for analysis, reacts slowly to environmental changes, and cannot accurately provide users with satisfactory decisions.

[0007] To achieve the above objectives, the present invention provides an information processing method based on sentiment analysis, comprising:

[0008] Collect data and information, including text data, user behavior data, and event information;

[0009] Performing natural language processing and sentiment analysis on the text data to identify sentiment features of the text data;

[0010] Analyze the user behavior data to determine the user's behavior pattern;

[0011] Analyze the event information to determine the event change trend;

[0012] Determining emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends;

[0013] Monitor information on online platforms and obtain user sentiment feedback;

[0014] Comprehensively analyzing the emotional state indicators and user emotional feedback to generate emotional analysis results;

[0015] Target output content is generated based on the emotion analysis result to respond to the user's emotional state.

[0016] Furthermore, to achieve the above-mentioned purpose, the present invention provides an information processing device based on sentiment analysis, comprising:

[0017] A data collection module is used to collect data and information, including text data, user behavior data, and event information;

[0018] A sentiment analysis module, configured to perform natural language processing and sentiment analysis on the text data to identify sentiment features of the text data;

[0019] A user behavior analysis module, configured to analyze the user behavior data and determine the user's behavior pattern;

[0020] An event analysis and trend prediction module is used to analyze the event information and determine the event change trend;

[0021] An emotional state analysis module, configured to determine emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends;

[0022] User emotional feedback collection module, used to monitor information on online platforms and obtain user emotional feedback;

[0023] The comprehensive emotion analysis module is used to comprehensively analyze the emotional state indicators and user emotional feedback to generate emotion analysis results;

[0024] An output generation module is used to generate target output content for responding to the user's emotional state based on the emotion analysis result.

[0025] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer device, which includes a memory, a processor, and an information processing program based on sentiment analysis stored in the memory and runnable on the processor. When the information processing program based on sentiment analysis is executed by the processor, the steps of the information processing method based on sentiment analysis as described above are implemented.

[0026] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which an information processing program based on sentiment analysis is stored. When the information processing program based on sentiment analysis is executed by a processor, the steps of the information processing method based on sentiment analysis as described above are implemented.

[0027] Beneficial effects: The present invention relates to the fields of artificial intelligence technology and financial technology, and discloses an information processing method based on sentiment analysis, which collects data and information, including text data, user behavior data, and event information; performs natural language processing and sentiment analysis on the text data to identify the sentiment characteristics of the text data; analyzes user behavior data to determine behavior patterns; analyzes event information to determine event change trends; generates emotional state indicators based on emotional characteristics, behavior patterns, and event change trends; monitors online platforms to obtain user emotional feedback; comprehensively analyzes emotional state indicators and user emotional feedback to generate sentiment analysis results; and generates target output content based on the sentiment analysis results. The present invention can capture the user's emotional state in real time, dynamically adjust decision recommendations, and effectively respond to market or environmental changes by combining sentiment analysis with behavioral data and event change trends. In addition, the accuracy of personalized recommendations is further improved by combining analysis of user emotional feedback. This not only enhances the system's ability to respond to external changes, but also better meets the personalized needs of users and optimizes the overall decision-making process. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0029] Figure 1 A schematic diagram of an application environment of an information processing method based on sentiment analysis in one embodiment of the present invention;

[0030] Figure 2 This is a flow chart of an embodiment of an information processing method based on sentiment analysis according to the present invention;

[0031] Figure 3 This is a schematic diagram of the functional modules of a preferred embodiment of the information processing device based on sentiment analysis of the present invention;

[0032] Figure 4 A schematic diagram of the structure of a computer device according to an embodiment of the present invention;

[0033] Figure 5 FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0034] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0035] The information processing method based on sentiment analysis provided by the embodiment of the present invention can be applied in Figure 1 In an application environment, a user terminal communicates with a server terminal via a network. The server terminal can collect data and information from the user terminal, including text data, user behavior data, and event information; perform natural language processing and sentiment analysis on the text data to identify the emotional characteristics of the text data; analyze user behavior data to determine behavioral patterns; analyze event information to determine event trends; generate emotional state indicators based on the emotional characteristics, behavioral patterns, and event trends; monitor online platforms to obtain user emotional feedback; comprehensively analyze the emotional state indicators and user emotional feedback to generate emotional analysis results; and generate target output content based on the emotional analysis results. By combining sentiment analysis with behavioral data and event trends, the present invention can capture the user's emotional state in real time, dynamically adjust decision recommendations, and effectively respond to market or environmental changes. Furthermore, incorporating analysis of user emotional feedback further improves the accuracy of personalized recommendations. This not only enhances the system's ability to respond to external changes, but also better meets the user's personalized needs and optimizes the overall decision-making process. The user terminal can include, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. The server terminal can be implemented as a standalone server or a server cluster consisting of multiple servers. The present invention is described in detail below through specific embodiments.

[0036] See also Figure 2 , Figure 2 This is a flow chart of an embodiment of an information processing method based on sentiment analysis provided by the present invention. It should be noted that although a logical order is shown in the flow chart, in some cases, the steps shown or described may be performed in a different order than that shown here.

[0037] like Figure 2 As shown, the information processing method based on sentiment analysis proposed in the present invention includes the following steps:

[0038] S10, collecting data and information, wherein the data and information include text data, user behavior data, and event information;

[0039] In this embodiment, different types of data are obtained from multiple sources, including text data, user behavior data, and event information. These data together form the basis for analyzing user sentiment and market sentiment.

[0040] Text data refers to information obtained from unstructured textual content. This text can include social media comments, news articles, user messages, or other publicly available content. Text data is often used for sentiment analysis, analyzing sentiment to understand the market or user mood. Data scraping tools (such as web crawlers) can be used to obtain text data from social media, news websites, and other channels. APIs are used to access relevant platforms in real time to collect user comments, forum posts, news articles, and other content. Text data is typically stored in an unstructured format in databases to prepare for subsequent natural language processing (NLP).

[0041] User behavior data covers user actions within a platform or application, including clicks, page dwell time, browsing paths, search history, and more. This data helps depict user behavior patterns and reflect potential emotional shifts. Embed tracking scripts in applications or websites to record user actions. Collect data through log analysis or behavior tracking systems (such as Google Analytics or custom tracking tools). Real-time behavior data can be transmitted and processed through streaming data processing frameworks (such as Kafka and Flink) to respond to user actions instantly.

[0042] Event information refers to the impact of external events on the market or user sentiment. This includes major news, policy changes, and market emergencies. This type of information is often triggered by external factors and can significantly influence market fluctuations and user sentiment. Market or industry-related event information can be captured through news platforms or policy release platforms. This information can be obtained by regularly accessing data sources or in real time through APIs. Use keyword monitoring systems (such as news monitoring tools) to track specific events or news topics. Once a specific keyword is triggered, the system automatically captures and stores the relevant data.

[0043] Develop or integrate web crawlers to capture user comments, news articles, forum posts, and other content on social media platforms. Establish a database to store structured and unstructured text, ensuring a clean data source for subsequent analysis. Use tracking technology to monitor user behavior on the platform in real time, collecting click data, page view time, and path, generating user behavior logs and storing them in a data warehouse in real time. Establish regular or real-time data collection mechanisms to capture news, policy releases, industry announcements, and other information. Use monitoring systems to quickly respond to market or environmental emergencies and ensure timely data.

[0044] Example: In the financial sector, social media comments can be used to capture investor sentiment towards a particular stock. User behavior data, such as buying and selling activity on trading platforms, can also reflect shifts in investor sentiment. Event information, such as new policy releases and corporate financial reports, can significantly impact market sentiment. By combining these three types of data, financial platforms can more accurately predict market trends and provide investors with more targeted decision-making advice.

[0045] By collecting multiple types of data, we can comprehensively capture user and market dynamics. Text data provides the foundation for sentiment analysis, user behavior data helps us understand users' actual operational tendencies, and event information reflects the impact of the external environment on user emotions. By integrating this data, we can generate more comprehensive sentiment analysis, providing data support for personalized recommendations and decision-making.

[0046] S20, performing natural language processing and sentiment analysis on the text data to identify sentiment features of the text data;

[0047] In this embodiment, natural language processing technology is used to analyze and process natural language content extracted from text data. This mainly includes steps such as text preprocessing, word segmentation, part-of-speech tagging, and entity recognition, aiming to extract useful information from unstructured text data and perform semantic understanding.

[0048] First, it is necessary to remove noise data (such as HTML tags, special symbols, repeated irrelevant words, etc.) from the text to ensure that subsequent text analysis can be more accurate. Segmenting a continuous text stream into independent words or phrases is an important step before conducting sentiment analysis. In the Chinese environment, word segmentation technology is particularly important, and commonly used tools include Jieba and THULAC. Part-of-speech tagging is performed on the segmented words to determine whether each word is a noun, verb, adjective, etc., to facilitate the subsequent identification of sentiment words. Identify specific entities in the text (such as names of people, companies, events, places, etc.), which may be crucial for understanding the context of sentiment analysis.

[0049] Sentiment analysis is the process of analyzing words, sentences, or paragraphs in text content to determine their emotional tendencies (such as positive, negative, or neutral). This helps us understand the emotional attitude or emotional intensity expressed in the text.

[0050] Using a sentiment dictionary, we determine whether sentiment words in a text are positive, negative, or neutral, and calculate the overall sentiment based on the weights of these words. For example, the words "growth" and "profit" can be considered positive, while "loss" and "bankruptcy" are negative. Pre-trained sentiment classification models (such as SVM, Naive Bayes, and BERT) can be used for sentiment analysis. Training datasets typically contain text data with sentiment labels, and supervised learning is used to train the model to enable sentiment classification.

[0051] Based on the distribution and intensity of sentiment words in a text, sentiment analysis systems can calculate sentiment polarity (positive, negative, or neutral) and sentiment intensity (how strongly the sentiment is expressed). Sentiment polarity is typically +1 (positive), -1 (negative), or 0 (neutral), while sentiment intensity is the strength of the sentiment (e.g., a value between 0 and 1).

[0052] Sentiment features refer to emotional information extracted from text, such as sentiment polarity and sentiment intensity. These sentiment features are used to quantify the emotional expression of the text, thereby helping subsequent decision-making or recommendation systems understand the emotional state expressed in the text.

[0053] Extract sentiment polarity from the text's vocabulary and context to determine whether it is positive, negative, or neutral. Analyze the number and distribution of sentiment words to assess the intensity of sentiment expression. Calculate sentiment intensity based on the frequency and weight of sentiment words. Finally, quantify sentiment polarity and intensity into numerical values ​​to facilitate subsequent analysis and assessment of sentiment states.

[0054] Text cleaning uses regular expressions and natural language processing libraries (such as NLTK and spaCy) to remove noise. For Chinese text processing, Jieba can be used for word segmentation; for English text processing, spaCy or NLTK can be used for word segmentation and part-of-speech tagging. Pre-trained models (such as NER models) are used to identify named entities in the text and correlate them with sentiment analysis results.

[0055] Use existing sentiment lexicons (such as SentiWordNet or the Chinese Sentiment Dictionary) to analyze sentiment terms in a text. Build or call a pre-trained sentiment classification model (such as BERT or SVM), input the text into the model, and obtain sentiment classification results. By analyzing the sentiment terms in the text, calculate the sentiment polarity and intensity of the text and quantify them into numerical values.

[0056] Example: In the financial sector, sentiment analysis of news comments or user discussions on social media can be used to assess investor sentiment towards specific stocks or events. For example, when analyzing a company's financial report, natural language processing can be used to extract positive terms such as "growth" and "profit" or negative terms such as "loss" and "bankruptcy" from user comments. Combined with the intensity of sentiment, this approach can assess investors' overall sentiment towards the event. This helps financial platforms predict changes in market sentiment and provide more targeted investment advice.

[0057] By performing natural language processing and sentiment analysis on text data, we can effectively extract emotional features from user-generated content, helping us understand the emotional state of users or the market. Combining sentiment polarity and intensity, the system can quantify the emotional expression in text, thereby supporting personalized recommendations or sentiment management. Furthermore, this analysis method can help the system more accurately predict user sentiment trends, improving the flexibility and real-time nature of decision-making.

[0058] S30, analyzing the user behavior data to determine the user's behavior pattern;

[0059] In this embodiment, user behavior data is collected by recording user actions and interactions on applications, websites, or other platforms. This data includes user clicks, browsing, purchasing behavior, visit frequency, and dwell time. Analysis of this behavior data can reveal user habits, preferences, and behavioral patterns.

[0060] Through embedded tracking code and behavioral analysis tools (such as Google Analytics), various user behavior records on the platform can be collected. This includes pages visited, dwell time, content clicked, and so on. Before analysis, behavioral data needs to be cleaned, including removing invalid data (such as invalid clicks, bot traffic, etc.) and normalizing the data (such as unifying the time format and normalizing user behavior). Using time series models to analyze user behavior data, by analyzing the temporal characteristics of user behavior, we can determine how user behavior patterns change over time.

[0061] Behavioral patterns refer to the patterns of user behavior extracted by analyzing user behavior data. Behavioral pattern identification is usually based on techniques such as frequency analysis, behavior sequence mining, and cluster analysis.

[0062] By statistically analyzing the frequency of user behavior, we can identify high-frequency user operations, such as frequent clicks on a certain type of product or frequent use of a specific function. Frequency analysis can be used to discover user preferences and predict their future behavior. By analyzing the sequence of user behavior, we can discover the logical associations between certain behaviors. For example, users may frequently compare products after searching for a specific product and then make a purchase decision. Common algorithms include sequential pattern mining algorithms (such as the GSP algorithm) or association rule mining (such as the Apriori algorithm). Clustering the behavioral data of a large number of users can identify common behavioral patterns of similar user groups. Common clustering algorithms include K-means, DBSCAN, etc. By dividing users into different groups, we can identify the behavioral habits of different types of users.

[0063] User behavior is dynamic, and the system needs to accumulate and analyze long-term data to promptly adjust its judgment of user behavior patterns. Dynamic analysis can help identify changes in user behavior patterns over time or in response to specific events or environmental factors.

[0064] Use time series analysis or moving average algorithms to track changing trends in user behavior patterns. For example, by analyzing the frequency of user behavior over a period of time, you can determine whether a user's activity has increased or their interests have changed.

[0065] Build a prediction model based on the user's past behavior data to predict the user's likely future behavior. Behavior prediction can be performed using regression analysis or machine learning models (such as LSTM and RNN).

[0066] User behavior data is collected by tracking users' clicks, page views, and purchases on the platform. This data is stored in log files or databases. Data cleaning tools are then used to remove invalid clicks and abnormal behavior. Frequency statistics are then performed on the cleaned user behavior data to identify high-frequency user behaviors. Sequential pattern mining methods are then used to analyze user behavior sequences and identify regular patterns.

[0067] Perform cluster analysis on multiple user data to identify groups of users with similar behavioral characteristics. Use the K-means clustering algorithm to group users based on behavioral similarity.

[0068] Use time series models to monitor changes in user behavior over time and identify shifts in behavioral trends. Use LSTM models to predict users' likely future behavior patterns and adjust personalized recommendations or provide corresponding prompts based on these predictions.

[0069] Example: In the financial sector, user behavior data can reflect investors' operating habits and preferences. For example, if a user frequently browses certain financial products or frequently clicks on analysis reports for certain stocks, the system can analyze these behaviors and identify the user's preferred investment areas. By mining behavioral patterns, it is possible to discover patterns in user actions after stock analysis, such as frequent trading after reading multiple financial news articles. Based on these behavioral patterns, financial platforms can deliver timely and relevant investment recommendations or market analysis reports to help users make more informed investment decisions.

[0070] By analyzing user behavior data, we can gain a deeper understanding of user habits and preferences, thereby identifying behavioral patterns. This behavioral pattern analysis not only helps the system predict future user actions but also provides fundamental data support for personalized recommendations and decision-making. Based on identified behavioral patterns, the system can dynamically adjust recommendation strategies or provide more targeted services, improving user experience and engagement.

[0071] S40, analyzing the event information to determine the event change trend;

[0072] In this example, event information refers to external factors that may influence the market, user sentiment, or system decision-making. Events can include economic policies, market emergencies, corporate financial report releases, news and public opinion, and so on. By analyzing this event information, useful features can be extracted and their potential impact on the market or user behavior can be identified.

[0073] Event information is collected from news platforms, policy release platforms, social media, and public announcements through automated data scraping tools and APIs. This information can be stored in structured or unstructured formats, including text, timestamps, and classification tags. Event data is preprocessed, including text cleaning, deduplication, and noise removal. This preprocessing step ensures the accuracy of subsequent analysis. For example, it filters out events relevant to the analysis through keyword extraction, named entity recognition, and removal of irrelevant information.

[0074] Event classification involves categorizing event information based on its nature and impact. Events can be categorized into policy changes, corporate actions (such as financial report releases and mergers and acquisitions), natural disasters, and social events. Feature extraction extracts key information from event data to facilitate subsequent trend analysis.

[0075] Event information can be categorized using machine learning text classification techniques (such as Naive Bayes, SVM, and BERT). The system automatically identifies and categorizes events based on training datasets. For example, a pre-trained BERT model can be used to identify news articles as events related to economic policies, market sentiment, or corporate announcements.

[0076] Using natural language processing to extract keywords or entities (such as company names, person names, and dates) from an event helps to more accurately understand the specific impact of the event. Using named entity recognition (NER) tools, you can automatically extract highly influential entities from text.

[0077] Event information analysis requires not only classification and feature extraction but also assessment of the scope and intensity of an event's impact. The system needs to comprehensively assess the potential impact of an event on user behavior or the market based on multiple dimensions, including its nature, scale, and frequency.

[0078] Based on the scale of the event, the time window in which it occurred, and the event category, the system can calculate the impact of the event on a specific field or user. For example, for major policy changes in the financial sector, the impact strength can be measured by the level of public discussion and related stock price fluctuations.

[0079] By analyzing sentiment in news reports and social media, we can assess the negative or positive impact of an event. For example, the release of a company’s financial report can trigger fluctuations in market sentiment, and the system can use sentiment analysis to assess the potential impact of the event on the market.

[0080] Event trend analysis compares historical events with current events to identify their trajectory and future direction. It also predicts event trends by analyzing event frequency, intensity, and time intervals.

[0081] Use time series analysis models (such as ARIMA models and LSTM) to analyze the temporal changes in event information. For example, by analyzing the time nodes of multiple related events, you can identify the changing trends of events and determine the direction of market or user sentiment fluctuations.

[0082] Based on the frequency of events, the intensity and type of their impact, statistical models or machine learning prediction models are constructed to infer possible future event trends. Historical event data can be used for training to predict the impact of similar events on the future.

[0083] Regularly collect event information such as major news, policy releases, and corporate announcements through web crawlers, API interfaces, or third-party data providers. Store event data by timestamp to facilitate subsequent time series analysis.

[0084] Use a machine learning-based text classification model to categorize event information into different types (such as policy changes, market sentiment, natural disasters, etc.) and extract keywords and important entities (such as company names, dates, and person names) from events through natural language processing technology.

[0085] Calculate the scope and intensity of an event's impact and assess its potential impact on the market or user sentiment. Use sentiment analysis techniques to analyze relevant reports or comments to determine the emotional orientation of the event (positive, negative, neutral).

[0086] Apply time series analysis models to analyze historical event data, identify changes in frequency and intensity of events, and predict future trends. Build statistical models to infer the future impact of events and provide a basis for system decision-making.

[0087] Example: In the financial sector, event information analysis can help financial platforms quickly identify market emergencies and policy changes. For example, when a company releases its financial report, the system can automatically analyze market sentiment and predict the company's stock price based on the report content and market feedback. Furthermore, policy changes (such as interest rate adjustments and fiscal policy shifts) can be analyzed through event trend analysis to predict future market sentiment fluctuations, helping investors make informed decisions in advance. This real-time event analysis can significantly enhance the decision-making flexibility and user service levels of financial platforms.

[0088] By analyzing event information and predicting trends, the system can promptly capture changing trends in market or user sentiment. Event classification and impact analysis can identify which events have a significant impact on user sentiment or market fluctuations, enabling the system to respond in real time. Event trend analysis can predict potential emotional fluctuations caused by future events, enabling the system to provide early warnings and adjust response strategies.

[0089] S50, determining an emotional state indicator based on the emotional characteristics, behavioral patterns, and event change trends;

[0090] In this embodiment, the sentiment feature is the sentiment information extracted from the text data, including the sentiment polarity (positive, negative or neutral) and the sentiment intensity. Quantifying the sentiment feature means converting this sentiment information into numerical values ​​for subsequent calculations.

[0091] Sentiment polarity can be expressed as a numerical value, such as +1 for positive sentiment, -1 for negative sentiment, and 0 for neutral sentiment. Based on the strength of the sentiment word, a floating value between 0 and 1 is set to indicate the strength of the sentiment expression. For example, the sentiment word "very good" might correspond to a high sentiment intensity (such as 0.8), while "good" might correspond to a low sentiment intensity (such as 0.4).

[0092] Behavioral patterns are patterns in user behavior displayed through a series of operations. These patterns can be quantified through behavior frequency and behavioral change trends, reflecting user operating habits and behavioral tendencies. By counting the frequency of user behaviors within a specific time period, such as the number of clicks, visits, and purchases per day, behavioral frequencies can be expressed as integers or decimals. By comparing behavioral frequency changes over different time periods, the degree of behavioral fluctuation can be calculated. For example, the standard deviation or rate of change (such as the ratio of change within a day) can be used to represent the magnitude of behavioral changes.

[0093] Event trends refer to the changes in the impact of external events on user sentiment or the market over time. These events can be news reports, policy changes, market emergencies, etc. Quantifying event trends means using numerical values ​​to represent the intensity and frequency of the impact of events. By analyzing the emotional tendency (positive or negative) and the scope of influence (such as market fluctuations, social media discussion, etc.) of the event, the impact of the event is converted into a numerical value. For example, a value from 0 to 1 can be used to represent the intensity of the event's impact, with 0 indicating no impact and 1 indicating the greatest impact. Through time series analysis, the changes in the frequency and intensity of event occurrences are calculated. A common practice is to calculate the number of occurrences of related events per unit time and the magnitude of the change in their influence.

[0094] The emotional state index is generated by comprehensively considering the quantitative values ​​of emotional characteristics, behavioral patterns, and event trends, reflecting the current overall emotional state of users or the market. Through weighted calculation, the influence of various factors is integrated to produce a numerical emotional state index. Each factor (emotional characteristics, behavioral patterns, and event trends) is assigned a weight, and the weights are adjusted according to the needs of different scenarios. For example, the impact of user behavior on emotional state might be set at 50%, emotional characteristics at 30%, and event changes at 20%. Using the weighted average method, the quantitative values ​​of emotional characteristics, behavioral patterns, and event trends are added together to calculate the final emotional state index.

[0095] A sentiment analysis model is used to process text data and extract sentiment polarity and intensity. Positive, negative, and neutral sentiment polarities are converted to numerical values, and sentiment intensity is determined by analyzing word weights and sentence structure. Finally, the sentiment features are converted to a numerical value.

[0096] Conduct frequency statistics on user behavior data to analyze changing trends in user operations. Quantify user behavior fluctuations using metrics such as standard deviation and frequency ratio.

[0097] Analyze event information using time series models to calculate the frequency and impact of events. Use sentiment analysis tools to further assess the emotional tendencies of events and quantify their impact on emotions.

[0098] By assigning different weights and combining the quantitative values ​​of emotional characteristics, behavioral patterns, and event trends, we perform a weighted average calculation to generate the final emotional state indicator. This indicator reflects the overall emotional state of the current user or market.

[0099] Example: In the financial sector, a comprehensive analysis of investor sentiment, trading behavior patterns, and external market events (such as policy changes or corporate earnings releases) can be used to calculate an indicator reflecting the overall market sentiment. When investor sentiment fluctuates significantly, market events occur frequently, and behavioral patterns indicate abnormal trading frequencies, the system can use this sentiment indicator to assess potential market risks and issue early warnings to investors, helping them make more rational and stable investment decisions.

[0100] By combining sentiment characteristics, user behavior, and event trends, the system can accurately assess the emotional state of users or the market. Emotional state indicators provide a basis for system decision-making, reflecting the magnitude and stability of user emotional fluctuations, helping the system make more personalized recommendations and decisions. Furthermore, the inclusion of event information makes sentiment analysis more comprehensive, taking into account the impact of external environments on user emotions, thereby improving the system's predictive accuracy and adaptability.

[0101] S60, monitors information on online platforms and obtains user sentiment feedback;

[0102] In this embodiment, online platforms refer to social media platforms, forums, news comment areas, e-commerce platforms, etc. where users are active, from which users' emotional reactions to specific events, products, or markets can be obtained. The purpose of monitoring information on online platforms is to collect user-generated content (UGC) in real time, such as comments, likes, shares, articles, discussions, etc. These data can reflect users' emotional attitudes towards specific events or products. By using web crawler tools, user comments, posts, articles, and other information on a specified platform are automatically captured. For example, user comments on social media, discussion posts in forums, etc.

[0103] Some platforms provide public APIs that can be used to obtain real-time user comments, feedback, likes, and other information. For example, the APIs of platforms such as Twitter and Reddit can be used to capture comments and discussion data.

[0104] Establish a regular data crawling mechanism to monitor and store user content from different platforms to ensure the timeliness and continuity of information collection.

[0105] User emotional feedback refers to the emotional reactions expressed by users through online platforms. These reactions can be sentimental in the text (e.g., positive, negative, neutral) or emotional attitudes reflected in non-textual behaviors (e.g., likes, shares, comments, etc.). User emotional feedback directly reflects the user's emotional state and is used for subsequent sentiment analysis and emotional state assessment.

[0106] Perform natural language processing (NLP) and sentiment analysis on captured text data (such as comments and articles) to identify the polarity and intensity of user emotions. For example, sentiment classification models (such as BERT and SVM) can be used to analyze the emotional tendencies of user comments and extract the user's emotional reactions.

[0107] By analyzing user interactions on the platform, such as likes, shares, and reposts, we can infer user emotional feedback. The number of likes and reposts a user gives to a piece of content often reflects their emotional attitude towards the content, especially positive or supportive emotions.

[0108] The sentiment polarity (positive, negative, neutral) and sentiment intensity in the analysis results are quantified and extracted into structured sentiment data for subsequent use.

[0109] Users' emotional feedback on multiple online platforms may differ, so it is necessary to integrate the emotional data on different platforms to generate a unified emotional view.

[0110] Data from different platforms should be cleaned, including deduplication, noise reduction, and format standardization. Data structure and quality may vary across platforms, so standardization is necessary to ensure data consistency. User sentiment feedback from multiple platforms should be aggregated and stored in a unified database. Big data storage frameworks such as Hadoop and Elasticsearch can be used to consolidate and manage this data.

[0111] Use web crawlers or API interfaces to capture user-generated content from online platforms such as social media, news websites, and forums in real time. Web crawlers are run regularly to ensure that the latest user sentiment feedback data is continuously collected. API interfaces are used for efficient extraction of real-time data. The acquired text data is input into the natural language processing system, and the sentiment analysis model is used to identify the sentiment polarity and sentiment intensity in the text. For non-text data, the user's sentiment attitude towards the content is inferred by analyzing behaviors such as the number of likes, reposts, and comments. After processing the sentiment data through the sentiment classification model, it is quantified into sentiment polarity (positive, negative, neutral) and intensity values ​​and stored as structured data. The sentiment data from different platforms are cleaned, denoised, and deduplicated to ensure consistency in format. The integrated sentiment data is stored in a unified database to provide support for subsequent emotional state analysis.

[0112] Example: In the financial sector, by monitoring investor comments on social media, discussions in financial forums, and user messages on news platforms, we can capture investor sentiment towards market dynamics, individual stock performance, or policy changes. For example, after a company releases its quarterly financial report, the system can identify investor sentiment by crawling social media and news comments. If investor comments are predominantly positive and receive a high number of likes and reposts, the system can infer that market sentiment is optimistic and provide investors with investment advice based on this sentiment.

[0113] By monitoring information from online platforms, we can capture comprehensive and real-time user emotional feedback. By combining sentiment analysis with both textual and non-textual behavioral data, the system can accurately grasp users' emotional reactions to events, products, or markets. By integrating and standardizing data from multiple platforms, the system can generate a more comprehensive view of emotional feedback, providing data support for subsequent emotional state assessment and personalized recommendations. This approach improves the coverage and accuracy of emotional feedback, empowering the system with stronger real-time analysis capabilities.

[0114] S70, comprehensively analyzing the emotional state indicator and the user's emotional feedback to generate an emotional analysis result;

[0115] In this embodiment, the emotional state index is a numerical value calculated based on emotional characteristics, user behavior patterns, and event change trends. It is used to quantify the overall emotional state of the current market or users. It reflects the combined impact of emotions, behaviors, and external events on user emotions.

[0116] The emotional state indicator has been generated through weighted calculations in the previous steps. Now, the system uses it as input data for comparison and comprehensive analysis with the user's real-time emotional feedback. The emotional state indicator can be a single numerical value or a multidimensional indicator vector, depending on the analysis requirements.

[0117] User sentiment feedback is derived through sentiment analysis of user-generated content (UGC) on online platforms. It includes users' emotional tendencies (positive, negative, neutral) and emotional intensity. It represents users' immediate reactions to current events, products, or markets. User sentiment feedback is generated by a sentiment analysis system and includes quantified results of sentiment polarity and intensity. Feedback data can be obtained from non-textual interactions such as social media comments, news comments, and likes, and converted into a standardized sentiment data format for analysis in conjunction with emotional state indicators.

[0118] Comprehensive analysis refers to the fusion of emotional state indicators and user emotional feedback, evaluating the correlation and difference between the two, and generating the final emotional analysis results. This process needs to consider the time dimension, emotional tendency and intensity of the two sets of data, and analyze whether the user emotional feedback is consistent with the current emotional state indicators. The correlation between emotional state indicators and user emotional feedback is evaluated through correlation analysis techniques (such as the Pearson correlation coefficient). If the two are highly correlated, it can be inferred that user emotions are consistent with the overall market sentiment or event sentiment changes. Calculate the deviation between the emotional state indicator and the user emotional feedback. If the deviation is large, it means that the user emotion is inconsistent with the overall sentiment, and further measures may need to be taken to adjust or understand the user's special emotional state. The deviation calculation formula can be a simple difference or a multidimensional difference analysis based on Euclidean distance.

[0119] During the comprehensive analysis, weights can be assigned based on the importance of the emotional state indicator and user emotional feedback. For example, if user emotional feedback is more representative than the emotional state indicator, a greater weight can be assigned to adjust the final results. The comprehensive analysis uses a weighted average method to calculate the combined results of the two.

[0120] Sentiment analysis results are the final output of a comprehensive analysis, used to describe the current user's emotional state or market sentiment fluctuations. They combine the correlation and variance of emotional state indicators with user emotional feedback to produce a quantitative assessment of emotional state. Based on the correlation and deviation analysis results, the system generates a sentiment analysis result that represents the current emotional state. Sentiment analysis results are typically a quantitative value or rating, indicating the intensity of current emotional fluctuations or the stability of the user's emotions. Based on the sentiment analysis results, the system can further determine the degree of user emotional fluctuations. For example, a large deviation may indicate unstable user emotions, while a high correlation indicates relatively stable user emotions. Sentiment analysis results can be used for subsequent personalized recommendations or market warnings.

[0121] The system obtains the emotional state indicator from the previous steps and obtains user emotional feedback through an API interface or a real-time data capture system. These two data are used as input for comprehensive analysis and stored in a database or memory, ready for subsequent analysis. First, the system performs a correlation analysis between the emotional state indicator and the user's emotional feedback, calculating the correlation coefficient between the two and evaluating their degree of association. If there is a high correlation between the two, the system will determine that the user's emotions are consistent with the overall trend. Next, the system performs a variance analysis, calculating the deviation between the emotional state indicator and the user's emotional feedback to determine whether the user's emotions are consistent with expectations. If the deviation is large, the system will further evaluate the user's emotional fluctuations. Finally, the system performs a weighted average calculation of the correlation and deviation results to generate a comprehensive emotional evaluation value, which represents the current user's emotional state or the degree of fluctuation in market sentiment.

[0122] Based on the comprehensive analysis, the system outputs sentiment analysis results, which are stored or passed to downstream modules for personalized recommendations or market analysis and early warning. Sentiment analysis results can be a specific value or level that describes the current emotional state.

[0123] Example: In the financial sector, the system can assess investor sentiment fluctuations in real time through a comprehensive analysis of sentiment indicators and user feedback. For example, sentiment indicators may indicate an overall optimistic market sentiment, but monitoring social media comments may reveal a rise in negative user feedback. Through this comprehensive analysis, the system can identify potential market sentiment fluctuations, provide investors with sentiment warnings, and adjust recommended investment strategies.

[0124] By comprehensively analyzing emotional state indicators and user feedback, the system can accurately assess users' emotional states in real time and identify emotional fluctuations. This not only improves the accuracy of sentiment analysis but also enhances the system's real-time adaptability. Sentiment analysis results provide strong data support for personalized recommendations, market warnings, user emotion management, and other functions, ensuring that the system can promptly adjust to user mood swings.

[0125] S80: Generate target output content for responding to the user's emotional state based on the emotion analysis result.

[0126] In this embodiment, sentiment analysis results are quantitatively derived by combining sentiment indicators and user sentiment feedback, representing the current sentiment state of users or the market. The sentiment analysis results are used to determine how the system generates personalized output content, such as recommendations, prompts, and warnings.

[0127] Sentiment analysis results can be expressed as a numerical value or a rating, representing the degree of emotional fluctuation, emotional intensity, or emotional stability. This result serves as the basic input for subsequent content generation and enters the system's output module, which is used to formulate appropriate content generation strategies.

[0128] Based on the sentiment analysis results, the system needs to generate targeted output content. This output can be personalized recommendations, prompts, warnings, or decision-making advice, depending on the user's emotional state and the analysis results. If the sentiment analysis results indicate that the user's mood is relatively stable and positive, the system can generate personalized recommendations (such as related products or news). In this case, the content generation module uses recommendation algorithms (such as collaborative filtering and content-based recommendations) to generate personalized output tailored to the user's emotional state. If the sentiment analysis results indicate significant emotional fluctuations or negative market sentiment, the system can generate prompts or warnings. Using a predefined rule engine, the system can generate appropriate warnings for users to help them avoid irrational behavior caused by emotional fluctuations. For example, in the financial sector, if sentiment analysis results indicate significant investor sentiment fluctuations, the system can generate a "Please invest with caution" reminder. Based on the sentiment analysis results, the system can also generate adjustment suggestions to guide users on how to balance their emotions or take action. For example, the system can recommend relevant emotional management tips or adjust the user's recommended content preferences to suit their current emotional state.

[0129] The generated output content needs to be presented in a form that users can understand and accept. Depending on the results of the sentiment analysis, the output content can take various forms, such as text, charts, warning boxes, and notifications. Based on different scenarios and emotional states, the system formats the generated content to ensure that it is concise and clear. For example, when the user's emotions are relatively stable, the system can generate detailed recommendations and related prompts; when emotions fluctuate significantly, warning content can be presented in more prominent forms such as pop-ups and notifications. The system can output content through various channels, such as mobile phone push, email, and platform notifications, to ensure that users can promptly access output content that matches their emotional state.

[0130] The generated target output content must be fed back into the system so that it can continuously optimize its content generation strategy and effectiveness. User feedback on the output content will be recorded and used to further adjust the content generation logic. The system can obtain user feedback data by monitoring user interactions with recommended content and prompt information (such as clicks, ignores, and positive or negative comments). Based on this feedback, the system can adjust its future output content generation strategy.

[0131] The system obtains the user's current emotional state data from the sentiment analysis results of the previous step. This result may be a quantified sentiment value, sentiment polarity, and sentiment intensity, etc., which is stored in memory or a database and serves as the core input data for the content generation module. If the sentiment analysis results show that the user's emotions are relatively positive and stable, the system generates personalized content (such as news, product recommendations, etc.) through the recommendation engine. Based on the user's emotional state and past behavior records, the system calls the recommendation algorithm to generate recommendation information that matches the user's emotions. When the sentiment analysis results show that the user's emotions fluctuate greatly or tend to be negative, the system can generate warnings or prompts based on preset rules to remind the user to be cautious when making decisions. Emergency notifications or intervention suggestions are generated by the system through predefined rules.

[0132] The presentation of output content is adjusted based on the user's emotional state. Recommendations are provided to users through the platform's notification or push system. Warnings and prompts are presented through pop-ups and alert boxes to ensure timely user awareness. The system monitors user interaction with output content and collects feedback. User behavioral data (such as click-through rate, dwell time, and emotional response) is used to further optimize content generation and emotional state assessment strategies.

[0133] Example: In the financial sector, sentiment analysis can be used to generate personalized investment recommendations for investors. When the system detects stable investor sentiment and optimistic market sentiment, it can recommend suitable investment products or strategies to help investors seize market opportunities. However, when investor sentiment fluctuates significantly and market sentiment becomes negative, the system can generate risk warnings or suggest suspending trading, helping investors avoid the potential risks associated with emotional decision-making. This sentiment analysis-based output generation mechanism can significantly improve the rationality of investor decision-making and reduce losses caused by market fluctuations.

[0134] By generating targeted output content based on sentiment analysis results, the system can better adapt to user mood swings, providing more personalized and dynamically adjusted services. For users with relatively stable emotions, the system can provide appropriate recommendations; for users with more volatile emotions, the system can issue timely warnings or prompts, helping them make more rational and robust decisions. This personalized content generation mechanism effectively improves the user experience and enhances the system's decision-making support capabilities.

[0135] The present invention relates to the fields of artificial intelligence technology and financial technology, and discloses an information processing method based on sentiment analysis, which collects data and information, including text data, user behavior data, and event information; performs natural language processing and sentiment analysis on the text data to identify the sentiment characteristics of the text data; analyzes user behavior data to determine behavior patterns; analyzes event information to determine event change trends; generates sentiment state indicators based on sentiment characteristics, behavior patterns, and event change trends; monitors online platforms to obtain user sentiment feedback; comprehensively analyzes sentiment state indicators and user sentiment feedback to generate sentiment analysis results; and generates target output content based on the sentiment analysis results. The present invention can capture the user's emotional state in real time, dynamically adjust decision recommendations, and effectively respond to market or environmental changes by combining sentiment analysis with behavior data and event change trends. In addition, the analysis of user sentiment feedback further improves the accuracy of personalized recommendations. This not only enhances the system's ability to respond to external changes, but also better meets the user's personalized needs and optimizes the overall decision-making process.

[0136] In one embodiment, the above S50 includes:

[0137] S501, generating a quantitative value of the emotional feature based on the emotional polarity and emotional intensity in the emotional feature;

[0138] S502, analyzing the behavior frequency and behavior change degree in the behavior pattern to generate a quantitative value of the behavior pattern;

[0139] S503, generating a quantitative value of the event change trend according to the event's impact intensity and the event's influence coefficient on the user's emotion;

[0140] S504: Perform weighted analysis on the quantitative values ​​of the emotional characteristics, behavioral patterns, and event change trends to generate the emotional state index.

[0141] In this embodiment, sentiment features are extracted from text data through natural language processing and sentiment analysis, including sentiment polarity (such as positive, negative, or neutral) and sentiment intensity (such as the intensity of the sentiment expression). These sentiment features are quantified into numerical values ​​to facilitate subsequent weighted calculations.

[0142] Based on the analysis results, sentiment polarity can be quantified as positive (+1), negative (-1), and neutral (0), indicating the user or market's emotional inclination towards a specific event. Sentiment intensity is based on the frequency and intensity of sentiment words and is set within a range of 0 to 1. For example, a stronger sentiment expression (such as "very happy") may correspond to a higher sentiment intensity value (such as 0.9), while a weaker sentiment expression (such as "a little good") may correspond to a lower sentiment intensity value (such as 0.4). Sentiment polarity and sentiment intensity are combined to generate a quantitative value of the sentiment feature, ultimately resulting in a comprehensive value representing the sentiment feature.

[0143] The quantification of behavioral patterns is achieved by analyzing dimensions such as the user's operation frequency and behavioral change trends. This can indicate the stability or volatility of user behavior over time or environmental changes. By counting the user's behavior frequency, such as the number of clicks per hour, the frequency of daily logins, etc., the behavior frequency can be used as a measure of behavioral activity and is usually quantified as an integer or floating point number. Behavioral changes are determined by analyzing the fluctuation trend of behavior, such as through standard deviation, variance, or behavioral change ratio (such as comparing the behavior in the past week with the previous week). A larger change amplitude indicates a larger fluctuation in the user's behavior pattern, and a smaller change amplitude indicates stable behavior. Combining the behavior frequency with the degree of behavior change generates a comprehensive behavior pattern quantification value that reflects the activity and stability of user behavior.

[0144] Quantifying event trends is based on the impact of external events on the market or users. This requires considering the intensity of the event (such as its importance and scale) and its impact on user sentiment. By analyzing factors such as the event's scale, duration, and influence on public opinion, the intensity of the event's impact can be quantified using a value between 0 and 1. For example, a major economic policy change might be assigned a higher impact intensity (such as 0.8 or 0.9), while a minor market fluctuation might be assigned a lower impact intensity (such as 0.2). By analyzing the impact of an event on user emotional feedback (such as user attention or emotional response to the event), an impact coefficient is assigned to the event. The emotional impact coefficient can be assigned based on the intensity of user emotional feedback, such as a range of 0 to 1, reflecting the influence of the event on user emotions. Combining the event's impact intensity and emotional impact coefficient generates a quantitative value for the event's trend. This value reflects the importance of the event in the market or user sentiment.

[0145] By weighting the quantitative values ​​of emotional characteristics, behavioral patterns, and event trends, an overall emotional state indicator is generated. Weighted analysis aims to balance the impact of different factors on emotional state. Different weights are assigned to emotional characteristics, behavioral patterns, and event trends. Specific weights can be adjusted dynamically based on the application scenario, for example, emotional characteristics could account for 50%, behavioral patterns for 30%, and event trends for 20%. This weighted analysis ultimately generates a numerical indicator reflecting the current emotional state of users or the market. This indicator can be used for subsequent personalized recommendations, market forecasts, or emotional early warning.

[0146] This embodiment uses quantified and weighted analysis of emotional characteristics, behavioral patterns, and event trends to generate an indicator that accurately reflects the current emotional state. This emotional state indicator not only reflects user or market emotional fluctuations in real time but also provides important data support for subsequent personalized recommendations and market warnings. The generation of this indicator helps the system better understand user emotional fluctuations, thereby optimizing decision-making and services.

[0147] In one embodiment, the above S60 includes:

[0148] S601, collecting user-generated text content from an online platform;

[0149] S602, extracting sentiment polarity and sentiment intensity from the text content based on a sentiment analysis module, and generating sentiment data according to the sentiment polarity and sentiment intensity;

[0150] S603: combining the emotion data with the user's behavior pattern and historical emotion data, and comprehensively analyzing and generating the user emotion feedback.

[0151] In this embodiment, user-generated content (UGC) in the online platform usually appears in the form of comments, posts, articles, etc. These text contents can directly reflect the user's emotional response to an event, product or market. Web crawler technology is used to automatically collect text data such as user comments, posts or articles from online platforms (such as social media, forums, news websites, etc.). Crawlers can crawl content regularly to ensure the real-time and integrity of the data. The API interface provided by the platform (such as TwitterAPI, RedditAPI, etc.) is used to obtain user-generated content in real time, and filter relevant text data according to preset topics or keywords.

[0152] The role of the sentiment analysis module is to analyze user-generated text content through natural language processing technology to identify the emotional tendency (such as positive, negative, or neutral) and the intensity of emotional expression in the text. Use a sentiment analysis model (such as dictionary-based sentiment analysis or a machine learning model) to process the text and identify the emotional polarity in the text. Polarity is usually expressed as positive, negative, or neutral. By analyzing the emotional vocabulary, context, and emotional expression in the text, the intensity of the emotion is quantified (such as a numerical range of 0 to 1) to indicate the strength of the emotional expression. For example, strong negative emotions may have higher intensity values. The extracted emotional polarity and emotional intensity are combined to generate structured emotional data, which is usually expressed as a combination of emotional polarity and intensity. For example, a text may be identified as having negative emotion (polarity of -1) and strong intensity (intensity of 0.8).

[0153] Emotional data needs to be combined with the user's behavior patterns (such as the user's operation frequency, behavior trends, etc.) and historical emotion data to form a more complete judgment of the user's emotions. The user's historical emotion data includes their past emotional reactions and behavior patterns, which can reflect the user's long-term emotional tendencies. By monitoring and analyzing the user's behavior patterns on the platform (such as browsing history, click frequency, comment interaction, etc.), combined with the current emotion data, the correlation between the user's emotions and behaviors can be judged. For example, if the user's behavior pattern shows positive behavior, but the emotion data is negative, it may be necessary to further analyze the cause of the emotional fluctuations. Compare the currently extracted emotion data with the user's historical emotion data to understand the user's emotional change trend. For example, if the user's historical emotion data is mostly positive, but the current emotion data shows negative emotions, the system can infer that the emotion may have fluctuated significantly.

[0154] User emotional feedback is a comprehensive emotional response derived from the integration of current emotional data, user behavior patterns, and historical emotional data. This feedback can reflect changes in the user's emotional state, mood fluctuations, or stability, helping the system make subsequent decisions. By combining sentiment analysis results, user behavior pattern data, and historical emotional data, user emotional feedback is generated using a weighted algorithm or machine learning model (such as a decision tree or neural network). Emotional feedback can indicate whether the user's current emotional state is positive, negative, or unstable. Based on the results of the comprehensive analysis, structured emotional feedback data is generated and stored in a database or used for subsequent personalized recommendations or emotional state assessment.

[0155] This embodiment collects user-generated text content from online platforms and performs sentiment analysis, enabling the system to accurately extract the polarity and intensity of user emotions. By combining current sentiment data with user behavior patterns and historical sentiment data, the system can generate more complete and accurate user sentiment feedback. This feedback not only helps the system understand the user's emotional state in real time but also provides powerful data support for personalized recommendations, sentiment management, and market analysis.

[0156] In one embodiment, the above S70 includes:

[0157] S701, analyzing the deviation between the emotional state indicator and the user's emotional feedback;

[0158] S702: Generate a sentiment analysis result based on the deviation, where the sentiment analysis result is used to reflect the degree of emotional fluctuation or emotional stability of the user.

[0159] In this embodiment, the emotional state index is quantified based on emotional characteristics, behavioral patterns, and event change trends, and represents the system's prediction or inference of the user's overall emotional state. User emotional feedback is generated through real-time emotional analysis and represents the user's current actual emotion. The deviation between the two reflects the difference between the emotional state predicted by the system and the user's current actual emotional state. The emotional state index obtained through weighted analysis is compared with the user's current emotional feedback. The deviation can be expressed in the form of a numerical difference. Common calculation methods include simple difference calculation or Euclidean distance calculation to quantify the degree of difference between the two.

[0160] The deviation can be calculated as follows: Deviation = |emotional state index - user emotional feedback|.

[0161] If the deviation value is large, it indicates that there is a significant difference between the user's current mood and the emotional state predicted by the system. If the deviation value is small, it indicates that the user's mood is relatively stable or consistent with the system prediction.

[0162] Sentiment analysis results are derived by analyzing the deviation between the emotional state indicator and the user's emotional feedback. They primarily reflect the user's current emotional fluctuation or emotional stability. A larger deviation indicates more pronounced emotional fluctuations, while a smaller deviation indicates more stable emotions. If the deviation is large, the system may determine that the user's emotions are fluctuating significantly. In this case, the sentiment analysis results will reflect the user's emotional fluctuations, potentially prompting the system to pay special attention to the user's emotional state or adjust subsequent personalized recommendation strategies. If the deviation is small, the user's emotions are relatively stable, and the sentiment analysis results will reflect that the user's emotions are becoming more consistent and stable. In this case, the system can continue to use the current recommendation or behavior pattern.

[0163] Ultimately, the system generates a sentiment analysis result based on the deviation between the emotional state indicator and the user's emotional feedback. This result can be expressed as a numerical value or a level. For example, the emotional fluctuation level can be divided into three levels: "high", "medium", and "low", or expressed as a specific emotional fluctuation value.

[0164] Sentiment analysis results can inform the system's subsequent personalized recommendations, prompt generation, and market analysis. The system can adjust its service strategies based on the analysis results, providing users with content or prompts more appropriate to their current emotional state. If sentiment analysis results indicate significant emotional fluctuations, the system can generate risk warnings or reduce the frequency of recommended content. If emotions are relatively stable, the system can continue to provide recommendations that match the user's emotional state. For users experiencing significant emotional fluctuations, the system can provide emotional management advice or guidance to help them alleviate or regulate their emotions.

[0165] This embodiment analyzes the deviation between the emotional state indicator and the user's emotional feedback, allowing the system to promptly determine the user's emotional fluctuations or emotional stability. This process helps the system more accurately understand the user's emotional state, thereby providing more targeted personalized services and content recommendations. Based on the results of emotional analysis, the system can adjust content generation strategies to avoid user experience degradation caused by emotional fluctuations and improve the system's intelligent response capabilities.

[0166] In one embodiment, before the above S20, the method further includes:

[0167] S201, collecting historical data and corresponding sentiment analysis results;

[0168] S202: Based on the historical data and the corresponding sentiment analysis results, a sentiment analysis model is trained and generated, where the sentiment analysis model is used to perform natural language processing and sentiment analysis on the text data.

[0169] In this embodiment, historical data includes previously collected user text data and its corresponding sentiment analysis results. This historical data can include known sentiment tendencies (positive, negative, neutral) and sentiment intensity. By collecting sufficient historical data, the system can use this data to train a sentiment analysis model.

[0170] Historical data can come from various online platforms, social media, news articles, user reviews, and other channels. Sentiment analysis results are derived from previous sentiment analysis operations or manually annotated sentiment tags. Collected historical data requires preprocessing, such as removing irrelevant content, cleaning invalid data, and standardizing data formats. The cleaned data should include each text segment and its corresponding sentiment polarity (e.g., positive, negative, neutral) and sentiment intensity (e.g., a quantitative value from 0 to 1).

[0171] Training a sentiment analysis model involves using machine learning algorithms to train a model capable of automatically identifying the emotional characteristics of text based on existing historical data and its corresponding sentiment analysis results. This model will be used for subsequent text sentiment analysis operations. Common sentiment analysis models can be based on traditional machine learning algorithms (such as support vector machines (SVMs) and Naive Bayes) or deep learning models (such as LSTMs and BERT). Choosing an appropriate model depends on the size and complexity of the dataset. The model is trained using supervised learning, using historical data as a training set. The model receives input text data and learns the mapping between text and sentiment based on the labeled sentiment polarity and intensity. After model training is complete, validation and test sets are used to evaluate the model's accuracy and generalization ability. If the model's performance is suboptimal, optimization may be necessary through hyperparameter tuning, data augmentation, or changes to the model architecture.

[0172] The trained sentiment analysis model is then used in actual sentiment recognition tasks. It performs natural language processing (NLP) on input text data and outputs sentiment characteristics of the text (such as sentiment polarity and intensity). Before inputting into the sentiment analysis model, the system first preprocesses the text data, such as token segmentation, stop word removal, and part-of-speech tagging. These operations help the model better understand the text content. The model analyzes the text's vocabulary, syntactic structure, and contextual information to extract the text's sentiment polarity and intensity. Sentiment polarity can be positive, negative, or neutral, while sentiment intensity quantifies the intensity of the emotional expression.

[0173] By collecting historical data and training a sentiment analysis model, this embodiment enables the system to more accurately analyze the sentiment of future text data. Automated sentiment recognition based on the sentiment analysis model significantly enhances the system's sentiment analysis capabilities, enabling it to quickly and accurately determine the user's emotional tendencies and intensity, and to make more precise emotional state assessments or personalized recommendations based on this information.

[0174] In one embodiment, after the above S50, the method further includes:

[0175] S505, setting a warning threshold for the emotional state indicator;

[0176] S506, when the emotional state indicator reaches the warning threshold, generating warning information;

[0177] S507: Send the warning information to the business processing end.

[0178] In this embodiment, the emotional state indicator is derived through a comprehensive analysis of emotional characteristics, behavioral patterns, and event trends. The purpose of setting an early warning threshold is to identify abnormal fluctuations in user or market sentiment. When the emotional state indicator reaches or exceeds this threshold, the system triggers an early warning mechanism.

[0179] Warning thresholds can be set based on historical data, industry standards, or system configuration. For example, the system can establish a benchmark based on past mood swings, triggering a warning when the mood swing exceeds a certain value. Specifically, upper and lower thresholds can be set, such as triggering a warning when the mood state indicator exceeds 0.8 or falls below 0.2. Warning thresholds can be static (fixed) or dynamically adjusted. Dynamic thresholds can automatically adjust based on real-time data analysis or machine learning model results to meet the needs of different scenarios.

[0180] When the emotional state indicator exceeds the preset threshold, the system generates an early warning message. This warning message typically includes details of the emotional anomaly, indicating excessive user emotional fluctuations, abnormal market sentiment, and so on. The system monitors the emotional state indicator in real time and immediately triggers the early warning mechanism when it exceeds the preset threshold. The early warning message generation module generates targeted early warning content based on the specific range exceeded and the type of emotional fluctuation. The early warning message includes the current value of the emotional state indicator, the threshold, and the system's interpretation of the emotional fluctuation. For example, "The current emotional state indicator is 0.9, exceeding the early warning threshold of 0.8, and the user's emotional fluctuations are significant." This information can be output as text, a chart, or an alert.

[0181] Early warning information needs to be delivered to the relevant business processing end in a timely manner so that the system or staff can take further actions, such as adjusting the recommendation strategy, sending warning messages or triggering automated processing processes.

[0182] The system sends the warning information to the relevant business processing end, which may be a content recommendation module, market analysis module or risk management system. The warning information can be transmitted to the corresponding business processing system through technical means such as API interface and message queue.

[0183] The system can send warning information to the business processing end or administrator via email, platform notification, SMS or other instant messaging tools. For automated business processing, warning information can be sent to the system's decision module through a message queue or API interface to trigger subsequent actions.

[0184] By setting a warning threshold and generating a warning message when the emotional state indicator reaches that threshold, this embodiment enables the system to monitor user or market emotional fluctuations in real time and respond promptly. Sending warning information to the business processing end enables the system or staff to quickly take countermeasures, such as sending prompts or adjusting recommendation strategies, to prevent the negative impact of excessive emotional fluctuations. This warning mechanism can significantly improve the system's emotional monitoring and response capabilities, ensuring the effectiveness of emotional management.

[0185] In one embodiment, the above S80 includes:

[0186] S801, obtaining the target user's risk preference information and expected results;

[0187] S802, generating personalized output content for the target user based on the risk preference information, expected results, and sentiment analysis results;

[0188] S803: Send the personalized output content to the target user.

[0189] In this embodiment, the system needs to obtain the user's risk preference and desired outcome from the information provided by the user. Risk preference refers to the user's tolerance for risk when making decisions, and is generally categorized as conservative, neutral, and aggressive. Desired outcome refers to the goal or result that the user hopes to achieve in a specific situation.

[0190] The system can obtain information about a user's risk preferences through historical behavior (such as trading history and past choices), completed risk assessment forms, or directly from the user's account settings. For example, in the financial sector, a user's risk preference can reflect their investment strategy preferences. Expected outcomes can be derived through questionnaires, user-defined settings, or system analysis. For example, a user may seek to maximize short-term profits or achieve stable long-term investment returns. This objective information can be stored in the user's account configuration.

[0191] The system generates personalized output based on the user's risk preferences, expected outcomes, and current sentiment analysis results. This output may include decision suggestions, product recommendations, risk warnings, etc., all tailored to the user's current emotional state and personalized needs.

[0192] The system generates user-specific output based on the user's risk appetite and desired outcomes, combined with current sentiment analysis results (such as the degree of emotional volatility). For example, if the user's emotional state is unstable, the system may recommend a more conservative option; whereas, if the user's emotions are stable and their risk appetite is high, the system may recommend an aggressive option.

[0193] The system uses recommendation algorithms or rule engines to generate personalized output content. Recommendation engines can generate personalized content based on user profiles, sentiment analysis results, and risk preferences, using collaborative filtering, content-based recommendations, and other technologies.

[0194] The output content can be text (such as prompt information, analysis report), chart (such as investment strategy chart, risk assessment chart) or other forms of suggestions. The specific type depends on the system design and user needs.

[0195] The system needs to send the generated personalized output content to users to ensure that users can receive relevant information in a timely manner. The delivery of this content can be achieved through different channels, such as platform messages, emails, mobile application push, etc.

[0196] Through the system's push notification function, personalized output content is sent to the user's account message center or mobile device in the form of notifications. The system can push this content to users on a regular basis or in real time.

[0197] The system can send personalized output content to users through various channels such as email, text messages, instant messaging applications, etc., to ensure that users can obtain this information in a timely manner.

[0198] Based on the user's device and preferences, the system can choose the appropriate content display method, such as a report with pictures and text, a concise prompt message or an interactive chart display.

[0199] This embodiment combines a user's risk preferences, desired outcomes, and sentiment analysis results to generate highly personalized output, helping users make rational decisions based on their current emotional state and preferences. This approach enhances the user experience, ensuring that the content provided by the system closely matches user needs, thereby improving the accuracy of the system's recommendations and user satisfaction.

[0200] In one embodiment, an information processing device based on sentiment analysis is provided, and the information processing device based on sentiment analysis corresponds one-to-one to the information processing method based on sentiment analysis in the above embodiment. Figure 3 , Figure 3 This is a schematic diagram of the functional modules of a preferred embodiment of the information processing device based on sentiment analysis of the present invention. It includes a data acquisition module 10, a sentiment analysis module 20, a user behavior analysis module 30, an event analysis and trend prediction module 40, an emotional state analysis module 50, a user emotional feedback acquisition module 60, a comprehensive emotional analysis module 70, and an output generation module 80. Each functional module is described in detail below:

[0201] A data collection module 10 is used to collect data and information, including text data, user behavior data, and event information;

[0202] Sentiment analysis module 20, used to perform natural language processing and sentiment analysis on the text data to identify the sentiment characteristics of the text data;

[0203] A user behavior analysis module 30 is used to analyze the user behavior data and determine the user's behavior pattern;

[0204] An event analysis and trend prediction module 40 is used to analyze the event information and determine the event change trend;

[0205] An emotional state analysis module 50 is used to determine an emotional state indicator based on the emotional characteristics, behavior patterns, and event change trends;

[0206] User emotional feedback collection module 60, used to monitor information on the online platform and obtain user emotional feedback;

[0207] The comprehensive emotion analysis module 70 is used to comprehensively analyze the emotional state indicators and user emotional feedback to generate an emotional analysis result;

[0208] The output generation module 80 is configured to generate target output content that responds to the user's emotional state based on the emotion analysis result.

[0209] In one embodiment, the emotional state analysis module 50 is specifically configured to:

[0210] generating a quantitative value of the emotional feature based on the emotional polarity and the emotional intensity in the emotional feature;

[0211] Analyzing the behavior frequency and the degree of behavior change in the behavior pattern to generate a quantitative value of the behavior pattern;

[0212] Generate a quantitative value of the event change trend based on the event's impact intensity and the event's impact coefficient on user emotions;

[0213] The emotional state index is generated by performing weighted analysis on the quantitative values ​​of emotional characteristics, behavioral patterns and event change trends.

[0214] In one embodiment, the user emotion feedback collection module 60 is specifically configured to:

[0215] Collect user-generated text content from online platforms;

[0216] Extracting sentiment polarity and sentiment intensity from the text content based on a sentiment analysis module, and generating sentiment data according to the sentiment polarity and sentiment intensity;

[0217] The emotional data is combined with the user's behavior pattern and historical emotional data, and comprehensively analyzed to generate the user emotional feedback.

[0218] In one embodiment, the comprehensive emotion analysis module 70 is specifically configured to:

[0219] Analyzing the deviation between the emotional state indicator and the user's emotional feedback;

[0220] A sentiment analysis result is generated based on the deviation, where the sentiment analysis result is used to reflect the degree of emotional fluctuation or emotional stability of the user.

[0221] In one embodiment, the sentiment analysis module 20 is specifically configured to:

[0222] Collect historical data and corresponding sentiment analysis results;

[0223] Based on the historical data and the corresponding sentiment analysis results, a sentiment analysis model is trained and generated, and the sentiment analysis model is used to perform natural language processing and sentiment analysis on text data.

[0224] In one embodiment, the emotional state analysis module 50 is specifically configured to:

[0225] Set warning thresholds for emotional state indicators;

[0226] When the emotional state indicator reaches the warning threshold, generating warning information;

[0227] The warning information is sent to the business processing end.

[0228] In one embodiment, the output generation module 80 is specifically configured to:

[0229] Obtain target users’ risk preference information and expected outcomes;

[0230] Generate personalized output content for the target user based on the risk preference information, expected results, and sentiment analysis results;

[0231] The personalized output content is sent to the target user.

[0232] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external user terminal via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the service side of an information processing method based on sentiment analysis.

[0233] In one embodiment, a computer device is provided. The computer device may be a user terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the user side of an information processing method based on sentiment analysis.

[0234] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0235] Collect data and information, including text data, user behavior data, and event information;

[0236] Performing natural language processing and sentiment analysis on the text data to identify sentiment features of the text data;

[0237] Analyze the user behavior data to determine the user's behavior pattern;

[0238] Analyze the event information to determine the event change trend;

[0239] Determining emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends;

[0240] Monitor information on online platforms and obtain user sentiment feedback;

[0241] Comprehensively analyzing the emotional state indicators and user emotional feedback to generate emotional analysis results;

[0242] Target output content is generated based on the emotion analysis result to respond to the user's emotional state.

[0243] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0244] Collect data and information, including text data, user behavior data, and event information;

[0245] Performing natural language processing and sentiment analysis on the text data to identify sentiment features of the text data;

[0246] Analyze the user behavior data to determine the user's behavior pattern;

[0247] Analyze the event information to determine the event change trend;

[0248] Determining emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends;

[0249] Monitor information on online platforms and obtain user sentiment feedback;

[0250] Comprehensively analyzing the emotional state indicators and user emotional feedback to generate emotional analysis results;

[0251] Target output content is generated based on the emotion analysis result to respond to the user's emotional state.

[0252] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the user side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0253] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0254] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0255] It should be noted that if any software tools or components other than those of the Company appear in the embodiments of this application, they are merely for illustration and do not represent actual use. The above embodiments are intended only to illustrate the technical solutions of the present invention, not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some of the technical features therein with equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. An information processing method based on sentiment analysis, characterized in that: The following steps are involved: Collect data and information, including text data, user behavior data, and event information; Performing natural language processing and sentiment analysis on the text data to identify sentiment features of the text data; Analyze the user behavior data to determine the user's behavior pattern; Analyze the event information to determine the event change trend; Determining emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends; Monitor information on online platforms and obtain user sentiment feedback; Analyzing the deviation between the emotional state indicator and the user's emotional feedback; generating a sentiment analysis result based on the deviation, wherein the sentiment analysis result is used to reflect the degree of emotional fluctuation or emotional stability of the user; Obtain target users’ risk preference information and expected outcomes; Generate personalized output content for the target user based on the risk preference information, expected results, and sentiment analysis results; The personalized output content is sent to the target user.

2. The information processing method based on sentiment analysis according to claim 1, characterized in that Based on the emotional characteristics, behavioral patterns, and event change trends, determine emotional state indicators, including: generating a quantitative value of the emotional feature based on the emotional polarity and the emotional intensity in the emotional feature; Analyzing the behavior frequency and the degree of behavior change in the behavior pattern to generate a quantitative value of the behavior pattern; Generate a quantitative value of the event change trend based on the event's impact intensity and the event's impact coefficient on user emotions; The emotional state index is generated by performing weighted analysis on the quantitative values ​​of emotional characteristics, behavioral patterns and event change trends.

3. The information processing method based on sentiment analysis according to claim 1, characterized in that Monitor information on online platforms and obtain user sentiment feedback, including: Collect user-generated text content from online platforms; Extracting sentiment polarity and sentiment intensity from the text content based on a sentiment analysis module, and generating sentiment data according to the sentiment polarity and sentiment intensity; The emotional data is combined with the user's behavior pattern and historical emotional data, and comprehensively analyzed to generate the user emotional feedback.

4. The information processing method based on sentiment analysis according to claim 1, characterized in that Before performing natural language processing and sentiment analysis on the text data and identifying the sentiment features of the text data, the method further includes: Collect historical data and corresponding sentiment analysis results; Based on the historical data and the corresponding sentiment analysis results, a sentiment analysis model is trained and generated, and the sentiment analysis model is used to perform natural language processing and sentiment analysis on text data.

5. The information processing method based on sentiment analysis according to claim 1, characterized in that: After determining the emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends, the following steps are also included: Set warning thresholds for emotional state indicators; When the emotional state indicator reaches the warning threshold, generating warning information; The warning information is sent to the business processing end.

6. An information processing device based on sentiment analysis, characterized in that: The information processing device based on sentiment analysis includes: A data collection module is used to collect data and information, including text data, user behavior data, and event information; A sentiment analysis module, configured to perform natural language processing and sentiment analysis on the text data to identify sentiment features of the text data; A user behavior analysis module, configured to analyze the user behavior data and determine the user's behavior pattern; An event analysis and trend prediction module is used to analyze the event information and determine the event change trend; An emotional state analysis module, configured to determine emotional state indicators based on the emotional characteristics, behavioral patterns, and event change trends; User emotional feedback collection module, used to monitor information on online platforms and obtain user emotional feedback; A comprehensive emotion analysis module is used to analyze the deviation between the emotional state indicator and the user's emotional feedback; generate an emotion analysis result based on the deviation, and the emotion analysis result is used to reflect the user's emotional fluctuation degree or emotional stability; The output generation module is used to obtain the risk preference information and expected results of the target user; generate personalized output content for the target user based on the risk preference information, expected results and sentiment analysis results; and send the personalized output content to the target user.

7. A computer device, characterized in that: The computer device includes a memory, a processor, and an information processing program based on sentiment analysis stored in the memory and capable of running on the processor. When the information processing program based on sentiment analysis is executed by the processor, the steps of the information processing method based on sentiment analysis as described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium, characterized in that The storage medium stores an information processing program based on sentiment analysis, and when the information processing program based on sentiment analysis is executed by the processor, the steps of the information processing method based on sentiment analysis as described in any one of claims 1 to 5 are implemented.

Images