Market emotion analysis method based on natural language processing

Abstract

The invention discloses a market emotion analysis method based on natural language processing, and the method comprises the steps: 1, combining market quantitative index data with a word bank to obtain an integrated database, 2, constructing a cloud processing system, 3, constructing a cloud recognition system, and 4, carrying out the system integration optimization to obtain a market emotion cloud analysis prediction system. Step 5, establishing an intelligent interaction APP and giving a market emotion prediction analysis result; the market emotion cloud analysis and prediction system is constructed by establishing the lexicon containing market communication terms and emotion vocabularies and integrating the lexicon with quantitative index data, and finally, the market emotion prediction and analysis result is given in combination with time sequence analysis, so that real-time monitoring and analysis can be realized, and instant market emotion feedback can be provided; the investor can be accurately and efficiently helped to better understand the multi-modal market emotion in real time, and the investor can be helped to make a more intelligent investment decision.

Description

Technical Field

[0001] The present invention relates to the technical field of financial market data analysis, and in particular to a market sentiment analysis method based on natural language processing. Background Art

[0002] In the financial market, investors' emotions have a significant impact on market fluctuations. Traditional market sentiment analysis methods mainly rely on quantitative indicators such as price and trading volume, but these methods are difficult to capture the market sentiment information contained in unstructured text data such as news reports and social media. Therefore, the present invention proposes a market sentiment analysis method based on natural language processing to solve the problems existing in the prior art. Summary of the invention

[0003] In view of the above problems, the purpose of the present invention is to propose a market sentiment analysis method based on natural language processing, which establishes a vocabulary library containing market communication terms and emotional vocabulary, and constructs a market sentiment cloud analysis and prediction system after integrating with quantitative indicator data. Finally, the market sentiment prediction analysis results are given in combination with time series analysis. It can monitor and analyze in real time and provide instant market sentiment feedback, which can accurately, efficiently and in real time help investors better understand multimodal market sentiment, and can help investors make more informed investment decisions.

[0004] To achieve the purpose of the present invention, the present invention is implemented by the following technical solution: a market sentiment analysis method based on natural language processing, comprising the following steps: Step 1: Obtain market quantitative indicator data based on network big data, build and import a vocabulary containing market communication terms and sentimental vocabulary, and combine the quantitative indicator data with the vocabulary to obtain an integrated database; Step 2: Based on the BERT pre-trained language model, natural language processing technology is used in combination with an integrated database to build a real-time data processing cloud system for feature extraction of text data, and transfer learning technology is introduced to switch the language model between multimodal markets to complete the construction of the cloud processing system; Step 3: Build a cloud-based recognition system for fine-grained emotion classification and intensity differentiation based on the regression model, and interact with the cloud-based processing system data; Step 4: Integrate the cloud processing system and the cloud recognition system into a multimodal market sentiment cloud analysis and prediction system, and introduce differential privacy technology and blockchain technology to optimize the system to obtain a market sentiment cloud analysis and prediction system; Step 5. Use Pixso software to build an intelligent interactive APP based on the user's mobile communication device, exchange data with the market sentiment cloud analysis and prediction system based on the user's input needs, and provide market sentiment forecast analysis results in combination with time series analysis.

[0005] Further improvements are as follows: the quantitative indicator data of the market in step one include product market price, product trading volume, average range index, Williams indicator and Bollinger Bands, and the vocabulary library containing market communication terms and emotional vocabulary is based on the vocabulary collected and sorted from network big data and is formed by differentiating multi-source data according to different modal markets.

[0006] A further improvement is that the transfer learning technology introduced in the step 2 specifically switches the BERT pre-trained language model to different modal markets according to user selection requirements, and uses the language model to extract information fragment features from the text data of the modal market for subsequent market sentiment analysis and prediction.

[0007] A further improvement is that the fine-grained emotion classification in step three specifically includes multi-dimensional emotions such as positive, negative, neutral, surprise and fear, and then a regression model is used to judge the low, medium and high intensity of the emotion in each dimension.

[0008] Further improvements are as follows: the differential privacy technology introduced in step 4 is to protect the data when some sensitive language data input by the user is required by differential privacy; the blockchain technology introduced is to store the record and verification of the sentiment analysis results in a block chain manner in an unalterable, secure and reliable manner.

[0009] A market sentiment analysis system based on natural language processing includes a cloud system and a user-end system. The cloud system includes a data acquisition module, a cloud processing module, a cloud identification module and an analysis and prediction module. The user-end system includes a customized input module, a dynamic update module and a prediction output module.

[0010] Further improvements are as follows: the data acquisition module acquires language data and quantitative indicator data of the multimodal market in real time based on network big data; the cloud processing module is used to identify the modal market and perform feature extraction on the language data based on the BERT pre-trained language model to obtain market sentiment semantic information fragments; the cloud recognition module identifies market sentiment and makes intensity judgments based on the extracted semantic information fragments; the analysis and prediction module performs product market analysis and trend prediction based on the identified market sentiment and intensity.

[0011] A further improvement is that the cloud system also includes a blockchain storage module, which stores the records and verifications of the sentiment analysis results in a tamper-proof, secure and reliable manner.

[0012] Further improvements are as follows: the customized input module provides users with personalized input of market investment needs based on the mobile intelligent interactive APP, the dynamic update module obtains multi-source data of the multimodal market in real time based on time series and network big data and performs real-time updates of market sentiment analysis and forecasts, and the forecast output module combines sentiment analysis forecasts with real-time market data to output visualization results.

[0013] A further improvement is that the user-side system also includes a differential privacy module and a fluctuation detection module. The differential privacy module protects some sensitive language data requirements input by the user, and the fluctuation detection module is used to detect fluctuations in market sentiment that the user is concerned about and issue an early warning when an abnormality occurs.

[0014] The beneficial effects of the present invention are as follows: the present invention establishes a vocabulary containing market communication terms and sentiment words, and integrates it with quantitative indicator data to build a market sentiment cloud analysis and prediction system, and finally combines time series analysis to give market sentiment prediction analysis results, which can monitor and analyze in real time and provide instant market sentiment feedback, and can accurately, efficiently and in real time help investors better understand multimodal market sentiment, and can help investors make more informed investment decisions; At the same time, the present invention processes, analyzes and predicts data on a cloud platform, which greatly reduces the server's operating load, improves server data processing and response speed, and ensures investors' privacy and data security through the introduction of differential privacy technology and blockchain technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of the method of Example 1 of the present invention.

[0016] Figure 2 This is a system architecture diagram of Example 2 of the present invention. DETAILED DESCRIPTION

[0017] In order to deepen the understanding of the present invention, the present invention will be further described in detail below in conjunction with examples. The examples are only used to explain the present invention and do not constitute a limitation on the protection scope of the present invention. Example

[0018] according to Figure 1 As shown, this embodiment provides a market sentiment analysis method based on natural language processing, comprising the following steps: Step 1: Based on the big data of the Internet, we obtain the market quantitative indicator data including product market price, product transaction volume, average range index, Williams index and Bollinger Bands, and at the same time, we build and import the word library containing market communication terms and emotional vocabulary, and use the data warehouse model to combine the quantitative indicator data and the word library to obtain an integrated database; The vocabulary database is based on the words collected and sorted from the Internet big data and is formed by differentiating multi-source data according to different modal markets. This type of differentiated database facilitates the subsequent data extraction and processing, and eliminates the need for secondary classification of data in the subsequent process.

[0019] Step 2: Based on the BERT pre-trained language model, natural language processing technology is used in combination with an integrated database to build a real-time data processing cloud system for feature extraction of text data, and transfer learning technology is introduced to switch the language model between multimodal markets to complete the construction of the cloud processing system. The cloud processing system here is organized and built on the cloud platform, which can reduce the operating load of the offline server; The transfer learning technology introduced here specifically switches the BERT pre-trained language model to different modal markets according to user selection needs, and uses the language model to extract information fragment features from the text data of the modal market for subsequent market sentiment analysis and prediction. The transfer learning method can provide a fast-response market sentiment analysis service that switches between multi-modal markets for each user who creates an account.

[0020] Step 3: Build a cloud-based recognition system for fine-grained emotion classification and intensity differentiation based on the regression model, and interact with the cloud-based processing system data; The fine-grained emotion classification specifically includes positive, negative, neutral, surprise and fear emotions in multiple dimensions. Then, a regression model is used to judge the low, medium and high intensity of emotions in each dimension. The distinction of multi-dimensional emotions and the division of the intensity of each emotion can enable more detailed and in-depth market sentiment analysis.

[0021] The cloud processing system here is also organized and built on the cloud platform, and has the same effect as the aforementioned cloud processing system, which is to reduce the operating load of offline servers.

[0022] Step 4: Integrate the cloud processing system and the cloud recognition system into a multimodal market sentiment cloud analysis and prediction system, and introduce differential privacy technology and blockchain technology to optimize the system to obtain a market sentiment cloud analysis and prediction system; The differential privacy technology introduced here is to protect some sensitive language data input by users by using differential privacy. After the cloud receives the user input data, it performs differential privacy protection on the input data information. The blockchain technology introduced is to store the record and verification of sentiment analysis results in a tamper-proof, secure and reliable manner in a block chain. After the cloud system outputs the final result, it stores the output result and verification information securely to prevent loss.

[0023] Step 5. Use Pixso software to build an intelligent interactive APP based on the user's mobile communication device, exchange data with the market sentiment cloud analysis and prediction system based on the user's input needs, and combine time series analysis to give market sentiment forecast analysis results; the application of time series can monitor and analyze the latest financial text data in real time, and provide instant market sentiment feedback.

[0024] The Smart Interaction APP allows users to customize sentiment analysis services based on their own investment portfolios and receive market sentiment updates related to their own holdings. It can also communicate with users through natural language to provide personalized market sentiment analysis and investment advice. Example

[0025] according to Figure 2 As shown, this embodiment provides a market sentiment analysis system based on natural language processing, including a cloud system and a user-side system. The cloud system includes a data acquisition module, a cloud processing module, a cloud identification module and an analysis and prediction module. The cloud system also includes a blockchain storage module. The user-side system includes a customized input module, a dynamic update module and a prediction output module. The user-side system also includes a differential privacy module and a fluctuation detection module.

[0026] Data acquisition module: Based on network big data, it can acquire language data of multimodal markets and market quantitative indicator data including product market price, product trading volume, average range index, Williams index and Bollinger Bands in real time; Cloud processing module: used to identify modal markets and extract features from language data based on the BERT pre-trained language model to obtain market sentiment semantic information fragments, that is, to use natural language processing technology to perform word segmentation, part-of-speech tagging, syntactic analysis, etc. on language data to achieve recognition and extraction of semantic information; Cloud recognition module: Identify market sentiment and judge its intensity based on the extracted semantic information fragments. According to the extracted semantics, determine whether the market sentiment belongs to the multi-dimensional emotions such as positive, negative, neutral, surprise and fear, and judge and distinguish its intensity; Analysis and forecasting module: Analyze the product market and forecast trends based on the identified market sentiment and strength; Blockchain storage module: The records and verifications of sentiment analysis results are stored in a blockchain format in an unalterable, secure and reliable manner, improving the transparency and unalterability of data.

[0027] Customized input module: Based on the mobile intelligent interactive APP, it provides users with personalized input of market investment needs, allowing users to customize sentiment analysis services according to their own investment portfolios and receive market sentiment updates related to their own positions; Dynamic update module: Based on time series and network big data, it can obtain multi-source data of multi-modal markets in real time and update market sentiment analysis and forecasts in real time. It can monitor and analyze the latest financial text data in real time and provide instant market sentiment feedback; Prediction output module: Combines sentiment analysis predictions with real-time market data to output visualization results, and can communicate with users through natural language to provide personalized market sentiment analysis and investment advice; Differential privacy module: protects some sensitive language data input by users; Fluctuation detection module: used to detect market sentiment fluctuations that users are concerned about and issue warnings when abnormalities occur. It can monitor sentiment changes in real time. When abnormal sentiment fluctuations are detected, timely warnings can be issued, which can greatly reduce investors' losses and achieve the effect of timely stopping losses.

[0028] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A market sentiment analysis method based on natural language processing, characterized in that: The following steps are involved: Step 1: Obtain market quantitative indicator data based on network big data, build and import a vocabulary containing market communication terms and sentiment vocabulary, and combine the quantitative indicator data with the vocabulary to obtain an integrated database; Step 2: Based on the BERT pre-trained language model, natural language processing technology is used in combination with an integrated database to build a real-time data processing cloud system for feature extraction of text data, and transfer learning technology is introduced to switch the language model between multimodal markets to complete the construction of the cloud processing system; Step 3: Build a cloud-based recognition system for fine-grained emotion classification and intensity differentiation based on the regression model, and interact with the cloud-based processing system data; Step 4: Integrate the cloud processing system and the cloud recognition system into a multimodal market sentiment cloud analysis and prediction system, and introduce differential privacy technology and blockchain technology to optimize the system to obtain a market sentiment cloud analysis and prediction system; Step 5. Use Pixso software to build an intelligent interactive APP based on the user's mobile communication device, exchange data with the market sentiment cloud analysis and prediction system based on the user's input needs, and provide market sentiment forecast analysis results in combination with time series analysis.

2. The market sentiment analysis method based on natural language processing according to claim 1, characterized in that: The quantitative indicator data of the market in step one include product market price, product trading volume, average range index, Williams indicator and Bollinger Bands. The vocabulary containing market communication terms and emotional vocabulary is based on the vocabulary collected and sorted from network big data and is formed by differentiating multi-source data according to different modal markets.

3. The market sentiment analysis method based on natural language processing according to claim 1, characterized in that: The transfer learning technology introduced in the step 2 specifically switches the BERT pre-trained language model to different modal markets according to user selection requirements, and uses the language model to extract information fragment features from the text data of the modal market for subsequent market sentiment analysis and prediction.

4. The market sentiment analysis method based on natural language processing according to claim 1, characterized in that: The fine-grained emotion classification in step three specifically includes multi-dimensional emotions such as positive, negative, neutral, surprise and fear, and then a regression model is used to judge the low, medium and high intensity of the emotion in each dimension.

5. The market sentiment analysis method based on natural language processing according to claim 1, characterized in that: The differential privacy technology introduced in the step 4 is to protect the data when some sensitive language data input by the user is required by differential privacy; the blockchain technology introduced is to store the record and verification of the sentiment analysis results in a block chain manner in an unalterable, secure and reliable manner.

6. A market sentiment analysis system based on natural language processing, characterized by: It includes a cloud system and a user-end system. The cloud system includes a data acquisition module, a cloud processing module, a cloud identification module and an analysis and prediction module. The user-end system includes a customized input module, a dynamic update module and a prediction output module.

7. The market sentiment analysis system based on natural language processing according to claim 6, characterized in that: The data acquisition module acquires language data and quantitative indicator data of the multimodal market in real time based on network big data. The cloud processing module is used to identify the modal market and extract features from the language data based on the BERT pre-trained language model to obtain market sentiment semantic information fragments. The cloud recognition module identifies market sentiment and makes intensity judgments based on the extracted semantic information fragments. The analysis and prediction module performs product market analysis and trend prediction based on the identified market sentiment and intensity.

8. The market sentiment analysis system based on natural language processing according to claim 6, characterized in that: The cloud system also includes a blockchain storage module, which records and verifies the sentiment analysis results in a blockchain-style storage in an unalterable, secure and reliable manner.

9. The market sentiment analysis system based on natural language processing according to claim 6, characterized in that: The customized input module provides users with personalized market investment demand input based on the mobile intelligent interactive APP. The dynamic update module obtains multi-source data of the multimodal market in real time based on time series and network big data and performs real-time updates of market sentiment analysis and prediction. The prediction output module combines sentiment analysis prediction and real-time market data to output visualization results.

10. The market sentiment analysis system based on natural language processing according to claim 6, characterized in that: The user-side system also includes a differential privacy module and a fluctuation detection module. The differential privacy module protects some sensitive language data requirements input by users, and the fluctuation detection module is used to detect fluctuations in market sentiment that users are concerned about and issue warnings when abnormalities occur.

Images