Multi-channel target information automatic collection and fusion method and system

Abstract

The present invention relates to the technical field of information collection and fusion, and discloses a multi-channel target information automatic collection and fusion method and system, which obtains target information from different channels to ensure the comprehensiveness of target information; cleans and pre-processes the acquired target information, and classifies the target information according to target characteristics to facilitate the subsequent fusion of the target information; safely fuses the classified target information to better utilize and protect the target information, comprehensively analyzes the target information, and improves the intelligence level; monitors the update and change of the target information in real time, and warns the target in real time to improve the stability and accuracy of the target prediction; and provides personalized target recommendations based on the user's historical behavior data and target information.

Description

Technical Field

[0001] The present invention relates to the technical field of information collection and fusion, and in particular to a method and system for automatically collecting and fusion of multi-channel target information. Background Art

[0002] At present, the automatic collection of target information is realized by using technologies such as natural language processing, machine learning, and data mining, which improves the efficiency and timeliness of obtaining target information. The target information can be comprehensively analyzed from multiple channels to accurately grasp the dynamic changes of the target. The target information here is business opportunity information, so as to analyze the dynamic changes of business opportunities. However, most of them have not solved how to obtain business opportunity information, classify target information according to business opportunity characteristics, and safely integrate it, and then monitor the changes of business opportunity information and warn of business opportunities while providing personalized target recommendations.

[0003] For example, a Chinese patent with publication number CN112069484A discloses a method and system for information collection based on multimodal interaction, the method comprising: identity authentication based on the face image and voice clip of the person to be verified; pre-identifying the attribute information related to the person to be verified through the face image and voice clip of the person to be verified; obtaining the demand information of the person to be verified; determining the content of the questionnaire according to the demand information and the attribute information; presenting the content of the questionnaire to the person to be verified in a multimodal interactive form to obtain the multimodal questionnaire answer of the person to be verified; fusing the multimodal questionnaire answer; and verifying the validity of the information according to the fusion result. Identity authentication is performed through the face image and voice clip of the person to be verified, using voice and face image information, starting from pattern recognition and multimodal fusion analysis methods, the association of multiple questionnaire identities of the user is completed from the authentication end, and then the information of multiple dimensions of the user is fully mined, and the validity of this information is judged through the interactive process. For information with questionable accuracy, the validity of the information is further confirmed through secondary interaction.

[0004] For example, the Chinese patent with the authorization announcement number CN109634407B discloses a control method for mobile synchronous acquisition and fusion based on multimodal brain-eye information fusion. In view of the requirements for human physiological information acquisition, visual tracking and brain-computer interface technology are used to collect human visual and consciousness signals as the input of the multimodal information channel; an augmented reality environment based on a digital helmet is developed, and an integrated sampling interface based on event triggering is designed to realize mobile synchronous acquisition and output of information flow; ReliefF is used to evaluate and induce ordered information aggregation operators to realize the fusion of human physiological information at the feature layer and the weight layer; under the condition of digital helmet integration, the coordinated control of the instruction set of the controlled object is realized under the multimodal information state. It solves the problem of insufficient robustness and stability of human-computer interaction information representation existing in the prior art.

[0005] The above patents have the problem raised by the background technology: they do not solve the problem of how to obtain business opportunity information, classify business opportunity information according to business opportunity characteristics, and safely integrate them, and then monitor changes in business opportunity information and warn of business opportunities while providing personalized target recommendations. To solve this problem, the present invention proposes a multi-channel target information automatic collection and integration method and system. Summary of the invention

[0006] In view of the problems existing in the above-mentioned existing multi-channel target information automatic collection and fusion method and system, the present invention is proposed.

[0007] Therefore, the object of the present invention is to provide a method and system for automatically collecting and fusing multi-channel target information.

[0008] In order to solve the above technical problems, the present invention provides a multi-channel target information automatic collection and fusion method:

[0009] Obtain target information from different channels;

[0010] Clean and preprocess the acquired target information, and classify the target information according to the target characteristics;

[0011] Securely fuse the classified target information;

[0012] Monitor updates and changes of target information in real time, and issue early warnings to targets in real time;

[0013] Provide personalized target recommendations based on the user's historical behavior data and target information.

[0014] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the strategy of safe fusion includes:

[0015] Divide the target information into data fragments, encrypt each data fragment, store the encrypted data fragments in different physical locations, and build a data index table to record the physical location and information of each data fragment;

[0016] The strategies for segmenting data segments include:

[0017] Determine the segmentation granularity of the target information and the division boundaries of the data segments, which include text, charts, images, and videos, and add a unique identifier to each data segment, which includes the data segment sequence number, division boundary, and target features, and encrypt each data segment;

[0018] Strategies for decentralized storage include:

[0019] An encrypted data fragment is stored at an initial physical location and the initial physical location is marked, and then other data fragments are stored and marked at physical locations adjacent to the initial physical location until all data fragments are stored.

[0020] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the target information includes text target information, image target information, chart target information and video target information;

[0021] Extract the target features, obtain the text feature vector, image feature vector, chart feature vector and video feature vector respectively, fuse the classified target information, calculate the similarity of the target information respectively, and convert the similarity of the target information into the attention weight of the target information. The function expression of the attention weight of the target information is as follows:

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] In the formula, Represents the attention weight of the text target information, The attention weight representing the target information of the image, The attention weight representing the target information of the graph, Represents the attention weight of the video target information, represents the normalized exponential function, Indicates the similarity of text target information, Represents the similarity of image target information, Indicates the similarity of the target information of the graph, Indicates the similarity of the video target information;

[0027] The attention weight of the target information is normalized, and the normalized function expression is as follows:

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] In the formula, represents the normalized attention weight of the text target information, represents the normalized attention weight of the image target information, represents the normalized attention weight of the target information of the chart, Represents the normalized attention weight of the video target information.

[0033] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the strategy of safe fusion also includes:

[0034] The fusion function expression of target information is as follows:

[0035] ;

[0036] In the formula, represents the fusion result of target information, represents the text feature vector, represents the image feature vector, represents the graph feature vector, Represents the video feature vector.

[0037] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the fusion result of the target information is monitored in real time and recorded in time, and the function expression of the predicted target is as follows:

[0038] ;

[0039] In the formula, Indicates the prediction time The fusion result of target information is represents a constant, , and represents the autoregressive coefficient, Indicates time The fusion result of target information is Indicates time The fusion result of target information is Indicates time The fusion result of target information is represents the error term.

[0040] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the function expression for predicting target changes is as follows:

[0041] ;

[0042] In the formula, Indicates the prediction time The target changes when Represents a neural network model.

[0043] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the process of providing personalized target recommendation based on the user's historical behavior data and target information is as follows:

[0044] S51, collecting the user's historical behavior data and target information, and preprocessing the user's historical behavior data;

[0045] S52, constructing a user goal matrix based on the user's historical behavior data and goal information, and generating user goal preferences;

[0046] The construction rules of the user target matrix are as follows:

[0047] The historical behavior data of different types of users are weighted, and the time decay factor is introduced to consider the decay trend of the user's behavior towards the target over time. The user's historical behavior data and target information are continuously monitored to dynamically update the user target matrix.

[0048] S53, generating a sequence table for personalized recommendation based on the user's target preference and the features of the target information, and recommending the target features as keywords;

[0049] S54. Update user target preferences in real time, collect user feedback on target recommendations, and optimize the sequence list.

[0050] As a preferred solution of the multi-channel target information automatic collection and fusion method of the present invention, the process of cleaning and preprocessing the collected target information is as follows:

[0051] For the text target information, special characters and stop words are removed from the text, and the text is segmented into words, and the frequency of the words in the text is counted;

[0052] For the image target information, remove the noise in the image, adjust the contrast and brightness of the image, and resize the image to a uniform size;

[0053] For the target information of the chart, the digital and text data in the chart are extracted, the missing values ​​in the chart are filled, the abnormal values ​​in the chart are processed, and the charts in different formats are converted into a unified format;

[0054] For the video target information, the video is segmented into individual frames, and the key frames in the video are extracted, and the data in the video is serialized.

[0055] Multi-channel target information automatic collection and fusion system, including: data collection module, data processing module, information fusion module, real-time monitoring module and personalized recommendation module;

[0056] The data acquisition module is used to obtain target information from different channels;

[0057] The data processing module is used to clean and preprocess the acquired target information and classify the target information according to the target characteristics;

[0058] The information fusion module is used to safely fuse the classified target information;

[0059] The real-time monitoring module is used to monitor the updates and changes of target information in real time and to warn the target in real time;

[0060] The personalized recommendation module is used to provide personalized target recommendations based on the user's historical behavior data and target information.

[0061] A computer device includes a memory for storing instructions and a processor for executing the instructions, so that the device executes a method for automatically collecting and fusing multi-channel target information.

[0062] A computer-readable storage medium stores a computer program, which, when executed, implements a method for automatically collecting and fusing multi-channel target information.

[0063] The beneficial effects of the present invention are as follows: the present invention obtains target information from different channels to ensure the comprehensiveness of target information; cleans and preprocesses the obtained target information, and classifies the target information according to target characteristics to facilitate the subsequent fusion of the target information; securely fuses the classified target information to better utilize and protect the target information, comprehensively analyze the target information, and improve the level of intelligence; monitors the update and change of target information in real time, and warns the target in real time to improve the stability and accuracy of target prediction; provides personalized target recommendations based on the user's historical behavior data and target information. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0065] Figure 1 A method flow chart of the multi-channel target information automatic collection and fusion method of the present invention;

[0066] Figure 2It is a business opportunity information fusion flow chart of the multi-channel target information automatic collection and fusion method of the present invention;

[0067] Figure 3 This is a system structure diagram of the multi-channel target information automatic collection and fusion system of the present invention. DETAILED DESCRIPTION

[0068] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0069] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0070] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0071] Example 1

[0072] In this embodiment, a method flow chart of a multi-channel target information automatic collection and fusion method is provided, such as Figure 1 As shown, the multi-channel target information automatic collection and fusion method includes:

[0073] S1. Obtain business opportunity information from different channels.

[0074] The business opportunity information includes text business opportunity information, image business opportunity information, chart business information and video business opportunity information.

[0075] In specific applications, business opportunity information is collected from different channels. Here, business opportunity information is mainly collected from four channels: text business opportunity information, image business opportunity information and video business opportunity information are collected from social media. Text business opportunity information in social media includes text content such as user posts, comments, and reposts. Image business opportunity information in social media includes image content such as pictures, posters, and product display pictures shared by users. Video business opportunity information in social media includes video content and advertising videos posted by users.

[0076] Collect text business opportunity information and image business opportunity information from news media. Text business opportunity information in news media includes news reports, industry analysis articles, reviews, etc. Image business opportunity information in news media includes news illustrations, event photos, etc.;

[0077] Collect text opportunity information and graphic opportunity information from industry reports and research. The text opportunity information in industry reports and research includes industry reports, market research, analysis reports, etc. The graphic opportunity information in industry reports and research includes statistical data, market trend charts, etc.;

[0078] Text business opportunity information is collected from customer feedback and surveys. The text business opportunity information in customer feedback and surveys includes customer feedback, questionnaire completion results, etc.

[0079] S2. Clean and pre-process the acquired business opportunity information, and classify the business opportunity information according to business opportunity characteristics.

[0080] The process of cleaning and preprocessing the acquired business opportunity information is as follows:

[0081] For text opportunity information, remove special characters and stop words from the text, split the text into words, and count the frequency of the words in the text;

[0082] For image business opportunity information, remove the noise in the image, adjust the contrast and brightness of the image, and resize the image to a uniform size;

[0083] For chart opportunity information, extract the numerical and text data in the chart, fill in the missing values ​​in the chart, process the abnormal values ​​in the chart, and convert charts of different formats into a unified format;

[0084] For video opportunity information, the video is segmented into individual frames, key frames in the video are extracted, and the data in the video is serialized.

[0085] In specific applications, for text business opportunity information, it is necessary to remove special characters and stop words in the text to reduce noise interference, segment the text into words, and perform stem extraction to reduce vocabulary variations and improve the consistency and comparability of the text. Then, the frequency of text words is counted to measure the importance of words in the text, and the topic model is used to extract the topic information in the text;

[0086] For image business opportunity information, image processing technology needs to be used to remove noise and blur in the image to improve image quality, adjust the contrast and brightness of the image, enhance the visual effect of the image, make the business opportunity information easier to identify and understand, and adjust all images to a uniform size to facilitate the subsequent fusion and association of business opportunity information. At the same time, a deep learning model is used to extract the feature representation of the image to capture the key information in the image;

[0087] For chart business opportunity information, it is necessary to use computer vision technology to extract digital and text data in the chart, use interpolation methods to fill in missing values ​​in the chart and process abnormal values ​​in the chart to ensure the accuracy and reliability of business opportunity information, and convert charts in different formats into a unified format;

[0088] For video business opportunity information, it is necessary to segment the video into individual frames and extract the key frames in the video, that is, the frames containing the most important information. At the same time, the data in the video needs to be serialized to facilitate time series analysis.

[0089] S3. Securely integrate the classified business opportunity information.

[0090] Security convergence strategies include:

[0091] The business opportunity information is divided into data fragments, and each data fragment is encrypted. The encrypted data fragments are dispersed and stored in different physical locations, and a data index table is constructed to record the physical location and information of each data fragment.

[0092] The strategies for segmenting data segments include:

[0093] Determine the segmentation granularity of business opportunity information and the boundaries of data segments. The boundaries include text, charts, images, and videos. Add a unique identifier to each data segment. The unique identifier includes the data segment sequence number, the boundaries, and the business opportunity characteristics. Each data segment is encrypted.

[0094] Strategies for decentralized storage include:

[0095] An encrypted data fragment is stored at an initial physical location and the initial physical location is marked, and then other data fragments are stored and marked at physical locations adjacent to the initial physical location until all data fragments are stored.

[0096] The features of business opportunity information are extracted to obtain text feature vectors, image feature vectors, chart feature vectors and video feature vectors respectively. The classified business opportunity information is integrated, and the similarity of the business opportunity information is calculated respectively. Then, the similarity of the business opportunity information is converted into the attention weight of the business opportunity information. The function expression of the attention weight of the business opportunity information is as follows:

[0097] ;

[0098] ;

[0099] ;

[0100] ;

[0101] In the formula, Represents the attention weight of text opportunity information, The attention weight representing the image business opportunity information, Represents the attention weight of the chart opportunity information, Represents the attention weight of the video opportunity information, represents the normalized exponential function, Indicates the similarity of text opportunity information, Represents the similarity of image business opportunity information, Indicates the similarity of the chart opportunity information. Indicates the similarity of video opportunity information.

[0102] The attention weight of business opportunity information is normalized, and the normalized function expression is as follows:

[0103] ;

[0104] ;

[0105] ;

[0106] ;

[0107] In the formula, represents the normalized attention weight of the text opportunity information, represents the normalized attention weight of the image business opportunity information, represents the normalized attention weight of the chart opportunity information, Represents the normalized attention weight of the video opportunity information.

[0108] Finally, the fusion function expression of business opportunity information is as follows:

[0109] ;

[0110] In the formula, represents the fusion result of business opportunity information, represents the text feature vector, represents the image feature vector, represents the graph feature vector, Represents the video feature vector.

[0111] It should be explained that the similarity of business opportunity information is calculated through the feature vector of each modality, the similarity of text business opportunity information is calculated through the text similarity algorithm, the similarity of chart business opportunity information is calculated through the Euclidean distance, and the similarity of image business opportunity information and video business opportunity information is calculated through the extracted features.

[0112] The similarity of the text business opportunity information is calculated by using the cosine similarity algorithm. For example, there are two text feature vectors A and B, and the text feature vectors are respectively represented in the form of feature vectors. Then, the calculation formula for the similarity of the text business opportunity information is as follows:

[0113] ;

[0114] In the formula, Represents the similarity results of text opportunity information, Represents text feature vector and The inner product of Represents text feature vector The norm of ; Represents text feature vector The norm of .

[0115] For example, there are two chart feature vectors C and D. The calculation formula for the similarity of chart business opportunity information is as follows:

[0116] ;

[0117] In the formula, Represents the similarity results of chart opportunity information, Represents the graph feature vector The data points, Represents the graph feature vector The data points, Represents the number of data points.

[0118] For example, there are two image feature vectors E and F. The calculation formula for the similarity of image business opportunity information is as follows:

[0119] ;

[0120] In the formula, Represents the similarity result of image business opportunity information, Represents the image feature vector and The inner product of Represents the image feature vector The norm of ; Represents the image feature vector The norm of .

[0121] For example, there are two video feature vectors G and H. The calculation formula for the similarity of video business opportunity information is as follows:

[0122] ;

[0123] In the formula, Represents the similarity results of video business opportunity information. and Respectively represent the height and width of the video frame, express The pixel position in the video frame is The pixel value at express The pixel position in the video frame is The pixel value at .

[0124] In specific applications, the process of business opportunity information fusion is as follows: Figure 2 As shown, first, the features of the business opportunity information are extracted respectively to obtain the text feature vector, image feature vector, chart feature vector and video feature vector. The similarity of the business opportunity information is calculated according to the feature vector of the business opportunity information, and the similarity of the business opportunity information is converted into the attention weight of the business opportunity information through the normalized exponential function. At this time, the attention weight is non-normalized, so it is necessary to normalize the attention weight of the business opportunity information. Finally, the normalized attention weight and the feature vector are weighted and calculated to obtain the fusion result of the business opportunity information. According to the characteristics of the input business opportunity information, the importance of different modes is adaptively calculated, so that the business opportunity information can be better utilized, the business opportunity information can be comprehensively analyzed, and the level of intelligence can be improved.

[0125] S4. Monitor the updates and changes of business opportunity information in real time, and issue early warnings of business opportunities in real time.

[0126] The fusion results of business opportunity information are monitored in real time and recorded in different time periods. The function expression for predicting business opportunities is as follows:

[0127] ;

[0128] In the formula, Indicates the prediction time The fusion result of business opportunity information at that time, represents a constant, , and represents the autoregressive coefficient, Indicates time The fusion result of business opportunity information at that time, Indicates time The fusion result of business opportunity information at that time, Indicates time The fusion result of business opportunity information at that time, represents the error term.

[0129] What needs to be explained is: constant It refers to the intercept of the function for predicting business opportunity changes. The autoregressive parameter indicates the The fusion result of current business opportunity information and its past The linear relationship between the fusion results of business opportunity information at different times. These autoregressive parameters measure the influence of the fusion results of past business opportunity information on the fusion results of current business opportunity information. The error term Indicates unexplained noise.

[0130] Then the function expression for predicting business opportunity changes is as follows:

[0131] ;

[0132] In the formula, Indicates the prediction time Business opportunities change over time. Represents a neural network model.

[0133] What needs to be explained is: Neural network model It uses the fusion results of business opportunity information predicted at different times to predict future changes in business opportunities. The neural network model here needs to be continuously trained.

[0134] In specific applications, time series models are used to build forecast business opportunities and obtain the fusion results of business opportunity information at a certain time. The fusion results of business opportunity information at multiple time periods are then used as input data. This can comprehensively consider the information required for business opportunity forecasting and use continuous training of neural network models to predict future business opportunity changes, thereby facilitating real-time early warning of business opportunities and improving the stability and accuracy of business opportunity forecasting.

[0135] S5. Provide personalized business opportunity recommendations based on users’ historical behavior data and business opportunity information.

[0136] The process of providing personalized business opportunity recommendations based on the user's historical behavior data and business opportunity information is as follows:

[0137] S51, collecting the user's historical behavior data and business opportunity information, and pre-processing the user's historical behavior data;

[0138] S52. Based on the user's historical behavior data and business opportunity information, a user business opportunity matrix is ​​constructed to generate the user's business opportunity preference;

[0139] The construction rules of the user opportunity matrix are as follows:

[0140] Perform weighted processing on the historical behavior data of different types of users, and introduce a time decay factor to consider the decay trend of users' behavior towards business opportunities over time. In addition, the historical behavior data and business opportunity information of users are constantly monitored to dynamically update the user business opportunity matrix.

[0141] S53, generating a sequence table for personalized recommendation based on the user's business opportunity preference and the characteristics of the business opportunity information, and using the business opportunity characteristics as keywords for recommendation;

[0142] S54. Update user business opportunity preferences in real time, collect user feedback on business opportunity recommendations, and optimize the sequence list.

[0143] In specific applications, users' attention to business opportunities is collected. The ways to analyze attention include browsing, searching and clicking behaviors, as well as interaction data on social media. Then, a collaborative filtering algorithm is used to analyze users' historical behavior data, and a user-business opportunity matrix is ​​constructed in combination with business opportunity information to generate user business opportunity preferences. A personalized recommendation sequence is generated by combining user business opportunity preferences and the characteristics of business opportunity information, and recommendations are made using business opportunity keywords. Finally, user business opportunity preferences are updated in real time to reflect users' latest interests and behavior changes, and user feedback on personalized business opportunity recommendations is collected to optimize the personalized recommendation sequence, which can provide users with more accurate and targeted business opportunity recommendations.

[0144] The secure storage and transmission of business opportunity information and user historical behavior data are guaranteed throughout the entire method process.

[0145] The rules for securely storing and transmitting business opportunity information and user historical behavior data are as follows:

[0146] First, business opportunity information and user historical behavior data are divided into data fragments, and each data fragment is encrypted; then the encrypted data fragments are dispersed and stored in different physical locations; secondly, a data index table is constructed to record the physical location and related information of each data fragment; finally, a role-based access mechanism is established.

[0147] In specific applications, business opportunity information and user historical behavior data are divided into multiple data fragments, and each data fragment is encrypted to ensure the security of data during transmission and storage; the encrypted data fragments are stored in different physical locations to increase data security and anti-attack capabilities; a data index table is built to record the physical location and related information of each data fragment, manage the storage and retrieval of data fragments, and ensure that data can be restored and accessed in a timely manner; a role-based access mechanism is established to manage data access rights according to user permissions.

[0148] Example 2

[0149] In this embodiment, a system structure diagram of a multi-channel target information automatic collection and fusion system is provided, such as Figure 3 As shown, the multi-channel target information automatic collection and fusion system includes a data collection module, a data processing module, an information fusion module, a real-time monitoring module and a personalized recommendation module.

[0150] The data collection module is used to obtain business opportunity information from different channels.

[0151] The data processing module is used to clean and pre-process the acquired business opportunity information and classify the business opportunity information according to business opportunity characteristics.

[0152] The information fusion module is used to securely fuse the classified business opportunity information.

[0153] The real-time monitoring module is used to monitor the updates and changes of business opportunity information in real time, and to provide real-time warnings of business opportunities.

[0154] The personalized recommendation module is used to provide personalized business opportunity recommendations based on the user's historical behavior data and business opportunity information.

[0155] The specific methods for implementing the functions of the above modules can be found in the above multi-channel target information automatic collection and fusion method, which will not be described here one by one.

[0156] Example 3

[0157] In this embodiment, a computer device is provided, including a memory and a processor, the memory is used to store instructions, and the processor is used to execute the instructions, so that the computer device executes the steps of implementing the above-mentioned multi-channel target information automatic collection and fusion method.

[0158] Example 4

[0159] In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed, the steps of the above-mentioned multi-channel target information automatic collection and fusion method are implemented.

[0160] The computer-readable storage medium includes: a U disk, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk, and other media for storing program codes.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A multi-channel target information automatic collection and fusion method, characterized by: include, Obtain target information from different channels; Clean and preprocess the acquired target information, and classify the target information according to the target characteristics; Securely fuse the classified target information; Security convergence strategies include: Divide the target information into data fragments, encrypt each data fragment, store the encrypted data fragments in different physical locations, and build a data index table to record the physical location and information of each data fragment; The strategies for segmenting data segments include: Determine the segmentation granularity of the target information and the division boundaries of the data segments, which include text, charts, images, and videos, and add a unique identifier to each data segment, which includes the data segment sequence number, division boundary, and target features, and encrypt each data segment; Strategies for decentralized storage include: The encrypted data fragment is stored at the initial physical location and the initial physical location is marked, and then other data fragments are stored and marked at the physical locations adjacent to the initial physical location until all the data fragments are stored; Security integration strategies also include: The fusion function expression of target information is as follows: ； In the formula, represents the fusion result of target information, represents the text feature vector, represents the image feature vector, represents the graph feature vector, represents the video feature vector; Monitor updates and changes of target information in real time, and issue early warnings to targets in real time; Provide personalized target recommendations based on the user's historical behavior data and target information; Based on the user's historical behavior data and target information, the process of providing personalized target recommendations is as follows: S51, collecting the user's historical behavior data and target information, and preprocessing the user's historical behavior data; S52, constructing a user goal matrix based on the user's historical behavior data and goal information, and generating user goal preferences; The construction rules of the user target matrix are as follows: The historical behavior data of different types of users are weighted, and the time decay factor is introduced to consider the decay trend of the user's behavior towards the target over time. The user's historical behavior data and target information are continuously monitored to dynamically update the user target matrix. S53, generating a sequence table for personalized recommendation based on the user's target preference and the features of the target information, and recommending the target features as keywords; S54. Update user target preferences in real time, collect user feedback on target recommendations, and optimize the sequence list.

2. The multi-channel target information automatic collection and fusion method as claimed in claim 1, characterized in that: The target information includes text target information, image target information, chart target information and video target information; Extract the target features, obtain the text feature vector, image feature vector, chart feature vector and video feature vector respectively, fuse the classified target information, calculate the similarity of the target information respectively, and convert the similarity of the target information into the attention weight of the target information. The function expression of the attention weight of the target information is as follows: ； ； ； ； In the formula, Represents the attention weight of the text target information, The attention weight representing the target information of the image, The attention weight representing the target information of the graph, Represents the attention weight of the video target information, represents the normalized exponential function, Indicates the similarity of text target information, Represents the similarity of image target information, Indicates the similarity of the target information of the graph, Indicates the similarity of the video target information; The attention weight of the target information is normalized, and the normalized function expression is as follows: ； ； ； ； In the formula, represents the normalized attention weight of the text target information, represents the normalized attention weight of the image target information, represents the normalized attention weight of the target information of the chart, Represents the normalized attention weight of the video target information.

3. The multi-channel target information automatic collection and fusion method as claimed in claim 2, characterized in that: The fusion results of the target information are monitored in real time and recorded in different time periods. The function expression of the predicted target is as follows: ； In the formula, Indicates the prediction time The fusion result of target information is represents a constant, , and represents the autoregressive coefficient, Indicates time The fusion result of target information is Indicates time The fusion result of target information is Indicates time The fusion result of target information is represents the error term.

4. The multi-channel target information automatic collection and fusion method as claimed in claim 3, characterized in that: The function expression for predicting target changes is as follows: ； In the formula, Indicates the prediction time The target changes when Represents a neural network model.

5. The multi-channel target information automatic collection and fusion method as claimed in claim 4, characterized in that: The process of cleaning and preprocessing the collected target information is as follows: For the text target information, special characters and stop words are removed from the text, and the text is segmented into words, and the frequency of the words in the text is counted; For the image target information, remove the noise in the image, adjust the contrast and brightness of the image, and resize the image to a uniform size; For the target information of the chart, the digital and text data in the chart are extracted, the missing values ​​in the chart are filled, the abnormal values ​​in the chart are processed, and the charts in different formats are converted into a unified format; For the video target information, the video is segmented into individual frames, and the key frames in the video are extracted, and the data in the video is serialized.

6. A multi-channel target information automatic collection and fusion system, used to implement the multi-channel target information automatic collection and fusion method according to any one of claims 1 to 5, characterized in that: include, Data collection module, data processing module, information fusion module, real-time monitoring module and personalized recommendation module; The data acquisition module is used to obtain target information from different channels; The data processing module is used to clean and preprocess the acquired target information and classify the target information according to the target characteristics; The information fusion module is used to safely fuse the classified target information; The real-time monitoring module is used to monitor the updates and changes of target information in real time and to warn the target in real time; The personalized recommendation module is used to provide personalized target recommendations based on the user's historical behavior data and target information.

7. A computer device, characterized in that: include, A memory for storing instructions; The processor is used to execute the instruction so that the device implements the multi-channel target information automatic collection and fusion method as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the multi-channel target information automatic collection and fusion method according to any one of claims 1 to 5 is implemented.

Images