Scenario-based multi-source data fusion analysis method and system and related device

Abstract

Disclosed in embodiments of the present invention are a scenario-based multi-source data fusion analysis method and system and a related device. The method comprises: acquiring multi-source data; preprocessing the multi-source data to obtain preprocessed data; performing feature fusion on the preprocessed data and then performing normalization fusion processing to form a multi-source data feature set; and performing data type division on feature data in the multi-source data feature set, and storing the feature data in the multi-source data feature set on the basis of the data types. By means of the method, multi-source data is classified and divided, contributing to data search, improving the efficiency of data utilization.

Description

A scenario-based multi-source data fusion analysis method, system and related equipment Technical Field

[0001] The present invention relates to the technical field of multi-source data fusion, and in particular to a scenario-based multi-source data fusion analysis method, system and related equipment. Background Art

[0002] In the information age, data has become an essential component of value. However, data is often scattered and distributed across different data sources. To improve data utility, multi-source data fusion technology has emerged. Existing multi-source data fusion methods typically store all source data in the same database without categorizing each source. This results in disorganized data storage, inconvenient data retrieval, and low data utilization efficiency. Summary of the Invention

[0003] The embodiments of the present invention provide a scenario-based multi-source data fusion analysis method, system and related equipment, which help to find data and improve the efficiency of data utilization by classifying and dividing source data.

[0004] In order to achieve the above-mentioned objectives, an embodiment of the present invention provides a scenario-based multi-source data fusion analysis method, including: acquiring multi-source data; preprocessing the multi-source data to obtain preprocessed data; fusing the preprocessed data with features and performing normalization and fusion processing to form a multi-source data feature set; dividing the feature data in the multi-source data feature set into data types, and storing the feature data in the multi-source data feature set according to the data types.

[0005] Furthermore, the preprocessing of the multi-source data includes: removing device identifiers carried in the multi-source data.

[0006] Furthermore, the acquiring of multi-source data includes: receiving original data sent by a plurality of terminal devices; and filtering out target type data from the original data of each terminal device, thereby obtaining multi-source data.

[0007] Furthermore, the method further includes: classifying multiple data types according to different application scenarios.

[0008] Furthermore, the data types include audio data and video data.

[0009] An embodiment of the present invention also provides a scenario-based multi-source data fusion analysis device, including: an acquisition module for acquiring multi-source data; a preprocessing module for preprocessing the multi-source data to obtain preprocessed data; a fusion module for fusing features of the preprocessed data and performing normalized fusion processing to form a multi-source data feature set; a storage module for dividing the feature data in the multi-source data feature set into data types and storing the feature data in the multi-source data feature set according to the data types.

[0010] Furthermore, the pre-processing module is used to remove the device identification carried by the multi-source data.

[0011] Furthermore, the acquisition module is used to: receive original data sent by multiple terminal devices; and filter out target type data from the original data of each terminal device, thereby obtaining multi-source data.

[0012] Furthermore, a classification module is included for classifying various data types according to different application scenarios.

[0013] Furthermore, the data types include audio data and video data.

[0014] An embodiment of the present invention also provides a scenario-based multi-source data fusion system, including an identification module and a processing module, wherein: the identification module is used to identify the current scene and digitize it to obtain digitized scene data; the processing module is used to process the digitized scene data.

[0015] Beneficial effects: The scenario-based multi-source data fusion analysis method of the present invention includes: acquiring multi-source data, preprocessing the multi-source data to obtain preprocessed data, fusing features of the preprocessed data and performing normalized fusion processing to form a multi-source data feature set, classifying the feature data in the multi-source data feature set into types, and storing the feature data in the multi-source data feature set according to the types. By classifying and dividing the multi-source data, it is helpful to find data and improve the efficiency of data utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The technical solutions and beneficial effects of the present invention will be made apparent by describing in detail the specific embodiments of the present invention in conjunction with the accompanying drawings.

[0017] FIG1 is a flow chart of the scenario-based multi-source data fusion analysis method provided by the present invention.

[0018] FIG2 is a schematic diagram of the structure of the scenario-based multi-source data fusion analysis device provided by the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0020] In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more features. In the description of the present application, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0021] In this application, the word "exemplary" is used to mean "serving as an example, illustration, or illustration." Any embodiment described in this application as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments. The following description is given to enable any person skilled in the art to implement and use the present application. In the following description, details are listed for the purpose of explanation. It should be understood that one of ordinary skill in the art can recognize that the present application can be implemented without using these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present application with unnecessary details. Therefore, the present application is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in this application.

[0022] Please refer to the drawings, in which the same component symbols represent the same components. The principles of the present invention are illustrated by implementing them in an appropriate computing environment. The following description is based on the specific embodiments of the present invention illustrated, and should not be construed as limiting the present invention to other specific embodiments not described in detail herein.

[0023] 1 , the scenario-based multi-source data fusion analysis method according to an embodiment of the present invention includes the following steps: Step S11 : acquiring multi-source data.

[0024] Specifically, raw data sent from multiple terminal devices is received, and target type data is filtered from the raw data of each terminal device, thereby obtaining multi-source data. Multi-source data can come from multiple terminal devices, such as mobile phones, computers, and tablets. Multiple terminal devices send different raw data. By filtering this raw data, data such as SMS and MMS is filtered out, and target type data such as audio data and video data is retained. That is, the target type data of this application can be audio data and video data.

[0025] Step S12: preprocessing the multi-source data to obtain preprocessed data.

[0026] The preprocessing process specifically involves removing the device identifiers carried by multi-source data. For example, data sent by a mobile phone will carry the device identifier of the mobile phone. By removing these device identifiers, only the data itself is retained.

[0027] Step S13: The pre-processed data is subjected to feature fusion and normalization processing to form a multi-source data feature set.

[0028] Step S14: dividing the feature data in the multi-source data feature set into data types, and storing the feature data in the multi-source data feature set according to the data types.

[0029] After the features of the multi-source data sent by all terminal devices are fused, the multi-source data feature set is formed, and the feature data in the multi-source data feature set is divided into data types. For example, the data type can be audio data class, video data class and other data types, among which audio feature data is classified as audio data class, and video feature data is classified as video data class.

[0030] The feature fusion process is used to fuse the pre-processed multi-source data features together and perform normalized fusion processing to form a multi-source data feature set in a specific scenario, and then classify, aggregate and integrate the multi-source data according to the feature information to generate a feature vector, including: training a group of classifiers with training samples of each feature, and using the probability density value output by the classifier as the feature vector after feature fusion.

[0031] In one embodiment, when the features to be fused are: S = [s1, s2, s3, s4.....sn] and T = [t1, t2, t3, t4....tm], the feature fusion process includes: using K features of the feature training samples to train n classifiers k: k1, k2....kn; calculating xi = ki(si), i = 1.2.3....n.

[0032] Use T features of the feature training samples to train m classifiers g: g1, g2...gn, and calculate yj=gj(tj), j=1.2.3....m.

[0033] Classifier k and classifier g are the same model, for example, they can both be Bayesian or SVM, and the output of classifier k and classifier g is the probability of a certain category, instead of setting the output of the classifier as the category, and using the obtained probability density value as the feature vector after feature fusion: C = [x1.x2.x3.....xn.y1.y2.y3......ym].

[0034] After obtaining the multi-source data feature set, it also includes: building an algorithm library of commonly used data fusion algorithms to support video text data analysis and image data analysis, as well as fusion analysis of structured data and unstructured data from multiple sources. Since the multi-source data has been previously subjected to feature extraction and division, the extracted feature information is already a sufficient representation or sufficient statistic of the original information, and the multi-source data is classified, aggregated and integrated according to the feature information to generate feature vectors. Therefore, some feature-level fusion methods are mainly used here to fuse these feature vectors and make attribute descriptions based on the fused feature vectors.

[0035] This fusion process is implemented using the SdstoosT algorithm. SdstoosT is an iterative algorithm whose core concept is to train different classifiers (weak classifiers) on the same training set and then combine these weak classifiers to form a stronger final classifier (strong classifier). The algorithm itself works by altering the data distribution. It weights each sample based on the correct classification of each sample in each training set and the overall classification accuracy of the previous training set. The new dataset with modified weights is then fed to the lower-level classifiers for training. Finally, the classifiers obtained from each training step are combined to form the final decision classifier. Using the sdstoosT classifier eliminates unnecessary training data features and focuses on the key training data. Currently, research and applications of the sdstoosT algorithm have primarily focused on classification problems, although some applications have also emerged in recent years for regression problems. The sdstoosT family of algorithms primarily addresses two-class problems, multi-class single-label problems, multi-class multi-label problems, large-class single-label problems, and regression problems. It uses all training samples for learning.

[0036] The principle of the SdstoosT algorithm is: input: training data set χ = ​​{(x1, y1), (x2, y2), ..., (xn, yn)}; output: the final strong classifier G(x).

[0037] Specifically, it may include: initializing the weight distribution of the training samples χ={(x1, y1), (x2, y2), ..., (xn, yn)} of the feature set.

[0038] Specifically, if there are N samples, each training sample point is initially assigned the same weight w = 1 / N.

[0039] The weight D1 of the training sample of each feature set in each iteration is: D1 = (w11, w12…w1i…, w1N), and the training sample set of the feature set with weight distribution Dm (m = 1, 2, 3…N) is used for learning to obtain a weak classifier; during the training process of the weak classifier, for the training samples of the feature set that have been accurately classified, the weight of the training samples combined with this feature is reduced when constructing the next training sample set; for the training samples of the feature set that have not been accurately classified, the weight of the training samples combined with this feature is increased when constructing the next training sample set; the training sample set after weight update is used to train the next classifier, and the entire training process is carried out iteratively; at the same time, the corresponding voice of the weak classifier is obtained according to the sum of the weights corresponding to the training samples x of the feature set that are misclassified by the weak classifier.

[0040] In the embodiment of the present invention, by classifying and dividing the feature data in the multi-source data feature set, corresponding data can be quickly found only by category when searching for data, thereby improving the efficiency of data utilization.

[0041] Step S15: Classify the various data types according to different application scenarios.

[0042] Application scenarios include advertising, legal education, job hunting, and communications. For example, if the feature data in the audio data class can be applied to advertising scenarios, the audio data class will be classified as advertising data. In practical applications, for example, if it is an outdoor LED display, advertising data can be quickly obtained based on the advertising category, and if it is an indoor TV, legal education data can be quickly obtained based on the legal education category, further improving data utilization efficiency.

[0043] 2 , the present invention further provides a multi-source data fusion device, comprising: an acquisition module 21 , a pre-processing module 22 , a fusion module 23 , a storage module 24 and a classification module 25 .

[0044] Among them, the acquisition module 21 is used to acquire multi-source data. The preprocessing module 22 is used to preprocess the multi-source data to obtain preprocessed data, wherein the preprocessing process is specifically used to remove the device identifier carried by the multi-source data. The fusion module 23 is used to perform feature fusion and normalization fusion processing on the preprocessed data to form a multi-source data feature set. The storage module 24 is used to classify the feature data in the multi-source data feature set into data types and store the feature data in the multi-source data feature set according to the data type. The data types include audio data class and video data class.

[0045] Furthermore, the acquisition module 21 is specifically configured to receive original data sent by a plurality of terminal devices; filter out target type data from the original data of each terminal device, thereby obtaining multi-source data.

[0046] The classification module 25 is used to classify various data types according to different application scenarios.

[0047] An embodiment of the present invention also provides a scenario-based multi-source data fusion system, including an identification module and a processing module, wherein: the identification module is used to identify the current scene and digitize it to obtain digitized scene data; the processing module is used to process the digitized scene data.

[0048] In the embodiment of the present invention, by classifying and dividing the feature data in the multi-source data feature set, corresponding data can be quickly found only by category when searching for data, thereby improving the efficiency of data utilization.

[0049] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0050] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

[0051] In specific implementation, the above units or structures can be implemented independently or in any combination, and can be implemented as one or several. The specific implementation of the above units or structures can refer to the previous method embodiments and will not be repeated here.

[0052] The specific implementation of the above operations can be found in the previous embodiments and will not be repeated here.

[0053] The above description is only an implementation method of the present application and does not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the description and drawings of this application, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A scenario-based multi-source data fusion analysis method, characterized in that: include: Acquire data from multiple sources; Preprocessing the multi-source data to obtain preprocessed data; After the preprocessed data is subjected to feature fusion and normalization, a multi-source data feature set is formed; The feature data in the multi-source data feature set are divided into data types, and the feature data in the multi-source data feature set are stored according to the data types.

2. The scenario-based multi-source data fusion analysis method according to claim 1 is characterized in that: The preprocessing of the multi-source data includes: removing device identifiers carried by the multi-source data.

3. The scenario-based multi-source data fusion analysis method according to claim 1 is characterized in that: The acquiring of multi-source data includes: receiving original data sent by a plurality of terminal devices; and filtering out data of a target type from the original data of each terminal device, thereby obtaining multi-source data.

4. The scenario-based multi-source data fusion analysis method according to claim 1 is characterized in that: The method also includes: classifying multiple data types according to different application scenarios.

5. The scenario-based multi-source data fusion analysis method according to claim 4 is characterized in that: The data types include audio data types and video data types.

6. A scenario-based multi-source data fusion analysis device, characterized in that: include: Acquisition module, used to acquire multi-source data; A preprocessing module, used for preprocessing the multi-source data to obtain preprocessed data; The fusion module is used to perform feature fusion and normalization processing on the pre-processed data to form a multi-source data feature set; The storage module is used to classify the feature data in the multi-source data feature set into data types, and store the feature data in the multi-source data feature set according to the data types.

7. The scenario-based multi-source data fusion analysis device according to claim 6 is characterized in that: The preprocessing module is used to remove the device identification carried by the multi-source data.

8. The scenario-based multi-source data fusion analysis device according to claim 6 is characterized in that: The acquisition module is used to: receive original data sent by multiple terminal devices; and filter out target type data from the original data of each terminal device, thereby obtaining multi-source data.

9. The scenario-based multi-source data fusion analysis device according to claim 6, characterized in that: It also includes a classification module for classifying various data types according to different application scenarios, and the data types include audio data class and video data class.

10. A scenario-based multi-source data fusion system, characterized in that: It includes an identification module and a processing module, wherein: The recognition module is used to recognize the current scene and digitize it, thereby obtaining digitized scene data; The processing module is used to process the digitized scene data.

Images