CN113220970A - E-commerce big data platform based on block chain - Google Patents
E-commerce big data platform based on block chain Download PDFInfo
- Publication number
- CN113220970A CN113220970A CN202110255398.3A CN202110255398A CN113220970A CN 113220970 A CN113220970 A CN 113220970A CN 202110255398 A CN202110255398 A CN 202110255398A CN 113220970 A CN113220970 A CN 113220970A
- Authority
- CN
- China
- Prior art keywords
- data
- module
- unit
- commerce
- platform
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/38—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/383—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
Abstract
The invention relates to the technical field of E-commerce big data analysis, in particular to an E-commerce big data platform based on a block chain. The E-commerce data analysis platform comprises a data feedback unit, a data feature extraction unit, an evidence extraction unit and a comparison unit. According to the invention, the evidence extraction unit is used for directly carrying out evidence collection operation on the e-commerce platform, and the customer service does not need to be contacted for feedback, and then the customer service communicates with the e-commerce platform for evidence collection, so that the working efficiency of evidence collection is greatly improved; in addition, the evaluation of the well-cut square block is sampled by utilizing the trained residual neural network, and then the model containing the attention pooling unit is used for classification, so that classification and comparison are realized through the classification of the image, and the accuracy and the comparison efficiency during comparison are improved.
Description
Technical Field
The invention relates to the technical field of E-commerce big data analysis, in particular to an E-commerce big data platform based on a block chain.
Background
With the development of online platforms, the e-commerce industry is gradually known by people, and provides convenience for customers in an online manner, specifically, the e-commerce refers to transaction activities and related service activities in an electronic transaction manner on the internet, an intranet and a value-added network, so that each link of the traditional business activities is electronized and networked.
However, many illegal merchants sell illegal commodities such as pirated books, pirated web courses and illegal commodities on the e-commerce platform, and reporting the illegal commodities requires that the illegal commodities are fed back to customer service first and then communicated with the merchants for evidence collection, so that the merchants have enough time to withdraw the sold illegal commodities, and the evidence collection efficiency is very low.
In addition, in the process of evidence collection, feedback data needs to be compared with evidence data, and general comparison directly compares the two data, but the speed of image data comparison is very slow, classification comparison cannot be performed, and the accuracy of comparison is very low.
Disclosure of Invention
The invention aims to provide a block chain-based e-commerce big data platform to solve the problems in the background art.
In order to achieve the purpose, the invention provides an e-commerce big data platform based on a block chain, which comprises an e-commerce data analysis platform, wherein the e-commerce data analysis platform comprises a data feedback unit, a data feature extraction unit, an evidence extraction unit and a comparison unit; the data feedback unit is used for receiving false data fed back to the e-commerce platform; the data feature extraction unit is used for extracting the information features of the fed back false data to form feature data; the evidence extraction unit is used for searching evidence data compared with the feedback false data according to the feedback false data and storing the evidence data in the block chain database; the comparison unit is used for comparing the characteristic data with the evidence data and generating a false evaluation report.
As a further improvement of the technical scheme, the data feature extraction unit comprises a character feature extraction module, a picture feature extraction module and an e-commerce platform feature extraction module; the character feature extraction module is used for extracting character features in the false data; the picture feature extraction module is used for extracting image features in the false data; the E-commerce platform feature extraction module is used for extracting information features of the E-commerce platform.
As a further improvement of the technical solution, the evidence extraction unit includes an information acquisition module, a search module and an interception module; the information acquisition module is used for acquiring the information characteristics extracted by the commercial platform characteristic extraction module; the searching module is used for searching the E-commerce platform according to the information characteristics and intercepting the corresponding evidence by using the intercepting module.
As a further improvement of the technical scheme, the comparison unit comprises a text data comparison module, a picture data comparison module and a summary module; the character data comparison module is used for comparing character characteristics with characters in the evidence intercepted by the interception module and outputting a comparison result; the image data comparison module is used for comparing the image characteristics with the image in the evidence intercepted by the interception module and outputting a comparison result; the summarizing module is used for summarizing and analyzing results output by the character data comparison module and the picture data comparison module and generating corresponding false evaluation reports.
As a further improvement of the technical solution, the image data comparison module further includes an image classification module for classifying the compared images.
As a further improvement of the technical solution, the image classification module includes an image classification algorithm, and the algorithm steps are as follows:
s1, dividing the image into square blocks with the side length of D, screening out the square blocks with the background range ratio exceeding a preset value, and sorting the screened square blocks into a first training set;
s2, selecting a residual error neural network trained on a natural image target detection database, modifying the output dimension of the last full-connection layer of the residual error neural network into M dimension, and adding a full-connection layer with M input dimension and N output dimension to obtain a first model;
s3, constructing a second model containing an attention pooling unit, wherein the second model comprises a feature acquisition unit, the feature acquisition unit is followed by an attention unit constructed by connecting two full-connection layers with different activation layers in parallel, and the attention unit is followed by a classification unit constructed by connecting two full-connection layers in series; the feature acquisition unit is used for acquiring features corresponding to each square block, calculating an attention coefficient by the attention unit according to the features, performing weighted summation operation on the attention coefficient and the features to obtain feature expression of the image, and classifying the feature expression by the classification unit;
it should be noted that M is 128, N is 2, the batch size when the first model is trained is 32, and the batch size when the second model is trained is 1;
s4, scoring all the squares by using the trained first model, screening the squares with the highest score and the lowest score in each image, and sorting the squares into a second training set, and training the second model by using the second training set to obtain a trained second model;
s5, dividing the images to be classified into square blocks with the side length of D, screening out square blocks with the background range exceeding a certain range, sorting the screened square blocks into a third training set, scoring all the square blocks in the third training set by adopting a trained first model, sorting the square blocks with the highest score and the square blocks with the lowest score in each image into a fourth training set, merging the third training set and the fourth training set to obtain a fifth training set, and finally classifying the fifth training set by adopting a trained second model to obtain the final classification result of the images to be classified.
As a further improvement of the technical solution, in S2, a supervised learning algorithm is used to train the first model, specifically, the supervised learning algorithm sets that all squares in a negative set are negative if at least one square in the positive set is positive.
As a further improvement of the technical solution, the training of the first model and the training of the second model both adopt a random parallel gradient descent method and an adaptive moment estimation optimizer.
Compared with the prior art, the invention has the beneficial effects that:
1. in the block chain-based E-commerce big data platform, the evidence collection operation is directly carried out on the E-commerce platform through the evidence extraction unit, the feedback of contact customer service is not needed, and then the customer service and the E-commerce platform are communicated to obtain evidence, so that the work efficiency of evidence collection is greatly improved.
2. In the block chain-based E-commerce big data platform, the evaluation of the well-cut square blocks is sampled by using the trained residual neural network, and then the model containing the attention pooling unit is used for classification, so that the accuracy of the model for classifying images is improved effectively when only a small amount of data can be used for training the model.
In addition, due to the limitation of computational resources, the existing attention pooling method can only use a two-stage model, namely, the features are pooled after the features are extracted, and the method evaluates the square blocks in advance and screens the square blocks according to scores, so that the screened square blocks can be directly used as the input of the model, so that the classification result after the attention pooling can be fed back to the feature acquisition process through the backward propagation of the gradient, the end-to-end attention pooling unit remarkably improves the image classification accuracy, and is combined with a supervised learning algorithm, namely, not only the square blocks with higher scores are screened, but also the square blocks with lower scores are screened, thus the influence of negative evidence in the image is synthesized, thereby being beneficial to improving the effect of model classification, and classification comparison is realized by utilizing the classification of the images, and the accuracy and the comparison efficiency during comparison are improved.
Drawings
FIG. 1 is a block diagram of an e-commerce data analysis platform module of the present invention;
FIG. 2 is a block diagram of a data feature extraction unit module according to the present invention;
FIG. 3 is a schematic diagram of an evidence extraction unit according to the present invention;
FIG. 4 is a block diagram of a comparison unit module according to the present invention.
The various reference numbers in the figures mean:
100. e-commerce data analysis platform;
110. a data feedback unit;
120. a data feature extraction unit; 121. a character feature extraction module; 122. a picture feature extraction module; 123. e-commerce platform feature extraction module;
130. an evidence extraction unit; 131. an information acquisition module; 132. searching a module; 133. an intercepting module;
140. a comparison unit; 141. a text data comparison module; 142. a picture data comparison module; 143. and a summarizing module.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Example 1
The invention provides an e-commerce big data platform based on a block chain, please refer to fig. 1-4, which comprises an e-commerce data analysis platform 100, wherein the e-commerce data analysis platform 100 comprises a data feedback unit 110, a data feature extraction unit 120, an evidence extraction unit 130 and a comparison unit 140; the data feedback unit 110 is used for receiving false data fed back to the e-commerce platform; the data feature extraction unit 120 is configured to extract information features of the fed back dummy data to form feature data; the evidence extraction unit 130 is used for searching evidence data compared with the fed-back dummy data according to the fed-back dummy data, and storing the evidence data in the block chain database; the comparison unit 140 is configured to compare the feature data with the evidence data, and generate a false evaluation report;
the dummy data and the evidence data include, but are not limited to, text data or picture data.
In this embodiment, the data feature extraction unit 120 includes a text feature extraction module 121, an image feature extraction module 122, and an e-commerce platform feature extraction module 123; the character feature extraction module 121 is configured to extract character features in the dummy data; the picture feature extraction module 122 is configured to extract image features in the dummy data; the e-commerce platform feature extraction module 123 is configured to extract information features of the e-commerce platform;
the extraction of the image features specifically comprises the extraction of video image features and the extraction of picture image features, and the information of the e-commerce platform specifically comprises name, contact and position information and the like of the e-commerce platform.
Further, the evidence extracting unit 130 includes an information obtaining module 131, a finding module 132, and an intercepting module 133; the information obtaining module 131 is configured to obtain the information features extracted by the provider platform feature extraction module 123; the searching module 132 is configured to search for the e-commerce platform according to the information characteristics, and intercept the corresponding evidence by using the intercepting module 133.
Specifically, the comparison unit 140 includes a text data comparison module 141, a picture data comparison module 142, and a summary module 143; the text data comparison module 141 is configured to compare text features with the text in the evidence intercepted by the interception module 133, and output a comparison result; the image data comparison module 142 is used for comparing the image features with the images in the evidence intercepted by the interception module 133 and outputting comparison results; the summarizing module 143 is configured to summarize and analyze the results output by the text data comparing module 141 and the picture data comparing module 142, and generate a corresponding false evaluation report.
The summarizing module 143, when analyzing specifically, includes the following steps:
setting a reference proportion of the result compared by the text data comparison module 141 and the picture data comparison module 142, wherein the reference proportion refers to the reference proportion of the result when the result is subjected to summary analysis;
and analyzing the output result by combining the reference proportion, and dividing the false evaluation report into three evaluation grades of non-similarity, similarity and extreme similarity, wherein the specific analysis results are non-similarity in 0-30%, similarity in 30-60% and extreme similarity in 60-100%.
The specific analysis is exemplified as follows:
if the reference proportion of the text data comparison module 141 and the picture data comparison module 142 is set to be 3: 7, the comparison result output by the text data comparison module 141 is 80%, and the result obtained by analysis is 59% when the picture data comparison module 142 is 50%, so that the results fall into similar regions, and false evaluation reports are similar, so that the evidence collection operation is directly performed on the e-commerce platform through the evidence extraction unit 130, the customer service feedback does not need to be contacted, and the customer service and the e-commerce platform are communicated for evidence collection, so that the work efficiency of evidence collection is greatly improved.
In addition, the image data comparing module 142 further includes an image classifying module for classifying the compared images.
In addition, the image classification module comprises an image classification algorithm, and the algorithm steps are as follows:
s1, dividing the image into square blocks with the side length of D, screening out the square blocks with the background range ratio exceeding a preset value, and sorting the screened square blocks into a first training set; specifically, a local recursive segmentation algorithm is adopted to calculate a threshold value of a background range and a foreground range for a square image under low resolution, the local recursive segmentation algorithm calculates a threshold value to enable the sum of variances in the foreground range and the background range to be the minimum value, then, the ratio of the background range to the whole square area is calculated under the threshold value, and if the ratio exceeds a preset value, the square is screened out;
s2, selecting a residual error neural network trained on a natural image target detection database, modifying the output dimension of the last full-connection layer of the residual error neural network into M dimension, and adding a full-connection layer with M input dimension and N output dimension to obtain a first model;
s3, constructing a second model containing an attention pooling unit, wherein the second model comprises a feature acquisition unit, the feature acquisition unit is followed by an attention unit constructed by connecting two full-connection layers with different activation layers in parallel, and the attention unit is followed by a classification unit constructed by connecting the two full-connection layers in series; the characteristic acquisition unit adopts a convolutional neural network, the basic constituent unit of the convolutional neural network is a bottleneck layer, each bottleneck layer comprises three convolutional layers, three batch regularization layers and an activation layer, the framework of the convolutional neural network comprises four parts, the first part comprises three bottleneck layers, the second part comprises four bottleneck layers, the third part comprises six bottleneck layers, and the fourth part comprises three bottleneck layers; then connecting a global pooling layer and two full-connection layers; the output dimension of the last full connection layer is M; the feature obtaining unit is used for obtaining features corresponding to each square block, calculating an attention coefficient by an attention unit according to the features, performing weighted summation operation on the attention coefficient and the features to obtain feature expression of the image, and finally classifying the feature expression by the classifying unit;
it should be noted that M is 128, N is 2, the batch size when training the first model is 32, and the batch size when training the second model is 1;
s4, scoring all the squares by using the trained first model, screening the squares with the highest score and the lowest score in each image, and sorting the squares into a second training set, and training the second model by using the second training set to obtain a trained second model;
s5, dividing the images to be classified into square blocks with the side length of D, screening out square blocks with the background range exceeding a certain range, sorting the screened square blocks into a third training set, grading all the square blocks in the third training set by adopting a trained first model, sorting the square blocks with the highest score and the square blocks with the lowest score in each image into a fourth training set, merging the third training set and the fourth training set to obtain a fifth training set, and finally classifying the fifth training set by adopting a trained second model to obtain the final classification result of the images to be classified.
Further, in S2, the first model is trained by using a supervised learning algorithm, specifically, the supervised learning algorithm sets that all squares in the negative set are negative if at least one square in the positive set is positive.
The specific training comprises the following steps:
before each training, evaluating all square blocks in a first training set by adopting a first model, screening the square block with the highest score in each image, and marking the screened square block with the same mark as the whole square block image;
all the screened square block-mark pairs are gathered to form a training set required by training, the first model is continuously trained, the operation is circulated until the accuracy of the first model on the verification set is not improved any more, and when the accuracy is calculated on the verification set, the classification result of the first model on the square block with the highest score is used as the classification result of the first model on the whole image;
when a first model is trained, inputting a square block in a training set into the first model, then obtaining an operation result by using a normalized exponential function for the output of the first model, collecting a maximum value in the operation result, obtaining a prediction result of an image by using a supervised learning algorithm according to the maximum value in the collected operation result, finally calculating the classification accuracy of the first model according to the prediction result, and if the accuracy of the first model on a verification set is not improved within a preset cycle number, stopping training and storing the model with the highest accuracy on the verification set.
When the image classification algorithm is used specifically, the trained residual neural network is used for sampling evaluation of the scribed square blocks, and then the model with the attention pooling unit is used for classification, so that the accuracy of the model for image classification is improved when only a small amount of data can be used for training the model.
In addition, due to the limitation of computing resources, the existing attention pooling method can only use a two-stage model, namely pooling the features after extracting the features, the method evaluates the square blocks in advance, and then screens the square blocks according to scores, so that the screened square blocks can be directly used as the input of the model, the classification result after the attention pooling can be fed back to the feature acquisition process through the reverse propagation of gradients, and the accuracy of image classification is remarkably improved by the end-to-end attention pooling unit.
According to the method, when the squares required by the attention pooling are screened, a supervised learning algorithm is combined, namely, the squares with higher scores are not only screened, but also the squares with lower scores are screened, so that the influence of negative evidence in the image is synthesized, the effect of model classification is favorably improved, classification comparison is realized by utilizing the classification of the image, and the accuracy and the comparison efficiency during comparison are improved.
Specifically, the training of the first model and the training of the second model both adopt a random parallel gradient descent method and an adaptive moment estimation optimizer.
The foregoing shows and describes the general principles, essential features, and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and the preferred embodiments of the present invention are described in the above embodiments and the description, and are not intended to limit the present invention. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. An e-commerce big data platform based on a block chain comprises an e-commerce data analysis platform (100), and is characterized in that: the e-commerce data analysis platform (100) comprises a data feedback unit (110), a data feature extraction unit (120), an evidence extraction unit (130) and a comparison unit (140); the data feedback unit (110) is used for receiving false data fed back to the e-commerce platform; the data feature extraction unit (120) is used for extracting the information features of the feedback false data to form feature data; the evidence extraction unit (130) is used for searching evidence data compared with the feedback spurious data according to the feedback spurious data and storing the evidence data in the block chain database; the comparison unit (140) is configured to compare the feature data with the evidence data and generate a false assessment report.
2. The blockchain-based e-commerce big data platform of claim 1, wherein: the data feature extraction unit (120) comprises a character feature extraction module (121), a picture feature extraction module (122) and an e-commerce platform feature extraction module (123); the character feature extraction module (121) is used for extracting character features in the false data; the picture feature extraction module (122) is used for extracting image features in the false data; the E-commerce platform feature extraction module (123) is used for extracting information features of the E-commerce platform.
3. The blockchain-based e-commerce big data platform of claim 1, wherein: the evidence extraction unit (130) comprises an information acquisition module (131), a searching module (132) and a intercepting module (133); the information acquisition module (131) is used for acquiring the information characteristics extracted by the commercial platform characteristic extraction module (123); the searching module (132) is used for searching the E-commerce platform according to the information characteristics and intercepting the corresponding evidence by using the intercepting module (133).
4. The blockchain-based e-commerce big data platform of claim 1, wherein: the comparison unit (140) comprises a text data comparison module (141), a picture data comparison module (142) and a summary module (143); the character data comparison module (141) is used for comparing character features with characters in the evidence intercepted by the interception module (133) and outputting a comparison result; the picture data comparison module (142) is used for comparing the image features with the images in the evidence intercepted by the interception module (133) and outputting comparison results; the summarizing module (143) is used for summarizing and analyzing results output by the text data comparison module (141) and the picture data comparison module (142) and generating corresponding false evaluation reports.
5. The blockchain-based e-commerce big data platform of claim 1, wherein: the picture data comparison module (142) further comprises an image classification module for classifying the compared images.
6. The blockchain-based e-commerce big data platform of claim 5, wherein: the image classification module comprises an image classification algorithm, and the algorithm steps are as follows:
s1, dividing the image into square blocks with the side length of D, screening out the square blocks with the background range ratio exceeding a preset value, and sorting the screened square blocks into a first training set;
s2, selecting a residual error neural network trained on a natural image target detection database, modifying the output dimension of the last full-connection layer of the residual error neural network into M dimension, and adding a full-connection layer with M input dimension and N output dimension to obtain a first model;
s3, constructing a second model containing an attention pooling unit, wherein the second model comprises a feature acquisition unit, the feature acquisition unit is followed by an attention unit constructed by connecting two full-connection layers with different activation layers in parallel, and the attention unit is followed by a classification unit constructed by connecting two full-connection layers in series; the feature acquisition unit is used for acquiring features corresponding to each square block, calculating an attention coefficient by the attention unit according to the features, performing weighted summation operation on the attention coefficient and the features to obtain feature expression of the image, and classifying the feature expression by the classification unit;
s4, scoring all the squares by using the trained first model, screening the squares with the highest score and the lowest score in each image, and sorting the squares into a second training set, and training the second model by using the second training set to obtain a trained second model;
s5, dividing the images to be classified into square blocks with the side length of D, screening out square blocks with the background range exceeding a certain range, sorting the screened square blocks into a third training set, scoring all the square blocks in the third training set by adopting a trained first model, sorting the square blocks with the highest score and the square blocks with the lowest score in each image into a fourth training set, merging the third training set and the fourth training set to obtain a fifth training set, and finally classifying the fifth training set by adopting a trained second model to obtain the final classification result of the images to be classified.
7. The blockchain-based e-commerce big data platform of claim 6, wherein: in S2, a supervised learning algorithm is used to train the first model, specifically, the supervised learning algorithm sets that all squares in a negative set are negative if at least one square in the positive set is positive.
8. The blockchain-based e-commerce big data platform of claim 6, wherein: and training the first model and the second model by adopting a random parallel gradient descent method and an adaptive moment estimation optimizer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110255398.3A CN113220970A (en) | 2021-03-09 | 2021-03-09 | E-commerce big data platform based on block chain |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110255398.3A CN113220970A (en) | 2021-03-09 | 2021-03-09 | E-commerce big data platform based on block chain |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113220970A true CN113220970A (en) | 2021-08-06 |
Family
ID=77084851
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110255398.3A Withdrawn CN113220970A (en) | 2021-03-09 | 2021-03-09 | E-commerce big data platform based on block chain |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113220970A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108694593A (en) * | 2018-05-28 | 2018-10-23 | 广州中国科学院软件应用技术研究所 | A kind of one key of commodity tracing information evidence obtaining supervisory systems and method |
| CN108921198A (en) * | 2018-06-08 | 2018-11-30 | 山东师范大学 | commodity image classification method, server and system based on deep learning |
| CN109165966A (en) * | 2018-07-13 | 2019-01-08 | 江苏恒宝智能系统技术有限公司 | A kind of electric business platform evaluation management system based on block chain |
| CN110321671A (en) * | 2019-06-26 | 2019-10-11 | 阿里巴巴集团控股有限公司 | Transaction system, method, apparatus and the electronic equipment of picture based on block chain |
| CN110704876A (en) * | 2019-09-02 | 2020-01-17 | 杭州趣链科技有限公司 | Webpage evidence obtaining and evidence fixing method based on block chain |
-
2021
- 2021-03-09 CN CN202110255398.3A patent/CN113220970A/en not_active Withdrawn
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108694593A (en) * | 2018-05-28 | 2018-10-23 | 广州中国科学院软件应用技术研究所 | A kind of one key of commodity tracing information evidence obtaining supervisory systems and method |
| CN108921198A (en) * | 2018-06-08 | 2018-11-30 | 山东师范大学 | commodity image classification method, server and system based on deep learning |
| CN109165966A (en) * | 2018-07-13 | 2019-01-08 | 江苏恒宝智能系统技术有限公司 | A kind of electric business platform evaluation management system based on block chain |
| CN110321671A (en) * | 2019-06-26 | 2019-10-11 | 阿里巴巴集团控股有限公司 | Transaction system, method, apparatus and the electronic equipment of picture based on block chain |
| CN110704876A (en) * | 2019-09-02 | 2020-01-17 | 杭州趣链科技有限公司 | Webpage evidence obtaining and evidence fixing method based on block chain |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106599155B (en) | Webpage classification method and system | |
| CN108734184B (en) | Method and device for analyzing sensitive image | |
| KR101856120B1 (en) | Discovery of merchants from images | |
| CN105653444B (en) | Software defect fault recognition method and system based on internet daily record data | |
| CN107657267B (en) | Product potential user mining method and device | |
| CN105426356A (en) | Target information identification method and apparatus | |
| CN114066242A (en) | Enterprise risk early warning method and device | |
| CN111291816A (en) | Method and device for carrying out feature processing aiming at user classification model | |
| CN110851176B (en) | Clone code detection method capable of automatically constructing and utilizing pseudo-clone corpus | |
| CN109635010B (en) | User characteristic and characteristic factor extraction and query method and system | |
| CN110414780A (en) | A kind of financial transaction negative sample generation method based on generation confrontation network | |
| CN107341508B (en) | Fast food picture identification method and system | |
| CN111242899B (en) | Image-based flaw detection method and computer-readable storage medium | |
| US20230360390A1 (en) | Transmission line defect identification method based on saliency map and semantic-embedded feature pyramid | |
| CN112258254B (en) | Internet advertisement risk monitoring method and system based on big data architecture | |
| CN109739844A (en) | Data classification method based on decaying weight | |
| CN109918648B (en) | Rumor depth detection method based on dynamic sliding window feature score | |
| CN104463601A (en) | Method for detecting users who score maliciously in online social media system | |
| CN110781333A (en) | Method for processing unstructured monitoring data of cable-stayed bridge based on machine learning | |
| CN113706291A (en) | Fraud risk prediction method, device, equipment and storage medium | |
| CN110991485B (en) | Performance evaluation method and system of target detection algorithm | |
| CN113283467A (en) | Weak supervision picture classification method based on average loss and category-by-category selection | |
| CN109409091B (en) | Method, device and equipment for detecting Web page and computer storage medium | |
| CN113220970A (en) | E-commerce big data platform based on block chain | |
| CN110808947A (en) | Automatic vulnerability quantitative evaluation method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210806 |
|
| WW01 | Invention patent application withdrawn after publication |