CN114595382A - Privacy decision recommendation system based on image data and deep learning - Google Patents
Privacy decision recommendation system based on image data and deep learning Download PDFInfo
- Publication number
- CN114595382A CN114595382A CN202210164785.0A CN202210164785A CN114595382A CN 114595382 A CN114595382 A CN 114595382A CN 202210164785 A CN202210164785 A CN 202210164785A CN 114595382 A CN114595382 A CN 114595382A
- Authority
- CN
- China
- Prior art keywords
- privacy
- image
- decision
- semantic
- deep learning
- 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.)
- Pending
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 48
- 238000013136 deep learning model Methods 0.000 claims abstract description 35
- 238000012545 processing Methods 0.000 claims abstract description 35
- 238000005516 engineering process Methods 0.000 claims abstract description 11
- 238000004364 calculation method Methods 0.000 claims description 36
- 238000000034 method Methods 0.000 claims description 31
- 239000011159 matrix material Substances 0.000 claims description 16
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 238000003066 decision tree Methods 0.000 claims description 10
- 238000000605 extraction Methods 0.000 claims description 9
- 238000002372 labelling Methods 0.000 claims description 9
- 238000012216 screening Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 8
- 238000013475 authorization Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000007781 pre-processing Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 5
- 230000006399 behavior Effects 0.000 claims description 4
- 230000009193 crawling Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 claims description 4
- 238000013075 data extraction Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 230000006378 damage Effects 0.000 description 2
- 208000003443 Unconsciousness Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 210000004072 lung Anatomy 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
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
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- 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
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5846—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using extracted text
-
- 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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Library & Information Science (AREA)
- Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mathematical Physics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a privacy decision recommendation system based on image data and deep learning. Model library: the deep learning model is used for learning a large number of social privacy samples in advance through a deep learning technology and generating a plurality of different privacy level judgments; and a semantic recognition module: the system comprises a display unit, a processing unit and a display unit, wherein the display unit is used for converting surface information of an image in image data into semantic information when receiving an instruction sent by the image data; a privacy identification module: the system is used for importing the semantic information into the model base and judging the privacy level of the image data; a decision module: for issuing a decision to the user feedback image after the privacy level is determined; wherein the image issuance decision comprises: instant issue decisions, inhibit issue decisions, and partial issue decisions.
Description
Technical Field
The invention relates to the technical field of privacy processing, in particular to a privacy decision recommendation system based on image data and deep learning.
Background
Currently, with the development of social networks and smart phone technologies, people can conveniently share images taken at will in the social networks. However, such sharing, if not controlled or directed, may cause various privacy leaks, serious or even personal injuries or infringements, various cyber crime or fraud. As indicated by relevant research studies, many teenagers prefer to share or dazzle their own, friends or classmates of private photos on social networks, and a great lack of knowledge in doing so can cause various privacy leaks, and in some cases even lifelong negative effects or injuries. It is also reported that 78% of network-related perpetrators use social networking media to locate victims. The perpetrator can inquire or deduce various information of the victim according to the social network picture (and the related auxiliary information). The current social network sites do not provide an effective privacy decision recommendation tool for common users (especially teenagers) to use, and remind the users to make decisions for use by the users or a social network access control system when the users publish social network images. Therefore, there is a need to develop a privacy decision recommendation method and system for social network digital image sharing.
In the prior art, privacy protection of images in a social network is mainly achieved through user-defined access control authority, anonymization processing, image encryption, burning after reading and the like. However, since the mobile terminal supports real-time sharing of instant shooting and instant transmission, a user is likely to be aware of the deficiency of the permission setting after uploading an image, or the user carelessly transmits an image with private information due to insanity, and at this time, the image is uploaded, which brings a privacy disclosure risk that is difficult to eliminate in a large environment of a social network.
Disclosure of Invention
The invention provides a privacy decision recommendation system based on image data and deep learning, which is used for solving the problem that a user is likely to realize the permission setting after uploading an image, or the user carelessly sends the image with private information due to insanity consciousness, and the image is uploaded at the moment, so that the privacy leakage risk which is difficult to eliminate is brought in the large environment of a social network. The case (1).
A privacy decision recommendation system based on image data and deep learning, comprising:
model library: the deep learning model is used for learning a large number of social privacy samples in advance through a deep learning technology and generating a plurality of different privacy level judgments;
a semantic recognition module: the system comprises a display unit, a processing unit and a display unit, wherein the display unit is used for converting surface information of an image in image data into semantic information when receiving an instruction sent by the image data;
a privacy identification module: the system is used for importing the semantic information into the model base and judging the privacy level of the image data;
a decision module: for issuing a decision to the user feedback image after the privacy level is determined; wherein,
the image issuance decision comprises: instant issue decisions, inhibit issue decisions, and partial issue decisions.
As an embodiment of the present invention: the system further comprises:
a privacy determination module: the setting standard used for obtaining the setting of different privacy levels set by the user according to the historical privacy setting information of the user; wherein
The setting criteria include: the method comprises the steps of obtaining picture content standards, social type standards and picture storage position standards;
a classifier building module: establishing a picture classifier based on picture information extraction according to the set standard; wherein,
the picture information includes: the image surface information classifier, the social type classifier and the picture storage classifier are used for classifying the image surface information;
the picture classifier includes: an information content classifier and a picture type classifier;
a sample screening module: screening historical privacy data of the user according to the picture classifier, and determining a social privacy sample;
a grading module: the privacy awareness and trust calculation method is used for carrying out privacy awareness calculation and privacy trust calculation on different social privacy samples, dividing privacy grades according to calculation results and determining the social privacy samples with different social privacy grades; wherein,
the social privacy sample comprises: low risk privacy samples, medium risk privacy samples, and high risk privacy samples.
As an embodiment of the present invention: the ranking module comprises:
a data extraction unit: the system comprises a social program used for determining the social program existing on a user terminal, and establishing an API crawling interface according to the social program to acquire user data; wherein,
the user data is image data generated in a social process;
a data preprocessing unit: the system is used for preprocessing the user data and determining semantic content of the user data; wherein,
the pretreatment comprises the following steps: image-text conversion processing, word segmentation processing, privacy labeling processing sum TF-IDF processing and image named entity classification processing;
a calculation unit: carrying out privacy consciousness calculation on the semantic contents to determine a privacy consciousness value, and carrying out privacy trust calculation on the semantic contents to determine a privacy trust value;
a grade division unit: establishing a binary decision tree according to the privacy trust value and the privacy consciousness value, determining the sensitive values of different image data, and dividing privacy grades according to the sensitive values; wherein,
the privacy classes include: low risk privacy, medium risk privacy, and high risk privacy;
the sensitive value has a corresponding sensitive value threshold according to the privacy level.
As an embodiment of the invention: the model library comprises:
deep learning model building unit: presetting a deep learning model based on different privacy levels; wherein,
the deep learning module comprises: a low risk privacy model, a medium risk privacy model, and a high risk privacy model;
deep learning training unit: the system is used for importing the social privacy samples into a deep learning model correspondingly, carrying out data iterative computation on the social privacy samples, and determining optimal parameters of the model after the iterative computation;
a model library unit: and the deep learning model library is used for generating deep learning according to the optimal parameters of the model.
As an embodiment of the present invention: the semantic recognition module comprises:
an instruction recognition unit: the system is used for carrying out privacy monitoring on social software of the user terminal and judging whether an image sending instruction exists or not;
a feedback unit: the image processing device is used for generating a feedback signal when an image sending instruction exists;
labeling unit: the image processing device is used for determining the image address of the image to be sent when the feedback signal is sensed, and performing semantic annotation on the surface information of the image according to the image address;
a word segmentation processing unit: the semantic annotation is used for segmenting words of the surface information of the image according to the semantic annotation to generate a vocabulary set;
a semantic conversion unit: and the semantic conversion module is used for performing semantic conversion on each image according to the vocabulary set to generate a semantic text of each image.
As an embodiment of the present invention: the privacy identification module includes:
a matching unit: the system is used for performing steady-state matching of a deep learning model according to the semantic information and determining a unique deep learning model corresponding to the semantic information; wherein,
the steady state matching comprises: performing steady state calculation on the semantic information and different deep learning models to determine steady state degree, and determining a unique depth recognition model according to the steady state degree;
an importing unit: the semantic information is imported into the unique deep learning model, and a privacy grade value is determined;
a rank determination unit: and the privacy level determining unit is used for determining the corresponding privacy level according to the privacy level value.
As an embodiment of the invention: the decision module comprises:
discrete attribute determination unit: the system is used for matching the corresponding discrete model according to the privacy level of the image data and determining a discrete value;
a continuous attribute determination unit: the device is used for matching the corresponding continuous model according to the privacy level of the image data and determining a continuous value;
a data set generation unit: generating a corresponding semantic generalization data set according to the discrete value and the continuous value;
adaptively allocating privacy budget units: performing privacy budgeting according to the generalized data set, constructing a privacy decision tree, and determining images which can be sent and images which cannot be sent in the image data;
a subdivision scheme selection unit: according to the privacy decision tree, carrying out privacy decision subdivision to generate an image sending scheme;
a decision unit: and sending the image sending scheme to a user terminal interface according to the image sending scheme, and acquiring a selection instruction.
As an embodiment of the present invention: the instant issuance decision further comprises: acquiring information of an image receiving end, judging whether the receiving end has privacy acquisition authorization, and executing an instant privacy decision when the image receiving end has the privacy acquisition authorization;
the prohibiting issuing a decision further comprises: recording an issuing instruction according to the risk level of the image data, generating an information mail with a decision forbidden issuing function, and sending the information mail to a user mailbox;
the partial issuance decision further comprises: and evaluating the risk level of the image in the image data, determining the image with low risk level according to the evaluation result, sending the image with low risk level, and storing the image with medium risk level and the image with high risk level to the corresponding space to be sent.
As an embodiment of the present invention, the system further includes:
a decision matrix unit: the decision matrix generating method comprises the steps of obtaining all decisions sent by an image, and establishing a hesitation fuzzy decision matrix based on the attribute of each decision;
a decision-making integrated value calculating unit: the comprehensive decision value of each decision behavior is determined according to the fuzzy decision matrix;
a sorting unit: the average weighting calculation is carried out on the comprehensive decision value to generate a decision sequence;
a decision scheme determination unit: and determining an optimal decision scheme according to the decision sequence.
As an embodiment of the present invention: the system further comprises:
privacy label setting module: a user submits a picture containing a protected object and a semantic label set correspondingly;
a tag image matching module: the method comprises the steps of identifying a picture to obtain semantic content of a protected object in the picture;
storing the recognized semantic content and the correspondingly set semantic tags into a semantic sample library together, and setting semantic information extraction models of different pictures based on the semantic sample library;
a rule setting module: selecting a corresponding semantic sample set from the semantic information extraction model by a user, and setting the privacy level of each semantic sample;
a privacy level matching module: and the user stores the new picture to a rule setting module for image recognition to obtain the privacy level of the new picture, and performs semantic sample calibration in the rule setting module according to the rule setting module of the new picture.
The invention has the beneficial effects that:
the privacy level of the image data is judged, the judgment result is filtered, and the image data with the top level is pushed to the user; performing semantic analysis on the image data, and combining the image data to be protected with corresponding semantic information; a large number of samples are used for learning the privacy data in the social network in early learning through a deep learning technology, and the reliability of image privacy decision is increased. The method is based on human-computer interaction, the deterministic text privacy rule is extracted, and based on the extracted deterministic text privacy rule, the privacy semantic relation is analyzed layer by layer, so that the quick matching of the complex image privacy relation is realized. The invention not only solves the problem of difficulty in automatic understanding and matching of the privacy semantics of the complex digital image. The invention provides a multi-source social network privacy decision-making model based on multi-mode deep learning, and solves the problem of difficulty in image privacy and semantic joint analysis in a complex and diversified social environment. The invention provides and designs a privacy protection method based on a random corresponding mechanism, and solves the problem of risk leakage of user training data privacy of a deep learning privacy decision model.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a system diagram of a privacy decision recommendation system based on image data and deep learning according to an embodiment of the present invention;
FIG. 2 is a sample processing diagram of a privacy decision recommendation system based on image data and deep learning according to an embodiment of the present invention;
fig. 3 is a new image calibration diagram of a privacy decision recommendation system based on image data and deep learning in an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described in conjunction with the accompanying drawings, and it will be understood that they are described herein for the purpose of illustration and explanation and not limitation.
According to the scheme, the privacy grade is automatically calculated through a deep learning method, and corresponding decision recommendation is carried out according to the calculation result of the privacy grade, so that the privacy grade of the image information subjected to the decision recommendation is higher.
Example 1:
as shown in fig. 1, the privacy decision recommendation system based on image data and deep learning of the present invention includes:
model library: the deep learning model is used for learning a large number of social privacy samples in advance through a deep learning technology and generating a plurality of different privacy level judgments;
a semantic recognition module: the system comprises a display unit, a processing unit and a display unit, wherein the display unit is used for converting surface information of an image in image data into semantic information when receiving an instruction sent by the image data;
a privacy identification module: the system is used for importing the semantic information into the model base and judging the privacy level of the image data;
a decision module: for issuing a decision to the user feedback image after the privacy level is determined; wherein,
the image issuance decision comprises: instant issue decisions, inhibit issue decisions, and partial issue decisions.
The principle of the technical scheme is as follows:
compared with the prior art, the privacy decision mode of the invention mainly differs in semantic conversion and deep model establishment, and the prior art has a single model, but the invention is mainly innovated in that a plurality of deep learning models without registration judgment are established, such as: a low-level privacy decision recommendation model, a medium-level privacy decision recommendation model and a high-level privacy decision recommendation model; different images are determined to correspond to different privacy levels in a semantic recognition mode through multiple models, different privacy levels are determined, different images can be sent out somewhat and cannot be sent out when being sent out, and the problem that parts of the images are sent out still exists when multiple images are sent out. In the prior art, a social contact mode mostly adopts a manual mode of a user to judge the privacy level of image data, and privacy information is leaked when the privacy protection capability of the user is not strong or the user is unconscious; the protection against image privacy only supports image surface information and is not combined with semantic information. The privacy samples are less learned in the early stage, so the reliability in the decision recommendation process is lower. The method comprises the steps of judging the privacy level of the image data, filtering the judgment result, and pushing the image data with the top level to a user; performing semantic analysis on the image data, and combining the image data to be protected with corresponding semantic information; a large number of samples are used for learning the privacy data in the social network in early learning through a deep learning technology, and the reliability of image privacy decision is increased. According to the method, the deterministic text privacy rule is extracted through human-computer interaction, and the privacy semantic relation is analyzed layer by layer on the basis of the deterministic text privacy rule, so that the quick matching of the complex image privacy relation is realized. The invention provides an image source-only social network decision model based on image-text matching deep learning, and solves the problems of automatic understanding and difficult matching of privacy semantics of complex digital images. The invention provides a multi-source social network privacy decision-making model based on multi-mode deep learning, and solves the problem of difficulty in image privacy and semantic joint analysis in a complex and diversified social environment. The invention provides and designs a privacy protection method based on a random corresponding mechanism, and solves the problem of risk leakage of user training data privacy of a deep learning privacy decision model.
The invention has the beneficial effects that:
the privacy level of the image data is judged, the judgment result is filtered, and the image data with the top level is pushed to the user; performing semantic analysis on the image data, and combining the image data to be protected with corresponding semantic information; a large number of samples are used for learning the privacy data in the social network in early learning through a deep learning technology, and the reliability of image privacy decision is increased. The method is based on human-computer interaction, the deterministic text privacy rule is extracted, and based on the extracted deterministic text privacy rule, the privacy semantic relation is analyzed layer by layer, so that the quick matching of the complex image privacy relation is realized. The invention not only solves the problem of difficulty in automatic understanding and matching of the privacy semantics of the complex digital image. The invention provides a multi-source social network privacy decision-making model based on multi-mode deep learning, and solves the problem of difficulty in image privacy and semantic joint analysis in a complex and diversified social environment. The invention provides and designs a privacy protection method based on a random corresponding mechanism, and solves the problem of risk leakage of user training data privacy of a deep learning privacy decision model.
Example 2:
as an embodiment of the present invention: as shown in fig. 2, the system further comprises:
a privacy determination module: the setting standard used for obtaining the setting of different privacy levels set by the user according to the historical privacy setting information of the user; wherein
The setting criteria include: the method comprises the steps of obtaining picture content standards, social type standards and picture storage position standards;
the historical privacy setting information of the user is information set by pre-acquiring the historical privacy of the user, and then judging the type and standard of the set privacy information; for example: when a user uses a life photo of the user as privacy, if the life photo of the user is related to the life photo of the user in the picture content, the life information is the standard of the picture content, the life information can be 2-level privacy, in addition, some pictures can be the work content of the user, the work of the user can be 1-level privacy content, the social type standard can be different social channels such as WeChat and relatives, the social channel can be 2-level privacy of the social type, encrypted places can exist in a mobile phone of the user, and the encrypted places can belong to 3-level privacy of the user;
a classifier building module: establishing a picture classifier based on picture information extraction according to the set standard; wherein,
the picture information includes: an image surface information classifier, a social type classifier and a picture storage classifier;
the image classifier for extracting the image information is also a classifier for dividing the information through the image; the image surface information classifier classifies information according to picture surface display elements. For example: the pictures are used for working in a working place, and the classification types are used for working; and when the picture surface display element is a forest, the classified type may be landscape. The social type classifier determines what type of social channel is according to the elements on the picture.
The picture classifier includes: an information content classifier and a picture type classifier;
a sample screening module: screening historical privacy data of the user according to the picture classifier, and determining a social privacy sample;
a grading module: the privacy awareness and trust calculation method is used for carrying out privacy awareness calculation and privacy trust calculation on different social privacy samples, dividing privacy grades according to calculation results and determining the social privacy samples with different social privacy grades; wherein,
the social privacy sample comprises: low risk privacy samples, medium risk privacy samples, and high risk privacy samples. The privacy consciousness calculation is that a user sets privacy characteristics according to a judgment standard for privacy, a screening model is established through the privacy characteristics, and privacy consciousness is determined through the screening model judgment; and then, calculating the trust degree through establishing a privacy entropy calculation and then calculating the trust degree through the privacy entropy value.
The principle of the technical scheme is as follows: in the prior art, the problem is that when the privacy samples are obtained, the sample data are few, so the scheme adopted by the invention is that the privacy data of different levels are divided in a classifier mode, classified storage is carried out after division, and all the privacy samples can be determined by calculating privacy consciousness and privacy trust based on user habits for the classified and stored different social privacy samples.
Example 3:
as an embodiment of the present invention: the ranking module comprises:
a data extraction unit: the system comprises a social program used for determining the social program existing on a user terminal, and establishing an API crawling interface according to the social program to acquire user data; wherein,
the user data is image data generated in a social process;
the API crawling interface is a predefined full-period interface and is used for comprehensively acquiring all data in user terminal equipment, so that all data in a mobile phone terminal of a user can be comprehensively detected.
A data preprocessing unit: the system is used for preprocessing the user data and determining semantic content of the user data; wherein,
the pretreatment comprises the following steps: the image-text conversion processing (namely converting the image content into characters), the word segmentation processing (if the characters exist in the image or the text, the characters are segmented and are divided into a plurality of technical nouns for processing), the privacy labeling processing (namely labeling the privacy content in the image), the TF-IDF processing (belonging to the text mining technology and being the privacy of five users in a user terminal) and the image named entity classification processing (namely after the image has specific classification, the image is classified by specific classification names);
a calculation unit: carrying out privacy consciousness calculation on the semantic contents to determine a privacy consciousness value, and carrying out privacy trust calculation on the semantic contents to determine a privacy trust value; this is different from the privacy awareness calculation of the privacy sample in embodiment 3, which is to calculate the privacy content for each privacy content, and further calculate the privacy trust value according to the privacy content. The privacy trust value is whether the private content is trustworthy or not, and whether the private content is truly private or not. The privacy awareness value is a value obtained by calculating the image content, and determining whether or not the image content is private, what the privacy level is, and what the privacy level value is the privacy awareness value.
A grade division unit: establishing a binary decision tree according to the privacy trust value and the privacy consciousness value, determining the sensitive values of different image data, and dividing privacy grades according to the sensitive values; the sensitive value is also a privacy value of the privacy content for the user, that is, a value with a low degree of privacy.
The privacy classes include: low risk privacy, medium risk privacy, and high risk privacy;
the sensitive value has a corresponding sensitive value threshold according to the privacy level.
According to the invention, the deep learning models with different privacy levels are required to be established, so that a plurality of samples are required to be obtained, the image data is obtained through establishing the API interface, the image data is processed, the privacy data is divided according to different privacy levels based on the processing result, and the privacy division is based on the sensitive value which is mainly divided artificially. The invention can not only analyze single or multiple images of the social network to obtain individual privacy, but also analyze single or multiple images (and environment information such as related texts) published in forums, social networks or other environments of the internet to obtain information in various fields such as national relevant politics, economy, military and the like, and the invention only establishes a model by an API interface.
In a particular embodiment, the method further comprises: determining a value of privacy awareness is calculated by:
step 1: judging whether the semantic content conforms to the privacy content or not according to the semantic content:
wherein, P (P)i) Represents a privacy value; q represents a history privacy type determination coefficient; w represents a history privacy content determination coefficient; k is a radical of formulaiA type feature representing the ith image data; p is a radical ofiSemantic features representing the ith image data; p (Q | k)i) The judging function is used for judging whether the type characteristics of the ith image data accord with the privacy type judging coefficient or not and determining a privacy type value; p (W | P)i) The device is used for judging whether the type characteristics of the ith image data accord with the privacy content judgment coefficient or not and determining a privacy content value; judging whether the content is the private content or not through the sum of the types of the content and the privacy; only in the case of both or only one match, but the sum must be greater than 1, otherwise neither is privacy worthy. i ∈ n, n representing the number of private data.
Step 2, content matching is carried out, and a privacy consciousness value is calculated:
wherein, X represents a correlation coefficient, and when X is 1, f (j) represents an image feature of the privacy image of the jth privacy class, which belongs to an image feature trained in advance for comparison.
When the technical scheme is put and the privacy consciousness is worth, a matching algorithm is carried out, then the technical scheme that the image characteristics of the privacy image are obtained through calculation of the matching algorithm and different levels of history is different, when the two are different, the correlation value is judged based on the inertia formula, the corresponding privacy consciousness value is determined according to the correlation value, and the user is more cautious when sending the image through the privacy consciousness value.
Example 4:
as an embodiment of the present invention: the model library comprises:
deep learning model building unit: presetting a deep learning model based on different privacy levels; wherein,
the deep learning module comprises: a low risk privacy model, a medium risk privacy model, and a high risk privacy model;
deep learning training unit: the system is used for importing the social privacy samples into a deep learning model correspondingly, carrying out data iterative computation on the social privacy samples, and determining optimal parameters of the model after the iterative computation;
the most important parameter of the model is that the privacy data is determined to be a reference parameter of low risk privacy, medium risk privacy or high risk privacy, the reference parameter is an excellent reference parameter, a threshold value can be obtained in actual calculation, and different threshold values exist in the low risk privacy, the medium risk privacy or the high risk privacy. The maximum value among the threshold values is the optimal model parameter;
a model library unit: and the deep learning model library is used for generating deep learning according to the optimal parameters of the model.
The principle of the technical scheme is as follows:
according to the invention, a plurality of different deep learning model libraries are required to be arranged, and the model libraries can not only judge the privacy grade, but also continuously optimize the identification of the privacy data in an iterative calculation manner, so that the most appropriate privacy data can be obtained as far as possible. At present, various privacy security threats may be brought by image sharing of a user in a social network, but an effective decision support tool cannot be provided by an existing social network application system, and many defects exist in the aspect of privacy decision aiming at image sharing of the social network in the existing theoretical model and method research. Aiming at the application requirements and the current situation of theoretical research, the invention comprehensively operates methods and means such as human-computer interaction, deep learning, differential privacy and the like, researches and designs a privacy decision model in network image sharing, explores and solves related scientific problems, and the lung and innovates the content and the method of multimedia information security research, thereby providing core algorithm and technical support for the privacy security application of social network image tray information. Therefore, the invention establishes a plurality of different deep learning models. At the present stage, some disadvantages exist, such as that most of the methods are based on manual marking, not based on automatic recognition of image semantics, so that the burden of users is increased, and many users are unwilling or even unable to perform manual labeling, so that the method is ineffective. If the user cannot complete effective manual marking under the drunk condition, part of old people or teenagers cannot complete accurate marking. Meanwhile, the privacy level analysis can be performed by using the image information, such as "the more people in the image, the higher the privacy level", and "the brighter the image, the lower the privacy level", but these rules have no universality.
Based on the consideration, the social network image privacy decision model is constructed by adopting the current rapidly developed deep learning technology, and the more accurate and effective privacy decision model is designed by utilizing the advantages of the deep learning technology in the aspects of image privacy semantic understanding and matching, so that the trouble of the problems is solved.
Example 5:
as an embodiment of the invention: the semantic recognition module comprises:
an instruction recognition unit: the system is used for carrying out privacy monitoring on social software of the user terminal and judging whether an image sending instruction exists or not;
because the invention mainly detects the image data, the image data in the user terminal can be monitored, and the image is mainly monitored and sent from the user terminal.
A feedback unit: the image processing device is used for generating a feedback signal when an image sending instruction exists; the feedback instruction refers to the feedback after the image instruction is generated, but the feedback instruction is not set in the prior art, and whether the image is sent or not can be judged only through real-time monitoring.
Labeling unit: the image processing device is used for determining the image address of the image to be sent when the feedback signal is sensed, and performing semantic annotation on the surface information of the image according to the image address;
the semantic annotation of the image surface information is carried out through the image address, namely, the type, the content and the specific image information of the image are annotated through the image address, and the annotation refers to the annotation and the type annotation of a specific content of the image.
A word segmentation processing unit: the semantic annotation processing module is used for segmenting words of the surface information of the image according to the semantic annotation to generate a vocabulary set; the word segmentation set has the main function of segmenting the key information of the image information.
A semantic conversion unit: and the semantic conversion module is used for performing semantic conversion on each image according to the vocabulary set to generate a semantic text of each image. The semantic text is an image specification for each image, and is a specific content description and type description for the image.
The beneficial effects of the above technical scheme are that: the method has the main capability of realizing semantic recognition and semantic labeling, realizes the assistance on deep learning by the way, can convert each image into semantic texts, realizes more accurate training by the semantic texts, and can also realize more accurate risk grade judgment.
Example 6:
as an embodiment of the present invention: the privacy identification module comprises:
a matching unit: the system is used for performing steady-state matching of a deep learning model according to the semantic information and determining a unique deep learning model corresponding to the semantic information; wherein,
the steady state matching comprises: performing steady state calculation on the semantic information and different deep learning models to determine steady state degree, and determining a unique depth recognition model according to the steady state degree;
the steady-state matching is that deep learning model recognition is carried out on the images, when the privacy level is determined, only one level can be recognized, one image cannot be recognized, two different levels can be recognized, and when the privacy level is ensured to be recognized, the only one privacy level can be provided.
An importing unit: the semantic information is imported into the unique deep learning model, and a privacy grade value is determined;
a rank determination unit: and the privacy level determining unit is used for determining the corresponding privacy level according to the privacy level value.
In the invention, when privacy identification is carried out, the main capability of the model is stable, because only the model is stable, the identification of the model to the depth information can be realized more quickly and directly, and in addition, different data can be subjected to risk grade evaluation through different deep learning models when image data are identified, so that the risk grade can be subdivided.
Example 7:
as an embodiment of the present invention: the decision module comprises:
discrete attribute determination unit: the system is used for matching the corresponding discrete model according to the privacy level of the image data and determining a discrete value; the discrete model is to represent the overall discreteness of the image data. The discrete type is mainly a discrete type which embodies different contents of image data, the image data comprises a plurality of images which comprise different privacy levels, different contents and different image types, and the discrete model is mainly used for embodying the discrete type.
A continuous attribute determination unit: the device is used for matching the corresponding continuous model according to the privacy level of the image data and determining a continuous value; the invention matches continuous model according to privacy level, the continuous model is in a privacy level, the model of image continuity and the image correlation have a certain degree of correlation, but the correlation is not complete, and is for example: the first image is that a person faces red light, the second image is that the person rushes the red light, the continuity is achieved, the person takes time as a reference axis, and the events are before and after the reference axis.
A data set generation unit: generating a corresponding semantic generalization data set according to the discrete value and the continuous value; semantically generalizing the data set; the semanteme generalization data set is that after the discrete value and the continuous value are determined, each discrete value and each continuous value can determine similar or same-class image data, and the semanteme set of the similar or same-class image data can be increased by determining the semanteme of the image data of the similar or same-class image data, so that the semanteme generalization data set can construct a privacy decision tree, has more semanteme samples when carrying out privacy budget, and has more detailed decision when carrying out privacy decision.
Adaptively allocating privacy budget units: performing privacy budgeting according to the generalized data set, constructing a privacy decision tree, and determining images which can be sent and images which cannot be sent in the image data;
a subdivision scheme selection unit: according to the privacy decision tree, carrying out privacy decision subdivision to generate an image sending scheme;
a decision unit: and sending the image sending scheme to a user terminal interface according to the image sending scheme, and acquiring a selection instruction. The image issuing scheme is an instruction which can be issued by the image and can not be issued by the image, and is a dialogue and a strategy for issuing the image.
When the decision is established, although the decision only has three modes of sending out the image, not sending out the image and sending out the partial image, if the decision is multiple images, the attribute and the continuity of each image need to be judged, and finally, the images are determined to be sent out through a subdivision scheme to generate the corresponding sending decision.
Example 8:
as an embodiment of the present invention: the instant issuance decision further comprises: acquiring information of an image receiving end, judging whether the receiving end has privacy acquisition authorization, and executing an instant privacy decision when the image receiving end has the privacy acquisition authorization; that is, only the receiving end of the image can receive the image when the receiving end has the authorization of the privacy data of the user, otherwise, the receiving end cannot receive the image with the privacy data.
The prohibiting issuing a decision further comprises: recording an issuing instruction according to the risk level of the image data, generating an information mail with a decision forbidden issuing function, and sending the information mail to a user mailbox; when the decision for prohibiting sending is made, people can invade and privacy is peeped, so that when the decision for prohibiting sending is made, the invention can record through the information mail, and can enable a user to monitor whether the terminal equipment is invaded or not.
The partial issuance decision further comprises: and evaluating the risk level of the images in the image data, determining the images with low risk level according to the evaluation result, sending the images with low risk level, and storing the images with medium risk level and the images with high risk level to the corresponding space to be sent. The space to be issued is a preparation space, and there is an issuing delay, and in the delay, the user can judge that the medium-risk and high-risk images cannot be issued.
As an embodiment of the present invention, the system further includes:
a decision matrix unit: the decision matrix generating method comprises the steps of obtaining all decisions sent by an image, and establishing a hesitation fuzzy decision matrix based on the attribute of each decision; the hesitation fuzzy decision matrix is used for judging decision attributes, a detection matrix with inaccurate decision possibly exists, various decision schemes exist during fuzzy decision, namely the decision-based attributes are determined by a trust value, if the privacy levels of a receiver and a sent image are matched, the trust value is high, and if the privacy levels of the receiver and the sent image are not matched, namely the authority of the receiver for receiving the privacy data is not enough, the trust value is low. The fuzzy decision matrix is formed by fusing trust values of a fuzzy algorithm and a decision scheme.
A decision-making integrated value calculating unit: the comprehensive decision value of each decision behavior is determined according to the fuzzy decision matrix; the comprehensive decision value is determined by the fuzzy decision value of each decision scheme obtained by the fuzzy decision matrix, the number of images sent by the decision behaviors and the privacy grade, the comprehensive decision coefficients of the images with different privacy grades are different, and the comprehensive decision value is obtained by multiplying the privacy images with different grades by the coefficients, then multiplying by the number of the images and dividing by the fuzzy decision value.
A sorting unit: the average weighting calculation is carried out on the comprehensive decision value to generate a decision sequence; the average weighting calculation is an existing calculation mode, and is mainly used for enabling different comprehensive decision values to have a large difference, so that an optimal decision scheme can be selected.
A decision scheme determination unit: and determining an optimal decision scheme according to the decision sequence.
When the invention executes the decision, for any image, there are issued, not issued and partially issued decisions, but the priority of each decision is different, so the invention actually determines which images can be issued and which images cannot be issued in the form of priority, and generates a decision sequence, and determines how to give the highest priority to the scheme through the decision sequence.
As an embodiment of the invention: as shown in fig. 3, the system further comprises:
privacy label setting module: a user submits a picture containing a protected object and a semantic label set correspondingly; the semantic tag is a tag of a picture of an object to be protected, and is used for explaining picture information of the protected object.
A tag image matching module: the method comprises the steps of identifying a picture to obtain semantic content of a protected object in the picture;
storing the recognized semantic content and the correspondingly set semantic tags into a semantic sample library together, and setting semantic information extraction models of different pictures based on the semantic sample library;
a rule setting module: selecting a corresponding semantic sample set from the semantic information extraction model by a user, and setting the privacy level of each semantic sample;
a privacy level matching module: and the user stores the new picture to a rule setting module for image recognition to obtain the privacy level of the new picture, and performs semantic sample calibration in the rule setting module according to the rule setting module of the new picture.
In the method, because the privacy is judged based on the actual condition of the user, new pictures may be added to the terminal equipment of the user, and certainly, a new data set is generated in the data set subjected to model training, so that for the new pictures, the method can perform calibration again in the semantic sample according to the label to set the privacy level for the new pictures.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A privacy decision recommendation system based on image data and deep learning, comprising:
model library: the deep learning model is used for learning a large number of social privacy samples in advance through a deep learning technology and generating a plurality of different privacy level judgments;
and a semantic recognition module: the system comprises a display unit, a processing unit and a display unit, wherein the display unit is used for converting surface information of an image in image data into semantic information when receiving an instruction sent by the image data;
a privacy identification module: the system is used for importing the semantic information into the model base and judging the privacy level of the image data;
a privacy identification module: for issuing a decision to the user feedback image after the privacy level is determined; wherein,
the image issuance decision comprises: instant issue decisions, inhibit issue decisions, and partial issue decisions.
2. The privacy decision recommendation system based on image data and deep learning of claim 1, further comprising:
a privacy determination module: the setting standard used for obtaining the setting of different privacy levels set by the user according to the historical privacy setting information of the user; wherein
The setting criteria include: the method comprises the steps of obtaining picture content standards, social type standards and picture storage position standards;
a classifier building module: establishing a picture classifier based on picture information extraction according to the set standard; wherein,
the picture information includes: an image surface information classifier, a social type classifier and a picture storage classifier;
the picture classifier includes: an information content classifier and a picture type classifier;
a sample screening module: screening historical privacy data of the user according to the picture classifier, and determining a social privacy sample;
a grading module: the privacy awareness and trust calculation method is used for carrying out privacy awareness calculation and privacy trust calculation on different social privacy samples, dividing privacy grades according to calculation results and determining the social privacy samples with different social privacy grades; wherein,
the social privacy sample comprises: low risk privacy samples, medium risk privacy samples, and high risk privacy samples.
3. The privacy decision recommendation system based on image data and deep learning of claim 2, wherein the ranking module comprises:
a data extraction unit: the system comprises a social program used for determining the social program existing on a user terminal, and establishing an API crawling interface according to the social program to acquire user data; wherein,
the user data is image data generated in a social process;
a data preprocessing unit: the system is used for preprocessing the user data and determining semantic content of the user data; wherein,
the pretreatment comprises the following steps: image-text conversion processing, word segmentation processing, privacy labeling processing sum TF-IDF processing and image named entity classification processing;
a calculation unit: carrying out privacy consciousness calculation on the semantic contents to determine a privacy consciousness value, and carrying out privacy trust calculation on the semantic contents to determine a privacy trust value;
a grade division unit: establishing a binary decision tree according to the privacy trust value and the privacy consciousness value, determining the sensitive values of different image data, and dividing privacy grades according to the sensitive values; wherein,
the privacy classes include: low risk privacy, medium risk privacy, and high risk privacy;
the sensitive value has a corresponding sensitive value threshold according to the privacy level.
4. The privacy decision recommendation system based on image data and deep learning according to claim 1, wherein the model library comprises:
deep learning model building unit: presetting a deep learning model based on different privacy levels; wherein,
the deep learning module comprises: a low risk privacy model, a medium risk privacy model, and a high risk privacy model;
deep learning training unit: the system is used for importing the social privacy samples into a deep learning model correspondingly, carrying out data iterative computation on the social privacy samples, and determining optimal parameters of the model after the iterative computation;
a model library unit: and the deep learning model library is used for generating deep learning according to the optimal parameters of the model.
5. The privacy decision recommendation system based on image data and deep learning according to claim 1, wherein the semantic recognition module comprises:
an instruction recognition unit: the system is used for carrying out privacy monitoring on social software of the user terminal and judging whether an image sending instruction exists or not;
a feedback unit: the image processing device is used for generating a feedback signal when an image sending instruction exists;
labeling unit: the image processing device is used for determining the image address of the image to be sent when the feedback signal is sensed, and performing semantic annotation on the surface information of the image according to the image address;
a word segmentation processing unit: the semantic annotation processing module is used for segmenting words of the surface information of the image according to the semantic annotation to generate a vocabulary set;
a semantic conversion unit: and the semantic conversion module is used for performing semantic conversion on each image according to the vocabulary set to generate a semantic text of each image.
6. The privacy decision recommendation system based on image data and deep learning of claim 1, wherein the privacy identification module comprises:
a matching unit: the deep learning model is used for performing steady-state matching on the deep learning model according to the semantic information and determining a unique deep learning model corresponding to the semantic information; wherein,
the steady state matching comprises: performing steady state calculation on the semantic information and different deep learning models to determine steady state degree, and determining a unique depth recognition model according to the steady state degree;
an importing unit: the semantic information is imported into the unique deep learning model, and a privacy grade value is determined;
a rank determination unit: and the privacy level determining module is used for determining the corresponding privacy level according to the privacy level value.
7. The privacy decision recommendation system based on image data and deep learning according to claim 1, wherein the decision module comprises:
discrete attribute determination unit: the system is used for matching the corresponding discrete model according to the privacy level of the image data and determining a discrete value;
a continuous attribute determination unit: the device is used for matching the corresponding continuous model according to the privacy level of the image data and determining a continuous value;
a data set generation unit: generating a corresponding semantic generalization data set according to the discrete value and the continuous value;
adaptively allocating privacy budget units: carrying out privacy budgeting according to the generalized data set, constructing a privacy decision tree, and determining images which can be sent out and images which can not be sent out in the image data;
a subdivision scheme selection unit: according to the privacy decision tree, carrying out privacy decision subdivision to generate an image sending scheme;
a decision unit: and sending the image sending scheme to a user terminal interface according to the image sending scheme, and acquiring a selection instruction.
8. The privacy decision recommendation system based on image data and deep learning of claim 1, wherein:
the instant issuance decision further comprises: acquiring information of an image receiving end, judging whether the receiving end has privacy acquisition authorization, and executing an instant privacy decision when the image receiving end has the privacy acquisition authorization;
the prohibiting issuing a decision further comprises: recording an issuing instruction according to the risk level of the image data, generating an information mail with a decision forbidden issuing function, and sending the information mail to a user mailbox;
the partial issuance decision further comprises: and evaluating the risk level of the image in the image data, determining the image with low risk level according to the evaluation result, sending the image with low risk level, and storing the image with medium risk level and the image with high risk level to the corresponding space to be sent.
9. The privacy decision recommendation system based on image data and deep learning of claim 1, further comprising:
a decision matrix unit: the decision matrix generating method comprises the steps of obtaining all decisions sent by an image, and establishing a hesitation fuzzy decision matrix based on the attribute of each decision;
a decision-making integrated value calculating unit: the comprehensive decision value of each decision behavior is determined according to the fuzzy decision matrix;
a sorting unit: the average weighting calculation is carried out on the comprehensive decision value to generate a decision sequence;
a decision scheme determination unit: and determining an optimal decision scheme according to the decision sequence.
10. The privacy decision recommendation system based on image data and deep learning of claim 1, further comprising:
privacy label setting module: a user submits a picture containing a protected object and a semantic label set correspondingly;
a tag image matching module: the method comprises the steps of identifying a picture to obtain semantic content of a protected object in the picture;
storing the recognized semantic content and the correspondingly set semantic tags into a semantic sample library together, and setting semantic information extraction models of different pictures based on the semantic sample library;
a rule setting module: selecting a corresponding semantic sample set from the semantic information extraction model by a user, and setting the privacy level of each semantic sample;
a privacy level matching module: and the user stores the new picture to a rule setting module for image recognition to obtain the privacy level of the new picture, and performs semantic sample calibration in the rule setting module according to the rule setting module of the new picture.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210164785.0A CN114595382A (en) | 2022-02-23 | 2022-02-23 | Privacy decision recommendation system based on image data and deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210164785.0A CN114595382A (en) | 2022-02-23 | 2022-02-23 | Privacy decision recommendation system based on image data and deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114595382A true CN114595382A (en) | 2022-06-07 |
Family
ID=81806836
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210164785.0A Pending CN114595382A (en) | 2022-02-23 | 2022-02-23 | Privacy decision recommendation system based on image data and deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114595382A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115396374A (en) * | 2022-08-12 | 2022-11-25 | 徐州恒佳电子科技有限公司 | Intelligent routing system special for priority data forwarding and method thereof |
| CN116956347A (en) * | 2023-07-28 | 2023-10-27 | 浙江大学 | Interactive micro data release system under privacy protection |
-
2022
- 2022-02-23 CN CN202210164785.0A patent/CN114595382A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115396374A (en) * | 2022-08-12 | 2022-11-25 | 徐州恒佳电子科技有限公司 | Intelligent routing system special for priority data forwarding and method thereof |
| CN115396374B (en) * | 2022-08-12 | 2023-12-22 | 徐州恒佳电子科技有限公司 | Routing system and method special for intelligent priority data forwarding |
| CN116956347A (en) * | 2023-07-28 | 2023-10-27 | 浙江大学 | Interactive micro data release system under privacy protection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Parra-Arnau et al. | Measuring the privacy of user profiles in personalized information systems | |
| KR101699653B1 (en) | Identifying and preventing leaks of sensitive information | |
| CN114595382A (en) | Privacy decision recommendation system based on image data and deep learning | |
| CA2504680A1 (en) | A method and system for managing confidential information | |
| CN116934285B (en) | Visual intelligent system and equipment for realizing digitization and entity file management | |
| CN109583228B (en) | Privacy information management method, device and system | |
| CN113313170A (en) | Full-time global training big data platform based on artificial intelligence | |
| Boehm et al. | Introducing iyouit | |
| CN109450882A (en) | A kind of security management and control system and method for the internet behavior merging artificial intelligence and big data | |
| CN115632839B (en) | Intelligent campus environment network supervision method and system | |
| Geradts | Digital, big data and computational forensics | |
| Horsman | The COLLECTORS ranking scale for ‘at‐scene’digital device triage | |
| Kumar et al. | Automatic detection of fake profiles in online social networks | |
| Laestadius et al. | Instagram revisited | |
| Wanjau et al. | Network intrusion detection systems: A systematic literature review of hybrid deep learning approaches | |
| CN112287209A (en) | Intelligent recommendation method and system for thousands of people and thousands of faces of portals | |
| US20200218804A1 (en) | System and method for detecting leaked documents on a computer network | |
| Koch et al. | Recognizability bias in citizen science photographs | |
| Canelón et al. | Unstructured data for cybersecurity and internal control | |
| Teymourlouei et al. | Dark data: managing cybersecurity challenges and generating benefits | |
| Garchery | User-centered intrusion detection using heterogeneous data | |
| Ramya et al. | SVM Based Fake Account Sign-In Detection | |
| Sreenivasulu et al. | Image Based Classification of Rumor Information from the Social Network Platform. | |
| Gioti | Advancements in Open Source Intelligence (OSINT) Techniques and the role of artificial intelligence in Cyber Threat Intelligence (CTI) | |
| Ji et al. | Adaptive Sensitive Information Recognition Based on Multimodal Information Inference in Social Networks |
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 |