CN113537206A - Pushed data detection method and device, computer equipment and storage medium - Google Patents
Pushed data detection method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN113537206A CN113537206A CN202010756055.0A CN202010756055A CN113537206A CN 113537206 A CN113537206 A CN 113537206A CN 202010756055 A CN202010756055 A CN 202010756055A CN 113537206 A CN113537206 A CN 113537206A
- Authority
- CN
- China
- Prior art keywords
- detected
- target object
- text
- pushing
- influence
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 210
- 238000000034 method Methods 0.000 claims abstract description 47
- 238000013145 classification model Methods 0.000 claims abstract description 30
- 230000002441 reversible effect Effects 0.000 claims description 134
- 238000012549 training Methods 0.000 claims description 130
- 238000004422 calculation algorithm Methods 0.000 claims description 42
- 238000004364 calculation method Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 15
- 238000013527 convolutional neural network Methods 0.000 claims description 12
- 235000014101 wine Nutrition 0.000 description 23
- 238000010586 diagram Methods 0.000 description 22
- 230000006870 function Effects 0.000 description 17
- 238000013528 artificial neural network Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 12
- 230000011218 segmentation Effects 0.000 description 12
- 238000011176 pooling Methods 0.000 description 10
- 238000010801 machine learning Methods 0.000 description 9
- 238000012015 optical character recognition Methods 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 8
- 239000000284 extract Substances 0.000 description 7
- 239000000047 product Substances 0.000 description 6
- 238000013473 artificial intelligence Methods 0.000 description 5
- 230000004927 fusion Effects 0.000 description 5
- 230000007246 mechanism Effects 0.000 description 5
- 239000013598 vector Substances 0.000 description 5
- 241000282414 Homo sapiens Species 0.000 description 4
- 230000004913 activation Effects 0.000 description 4
- 238000012550 audit Methods 0.000 description 4
- 238000002372 labelling Methods 0.000 description 4
- 238000007726 management method Methods 0.000 description 4
- 238000003058 natural language processing Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 238000003709 image segmentation Methods 0.000 description 3
- 238000007781 pre-processing Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 230000002860 competitive effect Effects 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 102100032202 Cornulin Human genes 0.000 description 1
- LFQSCWFLJHTTHZ-UHFFFAOYSA-N Ethanol Chemical compound CCO LFQSCWFLJHTTHZ-UHFFFAOYSA-N 0.000 description 1
- 241000282412 Homo Species 0.000 description 1
- 101000920981 Homo sapiens Cornulin Proteins 0.000 description 1
- 241001465754 Metazoa Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000001939 inductive effect Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002829 reductive effect Effects 0.000 description 1
- 230000002787 reinforcement Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 238000013519 translation 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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
- G06N3/04—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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic 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
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/50—Network services
- H04L67/55—Push-based network services
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Databases & Information Systems (AREA)
- Image Analysis (AREA)
Abstract
The application relates to an information push detection method, an information push detection device, computer equipment and a storage medium. The method comprises the following steps: acquiring information to be detected, wherein the information to be detected comprises an image to be detected and a text to be detected; extracting a target object region in an image to be detected, identifying a target object type in the target object region, and determining a first pushing influence characteristic corresponding to information to be detected according to the target object type; extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected; inputting the text to be detected into a text classification model to be detected, classifying the text to be detected to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected; and determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic. By adopting the method, the accuracy of information push detection can be improved.
Description
Technical Field
The present application relates to the field of internet technologies, and in particular, to a method and an apparatus for detecting pushed data, a computer device, and a storage medium.
Background
With the development of internet technology, people begin to learn about information through the internet. At present, information can be conveniently and quickly pushed through an internet platform, for example, videos are pushed through a video website, live videos are pushed through a live website, news is pushed through a news website, and friend circle messages and the like are sent through an instant messaging application. At present, information is pushed through the Internet, and the Internet platform can detect the pushed information, so that the information which is forbidden to be pushed is prevented from being pushed, and the information safety is ensured. At present, the pushed information is generally detected according to expert opinions, and the pushed information is pushed when the detection is passed, however, the method for detecting through the expert opinions has the problems of low detection efficiency and low accuracy.
Disclosure of Invention
In view of the above, it is necessary to provide an information push detection method, an information push detection apparatus, a computer device, and a storage medium, which can improve the information push detection efficiency and accuracy.
An information push detection method, the method comprising:
acquiring information to be detected, wherein the information to be detected comprises an image to be detected and a text to be detected;
extracting a target object region in an image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to information to be detected according to the category of the target object;
extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected;
inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
and determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
An information push detection device, the device comprising:
the information acquisition module is used for acquiring information to be detected, and the information to be detected comprises an image to be detected and a text to be detected;
the first characteristic determining module is used for extracting a target object region in the image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object;
the second characteristic obtaining module is used for extracting an image text in the image to be detected, matching the image text through a preset keyword library and obtaining a second pushing influence characteristic corresponding to the information to be detected;
the third characteristic obtaining module is used for inputting the text to be detected into the text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
and the result determining module is used for determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:
acquiring information to be detected, wherein the information to be detected comprises an image to be detected and a text to be detected;
extracting a target object region in an image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to information to be detected according to the category of the target object;
extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected;
inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
and determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:
acquiring information to be detected, wherein the information to be detected comprises an image to be detected and a text to be detected;
extracting a target object region in an image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to information to be detected according to the category of the target object;
extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected;
inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
and determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
According to the information push detection method, the information push detection device, the computer equipment and the storage medium, the class of the target object in the target object region of the image to be detected is identified, the first push influence characteristic is determined according to the class of the target object, the second push influence characteristic is obtained through the preset keyword library according to the image text in the image to be detected, the text to be detected is classified, the third push influence characteristic is obtained according to the class of the text to be detected, the push detection result corresponding to the information to be detected is determined according to the first push influence characteristic, the second push influence characteristic and the third push influence characteristic, the corresponding push influence characteristics are obtained according to different types of data, the push detection result corresponding to the information to be detected is determined according to the different push influence characteristics, and the information push detection efficiency and accuracy are improved.
Drawings
FIG. 1 is a diagram of an application environment of a method for push detection of information in an embodiment;
FIG. 2 is a flowchart illustrating an information push detection method according to an embodiment;
FIG. 3 is a schematic flow diagram illustrating a first push impact feature in one implementation;
FIG. 4 is a flow diagram illustrating obtaining a first push impact feature in an implementation;
FIG. 5 is a schematic flow chart illustrating the detection of an advertising image of the wine in one embodiment;
FIG. 6 is a schematic flow diagram illustrating the determination of a target object class in one implementation;
FIG. 7 is a schematic diagram of an attention mechanism in one embodiment;
FIG. 8 is a schematic flow chart of an implementation of obtaining a trained target object area detection network;
FIG. 9 is a schematic flow diagram of an implementation of obtaining a trained target object recognition network;
FIG. 10 is a flow diagram illustrating a second push impact feature in one implementation;
FIG. 11 is a schematic flow chart illustrating detection of text in an image of an alcohol advertisement in one embodiment;
FIG. 12 is a schematic flow diagram of an implementation of obtaining a trained target object recognition network;
FIG. 13 is a schematic flow chart of obtaining word frequency and reverse order file frequency in one implementation;
FIG. 14 is a diagram illustrating a network structure of a text classification model to be detected in an embodiment;
FIG. 15 is a schematic flow chart of push detection results in one implementation;
FIG. 16 is a flow diagram illustrating a method for push detection in an embodiment;
FIG. 17 is a block diagram of a method for push detection in an embodiment;
fig. 18 is a schematic view of an application scenario of an information push detection method in a specific implementation;
FIG. 19 is a block diagram of an information push detection device according to an embodiment;
FIG. 20 is a diagram illustrating an internal structure of a computer device in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Computer Vision technology (CV) Computer Vision is a science for researching how to make a machine "see", and further refers to that a camera and a Computer are used to replace human eyes to perform machine Vision such as identification, tracking and measurement on a target, and further image processing is performed, so that the Computer processing becomes an image more suitable for human eyes to observe or transmitted to an instrument to detect. As a scientific discipline, computer vision research-related theories and techniques attempt to build artificial intelligence systems that can capture information from images or multidimensional data. Computer vision technologies generally include image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content/behavior recognition, three-dimensional object reconstruction, 3D technologies, virtual reality, augmented reality, synchronous positioning, map construction, and other technologies, and also include common biometric technologies such as face recognition and fingerprint recognition.
Natural Language Processing (NLP) is an important direction in the fields of computer science and artificial intelligence. It studies various theories and methods that enable efficient communication between humans and computers using natural language. Natural language processing is a science integrating linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language that people use everyday, so it is closely related to the research of linguistics. Natural language processing techniques typically include text processing, semantic understanding, machine translation, robotic question and answer, knowledge mapping, and the like.
Machine Learning (ML) is a multi-domain cross discipline, and relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The special research on how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach for computers to have intelligence, and is applied to all fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and formal education learning.
The scheme provided by the embodiment of the application relates to the technologies of artificial intelligence, such as image processing, text processing, deep learning and the like, and is specifically explained by the following embodiments:
the information push detection method provided by the application can be applied to the application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The server 104 acquires information to be detected sent by the terminal 102, wherein the information to be detected comprises an image to be detected and a text to be detected; the server 104 extracts a target object area in the image to be detected, identifies the category of a target object in the target object area, and determines a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object; the server 104 extracts an image text in the image to be detected, and matches the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected; the server 104 inputs the text to be detected into the text classification model to be detected for text classification, so as to obtain a text category to be detected, and obtains a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected; the server 104 determines a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.
In an embodiment, as shown in fig. 2, an information push detection method is provided, which is described by taking the method as an example applied to the server in fig. 1, and it is understood that the method may also be applied to a terminal, and in this embodiment, the method includes the following steps:
step 202, information to be detected is obtained, and the information to be detected comprises an image to be detected and a text to be detected.
The information to be detected refers to information to be pushed, and may be, for example, advertisements, information of an instant messaging application friend circle, news of a news platform, postings of a video website, information of a social platform, and the like, the news platform is a platform for publishing news, and the news platform may be, for example, a cybersery news platform, a hundred-degree news platform, a UC news platform, various news websites, and the like. The social platform refers to a platform for performing social contact, for example, the social platform may be a microblog application platform, a wechat application platform, a video website platform, or the like. The information to be detected is multimodal, i.e. the information to be detected comprises information of a plurality of different forms. The image to be detected refers to an image included in the information to be detected, the number of the images to be detected can be multiple, and the image to be detected can also be an image obtained from a video in the information to be detected. The text to be detected refers to the text included in the information to be detected, and the text can be a title, a file and the like in the information to be detected.
Specifically, the server obtains information to be detected, where the information to be detected may be information that a sender needs to push to each receiver, and the sender may be an object that can push information, such as a user, an enterprise, and a platform. The receiving party may be an object for which a user or the like receives information. The server can acquire the sent information to be detected from the sender terminal. The information to be detected can also be stored in a database. The server can acquire the information to be detected from a database, the database can be the database of the server itself or the database in the third-party server, and when the database is the database in the third-party server, the server is indicated to push and detect the information to be detected, which needs to be pushed by the third party. In one embodiment, when the information to be detected is acquired, the image to be detected and the text to be detected in the information to be detected are preprocessed, for example, the image to be detected is converted into an image with a fixed size, the text to be detected is subjected to word segmentation and duplicate removal, and the like. And obtaining a preset image to be detected and a preset text to be detected, and performing subsequent processing by using the preset image to be detected and the preset text to be detected.
And 204, extracting a target object area in the image to be detected, identifying the category of the target object in the target object area, and determining a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object.
The target object refers to an object in an image to be detected, the number of the target objects can be multiple, and the target object can be various objects, such as various advertisement commodities, various animals and plants, a human body and the like. The target object region is an image region in which a target object exists. The category of the target object refers to a category to which the target object belongs, for example, white spirit belongs to wines, and different objects have different categories. The category of the target object may also be a specific target object identification, which may be a name, a number, etc. The push influence characteristic refers to a characteristic influencing a push detection result, and comprises a push forward influence characteristic and a push reverse influence characteristic. The push forward influence feature is a feature that can influence the push detection result on the push. The push reverse influence feature is a feature that can influence the push detection result to the prohibition of push. The first pushing influence characteristic refers to a pushing influence characteristic obtained according to the image to be detected.
Specifically, the server may extract a target object region in the image to be detected by using an image segmentation algorithm, where the image segmentation algorithm is used to segment the image into different regions and extract a region of interest, and the image segmentation algorithm may include a threshold-based segmentation method, a region-based segmentation method, an edge-based segmentation method, a specific theory-based segmentation method, a deep learning-based segmentation method, and the like. The server identifies the class of the target object in the target object region using a deep Neural network algorithm, for example, using a CNN (Convolutional Neural network) algorithm. And comparing the category of the target object with the preset category of the pushing influence object, and obtaining a first pushing influence characteristic corresponding to the information to be detected according to the comparison result. When the category of the target object is consistent with that of a preset pushing reverse influence object, the obtained first pushing influence feature is a pushing reverse influence feature, and when the category of the target object is inconsistent with that of the preset pushing reverse influence object, the obtained first pushing influence feature is a pushing forward influence feature.
And step 206, extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected.
The image text refers to a text included in the image to be detected, and the text may be an explanation of the image to be detected. The preset keyword library refers to a database of preset keywords affecting push, and the preset keyword library may be a keyword library affecting push in a forward direction or a keyword library affecting push in a reverse direction. The second push impact feature refers to a push impact feature derived from the image text.
Specifically, the server may extract the image text in the image to be detected by using an image Character Recognition algorithm, where the image Character Recognition algorithm may be an OCR (Optical Character Recognition) algorithm, a Character positioning Recognition algorithm in the image, a deep neural network algorithm, and the like. And then segmenting the image text to obtain a segmentation result, searching the segmentation result in a preset keyword library, determining a matching result according to the searching result, and obtaining a second pushing influence characteristic corresponding to the information to be detected according to the matching result.
And 208, inputting the text to be detected into the text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected.
The classification model of the text to be detected is used for identifying the pushing influence category corresponding to the text to be detected, and the pushing influence category comprises a pushing forward influence category and a pushing reverse influence category. And obtaining a pushing forward influence characteristic according to the pushing forward influence type, and obtaining a pushing reverse influence characteristic according to the pushing reverse influence type. The text classification model to be detected is obtained by training through a deep neural network algorithm. Wherein, the deep neural network algorithm may be a CNN algorithm.
Specifically, the server inputs the text to be detected into the text classification model to be detected for text classification, namely, the text to be detected is input into an input layer of the text classification model to be detected and converted into vectors, the vectors are input into a convolution layer of the text classification model to be detected for convolution calculation, the convolution calculation results are input into a pooling layer of the text classification model to be detected for pooling, and then the pooling results are input into a classification output layer in a full-connection mode for classification and output of classification results, so that the classification of the text to be detected is obtained. And then directly obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected.
Step 210, determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
The push detection result refers to a detection result of whether the information to be detected is pushed or not.
Specifically, the server may input the first push influence feature, the second push influence feature, and the third push influence feature into the feature fusion model to identify whether pushing is performed, so as to obtain a result output by the feature fusion model, that is, obtain the feature fusion model. The feature fusion model is obtained by training according to historical data by using a machine learning algorithm, the historical data comprises historical information to be detected and corresponding historical pushing detection results, historical first pushing influence features, historical second pushing influence features and historical third pushing influence features are obtained according to the historical information to be detected, then training is carried out by using the machine learning algorithm, and when the training is finished, the feature fusion model is obtained. The machine learning algorithm may be a linear regression algorithm, a random forest algorithm, a neural network algorithm, and the like. The server can also directly obtain a push detection result corresponding to the information to be detected according to the first push influence characteristic, the second push influence characteristic and the third push influence characteristic, determine the number of push reverse influence characteristics and the number of push forward influence characteristics from the first push influence characteristic, the second push influence characteristic and the third push influence characteristic, and determine the push detection result corresponding to the information to be detected according to the number of push reverse influence characteristics and the number of push forward influence characteristics.
According to the information push detection method, the class of the target object in the target object area of the image to be detected is identified, the first push influence characteristic is determined according to the class of the target object, the second push influence characteristic is obtained through the preset keyword library according to the image text in the image to be detected, the text of the text to be detected is classified at the same time, the third push influence characteristic is obtained according to the class of the text to be detected, the push detection result corresponding to the information to be detected is determined according to the first push influence characteristic, the second push influence characteristic and the third push influence characteristic, the corresponding push influence characteristics are obtained according to different types of data, the push detection result corresponding to the information to be detected is determined according to different push influence characteristics, and the information push detection efficiency and accuracy are improved.
In one embodiment, as shown in fig. 3, step 204, extracting a target object region in an image to be detected, identifying a category of a target object in the target object region, and determining a first push influence characteristic corresponding to information to be detected according to the category of the target object includes:
step 302, inputting an image to be detected into a target object detection model, wherein the target object detection model comprises a target object area detection network and a target object identification network.
The target object detection model is used for identifying the type of a target object in an image to be detected, and is obtained by training through a deep neural network according to a training image and the type of a corresponding training object. The target object area detection network is used for identifying an interested area from an image to be detected, is a partial network in a target object detection model, and can be trained independently. The target object recognition network is used for recognizing the category of a target object in an interested area in an image to be detected, the target object recognition network can be a partial network of a target object detection model red, and the target object recognition network can also be obtained by independent training.
Specifically, the server inputs the image to be detected into a target object detection model for detection, namely, the category of the target object is obtained after passing through a target object area detection network and a target object identification network.
And 304, detecting the image to be detected according to the target object area detection network to obtain a target object area.
The target object area detection network is obtained by training a deep neural network for target detection, wherein the deep neural network for target detection may be an ssd (single Shot multi box detector) network, a cascade-RCNN (a target detection algorithm) network, a YOLO (a target detection algorithm) network, a fast RCNN (a target detection algorithm) network, or the like. The target object area is an area in which a target object exists, and the target object area may be plural or only one.
Specifically, the server inputs the image to be detected into a target object region detection network to detect the region of interest, so as to obtain a target object region. The method has the advantages that the cascade-rcnn network is used for detection, so that the positioning accuracy of the region of interest can be improved, and the performance of the whole model is improved.
Step 306, inputting the target object area into a target object identification network for identification, and obtaining the category of the target object in the target object area.
The target object recognition network is obtained by training according to a deep neural network for object recognition. The deep neural network for object recognition may be a ResNet (residual error) network, an RCNN network, a Fast RCNN network, and the like.
Specifically, the server inputs the target object area into a target object identification network for identification, and the category of the target object in the target object area is obtained. When a plurality of target object areas exist, the plurality of target object areas are all input into a target object identification network for identification, and a target object type corresponding to each target object area is obtained. In one embodiment, the target object identification network is only used for identifying the class of the target object prohibited to be pushed, so that the number of the identified classes is reduced, and the identification efficiency and accuracy are improved.
In a specific embodiment, as shown in fig. 4, a schematic diagram of a network structure for detecting an image to be detected is shown, specifically: the image to be detected comprises an image text and a region of a target object, the detected image is input into an SSD network in a target object detection model, and a region of an induction interest, namely the region of the target object, is obtained from the image to be detected. And inputting the region of the target object into a ResNet network of a target object detection model, wherein the ResNet network introduces a CBAM module to obtain the class of the target object in the input target object region. For example, as shown in fig. 5, the schematic diagram of detecting an image in an liquor advertisement is shown, where when the image to be detected in the advertisement to be detected is detected by using the target object detection model, the area where the wine bottle is located is obtained through target object area detection network detection, and the area where the wine bottle is located is identified, so that the identification result is obtained as liquor.
And 308, matching the category of the target object with a preset pushing reverse influence object category, and when the matching is consistent, obtaining a first pushing influence characteristic corresponding to the information to be detected as a first pushing reverse influence characteristic.
And 310, when the matching is not consistent, obtaining a first pushing influence characteristic corresponding to the information to be detected as a first pushing forward influence characteristic.
Wherein the preset pushing reverse influence object type refers to a preset pushing prohibition object type, for example, objects such as liquor, watches, adult products and the like are prohibited to be pushed in the advertisement image,
specifically, the server matches all the obtained target object categories with preset pushing reverse influence object categories, and when the preset pushing reverse influence object categories which are consistent in matching exist, a first pushing influence characteristic corresponding to the information to be detected is obtained and is used as a first pushing reverse influence characteristic. When the preset pushing reverse influence object types which are matched consistently do not exist, the obtained first pushing influence characteristic corresponding to the information to be detected is the first pushing forward influence characteristic, namely when only the types of all the target objects are not the pushing reverse influence object types, the obtained first pushing influence characteristic corresponding to the information to be detected is the first pushing forward influence characteristic.
In the above embodiment, the target object detection model is used to detect the type of the corresponding target object in the image to be detected, so that the accuracy and efficiency of obtaining the type of the target object are improved, then the type of the target object is matched with the preset push reverse influence object type to obtain the first push influence feature, and the accuracy and efficiency of the obtained first push influence feature are improved.
In one embodiment, as shown in fig. 6, step 306, inputting the target object area into a target object recognition network for recognition, and obtaining the category of the target object in the target object area includes:
and step 602, calculating the regional characteristics, the channel attention characteristics and the spatial attention characteristics of the target object region through the target object recognition network.
The region feature is a feature of the extracted target region, the channel attention feature is a feature obtained from a relationship between channels of the region feature, and the spatial attention feature is a feature obtained from a spatial relationship of the region feature, and is a supplement to the channel attention feature.
Specifically, the server calculates the area characteristics corresponding to the target object area by using the area parameters through the target object identification network, calculates the channel attention characteristics by using the channel attention parameters in the attention mechanism module, and calculates the spatial attention characteristics by using the spatial attention parameters in the attention mechanism module.
And step 604, combining the channel attention feature, the space attention feature and the region feature, performing convolution calculation according to the combined feature, and classifying, identifying and determining the category of the target object in the target object region according to the convolution calculation result.
The combination refers to combining the features together, the features may be directly spliced, or the combined features may be obtained by performing operations between the features, such as addition, dot product, multiplication, and the like.
Specifically, the server splices the channel attention feature, the space attention feature and the region feature to obtain spliced features, inputs the spliced features into the convolution layer to perform convolution operation, performs classification and identification through the output layer according to convolution calculation results to obtain output probability, and determines the category of the target object in the target object region according to the output probability.
In a specific embodiment, the target object recognition network includes an Attention mechanism Module, that is, a CBAM (Convolutional Block Attention Module) Module, as shown in fig. 7, which is a schematic structural diagram of an Attention mechanism, specifically: and acquiring the output feature of the last volume block as an input feature F, calculating by using the input feature F and the channel attention Mc to obtain a channel attention feature F ', and calculating by using the channel attention feature F' and the spatial attention Ms to obtain a spatial attention feature F ". Then, the output characteristic of the attention memory module is obtained according to the spatial attention characteristic F', and the output characteristic is used as the input characteristic of the next volume block to continue calculation.
In the embodiment, the expression capability of the target object region is improved by calculating the region feature, the channel attention feature and the space attention feature, and then the category of the target object is calculated by using the region feature, the channel attention feature and the space attention feature, so that the acquired category of the target object is more accurate.
In one embodiment, as shown in fig. 8, the training of the target object area detection network includes the following steps:
step 802, acquiring a training image of the marked target object area.
Specifically, the server acquires a training image with a labeled target object region, where the labeled target object region is a target object labeled in the training image, and the target object may be an object prohibited from being pushed or an object capable of being pushed. The server may collect an unlabelled image in advance, and then obtain a training image of an labeled target object region after labeling, where the server may collect the unlabelled image from the internet for labeling, may also obtain the unlabelled image from a server database for labeling, and may also obtain the training image of the labeled target object region from a third-party data platform for providing the labeled training image.
Step 804, inputting the training image with the marked target object area into an initial target object area detection network for detection, and obtaining an initial target object area.
The initial target object area detection network refers to a target object area detection network with initialized network parameters. And the initial target object area uses the initialized network parameters to extract the area to obtain the target object area.
Specifically, the server inputs the training image with the marked target object area into an initial target object area detection network for detection, that is, the training image is vectorized, and the vectorized training image is subjected to image area division by using initialized network parameters to obtain an initial target object area.
Step 806, calculating area error information of the initial target object area and the labeled target object area, and updating the network parameters in the initial target object area detection network according to the area error information.
And 808, obtaining the trained target object area detection network until the area error information obtained by training meets the preset first training completion condition.
The region error information is used for representing the error between the target object region obtained by training and the labeled target object region. The preset first training completion condition refers to a preset condition for completing the network training of the target object area detection, and may include that the area error information is smaller than a preset threshold or the training iteration number exceeds a preset number.
Specifically, the server calculates the area error information of the initial target object area and the marked target object area by using a preset loss function, wherein the preset loss function may be a cross entropy loss function. And judging whether the area error information meets a preset first training completion condition, wherein the preset first training completion condition can be whether the error information is smaller than a preset threshold value. The preset first training completion condition may be whether or not the number of training iterations exceeds a preset number. And when the regional error information does not meet the preset first training completion condition, performing back propagation by using a back propagation algorithm according to the regional error information, namely updating network parameters in the initial target object regional detection network, and performing back propagation by using a gradient descent algorithm. And obtaining the target object area detection network with updated network parameters. And taking the target object area detection network with updated network parameters as an initial target object area detection network, returning to the step 802 to continue the iterative execution until the area error information obtained by training meets the preset first training completion condition, indicating that the training of the target object area detection network is completed, and obtaining the trained target object area detection network.
In one embodiment, as shown in FIG. 9, the training of the target object recognition network includes the steps of:
step 902, acquiring a training image of the labeled target object region, extracting the labeled target object region from the training image, and acquiring a target object type label corresponding to the labeled target object region.
The target object class label refers to a real target object class.
Specifically, the server acquires a training image of the labeled target object region, extracts the labeled target object region by using the trained target object region detection network, and may also directly cut the training image of the labeled target object region to obtain the labeled target object region. Meanwhile, the target object category label corresponding to the labeled target object area is obtained, which may be set when labeling the target object area. Or the target object region can be acquired from a database which guarantees that the category label exists according to the marked target object region.
And 904, inputting the marked target object region into an initial target object recognition network for classification to obtain output training target object class probability, and obtaining a training target object class according to the training target object class probability.
The initial target object identification network refers to a target object identification network with initialized network parameters, and the training target object class probability refers to the probability of a target object class obtained through training.
Specifically, the server inputs the labeled target object region into the initial target object recognition network for classification recognition, namely, the labeled target object region is vectorized, each target object class probability is calculated according to the vectorized labeled target object region and the initialized network parameters, and the target object class corresponding to the maximum training target object class probability is obtained and serves as the training target object class.
Step 906, calculating class error information of the target object class label and the training target object class, and updating the network parameters in the initial target object identification network based on the class error information.
And 908, obtaining the trained target object recognition network until the class error information obtained by training reaches a preset second training completion condition.
The category error information refers to an error between a real target object category and a trained target object category. The preset second training completion condition is a preset condition for completing the training of the target object recognition network, and may be whether the category error information is smaller than a preset category error threshold or whether the training frequency reaches the maximum training iteration frequency.
Specifically, the server calculates the class error information of the target object class label and the training target object class by using a preset loss function, wherein the loss function can be a cross entropy loss function. The server judges whether the category error information reaches a preset second training completion condition, when the category error information does not reach the preset second training completion condition, the back propagation is carried out by using a back propagation algorithm according to the category error information, namely, the network parameters in the initial target object recognition network are updated, the target object recognition network with the updated network parameters is obtained, the target object recognition network with the updated network parameters is used as the initial target object recognition network, the step 902 is returned to continue to carry out iteration execution, and when the category error information obtained by training reaches the preset second training completion condition, the training of the target object recognition network is completed, namely, the trained target object recognition network is obtained.
In the above embodiment, the target object area detection network and the target object recognition network in the target object detection model are trained respectively, so that the target object area detection network can be optimized in a targeted manner, for example, when the target object area detection network is underperforming, the target object area detection network can be optimized independently, or when the target object recognition network is underperforming, the target object recognition network can be optimized independently, and the training accuracy is improved.
In an embodiment, as shown in fig. 10, in step 206, extracting an image text in the image to be detected, and matching the image text with a preset keyword library to obtain a second push influence feature corresponding to the information to be detected, where the second push influence feature includes:
step 1002, extracting an image text in the image to be detected through an image text recognition algorithm, and segmenting the image text to obtain each word to be detected.
Wherein the image text recognition algorithm is used for recognizing the text in the image, and the image text recognition algorithm can comprise
Specifically, the server extracts a feature sequence in the image to be detected through an image text recognition algorithm, obtains real text distribution corresponding to the feature sequence, and converts the real text distribution corresponding to the feature sequence into each word to be detected through operations such as removing and integrating. Among them, the image text recognition algorithms include a CRNN OCR (convolution cyclic neural network optical character recognition) algorithm and an attention OCR (attention optical character recognition) algorithm.
And 1004, matching each word to be detected with the pushing reverse influence keywords in the preset keyword library.
The preset keyword library is a database for storing and transmitting reverse influence keywords. The pushing of the reverse influence keyword refers to whether the information to be detected is pushed with a reverse influence, namely, the keyword is prohibited from being pushed.
Specifically, the server searches each word to be detected in the push reverse influence keywords in the preset keyword library, and when the push reverse influence keywords consistent with the word to be detected are found, a result of consistent matching is obtained. And when all the words to be detected are not searched in the preset keyword library, obtaining a result of inconsistent matching.
And 1006, when the matching is consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing reverse influence characteristic.
And step 1008, when the matching is not consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing forward influence characteristic.
The second pushing reverse influence feature means that the image text contains a corresponding pushing influence feature when the pushing of the to-be-detected word is forbidden. The second pushing positive influence characteristic refers to that the image text does not contain a pushing influence characteristic corresponding to the situation that the pushing of the words to be detected is forbidden.
Specifically, when the matching is consistent, the second pushing influence characteristic corresponding to the information to be detected is obtained as a second pushing reverse influence characteristic. And when the matching is not consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing forward influence characteristic.
In a specific embodiment, as shown in fig. 11, a frame diagram for image text recognition is shown. The wine advertisement image is recognized through an OCR text recognition algorithm to obtain each text word including a brewing engineer, national goods and boutique, then the category of each text word is recognized through a preset keyword library, the key word is obtained for pushing reverse influence by the brewing engineer, and the second pushing influence characteristic is obtained and is a second pushing reverse influence characteristic.
In the embodiment, whether the image text has the pushing reverse influence keyword is detected through the preset keyword library, so that the second pushing influence characteristic is obtained, and the efficiency of obtaining the second pushing influence characteristic is improved.
In one embodiment, as shown in fig. 12, the building of the preset keyword library includes the following steps:
step 1202, acquiring a pushing reverse influence target object identifier, and acquiring a corresponding pushing reverse influence target object text according to the pushing reverse influence target object identifier;
the pushing of the reverse-impact target object refers to whether the information to be detected is pushed to a target object with a reverse impact, namely, the target object is prohibited from being pushed, the pushing of the reverse-impact target object is used for identifying, and the pushing of the reverse-impact target object is performed. The pushing of the reverse influence target object text refers to whether the information to be detected is pushed to have reverse influence, namely the target object text which is prohibited from being pushed.
Specifically, the server obtains the pushing reverse influence target object identifier, and can acquire the corresponding pushing reverse influence target object text from the internet according to the pushing reverse influence target object identifier, and also can directly obtain the stored pushing reverse influence target object text from the database. The number of the pushing reverse influence target object identifications can be multiple, and each pushing reverse influence target object identification obtains multiple pushing reverse influence target object texts.
And 1204, performing word segmentation on the pushed reverse influence target object text to obtain each text word.
Specifically, word segmentation is a process of recombining continuous word sequences into word sequences according to a certain specification. And the server performs word segmentation on each obtained push reverse influence target object text to obtain each text word corresponding to each push reverse influence target object text.
And 1206, calculating the word frequency and the reverse file frequency corresponding to each text word, and calculating the importance degree of each text word based on the word frequency and the reverse file frequency corresponding to each text word.
The word frequency refers to the frequency of the text words appearing in the pushed reverse influence target object text. The reverse file frequency refers to the frequency with which text words appear in all pushed reverse-affected target object text.
Specifically, the server calculates the word frequency and the reverse file frequency corresponding to each text word, and then calculates the product of the word frequency and the reverse file frequency corresponding to each text word to obtain the importance degree of each text word.
And 1208, selecting the text words with the target quantity as the pushing reverse influence keywords according to the importance degrees of the text words, and storing the pushing reverse influence keywords into a keyword library to obtain a preset keyword library.
Specifically, the server sorts the text words according to the importance degrees of the text words, and sequentially selects the text words with the target number from large to small according to the sorting result as the pushing reverse influence keywords. The method comprises the steps of obtaining pushing reverse influence keywords corresponding to pushing reverse influence target object identifications, storing the pushing reverse influence keywords corresponding to each pushing reverse influence target object identification in a keyword database, and obtaining a preset keyword database.
In the embodiment, the importance degree of each text word is obtained by calculating the word frequency and the reverse file frequency corresponding to each text word, and the pushed reverse influence keywords corresponding to the pushed reverse influence target object identifier are obtained according to the importance degree of each text word, so that the pushed reverse influence keywords are more accurate. And then, a keyword library is established, so that the subsequent use is facilitated.
In one embodiment, as shown in fig. 13, in step 1206, calculating a word frequency and a reverse file frequency corresponding to each text word includes:
step 1302, counting a total number of words corresponding to the text of the pushing reverse impact target object, and counting the occurrence frequency of each text word in the text of the pushing reverse impact target object.
Step 1304, calculating the ratio of the occurrence frequency corresponding to each text word to the total number of words to obtain the word frequency corresponding to each text word.
The total number of words refers to the number of text words contained in the text of the pushing reverse impact target object.
Specifically, the server counts the total number of words of each text word corresponding to the target text with reverse influence, counts the number of occurrences of each text word in the target text with reverse influence, and calculates the ratio of the number of occurrences of each text word to the total number of words to obtain the word frequency corresponding to each text word. The word frequency corresponding to each text word can be calculated using the following formula (1).
Wherein, TFwRefers to the word frequency corresponding to the text word.
Step 1306, counting the total number of texts corresponding to the texts of the pushed reverse influence target object, and counting the number of target texts corresponding to each text word, wherein the number of target texts is the number of texts containing the target text words in the pushed reverse influence target object text, and the target text words are obtained from each text word.
Specifically, the server counts the total number of texts of the pushing reverse influence target object corresponding to the collected pushing reverse influence target object identifier. Then, each text word is used as a target text word, and the number of texts containing the target text word is calculated.
Step 1308, calculating a ratio of the target text number corresponding to each text word to the total text number, and calculating a logarithm of the ratio to obtain a reverse file frequency corresponding to each text word.
Specifically, the server calculates the ratio of the target text number corresponding to each text word to the total text number, and then calculates the logarithm of the ratio to obtain the reverse file frequency corresponding to each text word. Wherein, the inverse document frequency corresponding to the text word can be calculated by using formula (2).
The IDF refers to the inverse document frequency corresponding to the text word, and the denominator is increased by 1 to avoid the denominator being 0.
In one embodiment, the training of the text classification model to be detected comprises the following steps:
acquiring a training text to be detected and a corresponding training text category to be detected; and taking the training text to be detected as input, taking the training text category to be detected as a label, and training by using a convolutional neural network, and obtaining a text classification model to be detected when the training is finished.
The training text type to be detected refers to a real text type to be detected during training, and the text type to be detected comprises a pushing type and a pushing prohibition type. Among them, Convolutional Neural Networks (CNN) is a kind of feedforward Neural network that contains convolution calculation and has a deep structure. The convolutional neural network comprises an input layer, a convolutional layer, a pooling layer, a full connection layer and a softmax layer.
Specifically, the server acquires a training text to be detected and a corresponding training text category to be detected, wherein the training text to be detected is acquired from the internet, and the training text category to be detected is obtained after class marking is performed on the training text to be detected. And then the server takes the training text to be detected as input, takes the training text category to be detected as a label and trains the label by using a convolutional neural network, wherein an activation function used for training can be a Relu activation function, an S-shaped activation function or a tanh activation function. And when the training is finished, obtaining a text classification model to be detected, wherein the training is finished when the value of the loss function reaches a preset threshold value or the training reaches the maximum iteration times. The loss function may be a cross entropy loss function, a square error loss function, a binary loss function, a KL divergence loss function, or the like.
In a specific embodiment, as shown in fig. 14, the convolutional neural network is a schematic structural diagram, which includes an input layer, a convolutional layer, a pooling layer, a full connection layer, and a softmax layer. And performing text classification on the 'brewing teacher, national goods and competitive products', and obtaining the representation of the text by the 'brewing teacher, national goods and competitive products' through an input layer to obtain a 3x100 vector matrix. Inputting the characters of the text into a convolutional layer for convolution to obtain characters of the text, pooling the characters of the text through a pooling layer, namely extracting the characters corresponding to the maximum value from the characters to be used as a pooling result, classifying the pooling result through a full connection layer and a softmax layer to obtain the probability that the class is a push class and the probability that the push class is forbidden, obtaining the class output by training, calculating a loss value according to the class output by the training and the push class and the forbidden push class by using a loss function, updating parameters in a network according to the loss value to complete a round of training, and when the loss value is smaller than a preset threshold value, completing the training to obtain a classification model of the text to be detected.
In the embodiment, the text to be detected and the corresponding training text category to be detected are trained to obtain the text classification model to be detected, and then the text classification model to be detected can be deployed, so that the subsequent use is facilitated.
In one embodiment, step 210, determining a push detection result corresponding to information to be detected based on the first push influence feature, the second push influence feature, and the third push influence feature, includes the steps of:
and combining the first pushing influence characteristic, the second pushing influence characteristic and the third pushing influence characteristic to obtain a combined characteristic, inputting the combined characteristic into a pushing influence detection model to carry out pushing influence detection, and obtaining a pushing detection result corresponding to the information to be detected.
The combined feature refers to a feature obtained by combining the first pushing influence feature, the second pushing influence feature and the third pushing influence feature, where the combination may be splicing, vector operation, and the like, and the vector operation may be addition, number multiplication, dot product, and the like. The push influence detection model is obtained by training according to historical push influence characteristic data and historical push detection results by using a machine learning algorithm, wherein the machine learning algorithm can be a linear regression algorithm, a neural network algorithm, a random forest algorithm and the like. The push detection result refers to a detection result of whether to push the information to be detected, and includes push and prohibition of push.
Specifically, the server may directly splice the first pushing influence characteristic, the second pushing influence characteristic, and the third pushing influence characteristic to obtain a combined characteristic, and input the combined investment into a pushing influence detection model to perform pushing influence detection to obtain an output pushing detection result, that is, a pushing detection result corresponding to the information to be detected.
In the embodiment, the pushing influence detection is performed by using the pushing influence detection model, so that the pushing detection result corresponding to the information to be detected is obtained, and the accuracy of obtaining the pushing detection result is improved.
In one embodiment, the push impact feature includes a push forward impact feature and a push reverse impact feature. As shown in fig. 15, step 210, determining a push detection result corresponding to information to be detected based on the first push influence feature, the second push influence feature, and the third push influence feature, includes:
and 1502, matching the first pushing influence characteristic, the second pushing influence characteristic and the third pushing influence characteristic with the pushing reverse influence characteristic, and counting the number of the characteristics which are matched with each other.
Specifically, the server determines a push reverse influence feature among the first push influence feature, the second push influence feature, and the third push influence feature, that is, matches the push reverse influence feature with the first push influence feature, the second push influence feature, and the third push influence feature, respectively, that is, whether the push reverse influence feature is the same as the first push influence feature, the second push influence feature, and the third push influence feature. And then, counting the number of the matched features, wherein the number of the matched features is the number of the first pushing influence feature, the second pushing influence feature and the third pushing influence feature, which is consistent with the pushing reverse influence feature, for example, when the first pushing influence feature is the pushing reverse influence feature, the second pushing influence feature is the pushing reverse influence feature and the third pushing influence feature is the pushing forward influence feature, the pushing reverse influence feature is consistent with the matching of the first pushing influence feature and the second pushing influence feature, and the counted number of the features is 2.
Step 1504, when the number of features exceeds the preset number, obtaining a push detection result corresponding to the information to be detected as push prohibition, and when the number of features does not exceed the preset number, obtaining a push detection result corresponding to the information to be detected as push.
Wherein the preset number may be a preset number threshold for the existence of the push reverse impact feature. The push prohibition means that the information to be checked means information for prohibiting push. The push means that the information to be detected is information capable of being pushed.
Specifically, the feature quantity is compared with a preset quantity, when the feature quantity exceeds the preset quantity as a comparison result, a pushing detection result corresponding to the information to be detected is obtained as pushing prohibition, and when the feature quantity does not exceed the preset quantity, a pushing detection result corresponding to the information to be detected is obtained as pushing. In an embodiment, the server may further count the number of the characteristics that are not matched, when the number of the characteristics that are matched with each other is greater than the number of the characteristics that are not matched with each other, the push detection result corresponding to the information to be detected is obtained as push prohibition, and when the number of the characteristics that are matched with each other is less than the number of the characteristics that are not matched with each other, the push detection result corresponding to the information to be detected is obtained as push.
In the above embodiment, the number of the features consistent with the matching of the pushing reverse influence features is counted, and the number of the features is compared with the preset number to obtain the pushing detection result corresponding to the information to be detected, so that the efficiency of obtaining the pushing detection result is improved.
In one embodiment, after step 210, after determining a push detection result corresponding to the information to be detected based on the first push influence feature, the second push influence feature, and the third push influence feature, the method further includes the steps of:
and when the pushing detection result is pushing, acquiring each receiver address, and sending the information to be detected to the receiver terminal according to each receiver address so that the receiver terminal can display the information to be detected.
And when the pushing detection result is that pushing is forbidden, generating alarm information, acquiring the address of the sender, and returning the alarm information to the terminal of the sender corresponding to the address of the sender.
Here, the receiving side refers to an object that receives the pushed information. The receiver address is used to indicate an address of the receiver terminal, and may be an IP (internet protocol) address and a MAC (physical) address, etc. The alarm information is used for prompting that the information to be detected is information forbidden to be pushed. The sender refers to an object for pushing information to be detected. The sender address is used to indicate the address of the sender terminal, and may be an IP address, a MAC address, and the like.
Specifically, when the server obtains the detection result of the information to be detected, the server performs subsequent processing according to the detection result, that is, when the push detection result is push, the server obtains addresses of all the receiving parties, where the addresses of all the receiving parties are addresses, where the information to be detected is to be pushed, and the addresses of all the receiving parties are preset addresses where the information to be detected is to be pushed. And sending the information to be detected to the corresponding receiver terminal according to each receiver address. And when the receiving party terminal receives the information to be detected, displaying the information to be detected. And when the pushing detection result is that pushing is forbidden, generating alarm information, acquiring a sender address, returning the alarm information to a sender terminal corresponding to the sender address, and when the sender terminal receives the alarm information, re-modifying the information to be detected according to the alarm information and then pushing. In one embodiment, when the file pushing detection result is that pushing is prohibited, a prompt that illegal information exists is prompted to the management terminal, so that the management terminal further checks the information to be detected corresponding to the pushing prohibition, further detection results are obtained, and the accuracy of the detection results is improved.
In the above embodiment, when the push detection result is push, the information to be detected is pushed, and when the push detection result is push prohibition, the alarm information is generated, so that the information to be detected can be subjected to subsequent processing according to the push detection result.
In a specific embodiment, as shown in fig. 16, the information push detection method specifically includes the following steps:
step 1602, information to be detected is obtained, and the information to be detected includes an image to be detected and a text to be detected.
Step 1604, inputting the image to be detected into a target object detection model, wherein the target object detection model comprises a target object area detection network and a target object identification network, and detecting the image to be detected according to the target object area detection network to obtain a target object area.
And 1606, calculating the regional characteristics, the channel attention characteristics and the spatial attention characteristics of the target object region through the target object identification network, combining the channel attention characteristics, the spatial attention characteristics and the regional characteristics, performing convolution calculation according to the combined characteristics, and performing classification and identification according to the convolution calculation result to determine the category of the target object in the target object region.
Step 1608, matching the category of the target object with a preset pushing reverse influence object category, when the matching is consistent, obtaining that the first pushing influence feature corresponding to the information to be detected is the first pushing reverse influence feature, and when the matching is not consistent, obtaining that the first pushing influence feature corresponding to the information to be detected is the first pushing forward influence feature.
Step 1610, extracting an image text in the image to be detected through an image text recognition algorithm, and segmenting the image text to obtain each word to be detected.
Step 1612, matching each word to be detected with the push reverse influence keywords in the preset keyword library, and when the matching is consistent, obtaining a second push influence characteristic corresponding to the information to be detected as a second push reverse influence characteristic; and when the matching is not consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing forward influence characteristic.
Step 1614, inputting the text to be detected into the text classification model to be detected for text classification, obtaining a text category to be detected, obtaining a third pushing influence characteristic corresponding to the information to be detected as a third pushing forward influence characteristic when the text category to be detected is a pushing category, and obtaining a third pushing influence characteristic corresponding to the information to be detected as a third pushing reverse influence characteristic when the text category to be detected is a pushing prohibition category.
Step 1616, matching the first pushing influence characteristic, the second pushing influence characteristic and the third pushing influence characteristic with the pushing reverse influence characteristic, counting the number of the characteristics which are consistent in matching, when the number of the characteristics exceeds a preset number, obtaining a pushing detection result corresponding to the information to be detected as forbidden to be pushed, and when the number of the characteristics does not exceed the preset number, obtaining a pushing detection result corresponding to the information to be detected as pushed.
In a specific embodiment, as shown in fig. 17, a schematic diagram of a framework of an information push detection method is provided, specifically:
the method comprises the steps that a server pushes wine advertisements to user terminals, at the moment, the server obtains wine advertisement information, the wine advertisement information comprises wine advertisement title texts and wine advertisement videos, wine advertisement images are intercepted from the wine advertisement videos, image texts are obtained by the wine advertisement images through OCR text recognition, the image texts are segmented to obtain text words, whether the image texts correspond to commodities forbidden to be pushed or not is judged through a preset keyword library by using the text words, texts related to wine exist in the image texts, and a detection result that the image texts are forbidden to be pushed is obtained. And then inputting the wine advertisement image into a trained SSD network to obtain an interested region, namely a wine bottle region, inputting the wine bottle region into a trained ResNet network introduced with a CBAM module, obtaining that the type of the wine bottle of the output wine bottle region is wine, the wine is a commodity prohibited to be pushed, and obtaining a detection result prohibited to be pushed. And then inputting the wine advertisement title text into a CNN text classification network, and obtaining a result of forbidding pushing through an input layer, a convolution layer, a pooling layer, a full connection layer and a softmax layer. And finally, obtaining that the wine advertising information is the information which is forbidden to be pushed according to the result that the three detection results are forbidden to be pushed, and generating alarm information to prompt the forbidden to be pushed.
The application also provides an application scene, and the application scene applies the information push detection method. Specifically, the information push detection method is applied to the application scene as follows:
the information push detection method is applied to a news website, the news website sends a piece of news, a news website server acquires the news to be sent, the news comprises news files and news pictures, a news object area in the news pictures is extracted, the category of the news object in the news object area is identified, and a first push influence characteristic corresponding to the news is determined according to the category of the news object; extracting picture texts in news pictures, and matching the picture texts through a preset keyword library to obtain second pushing influence characteristics corresponding to news; inputting the news documents into a text classification model to be detected for text classification to obtain news document classes, and obtaining third pushing influence characteristics corresponding to news according to the news document classes; and determining a push detection result corresponding to the news based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic. When the push detection result is push, the news can be pushed to the user, that is, when the user enters the news website, the news is pushed to the user terminal, so that the user terminal displays the news on the news website page. When the news has the pictures and words with the bloody fishy violence, the pushing detection result is that pushing is forbidden, and at the moment, prompt information such as the pictures and words with the bloody fishy violence in the news is generated, and is prompted to the management terminal, and the news is modified and then pushed through the management terminal.
The application further provides an application scenario applying the information push detection method. Specifically, the information push detection method is applied to the application scene as follows:
the information push detection method is applied to a wechat application platform, and as shown in fig. 18, is a schematic diagram of a framework for push detection of the wechat application platform, specifically:
the WeChat application platform acquires information to be detected, wherein the information to be detected can be information published by WeChat public numbers, advertisements in WeChat and the like. And performing data preprocessing on the information to be detected through a data processing server, namely preprocessing a text in the information to be detected and preprocessing an image in the information to be detected. And then, detecting through an intelligent auditing system, inputting the image to be detected into a target object detection model for detection, and obtaining a first pushing influence characteristic. And using ORC to identify and obtain an image text, segmenting the image text into words to obtain each word to be detected, and matching each word to be detected with the pushing reverse influence keywords in the preset keyword library to obtain a second pushing influence characteristic. And inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected. And determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic. And when the pushing detection result is pushing, pushing processing is carried out. And when the pushing detection result is that pushing is prohibited, sending the information to be detected, which is prohibited from being pushed, to a passing auditing system for further auditing, namely carrying out manual auditing, or detecting through an inspection system. And then, storing the audit record information through the audit snapshot server. And then, the stored audit record information is used for carrying out off-line model training including an image capability model and a text capability model, and a deep learning model is obtained through training. Detection is then performed by a deep learning model. Namely, a self-learning loop is performed to detect whether the information to be detected is pushed. The stored audit record information may also be used to perform data analysis through a data analysis system, for example, to analyze reasons for prohibiting push, percentage of prohibited push, and the like.
It should be understood that, although the individual steps in the flowcharts of fig. 2, 3, 6, 8-10, 12, 13, 15 and 16 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2, 3, 6, 8-10, 12, 13, 15, and 16 may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed in turn or alternately with other steps or at least a portion of the steps or stages in other steps.
In one embodiment, as shown in fig. 19, there is provided an information push detection apparatus 1900, which may be a part of a computer device using a software module or a hardware module, or a combination of the two, and specifically includes: an information obtaining module 1902, a first feature determining module 1904, a second feature obtaining module 1906, a third feature obtaining module 1908, and a result determining module 1910, wherein:
an information obtaining module 1902, configured to obtain information to be detected, where the information to be detected includes an image to be detected and a text to be detected;
the first feature determining module 1904 is configured to extract a target object region in the image to be detected, identify a category of a target object in the target object region, and determine a first pushing influence feature corresponding to information to be detected according to the category of the target object;
a second feature obtaining module 1906, configured to extract an image text in the image to be detected, and match the image text with a preset keyword library to obtain a second pushing influence feature corresponding to the information to be detected;
a third characteristic obtaining module 1908, configured to input the text to be detected into the text classification model to be detected for text classification, so as to obtain a text category to be detected, and obtain a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
a result determining module 1910 configured to determine, based on the first pushing influence feature, the second pushing influence feature, and the third pushing influence feature, a pushing detection result corresponding to the information to be detected.
In one embodiment, the first feature determination module 1904 includes:
the image input unit is used for inputting an image to be detected into a target object detection model, and the target object detection model comprises a target object area detection network and a target object identification network;
the area obtaining unit is used for detecting the image to be detected according to the target object area detection network to obtain a target object area;
the object identification unit is used for inputting the target object area into a target object identification network for identification to obtain the category of the target object in the target object area;
the category matching unit is used for matching the category of the target object with the category of a preset pushing reverse influence object, and when the matching is consistent, a first pushing influence characteristic corresponding to the information to be detected is obtained and is used as a first pushing reverse influence characteristic; and when the matching is not consistent, obtaining a first pushing influence characteristic corresponding to the information to be detected as a first pushing forward influence characteristic.
In one embodiment, the object recognition unit is further configured to calculate a regional feature, a channel attention feature, and a spatial attention feature of the target object region through the target object recognition network; and combining the channel attention feature, the space attention feature and the region feature, performing convolution calculation according to the combined feature, and classifying, identifying and determining the category of the target object in the target object region according to the convolution calculation result.
In one embodiment, the information push detection apparatus 1900 further includes:
the detection network training module is used for acquiring a training image of the marked target object area; inputting the training image with the marked target object area into an initial target object area detection network for detection to obtain an initial target object area; calculating the area error information of the initial target object area and the marked target object area, and updating the network parameters in the initial target object area detection network according to the area error information; and obtaining the trained target object area detection network until the area error information obtained by training meets the preset first training completion condition.
In one embodiment, the information push detection apparatus 1900 further includes:
the identification network training module is used for acquiring a training image of the marked target object area, extracting the marked target object area from the training image and acquiring a target object type label corresponding to the marked target object area; inputting the marked target object region into an initial target object recognition network for classification to obtain output training target object class probability, and obtaining a training object class according to the training target object class probability; calculating category error information of the target object category label and the training target object category, and updating network parameters in the initial target object identification network based on the category error information; and obtaining the trained target object recognition network until the class error information obtained by training reaches a preset second training completion condition.
In one embodiment, the second feature derivation module 1906 includes:
the text extraction unit is used for extracting the image text in the image to be detected through an image text recognition algorithm, and segmenting the image text into words to obtain each word to be detected;
the word matching unit is used for matching each word to be detected with the pushing reverse influence keywords in the preset keyword library; when the matching is consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing reverse influence characteristic; and when the matching is not consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing forward influence characteristic.
In one embodiment, the information push detection apparatus 1900 further includes:
the text acquisition module is used for acquiring the pushing reverse influence target object identifier and acquiring a corresponding pushing reverse influence target object text according to the pushing reverse influence target object identifier;
the word segmentation module is used for segmenting the text of the pushed reverse influence target object to obtain each text word;
the first calculation module is used for calculating the word frequency and the reverse file frequency corresponding to each text word;
the second calculation module is used for calculating the importance degree of each text word based on the word frequency and the reverse file frequency corresponding to each text word;
and the selecting module is used for selecting the text words with the target quantity as the pushing reverse influence keywords according to the importance degree of each text word, and storing the pushing reverse influence keywords into the keyword library to obtain the preset keyword library.
In one embodiment, the first calculation module is further configured to count a total number of words corresponding to the pushed reverse impact target object text, and count the number of occurrences of each text word in the pushed reverse impact target object text; calculating the ratio of the occurrence frequency corresponding to each text word to the total number of the words to obtain the word frequency corresponding to each text word; counting the total number of texts corresponding to the text of the pushed reverse influence target object, and counting the number of target texts corresponding to each text word, wherein the number of the target texts is the number of texts containing the target text words in the pushed reverse influence target object text, and the target text words are obtained from each text word; and calculating the ratio of the target text number corresponding to each text word to the total text number, and calculating the logarithm of the ratio to obtain the reverse file frequency corresponding to each text word.
In one embodiment, the information push detection apparatus 1900 further includes:
the classification model training module is used for acquiring a text to be trained and a corresponding text category to be trained; and taking the training text to be detected as input, taking the training text category to be detected as a label, and training by using a convolutional neural network, and obtaining a text classification model to be detected when the training is finished.
In an embodiment, the result determining module 1910 is further configured to combine the first pushing influence feature, the second pushing influence feature, and the third pushing influence feature to obtain a combined feature, and input the combined feature into a pushing influence detection model to perform pushing influence detection, so as to obtain a pushing detection result corresponding to the information to be detected.
In one embodiment, the push impact feature comprises a push forward impact feature and a push reverse impact feature; the result determining module 1910 is further configured to match the first pushing influence feature, the second pushing influence feature, and the third pushing influence feature with the pushing reverse influence feature, and count feature numbers that are consistent in matching; and when the characteristic quantity exceeds the preset quantity, obtaining a pushing detection result corresponding to the information to be detected as forbidden pushing, and when the characteristic quantity does not exceed the preset quantity, obtaining a pushing detection result corresponding to the information to be detected as pushing.
In one embodiment, the information push detection apparatus 1900 further includes:
the pushing module is used for acquiring each receiver address when the pushing detection result is pushing, and sending the information to be detected to the receiver terminal according to each receiver address so that the receiver terminal can display the information to be detected;
and the push forbidding module is used for generating alarm information when the push detection result is that push is forbidden, acquiring the address of the sender and returning the alarm information to the sender terminal corresponding to the address of the sender.
For specific limitations of the information push detection apparatus, reference may be made to the above limitations of the information push detection method, which is not described herein again. All or part of the modules in the information push detection device can be realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 20. The computer device includes a processor, a memory, and a network interface connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used for storing information to be detected and detection result data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement an information push detection method.
Those skilled in the art will appreciate that the architecture shown in fig. 20 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is further provided, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method embodiments when executing the computer program.
In an embodiment, a computer-readable storage medium is provided, in which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.
In one embodiment, a computer program product or computer program is provided that includes computer instructions stored in a computer-readable storage medium. The computer instructions are read by a processor of a computer device from a computer-readable storage medium, and the computer instructions are executed by the processor to cause the computer device to perform the steps in the above-mentioned method embodiments.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database or other medium used in the embodiments provided herein can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (15)
1. An information push detection method, the method comprising:
acquiring information to be detected, wherein the information to be detected comprises an image to be detected and a text to be detected;
extracting a target object region in the image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object;
extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected;
inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
determining a push detection result corresponding to the information to be detected based on the first push influence characteristic, the second push influence characteristic and the third push influence characteristic.
2. The method according to claim 1, wherein the extracting a target object region in the image to be detected, identifying a category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object comprises:
inputting the image to be detected into a target object detection model, wherein the target object detection model comprises a target object area detection network and a target object identification network;
detecting the image to be detected according to the target object area detection network to obtain the target object area;
inputting the target object area into the target object identification network for identification to obtain the category of the target object in the target object area;
matching the category of the target object with a preset pushing reverse influence object category, and when the matching is consistent, obtaining a first pushing influence characteristic corresponding to the information to be detected as a first pushing reverse influence characteristic;
and when the matching is not consistent, obtaining a first pushing influence characteristic corresponding to the information to be detected as a first pushing forward influence characteristic.
3. The method according to claim 2, wherein the inputting the target object area into the target object recognition network for recognition to obtain the category of the target object in the target object area comprises:
calculating the regional characteristics, the channel attention characteristics and the spatial attention characteristics of the target object region through the target object recognition network;
and combining the channel attention feature, the space attention feature and the region feature, performing convolution calculation according to the combined feature, and classifying, identifying and determining the category of the target object in the target object region according to the convolution calculation result.
4. The method of claim 2, wherein the training of the target object area detection network comprises the steps of:
acquiring a training image of a marked target object area;
inputting the training image of the marked target object area into an initial target object area detection network for detection to obtain an initial target object area;
calculating the area error information of the initial target object area and the marked target object area, and updating the network parameters in the initial target object area detection network according to the area error information;
and obtaining the trained target object area detection network until the area error information obtained by training meets the preset first training completion condition.
5. The method of claim 2, wherein the training of the target object recognition network comprises the steps of:
acquiring a training image of a marked target object area, extracting the marked target object area from the training image, and acquiring a target object type label corresponding to the marked target object area;
inputting the marked target object region into an initial target object recognition network for classification to obtain output training target object class probability, and obtaining a training target object class according to the training target object class probability;
calculating category error information of the target object category label and the training target object category, and updating network parameters in the initial target object identification network based on the category error information;
and obtaining the trained target object recognition network until the class error information obtained by training reaches a preset second training completion condition.
6. The method according to claim 1, wherein the extracting an image text in the image to be detected, and matching the image text through a preset keyword library to obtain a second pushing influence characteristic corresponding to the information to be detected comprises:
extracting an image text in the image to be detected through an image text recognition algorithm, and segmenting the image text to obtain each word to be detected;
matching each word to be detected with the pushing reverse influence keyword in the preset keyword library;
when the matching is consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing reverse influence characteristic;
and when the matching is not consistent, obtaining a second pushing influence characteristic corresponding to the information to be detected as a second pushing forward influence characteristic.
7. The method according to claim 1, wherein the establishing of the preset keyword library comprises the following steps:
acquiring a pushing reverse influence target object identifier, and acquiring a corresponding pushing reverse influence target object text according to the pushing reverse influence target object identifier;
segmenting the text of the pushed reverse influence target object to obtain each text word;
calculating the word frequency and the reverse file frequency corresponding to each text word;
calculating the importance degree of each text word based on the word frequency and the reverse file frequency corresponding to each text word;
and selecting a target number of text words as pushing reverse influence keywords according to the importance degree of each text word, and storing the pushing reverse influence keywords into a keyword library to obtain the preset keyword library.
8. The method of claim 7, wherein said calculating a word frequency and an inverse document frequency for each text word comprises:
counting the total number of words corresponding to the text of the pushing reverse influence target object, and counting the occurrence frequency of each text word in the text of the pushing reverse influence target object;
calculating the ratio of the occurrence frequency corresponding to each text word to the total number of the words to obtain the word frequency corresponding to each text word;
counting the total number of texts corresponding to the pushed reverse influence target object texts and counting the number of target texts corresponding to each text word, wherein the number of the target texts is the number of texts containing the target text words in the pushed reverse influence target object texts, and the target text words are obtained from each text word;
and calculating the ratio of the target text number corresponding to each text word to the total text number, and calculating the logarithm of the ratio to obtain the reverse file frequency corresponding to each text word.
9. The method according to claim 1, wherein the training of the text classification model to be detected comprises the following steps:
acquiring a training text to be detected and a corresponding training text category to be detected;
and taking the training text to be detected as input, taking the class of the training text to be detected as a label, training by using a convolutional neural network, and obtaining the classification model of the training text to be detected when the training is finished.
10. The method according to claim 1, wherein determining the push detection result corresponding to the information to be detected based on the first push influence feature, the second push influence feature, and the third push influence feature comprises:
and combining the first pushing influence characteristic, the second pushing influence characteristic and the third pushing influence characteristic to obtain a combined characteristic, inputting the combined characteristic into a pushing influence detection model to detect the pushing influence, and obtaining a pushing detection result corresponding to the information to be detected.
11. The method of claim 1, wherein pushing impact features comprises pushing forward impact features and pushing reverse impact features;
determining a push detection result corresponding to the to-be-detected information based on the first push influence feature, the second push influence feature and the third push influence feature, including:
matching the first pushing influence characteristic, the second pushing influence characteristic and the third pushing influence characteristic with the pushing reverse influence characteristic, and counting the number of the characteristics which are matched with each other;
and when the characteristic quantity exceeds a preset quantity, obtaining a pushing detection result corresponding to the information to be detected as forbidden pushing, and when the characteristic quantity does not exceed the preset quantity, obtaining a pushing detection result corresponding to the information to be detected as pushing.
12. The method according to claim 1, wherein after determining the push detection result corresponding to the information to be detected based on the first push influence feature, the second push influence feature, and the third push influence feature, the method further comprises:
when the push detection result is push, acquiring each receiver address, and sending the information to be detected to a receiver terminal according to each receiver address so that the receiver terminal can display the information to be detected;
and when the pushing detection result is that pushing is forbidden, generating alarm information, acquiring a sender address, and returning the alarm information to a sender terminal corresponding to the sender address.
13. An information push detection device, the device comprising:
the information acquisition module is used for acquiring information to be detected, and the information to be detected comprises an image to be detected and a text to be detected;
the first characteristic determining module is used for extracting a target object region in the image to be detected, identifying the category of a target object in the target object region, and determining a first pushing influence characteristic corresponding to the information to be detected according to the category of the target object;
the second characteristic obtaining module is used for extracting an image text in the image to be detected, matching the image text through a preset keyword library and obtaining a second pushing influence characteristic corresponding to the information to be detected;
the third characteristic obtaining module is used for inputting the text to be detected into a text classification model to be detected for text classification to obtain a text category to be detected, and obtaining a third pushing influence characteristic corresponding to the information to be detected according to the text category to be detected;
and a result determining module, configured to determine, based on the first pushing influence feature, the second pushing influence feature, and the third pushing influence feature, a pushing detection result corresponding to the to-be-detected information.
14. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 12.
15. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010756055.0A CN113537206B (en) | 2020-07-31 | 2020-07-31 | Push data detection method, push data detection device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010756055.0A CN113537206B (en) | 2020-07-31 | 2020-07-31 | Push data detection method, push data detection device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113537206A true CN113537206A (en) | 2021-10-22 |
| CN113537206B CN113537206B (en) | 2023-11-10 |
Family
ID=78094255
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010756055.0A Active CN113537206B (en) | 2020-07-31 | 2020-07-31 | Push data detection method, push data detection device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113537206B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113779308A (en) * | 2021-11-12 | 2021-12-10 | 冠传网络科技(南京)有限公司 | Short video detection and multi-classification method, device and storage medium |
| CN114399699A (en) * | 2021-12-06 | 2022-04-26 | 北京达佳互联信息技术有限公司 | Target recommendation object determination method and device, electronic equipment and storage medium |
| CN114496256A (en) * | 2022-01-28 | 2022-05-13 | 北京百度网讯科技有限公司 | Event detection method and device, electronic equipment and storage medium |
| CN116522011A (en) * | 2023-05-16 | 2023-08-01 | 深圳九星互动科技有限公司 | Big data-based pushing method and pushing system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108038209A (en) * | 2017-12-18 | 2018-05-15 | 深圳前海微众银行股份有限公司 | Answer system of selection, device and computer-readable recording medium |
| CN110555482A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Vulgar picture identification method and device based on artificial intelligence and electronic equipment |
| WO2020000847A1 (en) * | 2018-06-25 | 2020-01-02 | 中译语通科技股份有限公司 | News big data-based method and system for monitoring and analyzing risk perception index |
| CN111080628A (en) * | 2019-12-20 | 2020-04-28 | 湖南大学 | Image tampering detection method and device, computer equipment and storage medium |
| CN111104898A (en) * | 2019-12-18 | 2020-05-05 | 武汉大学 | Image scene classification method and device based on target semantics and attention mechanism |
| US10671878B1 (en) * | 2019-01-11 | 2020-06-02 | Capital One Services, Llc | Systems and methods for text localization and recognition in an image of a document |
| CN111310026A (en) * | 2020-01-17 | 2020-06-19 | 南京邮电大学 | Artificial intelligence-based yellow-related terrorism monitoring method |
-
2020
- 2020-07-31 CN CN202010756055.0A patent/CN113537206B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108038209A (en) * | 2017-12-18 | 2018-05-15 | 深圳前海微众银行股份有限公司 | Answer system of selection, device and computer-readable recording medium |
| WO2020000847A1 (en) * | 2018-06-25 | 2020-01-02 | 中译语通科技股份有限公司 | News big data-based method and system for monitoring and analyzing risk perception index |
| US10671878B1 (en) * | 2019-01-11 | 2020-06-02 | Capital One Services, Llc | Systems and methods for text localization and recognition in an image of a document |
| CN110555482A (en) * | 2019-09-06 | 2019-12-10 | 腾讯科技（深圳）有限公司 | Vulgar picture identification method and device based on artificial intelligence and electronic equipment |
| CN111104898A (en) * | 2019-12-18 | 2020-05-05 | 武汉大学 | Image scene classification method and device based on target semantics and attention mechanism |
| CN111080628A (en) * | 2019-12-20 | 2020-04-28 | 湖南大学 | Image tampering detection method and device, computer equipment and storage medium |
| CN111310026A (en) * | 2020-01-17 | 2020-06-19 | 南京邮电大学 | Artificial intelligence-based yellow-related terrorism monitoring method |
Non-Patent Citations (4)
| Title |
|---|
| QING-HUA CHEN ET.AL.: ""The Study of ROI Detection Based on Visual Attention Mechanism"", 《2010 6TH INTERNATIONAL CONFERENCE ON WIRELESS COMMUNICATIONS NETWORKING AND MOBILE COMPUTING (WICOM)》 * |
| 胡涛: "《面向城市疾病制图分析的地址匹配研究》", 吉林大学出版社, pages: 74 * |
| 范红超;李万志;章超权;: "基于Anchor-free的交通标志检测", 地球信息科学学报, no. 01, pages 92 - 103 * |
| 谢鑫: ""基于深度学习的不良信息检测技术的研究"", 《中国优秀硕士学位论文全文数据库信息科技辑》, vol. 2020, no. 07, pages 2 - 6 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113779308A (en) * | 2021-11-12 | 2021-12-10 | 冠传网络科技(南京)有限公司 | Short video detection and multi-classification method, device and storage medium |
| CN113779308B (en) * | 2021-11-12 | 2022-02-25 | 冠传网络科技(南京)有限公司 | Short video detection and multi-classification method, device and storage medium |
| CN114399699A (en) * | 2021-12-06 | 2022-04-26 | 北京达佳互联信息技术有限公司 | Target recommendation object determination method and device, electronic equipment and storage medium |
| CN114496256A (en) * | 2022-01-28 | 2022-05-13 | 北京百度网讯科技有限公司 | Event detection method and device, electronic equipment and storage medium |
| CN116522011A (en) * | 2023-05-16 | 2023-08-01 | 深圳九星互动科技有限公司 | Big data-based pushing method and pushing system |
| CN116522011B (en) * | 2023-05-16 | 2024-02-13 | 深圳九星互动科技有限公司 | Big data-based pushing method and pushing system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113537206B (en) | 2023-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110866140B (en) | Image feature extraction model training method, image searching method and computer equipment | |
| CN110837579B (en) | Video classification method, apparatus, computer and readable storage medium | |
| CN108694225B (en) | Image searching method, feature vector generating method and device and electronic equipment | |
| CN110737801B (en) | Content classification method, apparatus, computer device, and storage medium | |
| CN107526799B (en) | Knowledge graph construction method based on deep learning | |
| CN110472090B (en) | Image retrieval method based on semantic tags, related device and storage medium | |
| CN113537206B (en) | Push data detection method, push data detection device, computer equipment and storage medium | |
| CN111582409B (en) | Training method of image tag classification network, image tag classification method and device | |
| CN111444344B (en) | Entity classification method, entity classification device, computer equipment and storage medium | |
| CN112819023B (en) | Sample set acquisition method, device, computer equipment and storage medium | |
| CN112164391A (en) | Statement processing method and device, electronic equipment and storage medium | |
| US11966829B2 (en) | Convolutional artificial neural network based recognition system in which registration, search, and reproduction of image and video are divided between and performed by mobile device and server | |
| CN110796204A (en) | Video tag determination method and device and server | |
| CN111444387A (en) | Video classification method and device, computer equipment and storage medium | |
| CN113011172B (en) | Text processing method, device, computer equipment and storage medium | |
| CN113076905B (en) | Emotion recognition method based on context interaction relation | |
| CN113536784B (en) | Text processing method, device, computer equipment and storage medium | |
| CN111523421A (en) | Multi-user behavior detection method and system based on deep learning and fusion of various interaction information | |
| CN113704623B (en) | Data recommendation method, device, equipment and storage medium | |
| CN116226785A (en) | Target object recognition method, multi-mode recognition model training method and device | |
| CN112100377A (en) | Text classification method and device, computer equipment and storage medium | |
| CN113392179A (en) | Text labeling method and device, electronic equipment and storage medium | |
| CN113641797A (en) | Data processing method, device, equipment, storage medium and computer program product | |
| CN114329004A (en) | Digital fingerprint generation method, digital fingerprint generation device, data push method, data push device and storage medium | |
| Li et al. | Social context-aware person search in videos via multi-modal cues |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 40055190 Country of ref document: HK |
|
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |