CN114092106A - Method and device for identifying information false degree - Google Patents
Method and device for identifying information false degree Download PDFInfo
- Publication number
- CN114092106A CN114092106A CN202110141190.9A CN202110141190A CN114092106A CN 114092106 A CN114092106 A CN 114092106A CN 202110141190 A CN202110141190 A CN 202110141190A CN 114092106 A CN114092106 A CN 114092106A
- Authority
- CN
- China
- Prior art keywords
- user
- product
- node
- comment
- nodes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 71
- 230000002159 abnormal effect Effects 0.000 claims abstract description 68
- 238000012549 training Methods 0.000 claims abstract description 63
- 238000003062 neural network model Methods 0.000 claims abstract description 50
- 238000004364 calculation method Methods 0.000 claims abstract description 24
- 239000013598 vector Substances 0.000 claims description 40
- 239000013604 expression vector Substances 0.000 claims description 38
- 238000001914 filtration Methods 0.000 claims description 34
- 230000002776 aggregation Effects 0.000 claims description 27
- 238000004220 aggregation Methods 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 10
- 238000012552 review Methods 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 230000007246 mechanism Effects 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 238000001514 detection method Methods 0.000 abstract description 34
- 230000006870 function Effects 0.000 description 26
- 238000010586 diagram Methods 0.000 description 15
- 238000004891 communication Methods 0.000 description 6
- 239000011159 matrix material Substances 0.000 description 6
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000002547 anomalous effect Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000011144 upstream manufacturing Methods 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/018—Certifying business or products
- G06Q30/0185—Product, service or business identity fraud
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/907—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Library & Information Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method and a device for identifying information false degree, and relates to the technical field of computers. One embodiment of the method comprises: receiving an identification request, acquiring information to be identified, which is used for commenting on a product by a user, according to the identification request, and converting the information to be identified into an abnormal composition to be identified; calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in information to be identified; inputting the abnormal picture to be recognized, the initial characteristics of the user, the initial characteristics of the comment and the initial characteristics of the product into a trained neural network model to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees. According to the method and the device, information of a user commenting on a product can be converted into the heteromorphic graph, and then graph structure information and node characteristics in the graph are effectively fused by using the graph neural network model for detection, so that the false degree of the comment is determined.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method and a device for identifying information false degree.
Background
In the big data era, the need for anomaly detection of data is common. The abnormal detection is to find out abnormal behaviors or behavior sequences inconsistent with normal data by utilizing a computer data mining technology, and the abnormal detection technology can be applied to the fields of network security intrusion detection, E-commerce platform comment detection, Internet water force detection, false news detection and the like. Common anomaly detection methods include (1) graph structure model-based detection, and (2) graph neural network model-based detection.
In the process of implementing the invention, the prior art at least has the following problems:
(1) for the detection based on the graph structure model, only the structure information of the graph is usually considered, and the feature information of the fusion node is not available;
(2) for the detection based on the graph neural network model, the current model is mostly suitable for the same graph, and whether the adjacent nodes are disguised nodes or not is not considered, so that the detection result is inaccurate.
Disclosure of Invention
In view of this, embodiments of the present invention provide an information false degree identification method and apparatus, which can convert information to be identified, which is used by a user to comment on a product, into an abnormal graph, and further effectively fuse graph structure information and features of nodes in a graph for detection by using a graph neural network model, so as to determine a false degree of a comment.
To achieve the above object, according to an aspect of an embodiment of the present invention, there is provided a method for identifying a false degree of information, including:
receiving an identification request, acquiring information to be identified of a product commented by a user according to the identification request, and converting the information to be identified into an abnormal picture to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the information to be identified;
inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
Optionally, before receiving the identification request, the method further includes:
generating the identification request for all the comment information in a preset time period according to a preset frequency; or,
generating the identification request based on the received identification dimension information.
Optionally, converting the information to be identified into an abnormal image to be identified includes:
respectively determining user nodes, comment nodes and product nodes in the abnormal picture to be identified according to the user information, the comment content and the product information in the information to be identified; the comment node is a node to be identified, and the user node and the product node are adjacent nodes of the comment node;
determining edges of the user nodes and the comment nodes in the heterogeneous graph to be identified and edges of the comment nodes and the product nodes according to the incidence relation among the users, the comments and the products, which is indicated by the users commenting the products in the information to be identified;
and constructing the abnormal composition to be identified based on the user node, the comment node, the product node, the edges of the user node and the comment node and the edges of the comment node and the product node.
Optionally, the neural network model is obtained by training according to the following method, including:
calling historical information of commenting on a product by a user with a known false degree, and selecting a preset amount of information as training data;
converting the training data into a sample abnormal graph containing user nodes, comment nodes and product nodes to serve as a training sample; the comment nodes are to-be-identified nodes, and the user nodes and the product nodes are both adjacent nodes of the comment nodes; the false degree of the historical information in the training data is used as a sample label of the node to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the training data to serve as sample features;
and training according to the training sample, the sample label and the sample characteristic to obtain the neural network model.
Optionally, the training according to the training sample, the sample label, and the sample feature to obtain the neural network model includes:
for each comment node in the sample abnormal graph, determining an adjacent user node and an adjacent product node of the comment node;
for the adjacent user node, determining other comment nodes and other product nodes which can be reached by the node in the sample abnormal graph, and calling an aggregation function to generate a first user representation vector of the adjacent user node by combining the comment initial feature and the product initial feature; splicing the first user expression vector with the initial user characteristics of the adjacent user node to obtain a user expression vector of the adjacent user node;
for the adjacent product nodes, determining other comment nodes and other user nodes which can be reached by the nodes in the sample abnormal graph, and calling an aggregation function to generate a first product representation vector of the adjacent product nodes by combining the comment initial features and the user initial features; splicing the first product expression vector and the initial product characteristics of the adjacent product nodes to obtain the product expression vector of the adjacent product nodes;
and accessing the sample abnormal graph, the user expression vector, the product expression vector, the comment initial characteristic of the comment node and the sample label into a multilayer neural network, and training to obtain the neural network model.
Optionally, the invoking an aggregation function to generate a first user representation vector of the adjacent user node, or the invoking an aggregation function to generate a first product representation vector of the adjacent product node, includes:
generating the first user representation vector or first product representation vector using an attention mechanism in the aggregation function.
Optionally, for an adjacent user node of each comment node, after determining other user nodes that can be reached by the adjacent product node of the comment node in the sample heterogeneous graph, calculating user similarity between the adjacent user node and the other user nodes based on the initial characteristics of the user, determining a second user node from the other user nodes based on the user similarity by using a preset filtering method, and determining a corresponding second product node; determining a user representation vector of the adjacent user node according to the second user node and the corresponding second product node; and/or the presence of a gas in the gas,
for the adjacent product node of each comment node, after determining other product nodes which can be reached by the adjacent user node of the comment node in the sample abnormal graph, calculating the product similarity of the adjacent product node and the other product nodes based on the initial characteristics of the product, determining a third product node from the other product nodes based on the product similarity by adopting a preset filtering method, and determining a corresponding third user node; and determining a product representation vector of the adjacent product node according to the third product node and the corresponding third user node.
Optionally, the user-initiated characteristic comprises at least one of: the method comprises the following steps of (1) a user comment number characteristic, a user comment level characteristic, a user comment frequency characteristic and a user comment length characteristic;
the review initial characteristics include at least one of: the system comprises a grade characteristic, a grade deviation characteristic, a content characteristic and a characteristic of a corresponding product to be commented;
the product initial characteristics include at least one of: the comment processing system comprises a commented frequency characteristic, a commented grade characteristic, a commented content characteristic and a commented frequency characteristic.
According to still another aspect of the embodiments of the present invention, there is provided an apparatus for identifying a false degree of information, including:
the information acquisition module is used for receiving an identification request, acquiring information to be identified, which is used for commenting on a product by a user, according to the identification request, and converting the information to be identified into an abnormal composition to be identified;
the calculation characteristic module is used for calling a characteristic calculation method set and calculating the initial characteristics of the user, the initial characteristics of the comment and the initial characteristics of the product in the information to be identified;
the recognition module is used for inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model so as to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
Optionally, before receiving the identification request, the method further includes:
generating the identification request for all the comment information in a preset time period according to a preset frequency; or,
generating the identification request based on the received identification dimension information.
Optionally, converting the information to be identified into an abnormal image to be identified includes:
respectively determining user nodes, comment nodes and product nodes in the abnormal picture to be identified according to the user information, the comment content and the product information in the information to be identified; the comment node is a node to be identified, and the user node and the product node are adjacent nodes of the comment node;
determining edges of the user nodes and the comment nodes in the heterogeneous graph to be identified and edges of the comment nodes and the product nodes according to the incidence relation among the users, the comments and the products, which is indicated by the users commenting the products in the information to be identified;
and constructing the abnormal composition to be identified based on the user node, the comment node, the product node, the edges of the user node and the comment node and the edges of the comment node and the product node.
Optionally, the neural network model is obtained by training according to the following method, including:
calling historical information of commenting on a product by a user with a known false degree, and selecting a preset amount of information as training data;
converting the training data into a sample abnormal graph containing user nodes, comment nodes and product nodes to serve as a training sample; the comment nodes are to-be-identified nodes, and the user nodes and the product nodes are both adjacent nodes of the comment nodes; the false degree of the historical information in the training data is used as a sample label of the node to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the training data to serve as sample features;
and training according to the training sample, the sample label and the sample characteristic to obtain the neural network model.
Optionally, the training according to the training sample, the sample label, and the sample feature to obtain the neural network model includes:
for each comment node in the sample abnormal graph, determining an adjacent user node and an adjacent product node of the comment node;
for the adjacent user node, determining other comment nodes and other product nodes which can be reached by the node in the sample abnormal graph, and calling an aggregation function to generate a first user representation vector of the adjacent user node by combining the comment initial feature and the product initial feature; splicing the first user expression vector with the initial user characteristics of the adjacent user node to obtain a user expression vector of the adjacent user node;
for the adjacent product nodes, determining other comment nodes and other user nodes which can be reached by the nodes in the sample abnormal graph, and calling an aggregation function to generate a first product representation vector of the adjacent product nodes by combining the comment initial features and the user initial features; splicing the first product expression vector and the initial product characteristics of the adjacent product nodes to obtain the product expression vector of the adjacent product nodes;
and accessing the sample abnormal graph, the user expression vector, the product expression vector, the comment initial characteristic of the comment node and the sample label into a multilayer neural network, and training to obtain the neural network model.
Optionally, the invoking an aggregation function to generate a first user representation vector of the adjacent user node, or the invoking an aggregation function to generate a first product representation vector of the adjacent product node, includes:
generating the first user representation vector or first product representation vector using an attention mechanism in the aggregation function.
Optionally, for an adjacent user node of each comment node, after determining other user nodes that can be reached by the adjacent product node of the comment node in the sample heterogeneous graph, calculating user similarity between the adjacent user node and the other user nodes based on the initial characteristics of the user, determining a second user node from the other user nodes based on the user similarity by using a preset filtering method, and determining a corresponding second product node; determining a user representation vector of the adjacent user node according to the second user node and the corresponding second product node; and/or the presence of a gas in the gas,
for the adjacent product node of each comment node, after determining other product nodes which can be reached by the adjacent user node of the comment node in the sample abnormal graph, calculating the product similarity of the adjacent product node and the other product nodes based on the initial characteristics of the product, determining a third product node from the other product nodes based on the product similarity by adopting a preset filtering method, and determining a corresponding third user node; and determining a product representation vector of the adjacent product node according to the third product node and the corresponding third user node.
Optionally, the user-initiated characteristic comprises at least one of: the method comprises the following steps of (1) a user comment number characteristic, a user comment level characteristic, a user comment frequency characteristic and a user comment length characteristic;
the review initial characteristics include at least one of: the system comprises a grade characteristic, a grade deviation characteristic, a content characteristic and a characteristic of a corresponding product to be commented;
the product initial characteristics include at least one of: the comment processing system comprises a commented frequency characteristic, a commented grade characteristic, a commented content characteristic and a commented frequency characteristic.
According to another aspect of the embodiments of the present invention, there is provided an electronic device for identifying a false degree of information, including:
one or more processors;
a storage device for storing one or more programs,
when the one or more programs are executed by the one or more processors, the one or more processors implement the method for identifying the false degree of information provided by the invention.
According to a further aspect of the embodiments of the present invention, there is provided a computer-readable medium on which a computer program is stored, the program, when executed by a processor, implementing the method for identifying the degree of information falsification provided by the present invention.
One embodiment of the above invention has the following advantages or benefits: because the technical means of converting the information to be identified, which is used for commenting on a product by a user, into the heteromorphic graph, and then effectively fusing the graph structure information and the characteristics of the nodes in the graph by using the graph neural network model to detect and determine the false degree of the comment is adopted, the technical problems that the graph neural network model in the prior art is mostly suitable for the same composition or does not consider the characteristics of the graph nodes and is not suitable for the comment detection scene in the invention are solved, and the technical effect of commenting the false detection by using the graph neural network model in combination with the characteristics of the comment information is achieved.
Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
FIG. 1 is a schematic diagram illustrating the main flow of a method for identifying the level of information falsification according to a first embodiment of the present invention;
FIG. 2 is a diagram illustrating an abnormal pattern in a method for identifying the extent of information falsification according to a second embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating transformation of a heteromorphic graph during training of a neural network in an identification method of information false degree according to a second embodiment of the present invention;
FIG. 4 is a schematic diagram of the main blocks of an apparatus for identifying the extent of information falsification according to an embodiment of the present invention;
FIG. 5 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;
fig. 6 is a schematic block diagram of a computer system suitable for use in implementing a terminal device or server of an embodiment of the invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Fig. 1 is a schematic diagram of a main flow of a method for identifying information false degree according to a first embodiment of the present invention, as shown in fig. 1, including:
step S101, receiving an identification request, acquiring information to be identified, commented on a product by a user, according to the identification request, and converting the information to be identified into an abnormal picture to be identified;
step S102, calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the information to be identified;
step S103, inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
The scenario in the application can be that a user comments on a product (such as contents of goods/food/service). In practice, some merchants often hire the water force to give false goodwill on their goods in order to attract more users; at the same time, competitors may hire the water force to make false negative comments on a particular product, thereby interfering with the normal judgment of the user. These false comments may be considered abnormal comments relative to normal user comments. The anomaly detection algorithm identifies these anomalous samples from the normal samples. The anomaly detection method provided by the application can be suitable for the false comment detection, and can also be suitable for various application scenes such as credit card fraud, network security intrusion detection, medical image anomaly detection and the like.
With the rise and application of deep learning, the graph-based neural network model is widely used in anomaly detection. The embodiment of the invention provides a method and a device for identifying information false degree, which can convert information to be identified, which is used for commenting on a product by a user, into a heteromorphic graph, and further effectively fuse graph structure information and characteristics of nodes in a graph for detection by using a graph neural network model to determine the false degree of the comment.
In some embodiments, before receiving the identification request, further comprising:
generating the identification request for all the comment information in a preset time period according to a preset frequency; or,
generating the identification request based on the received identification dimension information.
The identification request in the present application may be generated based on a preset frequency set in advance (for example, a request is generated once per week, or a request is generated once per half month, etc.), or may be generated when an identification dimension of an external input is received; in some practical applications, the identification dimension may include information such as a user level, a product category, a time range, and the like, and the identification dimension may be input by a user or an identification dimension indicated in an identification instruction issued by another upstream system; can be flexibly set according to actual conditions.
In some embodiments, converting the information to be identified into an anomaly map to be identified includes:
respectively determining user nodes, comment nodes and product nodes in the abnormal picture to be identified according to the user information, the comment content and the product information in the information to be identified; the comment node is a node to be identified, and the user node and the product node are adjacent nodes of the comment node;
determining edges of the user nodes and the comment nodes in the heterogeneous graph to be identified and edges of the comment nodes and the product nodes according to the incidence relation among the users, the comments and the products, which is indicated by the users commenting the products in the information to be identified;
and constructing the abnormal composition to be identified based on the user node, the comment node, the product node, the edges of the user node and the comment node and the edges of the comment node and the product node.
The heteromorphic graph constructed according to the mode can present the information to be identified in a heteromorphic graph form completely, does not omit the association relation among nodes, and can be combined with the node characteristics conveniently.
In some embodiments, the neural network model is trained according to the following method, including:
calling historical information of commenting on a product by a user with a known false degree, and selecting a preset amount of information as training data;
converting the training data into a sample abnormal graph containing user nodes, comment nodes and product nodes to serve as a training sample; the comment nodes are to-be-identified nodes, and the user nodes and the product nodes are both adjacent nodes of the comment nodes; the false degree of the historical information in the training data is used as a sample label of the node to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the training data to serve as sample features;
and training according to the training sample, the sample label and the sample characteristic to obtain the neural network model.
In the history information of the product commented by the users with known false degree, the following content information can be contained in each comment: user-product-comment content-rating-comment time-false comment; wherein "comment is true" can be identified by 0 and "comment is false" can be identified by 1.
Fig. 2 is a schematic diagram of an abnormal pattern in a recognition method of information false degree according to a second embodiment of the present invention, and as shown in fig. 2, for a selected amount of determined training data, the converted abnormal pattern may include:
user node uiCommodity node pjComment node rkAnd user uiGive a comment of rkForm a class edge
In the commodity pjRemarks on rkForming another class of edges
In some embodiments, the training according to the training samples, the sample labels, and the sample features to obtain the neural network model includes:
for each comment node in the sample abnormal graph, determining an adjacent user node and an adjacent product node of the comment node;
for the adjacent user node, determining other comment nodes and other product nodes which can be reached by the node in the sample abnormal graph, and calling an aggregation function to generate a first user representation vector of the adjacent user node by combining the comment initial feature and the product initial feature; splicing the first user expression vector with the initial user characteristics of the adjacent user node to obtain a user expression vector of the adjacent user node;
for the adjacent product nodes, determining other comment nodes and other user nodes which can be reached by the nodes in the sample abnormal graph, and calling an aggregation function to generate a first product representation vector of the adjacent product nodes by combining the comment initial features and the user initial features; splicing the first product expression vector and the initial product characteristics of the adjacent product nodes to obtain the product expression vector of the adjacent product nodes;
and accessing the sample abnormal graph, the user expression vector, the product expression vector, the comment initial characteristic of the comment node and the sample label into a multilayer neural network, and training to obtain the neural network model.
FIG. 3 is a schematic diagram illustrating transformation of an abnormal pattern during training of a neural network in a recognition method of information false degree according to a second embodiment of the present invention, as shown in FIG. 3, r in the abnormal pattern of FIG. 2 is0Node and r1The conversion of the nodes comprises the following steps:
for r0The node, according to the map of FIG. 2, determines r0Adjacent node u of node0Node and p0A node; furthermore, for u0And the nodes determine the comment nodes and the corresponding product nodes which can be reached by the heterogeneous graph in the FIG. 2: u. of0-r1-p1，u0-r6-p2(ii) a For p0The nodes determine the comment nodes and the corresponding user nodes which can be reached by the heterogeneous graph in FIG. 2: none is determined in the heteromorphic graph of fig. 2; as shown in the upper half of figure 3.
Similarly, for r1The nodes are also transformed in the same way as shown in the lower half of figure 3.
In some practical applications, in consideration of facilitating computer processing, six connection matrices describing the sample heterogeneous graph can be generated according to the sample heterogeneous graph, and then calculation is performed based on the six connection matrices according to calculation requirements to obtain corresponding features or expression vectors.
In conjunction with fig. 2 above, the six connection matrices may include:
(1) connection matrix RU for comments to usersk，iAnd a connection matrix RP of reviews to goodsk，j(ii) a Wherein the k-th line represents the comment rkIs user uiFor a commodity pjDisclosed is a method for preparing a compound.
Corresponding RU in FIG. 2k，iIs [ [0 ]][0][1][2][2][3][0][3]]Corresponding RPk，jIs [ [0 ]][1][1][1][2][1][2][2]]；
(2) User to comment matrix URi，kAnd the user-to-merchandise matrix UPi，jB, carrying out the following steps of; wherein row i represents user uiIn the commodity pjPublished comment rk. Corresponding UR in FIG. 2i，kIs [ [0, 1, 6 ]][2][3，4][5，7]]Corresponding UPi，jIs [ [0, 1, 2 ]][1][1，2][1，2]]；
(3) Commodity-to-user matrix PUj，iAnd a matrix PR of goods to reviewsj，k(ii) a Wherein the jth row represents the item pjCorresponding user uiAnd comment rk. Corresponding PU in FIG. 2j，iIs [ [0 ]][0，1，2，3][0，2，3]]Corresponding PRj，kIs [ [0 ]][1，2，3，5][4，6，7]]。
Further, when the conversion of the heterogeneous map shown in fig. 3 is performed, the comment r may be determined firstkThe label 1 represents a false comment, and the label 0 represents a normal comment. According to RUk，iAnd RPk，jAnd finding out the corresponding user and commodity. To the user again according to URi，kAnd UPi，jAnd finding corresponding comments and products of the users. Subjecting the product to PRj，iAnd PUj，kFinding out corresponding comments and users to form the conversion data shown in fig. 3 so as to carry out feature splicing and obtain each expression vector.
The user expression vector, the product expression vector and the comment initial characteristics of the comment nodes obtained by the method can fully express the meaning of each node in detail, do not omit the association relation among the nodes and the characteristic information of the nodes, and further enable the trained model to be more accurate.
In some embodiments, invoking the aggregation function to generate the first user representation vector of the adjacent user node, or invoking the aggregation function to generate the first product representation vector of the adjacent product node, comprises:
generating the first user representation vector or first product representation vector using an attention mechanism in the aggregation function.
The use of an attention mechanism may enable the aggregated features to more accurately describe the reviews; the corresponding physical meaning is: when a user puts comments on a plurality of commodities, the comment information which is closer to the product corresponding to the current comment is more worth referring. Such as: the user gives comments on sports, food and electronic commodities, and the current comments are specific to the food, so that the comment information of the user in the food can be aggregated more by using an attention mechanism, and other irrelevant information is ignored, so that the aggregation characteristic can describe the comments more accurately, and further, the trained model is more accurate.
In some embodiments, for an adjacent user node of each comment node, after determining other user nodes that can be reached by the adjacent product node of the comment node in the sample heterogeneous graph, calculating user similarity between the adjacent user node and the other user nodes based on user initial characteristics, determining a second user node from the other user nodes based on the user similarity by using a preset filtering method, and determining a corresponding second product node; determining a user representation vector of the adjacent user node according to the second user node and the corresponding second product node; and/or the presence of a gas in the gas,
for the adjacent product node of each comment node, after determining other product nodes which can be reached by the adjacent user node of the comment node in the sample abnormal graph, calculating the product similarity of the adjacent product node and the other product nodes based on the initial characteristics of the product, determining a third product node from the other product nodes based on the product similarity by adopting a preset filtering method, and determining a corresponding third user node; and determining a product representation vector of the adjacent product node according to the third product node and the corresponding third user node.
The filtering method may include setting a filtering threshold, or setting a percentage threshold; the following are exemplified:
when the user similarity between one adjacent user node and other user nodes is calculated as: 1, 0.9, 0.8, 0.7, 0.2, if the filtering method is to set the filtering threshold, then when the filtering threshold is 0.6, the user nodes corresponding to 1, 0.9, 0.8, 0.7 can be selected for subsequent calculation; if the filtering method is to set a filtering percentage threshold, when the filtering threshold is 60%, the user nodes corresponding to 1, 0.9 and 0.8 can be selected for subsequent calculation; under different scene requirements, different filtering methods can be used and different filtering parameters (such as a filtering threshold value or a filtering percentage threshold value) can be set.
By using the method, the nodes which are more similar to the concerned nodes are screened out to determine the expression vector, more effective node information can be aggregated, the node characteristics can be expressed more accurately, and the model performance is improved.
In some embodiments, the user-initiated characteristic comprises at least one of: the method comprises the following steps of (1) a user comment number characteristic, a user comment level characteristic, a user comment frequency characteristic and a user comment length characteristic;
the review initial characteristics include at least one of: the system comprises a grade characteristic, a grade deviation characteristic, a content characteristic and a characteristic of a corresponding product to be commented;
the product initial characteristics include at least one of: the comment processing system comprises a commented frequency characteristic, a commented grade characteristic, a commented content characteristic and a commented frequency characteristic.
In practical application, the calculation method of each feature can be flexibly set according to actual conditions or updated according to actual requirements so as to select more appropriate features in different scenes, thereby improving the performance of the model.
Some examples of features are as follows:
TABLE 1 user characteristics
TABLE 2 product characteristics
TABLE 3 characteristics of comments
Fig. 4 is a schematic diagram of the main blocks of an information false degree identification apparatus according to an embodiment of the present invention, and as shown in fig. 4, the information false degree identification apparatus 400 includes:
an information obtaining module 401, configured to receive an identification request, obtain information to be identified, which is used for a user to comment on a product, according to the identification request, and convert the information to be identified into an abnormal composition to be identified;
a feature calculation module 402, configured to invoke a feature calculation method set, and calculate user initial features, comment initial features, and product initial features in the information to be identified;
the identification module 403 is configured to input the abnormal image to be identified, the initial user characteristic, the initial review characteristic, and the initial product characteristic into a trained neural network model, so as to obtain a false degree identification result of the information to be identified; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
The scenario in the application can be that a user comments on a product (such as contents of goods/food/service). In practice, some merchants often hire the water force to give false goodwill on their goods in order to attract more users; at the same time, competitors may hire the water force to make false negative comments on a particular product, thereby interfering with the normal judgment of the user. These false comments may be considered abnormal comments relative to normal user comments. The anomaly detection algorithm identifies these anomalous samples from the normal samples. The anomaly detection method provided by the application can be suitable for the false comment detection, and can also be suitable for various application scenes such as credit card fraud, network security intrusion detection, medical image anomaly detection and the like.
With the rise and application of deep learning, the graph-based neural network model is widely used in anomaly detection. The embodiment of the invention provides a method and a device for identifying information false degree, which can convert information to be identified, which is used for commenting on a product by a user, into a heteromorphic graph, and further effectively fuse graph structure information and characteristics of nodes in a graph for detection by using a graph neural network model to determine the false degree of the comment.
In some embodiments, before receiving the identification request, further comprising:
generating the identification request for all the comment information in a preset time period according to a preset frequency; or,
generating the identification request based on the received identification dimension information.
The identification request in the present application may be generated based on a preset frequency set in advance (for example, a request is generated once per week, or a request is generated once per half month, etc.), or may be generated when an identification dimension of an external input is received; in some practical applications, the identification dimension may include information such as a user level, a product category, a time range, and the like, and the identification dimension may be input by a user or an identification dimension indicated in an identification instruction issued by another upstream system; can be flexibly set according to actual conditions.
In some embodiments, converting the information to be identified into an anomaly map to be identified includes:
respectively determining user nodes, comment nodes and product nodes in the abnormal picture to be identified according to the user information, the comment content and the product information in the information to be identified; the comment node is a node to be identified, and the user node and the product node are adjacent nodes of the comment node;
determining edges of the user nodes and the comment nodes in the heterogeneous graph to be identified and edges of the comment nodes and the product nodes according to the incidence relation among the users, the comments and the products, which is indicated by the users commenting the products in the information to be identified;
and constructing the abnormal composition to be identified based on the user node, the comment node, the product node, the edges of the user node and the comment node and the edges of the comment node and the product node.
The heteromorphic graph constructed according to the mode can present the information to be identified in a heteromorphic graph form completely, does not omit the association relation among nodes, and can be combined with the node characteristics conveniently.
In some embodiments, the neural network model is trained according to the following method, including:
calling historical information of commenting on a product by a user with a known false degree, and selecting a preset amount of information as training data;
converting the training data into a sample abnormal graph containing user nodes, comment nodes and product nodes to serve as a training sample; the comment nodes are to-be-identified nodes, and the user nodes and the product nodes are both adjacent nodes of the comment nodes; the false degree of the historical information in the training data is used as a sample label of the node to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the training data to serve as sample features;
and training according to the training sample, the sample label and the sample characteristic to obtain the neural network model.
In the history information of the product commented by the users with known false degree, the following content information can be contained in each comment: user-product-comment content-rating-comment time-whether or not comment is false.
In some embodiments, the training according to the training samples, the sample labels, and the sample features to obtain the neural network model includes:
for each comment node in the sample abnormal graph, determining an adjacent user node and an adjacent product node of the comment node;
for the adjacent user node, determining other comment nodes and other product nodes which can be reached by the node in the sample abnormal graph, and calling an aggregation function to generate a first user representation vector of the adjacent user node by combining the comment initial feature and the product initial feature; splicing the first user expression vector with the initial user characteristics of the adjacent user node to obtain a user expression vector of the adjacent user node;
for the adjacent product nodes, determining other comment nodes and other user nodes which can be reached by the nodes in the sample abnormal graph, and calling an aggregation function to generate a first product representation vector of the adjacent product nodes by combining the comment initial features and the user initial features; splicing the first product expression vector and the initial product characteristics of the adjacent product nodes to obtain the product expression vector of the adjacent product nodes;
and accessing the sample abnormal graph, the user expression vector, the product expression vector, the comment initial characteristic of the comment node and the sample label into a multilayer neural network, and training to obtain the neural network model.
The user expression vector, the product expression vector and the comment initial characteristics of the comment nodes obtained by the method can fully express the meaning of each node in detail, do not omit the association relation among the nodes and the characteristic information of the nodes, and further enable the trained model to be more accurate.
Further, in some embodiments, invoking an aggregation function to generate a first user representation vector for the adjacent user node, or invoking an aggregation function to generate a first product representation vector for the adjacent product node, comprises:
generating the first user representation vector or first product representation vector using an attention mechanism in the aggregation function.
The use of an attention mechanism may enable the aggregated features to more accurately describe the reviews; the corresponding physical meaning is: when a user puts comments on a plurality of commodities, the comment information which is closer to the product corresponding to the current comment is more worth referring. Such as: the user gives comments on sports, food and electronic commodities, and the current comments are specific to the food, so that the comment information of the user in the food can be aggregated more by using an attention mechanism, and other irrelevant information is ignored, so that the aggregation characteristic can describe the comments more accurately, and further, the trained model is more accurate.
In some embodiments, for an adjacent user node of each comment node, after determining other user nodes that can be reached by the adjacent product node of the comment node in the sample heterogeneous graph, calculating user similarity between the adjacent user node and the other user nodes based on user initial characteristics, determining a second user node from the other user nodes based on the user similarity by using a preset filtering method, and determining a corresponding second product node; determining a user representation vector of the adjacent user node according to the second user node and the corresponding second product node; and/or the presence of a gas in the gas,
for the adjacent product node of each comment node, after determining other product nodes which can be reached by the adjacent user node of the comment node in the sample abnormal graph, calculating the product similarity of the adjacent product node and the other product nodes based on the initial characteristics of the product, determining a third product node from the other product nodes based on the product similarity by adopting a preset filtering method, and determining a corresponding third user node; and determining a product representation vector of the adjacent product node according to the third product node and the corresponding third user node.
The filtering method may include setting a filtering threshold, or setting a percentage threshold; the following are exemplified:
when the user similarity between one adjacent user node and other user nodes is calculated as: 1, 0.9, 0.8, 0.7, 0.2, if the filtering method is to set the filtering threshold, then when the filtering threshold is 0.6, the user nodes corresponding to 1, 0.9, 0.8, 0.7 can be selected for subsequent calculation; if the filtering method is to set a filtering percentage threshold, when the filtering threshold is 60%, the user nodes corresponding to 1, 0.9 and 0.8 can be selected for subsequent calculation; under different scene requirements, different filtering methods can be used and different filtering parameters (such as a filtering threshold value or a filtering percentage threshold value) can be set.
By using the method, the nodes which are more similar to the concerned nodes are screened out to determine the expression vector, more effective node information can be aggregated, the node characteristics can be expressed more accurately, and the model performance is improved.
In some embodiments, the user-initiated characteristic comprises at least one of: the method comprises the following steps of (1) a user comment number characteristic, a user comment level characteristic, a user comment frequency characteristic and a user comment length characteristic;
the review initial characteristics include at least one of: the system comprises a grade characteristic, a grade deviation characteristic, a content characteristic and a characteristic of a corresponding product to be commented;
the product initial characteristics include at least one of: the comment processing system comprises a commented frequency characteristic, a commented grade characteristic, a commented content characteristic and a commented frequency characteristic.
Fig. 5 shows an exemplary system architecture 500 to which the method for identifying the degree of information falsification or the apparatus for identifying the degree of information falsification of the embodiment of the present invention can be applied.
As shown in fig. 5, the system architecture 500 may include terminal devices 501, 502, 503, a network 504, and a server 505. The network 504 serves to provide a medium for communication links between the terminal devices 501, 502, 503 and the server 505. Network 504 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The user may use the terminal devices 501, 502, 503 to interact with a server 505 over a network 504 to receive or send messages or the like. The terminal devices 501, 502, 503 may have various client applications installed thereon, which have a requirement for identifying the extent of falsification of information.
The terminal devices 501, 502, 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The server 505 may be a server providing various services, such as a background management server supporting the user's false degree of identification information using the terminal devices 501, 502, 503. The background management server can analyze and process the received identification request and feed back the processing result to the terminal equipment.
It should be noted that the method for identifying the information falsification degree provided by the embodiment of the present invention is generally executed by the server 505, and accordingly, the apparatus for identifying the information falsification degree is generally disposed in the server 505.
It should be understood that the number of terminal devices, networks, and servers in fig. 5 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring now to FIG. 6, a block diagram of a computer system 600 suitable for use with a terminal device implementing an embodiment of the invention is shown. The terminal device shown in fig. 6 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 6, the computer system 600 includes a Central Processing Unit (CPU)601 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage section 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data necessary for the operation of the system 600 are also stored. The CPU 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.
The following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, and the like; an output portion 607 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 608 including a hard disk and the like; and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the internet. The driver 610 is also connected to the I/O interface 605 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 610 as necessary, so that a computer program read out therefrom is mounted in the storage section 608 as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 609, and/or installed from the removable medium 611. The computer program performs the above-described functions defined in the system of the present invention when executed by the Central Processing Unit (CPU) 601.
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The modules described in the embodiments of the present invention may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor comprises an information acquisition module, a feature calculation module and an identification module. Wherein the names of the modules do not in some cases constitute a limitation of the module itself.
As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: step S101, receiving an identification request, acquiring information to be identified, commented on a product by a user, according to the identification request, and converting the information to be identified into an abnormal picture to be identified; step S102, calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the information to be identified; step S103, inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
According to the technical scheme of the embodiment of the invention, because the technical means of converting the information to be identified, which is used for commenting on a product by a user, into the heteromorphic graph is adopted, the graph neural network model is used for effectively fusing the graph structure information and the characteristics of the nodes in the graph for detection, and the false degree of the comment is determined, the technical problems that the graph neural network model in the prior art is mostly suitable for the homographic graph or does not consider the characteristics of the graph nodes and is not suitable for the comment detection scene in the invention are solved, and the technical effect of commenting the false detection by using the graph neural network model in combination with the characteristics of the comment information is achieved.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (11)
1. A method for identifying the false degree of information is characterized by comprising the following steps:
receiving an identification request, acquiring information to be identified of a product commented by a user according to the identification request, and converting the information to be identified into an abnormal picture to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the information to be identified;
inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
2. The method of claim 1, prior to receiving the identification request, further comprising:
generating the identification request for all the comment information in a preset time period according to a preset frequency; or,
generating the identification request based on the received identification dimension information.
3. The method according to claim 1, wherein converting the information to be identified into an anomaly map to be identified comprises:
respectively determining user nodes, comment nodes and product nodes in the abnormal picture to be identified according to the user information, the comment content and the product information in the information to be identified; the comment node is a node to be identified, and the user node and the product node are adjacent nodes of the comment node;
determining edges of the user nodes and the comment nodes in the heterogeneous graph to be identified and edges of the comment nodes and the product nodes according to the incidence relation among the users, the comments and the products, which is indicated by the users commenting the products in the information to be identified;
and constructing the abnormal composition to be identified based on the user node, the comment node, the product node, the edges of the user node and the comment node and the edges of the comment node and the product node.
4. The method of claim 1, wherein the neural network model is trained according to a method comprising:
calling historical information of commenting on a product by a user with a known false degree, and selecting a preset amount of information as training data;
converting the training data into a sample abnormal graph containing user nodes, comment nodes and product nodes to serve as a training sample; the comment nodes are to-be-identified nodes, and the user nodes and the product nodes are both adjacent nodes of the comment nodes; the false degree of the historical information in the training data is used as a sample label of the node to be identified;
calling a feature calculation method set, and calculating user initial features, comment initial features and product initial features in the training data to serve as sample features;
and training according to the training sample, the sample label and the sample characteristic to obtain the neural network model.
5. The method according to any one of claims 4, wherein the training to obtain the neural network model according to the training samples, the sample labels and the sample features comprises:
for each comment node in the sample abnormal graph, determining an adjacent user node and an adjacent product node of the comment node;
for the adjacent user node, determining other comment nodes and other product nodes which can be reached by the node in the sample abnormal graph, and calling an aggregation function to generate a first user representation vector of the adjacent user node by combining the comment initial feature and the product initial feature; splicing the first user expression vector with the initial user characteristics of the adjacent user node to obtain a user expression vector of the adjacent user node;
for the adjacent product nodes, determining other comment nodes and other user nodes which can be reached by the nodes in the sample abnormal graph, and calling an aggregation function to generate a first product representation vector of the adjacent product nodes by combining the comment initial features and the user initial features; splicing the first product expression vector and the initial product characteristics of the adjacent product nodes to obtain the product expression vector of the adjacent product nodes;
and accessing the sample abnormal graph, the user expression vector, the product expression vector, the comment initial characteristic of the comment node and the sample label into a multilayer neural network, and training to obtain the neural network model.
6. The method of claim 5, wherein invoking an aggregation function to generate the first user representation vector for the neighboring user node or invoking an aggregation function to generate the first product representation vector for the neighboring product node comprises:
generating the first user representation vector or first product representation vector using an attention mechanism in the aggregation function.
7. The method of claim 5, further comprising:
for the adjacent user node of each comment node, after determining other user nodes which can be reached by the adjacent product node of the comment node in the sample heterogeneous graph, calculating the user similarity between the adjacent user node and the other user nodes based on the initial characteristics of the user, determining a second user node from the other user nodes based on the user similarity by adopting a preset filtering method, and determining a corresponding second product node; determining a user representation vector of the adjacent user node according to the second user node and the corresponding second product node; and/or the presence of a gas in the gas,
for the adjacent product node of each comment node, after determining other product nodes which can be reached by the adjacent user node of the comment node in the sample abnormal graph, calculating the product similarity of the adjacent product node and the other product nodes based on the initial characteristics of the product, determining a third product node from the other product nodes based on the product similarity by adopting a preset filtering method, and determining a corresponding third user node; and determining a product representation vector of the adjacent product node according to the third product node and the corresponding third user node.
8. The method according to claim 1 or 4,
the user initial characteristics include at least one of: the method comprises the following steps of (1) a user comment number characteristic, a user comment level characteristic, a user comment frequency characteristic and a user comment length characteristic;
the review initial characteristics include at least one of: the system comprises a grade characteristic, a grade deviation characteristic, a content characteristic and a characteristic of a corresponding product to be commented;
the product initial characteristics include at least one of: the comment processing system comprises a commented frequency characteristic, a commented grade characteristic, a commented content characteristic and a commented frequency characteristic.
9. An apparatus for identifying the extent of information falsification, comprising:
the information acquisition module is used for receiving an identification request, acquiring information to be identified, which is used for commenting on a product by a user, according to the identification request, and converting the information to be identified into an abnormal composition to be identified;
the calculation characteristic module is used for calling a characteristic calculation method set and calculating the initial characteristics of the user, the initial characteristics of the comment and the initial characteristics of the product in the information to be identified;
the recognition module is used for inputting the abnormal picture to be recognized, the initial features of the user, the initial features of the comment and the initial features of the product into a trained neural network model so as to obtain a false degree recognition result of the information to be recognized; the neural network model is obtained by training according to historical information of the product commented by users with known false degrees.
10. An electronic device for order management, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-8.
11. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110141190.9A CN114092106A (en) | 2021-02-01 | 2021-02-01 | Method and device for identifying information false degree |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110141190.9A CN114092106A (en) | 2021-02-01 | 2021-02-01 | Method and device for identifying information false degree |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114092106A true CN114092106A (en) | 2022-02-25 |
Family
ID=80295912
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110141190.9A Pending CN114092106A (en) | 2021-02-01 | 2021-02-01 | Method and device for identifying information false degree |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114092106A (en) |
-
2021
- 2021-02-01 CN CN202110141190.9A patent/CN114092106A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11461999B2 (en) | Image object detection method, device, electronic device and computer readable medium | |
| US20180089627A1 (en) | System and method for advanced candidate screening | |
| CN110929799B (en) | Method, electronic device, and computer-readable medium for detecting abnormal user | |
| CN114663952A (en) | Object classification method, deep learning model training method, device and equipment | |
| CN111160410B (en) | Object detection method and device | |
| CN111782850B (en) | Object searching method and device based on hand drawing | |
| WO2022152018A1 (en) | Method and device for identifying multiple accounts belonging to the same person | |
| US20230281696A1 (en) | Method and apparatus for detecting false transaction order | |
| CN110895811A (en) | Image tampering detection method and device | |
| CN109542743B (en) | Log checking method and device, electronic equipment and computer readable storage medium | |
| CN111415168A (en) | Transaction warning method and device | |
| US20170249697A1 (en) | System and method for machine learning based line assignment | |
| CN116204740A (en) | Label determining method, information recommending method, device, equipment and storage medium | |
| CN114066603A (en) | Post-loan risk early warning method and device, electronic equipment and computer readable medium | |
| CN111428536A (en) | Training method and device for detection network for detecting article category and position | |
| CN114780847A (en) | Object information processing and information pushing method, device and system | |
| CN114092106A (en) | Method and device for identifying information false degree | |
| CN112256566B (en) | Fresh-keeping method and device for test cases | |
| CN113655942A (en) | Chart data display method and device | |
| CN113342998A (en) | Multimedia resource recommendation method and device, electronic equipment and storage medium | |
| CN109657523B (en) | Driving region detection method and device | |
| CN112308678A (en) | Price information processing method, device, equipment and medium based on image recognition | |
| CN112861684A (en) | Article display method and device | |
| CN107368597B (en) | Information output method and device | |
| CN116880899B (en) | Task information association method, device, electronic equipment and computer readable medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |