CN113706173A - Information management method and device, electronic equipment and storage medium - Google Patents

Information management method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN113706173A
CN113706173A CN202111017416.0A CN202111017416A CN113706173A CN 113706173 A CN113706173 A CN 113706173A CN 202111017416 A CN202111017416 A CN 202111017416A CN 113706173 A CN113706173 A CN 113706173A
Authority
CN
China
Prior art keywords
image
house
information
target
similar
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
Application number
CN202111017416.0A
Other languages
Chinese (zh)
Other versions
CN113706173B (en
Inventor
饶卿寻
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing 58 Information Technology Co Ltd
Original Assignee
Beijing 58 Information Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing 58 Information Technology Co Ltd filed Critical Beijing 58 Information Technology Co Ltd
Priority to CN202111017416.0A priority Critical patent/CN113706173B/en
Publication of CN113706173A publication Critical patent/CN113706173A/en
Application granted granted Critical
Publication of CN113706173B publication Critical patent/CN113706173B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/16—Real estate

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Tourism & Hospitality (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Primary Health Care (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Human Resources & Organizations (AREA)
  • Evolutionary Biology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Finance (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention provides an information management method, an information management device, electronic equipment and a storage medium, and relates to the technical field of computers. The method comprises the following steps: acquiring an image and related information of a target house corresponding to the target post; under the condition that similar images matched with the images exist in the house source library, acquiring a characteristic information set of the similar images; updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house released by the same releasing user; determining the stealing probability of the similar images according to each feature information in the updated feature information set and the weight corresponding to each feature information; and in the case that the image is determined to be a stolen image according to the stealing probability, putting down the target post. The invention effectively reduces the probability of spreading the false information on the house trading platform and improves the accuracy of the house information displayed in the house trading platform.

Description

Information management method and device, electronic equipment and storage medium
Technical Field
The present invention relates to the field of computer technologies, and in particular, to an information management method and apparatus, an electronic device, and a storage medium.
Background
With the development of internet technology, more and more house transaction platforms are mature day by day. On the house transaction platform, all house users can issue posts for recording various house transaction information, such as house buying and selling information, renting information and the like through the platform. The house demand user can find a satisfied house by browsing posts of various house transaction information issued by all users of the house, so that house transaction is carried out with landlord users or broker users.
However, the various types of posts in the house trading platform are numerous and complicated, and posts which issue a pirate picture including pirate pictures of other house sources may exist. Therefore, a method for identifying posts distributed with false information such as a pirate graph is needed to avoid unnecessary trouble for trading real estate caused by browsing the posts by users, and influence on user viscosity of a house trading platform.
Disclosure of Invention
In view of this, the present application provides an information management method, an information management apparatus, an electronic device, and a storage medium, which are used to identify whether an image of a target house corresponding to a target post is a pirate image, so as to effectively reduce the probability of spreading false information on a house trading platform, improve the accuracy of property information displayed in the house trading platform, avoid unnecessary troubles caused by the issuance of posts with false information to users for trading properties, and improve the user viscosity of the house trading platform.
According to a first aspect of the present application, there is provided an information management method, the method including:
acquiring an image and related information of a target house corresponding to the target post;
under the condition that a similar image matched with the image exists in a room source library, acquiring a characteristic information set of the similar image, wherein the characteristic information set is at least used for reflecting the exposure degree of the similar image;
updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house released by the same releasing user;
determining the probability of stealing the similar images according to each feature information in the updated feature information set and the weight corresponding to each feature information;
and under the condition that the image is determined to be a stolen image according to the stealing probability, the target post is placed down.
According to a second aspect of the present application, there is provided an information management apparatus, the apparatus including:
the acquisition module is used for acquiring the image and the related information of the target house corresponding to the target post; the method comprises the steps of obtaining a characteristic information set of a similar image under the condition that the similar image matched with the image exists in a house source library, wherein the characteristic information set is at least used for reflecting the exposure degree of the similar image;
the updating module is used for updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house published by the same publishing user or not;
the determining module is used for determining the stealing probability of the similar images according to each piece of feature information in the updated feature information set and the weight corresponding to each piece of feature information;
and the control module is used for setting down the target post under the condition that the image is determined to be a stolen image according to the stealing probability.
In a third aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the steps of the information management method according to the first aspect.
In a fourth aspect, the present application provides an electronic device comprising: a processor, a memory and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the information management method according to the first aspect when executing the program.
Aiming at the prior art, the method has the following advantages:
according to the information management method, the information management device, the electronic equipment and the storage medium, whether the similar image of the target house exists in the house source library or not is determined through matching of the image of the target house corresponding to the target post with the house image in the house source library. And under the condition that the similar images exist in the house source library, determining the stealing probability of the similar images according to the feature information in the updated feature set corresponding to the similar images. Therefore, based on the principle that the higher the stealing probability of the similar image is, the higher the probability that the image of the target house similar to the similar image is the stolen image is, the authenticity identification of the image of the target house corresponding to the target post is realized. And the updated feature information in the feature set is at least used for indicating whether the similar house and the target house are from the same house released by the same releasing user. However, in the case where the similar house corresponding to the similar image and the target house are the same house issued by the same issuer, the determination of whether the image of the target house is a stolen image is greatly affected. Therefore, the influence of the image of the same house issued by the same issuing user on the image of the authentication target house is considered, and the accuracy of the authentication of whether the image of the target house is the stolen image is further improved. The probability of spreading the false information on the house trading platform is effectively reduced, the accuracy of the house property information displayed in the house trading platform is improved, unnecessary troubles brought by posts distributed with the false information to users for trading house properties are avoided, and the user viscosity of the house trading platform is improved.
Drawings
Fig. 1 is a schematic diagram of an implementation environment of an information management method according to an embodiment of the present application;
fig. 2 is a flowchart of an information management method provided in an embodiment of the present application;
fig. 3 is a flowchart of a similar image determining method provided in an embodiment of the present application;
FIG. 4 is a flow chart of another information management method provided by an embodiment of the present application;
FIG. 5 is a flow chart of another similar image determination method provided in the embodiments of the present application;
fig. 6 is a block diagram of an information management apparatus according to an embodiment of the present application;
fig. 7 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Please refer to fig. 1, which illustrates an implementation environment diagram of an information management method according to an embodiment of the present application. As shown in fig. 1, the implementation environment includes: a terminal 101 and a server 102. The terminal 101 and the server 102 are connected through a wired network or a wireless network. It should be noted that one or more terminals may be included in the implementation environment, and fig. 1 illustrates an example in which the implementation environment includes only one terminal.
The terminal 101 can log in a house transaction platform provided with services by the server 102. For example, the terminal 101 may be installed with a house transaction client that the server 102 provides services, the terminal 101 may be installed with a house transaction application that the server 102 provides services, or the terminal 101 may display a house transaction web page that the server 102 provides services. The terminal 101 may be a terminal used by a publishing user. The publishing user can publish posts recording various house transaction information such as house buying and selling information (such as new house buying and selling information, second-hand house buying and selling information and the like) and renting information on the platform so as to achieve the purposes of selling houses, renting houses and the like. Wherein the publishing user may be the owner of the house, i.e. the landlord. The publishing user may also be one or more house brokers. A house broker trades a house by accepting a commitment from the house owner. For example, the terminal 101 may be an electronic device such as a mobile phone, a tablet computer, or a wearable device. The server 102 may be a single server or a service cluster formed by a plurality of servers.
It should be noted that the implementation environment may further include: another terminal than terminal 101. The other terminal may also log in to the house transaction platform provided by the service end 102, and the other terminal may be a terminal used by the counsel user. The counsel user may also be referred to as a house requisition user. The consultant user may browse the posts posted by the posting user to find his or her own satisfied house for house trading with the landlord user or broker user.
Please refer to fig. 2, which shows a flowchart of an information management method according to an embodiment of the present application. The information management method is applied to a server side in the implementation environment shown in fig. 1. As shown in fig. 2, the method includes:
step 201, obtaining an image and related information of a target house corresponding to the target post.
In the embodiment of the application, the target post may refer to one or more online posts published on a house trading platform provided by a server. The online post refers to a post that the terminal can browse on the house trading platform. Alternatively, the targeted post may also refer to a post in an audit that has been submitted on the house trading platform but has not been formally published. The post in the audit refers to a post which can be browsed by staff of the house trading platform, and after the audit is passed for the post, the post can be browsed on the house trading platform by the terminal. The target house to which the target post corresponds may refer to a house to be traded that is exposed in the target post. The image of the target house refers to the related image of the target house shown in the target post. The image may include a picture, video, or the like. Wherein, the number of the images of the target house in the target post can be one or more. The relevant information of the target house acquired by the server may include: address information indicating the target premises, and identity information indicating the publishing user of the target post.
Optionally, the related information may include at least: the identity of the target house and the identity of the publishing user. For example, the identity of the target house may be address information of the target house, or an identification number assigned by the server to a house corresponding to each post on the house trading platform, where the identification number is used to uniquely identify one house. For example, the address information of the target premises may include one or more of: and the address information with finer granularity, such as the geographic coordinate information of the target house, the information of the number plate of the street where the target house is located, the cell address of the target house, or the cell floor address of the target house, and the like. In addition, the identity of the publishing user may be publishing user information. For example, publishing user information may include: and issuing the user name, account number and the like of the user. In one possible scenario, the target house may be published by multiple different house broker agents belonging to the same company. The publishing user information may be information of a company where the house broker is located. In another possible scenario, the target house may not be allowed to be published by multiple different house broker agents belonging to the same company. The posting user information may be the house broker that actually posted the targeted post. It is understood that the specific identity granularity of the published user information may be set according to the actual scene requirement, which is not limited in the embodiment of the present application.
It should be noted that the related information may further include one or more of the following items: basic information of the house, transaction information of the house, etc. Wherein the basic information comprises one or more of: area information, orientation information, functional partition information and decoration facility information of the house. The transaction information includes one or more of: price information of the house, payment means information, tradable time, etc. Optionally, in a house renting scenario, the tradeable time may be a house rentable date. In a house buying and selling scenario, the tradeable time may be a house sellable date.
Step 202, under the condition that it is determined that similar images matched with the images exist in the house source library, acquiring a characteristic information set of the similar images, wherein the characteristic information set is at least used for reflecting the exposure degrees of the similar images.
In the embodiment of the application, the server side can judge whether the similar image matched with the image of the target house exists in the house source library or not so as to determine whether the image of the target house in the house source library is possibly published for multiple times or not, and therefore whether the image is the probability of stealing images of other house images is judged.
And under the condition that the similar image matched with the image of the target house does not exist in the house source library, the image of the target house is shown to be released on the house trading platform for the first time, and the probability that the image of the target house is a stolen image is low. The server may not perform the off-shelf operation on the target post. Optionally, in the case that the target post is an online post, no operation may be performed on the target post to continue the online target post. When the target post is the post in the audit, other audit operations can be performed on the target post, or the target post is considered to pass the audit, and an online operation is performed on the target post, so that the target post is the online target post.
In the case that it is determined that a similar image matching the image of the target house exists in the house source library, it indicates that the image of the target house may be released on the house trading platform for many times, and the probability that the image of the target house is a stolen image is high. The server can obtain a feature information set of the similar image, wherein the feature information set is at least used for reflecting the exposure degree of the similar image. So as to perform subsequent operations according to the feature information sets of the similar images, thereby determining whether the image of the target house is a pirate image. And executes corresponding processing according to the determination result. In the embodiment of the present application, the implementation manner of the server determining whether there is a similar image matching with the image in the house source library may be described in detail later, which is not described herein for a while.
Optionally, the feature information set of the similar image acquired by the server may at least include: identity of similar premises and identity of the publishing user. The similar house refers to the real house corresponding to the similar image, namely the shooting place of the similar image. In an alternative implementation, since the trueness of posting a new house for the first time is generally high, the similar house corresponding to the similar image may refer to the house corresponding to the first posting, and the first posting refers to posting a similar image for the first time. The identity of the similar house may refer to the identity of the house to which the first post corresponds. The identity of the publishing user may refer to the identity of the publishing user that originated the post.
For example, the identity of the similar house may be address information of the similar house, or an identification number assigned by the server to the house corresponding to each post on the house trading platform, where the identification number is used to uniquely identify one house. For example, address information for similar premises may include one or more of: and the address information with finer granularity, such as the geographic coordinate information of the target house, the information of the number plate of the street where the target house is located, the cell address of the target house, or the cell floor address of the target house, and the like. In addition, the identity of the publishing user may be publishing user information. For example, publishing user information may include: and issuing the user name, account number and the like of the user. In one possible scenario, similar houses may be published by multiple different house broker agents belonging to the same company. The publishing user information may be information of a company where the house broker is located. In another possible scenario, similar houses may not be allowed to be released by multiple different house broker agents belonging to the same company. The posting user information may be the house broker that actually posted the targeted post. It is understood that the specific identity granularity of the published user information may be set according to the actual scene requirement, which is not limited in the embodiment of the present application. The expression mode of each feature information in the feature information set and the corresponding related information, such as address information and information of a publishing user, should be the same, so that the service end can process the two information conveniently.
In this embodiment, the feature information set may further include at least one of: the total number of posts distributed with similar images (also called post volume), and the number of different addresses (also called address volume) in the address information of the house recorded by each post distributed with similar images. For example, in the case where the address information of the similar house is the cell address of the similar house, the address amount may refer to the number of different cells in the cell address of the house recorded in each post among all posts where the similar image is issued.
And step 203, updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house released by the same releasing user.
In an alternative implementation manner, the process of the server updating the feature information set according to the related information may include the following steps 2031 to 2032.
Step 2031, consistency matching is performed between the address information in the related information and the address information in the feature information set, the address information in the feature information set is updated to a first feature value if matching is successful, and the address information in the feature information set is updated to a second feature value if matching is unsuccessful.
In the embodiment of the application, the server side can perform consistency matching on the address information in the related information and the address information in the feature information set. And under the condition that the address information in the related information is successfully matched with the address information in the characteristic information set, indicating that the target house and the similar house belong to the same house. The server may update the content of the address information in the feature information set to the first feature value. The first eigenvalue is used to represent that the target house originates from the same house as the similar house. And under the condition that the address information in the related information is not successfully matched with the address information in the characteristic information set, indicating that the target house and the similar house do not belong to the same house. The server may update the content of the address information in the feature information set to the second feature value. The second feature value is used to indicate that the similar house and the target house do not originate from the same house.
For example, the server may compare address information in the related information with address information in the feature information set. When the address information of the two is the same, the matching is successful. The server side updates the content of the address information in the characteristic information set to be the first characteristic value. When the two address information are not the same, the two are not matched successfully. The server side updates the content of the address information in the characteristic information set to a second characteristic value. Wherein the first characteristic value may be smaller than the second characteristic value. Therefore, under the condition that the theft probability of the similar image is determined subsequently according to each feature information in the updated feature information set and the weight corresponding to each feature information, the influence of the theft probability on the determination that the target house is the theft graph is small when the house is similar to the target house. The greater the probability of theft of a similar house, the greater the probability that the target house is a theft graph. For example, the address information in the updated feature information set may be a boolean value, the first feature value is 0, and the second feature value is 1.
Step 2032, consistency matching is performed between the publishing user information in the related information and the publishing user information in the feature information set, the publishing user information in the feature information set is updated to a first feature value when matching is successful, and the publishing user information in the feature information set is updated to a second feature value when matching is unsuccessful.
In this embodiment of the application, the process of the server performing consistency matching on the published user information in the relevant information and the published user information in the feature information set, and updating the content of the published user information in the feature information set according to the matching result may refer to the process of the server updating the content of the address information in the feature information set according to the consistency matching result between the address information in the relevant information and the address information in the feature information set in step 2031, which is not described in detail in this embodiment of the application.
In another optional implementation manner, the process of the server updating the feature information set according to the related information may include: and the server inputs the relevant information and each characteristic information in the characteristic information set into the information matching model. And updating to the characteristic information set according to the output information of the information matching model. The information matching model may be configured to output feature information at least indicating whether the similar house and the target house are from the same house released by the same releasing user according to the identity of the target house and the identity of the releasing user in the related information, and the identity of the similar house and the identity of the releasing user in the feature information set. For example, the information matching model may be used to output a matching result vector
Figure BDA0003240416460000091
u1 indicates whether the similar house and the target house originate from the same house.u2 indicates whether similar houses and target houses are the houses released by the same releasing user. u1 and u2 may be the first eigenvalue or the second eigenvalue.
And 204, determining the stealing probability of the similar images according to the feature information in the updated feature information set and the weight corresponding to the feature information.
Optionally, the process of determining the theft probability of the similar image according to each feature information in the updated feature information set and the weight corresponding to each feature information may include: and the server side carries out weighted summation operation on each feature information in the updated feature information set according to the weight corresponding to each feature information to obtain the stealing probability of the similar images. Or the server side carries out weighted average operation on each feature information in the updated feature information set according to the weight corresponding to each feature information to obtain the stealing probability of the similar images. Wherein the stealing probability of the similar images is used for representing the stealing probability of the similar images. The greater the probability of stealing the similar image, the greater the probability that the image of the target house corresponding to the target post is a stolen image.
In an optional implementation manner, if each piece of feature information in the updated feature information set is non-numerical information, the server may convert each piece of feature information into a corresponding feature value, and determine, according to the feature value corresponding to each piece of feature information, a theft probability of the similar image according to a weight corresponding to each piece of feature information. Wherein the characteristic value is used for characterizing the content of the characteristic information.
Optionally, for any feature information, the server may store in advance a characterization value corresponding to the feature information when the feature information is different content. The value of the characteristic value can be determined based on actual conditions. For example, the value of the characterization value may be determined according to the degree of influence on the theft probability of the similar image when the characteristic information is different contents, or the degree of influence on the theft graph of the image of the target house determined based on the theft probability of the similar image. For example, assume that the updated feature information set includes target feature information indicating that similar houses and target houses originate from the same house released by the same releasing user. The server can query a feature value 0 corresponding to the target feature information when the target feature information is used for indicating that the similar house and the target house are from the same house issued by the same issuing user. And converting the target characteristic information into a characteristic value 0 in the updated characteristic set.
In an example, the theft probability is taken as each feature information in the updated feature information set, and the weighting summation operation is performed according to the weight corresponding to each feature information to determine the feature information. The server side can generate attribute feature vectors corresponding to the similar images according to the feature information in the updated feature set. The weights corresponding to the characteristic information form a weight matrix. And the server performs dot product operation on the attribute feature vector and the weight matrix to obtain the stealing probability of the similar image.
For example, the set of feature information of similar images includes: address information a1, post user information a2, post volume A3, and address volume a 4. With the first implementation manner of step 203, the feature information of the similar images is updated, so that the updated feature information set includes: the updated feature set of the similar images comprises: address information B1, posting user information B2, posting volume B3, and address volume B4. The values of the address information B1 and the issuing user information B2 are a first characteristic value or a second characteristic value. Then the attribute feature vector
Figure BDA0003240416460000101
The weight corresponding to the address information is W1, the weight corresponding to the posting user information is W2, the weight corresponding to the posting volume is W3, and the weight corresponding to the address volume is W4. The permutation matrix W of the weight matrixT＝[W1,W2,W3,W4]. The probability of theft of similar images
Figure BDA0003240416460000102
In this embodiment of The application, The weight corresponding to each feature information used by The server to calculate The theft probability of The similar image may be pre-stored, or The weight corresponding to each feature information may also be determined by using The Analytic Hierarchy Process (AHP). In the embodiment of the present application, the example in which the weight corresponding to each feature information is determined by using AHP is described. The method may further include the following steps S1 to S4 before step 204.
Step S1, the server compares the initial weights of any two pieces of feature information in the feature information sets of the similar images to generate a judgment matrix (also called a pair comparison matrix). And the value of each element in the judgment matrix is the initial weight ratio after comparing every two pieces of feature information in the feature information of the similar house.
In the embodiment of the application, the initial weight ratio of the two pieces of feature information after comparison in each piece of feature information is the relative weight of the two pieces of feature information. The relative weight indicates the respective degree of importance levels of the two characteristic information.
For example, if the initial weight ratio of the feature information a to the feature information B is 3, this indicates that the initial weight value of the feature information a is three times as large as the initial weight value of the feature information B. Optionally, the initial weight ratio of the two pieces of feature information in each piece of feature information after comparison may be determined according to the degree of influence of each piece of feature information on the final decision result for determining that the image of the target house belongs to the stealing picture. For example, if the initial weight ratio of the feature information a to the feature information B is 3, the feature information a is considered to be three times as large as the degree of influence of the feature information B on the decision result. On the contrary, if the initial weight ratio of the feature information a to the feature information B is 1/3, the feature information B is considered to be three times as large as the influence degree of the feature information a on the decision result.
Optionally, the value of each element in the determination matrix, that is, the corresponding importance degree level of any two pieces of feature information, may be represented by a natural number. The value of the natural number can be determined according to actual conditions. In the examples of the present application, natural numbers 1 to 9 are used. Wherein 1 indicates that one of the two pieces of feature information compared with each other is the same in relative importance with respect to the other, and 9 indicates that one of the two pieces of feature information compared with each other is the largest in relative importance with respect to the other, and vice versa. Referring to table 1, it is listed in table 1 that natural numbers 1 to 9 indicate the importance of two characteristic information compared with each other.
TABLE 1
Figure BDA0003240416460000111
For example, it is assumed that the characteristic information of similar houses includes: feature information B1, feature information B2, feature information B3, and feature information B4. The matrix table corresponding to the judgment matrix Aij generated by the server is table 2.
TABLE 2
B B1 B2 B3 B4
B1 1 9 7 5
B2 1/9 1 1/3 1/5
B3 1/7 3 1 1/3
B4 1/5 5 3 1
And step S2, the server determines the relative weight of each characteristic information based on the judgment matrix.
In this embodiment of the application, the process of the server determining the relative weight of each feature information may include the following steps S21 to S24.
And step S21, the server side respectively carries out summation operation on each column of the judgment matrix to obtain a processed first matrix.
For example, continuing with the example of step S1, as shown in table 3, table 3 shows that the summation result of each column in the determination matrix is the matrix table corresponding to the first matrix. The Ai1(B1) column is summed, i.e., 1+1/9+1/7+1/5, resulting in a SUM of 1.454.
TABLE 3
B B1 B2 B3 B4
B1 1 9 7 5
B2 1/9 1 1/3 1/5
B3 1/7 3 1 1/3
B4 1/5 5 3 1
SUM 1.454 18.000 11.333 6.533
And step S22, the server side normalizes each column in the first matrix to obtain a processed second matrix.
For example, continuing with the example of step S21, as shown in table 4, table 4 shows that the result of normalization processing performed on each column in the first matrix is the matrix table corresponding to the second matrix. The Ai1(B1) columns were normalized to 0.688, 0.076, 0.098, 0.138, 1.
TABLE 4
B B1 B2 B3 B4
B1 0.688 0.500 0.618 0.0765
B2 0.076 0.056 0.029 0.031
B3 0.098 0.0167 0.088 0.051
B4 0.138 0.278 0.265 0.153
SUM 1 1.001 1.000 1.000
And step S23, the server-side respectively carries out summation operation on each row of the second matrix to obtain a processed third matrix.
Illustratively, continuing with the example of step S22, as shown in table 5, table 5 shows summation results of rows in the second matrix, that is, shows a matrix table corresponding to the third matrix. The A1j (B1) rows are summed, i.e., 0.688+0.500+0.618+0.0765, resulting in a SUM 2.571.
TABLE 5
B B1 B2 B3 B4 SUM
B1 0.688 0.500 0.618 0.0765 2.571
B2 0.076 0.056 0.029 0.031 0.192
B3 0.098 0.0167 0.088 0.051 0.4.4
B4 0.138 0.278 0.265 0.153 0.834
SUM 1 1.001 1.000 1.000 4.001
And step S24, the server side performs normalization processing on each row in the third matrix to obtain a processed fourth matrix.
For example, continuing with the example of step S23, as shown in table 6, table 6 shows the result of normalization processing performed on Ai5(SUM) column in the third matrix, where the result is located in Ai6(W) column, that is, the matrix table corresponding to the fourth matrix. The Ai5(SUM) columns were normalized to 0.643, 0.048, 0.101, 0.208, and 1.
TABLE 6
Figure BDA0003240416460000131
Figure BDA0003240416460000141
And step S3, the server side carries out consistency check on the judgment matrix. In the case where the verification is passed, step S4 is executed; in the case where the verification fails, step S1 is performed to reconstruct the determination matrix.
In the embodiment of the present application, in the process of generating the determination matrix for the initial weight comparison result of any two pieces of feature information between each piece of feature information in step S1, if it is determined that the importance degree of the feature information a is greater than that of the feature information B, and the importance degree of the feature information B is greater than that of the feature information C. The degree of importance of the feature information a needs to be greater than that of the feature information C. If the importance degree of the finally obtained feature information a in the generated judgment matrix needs to be smaller than that of the feature information C, the consistency of the judgment matrix is not established, that is, the verification fails.
For example, if the initial weight ratio of the feature information a to the feature information B is 3, and the initial weight ratio of the feature information B to the feature information C is 3, the initial weight ratio of the feature information a to the feature information C should be 6. Therefore, if the initial weight ratio of the feature information a to the feature information C in the generated determination matrix is not 6, even if it is 5 or 7 which is close to 6, the consistency of the determination matrix is not established, that is, the verification fails.
The process of the server side performing consistency check on the determination matrix may include steps S31 to S33.
And step S31, the server calculates the maximum characteristic root of the judgment matrix.
And step S32, the server calculates the consistency index of the judgment matrix.
In the embodiment of the application, the consistency index c.i. satisfies:
Figure BDA0003240416460000142
wherein λ ismaxIn order to judge the maximum characteristic root of the matrix, n is the order of the judgment matrix.
And step S33, the server calculates the consistency ratio of the judgment matrix according to the consistency index.
In the embodiments of the present application, the consistency ratio c.r. satisfies:
Figure BDA0003240416460000151
wherein, C.I. is the consistency index of the judgment matrix, and R.I. is the average random consistency index. The average random consistency index is a constant and is related to the order of the judgment matrix, and can be obtained by inquiring in a scale according to the order of the matrix.
Step S34, the server determines whether the consistency ratio of the judgment matrix is less than 0.1. When the consistency ratio is less than 0.1, determining that the verification is passed; and determining that the verification is failed when the consistency ratio is not less than 0.1.
For example, the server compares the consistency ratio of the judgment matrix with the size of 0.1, so as to determine whether the consistency ratio is less than 0.1.
Step S4, the server determines the weight corresponding to each feature information.
In the embodiment of the present application, when the server determines that the determination matrix passes the consistency check, the server uses the numerical values of Ai7 (i.e., W%) columns as the weights corresponding to the feature information represented by the corresponding rows. As an example, a matrix table corresponding to the fourth matrix shown in table 6 is taken as an example. A17: 64.3% is the weight corresponding to the characteristic information B1. A27: 4.80% is the weight corresponding to the characteristic information B2. A37: 10.1% is the weight corresponding to the characteristic information B3. A47: 20.8% is the weight corresponding to the characteristic information B4.
And step 205, putting down the target post under the condition that the image is determined to be a stolen image according to the stealing probability.
In the embodiment of the application, the server side can determine whether the image of the target house is a stolen image or not according to the relationship between the stolen probability and the first numerical threshold. And under the condition that the stealing probability is greater than the first numerical threshold, the stealing probability of the similar image is high, the probability that the image of the target house similar to the similar image is a stealing image is high, the image of the target house is determined to be the stealing image, and the server can put down the target post. The method comprises the steps that when a target post is an online post, a server-side shelf target post can mean that the server-side performs offline processing on the online target post, and when the target post is an audit post, the server-side shelf target post can mean that the server-side determines that the audit post is not approved and prohibits the target post from being added into a house source library. And under the condition that the stealing probability is not greater than the first numerical threshold, the stealing probability of the similar image is low, the probability that the image of the target house similar to the similar image is a stealing image is low, and the server side can not execute the operation of putting down the shelf on the target post if the image of the target house is determined not to be the stealing image. Optionally, in the case that the target post is an online post, no operation may be performed on the target post to continue the online target post. When the target post is the post in the audit, other audit operations can be performed on the target post, or the target post is considered to pass the audit, and an online operation is performed on the target post, so that the target post is the online target post.
It should be noted that, in an alternative case, the server determines that the house source library has a plurality of similar images matching the image of the target house. The server can acquire the feature information set of the similar images for any one of the plurality of similar images. And updating the characteristic information set according to the related information. And determining the stealing probability of the similar images according to the feature information in the updated feature information set and the weight corresponding to the feature information. The process that the server determines the image as the stealing graph according to the stealing probability can include: and the server determines the comprehensive probability that the image of the target house is the stealing image according to the corresponding stealing probability of each similar image. And determining the image as a stolen image under the condition that the comprehensive probability is greater than a second numerical threshold. Correspondingly, in the case that the comprehensive probability is not greater than the second numerical threshold, the image is determined not to be a stolen image, and the server side may not perform any operation on the target post to continue the online target post. The comprehensive probability may be the sum of the theft probabilities corresponding to the similar images, or may also be the average value of the theft probabilities corresponding to the similar images.
The comprehensive probability can be the sum of the corresponding theft probabilities of the similar images. Alternatively, the integrated probability may be an average of the theft probabilities corresponding to the respective similar images. Or, the server can also take the corresponding weighted sum of the stolen probabilities as the integrated probability. Or, the server can also take the weighted average of the corresponding stolen probabilities as the comprehensive probability. Wherein, the weight of each theft probability can be related to the similarity degree between the similar image corresponding to the theft probability and the image of the target house. The higher the degree of similarity between the similar image and the image of the target house, the greater the weight of the theft probability corresponding to the similar image.
In another alternative, the number of the images of the target house is multiple, and the server determines that similar images matching the multiple images exist in the house source library. And matching the image of the target house with the similar images one by one. For any image in the plurality of images, the server may acquire a feature information set of a similar image when determining that the similar image matching the image exists in the house source library. And updating the characteristic information set according to the related information. And determining the stealing probability of the similar images according to the feature information in the updated feature information set and the weight corresponding to the feature information. I.e. the probability of theft of a similar image matching each image. The server determines the image as a stealing graph according to the stealing probability, and the method comprises the following steps: and the server determines the target probability that the image of the target post is the stolen image according to the stealing probability of each similar image. And under the condition that the target probability is greater than the numerical threshold, determining that the images of the target posts are the stealing images. Correspondingly, under the condition that the target probability is not greater than the numerical threshold, the images of the target posts are determined not to be stolen images.
The process of determining the target probability that the image of the target post is the stolen image by the server side may include: the server side can take the sum of the stealing probabilities of the similar images as the target probability. Or, the server may use the average value of the stealing probabilities of the similar images as the target probability.
In yet another alternative, the number of images of the target house is plural, and the server determines that similar images matching the plural images exist in the house source library. And for each image in the images of the target houses, the server determines that the number of similar images matched with the images in the house source library is multiple. For any image in the plurality of images, the server may determine the theft probability of each similar image in the plurality of similar images matching the image. And then, according to the stealing probability of each similar image, determining the comprehensive probability that the image is a stealing graph. And determining the target probability that the image of the target post is a stolen image according to the comprehensive probability of each image in the plurality of images of the target house.
The implementation manner of the server side for determining the stealing probability of each similar image can refer to the server side to acquire the feature information set of the similar images. And updating the characteristic information set according to the related information. And determining the implementation mode of the stealing probability of the similar images according to the feature information in the updated feature information set and the weight corresponding to the feature information. And the target probability may be the sum of the integrated probabilities of the respective images, or the target probability may be the average of the integrated probabilities of the respective images.
It should be noted that, in any of the foregoing three optional cases, the server acquires the feature information set of the similar image. And updating the characteristic information set according to the related information. According to each feature information in the updated feature information set and the weight corresponding to each feature information, an explanation and implementation manner of the theft probability of the similar image is determined, which may refer to the relevant explanation and implementation manners in the foregoing step 202 to step 204, and details of this embodiment of the present application are not repeated.
In the embodiment of the application, whether a similar image of the target house exists in the house source library or not is determined by matching the image of the target house corresponding to the target post with the house image in the house source library. And under the condition that the similar images exist in the house source library, determining the stealing probability of the similar images according to the feature information in the updated feature set corresponding to the similar images. Therefore, based on the principle that the higher the stealing probability of the similar image is, the higher the probability that the image of the target house similar to the similar image is the stolen image is, the authenticity identification of the image of the target house corresponding to the target post is realized. Since each feature information in the updated feature set is at least used to indicate whether the similar house and the target house are from the same house issued by the same issuing user, and whether the similar house and the target house corresponding to the similar image are the same house issued by the same issuing user, the determination of whether the image of the target house is a stolen image is greatly influenced. (for example, if the similar house and the target house are the same house released by the same publisher, the probability that the image of the target house steals the similar image is small even if the exposure degree of the similar image is large, and if the similar house and the target house are not the same house released by the same publisher, the probability that the image of the target house steals the similar image is large even if the exposure degree of the similar image is small) therefore, the influence of the image of the same house released by the same publisher on the identification of the image of the target house as a stolen image is taken into consideration, and the accuracy of the identification of whether the image of the target house is a stolen image is further improved. The probability of spreading the false information on the house trading platform is effectively reduced, the accuracy of the house property information displayed in the house trading platform is improved, unnecessary troubles brought by posts distributed with the false information to users for trading house properties are avoided, and the user viscosity of the house trading platform is improved.
In this embodiment of the present application, before step 202, the method further includes: the server side judges whether similar images matched with the images exist in the house source library or not. In the case that it is determined that similar images matched with the images exist in the house source library, the server executes step 202; in the case that it is determined that a similar image matching the image does not exist in the house source library, the server may not perform a shelving operation on the target post.
Optionally, as shown in fig. 3, the process of the server determining whether a similar image matching the image exists in the house source library may include:
step 301, the server side extracts first feature information of the image.
Alternatively, the server may perform Discrete Cosine Transform (DCT) on the image of the target house by using an image perception (pHash) algorithm to convert the image from a spatial domain to a frequency domain, so as to obtain a transformed image. And taking the DCT mean value of the transformed image as the first characteristic information of the image. The first feature information is also called image fingerprint.
Or, the process of extracting the first feature information of the image by the server may include: the server side inputs the image into the image feature extraction model to obtain first feature information. Wherein, the first feature information can be deep features of the image feature extraction model. For example, the first feature information may be feature information of a last layer of the image feature extraction model. The image feature extraction model is obtained by training a plurality of sample data. The sample data includes: a similar image pair, a dissimilar image pair, a first label corresponding to the similar image pair, and a second label corresponding to the dissimilar image pair. The first label is used for indicating that the images included in the similar image pair are similar images to each other. The second label is used to indicate that the dissimilar image pair comprises images that are not similar to each other. For example, the content of the first tag may be 1; the content of the second tag may be 0.
Optionally, the image feature extraction model may include, but is not limited to: ResNet18 network (residual deep learning model, a convolutional neural network structure based on residual learning), convolutional neural network. In the case where the image feature extraction model is a ResNet18 network, the ResNet18 network may include at least 15 convolutional layers. For example, a ResNet18 network may include 15 convolutional layers. Alternatively, the ResNet18 network may include 16 convolutional layers; alternatively, the ResNet18 network may include 17 convolutional layers and 1 fully-connected layer. When the ResNet18 network includes 17 convolutional layers and 1 fully-connected layer, the feature information of the image output by the ResNet18 network is a 512-dimensional feature. And in the case that the ResNet18 network does not include the last two convolutional layers and the last fully connected layer, that is, the ResNet18 network includes 15 convolutional layers, the feature information of the image output by the ResNet18 network is a 128-dimensional feature. Therefore, the ResNet18 network comprising 15 convolution layers outputs the first characteristic information, so that the output dimensionality is reduced on the basis of ensuring that the characteristic effect of the input image can be fully expressed, the difficulty degree of subsequent operation by adopting the dimensionality characteristic is reduced, and the subsequent processing efficiency is improved.
In this embodiment of the application, before the server determines whether a similar image matching the image exists in the house source library, the method may further include: and the server executes the training process of the image feature extraction model. Take the example where the image feature extraction network is the ResNet18 network. The training process may include: the server inputs sample data comprising a plurality of image pairs into the ResNet18 network, resulting in a category for each image pair identifying whether the image pair belongs to a similar image pair or a dissimilar image pair. Wherein the image pairs are a combination of similar and dissimilar image pairs. The label corresponding to each image pair, and the class output by the ResNet18 network, are input into a loss function to determine a loss value. Under the condition that the loss value does not reach a preset threshold value, continuously inputting sample data according to parameters in the ResNet18 network adjusted by the optimizer; and in the case that the loss value reaches a preset threshold value, taking the ResNet18 network as an image feature extraction model.
Optionally, the loss function of the image feature extraction model is a euclidean distance loss function, and a power factor in the euclidean distance loss function may be 8. The Euclidean distance loss function satisfies:
Figure BDA0003240416460000201
wherein E iskLoss values for the model are extracted for the image features.
Figure BDA0003240416460000202
A category of the image pair output for the image feature extraction model. y isjLabels for corresponding pairs of images. l is the total number of image pairs in the training data.
In the embodiment of the application, by using the mode of extracting the image by using the image feature extraction model, deeper and abstract feature expressions in the image can be automatically learned from the image of the provided sample data, the method is a global feature extraction method, has adaptability to most image processing operations, does not need to manually extract features, and has higher flexibility. Moreover, when the loss function is the Euclidean distance loss function, and the power factor in the Euclidean distance loss function is 8, that is, the loss function is the power of 8 of the Euclidean distance, the training weight of the difficult sample is increased, and the feature extraction performance of the image with large angle change is improved.
Step 302, the server determines a target image set in the house source library, which belongs to the same category as the images, according to the first characteristic information.
In an alternative implementation, the first feature information may be a first feature vector. For example, the server may extract a 128-dimensional floating point type vector for the image through the ResNet18 network. The server side can determine a target image set by adopting an IndexIVFFlat retrieval method. For example, the process that the server determines, according to the first feature information, a target image set in the house source library that belongs to the same category as the image may include: and the server determines a target image set from each image set according to the first characteristic vector and the distance between the clustering centers corresponding to each image set in the house source library, wherein each image set is determined by adopting a k-means clustering algorithm. The k-means clustering algorithm is also called k-means clustering algorithm (k-means clustering algorithm) which is a clustering analysis algorithm for iterative solution.
The server side can determine the target image set according to the first characteristic vector and Euclidean distances between clustering centers corresponding to the image sets included in the house source library. For example, the server may sequentially calculate the euclidean distance between the first feature vector and the cluster center corresponding to each image set. And taking the image set with the minimum Euclidean distance as a target image set. Or, the server may sequentially calculate the euclidean distance between the first feature vector and the clustering center corresponding to each image set. And after sequencing the Euclidean distances from small to large, all the image sets corresponding to the Euclidean distances of the front appointed number are used as target image sets.
In the embodiment of the application, each image set included in the house source library can be determined by adopting a k-means clustering algorithm. Alternatively, the sets of images included in the house source library may be determined in other ways. For example, human division is based on actual applications. In the case that each image set can be determined by using a k-means clustering algorithm, before the server determines whether similar images matching the images exist in the house source library, the method may further include: and the server side acquires all images in the house source library. And (5) clustering all the images by adopting a k-means clustering algorithm to establish k image sets (clusters). k is a positive integer.
The server side clusters all the images by using a k-means clustering algorithm, and the process of establishing k image sets may include the following steps B10 to B40.
And step B10, randomly selecting the feature vectors of the k images from the house source library as initial clustering centers.
And step B20, respectively calculating the Euclidean distance between the feature vector of each image in the room source library and each cluster center. And dividing the image into image sets corresponding to the clustering centers closest to the image sets to obtain divided image sets.
And step B30, recalculating the mean value of the feature vector of each image in each image set as a new clustering center.
And step B40, judging whether the new clustering center is the same as the previous clustering center. If yes, the calculation is terminated, and each image set is determined. If not, step B20 is repeated.
In another optional implementation manner, the server may further divide the images with the similarity belonging to the same setting range in the multiple setting ranges into one image set according to the similarity between any two images in the room source library, so as to obtain multiple image sets. And aiming at each image set, the server calculates the Euclidean distance between the feature vector of any image in the image set and the first feature vector. And taking the image set with the Euclidean distance smaller than a second set distance threshold value as a target image set. Of course, there may be other ways for the server to determine, according to the first feature information, a target image set in the house source library that belongs to the same category as the image, which is not described in detail in this embodiment of the present application.
Step 303, the server determines that similar images matched with the images exist in the house source library under the condition that similar images exist in the target image set according to the matching result of the first characteristic information and the second characteristic information corresponding to each image in the target image set.
In this embodiment of the application, the server may determine whether a similar image exists in the target image set according to a matching result of the first feature information and the second feature information corresponding to each image in the target image set. And in the case that similar images exist in the target image set, determining that similar images matched with the images exist in the house source library. And under the condition that the similar images do not exist in the target image set, determining that the similar images matched with the images do not exist in the house source library.
Optionally, the first feature information includes a first feature vector, and the second feature information includes a second feature vector. The process of determining that similar images exist in the target image set by the server according to the matching result of the first feature information and the second feature information corresponding to each image in the target image set may include: and the server side takes the image in the target image set, wherein the distance between the second characteristic vector and the first characteristic vector is smaller than a set distance threshold value, as a similar image.
The server side can calculate the Euclidean distance between the second characteristic vector and the first characteristic vector of each image in the target image set. And determining that similar images exist in the target image set when the Euclidean distance between any second feature vector and the first feature vector is smaller than a set distance threshold. And taking the image of which the Euclidean distance between the second characteristic vector and the first characteristic vector is smaller than a set distance threshold value as a similar image. And under the condition that the Euclidean distance between the second characteristic vector and the first characteristic vector is not less than a set distance threshold, determining that no similar image exists in the target image set.
In this way, the server side can determine the target image set in the house source library, which belongs to the same category as the images, according to the first characteristic information, so as to search for similar images in the target image set. Compared with the method for directly searching similar images in all images in the house source library, the method reduces the waste of resources and time caused by direct violent searching and improves the searching efficiency.
In summary, the information management method provided by the embodiment of the present application obtains the image of the target house and the related information corresponding to the target post. And under the condition that the similar images matched with the images exist in the house source library, acquiring a characteristic information set of the similar images, wherein the characteristic information set is at least used for reflecting the exposure degree of the similar images. And updating the feature information set according to the related information, and determining the stealing probability of the similar images according to the feature information in the updated feature information set and the weight corresponding to the feature information. And in the case that the image is determined to be a stolen image according to the stealing probability, putting down the target post.
According to the technical scheme, whether similar images of the target house exist in the house source library or not is determined by matching the images of the target house corresponding to the target posts with the images of the house in the house source library. And under the condition that the similar images exist in the house source library, determining the stealing probability of the similar images according to the feature information in the updated feature set corresponding to the similar images. Therefore, based on the principle that the higher the stealing probability of the similar image is, the higher the probability that the image of the target house similar to the similar image is the stolen image is, the authenticity identification of the image of the target house corresponding to the target post is realized. Since each feature information in the updated feature set is at least used to indicate whether the similar house and the target house are from the same house issued by the same issuing user, and whether the similar house and the target house corresponding to the similar image are the same house issued by the same issuing user, the determination of whether the image of the target house is a stolen image is greatly influenced. (for example, if the similar house and the target house are the same house released by the same publisher, the probability that the image of the target house steals the similar image is small even if the exposure degree of the similar image is large, and if the similar house and the target house are not the same house released by the same publisher, the probability that the image of the target house steals the similar image is large even if the exposure degree of the similar image is small) therefore, the influence of the image of the same house released by the same publisher on the identification of the image of the target house as a stolen image is taken into consideration, and the accuracy of the identification of whether the image of the target house is a stolen image is further improved. The probability of spreading the false information on the house trading platform is effectively reduced, the accuracy of the house property information displayed in the house trading platform is improved, unnecessary troubles brought by posts distributed with the false information to users for trading house properties are avoided, and the user viscosity of the house trading platform is improved.
The embodiment of the present application further describes an information management method provided by the present application by taking the following example as an example. As shown in fig. 4, the information management method may be applied to the implementation environment shown in fig. 1, and executed by the server in fig. 1. The information management method comprises the following steps:
step 401, obtaining a picture of a target house, cell address information and publishing user information corresponding to the target post.
The explanation and implementation of step 401 may refer to the explanation and implementation of step 201, which is not described in detail in this embodiment of the present application.
Step 402, determining the image fingerprint of the picture of the target house by using the ResNet18 model.
The explanation and implementation of step 402 may refer to the extraction model of the image feature input by the server in step 301 to obtain an explanation and implementation of the first feature information, which is not described in detail in this embodiment of the present application.
And step 403, acquiring all pictures in the house source library.
And step 404, adopting IndexIVFFlat retrieval to judge whether similar pictures exist in the house source library or not according to the image fingerprints. If not, go to step 405; if yes, go to step 406.
In this embodiment, the following method shown in fig. 5 may be referred to in the process that the server searches whether similar pictures exist in the house source library by using indexivflat according to the image fingerprint. The embodiments of the present application are not described in detail herein.
And step 405, not processing, and returning a room source detection result.
And step 406, calculating the score of the stolen picture according to the cell address information, the release user information and the like of the similar picture.
In the embodiment of the application, the stealing graph score can be the stealing probability of the similar image. In the case that there are a plurality of similar pictures matching the picture of the target house, the score of the stolen picture may be a comprehensive probability that the image of the target house is the stolen picture. In the case that similar pictures matching the pictures of a plurality of target houses exist in the house source library, the score of the stolen image can be the target probability that the image of the target post is the stolen image. In the case that similar pictures matched with the pictures of the target houses exist in the house source library, and for each picture of the pictures of the target houses, the similar pictures matched with the pictures exist in the house source library, the picture stealing score can be the target probability that the image of the target post is the picture stealing. For the explanation and implementation of step 406, reference may be made to the explanation and implementation in step 202 to step 204, which is not described in detail in this embodiment of the application.
Step 407, judging whether the house source comprehensive stolen graph score is larger than a threshold value. If not, go to step 405; if yes, go to step 408.
And step 408, off-shelf processing is carried out on the house resources.
The explanation and implementation of step 407 and step 408 may refer to the explanation and implementation of step 204, which is not described in detail in this embodiment of the application.
The embodiment of the present application further describes, by taking the following example as an example, whether similar images exist in the house source library in the information management method provided by the present application. As shown in fig. 5, the information management method may be applied to the implementation environment shown in fig. 1, and executed by the server in fig. 1. The information management method comprises the following steps:
and step 501, acquiring all pictures in the house source library.
Step 502, clustering all images by using a k-means clustering algorithm, and establishing k clusters.
The explanation and implementation of step 502 may refer to the process in which the server uses a k-means clustering algorithm to cluster all the images in step 302 and establish k image sets, which is not described in detail in this embodiment of the present application.
And 503, searching the cluster with the closest distance between the cluster center and the characteristic vector of the picture of the target house from the k clusters.
The explanation and implementation of step 503 may refer to the process of determining the target image set in the room source library, which belongs to the same category as the image in step 302, which is not described in detail in this embodiment of the present application.
And step 504, searching a target vector of the picture closest to the characteristic vector under the cluster.
And 505, judging whether the distance between the searched target vector and the feature vector is smaller than a threshold value. If not, go to step 506. If yes, go to step 507.
The explanation and implementation of step 504 and step 505 may refer to the explanation and implementation of step 303, which is not described in detail in this embodiment of the present application.
Step 506, similar pictures do not exist in the house source library.
And step 507, similar pictures exist in the house source library.
In the embodiment of the application, whether a similar image of the target house exists in the house source library or not is determined by matching the image of the target house corresponding to the target post with the house image in the house source library. And under the condition that the similar images exist in the house source library, determining the stealing probability of the similar images according to the feature information in the updated feature set corresponding to the similar images. Therefore, based on the principle that the higher the stealing probability of the similar image is, the higher the probability that the image of the target house similar to the similar image is the stolen image is, the authenticity identification of the image of the target house corresponding to the target post is realized. Since each feature information in the updated feature set is at least used to indicate whether the similar house and the target house are from the same house issued by the same issuing user, and whether the similar house and the target house corresponding to the similar image are the same house issued by the same issuing user, the determination of whether the image of the target house is a stolen image is greatly influenced. (for example, if the similar house and the target house are the same house released by the same publisher, the probability that the image of the target house steals the similar image is small even if the exposure degree of the similar image is large, and if the similar house and the target house are not the same house released by the same publisher, the probability that the image of the target house steals the similar image is large even if the exposure degree of the similar image is small) therefore, the influence of the image of the same house released by the same publisher on the identification of the image of the target house as a stolen image is taken into consideration, and the accuracy of the identification of whether the image of the target house is a stolen image is further improved. The probability of spreading the false information on the house trading platform is effectively reduced, the accuracy of the house property information displayed in the house trading platform is improved, unnecessary troubles brought by posts distributed with the false information to users for trading house properties are avoided, and the user viscosity of the house trading platform is improved.
Referring to fig. 6, a block diagram of an information management apparatus according to an embodiment of the present application is shown. As shown in fig. 6, the information management apparatus 600 includes: an acquisition module 601, an update module 602, a determination module 603, and a control module 604.
The obtaining module 601 is configured to obtain an image and related information of a target house corresponding to the target post; the method comprises the steps of obtaining a characteristic information set of similar images under the condition that the similar images matched with the images exist in a house source library, wherein the characteristic information set is at least used for reflecting the exposure degrees of the similar images;
an updating module 602, configured to update the feature information set according to the relevant information, so that the updated feature information set is further used to reflect whether the target house and the similar house corresponding to the similar image originate from the same house issued by the same issuing user;
a determining module 603, configured to determine, according to each piece of feature information in the updated feature information set and a weight corresponding to each piece of feature information, a probability of stealing similar images;
and the control module 604 is used for placing the targeted post in the case that the image is determined to be a stolen image according to the stealing probability.
Optionally, the weight corresponding to each feature information is determined by using AHP.
Optionally, the related information at least includes: address information of the target house and issuing user information; the feature information set at least includes: address information of similar houses and issuing user information; the set of feature information further comprises at least one of: the total number of posts distributed with similar images and the number of different addresses in the address information of the house recorded by each post distributed with similar images.
Optionally, the control module 604 is further configured to:
determining the target probability that the image of the target post is the stolen image according to the stealing probability of each similar image;
and under the condition that the target probability is greater than the numerical threshold, determining that the images of the target posts are the stealing images.
Optionally, the apparatus further comprises: a matching module further configured to:
extracting first characteristic information of an image;
determining a target image set which belongs to the same category as the images in the house source library according to the first characteristic information;
and determining that similar images matched with the images exist in the house source library under the condition that the similar images exist in the target image set according to the matching result of the first characteristic information and the second characteristic information corresponding to the images in the target image set.
Optionally, the matching module is further configured to: inputting an image into an image feature extraction model to obtain first feature information, wherein the image feature extraction model is obtained by training a plurality of sample data, and the sample data comprises: a similar image pair, a dissimilar image pair, a first label corresponding to the similar image pair, and a second label corresponding to the dissimilar image pair.
Optionally, the image feature extraction model includes: a ResNet18 network including at least 15 convolutional layers; the loss function of the image feature extraction model is an Euclidean distance loss function, and the power factor in the Euclidean distance loss function is 8.
Optionally, the first feature information includes a first feature vector, and the matching module is further configured to:
and determining a target image set from each image set according to the first characteristic vector and the distance between the clustering centers corresponding to each image set in the house source library, wherein each image set is determined by adopting a k-means clustering algorithm.
Optionally, the first feature information includes a first feature vector, the second feature information includes a second feature vector, and the matching module is further configured to: and taking the image in the target image set, wherein the distance between the second characteristic vector and the first characteristic vector is smaller than a set distance threshold value, as a similar image.
Optionally, the updating module 602 is further configured to:
carrying out consistency matching on the address information in the related information and the address information in the feature information set, updating the address information in the feature information set to a first feature value under the condition of successful matching, and updating the address information in the feature information set to a second feature value under the condition of unsuccessful matching;
and carrying out consistency matching on the release user information in the related information and the release user information in the characteristic information set, updating the release user information in the characteristic information set to a first characteristic value under the condition of successful matching, and updating the release user information in the characteristic information set to a second characteristic value under the condition of unsuccessful matching.
In the embodiment of the application, whether a similar image of the target house exists in the house source library or not is determined by matching the image of the target house corresponding to the target post with the house image in the house source library. And under the condition that the similar images exist in the house source library, determining the stealing probability of the similar images according to the feature information in the updated feature set corresponding to the similar images. Therefore, based on the principle that the higher the stealing probability of the similar image is, the higher the probability that the image of the target house similar to the similar image is the stolen image is, the authenticity identification of the image of the target house corresponding to the target post is realized. Since each feature information in the updated feature set is at least used to indicate whether the similar house and the target house are from the same house issued by the same issuing user, and whether the similar house and the target house corresponding to the similar image are the same house issued by the same issuing user, the determination of whether the image of the target house is a stolen image is greatly influenced. Therefore, the influence of the image of the same house issued by the same issuing user on the image of the authentication target house is considered, and the accuracy of the authentication of whether the image of the target house is the stolen image is further improved. The probability of spreading the false information on the house trading platform is effectively reduced, the accuracy of the house property information displayed in the house trading platform is improved, unnecessary troubles brought by posts distributed with the false information to users for trading house properties are avoided, and the user viscosity of the house trading platform is improved.
The information management device provided by the embodiment of the application is provided with a functional module corresponding to the execution of the information management method, can execute the information management method provided by any embodiment of the application, and can achieve the same beneficial effects.
In another embodiment provided by the present application, there is also provided an electronic device, which may include: the processor executes the program to realize the processes of the information management method embodiment, and can achieve the same technical effects, and the details are not repeated here in order to avoid repetition.
For example, as shown in fig. 7, the electronic device may specifically include: a processor 701, a storage device 702, a display screen 703 with touch functionality, an input device 704, an output device 705, and a communication device 706. The number of the processors 701 in the electronic device may be one or more, and one processor 701 is taken as an example in fig. 7. The processor 701, the storage means 702, the display 703, the input means 704, the output means 705 and the communication means 706 of the electronic device may be connected by a bus or other means.
In yet another embodiment provided by the present application, a computer-readable storage medium is further provided, which stores instructions that, when executed on a computer, cause the computer to perform the information management method described in any of the above embodiments.
In yet another embodiment provided by the present application, there is also provided a computer program product containing instructions which, when run on a computer, cause the computer to perform the information management method of any of the above embodiments.
It is noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present invention.
While the present invention has been described with reference to the embodiments shown in the drawings, the present invention is not limited to the embodiments, which are illustrative and not restrictive, and it will be apparent to those skilled in the art that various changes and modifications can be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a U disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (13)

1. An information management method, characterized in that the method comprises:
acquiring an image and related information of a target house corresponding to the target post;
under the condition that a similar image matched with the image exists in a room source library, acquiring a characteristic information set of the similar image, wherein the characteristic information set is at least used for reflecting the exposure degree of the similar image;
updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house released by the same releasing user;
determining the probability of stealing the similar images according to each feature information in the updated feature information set and the weight corresponding to each feature information;
and under the condition that the image is determined to be a stolen image according to the stealing probability, the target post is placed down.
2. The method of claim 1, wherein the weight corresponding to each of the feature information is determined using an Analytic Hierarchy Process (AHP).
3. The method of claim 1, wherein the related information comprises at least: address information of the target house and issuing user information; the feature information set at least includes: address information of the similar house and issuing user information; the set of feature information further comprises at least one of: the number of different addresses in the total number of posts which issue the similar images and the address information of the house recorded by each post which issues the similar images.
4. The method of claim 1, wherein similar images matching a plurality of images exist in the house source library, and wherein determining the images as a pirate graph according to the pirate probability comprises:
determining the target probability that the image of the target post is a stolen image according to the stealing probability of each similar image;
and under the condition that the target probability is greater than a numerical threshold, determining that the images of the target posts are stolen images.
5. The method of claim 1, wherein determining that a similar image matching the image exists in the house-source library comprises:
extracting first characteristic information of the image;
determining a target image set which belongs to the same category as the images in the room source library according to the first characteristic information;
and determining that similar images matched with the images exist in the house source library under the condition that the similar images exist in the target image set according to the matching result of the first characteristic information and second characteristic information corresponding to the images in the target image set.
6. The method of claim 5, wherein the extracting first feature information of the image comprises:
inputting the image into an image feature extraction model to obtain the first feature information, wherein the image feature extraction model is obtained by training a plurality of sample data, and the sample data comprises: a similar image pair, a dissimilar image pair, a first label to which the similar image pair corresponds, and a second label to which the dissimilar image pair corresponds.
7. The method of claim 6, wherein the image feature extraction model comprises: a ResNet18 network including at least 15 convolutional layers; the loss function of the image feature extraction model is a Euclidean distance loss function, and the power factor in the Euclidean distance loss function is 8.
8. The method of claim 5, wherein the first feature information comprises a first feature vector, and wherein determining a set of target images in the house source library that belong to the same category as the image according to the first feature information comprises:
and determining the target image set from each image set according to the first characteristic vector and the distance between the clustering centers corresponding to each image set in the house source library, wherein each image set is determined by adopting a k-means clustering algorithm.
9. The method according to claim 5, wherein the first feature information includes a first feature vector, the second feature information includes a second feature vector, and the determining that the similar image exists in the target image set according to a matching result of the first feature information and second feature information corresponding to each image in the target image set comprises:
and taking the image of which the distance between the second characteristic vector and the first characteristic vector in the target image set is smaller than a set distance threshold value as the similar image.
10. The method of claim 3, wherein the updating the feature information set according to the related information comprises:
carrying out consistency matching on the address information in the related information and the address information in the feature information set, updating the address information in the feature information set to a first feature value under the condition of successful matching, and updating the address information in the feature information set to a second feature value under the condition of unsuccessful matching;
and carrying out consistency matching on the release user information in the related information and the release user information in the characteristic information set, updating the release user information in the characteristic information set to a first characteristic value under the condition of successful matching, and updating the release user information in the characteristic information set to a second characteristic value under the condition of unsuccessful matching.
11. An information management apparatus, characterized in that the apparatus comprises:
the acquisition module is used for acquiring the image and the related information of the target house corresponding to the target post; the method comprises the steps of obtaining a characteristic information set of a similar image under the condition that the similar image matched with the image exists in a house source library, wherein the characteristic information set is at least used for reflecting the exposure degree of the similar image;
the updating module is used for updating the feature information set according to the related information, so that the updated feature information set is also used for reflecting whether the target house and the similar house corresponding to the similar image are from the same house published by the same publishing user or not;
the determining module is used for determining the stealing probability of the similar images according to each piece of feature information in the updated feature information set and the weight corresponding to each piece of feature information;
and the control module is used for setting down the target post under the condition that the image is determined to be a stolen image according to the stealing probability.
12. An electronic device, comprising: processor, memory and a computer program stored on the memory and executable on the processor, the computer program, when executed by the processor, implementing the steps of the information management method according to any one of claims 1 to 10.
13. A computer-readable storage medium, characterized in that a computer program is stored thereon, which computer program, when being executed by a processor, carries out the steps of the information management method according to any one of claims 1 to 10.
CN202111017416.0A 2021-08-31 2021-08-31 Information management method and device, electronic equipment and storage medium Active CN113706173B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111017416.0A CN113706173B (en) 2021-08-31 2021-08-31 Information management method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111017416.0A CN113706173B (en) 2021-08-31 2021-08-31 Information management method and device, electronic equipment and storage medium

Publications (2)

Publication Number Publication Date
CN113706173A true CN113706173A (en) 2021-11-26
CN113706173B CN113706173B (en) 2022-07-08

Family

ID=78658438

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111017416.0A Active CN113706173B (en) 2021-08-31 2021-08-31 Information management method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN113706173B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114970891A (en) * 2022-05-11 2022-08-30 上海霖承环保科技有限公司 Transaction platform system and method for recycling recovered household appliances under big data

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017032117A1 (en) * 2015-08-25 2017-03-02 中兴通讯股份有限公司 Image processing method and apparatus
CN109447404A (en) * 2018-09-20 2019-03-08 深圳供电局有限公司 Appraisal procedure, device, computer equipment and the storage medium of Research on Housing Vacancy Rate
CN109583909A (en) * 2018-10-26 2019-04-05 深圳壹账通智能科技有限公司 Houseclearing discriminating method, device, equipment and medium based on image recognition
WO2019200737A1 (en) * 2018-04-20 2019-10-24 平安科技（深圳）有限公司 Real estate data uploading method and apparatus, computer device, and storage medium
CN110633381A (en) * 2018-12-25 2019-12-31 北京时光荏苒科技有限公司 Method and device for identifying false house source, storage medium and electronic equipment
WO2020135038A1 (en) * 2018-12-29 2020-07-02 上海掌门科技有限公司 Image processing method, electronic apparatus, and computer readable medium
CN112419312A (en) * 2020-12-11 2021-02-26 五八有限公司 Similar house source information detection method and device, electronic equipment and readable medium

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017032117A1 (en) * 2015-08-25 2017-03-02 中兴通讯股份有限公司 Image processing method and apparatus
WO2019200737A1 (en) * 2018-04-20 2019-10-24 平安科技（深圳）有限公司 Real estate data uploading method and apparatus, computer device, and storage medium
CN109447404A (en) * 2018-09-20 2019-03-08 深圳供电局有限公司 Appraisal procedure, device, computer equipment and the storage medium of Research on Housing Vacancy Rate
CN109583909A (en) * 2018-10-26 2019-04-05 深圳壹账通智能科技有限公司 Houseclearing discriminating method, device, equipment and medium based on image recognition
CN110633381A (en) * 2018-12-25 2019-12-31 北京时光荏苒科技有限公司 Method and device for identifying false house source, storage medium and electronic equipment
WO2020135038A1 (en) * 2018-12-29 2020-07-02 上海掌门科技有限公司 Image processing method, electronic apparatus, and computer readable medium
CN112419312A (en) * 2020-12-11 2021-02-26 五八有限公司 Similar house source information detection method and device, electronic equipment and readable medium

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
李振豪等: "基于层次分析法的制图综合相似性评价模型研究", 《测绘与空间地理信息》 *
鲁红英 等: "基于层次分析法的商品房选购满意度评价研究", 《四川文理学院学报》 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114970891A (en) * 2022-05-11 2022-08-30 上海霖承环保科技有限公司 Transaction platform system and method for recycling recovered household appliances under big data

Also Published As

Publication number Publication date
CN113706173B (en) 2022-07-08

Similar Documents

Publication Publication Date Title
TWI712981B (en) Risk identification model training method, device and server
CN107066616B (en) Account processing method and device and electronic equipment
WO2021174944A1 (en) Message push method based on target activity, and related device
US20140379617A1 (en) Method and system for recommending information
US20160364393A1 (en) Search method and system
CN110689457A (en) Intelligent reception method for online clients in real estate industry, server and storage medium
CN111666492A (en) Information pushing method, device and equipment based on user behaviors and storage medium
CN108269122B (en) Advertisement similarity processing method and device
CN110148053B (en) User credit line evaluation method and device, electronic equipment and readable medium
CN111612038A (en) Abnormal user detection method and device, storage medium and electronic equipment
CN108961019B (en) User account detection method and device
CN111966886A (en) Object recommendation method, object recommendation device, electronic equipment and storage medium
CN112507212A (en) Intelligent return visit method and device, electronic equipment and readable storage medium
CN111582932A (en) Inter-scene information pushing method and device, computer equipment and storage medium
CN113706173B (en) Information management method and device, electronic equipment and storage medium
CN107871277B (en) Server, client relationship mining method and computer readable storage medium
CN117114514A (en) Talent information analysis management method, system and device based on big data
CN111538909A (en) Information recommendation method and device
CN112036987B (en) Method and device for determining recommended commodity
CN112348300A (en) Method and device for pushing information
US20220172271A1 (en) Method, device and system for recommending information, and storage medium
CN115758271A (en) Data processing method, data processing device, computer equipment and storage medium
KR20190061317A (en) Method of providing stock information
KR102161828B1 (en) Method and apparatus for controlling merchant management
CN111784381A (en) Privacy protection and SOM network-based power customer segmentation method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant