Method and apparatus for predicting information

Abstract

The present application discloses a method and device for predicting information, which relates to the field of machine learning. The specific implementation scheme is as follows: obtaining order data and property data matching the order data; extracting features from the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address; parsing the property data to extract the household address and the number of associated persons corresponding to the household address; associating the delivery address with the household address, and classifying the associated delivery address according to the number of associated persons corresponding to the resident address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons; using the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, and predicting the number of associated persons of the corresponding address in the second address set. This scheme improves the accuracy of information prediction and makes it more convenient and accurate to obtain community associated person information.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, specifically to the field of machine learning, and more particularly to methods and devices for predicting information. Background Art

[0002] Community-associated person analysis refers to detecting abnormal rental behavior within the community, and given a community house number, predicting the number of people living in the house number, so as to find people in close contact. Illegal group rentals in cities are usually rented by a large number of people because of their low prices, but they pose great risks of infectious diseases and fire hazards. Detecting abnormal group rentals and timely discovering clustering behaviors are of great significance for saving community management costs and controlling the spread of the epidemic. There are currently two main methods for community-associated person analysis: one is through reporting and manual statistics by community staff; the other is to predict the association relationship of all other residents in the community based on a small amount of community-associated data. Summary of the invention

[0003] The present application proposes a method, apparatus, device and storage medium for predicting information.

[0004] According to a first aspect, the present application provides a method for predicting information, the method comprising: obtaining order data and property data matching the order data; performing feature extraction on the order data to obtain a delivery address in the order data and a household feature set corresponding to the delivery address; performing content parsing on the property data to extract the household address and the number of associated people corresponding to the resident address; associating the delivery address with the household address, and classifying the associated delivery addresses according to the number of associated people corresponding to the resident address, to obtain a first address set with a determined number of associated people and a second address set with an undetermined number of associated people; using a learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, to predict the number of associated people of the corresponding address in the second address set, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

[0005] In some embodiments, the associated person prediction model is obtained by the following learning method: taking the first address set as the initial labeled sample data set, and using the machine learning model to train the household feature set corresponding to each address in the initial labeled sample data set to obtain the initial associated person prediction model; judging whether all addresses in the second address set meet the iteration stop condition, if all addresses in the second address set do not meet the iteration stop condition, based on the current associated person prediction model and selection strategy, select addresses from the second address set as addresses to be labeled and perform manual labeling, wherein the selection strategy is to perform threshold judgment on the uncertainty of each address in the second address set, and the manual labeling is to manually label the number of associated people of the addresses to be labeled; updating the labeled sample data set based on the manually labeled addresses to be labeled, and training the household feature set corresponding to each address in the updated labeled sample data set to update the associated person prediction model; if all addresses in the second address set meet the iteration stop condition, determining the current associated person prediction model as the final associated person prediction model.

[0006] In some embodiments, after performing feature extraction on the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address, it also includes: based on the household feature set corresponding to the delivery address, using an abnormal household determination method to determine whether the delivery address is an abnormal resident; if the delivery address is an abnormal resident, outputting the determination result of the corresponding delivery address; if the delivery address is not an abnormal resident, continuing to execute.

[0007] In some embodiments, based on the current associated person prediction model and selection strategy, an address is selected from the second address set as the address to be marked and manually marked, including: based on the current associated person prediction model, calculating the information entropy of each address in the second address set; determining whether the information entropy of each address is greater than a preset threshold; in response to the information entropy being greater than the preset threshold, determining the address corresponding to the information entropy as the address to be marked and manually marking it.

[0008] In some embodiments, based on the household feature set corresponding to the delivery address, an abnormal household determination method is used to determine whether the delivery address is an abnormal resident, including: judging whether the delivery address is an abnormal resident based on the household feature set corresponding to the delivery address according to the classification of the abnormal resident and the type of the delivery address.

[0009] In some embodiments, the feature categories in the household feature set include address entity features, portrait features, and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of personnel of the household, and the consumption features are used to indicate the quantity of different consumer items of the household.

[0010] In some embodiments, the method further includes: performing semantic processing on the delivery address and the household address to obtain a processed delivery address and a processed household address.

[0011] In a second aspect, the present application provides a device for predicting information, the device comprising: an acquisition unit, configured to acquire order data and property data matching the order data; an analysis unit, configured to perform feature extraction on the order data to obtain a delivery address in the order data and a household feature set corresponding to the delivery address; perform content parsing on the property data to extract the household address and the number of associated people corresponding to the household address; a classification unit, configured to associate the delivery address with the household address, and classify the associated delivery addresses according to the number of associated people corresponding to the resident addresses, to obtain a first address set with a determined number of associated people and a second address set with an undetermined number of associated people; a prediction unit, configured to use the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated people of the corresponding address in the second address set, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

[0012] In some embodiments, the associated person prediction model in the prediction unit is learned by the following units: a training unit, configured to use the first address set as the initial labeled sample data set, and use the machine learning model to train the household feature set corresponding to each address in the initial labeled sample data set to obtain the initial associated person prediction model; a selection unit, configured to determine whether all addresses in the second address set meet the iteration stop condition. If all addresses in the second address set do not meet the iteration stop condition, based on the current associated person prediction model and selection strategy, select addresses from the second address set as addresses to be labeled and perform manual labeling, wherein the selection strategy is to perform threshold judgment on the uncertainty of each address in the second address set, and the manual labeling is to manually label the number of associated people of the addresses to be labeled; an updating unit, configured to update the labeled sample data set based on the manually labeled addresses to be labeled, and train the household feature set corresponding to each address in the updated labeled sample data set to update the associated person prediction model; a determination unit, configured to determine that the current associated person prediction model is the final associated person prediction model if all addresses in the second address set meet the iteration stop condition.

[0013] In some embodiments, the device also includes: an abnormality judgment unit, which is configured to judge whether the delivery address is an abnormal resident based on the household feature set corresponding to the delivery address and using an abnormal resident judgment method; if the delivery address is an abnormal resident, output the judgment result of the corresponding delivery address; if the delivery address is not an abnormal resident, continue to execute.

[0014] In some embodiments, the selection unit includes: a calculation module, configured to calculate the information entropy of each address in the second address set based on the current associated person prediction model; a judgment module, configured to judge whether the information entropy of each address is greater than a preset threshold; a determination module, configured to determine that the address corresponding to the information entropy is an address to be marked and perform manual marking in response to the information entropy being greater than the preset threshold.

[0015] In some embodiments, the abnormality judgment unit is further configured to judge whether the delivery address is an abnormal resident based on the classification of the abnormal resident and the type of the delivery address and based on the resident feature set corresponding to the delivery address.

[0016] In some embodiments, the feature categories in the household feature set in the analysis unit include address entity features, portrait features, and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of personnel of the household, and the consumption features are used to indicate the quantity of different consumer items of the household.

[0017] In some embodiments, the device further includes: a processing unit configured to perform semantic processing on the delivery address and the household address to obtain a processed delivery address and a processed household address.

[0018] In a third aspect, the present application provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a method described in any implementation manner in the first aspect.

[0019] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium storing computer instructions, characterized in that the computer instructions are used to cause a computer to execute a method as described in any implementation manner in the first aspect.

[0020] According to the technology of the present application, this method obtains order data and property data matching the order data, extracts features from the order data, obtains the delivery address in the order data and the household feature set corresponding to the delivery address, performs content analysis on the property data, extracts the household address and the number of associated people corresponding to the resident address, associates the delivery address with the household address, and classifies the associated delivery address according to the number of associated people corresponding to the resident address, obtains a first address set with a determined number of associated people and a second address set with an undetermined number of associated people, uses the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, and predicts the number of associated people of the corresponding address in the second address set, solving the problem in the prior art that the number of community associated data is small and the overall prediction cannot be accurately performed through modeling methods, improves the accuracy of information prediction, and avoids possible omissions and large labor cost expenditures caused by manual investigation, making it more convenient and accurate to obtain community associated person information. The solution provided by this application is for legal purposes only.

[0021] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present application.

[0023] Figure 1 is a schematic diagram of a first embodiment of a method for predicting information according to the present application;

[0024] Figure 2 is a scene diagram that can implement the method for predicting information in an embodiment of the present application;

[0025] Figure 3 is a schematic diagram of a second embodiment of a method for predicting information according to the present application;

[0026] Figure 4 is a schematic structural diagram of an embodiment of a device for predicting information according to the present application;

[0027] Figure 5 It is a block diagram of an electronic device used to implement the method for predicting information according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0029] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0030] Figure 1 A schematic diagram 100 of a first embodiment of a method for predicting information according to the present application is shown. The method for predicting information comprises the following steps:

[0031] Step 101, obtaining order data and property data matching the order data.

[0032] In this embodiment, the executor of the method for predicting information can obtain order data and property data matching the order data from other electronic devices or locally through a wired connection or a wireless connection according to a business request, wherein the property data can be obtained based on the business request, and the property data is in the same geographical scope or the same household address as the order data.

[0033] Step 102, extract features from the order data to obtain the delivery address in the order data and the resident feature set corresponding to the delivery address, perform content analysis on the property data, and extract the resident address and the number of associated people corresponding to the resident address.

[0034] In this embodiment, the above-mentioned execution entity can perform feature extraction on the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address. At the same time, it can perform content analysis on the property data to extract the household address in the property data and the number of associated people corresponding to the household address (i.e., the number of households in the household address).

[0035] In some optional implementations of this embodiment, the feature categories in the household feature set include address entity features, portrait features, and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of residents, and the consumption features are used to indicate the quantity of different consumer items of the residents.

[0036] Step 103, associating the delivery address with the household address, and classifying the associated delivery addresses according to the number of associated persons corresponding to the household address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons.

[0037] In this embodiment, the execution subject may align and associate the household address with the delivery address to obtain the associated delivery address, and classify the associated delivery address according to the number of associated persons corresponding to the household address obtained in step 102 to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons. The first address set is used to represent address information with an associated number of persons confirmed by property data, and the second address set is used to represent address information with an associated number of persons not confirmed by property data.

[0038] Step 104, using the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, and predicting the number of associated persons of the corresponding address in the second address set.

[0039] In this embodiment, the execution subject can use the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated persons of the corresponding address in the second address set. The associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

[0040] To further explain, for example, based on the user's order data, the characteristics of the household can be extracted, which mainly include three parts: First, the address entity characteristics: including the number of associated accounts and the number of address aliases. The more accounts / address aliases associated with the same delivery address, the more associated people may be; Second, the portrait characteristics: including whether there are elderly people, children, couples, etc. According to the name of the product, we defined the keywords of children, elderly people, boys, girls, and couples respectively. For example, if the purchased goods contain words such as "baby" and "children", it means that there are children among the associated people. Similarly, if there are words such as "dad shoes" and "mom clothes", it means that there are elderly people among the associated people. If there are children, the number of closely associated people is >= 3, if there are elderly people, the number of closely associated people is +1, and so on; Third, the consumption characteristics: including the quantity of different consumer items, such as the consumption of drinking water / edible oil / rice / noodles / toothpaste / toothbrush. If the user buys a large amount of edible oil, toothpaste, toothbrush, etc., it means that there are many closely associated people.

[0041] It should be noted that technicians can set the model structure of the above-mentioned associated person prediction model according to actual needs, and the embodiments of the present disclosure are not limited to this.

[0042] Continue to see Figure 2, the method 200 for predicting information in this embodiment runs in an electronic device 201. The electronic device 201 first obtains order data and property data 202 that matches the order data, then the electronic device 201 extracts features from the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address, performs content analysis on the property data, extracts the household address and the number of associated persons corresponding to the household address 203, then the electronic device 201 associates the delivery address with the household address, and classifies the associated delivery address according to the number of associated persons corresponding to the resident address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons 204, finally the electronic device 201 uses the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, and predicts the number of associated persons of the corresponding address in the second address set 205, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

[0043] The method provided by the above-mentioned embodiments of the present application obtains order data and property data matching the order data, extracts features of the order data, obtains the delivery address in the order data and the household feature set corresponding to the delivery address, parses the property data, extracts the household address and the number of associated people corresponding to the resident address, associates the delivery address and the household address, and classifies the associated delivery addresses according to the number of associated people corresponding to the resident address, to obtain a first address set with a determined number of associated people and a second address set with an undetermined number of associated people, and uses the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated people of the corresponding addresses in the second address set, thereby solving the problem in the prior art that the amount of community associated data is small and the overall prediction cannot be accurately performed through modeling methods, thereby improving the accuracy of information prediction and avoiding possible omissions and large labor costs caused by manual investigation, making it more convenient and accurate to obtain community associated person information.

[0044] Further references Figure 3 , which shows a schematic diagram 300 of a second embodiment of a method for predicting information. The process of the method comprises the following steps:

[0045] Step 301, obtaining order data and property data matching the order data.

[0046] Step 302, extract features from the order data to obtain the delivery address in the order data and the resident feature set corresponding to the delivery address, perform content analysis on the property data, and extract the resident address and the number of associated people corresponding to the resident address.

[0047] In some optional implementations of this embodiment, the method further includes: semantically processing the delivery address and the household address to obtain a processed delivery address and a processed household address. Different users may have different habits of writing delivery addresses. For example, "Room 101, Unit 1, Building 1", "Room 101, Unit 1, Building 1", and "1-1-101" may represent the same delivery address. Therefore, it is necessary to perform semantic disambiguation on the information representing the same delivery address. Predicting information for the semantically processed address improves the accuracy of information prediction.

[0048] Step 303, based on the household feature set corresponding to the delivery address, use the abnormal household determination method to determine whether the delivery address is an abnormal household. If the delivery address is an abnormal household, output the determination result of the corresponding delivery address. If the delivery address is not an abnormal household, continue to execute.

[0049] In this embodiment, the executing entity can use a predefined abnormal resident determination method to search in the household feature set corresponding to the delivery address, and determine whether the delivery address is an abnormal resident based on the search results. If the delivery address is determined to be an abnormal resident, the determination result of the corresponding delivery address is output; if the delivery address is determined not to be an abnormal resident, execution continues.

[0050] In some optional implementations of this embodiment, based on the household feature set corresponding to the delivery address, an abnormal household determination method is used to determine whether the delivery address is an abnormal household, including: judging whether the delivery address is an abnormal household based on the household feature set corresponding to the delivery address according to the classification of abnormal households and the type of the delivery address. Among them, the types of delivery addresses include: youth apartments, entrepreneurial apartments, elderly apartments, single apartments, basements, supermarkets, etc., and the type of each delivery address can be determined by the name of the delivery address. The classification of abnormal residents includes: youth / entrepreneurship / single apartments with elderly / children; elderly apartments with children; basements with a large number of residents (i.e., a large number of associated accounts), or elderly and children; supermarkets with a large number of residents upstairs (i.e., many associated accounts or too many daily necessities purchased).

[0051] Step 304, associating the delivery address with the household address, and classifying the associated delivery addresses according to the number of associated persons corresponding to the household address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons.

[0052] Step 305, using the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input, and predicting the number of associated persons of the corresponding address in the second address set.

[0053] In this embodiment, the execution subject may use the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated persons of the corresponding address in the second address set.

[0054] In some optional implementations of this embodiment, the associated person prediction model is obtained by the following learning method: taking the first address set as the initial labeled sample data set, using the machine learning model to train the household feature set corresponding to each address in the initial labeled sample data set, and obtaining the initial associated person prediction model; judging whether all addresses in the second address set meet the iteration stop condition, if all addresses in the second address set do not meet the iteration stop condition, based on the current associated person prediction model and selection strategy, select addresses from the second address set as addresses to be labeled and perform manual labeling, wherein the selection strategy is to perform threshold determination on the uncertainty of each address in the second address set, and the manual labeling is to manually label the number of associated persons of the addresses to be labeled; updating the labeled sample data set based on the manually labeled addresses to be labeled, and training the household feature set corresponding to each address in the updated labeled sample data set, and updating the associated person prediction model; if all addresses in the second address set meet the iteration stop condition, determining the current associated person prediction model as the final associated person prediction model. Using the active learning method for model training can better process large training data sets and reduce manual labeling costs.

[0055] In some optional implementations of this embodiment, based on the current associated person prediction model and selection strategy, an address is selected from the second address set as the address to be marked and manually marked, including: calculating the information entropy of each address in the second address set based on the current associated person prediction model; judging whether the information entropy of each address is greater than a preset threshold; in response to the information entropy being greater than the preset threshold, determining the address corresponding to the information entropy as the address to be marked and manually marking it. Using active learning technology, residents with greater uncertainty are found and targetedly marked, which can not only discover abnormal associated behaviors, but also realize the construction of a global associated behavior prediction model.

[0056] In this embodiment, the specific operations of steps 301, 302 and 304 are the same as Figure 1 The operations of steps 101, 102 and 103 in the illustrated embodiment are substantially the same and will not be described in detail herein.

[0057] from Figure 3 It can be seen that Figure 1Compared with the corresponding embodiments, the schematic diagram 300 of the method for predicting information in this embodiment adopts a household feature set corresponding to the delivery address, and uses an abnormal household determination method to determine whether the delivery address is an abnormal household. If the delivery address is an abnormal household, the determination result of the corresponding delivery address is output; if the delivery address is not an abnormal household, the execution continues downward, thereby realizing the identification of abnormal cohabitation associated behaviors. By first identifying the abnormal household and then predicting the person associated with the address, the amount of data for the prediction of the associated person is reduced, and the efficiency of information prediction is further improved.

[0058] Further Figure 4 As shown above, Figure 1 The present application provides an embodiment of a device for predicting information. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0059] The device 400 for predicting information in this embodiment includes: an acquisition unit 401, an analysis unit 402, a classification unit 403 and a prediction unit 404, wherein the acquisition unit 401 is configured to acquire order data and property data matching the order data; the analysis unit 402 is configured to perform feature extraction on the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address; the property data is content parsed to extract the household address and the number of associated persons corresponding to the resident address; the classification unit 403 is configured to associate the delivery address with the household address, and classify the associated delivery address according to the number of associated persons corresponding to the resident address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons; the prediction unit 404 is configured to use the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated persons of the corresponding addresses in the second address set, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

[0060] In this embodiment, the specific processing of the acquisition unit 401, the analysis unit 402, the classification unit 403 and the prediction unit 404 of the device for predicting information and the technical effects thereof can be referred to in Figure 1 The relevant descriptions of steps 101 to 104 in the corresponding embodiment are not repeated here.

[0061] In some optional implementations of the present embodiment, the associated person prediction model in the prediction unit is learned by the following units: a training unit, configured to use the first address set as an initial labeled sample data set, and use a machine learning model to train the household feature set corresponding to each address in the initial labeled sample data set to obtain an initial associated person prediction model; a selection unit, configured to determine whether all addresses in the second address set meet the iteration stop condition, if all addresses in the second address set do not meet the iteration stop condition, based on the current associated person prediction model and selection strategy, select addresses from the second address set as addresses to be labeled and perform manual labeling, wherein the selection strategy is to perform threshold determination on the uncertainty of each address in the second address set, and the manual labeling is to manually label the number of associated people of the addresses to be labeled; an updating unit, configured to update the labeled sample data set based on the manually labeled addresses to be labeled, and train the household feature set corresponding to each address in the updated labeled sample data set to update the associated person prediction model; a determination unit, configured to determine that the current associated person prediction model is the final associated person prediction model if all addresses in the second address set meet the iteration stop condition.

[0062] In some optional implementations of this embodiment, the device also includes: an abnormality judgment unit, which is configured to judge whether the delivery address is an abnormal resident based on the household feature set corresponding to the delivery address and using an abnormal resident judgment method; if the delivery address is an abnormal resident, output the judgment result of the corresponding delivery address; if the delivery address is not an abnormal resident, continue to execute.

[0063] In some optional implementations of this embodiment, the selection unit includes: a calculation module, configured to calculate the information entropy of each address in the second address set based on the current associated person prediction model; a judgment module, configured to judge whether the information entropy of each address is greater than a preset threshold; a determination module, configured to determine that the address corresponding to the information entropy is an address to be marked and perform manual marking in response to the information entropy being greater than the preset threshold.

[0064] In some optional implementations of this embodiment, the abnormality judgment unit is further configured to judge whether the delivery address is an abnormal resident based on the classification of the abnormal resident and the type of the delivery address and based on the resident feature set corresponding to the delivery address.

[0065] In some optional implementations of this embodiment, the feature categories in the household feature set in the analysis unit include address entity features, portrait features, and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of residents, and the consumption features are used to indicate the quantity of different consumer items of the residents.

[0066] In some optional implementations of this embodiment, the device also includes: a processing unit configured to perform semantic processing on the delivery address and the household address to obtain a processed delivery address and a processed household address.

[0067] According to an embodiment of the present application, the present application also provides an electronic device and a readable storage medium.

[0068] like Figure 5 , is a block diagram of an electronic device for predicting information according to an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0069] like Figure 5 As shown, the electronic device includes: one or more processors 501, a memory 502, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected using different buses and can be installed on a common mainboard or installed in other ways as needed. The processor can process instructions executed in the electronic device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In other embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 501 is taken as an example.

[0070] The memory 502 is a non-transient computer-readable storage medium provided in the present application. The memory stores instructions executable by at least one processor to enable at least one processor to perform the method for predicting information provided in the present application. The non-transient computer-readable storage medium of the present application stores computer instructions, which are used to enable a computer to perform the method for predicting information provided in the present application.

[0071] The memory 502 is a non-transient computer-readable storage medium that can be used to store non-transient software programs, non-transient computer executable programs and modules, such as program instructions / modules corresponding to the method for predicting information in the embodiment of the present application (for example, the attached Figure 4 The processor 501 executes various functional applications and data processing of the server by running the non-transient software programs, instructions and modules stored in the memory 502, that is, implements the method for predicting information in the above method embodiment.

[0072] The memory 502 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device for predicting information, etc. In addition, the memory 502 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 502 may optionally include a memory remotely arranged relative to the processor 501, and these remote memories may be connected to the electronic device for predicting information via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0073] The electronic device for predicting information may further include: an input device 503 and an output device 504. The processor 501, the memory 502, the input device 503 and the output device 504 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.

[0074] The input device 503 can receive input digital or character information, and generate key signal input related to user settings and function control of the electronic device for predicting information, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator rod, one or more mouse buttons, a trackball, a joystick, and other input devices. The output device 504 may include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor), etc. The display device may include, but is not limited to, a liquid crystal display (LCD), a light emitting diode (LED) display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0075] Various implementations of the systems and techniques described herein can be realized in digital electronic circuit systems, integrated circuit systems, dedicated ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0076] These computer programs (also referred to as programs, software, software applications, or code) include machine instructions for programmable processors and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or means (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0077] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0078] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0079] A computer system may include clients and servers. Clients and servers are generally remote from each other and usually interact through a communication network. The relationship of client and server is generated by computer programs running on respective computers and having a client-server relationship to each other.

[0080] According to the technical solution of the embodiment of the present application, by acquiring order data and property data matching the order data, feature extraction is performed on the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address, content analysis is performed on the property data to extract the household address and the number of associated people corresponding to the resident address, the delivery address and the resident address are associated, and the associated delivery addresses are classified according to the number of associated people corresponding to the resident address to obtain a first address set with a determined number of associated people and a second address set with an undetermined number of associated people, and the learned associated person prediction model is used to take the household feature set corresponding to each address in the second address set as input to predict the number of associated people of the corresponding address in the second address set, thereby solving the problem in the prior art that The small amount of community-related data makes it impossible to accurately make overall predictions through modeling methods. This improves the accuracy of information predictions, avoids omissions that may be caused by manual investigations, and avoids large labor costs, making it more convenient and accurate to obtain community-related person information. The household feature set corresponding to the delivery address is used, and the abnormal household determination method is used to determine whether the delivery address is an abnormal resident. If the delivery address is an abnormal resident, the determination result of the corresponding delivery address is output. If the delivery address is not an abnormal resident, the execution continues, realizing the identification of abnormal cohabitation-related behaviors. By first identifying abnormal residents and then predicting address-related persons, the amount of data for related person predictions is reduced, further improving the efficiency of information prediction.

[0081] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this application can be executed in parallel, sequentially or in different orders, as long as the expected results of the technical solution disclosed in this application can be achieved, and this document is not limited here.

[0082] The above specific implementations do not constitute a limitation on the protection scope of this application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of this application should be included in the protection scope of this application.

Claims

1. A method for predicting information, the method comprising: Obtaining order data and property data matching the order data; Performing feature extraction on the order data to obtain a delivery address in the order data and a resident feature set corresponding to the delivery address; performing content analysis on the property data to extract a resident address and a number of associated persons corresponding to the resident address; Associating the delivery address with the household address, and classifying the associated delivery addresses according to the number of associated persons corresponding to the household address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons; The learned associated person prediction model is used to take the household feature set corresponding to each address in the second address set as input to predict the number of associated people of the corresponding address in the second address set, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

2. The method according to claim 1, wherein: The associated person prediction model is obtained by the following learning method: The first address set is used as an initial labeled sample data set, and a machine learning model is used to train a household feature set corresponding to each address in the initial labeled sample data set to obtain an initial associated person prediction model; Determine whether all addresses in the second address set meet the iteration stop condition. If all addresses in the second address set do not meet the iteration stop condition, select addresses from the second address set as addresses to be marked based on the current associated person prediction model and selection strategy and perform manual marking, wherein the selection strategy is to perform threshold determination on the uncertainty of each address in the second address set, and the manual marking is to manually mark the number of people associated with the address to be marked; Based on the manually labeled addresses to be labeled, the labeled sample data set is updated, and the household feature set corresponding to each address in the updated labeled sample data set is trained to update the associated person prediction model; If all addresses in the second address set meet the iteration stop condition, the current associated person prediction model is determined to be the final associated person prediction model.

3. The method according to claim 1, wherein: After extracting features from the order data to obtain the delivery address in the order data and the household feature set corresponding to the delivery address, the method further includes: Based on the household feature set corresponding to the delivery address, use the abnormal household determination method to determine whether the delivery address is an abnormal household; if the delivery address is an abnormal household, output the determination result of the corresponding delivery address; if the delivery address is not an abnormal household, continue to execute.

4. The method according to claim 2, wherein: The step of selecting an address from the second address set as an address to be marked and manually marking the address based on the current associated person prediction model and selection strategy includes: Calculate the information entropy of each address in the second address set based on the current associated person prediction model; Determine whether the information entropy of each address is greater than a preset threshold; In response to the information entropy being greater than a preset threshold, determining that the address corresponding to the information entropy is an address to be marked and manually marking it.

5. The method according to claim 3, wherein: The determining whether the delivery address is an abnormal resident by using an abnormal resident determination method based on the resident feature set corresponding to the delivery address includes: According to the classification of the abnormal household and the type of the delivery address, based on the household feature set corresponding to the delivery address, it is determined whether the delivery address is an abnormal household.

6. The method according to claim 1, wherein: The feature categories in the household feature set include address entity features, portrait features and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of residents, and the consumption features are used to indicate the quantity of different consumer items of the residents.

7. The method according to claim 1, further comprising: The delivery address and the household address are semantically processed to obtain a processed delivery address and a processed household address.

8. A device for predicting information, the device comprising: An acquisition unit, configured to acquire order data and property data matching the order data; An analysis unit is configured to perform feature extraction on the order data to obtain a delivery address in the order data and a household feature set corresponding to the delivery address; Parsing the content of the property data to extract the resident address and the number of people associated with the resident address; a classification unit configured to associate the delivery address with the household address, and classify the associated delivery addresses according to the number of associated persons corresponding to the household address, to obtain a first address set with a determined number of associated persons and a second address set with an undetermined number of associated persons; The prediction unit is configured to use the learned associated person prediction model to take the household feature set corresponding to each address in the second address set as input to predict the number of associated people of the corresponding address in the second address set, wherein the associated person prediction model is obtained by pre-training the household feature set corresponding to each address in the first address set and the household feature set corresponding to each address in the second address set.

9. The device according to claim 8, wherein: The associated person prediction model in the prediction unit is learned by the following units: A training unit is configured to use the first address set as an initial labeled sample data set, and use a machine learning model to train a household feature set corresponding to each address in the initial labeled sample data set to obtain an initial associated person prediction model; a selection unit configured to determine whether all addresses in the second address set meet an iteration stop condition, and if all addresses in the second address set do not meet the iteration stop condition, select addresses from the second address set as addresses to be marked and perform manual marking based on a current associated person prediction model and a selection strategy, wherein the selection strategy is to perform a threshold determination on the uncertainty of each address in the second address set, and the manual marking is to manually mark the number of associated persons of the addresses to be marked; An updating unit is configured to update the labeled sample data set based on the manually labeled addresses to be labeled, and train the household feature set corresponding to each address in the updated labeled sample data set to update the associated person prediction model; The determination unit is configured to determine the current associated person prediction model as the final associated person prediction model if all addresses in the second address set meet the iteration stop condition.

10. The device according to claim 8, wherein: The device also includes: The abnormality judgment unit is configured to judge whether the delivery address is an abnormal resident based on the resident feature set corresponding to the delivery address and using an abnormal resident judgment method; if the delivery address is an abnormal resident, output the judgment result of the corresponding delivery address; if the delivery address is not an abnormal resident, continue to execute.

11. The device according to claim 9, wherein: The selection unit comprises: A calculation module, configured to calculate the information entropy of each address in the second address set based on the current associated person prediction model; A determination module is configured to determine whether the information entropy of each address is greater than a preset threshold; The determination module is configured to determine, in response to the information entropy being greater than a preset threshold, that the address corresponding to the information entropy is an address to be marked and perform manual marking.

12. The device according to claim 10, wherein: The abnormality judgment unit is further configured to judge whether the delivery address is an abnormal resident based on the classification of the abnormal resident and the type of the delivery address and based on the resident feature set corresponding to the delivery address.

13. The device according to claim 8, wherein: The feature categories in the household feature set in the analysis unit include address entity features, portrait features and consumption features. The address entity features are used to indicate the number of associated accounts and the number of address aliases, the portrait features are used to indicate the type of residents, and the consumption features are used to indicate the quantity of different consumer items of the residents.

14. The device according to claim 8, further comprising: The processing unit is configured to perform semantic processing on the delivery address and the household address to obtain a processed delivery address and a processed household address.

15. An electronic device, characterized in that: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

16. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

Images