WO2020173043A1 - 用户群的优化方法及装置、计算机非易失性可读存储介质 - Google Patents

用户群的优化方法及装置、计算机非易失性可读存储介质 Download PDF

Info

Publication number
WO2020173043A1
WO2020173043A1 PCT/CN2019/097890 CN2019097890W WO2020173043A1 WO 2020173043 A1 WO2020173043 A1 WO 2020173043A1 CN 2019097890 W CN2019097890 W CN 2019097890W WO 2020173043 A1 WO2020173043 A1 WO 2020173043A1
Authority
WO
WIPO (PCT)
Prior art keywords
user group
target user
target
grouping
grouping rule
Prior art date
Application number
PCT/CN2019/097890
Other languages
English (en)
French (fr)
Inventor
邓悦
金戈
徐亮
Original Assignee
平安科技(深圳)有限公司
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 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2020173043A1 publication Critical patent/WO2020173043A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/24Classification techniques
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00Reducing energy consumption in communication networks
    • Y02D30/70Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • This application relates to the field of data processing technology, and in particular to a method and device for optimizing a user group, and a computer non-volatile readable storage medium.
  • the present application provides a user group optimization method and device, storage medium, and computer equipment, which can increase the coverage of the target classification population without remaking the rules of the decision tree classification model, and improve the optimization efficiency of the user group.
  • a method for optimizing a user group including:
  • the target user group set and the user group total set obtained by grouping the decision tree model, wherein the target user group set includes a first user group corresponding to each leaf node of the decision tree model, and the user group total The set includes a second user group corresponding to each node of the decision tree model;
  • the second user group corresponding to the parent node generates a new target user group set, wherein the number of the target quality users in the new target user group set is greater than or equal to the preset coverage threshold;
  • the final user group is determined.
  • a user group optimization device including:
  • the user group set establishment module is used to obtain the target user group set and the user group total set obtained by the decision tree model grouping, wherein the target user group set includes a first corresponding to each leaf node of the decision tree model.
  • a user group, the total set of user groups includes a second user group corresponding to each node of the decision tree model;
  • the user group optimization module is configured to: if the number of target quality users in the target user group set is less than a preset coverage threshold, then according to the number of grouping rules corresponding to the first user group and the total number of users in the user group The second user group corresponding to the parent node of the first user group node generates a new target user group set, wherein the number of the target quality users in the new target user group set is greater than or equal to the preset Set coverage threshold;
  • the user group determination module is used to determine the final user group according to the new target user group set.
  • a computer non-volatile readable storage medium on which computer readable instructions are stored, and when the program is executed by a processor, the following steps are implemented:
  • the target user group set and the user group total set obtained by grouping the decision tree model, wherein the target user group set includes a first user group corresponding to each leaf node of the decision tree model, and the user group total The set includes a second user group corresponding to each node of the decision tree model;
  • the second user group corresponding to the parent node generates a new target user group set, wherein the number of the target quality users in the new target user group set is greater than or equal to the preset coverage threshold;
  • the final user group is determined.
  • a computer device including a storage medium, a processor, and computer-readable instructions stored on the storage medium and executable on the processor, and the processor executes the computer-readable instructions When implementing the following steps:
  • the target user group set and the user group total set obtained by grouping the decision tree model, wherein the target user group set includes a first user group corresponding to each leaf node of the decision tree model, and the user group total The set includes a second user group corresponding to each node of the decision tree model;
  • the second user group corresponding to the parent node generates a new target user group set, wherein the number of the target quality users in the new target user group set is greater than or equal to the preset coverage threshold;
  • the final user group is determined.
  • this application provides a user group optimization method and device, storage medium, and computer equipment. Compared with the existing user group optimization method, this application is based on the user grouping method based on the decision tree model. , Continue to create a target user group set and a user group total set, where the user groups included in the target user group set correspond to the leaf nodes of the decision tree model, and the user groups included in the user group total set correspond to the nodes of the decision tree model.
  • the number of target quality users in the target user group set is less than the preset coverage threshold, as the number of grouping rules corresponding to the user group in the target user group set is larger, the more target quality users may be eliminated when grouping from the parent node.
  • the parent node user group in the total user group can be combined with the parent node user group in the total user group to find more target quality users, and then according to the number of grouping rules corresponding to the user group in the target user group and the user group
  • the two important factors of the parent node user group in the total user group set can generate a new target user group set with the number of target quality users greater than or equal to the preset coverage threshold, which is the final optimized grouping result.
  • the entire optimization process is simple, without re-establishing grouping rules and retraining decision trees, it can effectively increase the number of target quality users in the target user group set, making the optimization of the user group faster and more accurate, thereby improving the efficiency of user grouping .
  • FIG. 1 shows a schematic flowchart of a method for optimizing a user group provided by an embodiment of the present application
  • FIG. 2 shows a schematic flowchart of another user group optimization method provided by an embodiment of the present application
  • FIG. 3 shows a schematic structural diagram of a user group optimization apparatus provided by an embodiment of the present application
  • FIG. 4 shows a schematic structural diagram of another user group optimization apparatus provided by an embodiment of the present application.
  • Fig. 5 shows a schematic diagram of the physical structure of a computer device provided by an embodiment of the present application.
  • a method for optimizing a user group includes:
  • Step 101 Obtain a target user group set and a user group total set obtained by grouping the decision tree model, where the target user group set includes the first user group corresponding to each leaf node of the decision tree model, and the user group total set includes Each node of the decision tree model corresponds to the second user group.
  • a decision tree is used to group sample users. Specifically, a decision tree classification model is constructed using pre-made grouping rules, and the sample users are grouped using the model. Obtain multiple user groups, where each node of the decision tree corresponds to a user group.
  • the user group corresponding to each leaf node of the decision tree is used to construct the target user group, where all the sample user individuals included in the target user group are different.
  • the user group corresponding to each node of the decision tree is used to construct the user group total set, which facilitates the subsequent use of the user group total set to optimize the target user group.
  • Step 102 If the number of target quality users in the target user group set is less than the preset coverage threshold, then according to the number of grouping rules corresponding to the first user group and the first user group corresponding to the parent node of the first user group node in the total user group set 2. User group, generating a new target user group set, where the number of target quality users in the new target user group set is greater than or equal to the preset coverage threshold.
  • the number of target quality users in the first user group included in the target user group set is less than the preset coverage threshold, it means that the current target user group set does not meet the target population coverage.
  • the number of users should be optimized to make it meet the expectations . Specifically, based on two important factors: the number of grouping rules corresponding to the user group in the target user group set and the parent node user group of the user group in the total user group set, the number of target quality users can be generated greater than or equal to the preset A new target user group set that covers the threshold.
  • the number of grouping rules corresponding to the first user group is large, it means that the group corresponds to The grouping rules are more detailed, so this group is likely to become a user group with a higher proportion of target quality users.
  • the grouping rules for this type of user group it is often ignored because of over-consideration of how to ensure that the target quality is increased. The number of users covered by the target quality, therefore, priority is given to optimizing this user group. After determining the first user group that needs to be optimized, how to optimize it should also be considered.
  • the method provided in this application is to find the node of the first user group in the decision tree model, and use the corresponding parent node of the first user group node.
  • the second user group replaces the first user group that needs to be optimized. This is because the first user group that needs to be optimized is obtained based on the second user group corresponding to the parent node, so the second user group corresponding to the parent node.
  • the number of target quality users included must be greater than or equal to the first user group that needs to be optimized. If replacement is performed, the number of target quality users contained in the new target user group set must be greater than or equal to the original target user group set. Among them, the second user group corresponding to the parent node needs to be found in the total set of user groups.
  • optimization needs to be continued until the number of target quality users is greater than or equal to the preset coverage threshold to achieve the optimization goal.
  • Step 103 Determine the final user group according to the new target user group set.
  • all user groups included in the target user group set can be determined as the final optimized user group to achieve the optimization goal.
  • this application continues to create a target user group set and a total user group set based on the user grouping method based on the decision tree model, wherein the target user group
  • the user groups included in the set correspond to the leaf nodes of the decision tree model
  • the user groups included in the total set of user groups correspond to the nodes of the decision tree model.
  • the parent node user group in the total user group can be combined with the parent node user group in the total user group to find more target quality users, and then according to the number of grouping rules corresponding to the user group in the target user group and the user group
  • the two important factors of the parent node user group in the total user group set can generate a new target user group set with the number of target quality users greater than or equal to the preset coverage threshold, which is the final optimized grouping result.
  • the entire optimization process is simple, without re-establishing grouping rules and retraining decision trees, it can effectively increase the number of target quality users in the target user group set, making the optimization of the user group faster and more accurate, thereby improving the efficiency of user grouping .
  • the method includes:
  • Step 201 Obtain a target user group set and a user group total set obtained by grouping the decision tree model, where the target user group set includes the first user group corresponding to each leaf node of the decision tree model, and the user group total set includes Each node of the decision tree model corresponds to the second user group.
  • the decision tree model is used to group users, and the target user group set and the user group total set are established according to the obtained user group, so that the target user group set is optimized by the user group total set.
  • Step 202 If the number of target quality users in the target user group set is less than the preset coverage threshold, determine the first target user group according to the number of grouping rules corresponding to the first user group, where the first target user group is the target user The first user group with the largest number of grouping rules in the group set.
  • any one of the first user group with the largest number of grouping rules is determined as the first target user group.
  • the first user group with the most grouping rules is searched in the target user group set. Among them, if there are multiple user groups with the most grouping rules, you can Choose any one of them as the first target user group.
  • the user group with the smallest number of users in the user group can also be selected as the first target user group.
  • the above two methods for selecting the first target user group are not limited in this application, and those skilled in the art can arbitrarily select or adopt other methods to determine the first target user group.
  • Step 203 Search for the second target user group in the user group total set according to the first target user group, where the node corresponding to the second target user group is the parent node of the node corresponding to the first target user group.
  • any one of the first user group and the second user group corresponds to a grouping rule or a plurality of grouping rules arranged in sequence.
  • Each user group corresponds to one or more grouping rules, and the order of the grouping rules is consistent with the order of the grouping rules of nodes on the corresponding decision tree model.
  • the classification rules corresponding to a node of the decision tree classification model are from top to bottom: users whose occupation is white-collar, users with undergraduate education or above, users who purchase wealth management products more than 10,000 yuan each year... Then, the grouping rules for the corresponding user groups of this node are, in order, users with white-collar occupations, users with undergraduate education or above, users who purchase wealth management products more than 10,000 yuan each year...
  • the foregoing step 203 may include:
  • Step 2031 Determine the grouping rule corresponding to the first target user group as the first grouping rule, and delete the last one of the first grouping rule to obtain the second grouping rule.
  • the manner of determining the first grouping rule may include:
  • the grouping rule corresponding to the first target user group is determined as the first grouping rule, where
  • the proportion of target quality users is the ratio of the number of target quality users in the first target user group to the number of all users in the first target user group;
  • the re-determined grouping rule corresponding to the first target user group is determined as the first grouping rule.
  • the first target user group On the basis of screening the first target user group by the number of grouping rules, the first target user group also needs to meet the condition of the proportion of target quality users. Specifically, after determining the first target user group with the largest number of grouping rules, calculate the first target The ratio of the number of target quality users in the user group to the number of all users in the first target user group. If this ratio is greater than or equal to the preset proportion threshold, it means that the first target user group can be optimized, and the first target user is determined The first grouping rule corresponding to the group, so as to use the first grouping rule to optimize the first target user group.
  • the grouping rules of the user group are more detailed and the quality should be Higher, so if the grouping rule is deleted based on the grouping rule corresponding to the first target user group, the quality of the user group corresponding to the deleted grouping rule will be lower than that of the first target user group , Because the decision tree model node corresponding to the user group that satisfies the reduced grouping rule is the parent node of the decision tree model node corresponding to the first target user group, and user group optimization must ensure the overall target quality and the number of users covered. It is necessary to ensure the proportion of target quality users in each user group as much as possible. At this time, it is not appropriate to still select the first target user group, but to re-determine a higher quality user group as the first target user group.
  • the newly determined first target user group should meet the following conditions: first, it is different from the original first target user group; second, the number of grouping rules is the largest; third, the proportion of target quality users is greater than or equal to Preset the proportion threshold. After the new first target user group is searched according to the above conditions, the first grouping rule is determined according to the new first target user group.
  • Step 2032 According to the second grouping rule, search for a second target user group corresponding to the second grouping rule in the total set of user groups.
  • the method of searching for the second target user group specifically includes:
  • the second user group corresponding to the second grouping rule is determined as the second target user group, where the proportion of target quality users in the second target user group is the number of target quality users in the second target user group and The ratio of the number of all users in the second target user group;
  • the second grouping rule search for the corresponding user group in the total set of user groups, and calculate the ratio of the number of target quality users to the total number of users in the user group. If the ratio is greater than or equal to the preset percentage threshold, the user The quality of the group is good, and the user group is the user group corresponding to the parent node of the first target user group. The total number of people in the user group should be higher than that of the first target user group. If it is a replacement user for the first target user group Groups help to increase the number of target quality users in the target user group set.
  • the target user group A new first target user group different from the original first target user group is re-determined in the group set, and then the first grouping rule and the second grouping rule are determined, and the second target user group is re-determined according to the above rules.
  • Step 204 Replace the first target user group in the target user group set with the second target user group to generate a new target user group set.
  • the node corresponding to the second target user group is the first target user
  • the parent node of the node corresponding to the group, and both the first target user group and the second target user group meet the target quality user ratio is greater than or equal to the preset user group target sample ratio threshold, so in the target user group set after replacement,
  • the coverage of target quality users will be increased, that is, the quality of the target user group set will be improved, and the proportion of target quality users in the second target user group is also greater than or equal to the preset proportion threshold, so the first
  • the replacement of a target user group with a second target user group can also ensure a higher quality of the user group.
  • Step 205 Query grouping rules corresponding to any first user group in the new target user group set, and delete the target user group when grouping rules corresponding to other first user groups include grouping rules corresponding to any first user group The other first user groups in the set.
  • Replacing the first target user group with the second target user group essentially replaces the user group corresponding to the parent node in the decision tree model with the user group corresponding to the child node, and a parent node can correspond to one or more child nodes. If there are multiple sub-nodes corresponding to the node corresponding to the second target user group, the users included in the multiple user groups may overlap each other in the target user group set. Therefore, overlapping users need to be deduplicated. Specifically, the user group users corresponding to the parent node must include the user group users corresponding to all the child nodes.
  • the user group corresponding to the child node is deleted, or That is, if the grouping rules corresponding to any user group are all included in the grouping rules corresponding to another user group, the other user group is deleted from the target user group set, thereby removing duplicate users in the target user group set.
  • Step 206 Calculate the number of the first user group included in the new target user group set.
  • the proportion of users reaching the preset proportion threshold and the total number of target quality users included in all user groups reaching the preset In addition to the coverage threshold, a certain number of user groups needs to be guaranteed. Therefore, before determining the final user group, it is also necessary to calculate whether the number of user groups in the target user group set meets the expected number.
  • Step 207 If the number of the first user group is greater than or equal to the preset user group threshold, calculate the number of target quality users in the new target user group set.
  • the number of user groups in the target user group set is greater than or equal to the preset target user group number threshold, indicating that the number of user groups meets the expected number of groups at this time, then the number of target quality users covered by the target user group set can be further judged Whether to reach the preset target user group quantity threshold, so as to realize the optimization of the user group.
  • the first grouping rule is determined according to the new first target user group, the second grouping rule and the corresponding second target user group are determined, and the user group replacement is performed again.
  • the user groups included in the target user group set meet the conditions of the preset target user group quantity threshold, and obtain a higher quality user group.
  • Step 208 If the number of target quality users in the new target user group set is less than the preset coverage threshold, return to the step of determining the first target user group to re-determine the first target user group until the target instruction in the new target user group set The number of users is greater than or equal to the preset coverage threshold, where the newly determined first target user group is the first user group with the largest number of grouping rules different from the original first target user group.
  • the number of target quality users in the target user group set is still less than the preset coverage threshold, it means that it is necessary to continue to optimize the target user group set, then return to the first target user group search step, and search in the target user group set with the original A new first target user group whose first target user group is different and has the largest number of corresponding grouping rules, and the proportion of target quality users is greater than or equal to the preset user group target sample proportion threshold. Thereby, re-optimization is performed according to the new first target user group, so that the target user group set meets expectations, and the coverage of target quality users is improved.
  • the existing user groups and their corresponding grouping rules are used to recombine the existing user groups and use user groups with fewer grouping rules.
  • the user group nodes with fewer grouping rules mentioned above are the parent nodes of user group nodes with too detailed grouping rules to increase the coverage of target quality users, and the number of user groups added is lower than expected.
  • a protection mechanism with a threshold for the number of target user groups is set to ensure that the number of user groups is guaranteed, and the number of users covered by the target quality is met, and the optimization of the user group is realized.
  • an embodiment of the present application provides a user group optimization device.
  • the device includes: a user group set establishment module 31, a user group optimization module 32, and a user group Determine module 33.
  • the user group set establishment module 31 is used to obtain the target user group set and the user group total set obtained by grouping the decision tree model, wherein the target user group set includes the first user group corresponding to the leaf node of the decision tree model, and the user group total The set includes a second user group corresponding to each node of the decision tree model;
  • the user group optimization module 32 is configured to, if the number of target quality users in the target user group set is less than the preset coverage threshold, according to the number of grouping rules corresponding to the first user group and the number of the first user group nodes in the total set of user groups The second user group corresponding to the parent node generates a new target user group set, where the number of target quality users in the new target user group set is greater than or equal to the preset coverage threshold;
  • the user group determining module 33 is used to determine the final user group according to the new target user group set.
  • the user group optimization module 32 specifically includes: a first target user group determining unit 321, a second target user group determining unit 322, and a replacement unit 323.
  • the first target user group determining unit 321 is configured to determine the first target user group according to the number of grouping rules corresponding to the first user group, where the first target user group is the first user with the largest number of grouping rules in the target user group set group;
  • the second target user group determining unit 322 is configured to search for the second target user group in the total set of user groups according to the first target user group, where the node corresponding to the second target user group is the node corresponding to the first target user group The parent node;
  • the replacement unit 323 is configured to replace the first target user group in the target user group set with the second target user group to generate a new target user group set;
  • the first target user group determining unit 321 is further configured to return to the step of determining the first target user group to re-determine the first target user group if the number of target quality users in the new target user group set is less than the preset coverage threshold.
  • the number of target instruction users in the new target user group set is greater than or equal to the preset coverage threshold, where the newly determined first target user group is the first user with the largest number of grouping rules different from the original first target user group group.
  • any one of the first user group and the second user group corresponds to a grouping rule or a plurality of grouping rules arranged in sequence.
  • the second target user group determination unit 322 specifically includes: a grouping rule determination subunit 3221, a second target user group determination subunit 3222.
  • the grouping rule determination subunit 3221 is configured to determine the grouping rule corresponding to the first target user group as the first grouping rule, and delete the last item of the first grouping rule to obtain the second grouping rule;
  • the second target user group determination subunit 3222 is configured to search for the second target user group corresponding to the second grouping rule in the total set of user groups according to the second grouping rule.
  • the grouping rule determination subunit 3221 is further configured to determine the grouping rule corresponding to the first target user group as the first grouping rule if the proportion of the target quality users in the first target user group is greater than or equal to the preset proportion threshold, where , The proportion of target quality users in the first target user group is the ratio of the number of target quality users in the first target user group to the number of all users in the first target user group;
  • the proportion of target quality users in the first target user group is less than the preset proportion threshold, return to the step of determining the first target user group to re-determine the first target user group until the re-determined target quality users in the first target user group After the proportion is greater than or equal to the preset proportion threshold, the re-determined grouping rule corresponding to the first target user group is determined as the first grouping rule.
  • the user group optimization module 32 further includes: a deleting unit 324, a user group number calculation unit 325, and a target quality user number calculation unit 326.
  • the deleting unit 324 is configured to replace the first target user group in the target user group set with the second target user group, and after a new target user group set is generated, query the correspondence of any first user group in the new target user group set When grouping rules corresponding to other first user groups include grouping rules corresponding to any one of the first user groups, delete other first user groups in the target user group set.
  • the user group quantity calculation unit 325 is configured to calculate the number of first user groups included in the new target user group set after deleting other first user groups in the target user group set;
  • the target quality user quantity calculation unit 326 is configured to calculate the number of target quality users in the new target user group set if the number of the first user group is greater than or equal to the preset user group threshold.
  • the second target user group determination subunit 3222 is specifically configured to search for the second user group corresponding to the second grouping rule in the user group total set, if the target quality user in the second user group corresponding to the second grouping rule If the proportion is greater than or equal to the preset proportion threshold, the second user group corresponding to the second grouping rule is determined as the second target user group, where the proportion of the target quality users in the second target user group is the second The ratio of the number of target quality users in the target user group to the number of all users in the second target user group;
  • an embodiment of the present application also provides a storage medium on which a computer program is stored.
  • the program is executed by a processor, the above-mentioned steps shown in Figures 1 and 2 are implemented.
  • an embodiment of the present application also provides a computer non-volatile readable storage medium, on which computer readable instructions are stored, when the computer readable instructions are executed by the processor
  • the following steps are achieved: Obtain the target user group set and the user group total set obtained by the decision tree model grouping, wherein the target user group set includes the first user group corresponding to each leaf node of the decision tree model, and the user group total set includes The second user group corresponding to each node of the decision tree model; if the number of target quality users in the target user group set is less than the preset coverage threshold, according to the number of grouping rules corresponding to the first user group and the total set of user groups In the second user group corresponding to the parent node of the first user group node, a new target user group set is generated, where the number of target quality users in the new target user group set is greater than or equal to the preset coverage threshold; Set the target user group to determine the final user group.
  • the technical solution of this application can be embodied in the form of a software product.
  • the software product can be stored in a non-volatile storage medium (which can be a CD-ROM, U disk, mobile hard disk, etc.), including several
  • the instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute the methods described in each implementation scenario of this application.
  • the computer device includes: The processor 41, the memory 42, and computer-readable instructions stored on the memory 42 and capable of running on the processor, wherein the memory 42 and the processor 41 are both set on the bus 43, and the processor 41 implements the following steps when the processor 41 executes the program: Get decision The target user group set and the user group total set obtained by grouping the tree model, where the target user group set includes the first user group corresponding to each leaf node of the decision tree model, and the user group total set includes each user group corresponding to the decision tree model.
  • Each node corresponds to the second user group; if the number of target quality users in the target user group set is less than the preset coverage threshold, then according to the number of grouping rules corresponding to the first user group and the total set of user groups with the first user group
  • the second user group corresponding to the parent node of the node generates a new target user group set, where the number of target quality users in the new target user group set is greater than or equal to the preset coverage threshold; according to the new target user group set, determine The ultimate user base.
  • the computer device also includes a bus 43 configured to couple the processor 41 and the memory 42.
  • the computer device may also include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a WI-FI module, and so on.
  • the user interface may include a display screen (Display), an input unit such as a keyboard (Keyboard), etc., and the optional user interface may also include a USB interface, a card reader interface, etc.
  • the network interface can optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a WI-FI interface), etc.
  • the memory may also include an operating system and a network communication module.
  • the operating system is a program that manages the hardware and software resources of computer equipment, and supports the operation of information processing programs and other software and/or programs.
  • the network communication module is used to realize the communication between the various components in the memory and the communication with other hardware and software in the physical device.

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本申请公开了一种用户群的优化方法及装置、计算机非易失性可读存储介质,该方法包括:获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的每个叶子节点分别对应的第一用户群,用户群总集合包括与决策树模型的每个节点分别对应的第二用户群;若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量和用户群总集合中与第一用户群节点的父节点对应的第二用户群,生成新的目标用户群集合,其中,新的目标用户群集合中目标质量用户的数量大于或等于预设覆盖阈值;根据新的目标用户群集合,确定最终的用户群。本申请的用户群优化方法简单有效。

Description

用户群的优化方法及装置、计算机非易失性可读存储介质
本申请要求与2019年2月28日提交中国专利局、申请号为201910152473.6、申请名称为“用户群的优化方法及装置、存储介质、计算机设备”的中国专利申请的优先权,其全部内容通过引用结合在申请中。
技术领域
本申请涉及数据处理技术领域,尤其是涉及到一种用户群的优化方法及装置、计算机非易失性可读存储介质。
背景技术
现有的基于决策树的用户分群方法中,为了保证每个用户群的目标分类人群占比,通常会对决策树分类模型设置很多分群规则,使得每个用户群都对应有多条分群规则,但是从分群的结果来看,这种过于细致的分群规则会导致每个用户群保留的用户数量较少,从整体来看全部的用户群所覆盖的目标分类人群达不到预期数量。
而现有的用户群优化方式中,通常通过重新建立分群规则,重新训练决策树的方法解决上述问题,这种方法在分群规则复杂、用户数据量较大时会耗费很多时间。目前,还没有一种能够快速有效进行用户群优化的方法。
发明内容
有鉴于此,本申请提供了一种用户群的优化方法及装置、存储介质、计算机设备,无需重新制定决策树分类模型的规则即可提高目标分类人群覆盖量,提升了用户群的优化效率。
根据本申请的一个方面,提供了一种用户群的优化方法,包括:
获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
若在所述第一用户群中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
根据所述新的目标用户群集合,确定最终的用户群。
根据本申请的另一方面,提供了一种用户群的优化装置,包括:
用户群集合建立模块,用于获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
用户群优化模块,用于若在所述目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
用户群确定模块,用于根据所述新的目标用户群集合,确定最终的用户群。
依据本申请又一个方面,提供了一种计算机非易失性可读存储介质,其上存储有计算机可读指令,所述程序被处理器执行时实现以下步骤:
获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
若在所述第一用户群中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
根据所述新的目标用户群集合,确定最终的用户群。
依据本申请再一个方面,提供了一种计算机设备,包括存储介质、处理器及存储在存储介质上并可在处理器上运行的计算机可读指令,所述处理器执行所述计算机可读指令时实现以下步骤:
获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
若在所述第一用户群中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
根据所述新的目标用户群集合,确定最终的用户群。
借由上述技术方案,本申请提供的一种用户群的优化方法及装置、存储介质、计算机设备,与现有的用户群优化方式相比,本申请在基于决策树模型的用户分群方式基础上, 继续创建目标用户群集合和用户群总集合,其中目标用户群集合中包括的用户群对应决策树模型的叶子节点,而用户群总集合包括的用户群对应决策树模型的节点。在目标用户群集合中目标质量用户的数量小于预设覆盖阈值时,由于在目标用户群集合中用户群对应的分群规则数量越多,其相应从父节点分群时可能剔除的目标质量用户越多,因此为了提高目标质量用户的数量可结合其在用户群总集合中的父节点用户群进行查找更多的目标质量用户,进而根据目标用户群集合中用户群对应的分群规则数量和该用户群在用户群总集合中的父节点用户群这两个重要因素,可生成目标质量用户的数量大于或等于预设覆盖阈值的新目标用户群集合,即为最终优化的分群结果。整个优化方案过程简单,无需重新建立分群规则和重新训练决策树,即可有效提高目标用户群集合中的目标质量用户的数量,使得用户群的优化变得更加快捷准确,从而提高了用户分群效率。
上述说明仅是本申请技术方案的概述,为了能够更清楚了解本申请的技术手段,而可依照说明书的内容予以实施,并且为了让本申请的上述和其它目的、特征和优点能够更明显易懂,以下特举本申请的具体实施方式。
附图说明
此处所说明的附图用来提供对本申请的进一步理解,构成本申请的一部分,本申请的示意性实施例及其说明用于解释本申请,并不构成对本申请的不当限定。在附图中:
图1示出了本申请实施例提供的一种用户群的优化方法的流程示意图;
图2示出了本申请实施例提供的另一种用户群的优化方法的流程示意图;
图3示出了本申请实施例提供的一种用户群的优化装置的结构示意图;
图4示出了本申请实施例提供的另一种用户群的优化装置的结构示意图;
图5示出了本申请实施例提供的一种计算机设备的实体结构示意图。
具体实施方式
下文中将参考附图并结合实施例来详细说明本申请。需要说明的是,在不冲突的情况下,本申请中的实施例及实施例中的特征可以相互组合。
在本实施例中提供了一种用户群的优化方法,如图1所示,该方法包括:
步骤101,获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的每个叶子节点分别对应的第一用户群,用户群总集合包括与决策树模型的每个节点分别对应的第二用户群。
在本申请中,为了得到互相之间没有交集的用户群,采用决策树对样本用户进行分群,具体地,利用事先制定好的分群规则构建决策树分类模型,并利用模型对样本用户进行分群,得到多个用户群,其中,决策树的每个节点都对应有一个用户群。
应用预设决策树分类模型对样本用户进行分类后,利用决策树的每个叶子节点对应的用户群构建目标用户群,其中,目标用户群中包含的所有样本用户个体各不相同。另外,利用决策树的每个节点对应的用户群构建用户群总集合,从而便于后续利用用户群总集合对目标用户群进行优化。
步骤102,若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量和用户群总集合中与第一用户群节点的父节点对应的第二用户群,生成新的目标用户群集合,其中,新的目标用户群集合中目标质量用户的数量大于或等于预设覆盖阈值。
如果目标用户群集合中所包含的第一用户群中目标质量用户的数量小于预设覆盖阈值,说明目前得到的目标用户群集合不符合目标人群覆盖预期人数应进行用户群优化以使其满足预期。具体地,可以根据目标用户群集合中用户群对应的分群规则数量和该用户群在用户群总集合中的父节点用户群这两个重要因素,可生成目标质量用户的数量大于或等于预设覆盖阈值的新目标用户群集合。
例如,可以根据每个第一用户群对应的分群规则数量寻找需要被优化的第一用户群,详细来说,如果第一用户群对应的分群规则数量较多,也就意味着该群对应的分群规则更为细致,那么该群则很可能成为目标质量用户占比较高的用户群,而在设置这类用户群的分群规则时往往会因为过度考虑如何保证提高目标质量用户占比而忽略了目标质量用户的覆盖数量,因此优先考虑对这类用户群进行优化。进而确定需要优化的第一用户群之后,还应考虑如何对其进行优化,本申请提供的方法为找到第一用户群在决策树模型中的节点,利用第一用户群节点的父节点对应的第二用户群对需要优化的第一用户群进行替换,这是因为由于需要优化的第一用户群是根据该父节点对应的第二用户群得来的,所以父节点对应的第二用户群所包含的目标质量用户的数量一定大于或等于需要优化的第一用户群,若进行替换,则新的目标用户群集合中所包含的目标质量用户的数量一定大于或等于原有的目标用户群集合。其中,需要在用户群总集合中寻找父节点对应的第二用户群。
另外,如果一次优化后,目标用户群集合包含的目标质量用户的数量仍然小于预设覆盖阈值,则需要继续进行优化,直到目标质量用户的数量大于或等于预设覆盖阈值,以达到优化目的。
步骤103,根据新的目标用户群集合,确定最终的用户群。
目标用户群集合包含的目标质量用户的数量大于或等于预设覆盖阈值后,则可以将目标用户群集合中包含的全部用户群确定为最终的优化后的用户群,实现优化目的。
通过应用本实施例的技术方案,与现有的用户群优化方式相比,本申请在基于决策树模型的用户分群方式基础上,继续创建目标用户群集合和用户群总集合,其中目标用户群集合中包括的用户群对应决策树模型的叶子节点,而用户群总集合包括的用户群对应决策树模型的节点。在目标用户群集合中目标质量用户的数量小于预设覆盖阈值时,由于在目标用户群集合中用户群对应的分群规则数量越多,其相应从父节点分群时可能剔除的目标质量用户越多,因此为了提高目标质量用户的数量可结合其在用户群总集合中的父节点用 户群进行查找更多的目标质量用户,进而根据目标用户群集合中用户群对应的分群规则数量和该用户群在用户群总集合中的父节点用户群这两个重要因素,可生成目标质量用户的数量大于或等于预设覆盖阈值的新目标用户群集合,即为最终优化的分群结果。整个优化方案过程简单,无需重新建立分群规则和重新训练决策树,即可有效提高目标用户群集合中的目标质量用户的数量,使得用户群的优化变得更加快捷准确,从而提高了用户分群效率。
进一步的,作为上述实施例具体实施方式的细化和扩展,为了完整说明本实施例的具体实施过程,提供了另一种用户群的优化方法,如图2所示,该方法包括:
步骤201,获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的每个叶子节点分别对应的第一用户群,用户群总集合包括与决策树模型的每个节点分别对应的第二用户群。
利用决策树模型对用户进行分群,并根据所得用户群建立目标用户群集合以及用户群总集合,从而利用用户群总集合对目标用户群集合进行优化。
步骤202,若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量,确定第一目标用户群,其中,第一目标用户群为目标用户群集合中分群规则数量最多的第一用户群。
在上述实施例中,具体地,若分群规则数量最多的第一用户群包括多个,则确定多个分群规则数量最多的第一用户群中的任意一个为第一目标用户群。
如果目标用户群集合的目标质量用户数量低于预设覆盖阈值,则在目标用户群集合中,搜索分群规则最多的第一用户群,其中,如果分群规则数量最多的用户群有多个,可以选其中任意一个作为第一目标用户群。
需要说明的是,如果分群规则数量最多的用户群包括多个,还可以选择用户群中人数最少的用户群作为第一目标用户群。上述两种选择第一目标用户群的方式本申请在此不做限定,本领域技术人员可以任意选择或采用其他方法确定第一目标用户群。
步骤203,根据第一目标用户群,在用户群总集合中搜索第二目标用户群,其中,第二目标用户群对应的节点为第一目标用户群对应的节点的父节点。
在上述实施例中,具体地,第一用户群和第二用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则。
每个用户群都对应有一条或者多条分群规则,分群规则的顺序与对应的决策树模型上的节点的分群规则顺序一致。
例如,决策树分类模型的一个节点对应的分类规则自上而下分别为:职业为白领的用户、受教育程度为大学本科及本科以上的用户、每年购买理财产品超过1万元的用户……那么,该节点的对应用户群的分群规则按照顺序依次分别为职业为白领的用户、受教育程度为大学本科及本科以上的用户、每年购买理财产品超过1万元的用户……
作为本申请的一个具体实施例,上述步骤203可以包括:
步骤2031,将第一目标用户群对应的分群规则确定为第一分群规则,并将第一分群规则的最后一条删除得到第二分群规则。
确定第一目标用户群对应的第一分群规则后,将第一分群规则删除最后一条得到第二分群规则,这是因为每个用户群的分群规则都是按照顺序排列的,因此将第一分群规则删除最后一条后剩余的分群规则即为第一目标用户群节点的父节点对应的分群规则。
另外,具体地,确定第一分群规则的方式可以包括:
第一,若第一目标用户群中目标质量用户占比大于或等于预设占比阈值,则将第一目标用户群对应的分群规则确定为第一分群规则,其中,第一目标用户群中目标质量用户占比为第一目标用户群中目标质量用户的数量与第一目标用户群中全部用户的数量的比值;
第二,若第一目标用户群中目标质量用户占比小于预设占比阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至重新确定的第一目标用户群中目标质量用户占比大于或等于预设占比阈值后,将重新确定的第一目标用户群对应的分群规则确定为第一分群规则。
在利用分群规则数量筛选第一目标用户群的基础上,第一目标用户群还需要满足目标质量用户占比条件,具体地,确定分群规则数量最多的第一目标用户群后,计算第一目标用户群中的目标质量用户数量与第一目标用户群中全部用户数量的比值,如果这个比值大于或等于预设占比阈值,说明可以对第一目标用户群进行优化,则确定第一目标用户群对应的第一分群规则,以便利用第一分群规则实现第一目标用户群的优化。
如果上述比值小于预设占比阈值,说明该第一目标用户群质量较低,而相较于这个用户群对应的父节点的用户群来说,该用户群的分群规则更为细致,质量应该更高,所以如果仍然以该第一目标用户群对应的分群规则作为基础进行分群规则的删减,将会导致删减后的分群规则所对应的用户群质量比第一目标用户群质量更低,因为满足删减后的分群规则的用户群对应的决策树模型节点是第一目标用户群对应的决策树模型节点的父节点,而用户群优化除了要保证整体的目标质量用户覆盖数量,还需要尽可能保证每个用户群中目标质量用户占比,则此时不宜仍选用该第一目标用户群,而应重新确定质量更高的用户群作为第一目标用户群。
具体地,重新确定的第一目标用户群应该满足的条件为:第一,与原有的第一目标用户群不同;第二,分群规则数量最多;第三,目标质量用户占比大于或等于预设占比阈值。按照上述条件搜索到新的第一目标用户群之后,根据新的第一目标用户群确定第一分群规则。
步骤2032,根据第二分群规则,在用户群总集合中搜索与第二分群规则对应的第二目标用户群。
另外,具体地,搜索第二目标用户群的方式具体包括:
第一,在用户群总集合中搜索与第二分群规则对应的第二用户群,若与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值,则将与第二分群规则对 应的第二用户群确定为第二目标用户群,其中,第二目标用户群中的目标质量用户占比为第二目标用户群中目标质量用户的数量与第二目标用户群中全部用户的数量的比值;
第二,若与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值。
根据第二分群规则在用户群总集合中搜索对应的用户群,并计算该用户群中目标质量用户数量与全部用户数量的比值,如果该比值大于或等于预设占比阈值,说明该用户群的质量较好,而该用户群为第一目标用户群的父节点对应的用户群,该用户群包含的总人数应高于第一目标用户群,如果作为第一目标用户群的替换用户群,有助于提高目标用户群集合中目标质量用户的数量。
另外,在本实施例中,与第一目标用户群需满足的条件相似的,若与第二分群规则对应的用户群中目标质量用户占比小于预设占比阈值,则应在目标用户群集合中重新确定与原有的第一目标用户群不同的新的第一目标用户群,进而确定第一分群规则、第二分群规则,并按照上述规则重新确定第二目标用户群。
步骤204,将目标用户群集合中的第一目标用户群替换为第二目标用户群,生成新的目标用户群集合。
确定第二目标用户群之后,将目标用户群集合中的第一目标用户群替换为第二目标用户群,由于对于决策树分类模型来说,第二目标用户群对应的节点为第一目标用户群对应的节点的父节点,并且第一目标用户群与第二目标用户群都满足目标质量用户占比大于或等于预设用户群目标样本占比阈值,因此替换后的目标用户群集合中,目标质量用户的覆盖数量将会有所提高,也即提高了目标用户群集合的质量,并且第二目标用户群中目标质量用户的占比也是大于或等于预设占比阈值的,因此将第一目标用户群替换为第二目标用户群也能够保证用户群的较高质量。
步骤205,查询新的目标用户群集合中任意一个第一用户群对应的分群规则,并在其他第一用户群对应的分群规则包括任意一个第一用户群对应的分群规则时,删除目标用户群集合中的其他第一用户群。
将第一目标用户群替换为第二目标用户群,本质上是将决策树模型中父节点对应的用户群替换为子节点对应的用户群,而一个父节点又可以对应一个或多个子节点,若与第二目标用户群相应的节点对应有多个子节点,那么目标用户群集合中将可能出现多个用户群包括的用户相互重叠的现象,因此需要将重叠的用户进行去重。具体地,父节点对应的用户群用户一定包括所有子节点对应的用户群用户,因此在目标用户群中父节点和子节点对应的用户群同时存在的情况下,删除子节点对应的用户群,或者说,如果任一用户群对应的分群规则全部包含在另一个用户群对应的分群规则内,则将另一个用户群从目标用户群集合中删除,从而去掉目标用户群集合中的重复用户。
步骤206,计算新的目标用户群集合包含的第一用户群的数量。
对于目标用户群集合中所包含的用户群来说,除了要尽量满足每个用户群的目标质量用户占比达到预设占比阈值以及所有用户群所包含的目标质量用户的总数量达到预设覆盖阈值以外,还需保证一定的用户群数量,因此在确定最终的用户群之前,还需要计算目标用户群集合中的用户群数量是否符合预期数量。
步骤207,若第一用户群的数量大于或等于预设用户群阈值,则计算新的目标用户群集合中目标质量用户的数量。
如果目标用户群集合中的用户群数量大于或等于预设目标用户群数量阈值,说明此时用户群的数量满足分群期望数量,则可以进一步地判断目标用户群集合所涵盖的目标质量用户的数量是否达到预设目标用户群数量阈值,从而实现用户群的优化。
另外,在本申请的实施例中,具体地,若第一用户群的数量大于或等于预设用户群阈值,则返回搜索第一目标用户群的步骤重新确定第一目标用户群。从而根据新的第一目标用户群确定第一分群规则,进而确定第二分群规则以及对应的第二目标用户群,并重新进行用户群的替换。以使目标用户群集合所包含的用户群满足预设目标用户群数量阈值的条件,得到更优质的用户群。
步骤208,若新的目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至新的目标用户群集合中目标指令用户的数量大于或等于预设覆盖阈值,其中,重新确定的第一目标用户群为与原有的第一目标用户群不同的分群规则数量最多的第一用户群。
如果目标用户群集合中的目标质量用户的数量仍然小于预设覆盖阈值,说明需要继续对目标用户群集合进行优化,则返回第一目标用户群搜索步骤,在目标用户群集合中搜索与原有的第一目标用户群不同的,且对应的分群规则数量最多的,同时目标质量用户占比大于或等于预设用户群目标样本占比阈值的新的第一目标用户群。从而根据新的第一目标用户群重新进行优化,以使目标用户群集合符合预期,提高目标质量用户的覆盖数量。
通过应用本实施例的技术方案,无需重新制定决策树分类模型的规则,利用已有的用户群及其对应的分群规则,将已有的用户群进行重新组合,用分群规则较少的用户群替代分群规则过于细致的用户群,其中上述分群规则较少的用户群节点为上述分群规则过于细致的用户群节点的父节点,以提高目标质量用户的覆盖数量,并加入用户群数量低于预设目标用户群数量阈值的防护机制,保证用户群数量的基础上,满足了目标质量用户的覆盖数量,实现了用户群的优化。
进一步的,作为图1方法的具体实现,本申请实施例提供了一种用户群的优化装置,如图3所示,该装置包括:用户群集合建立模块31、用户群优化模块32、用户群确定模块33。
用户群集合建立模块31,用于获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的叶子节点对应的第一用户群,用户群总集合包括与决策树模型的每个节点对应的第二用户群;
用户群优化模块32,用于若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量和用户群总集合中与第一用户群节点的父节点对应的第二用户群,生成新的目标用户群集合,其中,新的目标用户群集合中目标质量用户的数量大于或等于预设覆盖阈值;
用户群确定模块33,用于根据新的目标用户群集合,确定最终的用户群。
在具体的应用场景中,如图4所示,用户群优化模块32,具体包括:第一目标用户群确定单元321、第二目标用户群确定单元322、替换单元323。
第一目标用户群确定单元321,用于根据第一用户群对应的分群规则数量,确定第一目标用户群,其中,第一目标用户群为目标用户群集合中分群规则数量最多的第一用户群;
第二目标用户群确定单元322,用于根据第一目标用户群,在用户群总集合中搜索第二目标用户群,其中,第二目标用户群对应的节点为第一目标用户群对应的节点的父节点;
替换单元323,用于将目标用户群集合中的第一目标用户群替换为第二目标用户群,生成新的目标用户群集合;
第一目标用户群确定单元321,还用于若新的目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至新的目标用户群集合中目标指令用户的数量大于或等于预设覆盖阈值,其中,重新确定的第一目标用户群为与原有的第一目标用户群不同的分群规则数量最多的第一用户群。
在本申请的实施例中,具体地,第一用户群和第二用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则。
在具体的应用场景中,如图4所示,第二目标用户群确定单元322,具体包括:分群规则确定子单元3221、第二目标用户群确定子单元3222。
分群规则确定子单元3221,用于将第一目标用户群对应的分群规则确定为第一分群规则,并将第一分群规则的最后一条删除得到第二分群规则;
第二目标用户群确定子单元3222,用于根据第二分群规则,在用户群总集合中搜索与第二分群规则对应的第二目标用户群。
分群规则确定子单元3221,还用于若第一目标用户群中目标质量用户占比大于或等于预设占比阈值,则将第一目标用户群对应的分群规则确定为第一分群规则,其中,第一目标用户群中目标质量用户占比为第一目标用户群中目标质量用户的数量与第一目标用户群中全部用户的数量的比值;
若第一目标用户群中目标质量用户占比小于预设占比阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至重新确定的第一目标用户群中目标质量用户占比大于或等于预设占比阈值后,将重新确定的第一目标用户群对应的分群规则确定为第一分群规则。
在具体的应用场景中,如图4所示,用户群优化模块32,还包括:删除单元324、用户群数量计算单元325、目标质量用户数量计算单元326。
删除单元324,用于将目标用户群集合中的第一目标用户群替换为第二目标用户群,生成新的目标用户群集合之后,查询新的目标用户群集合中任意一个第一用户群对应的分群规则,并在其他第一用户群对应的分群规则包括任意一个第一用户群对应的分群规则时,删除目标用户群集合中的其他第一用户群。
用户群数量计算单元325,用于删除目标用户群集合中的其他第一用户群之后,计算新的目标用户群集合包含的第一用户群的数量;
目标质量用户数量计算单元326,用于若第一用户群的数量大于或等于预设用户群阈值,则计算新的目标用户群集合中目标质量用户的数量。
第二目标用户群确定子单元3222,具体用于在用户群总集合中搜索与第二分群规则对应的第二用户群,若与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值,则将与第二分群规则对应的第二用户群确定为第二目标用户群,其中,第二目标用户群中的目标质量用户占比为第二目标用户群中目标质量用户的数量与第二目标用户群中全部用户的数量的比值;
若与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值,则返回确定第一目标用户群的步骤重新确定第一目标用户群,直至与第二分群规则对应的第二用户群中目标质量用户占比大于或等于预设占比阈值。
需要说明的是,本申请实施例提供的一种用户群的优化装置所涉及各功能单元的其他相应描述,可以参考图1和图2中的对应描述,在此不再赘述。
基于上述如图1和图2所示方法,相应的,本申请实施例还提供了一种存储介质,其上存储有计算机程序,该程序被处理器执行时实现上述如图1和图2所示的用户群的优化方法。
基于上述如图1所示方法,相应的,本申请实施例还提供了一种计算机非易失性可读存储介质,其上存储有计算机可读指令,该计算机可读指令被处理器执行时实现以下步骤:获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的每个叶子节点分别对应的第一用户群,用户群总集合包括与决策树模型的每个节点分别对应的第二用户群;若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量和用户群总集合中与第一用户群节点的父节点对应的第二用户群,生成新的目标用户群集合,其中,新的目标用户群集合中目标质量用户的数量大于或等于预设覆盖阈值;根据新的目标用户群集合,确定最终的用户群。
基于这样的理解,本申请的技术方案可以以软件产品的形式体现出来,该软件产品可以存储在一个非易失性存储介质(可以是CD-ROM,U盘,移动硬盘等)中,包括若干指令用以使得一台计算机设备(可以是个人计算机,服务器,或者网络设备等)执行本申请各个实施场景所述的方法。
基于上述如图1所示方法和如图3所示产品数据推送装置的实施例,本申请实施例还提供了一种计算机设备的实体结构图,如图5所示,该计算机设备包括:处理器41、存储器42、及存储在存储器42上并可在处理器上运行的计算机可读指令,其中存储器42和处理器41均设置在总线43上处理器41执行程序时实现以下步骤:获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,目标用户群集合包括与决策树模型的每个叶子节点分别对应的第一用户群,用户群总集合包括与决策树模型的每个节点分别对应的第二用户群;若在目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据第一用户群对应的分群规则数量和用户群总集合中与第一用户群节点的父节点对应的第二用户群,生成新的目标用户群集合,其中,新的目标用户群集合中目标质量用户的数量大于或等于预设覆盖阈值;根据新的目标用户群集合,确定最终的用户群。该计算机设备还包括:总线43,被配置为耦接处理器41及存储器42。
可选地,该计算机设备还可以包括用户接口、网络接口、摄像头、射频(Radio Frequency,RF)电路,传感器、音频电路、WI-FI模块等等。用户接口可以包括显示屏(Display)、输入单元比如键盘(Keyboard)等,可选用户接口还可以包括USB接口、读卡器接口等。网络接口可选的可以包括标准的有线接口、无线接口(如蓝牙接口、WI-FI接口)等。
存储器中还可以包括操作系统、网络通信模块。操作系统是管理计算机设备硬件和软件资源的程序,支持信息处理程序以及其它软件和/或程序的运行。网络通信模块用于实现存储器内部各组件之间的通信,以及与该实体设备中其它硬件和软件之间通信。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到本申请可以借助软件加必要的通用硬件平台的方式来实现,也可以通过硬件实现通过应用本实施例的技术方案。

Claims (20)

  1. 一种用户群的优化方法,其特征在于,包括:
    获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
    若在所述目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
    根据所述新的目标用户群集合,确定最终的用户群。
  2. 根据权利要求1所述的方法,其特征在于,所述根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,具体包括:
    根据所述第一用户群对应的分群规则数量,确定第一目标用户群,其中,所述第一目标用户群为所述目标用户群集合中所述分群规则数量最多的所述第一用户群;
    根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,其中,所述第二目标用户群对应的节点为所述第一目标用户群对应的节点的父节点;
    将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合;
    若所述新的目标用户群集合中所述目标质量用户的数量小于所述预设覆盖阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至所述新的目标用户群集合中所述目标指令用户的数量大于或等于所述预设覆盖阈值,其中,重新确定的所述第一目标用户群为与原有的所述第一目标用户群不同的所述分群规则数量最多的所述第一用户群。
  3. 根据权利要求2所述的方法,其特征在于,所述第一用户群和所述第二用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则;
    所述根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,具体包括:
    将所述第一目标用户群对应的分群规则确定为第一分群规则,并将所述第一分群规则的最后一条删除得到第二分群规则;
    根据所述第二分群规则,在所述用户群总集合中搜索与所述第二分群规则对应的第二目标用户群。
  4. 根据权利要求3所述的方法,其特征在于,所述将所述第一目标用户群对应的分群规则确定为第一分群规则,具体包括:
    若所述第一目标用户群中所述目标质量用户占比大于或等于预设占比阈值,则将所述第一目标用户群对应的分群规则确定为所述第一分群规则,其中,所述第一目标用户群中所述目标质量用户占比为所述第一目标用户群中所述目标质量用户的数量与所述第一目标用户群中全部用户的数量的比值;
    若所述第一目标用户群中所述目标质量用户占比小于所述预设占比阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至重新确定的所述第一目标用户群中所述目标质量用户占比大于或等于预设占比阈值后,将重新确定的所述第一目标用户群对应的分群规则确定为所述第一分群规则。
  5. 根据权利要求3所述的方法,其特征在于,所述将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合之后,所述方法还包括:
    查询所述新的目标用户群集合中任意一个第一用户群对应的分群规则,并在其他第一用户群对应的分群规则包括所述任意一个第一用户群对应的分群规则时,删除所述目标用户群集合中的所述其他第一用户群。
  6. 根据权利要求5所述的方法,其特征在于,所述删除所述目标用户群集合中的所述其他第一用户群之后,所述方法还包括:
    计算所述新的目标用户群集合包含的第一用户群的数量;
    若所述第一用户群的数量大于或等于预设用户群阈值,则计算所述新的目标用户群集合中所述目标质量用户的数量。
  7. 根据权利要求3所述的方法,其特征在于,所述根据所述第二分群规则,在所述用户群总集合中搜索与所述第二分群规则对应的第二目标用户群,具体包括:
    在所述用户群总集合中搜索与所述第二分群规则对应的所述第二用户群,若与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于所述预设占比阈值,则将与所述第二分群规则对应的所述第二用户群确定为所述第二目标用户群,其中,所述第二目标用户群中的所述目标质量用户占比为所述第二目标用户群中所述目标质量用户的数量与所述第二目标用户群中全部用户的数量的比值;
    若与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于 所述预设占比阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于所述预设占比阈值。
  8. 一种用户群的优化装置,其特征在于,包括:
    用户群集合建立模块,用于获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
    用户群优化模块,用于若在所述目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
    用户群确定模块,用于根据所述新的目标用户群集合,确定最终的用户群。
  9. 根据权利要求8所述的装置,其特征在于,所述用户群优化模块,具体包括:
    第一目标用户群确定单元,用于根据所述第一用户群对应的分群规则数量,确定第一目标用户群,其中,所述第一目标用户群为所述目标用户群集合中所述分群规则数量最多的所述第一用户群;
    第二目标用户群确定单元,用于根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,其中,所述第二目标用户群对应的节点为所述第一目标用户群对应的节点的父节点;
    替换单元,用于将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合;
    第一目标用户群确定单元,还用于若所述新的目标用户群集合中所述目标质量用户的数量小于所述预设覆盖阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至所述新的目标用户群集合中所述目标指令用户的数量大于或等于所述预设覆盖阈值,其中,重新确定的所述第一目标用户群为与原有的所述第一目标用户群不同的所述分群规则数量最多的所述第一用户群。
  10. 根据权利要求9所述的装置,其特征在于,所述第一用户群和所述第二用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则;
    所述第二目标用户群确定单元,具体包括:
    分群规则确定子单元,用于将所述第一目标用户群对应的分群规则确定为第一分群规则,并将所述第一分群规则的最后一条删除得到第二分群规则;
    第二目标用户群确定子单元,用于根据所述第二分群规则,在所述用户群总集合中搜索与所述第二分群规则对应的第二目标用户群。
  11. 根据权利要求10所述的装置,其特征在于,所述第二目标用户群确定子单元,具体用于:
    若所述第一目标用户群中所述目标质量用户占比大于或等于预设占比阈值,则将所述第一目标用户群对应的分群规则确定为所述第一分群规则,其中,所述第一目标用户群中所述目标质量用户占比为所述第一目标用户群中所述目标质量用户的数量与所述第一目标用户群中全部用户的数量的比值;
    若所述第一目标用户群中所述目标质量用户占比小于所述预设占比阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至重新确定的所述第一目标用户群中所述目标质量用户占比大于或等于预设占比阈值后,将重新确定的所述第一目标用户群对应的分群规则确定为所述第一分群规则。
  12. 根据权利要求10所述的装置,其特征在于,所述用户群优化模块,还包括:
    删除单元,用于所述将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合之后,查询所述新的目标用户群集合中任意一个第一用户群对应的分群规则,并在其他第一用户群对应的分群规则包括所述任意一个第一用户群对应的分群规则时,删除所述目标用户群集合中的所述其他第一用户群。
  13. 根据权利要求12所述的装置,其特征在于,所述用户群优化模块,还包括:
    用户群数量计算单元,用于删除所述目标用户群集合中的所述其他第一用户群之后,计算所述新的目标用户群集合包含的第一用户群的数量;
    目标质量用户数量计算单元,用于若所述第一用户群的数量大于或等于预设用户群阈值,则计算所述新的目标用户群集合中所述目标质量用户的数量。
  14. 根据权利要求10所述的装置,其特征在于,所述第二目标用户群确定子单元,具体用于,具体用于:
    在所述用户群总集合中搜索与所述第二分群规则对应的所述第二用户群,若与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于所述预设占比阈值,则将与所述第二分群规则对应的所述第二用户群确定为所述第二目标用户群,其中,所述第二目标用户群中的所述目标质量用户占比为所述第二目标用户群中所述目标质量用户的数量与所述第二目标用户群中全部用户的数量的比值;
    若与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于所述预设占比阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直 至与所述第二分群规则对应的所述第二用户群中所述目标质量用户占比大于或等于所述预设占比阈值。
  15. 一种计算机非易失性可读存储介质,其上存储有计算机可读指令,其特征在于,所述计算机可读指令被处理器执行时实现用户群的优化方法,包括:
    获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
    若在所述目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
    根据所述新的目标用户群集合,确定最终的用户群。
  16. 根据权利要求15所述的计算机非易失性可读存储介质,其特征在于,所述计算机可读指令被处理器执行时实现所述根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,具体包括:
    根据所述第一用户群对应的分群规则数量,确定第一目标用户群,其中,所述第一目标用户群为所述目标用户群集合中所述分群规则数量最多的所述第一用户群;
    根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,其中,所述第二目标用户群对应的节点为所述第一目标用户群对应的节点的父节点;
    将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合;
    若所述新的目标用户群集合中所述目标质量用户的数量小于所述预设覆盖阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至所述新的目标用户群集合中所述目标指令用户的数量大于或等于所述预设覆盖阈值,其中,重新确定的所述第一目标用户群为与原有的所述第一目标用户群不同的所述分群规则数量最多的所述第一用户群。
  17. 根据权利要求16所述的计算机非易失性可读存储介质,其特征在于,所述第一用户群和所述第二用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则;
    所述计算机可读指令被处理器执行时实现所述根据所述第一目标用户群,在所述用户 群总集合中搜索第二目标用户群,具体包括:
    将所述第一目标用户群对应的分群规则确定为第一分群规则,并将所述第一分群规则的最后一条删除得到第二分群规则;
    根据所述第二分群规则,在所述用户群总集合中搜索与所述第二分群规则对应的第二目标用户群。
  18. 一种计算机设备,包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机可读指令,其特征在于,所述处理器执行所述计算机可读指令时实现用户群的优化方法,包括:
    获取决策树模型分群得到的目标用户群集合和用户群总集合,其中,所述目标用户群集合包括与所述决策树模型的每个叶子节点分别对应的第一用户群,所述用户群总集合包括与所述决策树模型的每个节点分别对应的第二用户群;
    若在所述目标用户群集合中目标质量用户的数量小于预设覆盖阈值,则根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,其中,所述新的目标用户群集合中所述目标质量用户的数量大于或等于所述预设覆盖阈值;
    根据所述新的目标用户群集合,确定最终的用户群。
    根据权利要求18所述的计算机设备,其特征在于,所述处理器执行所述计算机可读指令时实现所述根据所述第一用户群对应的分群规则数量和所述用户群总集合中与所述第一用户群节点的父节点对应的所述第二用户群,生成新的目标用户群集合,具体包括:
    根据所述第一用户群对应的分群规则数量,确定第一目标用户群,其中,所述第一目标用户群为所述目标用户群集合中所述分群规则数量最多的所述第一用户群;
    根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,其中,所述第二目标用户群对应的节点为所述第一目标用户群对应的节点的父节点;
    将所述目标用户群集合中的所述第一目标用户群替换为所述第二目标用户群,生成新的目标用户群集合;
  19. 若所述新的目标用户群集合中所述目标质量用户的数量小于所述预设覆盖阈值,则返回所述确定第一目标用户群的步骤重新确定第一目标用户群,直至所述新的目标用户群集合中所述目标指令用户的数量大于或等于所述预设覆盖阈值,其中,重新确定的所述第一目标用户群为与原有的所述第一目标用户群不同的所述分群规则数量最多的所述第一用户群。
  20. 根据权利要求19所述的计算机设备,其特征在于,所述第一用户群和所述第二 用户群中的任意一个用户群对应有一条分群规则或按顺序排列的多条分群规则;
    所述处理器执行所述计算机可读指令时实现所述根据所述第一目标用户群,在所述用户群总集合中搜索第二目标用户群,具体包括:
    将所述第一目标用户群对应的分群规则确定为第一分群规则,并将所述第一分群规则的最后一条删除得到第二分群规则;
    根据所述第二分群规则,在所述用户群总集合中搜索与所述第二分群规则对应的第二目标用户群。
PCT/CN2019/097890 2019-02-28 2019-07-26 用户群的优化方法及装置、计算机非易失性可读存储介质 WO2020173043A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910152473.6A CN109992699B (zh) 2019-02-28 2019-02-28 用户群的优化方法及装置、存储介质、计算机设备
CN201910152473.6 2019-02-28

Publications (1)

Publication Number Publication Date
WO2020173043A1 true WO2020173043A1 (zh) 2020-09-03

Family

ID=67129955

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/097890 WO2020173043A1 (zh) 2019-02-28 2019-07-26 用户群的优化方法及装置、计算机非易失性可读存储介质

Country Status (2)

Country Link
CN (1) CN109992699B (zh)
WO (1) WO2020173043A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112016979A (zh) * 2020-09-08 2020-12-01 平安科技(深圳)有限公司 用户分群方法、装置、设备和计算机可读存储介质

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109992699B (zh) * 2019-02-28 2023-08-11 平安科技(深圳)有限公司 用户群的优化方法及装置、存储介质、计算机设备
CN110929752B (zh) * 2019-10-18 2023-06-20 平安科技(深圳)有限公司 基于知识驱动和数据驱动的分群方法及相关设备
CN111105266B (zh) * 2019-11-11 2023-10-27 建信金融科技有限责任公司 基于改进决策树的客户分群方法及装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103729478A (zh) * 2014-01-26 2014-04-16 重庆邮电大学 基于MapReduce的LBS兴趣点发现方法
CN105808698A (zh) * 2016-03-03 2016-07-27 江苏大学 一种面向物联网用户查询请求的TOP-k位置兴趣点推荐方法
CN109992699A (zh) * 2019-02-28 2019-07-09 平安科技(深圳)有限公司 用户群的优化方法及装置、存储介质、计算机设备

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105245265B (zh) * 2014-06-20 2021-01-01 中兴通讯股份有限公司 3d波束下的用户服务模式选择方法、导频信息发送及装置
CN108256894B (zh) * 2017-02-07 2022-03-29 阿里巴巴(中国)有限公司 广告展示控制方法、设备、服务器及客户端
CN108810060B (zh) * 2017-05-05 2022-03-25 阿里巴巴集团控股有限公司 业务信息的提供、接收、用户聚合方法、服务器及客户端
CN109101562B (zh) * 2018-07-13 2023-07-21 中国平安人寿保险股份有限公司 寻找目标群体的方法、装置、计算机设备及存储介质
CN109242012A (zh) * 2018-08-27 2019-01-18 平安科技(深圳)有限公司 分组归纳方法及装置、电子装置及计算机可读存储介质
CN109271592B (zh) * 2018-09-30 2021-02-23 东软集团股份有限公司 一种物品推荐方法、装置、存储介质及程序产品

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103729478A (zh) * 2014-01-26 2014-04-16 重庆邮电大学 基于MapReduce的LBS兴趣点发现方法
CN105808698A (zh) * 2016-03-03 2016-07-27 江苏大学 一种面向物联网用户查询请求的TOP-k位置兴趣点推荐方法
CN109992699A (zh) * 2019-02-28 2019-07-09 平安科技(深圳)有限公司 用户群的优化方法及装置、存储介质、计算机设备

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHAO, HUA; SONG, SHUNLIN: "Application of Improved Decision Tree Algorithm in Obtaining Potential Customers", COMPUTER ENGINEERING AND APPLICATIONS, no. 11, 11 April 2005 (2005-04-11), CN, pages 196 - 198, XP009522949 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112016979A (zh) * 2020-09-08 2020-12-01 平安科技(深圳)有限公司 用户分群方法、装置、设备和计算机可读存储介质
CN112016979B (zh) * 2020-09-08 2023-07-18 平安科技(深圳)有限公司 用户分群方法、装置、设备和计算机可读存储介质

Also Published As

Publication number Publication date
CN109992699A (zh) 2019-07-09
CN109992699B (zh) 2023-08-11

Similar Documents

Publication Publication Date Title
WO2020173043A1 (zh) 用户群的优化方法及装置、计算机非易失性可读存储介质
US10410128B2 (en) Method, device, and server for friend recommendation
US9633311B2 (en) Decision tree learning
JP5755822B1 (ja) 類似度算出システム、類似度算出方法およびプログラム
US11200444B2 (en) Presentation object determining method and apparatus based on image content, medium, and device
KR102654884B1 (ko) 대화형 정보 제공 서비스 방법 및 이를 위한 장치
WO2019134274A1 (zh) 兴趣探索方法、存储介质、电子设备及系统
WO2019120023A1 (zh) 性别预测方法、装置、存储介质及电子设备
WO2020155921A1 (zh) 以图搜图的方法、计算机可读存储介质及服务器
WO2022179384A1 (zh) 一种社交群体的划分方法、划分系统及相关装置
WO2012149789A1 (zh) Vpc迁移中的资源匹配方法及装置
US20180300413A1 (en) Method, device, server and storage medium of searching a group based on social network
CN105574030B (zh) 一种信息搜索方法及装置
US9582586B2 (en) Massive rule-based classification engine
US9026643B2 (en) Contents' relationship visualizing apparatus, contents' relationship visualizing method and its program
US20210035025A1 (en) Systems and methods for optimizing machine learning models by summarizing list characteristics based on multi-dimensional feature vectors
CN109582967B (zh) 舆情摘要提取方法、装置、设备及计算机可读存储介质
KR102197739B1 (ko) 연기자 정보 제공 방법
TW201531866A (zh) 主動偏好學習方法與系統
WO2024160255A1 (zh) 页面交互方法、装置、设备及存储介质
JPWO2014020834A1 (ja) 単語潜在トピック推定装置および単語潜在トピック推定方法
CN110580304A (zh) 数据融合方法、装置、计算机设备及计算机存储介质
WO2018064959A1 (zh) 确定多媒体资源的标题的方法及装置
KR101568800B1 (ko) 실시간 이슈 검색어 선별 방법 및 시스템
JP2014174781A (ja) アイテム推薦システム、アイテム推薦方法およびアイテム推薦プログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19916780

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19916780

Country of ref document: EP

Kind code of ref document: A1