Marketing recommendation method and system

Abstract

The invention relates to a marketing recommendation method and system, and the system extracts communication feature data and position feature data from the collected multi-domain data of an operator, carries out the fusion processing of the multi-domain data according to the communication feature data and the position feature data, and carries out the marketing recommendation. According to a modeling theory, performing theme domain model construction taking the family customer as a core on the fused multi-domain data, according to a preset business rule and the theme domain model, constructing an emotional family taking the family customer and the communication circle as double cores, performing physical division on the emotional family from a time dimension and a space dimension to obtain a physical family, and storing the physical family in a database; doorplate address labeling is performed on the physical families, full-link demand preference recognition including family demand preference, building demand preference and community demand preference is performed on the labeled physical families, corresponding full-link demand preference is obtained, and marketing recommendation is performed. Therefore, effective marketing is realized, and the marketing effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a marketing recommendation method and system. Background Art

[0002] As the urbanization process continues to accelerate and deepen, residential communities have become extremely important living places, with numerous community buildings and a considerable population size, thus forming a huge market with great potential. At the same time, as the communications market becomes increasingly saturated, it is difficult for the number of users to achieve significant growth, and the home market has become a key focus of competition for operators. Faced with the current market situation, how operators can conduct effective communications marketing has become an important issue at present. Summary of the invention

[0003] The technical problem to be solved by the present invention is: the present invention provides a marketing recommendation method and system to achieve effective marketing for operators and improve marketing effects.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] In a first aspect, the present invention provides a marketing recommendation method, comprising:

[0006] Collecting multi-domain data of operators, extracting communication feature data and location feature data from the multi-domain data, and fusing the multi-domain data according to the communication circle feature data and the location feature data to obtain fused multi-domain data;

[0007] According to the modeling theory, a subject domain model with family customers as the core is constructed for the fused multi-domain data to obtain a subject domain model, a dual-core emotional family is constructed according to preset business rules and the subject domain model, and the emotional family is physically divided from the time dimension and the space dimension to obtain a physical family, wherein the dual cores are family customers and social circles;

[0008] The physical household is labeled with a house address to obtain a labeled physical household, full-link demand preference identification is performed on the labeled physical household to obtain a corresponding full-link demand preference, and marketing recommendations are made based on the full-link demand preference, wherein the full-link demand preference includes household demand preference, building demand preference and community demand preference.

[0009] The beneficial effects of the present invention are as follows: based on the multi-domain data of the operator, the comprehensiveness and extensiveness of the data are guaranteed; after the multi-domain data is integrated and processed according to the communication feature data and the location feature data, a subject domain model with family customers as the core is constructed, so as to improve the accuracy of the subject domain model while grasping the current market status with family customers as the competition; and the obtained physical family is obtained by physically dividing it from the time dimension and the space dimension on the basis of the emotional family with family customers and communication circles as the dual cores obtained according to the subject domain model and preset business rules, that is, the emotional family with family customers and communication circles as the core is further divided in time and space, so as to ensure the accuracy of the obtained physical family; and the full-link demand preference of the physical family is identified, so as to obtain the full-link demand preference including family demand preference, building demand preference and community demand preference, so as to comprehensively grasp the market demand that the current operator needs to grasp when making marketing recommendations, so that when the operator makes marketing recommendations according to the full-link demand preference, it can achieve effective marketing and improve the marketing effect.

[0010] Optionally, the collecting of multi-domain data of the operator includes:

[0011] The multi-domain data is preprocessed by using an ETL tool to obtain preprocessed multi-domain data, wherein the preprocessing includes data cleaning processing, formatting processing and standardization processing.

[0012] According to the above description, data cleaning, formatting and standardization are performed on multi-domain data to improve the accuracy of subsequent fusion processing.

[0013] Optionally, the multi-domain data includes business domain data, operation domain data and professional domain data, the operation domain data includes social circle data and location data, and extracting social feature data and location feature data from the multi-domain data includes:

[0014] Interaction feature data including interaction frequency features and interaction duration features are extracted from the interaction circle data, and location feature data including nighttime base station location overlap features and permanent base station features are extracted from the location data.

[0015] According to the above description, the interaction feature data includes interaction frequency features and interaction duration features, and the location feature data includes nighttime base station location overlap features and permanent base station features, ensuring the rationality and accuracy of the subsequent fusion processing based on the interaction feature data and location feature data.

[0016] Optionally, the modeling theory includes paradigm modeling and dimensional modeling, and the subject domain model with household customers as the core is constructed on the fused multi-domain data according to the modeling theory, and the obtained subject domain model includes:

[0017] Through the ER diagram, the fused multi-domain data is subjected to entity recognition to obtain the business ordering entity, location information entity, device connection entity and household customer entity;

[0018] Associating the business subscription attribute of the business subscription entity with the family customer attribute of the family customer entity to obtain a subscription class association table;

[0019] Associating the location attribute of the location information entity with the family customer attribute of the family customer entity to obtain a location class association table;

[0020] Associating the device attribute of the device connection entity with the home customer attribute of the home customer entity to obtain a communication class association table;

[0021] Normalizing the subscription class association table, the location class association table, and the communication class association table by using paradigm modeling to obtain a normalized subscription class association table, a normalized location class association table, and a normalized communication class association table;

[0022] Dimensional modeling is used to perform data modeling on the normalized ordering class association table, the normalized location class association table, and the normalized communication class association table to obtain the corresponding ordering class subject domain model, location class subject domain model, and communication class subject domain model.

[0023] According to the above description, after entity recognition of the fused multi-domain data is performed through the ER diagram, the corresponding ordering class association table, location class association table and communication class association table are established respectively. After normalization through paradigm modeling, data modeling is performed through dimensional modeling to improve the accuracy and standardization of the obtained ordering class subject domain model, location class subject domain model and communication class subject domain model.

[0024] Optionally, physically dividing the emotional family from the time dimension and the space dimension to obtain a physical family includes:

[0025] Acquire the base station data of each emotional member in the emotional family, divide the base station data into night base stations and holiday base stations from the time dimension, calculate the night overlap degree and the number of night overlap days of the night base stations, and calculate the holiday overlap degree and the number of holiday overlap days of the holiday base stations;

[0026] The emotional members whose nighttime overlap exceeds a first threshold and whose nighttime overlap days exceed a second threshold are divided into the same physical family, or the emotional members whose holiday overlap exceeds a third threshold and whose holiday overlap days exceed a fourth threshold are divided into the same physical family;

[0027] The night base station is divided into night permanent base stations from a spatial dimension, and it is determined whether the night permanent base stations are consistent. If they are consistent, the corresponding emotional members are divided into the same physical family. If they are inconsistent, the base station distance of the night permanent base station is calculated. When the base station distance does not exceed the distance threshold, the emotional members corresponding to the base station distance not exceeding the distance threshold are divided into the same physical family.

[0028] According to the above description, for the emotional family with family customers and social circles as the dual core, it can be further divided from the time dimension and space dimension. In the time dimension, the night base station and the holiday base station are used as the benchmark. In the space dimension, the night permanent base station and the base station distance of the night permanent base station are used as the benchmark for dividing the same physical family to ensure the rationality and accuracy of the obtained physical family.

[0029] Optionally, the acquiring base station data of each emotional member in the emotional family includes:

[0030] Determine whether there is a blank different network user in the base station data, and if so, select a local network service user who has a conversation with the different network user;

[0031] Acquire the call parking base station corresponding to the calling call in the call behavior of the service class user of the local network, and select the call parking base stations whose parking times in the same month exceed the parking threshold from the call parking base stations as the base station data of the user of the different network;

[0032] The service users of this website include takeaway users and express delivery users.

[0033] According to the above description, it can be known that for users of other networks, they are not filtered directly. Instead, service users of this network who have conversations with users of other networks are selected. Starting from the call parking base stations corresponding to the calling calls of service users of this network in their conversations, the call parking base stations whose parking times exceed the parking threshold in the same month are taken as the base station data of users of other networks. The data are comprehensively considered from the perspective of time and number of times to improve the accuracy and rationality of the base station data of users of other networks.

[0034] Optionally, the physical household is marked with a house number address, and the marked physical household includes:

[0035] Acquire address data of each physical member of the physical household, score the credibility of the address data according to the source type of the address data, obtain the address score of each address data, and mark the address data with the highest address score as the house number address of the physical household to obtain the marked physical household;

[0036] It is determined whether there is a physical household with the same house number address among the marked physical households. If so, the physical households with the same house number address are merged to obtain a merged physical household.

[0037] According to the above description, when marking house number addresses, the credibility of the address data is scored based on the source type of the address data, so that the address data with the highest address score is used as the house number address of the physical household, thereby improving the accuracy of the house number address. At the same time, physical households with the same house number address are merged to further ensure the integrity of the physical household.

[0038] Optionally, the acquiring the address data of each physical member of the physical household includes:

[0039] The address data is standardized using the public security house number as a standard to obtain standardized address data.

[0040] According to the above description, address data is standardized based on the public security house number plate, which ensures the integrity of the obtained standardized address data.

[0041] Optionally, the full-link demand preference identification is performed on the labeled physical household to obtain the corresponding full-link demand preference, including:

[0042] The labeled physical households are used to construct household demand portraits based on house numbers. At the same time, the labeled physical households are recursively constructed into building demand portraits and community demand portraits based on house numbers.

[0043] In the family demand portrait, a family demand structure is drawn according to the communication characteristics, Internet access characteristics, broadband usage characteristics and TV usage characteristics of each physical member in the marked physical family, and corresponding family demand preferences are obtained according to the family demand structure;

[0044] In the building demand portrait, a building demand structure is drawn according to the commonalities of the household demand structure, and corresponding building demand preferences are obtained according to the building demand structure;

[0045] In the community demand portrait, a community demand structure is drawn according to the building demand structure and the family demand structure, and the community demand preference is obtained according to the community demand structure.

[0046] According to the above description, when performing full-link demand preference identification, the family demand portrait is constructed based on the house number address of the physical family, and the family demand structure is drawn according to the physical members and the communication characteristics, Internet characteristics, broadband usage characteristics and TV usage characteristics to obtain the family demand preferences, and the accuracy of the obtained family demand preferences is ensured. The building demand portrait and community demand portrait are recursively constructed upward according to the house number address, and the building demand structure is drawn according to the commonality of the family demand structure to obtain the corresponding building demand preferences, and the rationality and accuracy of the obtained building demand preferences are ensured. The community demand structure is drawn according to the building demand structure and the family demand structure to obtain the corresponding community demand preferences, and the comprehensiveness and accuracy of the obtained community demand preferences are ensured.

[0047] In a second aspect, the present invention provides a marketing recommendation system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a marketing recommendation method described in the first aspect when executing the computer program.

[0048] Among them, the technical effect corresponding to the marketing recommendation system provided by the second aspect refers to the relevant description of the marketing recommendation method provided by the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A flowchart of a marketing recommendation method provided in this embodiment;

[0050] Figure 2 A schematic diagram of the overall process of a marketing recommendation method provided in this embodiment;

[0051] Figure 3 A schematic diagram of the structure of a marketing recommendation system provided in this embodiment.

[0052] [Description of Reference Numerals]

[0053] 1. A marketing recommendation system;

[0054] 2. Processor;

[0055] 3. Memory. DETAILED DESCRIPTION

[0056] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0057] Embodiment 1

[0058] Please refer to Figure 1 to Figure 2 The present invention provides a marketing recommendation method, comprising the steps of:

[0059] S1. Collect multi-domain data of operators, extract communication feature data and location feature data from the multi-domain data, and fuse the multi-domain data according to the communication feature data and the location feature data to obtain fused multi-domain data;

[0060] In this embodiment, if Figure 2 As shown, multi-domain data of operators are collected, and the multi-domain data includes business domain data, operation domain data and professional domain data. Business domain data refers to business subscription relationship type data that reflects the consumption mode of household customers, such as: household business subscription data, smart networking business data, group broadband business data, shared resource business data, etc. Operation domain data refers to information data that reflects the social circle and location of household customers, such as: resident community, base station information, broadband installation address, etc. Professional domain data refers to data that reflects the connection status of household customer equipment, such as: connection information of smart home equipment, family information data registered in the community, home security data, etc. Communication feature data and location feature data are extracted from the collected multi-domain data, and the multi-domain data are fused according to the communication feature data and the location feature data to obtain the fused multi-domain data. In the process of fusion processing, the purpose is to fuse the communication feature data and the location feature data to construct multi-domain data that reflects the characteristics of family relationships.

[0061] At this time, the multi-domain data collected from the operator in step S1 includes:

[0062] S11 preprocesses the multi-domain data by using an ETL tool to obtain preprocessed multi-domain data, wherein the preprocessing includes data cleaning, formatting and standardization.

[0063] In this embodiment, if Figure 2 As shown, the multi-domain data is cleaned, formatted and standardized. During the data cleaning process, duplicate data, erroneous data and invalid data are deleted. During the formatting process, the multi-domain data is formatted into a formatted type data suitable for the later construction of the subject domain model, that is, suitable for the deep learning model. During the standardization process, the non-numeric data is encoded, such as converting text data into numeric data.

[0064] At this time, the multi-domain data in step S1 includes business domain data, operation domain data and professional domain data, the operation domain data includes social circle data and location data, and the extraction of social feature data and location feature data from the multi-domain data includes:

[0065] S12, extracting communication feature data including communication frequency features and communication duration features from the communication circle data, and extracting location feature data including nighttime base station location overlap features and permanent base station features from the location data.

[0066] In this embodiment, the operation data of the multi-domain data includes social circle data and location data. Therefore, the interaction feature data including the interaction frequency feature and the interaction duration feature are extracted from the social circle data, and the location feature data including the night base station location overlap feature and the permanent base station feature are extracted from the location data.

[0067] S2. According to the modeling theory, a subject domain model with family customers as the core is constructed for the fused multi-domain data to obtain a subject domain model, and a dual-core emotional family is constructed according to preset business rules and the subject domain model, and the emotional family is physically divided from the time dimension and the space dimension to obtain a physical family, wherein the dual cores are family customers and social circles;

[0068] In this embodiment, if Figure 2As shown in the figure, according to the modeling theory, a subject domain model with household customers as the core is constructed for the fused multi-domain data. In order to improve the accuracy of the subject domain model construction, the fused multi-domain data is first subject-classified before the subject domain model is constructed. The subjects include: participant subject, service subject, event subject, marketing subject, account subject, resource subject, marketing subject, and financial subject. Among them, participant subject: mainly describes the relevant information of the roles played by various participants, including individuals, groups and organizations in the operator's business activities, including information categories such as customers, competitors, and partners; service subject: mainly describes the main business and products provided by the operator to the customers, as well as the customer's selection and customization of products; event subject: mainly describes the event records generated by the participants in the process of participating in and using various businesses in the operator's field, such as various lists, logs, orders and customer interaction records; marketing Topic: mainly describes the marketing, promotion and other plans and activities carried out by operators for specific market environments and customer groups; Account topic: mainly describes the accounts, bills and payment records generated by users' use of services; Resource topic: mainly describes all carriers owned by operators for providing services to customers, including service resources, network resources, regional resources and grids; Financial topic: mainly describes the cost information of various expenditures of operators in providing various services to customers and in daily operations. After completing the topic classification, a topic domain model with family customers as the core is constructed. According to the obtained topic domain model and preset business rules, an emotional family with family customers and social circles as the dual core is constructed. Then, the emotional family is physically divided from the time dimension and space dimension to obtain a physical family. The preset business rules include family account rules and social rules. Specifically, the family account rules are as follows:

[0069] 1. One device with two cards and one ID with multiple numbers, all the numbers included are included in the same emotional family, one of which refers to the ID card;

[0070] 2. The numbers included in the same family resource sharing, family service subscription and group broadband or smart networking family will be included in the same emotional family. The smart networking family refers to using the same broadband and using it for 10 days or more per month.

[0071] Rules of engagement:

[0072] Based on the call data, if there is a member who has a communication relationship with no less than one-third of the emotional members in the same emotional family and the communication relationship meets the communication requirements, he or she will be included in the emotional family, where the communication requirements are: the call frequency in the month is not less than the first threshold or the SMS interaction volume in the month is not less than the second threshold.

[0073] The above interaction rules and family rules have an or relationship. After constructing an emotional family with family customers and interaction circles as dual cores according to the subject domain model and preset business rules, the emotional family will be expanded through the connection information of smart home devices, family information data registered in the community, home security data, etc.

[0074] At this time, in step S2, the fused multi-domain data is subjected to a subject domain model with household customers as the core according to the modeling theory, and the obtained subject domain model includes:

[0075] S21, performing entity recognition on the fused multi-domain data through an ER diagram to obtain a business ordering entity, a location information entity, a device connection entity, and a household customer entity;

[0076] S22, associating the service subscription attribute of the service subscription entity with the family customer attribute of the family customer entity to obtain a subscription class association table;

[0077] S23, associating the location attribute of the location information entity with the family customer attribute of the family customer entity to obtain a location class association table;

[0078] S24, associating the device attribute of the device connection entity with the home customer attribute of the home customer entity to obtain a communication class association table;

[0079] S25, normalizing the subscription class association table, the location class association table, and the communication class association table by using paradigm modeling to obtain a normalized subscription class association table, a normalized location class association table, and a normalized communication class association table;

[0080] S26. Use dimensional modeling to perform data modeling on the normalized ordering class association table, the normalized location class association table, and the normalized communication class association table to obtain corresponding ordering class subject domain model, location class subject domain model, and communication class subject domain model.

[0081] In this embodiment, if Figure 2As shown, entity types and attributes corresponding to each entity type are predefined, wherein the entity types include: business subscription entity, location information entity, device connection entity and family customer entity, wherein the business subscription entity is related to business domain data, the location information entity is related to operation domain data, the device connection entity is related to professional domain data, and the family customer entity is related to business domain data, professional domain data and operation domain data; the attributes corresponding to each entity type, the family customer attributes of the family customer entity: including basic information such as family number, family name, family member information, etc., used to determine the identity and member composition of the family, wherein the family number is a unique identifier; the business subscription attributes of the business subscription entity: business subscription number, business name, subscription status, subscription time, validity period and other business consumption details, wherein the business subscription number is a unique identifier; the location attributes of the location information entity: location number, location name, detailed address, associated family number, etc., used to accurately characterize the location characteristics of the family customer, wherein the location number is a unique identifier; the device attributes of the device connection entity: device number, device type, connection status, family customer number to which it belongs, device usage frequency, etc., to show the usage of smart home devices in the family, wherein the device number is a unique identifier.

[0082] Therefore, the fused multi-domain data can be identified through the ER diagram according to the pre-defined entity types and the attributes corresponding to the entity types, and the business subscription entity, location information entity, device connection entity and home customer entity can be obtained, and the business subscription number in the business subscription attribute of the business subscription entity is associated with the home number in the home customer attribute of the home customer entity to obtain a subscription class association table; the location number in the location attribute of the location information entity is associated with the home number in the home customer attribute of the home customer entity to obtain a location class association table; the device number in the device attribute of the device connection entity is associated with the home number in the home customer attribute of the home customer entity to obtain a communication class association table; paradigm modeling is used to The ordering class association table, location class association table and communication class association table are normalized to obtain the normalized ordering class association table, normalized location class association table and normalized communication class association table, and then dimensional modeling is used to perform data modeling on the normalized ordering class association table, normalized location class association table and normalized communication class association table. Dimensional modeling will reintegrate and reconstruct each association table to obtain the corresponding ordering class subject domain model, location class subject domain model and communication class subject domain model; after obtaining each subject domain model, integrity and consistency checks will be performed to ensure that all key data are included in the corresponding subject domain to avoid data conflicts and inconsistencies.

[0083] At this time, the emotional family is physically divided from the time dimension and the space dimension in step S2, and the physical family includes:

[0084] S27, obtaining the base station data of each emotional member in the emotional family, dividing the base station data into night base stations and holiday base stations from the time dimension, calculating the night overlap degree and the number of night overlap days of the night base stations, and calculating the holiday overlap degree and the number of holiday overlap days of the holiday base stations;

[0085] S28, classifying the emotional members whose nighttime overlap exceeds the first threshold and whose nighttime overlap days exceed the second threshold into the same physical family, or classifying the emotional members whose holiday overlap exceeds the third threshold and whose holiday overlap days exceed the fourth threshold into the same physical family;

[0086] In this embodiment, if Figure 2 As shown, the base station data of each emotional member in the emotional family is obtained, and the base station data is divided into night base stations and holiday base stations from the time dimension. The night time period of the night base station is: 22:00-06:00, and the holidays are mainly national statutory holidays, so as to calculate the night overlap degree and the number of night overlap days of the night base station, the holiday overlap degree and the number of holiday overlap days of the holiday base station, and divide the emotional members whose night overlap degree exceeds the first threshold and the number of night overlap days exceeds the second threshold into the same physical family, or divide the emotional members whose holiday overlap degree exceeds the third threshold and the number of holiday overlap days exceeds the fourth threshold into the same physical family, wherein the first threshold and the third threshold are both 0.6, the second threshold is 5 days, and the fourth threshold is 2 days, which can be adjusted according to actual conditions.

[0087] At this time, the step S25 of acquiring the base station data of each emotional member in the emotional family includes:

[0088] S251, determining whether there is a blank different network user in the base station data, and if so, selecting a local network service user who has a conversation with the different network user;

[0089] S252, obtaining the call parking base station corresponding to the calling call in the call behavior of the service class user of the local network, and selecting the call parking base stations whose parking times in the same month exceed the parking threshold from the call parking base stations as the base station data of the user of the different network;

[0090] The service users of this website include takeaway users and express delivery users.

[0091] In this embodiment, for users of other networks with blank base station data, they are not directly filtered and deleted. Instead, service users of this network who have conversations with users of other networks, that is, takeaway users and express users, are selected, and the call retention base stations corresponding to when the service users of this network are calling in the call behavior are obtained. The call retention base stations whose retention times in the same month exceed the retention threshold, the retention threshold is 3, that is, the call retention base stations with more than 3 retention times in the same month are used as the base station data of users of other networks.

[0092] S29. Divide the night base station into night permanent base stations from the spatial dimension, and determine whether the night permanent base stations are consistent. If they are consistent, the corresponding emotional members are divided into the same physical family. If they are inconsistent, calculate the base station distance of the night permanent base station. When the base station distance does not exceed the distance threshold, the emotional members whose base station distance does not exceed the distance threshold are divided into the same physical family.

[0093] In this embodiment, if Figure 2 As shown, the night base stations are divided into night permanent base stations. If the night permanent base stations are consistent, the corresponding emotional members are divided into the same physical family. If they are inconsistent, the base station distance of the night permanent base stations is calculated. If the base station distance does not exceed the distance threshold, the distance threshold is 200 meters, then the emotional members corresponding to the distance of no more than 200 meters are divided into the same physical family.

[0094] S3. Mark the house address of the physical household to obtain the marked physical household, identify the full-link demand preference of the marked physical household to obtain the corresponding full-link demand preference, and make marketing recommendations based on the full-link demand preference, wherein the full-link demand preference includes household demand preference, building demand preference and community demand preference.

[0095] At this time, the physical household is marked with the house number address in step S3, and the marked physical household includes:

[0096] S31, obtaining address data of each physical member of the physical family, scoring the credibility of the address data according to the source type of the address data, obtaining an address score for each address data, marking the address data with the highest address score as the house number address of the physical family, and obtaining a marked physical family;

[0097] In this embodiment, if Figure 2As shown, the address data of each family member in the physical family is obtained, and the credibility of the address data is scored according to the source type of the address data. The source types of address data include: broadband installation address, community registration address and network identity address. The credibility score of the broadband installation address is: 0.5, the credibility score of the community registration address is: 0.4, and the credibility score of the network identity address is 0.3, thereby obtaining the address score of each address data, and the address data with the highest address score is used as the house number address of the physical family, so as to mark the house number address, and obtain the marked physical family. If the address score is consistent, the address data with the same address score are all used as the house number address for house number address marking. If the same house number address corresponds to more than one physical family, the physical families with the same house number address are merged.

[0098] At this time, the step S31 of obtaining the address data of each physical member of the physical family includes:

[0099] S311, performing standardization processing on the address data based on the public security house number plate as the standard to obtain the standardized address data.

[0100] In this embodiment, the address data will be standardized based on the public security house number plate. During the standardization process, the address data is disassembled level by level according to the standard address format of the public security house number plate to obtain the disassembled address data; a standard address entry library of the provincial resident network community is collected in advance, specifically covering the province, city, district / county, street office / township or town, street / road / administrative village / residents' committee, house number, resident network community information, and a place name address change mapping relationship table is pre-constructed; the disassembled address data is matched level by level with the standard address entry library, and the place name address change mapping relationship table is simultaneously referenced during the level-by-level matching process. Noise words that appear during the level-by-level matching process are deleted or replaced, and the missing of one or several levels in the disassembled address data are automatically supplemented.

[0101] In a specific embodiment, taking Fujian Province as an example, a pre-built partial standard address entry library of a resident network cell in Fujian Province is shown in Table 1:

[0102] Table 1. Some standard address entries in the Fujian Province resident network community

[0103]

[0104] S32. Determine whether there is a physical household with the same house number address among the marked physical households. If so, merge the physical households with the same house number address to obtain a merged physical household.

[0105] In this embodiment, physical households with the same house address are merged to obtain a merged physical household.

[0106] At this time, the full-link demand preference identification is performed on the marked physical household in step S3, and the corresponding full-link demand preference is obtained, including:

[0107] S33, constructing a household demand portrait for the marked physical household based on the house number address, and recursively constructing a building demand portrait and a community demand portrait for the marked physical household based on the house number address;

[0108] S34, in the family demand portrait, drawing a family demand structure according to the communication characteristics, Internet access characteristics, broadband usage characteristics, and television usage characteristics of each physical member in the labeled physical family, and obtaining corresponding family demand preferences according to the family demand structure;

[0109] S35, drawing a building demand structure in the building demand portrait according to the commonalities of the household demand structure, and obtaining a corresponding building demand preference according to the building demand structure;

[0110] S36. Draw a community demand structure in the community demand portrait according to the building demand structure and the family demand structure, and obtain the community demand preference according to the community demand structure.

[0111] In this embodiment, if Figure 2As shown, the marked physical family is used as a unit to construct a family demand portrait, and the building address is recursively derived from the house address, and the building demand portrait is constructed according to the building address. The community address is recursively derived from the building address, and the community demand portrait is constructed according to the community address. In the family demand portrait, the family demand structure is drawn according to the communication characteristics, Internet characteristics, broadband usage characteristics and TV usage characteristics of each physical member. Specifically, the communication characteristics include: call frequency, call time distribution, communication contact type, etc.; Internet characteristics include: Internet time length, online applications used, and types of websites browsed, such as: educational websites, entertainment websites; broadband usage characteristics include: bandwidth requirements, broadband usage time periods, and the number of device connections; TV usage characteristics include: TV usage time, whether to use smart TV functions, and types of channels watched, such as: news, sports, entertainment, etc. The family demand structure includes but is not limited to: a family demand structure dominated by the elderly, a family demand structure dominated by children or teenagers, and a family demand structure dominated by multiple smart home devices. According to the family demand structure, the corresponding family demand preference can be obtained according to the preset family demand preference judgment standard. The building demand structure is drawn based on the commonality of the household demand structure at the same building address. The building demand structure includes but is not limited to: a building demand structure dominated by young office workers, a building demand structure dominated by tenants, and a building demand structure dominated by commercial and residential use. The corresponding building demand preferences can be obtained according to the building demand structure and the preset building demand preference judgment criteria. The community demand structure can be drawn based on all building demand structures and family demand structures. The corresponding community demand preferences can be obtained according to the community demand structure and the preset community demand preference judgment criteria. Among them, family demand preferences, building demand preferences, and community demand preferences include but are not limited to: health care demand preferences, gigabit broadband demand preferences, ITV TV demand preferences, terminal demand preferences, number card demand preferences, and cloud computer demand preferences. Corresponding marketing recommendations are made according to different demand preferences.

[0112] Embodiment 2

[0113] Please refer to Figure 3 The present invention provides a marketing recommendation system 1, comprising a memory 3, a processor 2, and a computer program stored in the memory 3 and executable on the processor 2, wherein the processor 2 implements the steps in the first embodiment when executing the computer program.

[0114] Since the system / device described in the above embodiments of the present invention is a system / device used to implement the method of the above embodiments of the present invention, a person skilled in the art can understand the specific structure and deformation of the system / device based on the method described in the above embodiments of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiments of the present invention belong to the scope of protection of the present invention.

[0115] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0116] The present invention is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0117] It should be noted that in the claims, any reference numerals placed between brackets shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In the claims enumerating several means, several of these means may be embodied by the same hardware. The use of the words first, second, third, etc., is for convenience of expression only and does not indicate any order. These words may be understood as part of the component name.

[0118] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0119] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments after knowing the basic creative concept. Therefore, the claims should be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present invention.

[0120] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention should also include these modifications and variations.

Claims

1. A marketing recommendation method, characterized in that: include: Collecting multi-domain data of operators, extracting communication feature data and location feature data from the multi-domain data, and fusing the multi-domain data according to the communication feature data and the location feature data to obtain fused multi-domain data; According to the modeling theory, a subject domain model with family customers as the core is constructed for the fused multi-domain data to obtain a subject domain model, a dual-core emotional family is constructed according to preset business rules and the subject domain model, and the emotional family is physically divided from the time dimension and the space dimension to obtain a physical family, wherein the dual cores are family customers and social circles; The physical household is labeled with a house address to obtain a labeled physical household, full-link demand preference identification is performed on the labeled physical household to obtain a corresponding full-link demand preference, and marketing recommendations are made based on the full-link demand preference, wherein the full-link demand preference includes household demand preference, building demand preference and community demand preference.

2. A marketing recommendation method as claimed in claim 1, characterized in that: The multi-domain data collected from operators includes: The multi-domain data is preprocessed by using an ETL tool to obtain preprocessed multi-domain data, wherein the preprocessing includes data cleaning processing, formatting processing and standardization processing.

3. A marketing recommendation method as claimed in claim 1, characterized in that: The multi-domain data includes business domain data, operation domain data and professional domain data, the operation domain data includes social circle data and location data, and the extraction of social feature data and location feature data from the multi-domain data includes: Interaction feature data including interaction frequency features and interaction duration features are extracted from the interaction circle data, and location feature data including nighttime base station location overlap features and permanent base station features are extracted from the location data.

4. A marketing recommendation method as claimed in claim 1, characterized in that: The modeling theory includes paradigm modeling and dimensional modeling. The subject domain model with household customers as the core is constructed for the fused multi-domain data according to the modeling theory, and the obtained subject domain model includes: Through the ER diagram, the fused multi-domain data is subjected to entity recognition to obtain the business ordering entity, location information entity, device connection entity and household customer entity; Associating the business subscription attribute of the business subscription entity with the family customer attribute of the family customer entity to obtain a subscription class association table; Associating the location attribute of the location information entity with the family customer attribute of the family customer entity to obtain a location class association table; Associating the device attribute of the device connection entity with the home customer attribute of the home customer entity to obtain a communication class association table; Normalizing the subscription class association table, the location class association table, and the communication class association table by using paradigm modeling to obtain a normalized subscription class association table, a normalized location class association table, and a normalized communication class association table; Dimensional modeling is used to perform data modeling on the normalized ordering class association table, the normalized location class association table, and the normalized communication class association table to obtain the corresponding ordering class subject domain model, location class subject domain model, and communication class subject domain model.

5. A marketing recommendation method as claimed in claim 1, characterized in that: The physical division of the emotional family from the time dimension and the space dimension to obtain the physical family includes: Acquire the base station data of each emotional member in the emotional family, divide the base station data into night base stations and holiday base stations from the time dimension, calculate the night overlap degree and the number of night overlap days of the night base stations, and calculate the holiday overlap degree and the number of holiday overlap days of the holiday base stations; The emotional members whose nighttime overlap exceeds a first threshold and whose nighttime overlap days exceed a second threshold are divided into the same physical family, or the emotional members whose holiday overlap exceeds a third threshold and whose holiday overlap days exceed a fourth threshold are divided into the same physical family; The night base station is divided into night permanent base stations from a spatial dimension, and it is determined whether the night permanent base stations are consistent. If they are consistent, the corresponding emotional members are divided into the same physical family. If they are inconsistent, the base station distance of the night permanent base station is calculated. When the base station distance does not exceed the distance threshold, the emotional members corresponding to the base station distance not exceeding the distance threshold are divided into the same physical family.

6. A marketing recommendation method as claimed in claim 5, characterized in that: The step of obtaining base station data of each emotional member in the emotional family includes: Determine whether there is a blank different network user in the base station data, and if so, select a local network service user who has a conversation with the different network user; Acquire the call parking base station corresponding to the calling call in the call behavior of the service class user of the local network, and select the call parking base stations whose parking times in the same month exceed the parking threshold from the call parking base stations as the base station data of the user of the different network; The service users of this website include takeaway users and express delivery users.

7. A marketing recommendation method as claimed in claim 1, characterized in that: The physical household is marked with a house number address, and the marked physical household includes: Acquire address data of each physical member of the physical household, score the credibility of the address data according to the source type of the address data, obtain the address score of each address data, and mark the address data with the highest address score as the house number address of the physical household to obtain the marked physical household; It is determined whether there is a physical household with the same house number address among the marked physical households. If so, the physical households with the same house number address are merged to obtain a merged physical household.

8. A marketing recommendation method as claimed in claim 7, characterized in that: The acquiring of address data of each physical member of the physical household comprises: The address data is standardized using the public security house number as a standard to obtain standardized address data.

9. A marketing recommendation method as claimed in claim 1, characterized in that: The full-link demand preference identification is performed on the labeled physical household to obtain the corresponding full-link demand preference, including: The labeled physical households are used to construct household demand portraits based on house numbers. At the same time, the labeled physical households are recursively constructed into building demand portraits and community demand portraits based on house numbers. In the family demand portrait, a family demand structure is drawn according to the communication characteristics, Internet access characteristics, broadband usage characteristics and TV usage characteristics of each physical member in the marked physical family, and corresponding family demand preferences are obtained according to the family demand structure; In the building demand portrait, a building demand structure is drawn according to the commonalities of the household demand structure, and corresponding building demand preferences are obtained according to the building demand structure; In the community demand portrait, a community demand structure is drawn according to the building demand structure and the family demand structure, and the community demand preference is obtained according to the community demand structure.

10. A marketing recommendation system, characterized in that: The method comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 9 when executing the computer program.

Images