CN116757779A - Recommendation method based on user portrait - Google Patents
Recommendation method based on user portrait Download PDFInfo
- Publication number
- CN116757779A CN116757779A CN202310748612.8A CN202310748612A CN116757779A CN 116757779 A CN116757779 A CN 116757779A CN 202310748612 A CN202310748612 A CN 202310748612A CN 116757779 A CN116757779 A CN 116757779A
- Authority
- CN
- China
- Prior art keywords
- data
- user
- analysis
- processing
- recommendation method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000012545 processing Methods 0.000 claims abstract description 122
- 238000004458 analytical method Methods 0.000 claims abstract description 108
- 238000004141 dimensional analysis Methods 0.000 claims abstract description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 238000005065 mining Methods 0.000 claims description 15
- 238000011263 association rule mining technique Methods 0.000 claims description 13
- 238000010276 construction Methods 0.000 claims description 12
- 238000003860 storage Methods 0.000 claims description 12
- 238000012216 screening Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000013079 data visualisation Methods 0.000 claims description 6
- 238000004140 cleaning Methods 0.000 claims description 5
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 abstract description 6
- 238000006243 chemical reaction Methods 0.000 abstract description 5
- 230000006399 behavior Effects 0.000 description 12
- 238000000605 extraction Methods 0.000 description 11
- 230000008569 process Effects 0.000 description 10
- 239000013598 vector Substances 0.000 description 8
- 230000002776 aggregation Effects 0.000 description 5
- 238000004220 aggregation Methods 0.000 description 5
- 230000000007 visual effect Effects 0.000 description 5
- 230000010354 integration Effects 0.000 description 4
- 238000009826 distribution Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 230000000386 athletic effect Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000007405 data analysis Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000013475 authorization Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000001502 supplementing effect Effects 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Data Mining & Analysis (AREA)
- Game Theory and Decision Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to a recommendation method based on user portraits, which belongs to the field of data classification, and comprises the following steps: acquiring user data; performing data processing on the user data to obtain processed data; constructing a data warehouse based on the processed data; performing multidimensional analysis on the processing data stored in the data warehouse, and establishing a multidimensional label model based on the processing data; inputting real-time data sent by a user into a multi-dimensional label model to determine the portrait label of the user; and recommending the target product to the user based on the portrait tag. According to the method, the multi-dimensional analysis is carried out on the user data, and a multi-dimensional label model is established, so that the user's interests, preferences, requirements and other information can be known in depth, and products meeting the requirements of the user can be recommended more accurately; meanwhile, portrait tag identification is performed by inputting user data in real time, so that behavior change of a user can be adapted in time, and accuracy of recommendation and purchase conversion rate of the user are improved effectively.
Description
Technical Field
The application relates to the field of data classification, in particular to a recommendation method based on user portraits.
Background
Online shopping refers to a manner of purchasing and transacting goods through an online platform such as the internet or a mobile network. When the consumer performs online shopping, the consumer can firstly log in the e-commerce website or APP, browse detailed information of the commodity, including information such as pictures, prices, specifications and after-sales services of the commodity, select the commodity required by the consumer, confirm information such as the quantity and amount of the commodity, and finally confirm order information to finish payment.
Current online shopping platforms typically set up merchandise recommendation algorithms for recommending items to users. The commodity recommendation algorithm mainly discovers interests and preferences of a user by analyzing historical browsing, clicking, purchasing and other behavior data of the user, and recommends commodities meeting the requirements of the user to the user. Because the demands and preferences of users are dynamically changed, the conventional commodity recommendation algorithm cannot timely identify and adapt to the behavior change of the users, so that commodity recommendation does not meet the demands of the users, and the applicant believes that the conventional commodity recommendation algorithm can cause low yield of commodity purchase of the users.
Disclosure of Invention
In order to effectively improve the commodity purchase success rate of users, the application provides a recommendation method based on user portraits.
The recommendation method based on the user portrait provided by the application adopts the following technical scheme:
a user portrayal-based recommendation method, comprising:
acquiring user data;
performing data processing on the user data to obtain processed data;
constructing a data warehouse based on the processing data;
performing multidimensional analysis on the processing data stored in the data warehouse, and establishing a multidimensional label model based on the processing data;
inputting real-time data sent by a user into the multi-dimensional label model, and determining the portrait label of the user;
and recommending the target product to the user based on the portrait tag.
By adopting the technical scheme, the multidimensional label model is established by carrying out multidimensional analysis on the user data, so that the information such as interests, preferences, demands and the like of the user can be known in depth, and products meeting the demands of the user can be recommended more accurately; meanwhile, portrait tag identification is performed by inputting user data in real time, so that behavior change of a user can be adapted in time, and accuracy of recommendation and purchase conversion rate of the user are improved effectively.
Optionally, the data processing of the user data to obtain processed data includes:
performing data cleaning on the user data to obtain first processing data;
aggregating the first processing data to obtain second processing data;
and carrying out streaming processing on the second processing data to obtain the processing data.
By adopting the technical scheme, the user data is subjected to data processing, so that the user data is more standard and is easier to process, and data support is better provided for subsequent multidimensional analysis and modeling.
Optionally, the constructing a data warehouse based on the processing data includes:
acquiring a construction rule of the data warehouse;
determining a storage type of the data warehouse based on the construction rule;
and acquiring target data conforming to the storage type from the processing data, and generating the data warehouse according to the target data.
By adopting the technical scheme, the storage type of the data warehouse is determined according to the construction rule, and the target data conforming to the storage type is acquired from the processed data, so that the accuracy and normalization of the data in the data warehouse are conveniently and effectively ensured.
Optionally, the multidimensional analysis of the processing data stored in the data warehouse includes:
determining analysis indexes of the processing data according to a preset analysis target;
extracting processing data conforming to the analysis index from the data warehouse according to the analysis index;
and taking the processed data as analysis data to carry out multidimensional analysis.
By adopting the technical scheme, the analysis index of the processing data is firstly determined, the processing data conforming to the analysis index is extracted from the data warehouse according to the analysis index, and finally the processing data is used as the analysis data for multidimensional analysis, so that the subsequent recommendation of products conforming to the requirements of the user is more accurately facilitated.
Optionally, the multi-dimensional analysis of the processing data as analysis data includes:
taking the processing data as analysis data, and carrying out association rule mining analysis on the analysis data;
performing data perspective analysis on the analysis data;
and carrying out data visualization analysis on the analysis data.
By adopting the technical scheme, potential rules and association relations contained in analysis data can be mined by carrying out association rule mining analysis on the analysis data, so that the interests and demands of users can be known in depth; the data perspective analysis facilitates in-depth knowledge of the behavior and preferences of the user in different scenarios; the data visualization analysis can present the analysis result in a visual mode, so that the analysis result is more visual and easy to understand.
Optionally, the performing association rule mining analysis on the analysis data includes:
performing association rule mining on the analysis data based on a preset association rule mining algorithm to obtain a plurality of frequent item sets and a plurality of association rules;
screening the frequent item sets and the association rules based on a preset support degree and a preset confidence degree to obtain a plurality of output frequent item sets and a plurality of output association rules;
and outputting a plurality of output frequent item sets and a plurality of output association rules to finish association rule mining analysis.
By adopting the technical scheme, the association rule mining is carried out on the analysis data, so that the association rule of the analysis data can be conveniently known, and further the interest and the demand of the user can be conveniently and deeply known.
Optionally, the establishing a multi-dimensional label model based on the processing data includes:
extracting the characteristics of the processed data to obtain a plurality of user characteristics;
generating a plurality of user tags based on a plurality of the user features; each of the user tags represents a user characteristic of a user;
and combining a plurality of user labels to obtain a multi-dimensional label model.
By adopting the technical scheme, the multi-dimensional label model is established to combine a plurality of user labels of the user, so that the interests and the demands of the user are more comprehensively and accurately described, the follow-up recommendation of products meeting the demands of the user is facilitated, and the commodity purchasing success rate of the user is effectively improved.
Optionally, the recommending the target product to the user based on the portrait tag includes:
obtaining product labels of all products;
matching calculation is carried out on the portrait label and the product label, so that interest scores of users on each product are obtained;
and determining target products in all products according to the interest scores, and recommending the target products to a user.
By adopting the technical scheme, the target product is determined through the interest score of the user on each product, so that the target product is recommended according to the user requirement, and the commodity purchasing success rate of the user is effectively improved.
In summary, the application has at least one of the following beneficial technical effects:
1. through multidimensional analysis on the user data, a multidimensional label model is established, so that the user's interests, preferences, demands and other information can be known in depth, and products meeting the demands can be recommended to the user more accurately; meanwhile, portrait tag identification is performed by inputting user data in real time, so that behavior change of a user can be adapted in time, and accuracy of recommendation and purchase conversion rate of the user are improved effectively.
2. Firstly, determining analysis indexes of the processing data, extracting the processing data which accords with the analysis indexes from a data warehouse according to the analysis indexes, and finally, taking the processing data as the analysis data to carry out multidimensional analysis, so that the subsequent recommendation of products which accord with the requirements of the user is more accurately facilitated.
3. By establishing a multidimensional label model, a plurality of user labels of the user are combined, so that interests and demands of the user are described more comprehensively and accurately, product recommendation meeting the demands of the user is conveniently provided for the user, and the commodity purchasing success rate of the user is effectively improved.
Drawings
FIG. 1 is a flow chart of a recommendation method based on user portraits according to an embodiment of the present application.
FIG. 2 is a flow chart of a recommendation method based on user portraits according to an embodiment of the application.
FIG. 3 is a flow chart illustrating a recommendation method based on user portraits according to an embodiment of the present application.
FIG. 4 is a flow chart of a recommendation method based on user portraits according to an embodiment of the application.
FIG. 5 is a flow chart of a recommendation method based on user portraits according to an embodiment of the present application.
FIG. 6 is a flow chart illustrating a recommendation method based on user portraits according to an embodiment of the present application.
FIG. 7 is a flow chart illustrating a recommendation method based on user portraits according to an embodiment of the present application.
FIG. 8 is a flow chart illustrating a recommendation method based on user portraits according to an embodiment of the present application.
Detailed Description
The application is described in further detail below with reference to fig. 1 to 8.
The embodiment of the application discloses a recommendation method based on user portraits.
Referring to fig. 1, a recommendation method based on a user portrait includes the steps of:
s101, acquiring user data.
In this embodiment, the user data is obtained through multiple data sources, and the data sources are not limited to social media, search engines, mobile applications, e-commerce platforms, and the like. The social media platform provides an API interface, and data of the user on the social media platform, such as personal information, posted content, comments and the like of the user, can be acquired through the API; the search engine provides a crawler tool, and behavior data of the user on the website, such as search keywords, clicking behaviors and the like of the user, can be obtained through the crawler tool; the mobile application can acquire the data of the user in a user authorization mode, such as the position information, the equipment information, the application use behavior and the like of the user; the e-commerce platform provides an API interface through which data of the user on the e-commerce platform, such as purchase records, browse records, ratings, etc., of the user can be obtained.
S102, performing data processing on the user data to obtain processed data.
Data processing refers to the process of sorting and analyzing and processing a range, amount, or type of data. In this embodiment, the data processing includes data cleansing, deduplication, missing value processing, and the like, and the user ensures the accuracy and integrity of the user data.
S103, constructing a data warehouse based on the processing data.
After the processing data is obtained, the processing data can be integrated to generate a unified data warehouse. Specifically, the data integration may use tools such as ETL tools, data integration platforms, and the like. Wherein the ETL tool is a software tool for automating the implementation of the data integration process.
A data warehouse is a system for storing, managing, and analyzing enterprise-level data that provides a unified, consistent view of data by integrating and transforming data from different data sources. A data warehouse typically includes four major components, a data source layer, a data integration layer, a data storage layer, and a data application layer.
In this embodiment, when a data warehouse is constructed, processing data is first loaded into the data warehouse, and the loading mode may be a batch loading or incremental loading mode; and finally, storing the loaded processing data into a data warehouse, wherein the data warehouse can adopt a relational database or a columnar database and the like.
S104, performing multidimensional analysis on the processing data stored in the data warehouse, and establishing a multidimensional label model based on the processing data.
Multidimensional analysis refers to multidimensional data processing on the processed data so as to facilitate mining of association and rules inside the processed data, thereby facilitating better understanding and analysis of characteristics and trends of the processed data by users. In multidimensional analysis, the processed data may be segmented and aggregated, such as into four categories, time, region, product, customer, etc., while the processed data may be calculated and compared, such as calculating averages, calculating growth rates, calculating duty cycles, etc.
In this embodiment, the processing data is first converted into numerical features that can be processed by a computer by using techniques such as text mining, image recognition, and speech recognition. Then, through algorithms such as machine learning or deep learning, the numerical characteristics are analyzed and modeled, and a multi-dimensional label model can be generated. The multidimensional label model is a vector model that converts processing data into a plurality of labels and combines the labels into one multidimensional. The multi-dimensional tag model is used to describe various characteristics of the user, such as interest preferences, consumption levels, geographic locations, social relationships, etc., so that the behavior and needs of the user can be more accurately described and predicted.
S105, inputting real-time data sent by the user into a multi-dimensional label model to determine the portrait label of the user.
After the multi-dimensional label model is generated, key features of real-time data sent by a user are firstly extracted, and the key features are input into the multi-dimensional label model, so that portrait labels of the user can be output.
S106, recommending the target product to the user based on the portrait tag.
The portrait tags of the users are known, namely target products can be recommended to the users according to the portrait tags, for example, the portrait tags comprise a plurality of dimensions of basic information, behavior preferences, consumption habits and the like of the users, and target products matched with the portrait tags can be found by analyzing the portrait tags of the users, so that the target products can be recommended to the users. For example, assuming that user A prefers to purchase athletic shoes, user A may be recommended to the user by analyzing user A's portrayal labels, such as favorite brands, colors, etc., to find the athletic shoes that match.
The implementation principle of the embodiment is as follows: through multidimensional analysis on the user data, a multidimensional label model is established, so that the user's interests, preferences, demands and other information can be known in depth, and products meeting the demands can be recommended to the user more accurately; meanwhile, portrait tag identification is performed by inputting user data in real time, so that behavior change of a user can be adapted in time, and accuracy of recommendation and purchase conversion rate of the user are improved effectively.
In one implementation manner of this embodiment, referring to fig. 2, data processing is performed on user data to obtain processed data, including the following steps:
s201, data cleaning is conducted on the user data to obtain first processing data.
Data cleaning refers to the process of preprocessing and correcting user data to solve the data quality problem and improve the data value. The data cleaning comprises the steps of data deduplication, data error correction, data formatting, data screening, data standardization, data complement and the like.
Wherein, the data deduplication refers to merging or deleting repeated data records; data error correction refers to detecting and correcting errors, deletions, anomalies and other problems in data, such as spelling errors, format errors, null values, outliers and the like; data formatting refers to formatting data to make the data consistent and standardized, such as date format, unit conversion, numerical precision, etc.; the data screening means screening and filtering the data and selecting the data meeting preset conditions; data standardization means standardization and normalization processing of data to meet preset data standards and specifications, such as data elements, data dictionaries and the like; data complementation refers to supplementing or estimating missing or incomplete information in data to ensure the integrity and continuity of the data.
S202, aggregating the first processing data to obtain second processing data.
Aggregation refers to the process of summarizing, counting and calculating data to obtain the overall characteristics and regularity of the data. In this embodiment, the implementation is implemented by using tools or languages such as SQL. Specifically, the data are aggregated, and first processing data are grouped according to preset attributes, such as grouping according to time, region, product category and the like; secondly, carrying out aggregation calculation on the data in each group, such as summation, average value, maximum value, minimum value and the like; filtering the calculation result, such as screening out ten products before sales, filtering out abnormal data, etc.; and finally, sequencing the aggregation result to finish the aggregation process.
S203, performing streaming processing on the second processing data to obtain processing data.
The streaming of the second processed data refers to a data processing mode of real-time generation, real-time processing, real-time analysis and real-time response of the data. Unlike batch processing, streaming processing does not need to wait for the data to be processed after all arrives, but rather can process the data batchwise or one by one to achieve real-time requirements. The Streaming processing can decompose the second processing data into a series of ordered data streams, and each data stream is processed in real time so as to discover the abnormality or change in the second processing data in time, and the Streaming processing technology can be Storm, spark Streaming, flink and the like.
It should be noted that, the data processing manners of the stream processing platform quasi-real-time data processing in steps S201 to S203 include, in addition to the data processing manners of the stream processing platform quasi-real-time data processing, a data processing manner of the client and server side real-time data processing and a data processing manner of the large data platform offline data processing, that is, the data processing of the user data is three data processing manners, and the three data processing manners are sequentially the client and server side real-time data processing, the stream processing platform quasi-real-time data processing and the large data platform offline data processing according to the real-time ranking. With the real-time property decreasing from strong to weak, the mass data processing capacity of the three modes is from weak to strong. The aggregation and processing of user data for multiple data sources is achieved through a variety of different data processing modes.
The first two steps of the 3 steps of the data processing modes of the real-time data processing of the client and the server are consistent with the step S201 and the step S202, and the third step is to analyze the aggregated data so as to extract the value data; the first two steps of the data processing modes of the offline data processing of the big data platform are consistent with the step S201 and the step S202, and the third step is to perform offline processing on the aggregated second processing data and realize the offline data processing through the distributed computing capacity of the big data platform. In offline processing, the processing and analysis of data is performed in an offline environment.
According to the recommendation method based on the user portraits, data processing is conducted on the user data, so that the user data are more standard and easier to process, and data support is better provided for subsequent multidimensional analysis and modeling.
In one implementation of this embodiment, referring to fig. 3, a data warehouse is constructed based on the processing data, including the steps of:
s301, acquiring a construction rule of a data warehouse.
The construction rule is preset for people, and the construction rule is set based on the use scene and the target of the data warehouse.
S302, determining the storage type of the data warehouse based on the construction rule.
The storage types comprise data sources, data types, data structures and data formats, wherein each construction rule is preset, and after the construction rule is known, the data sources, the data types, the data structures and the data formats which need to be acquired can be determined according to the use scene and the target of the data warehouse.
S303, acquiring target data conforming to the storage type from the processing data, and generating a data warehouse according to the target data.
And screening target data consistent with the storage type in the processed data, namely generating a data warehouse according to the target data.
According to the recommendation method based on the user portraits, which is provided by the embodiment, the storage type of the data warehouse is determined according to the construction rule, and the target data which accords with the storage type is acquired from the processed data, so that the accuracy and normalization of the data in the data warehouse are conveniently and effectively ensured.
In one implementation of this embodiment, referring to fig. 4, the multidimensional analysis of the process data stored in the data warehouse includes the steps of:
s401, determining analysis indexes of the processing data according to a preset analysis target.
The analysis target is preset, namely, the purpose of analyzing the data, such as sales increase, user activity improvement and the like, is the analysis target, and the analysis index refers to the index which needs to analyze the processed data, such as sales, order quantity, user access quantity and the like. The analysis index may be determined when the analysis target is determined, for example, when the analysis target is determined to be sales increase, the analysis index is sales, number of orders, user access, and the like.
S402, extracting processing data which accords with the analysis index from the data warehouse according to the analysis index.
When the analysis index is known, the processing data conforming to the analysis index can be screened from all the processing data stored in the data warehouse, for example, the analysis index is sales, order number, user access amount and the like, and the screened processing data is sales, order number, user access amount and the like.
S403, performing multidimensional analysis by taking the processed data as analysis data.
After the processed data is screened, multidimensional analysis is performed on the processed data.
According to the recommendation method based on the user portraits, firstly, the analysis index of the processing data is determined, the processing data which accords with the analysis index is extracted from the data warehouse according to the analysis index, and finally, the processing data is used as the analysis data to carry out multidimensional analysis, so that the subsequent recommendation of products which accord with the requirements of the user can be more accurately achieved.
In one implementation of this embodiment, referring to fig. 5, the multi-dimensional analysis is performed using the processing data as analysis data, including the steps of:
s501, processing data is used as analysis data, and association rule mining analysis is carried out on the analysis data.
Association rule mining is a data mining technique that can be used to mine frequent item sets and association rules in analytical data to discover potential associations in the data set. The association rule mining analysis can be performed on the analysis data through a preset association rule mining algorithm, such as an Apriori algorithm, an FP-Growth algorithm and the like.
S502, performing data perspective analysis on the analysis data.
Data perspective is a data analysis technique used to quickly aggregate, and present large amounts of data so that users can better understand the characteristics and rules of the data. The data perspective is usually performed in an interactive manner, and the data can be screened, ordered, grouped and summarized according to the requirement, and the data perspective tools which can be used in the embodiment include Excel, tableau, power BI and the like.
S503, performing data visualization analysis on the analysis data.
Visual analysis is a method of visually presenting data and presenting and analyzing the data in the form of charts, images, maps, and the like. The analysis data may be subjected to data visualization analysis by a visualization analysis tool, such as Tableau, power BI, excel, and the like.
According to the recommendation method based on the user portraits, through carrying out association rule mining analysis on the analysis data, potential rules and association relations contained in the analysis data can be mined, so that the interests and the demands of users can be known in depth; the data perspective analysis facilitates in-depth knowledge of the behavior and preferences of the user in different scenarios; the data visualization analysis can present the analysis result in a visual mode, so that the analysis result is more visual and easy to understand.
In one implementation manner of the present embodiment, referring to fig. 6, performing association rule mining analysis on analysis data includes the following steps:
s601, carrying out association rule mining on analysis data based on a preset association rule mining algorithm to obtain a plurality of frequent item sets and a plurality of association rules.
The association rule mining algorithm can be an Apriori algorithm or an FP-Growth algorithm, etc., wherein the Apriori algorithm is an algorithm based on candidate item sets, and the association rule is generated by iteratively searching frequent item sets; the FP-Growth algorithm is a data structure based algorithm that mines frequent item sets and generates association rules by building FP-trees. And carrying out association rule mining on the analysis data according to an association rule mining algorithm to obtain a plurality of frequent item sets and a plurality of association rules.
S602, screening the frequent item sets and the association rules based on preset support and preset confidence to obtain a plurality of output frequent item sets and a plurality of output association rules.
In association rule mining, a threshold value for the support degree and the confidence degree needs to be set. The support degree refers to the ratio of the number of transactions containing the association rule to the total number of transactions, and the confidence degree refers to the ratio of the number of transactions satisfying the precondition to the number of transactions satisfying the conclusion condition at the same time. The precondition refers to a hypothesis condition, namely an original hypothesis, which is made when statistical inference is performed. Confidence is the result of statistical inference on the sample data based on this original assumption. Conclusion conditions refer to events or conditions of interest in performing confidence calculations, and generally refer to questions that require inference or judgment, such as whether the mean value of a sample meets a certain requirement, etc.
If the support and confidence of the frequent item set or the association rule exceed the preset corresponding threshold values, the frequent item set or the association rule is considered to be reliable, so that a plurality of output frequent item sets and a plurality of output association rules are output. For example, if the support threshold is set to 0.5 and the confidence threshold is set to 0.7, only frequent item sets and association rules with support not less than 0.5 and confidence not less than 0.7 will be output. Frequent item sets refer to item sets that occur frequently in a dataset, and association rules refer to relationships between two or more items.
S603, outputting a plurality of output frequent item sets and a plurality of output association rules, and completing association rule mining analysis.
And outputting a plurality of output frequent item sets and a plurality of output association rules, namely indicating that association rule mining analysis is completed.
According to the recommendation method based on the user portraits, association rules are mined on the analysis data, so that the association rules of the analysis data can be conveniently known, and further the interests and the demands of the users can be conveniently and deeply known.
In one implementation of this embodiment, referring to fig. 7, a multi-dimensional label model is built based on processing data, including the steps of:
s701, extracting characteristics of the processed data to obtain a plurality of user characteristics.
Feature extraction is the process of converting raw data into features with interpretability, with the aim of better describing and analyzing the data. The feature extraction method can be statistical feature extraction method, frequency feature extraction method, time feature extraction method or text feature extraction method. The statistical feature extraction comprises mean, variance, standard deviation, maximum value, minimum value, median and the like, and is used for reflecting the distribution and trend of the data; the frequency feature extraction includes the frequency of occurrence, the frequency distribution, etc., for reflecting the occurrence law and probability distribution of the data. The time feature extraction includes time stamps, time intervals, periodicity, trending, etc. for reflecting the time features and trend of the data. Text feature extraction includes word frequency, keywords, text length, language features, etc., for reflecting the content and characteristics of the text data.
S702, generating a plurality of user labels based on a plurality of user characteristics; each user tag represents a user characteristic of the user.
Based on the result of feature extraction, i.e. a number of user features, a plurality of user tags are generated, each representing a user feature of the user, such as age, gender, interest preferences, etc.
S703, combining the plurality of user labels to obtain a multi-dimensional label model.
And combining the generated plurality of user labels into a multi-dimensional vector model, namely a multi-dimensional label model. In this embodiment, each user tag is regarded as a word, and a plurality of user tags are combined into a word bag, and the number of occurrences of each word in the word bag is represented by a vector, namely, a word bag model is provided, and the word bag model is a multidimensional tag model.
According to the recommendation method based on the user portrait, the multidimensional label model is established, and a plurality of user labels of the user are combined, so that interests and requirements of the user are described more comprehensively and accurately, subsequent product recommendation meeting the requirements of the user is conveniently provided for the user, and the commodity purchasing success rate of the user is effectively improved.
In one implementation manner of the present embodiment, referring to fig. 8, recommending a target product to a user based on an portrait tag includes the following steps:
s801, obtaining product labels of all products.
Product labels include type, brand, price, function, etc. of the product.
S802, matching calculation is carried out on the portrait label and the product label, and the interest score of the user on each product is obtained.
Matching calculation is carried out on the portrait label and the product label, and the interest score of the user on each product is obtained by the following steps:
1. encoding the portrait label and the product label of the user, for example, using One-Hot encoding or word bag model, and mapping each label into a vector. The portrait labels are mapped to portrait vectors, and the product labels are mapped to label vectors.
2. And for each product, calculating the similarity between the label vector and the user portrait vector, and obtaining the interest score of the user on the product label by adopting a cosine similarity method. The interest score is the calculated similarity.
3. And carrying out weighted average on the interest scores of the users on each product label to obtain the overall interest score of the users on the product.
S803, determining target products in all products according to the interest scores, and recommending the target products to the user.
And selecting a target product with the highest score from all products according to the interest scores of the users, generating a recommendation list, displaying the recommendation list through a recommendation system, and recommending the target product to the users by pushing the recommendation product in an APP home page or mail for example.
According to the recommendation method based on the user portrait, the target product is determined through the interest score of the user on each product, so that the target product can be recommended according to the user requirement, and the commodity purchasing success rate of the user is effectively improved.
The above embodiments are not intended to limit the scope of the present application, so: all equivalent changes in structure, shape and principle of the application should be covered in the scope of protection of the application.
Claims (8)
1. A user portrayal-based recommendation method, comprising:
acquiring user data;
performing data processing on the user data to obtain processed data;
constructing a data warehouse based on the processing data;
performing multidimensional analysis on the processing data stored in the data warehouse, and establishing a multidimensional label model based on the processing data;
inputting real-time data sent by a user into the multi-dimensional label model, and determining the portrait label of the user;
and recommending the target product to the user based on the portrait tag.
2. The recommendation method based on user portraits of claim 1, wherein the data processing of the user data to obtain processed data comprises:
performing data cleaning on the user data to obtain first processing data;
aggregating the first processing data to obtain second processing data;
and carrying out streaming processing on the second processing data to obtain the processing data.
3. The user portrayal-based recommendation method according to claim 1, wherein said constructing a data warehouse based on said processing data comprises:
acquiring a construction rule of the data warehouse;
determining a storage type of the data warehouse based on the construction rule;
and acquiring target data conforming to the storage type from the processing data, and generating the data warehouse according to the target data.
4. The user portrayal-based recommendation method according to claim 1, wherein said multi-dimensional analysis of said processed data stored in said data warehouse comprises:
determining analysis indexes of the processing data according to a preset analysis target;
extracting processing data conforming to the analysis index from the data warehouse according to the analysis index;
and taking the processed data as analysis data to carry out multidimensional analysis.
5. The user representation-based recommendation method according to claim 4, wherein the multi-dimensional analysis of the processing data as analysis data comprises:
taking the processing data as analysis data, and carrying out association rule mining analysis on the analysis data;
performing data perspective analysis on the analysis data;
and carrying out data visualization analysis on the analysis data.
6. The user profile-based recommendation method of claim 5, wherein said performing an association rule mining analysis on said analysis data comprises:
performing association rule mining on the analysis data based on a preset association rule mining algorithm to obtain a plurality of frequent item sets and a plurality of association rules;
screening the frequent item sets and the association rules based on a preset support degree and a preset confidence degree to obtain a plurality of output frequent item sets and a plurality of output association rules;
and outputting a plurality of output frequent item sets and a plurality of output association rules to finish association rule mining analysis.
7. The user portrayal-based recommendation method according to claim 4, wherein said creating a multi-dimensional label model based on said processing data comprises:
extracting the characteristics of the processed data to obtain a plurality of user characteristics;
generating a plurality of user tags based on a plurality of the user features; each of the user tags represents a user characteristic of a user;
and combining a plurality of user labels to obtain a multi-dimensional label model.
8. The recommendation method based on the portrait of the user according to claim 1, wherein the recommending the target product to the user based on the portrait tag includes:
obtaining product labels of all products;
matching calculation is carried out on the portrait label and the product label, so that interest scores of users on each product are obtained;
and determining target products in all products according to the interest scores, and recommending the target products to a user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310748612.8A CN116757779A (en) | 2023-06-21 | 2023-06-21 | Recommendation method based on user portrait |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310748612.8A CN116757779A (en) | 2023-06-21 | 2023-06-21 | Recommendation method based on user portrait |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116757779A true CN116757779A (en) | 2023-09-15 |
Family
ID=87947581
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310748612.8A Pending CN116757779A (en) | 2023-06-21 | 2023-06-21 | Recommendation method based on user portrait |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116757779A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117743848A (en) * | 2023-12-06 | 2024-03-22 | 暗物质(北京)智能科技有限公司 | User portrait generation method and device, electronic equipment and storage medium |
| CN117742223A (en) * | 2024-02-20 | 2024-03-22 | 深圳市凯度电器有限公司 | Control method and device of embedded remote water purification system |
-
2023
- 2023-06-21 CN CN202310748612.8A patent/CN116757779A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117743848A (en) * | 2023-12-06 | 2024-03-22 | 暗物质(北京)智能科技有限公司 | User portrait generation method and device, electronic equipment and storage medium |
| CN117742223A (en) * | 2024-02-20 | 2024-03-22 | 深圳市凯度电器有限公司 | Control method and device of embedded remote water purification system |
| CN117742223B (en) * | 2024-02-20 | 2024-04-26 | 深圳市凯度电器有限公司 | Control method and device of embedded remote water purification system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106844787B (en) | Recommendation method for searching target users and matching target products for automobile industry | |
| CN107220365B (en) | Accurate recommendation system and method based on collaborative filtering and association rule parallel processing | |
| US10095782B2 (en) | Summarization of short comments | |
| US20180158078A1 (en) | Computer device and method for predicting market demand of commodities | |
| CN111709812A (en) | E-commerce platform commodity recommendation method and system based on user dynamic classification | |
| US20170249389A1 (en) | Sentiment rating system and method | |
| CN116757779A (en) | Recommendation method based on user portrait | |
| CN108805598B (en) | Similarity information determination method, server and computer-readable storage medium | |
| CA2923600A1 (en) | Review sentiment analysis | |
| CN114372573B (en) | User portrait information recognition method and device, computer equipment and storage medium | |
| US11526261B1 (en) | System and method for aggregating and enriching data | |
| CN111737418B (en) | Method, apparatus and storage medium for predicting relevance of search term and commodity | |
| CN114254201A (en) | Recommendation method for science and technology project review experts | |
| CN114266443A (en) | Data evaluation method and device, electronic equipment and storage medium | |
| CN111429161B (en) | Feature extraction method, feature extraction device, storage medium and electronic equipment | |
| CN114997916A (en) | Prediction method, system, electronic device and storage medium of potential user | |
| Alamdari et al. | An image-based product recommendation for E-commerce applications using convolutional neural networks | |
| Weber | Artificial Intelligence for Business Analytics: Algorithms, Platforms and Application Scenarios | |
| US9619757B2 (en) | Nominal feature transformation using likelihood of outcome | |
| CN115760295A (en) | Commodity recommendation method based on neighbor users and comment information | |
| KR102405503B1 (en) | Method for creating predictive market growth index using transaction data and social data, system for creating predictive market growth index using the same and computer program for the same | |
| Bhuvaneswari et al. | Predicting periodical sales of products using a machine learning algorithm | |
| CN114491277A (en) | Knowledge tag endowing and applying method based on data cleaning | |
| CN113254775A (en) | Credit card product recommendation method based on client browsing behavior sequence | |
| CN115409630B (en) | Insurance product accurate recommendation method based on mixed recommendation algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |