TW201822101A - Personalized commodity recommendation method using the potential purchase intension of the consumer as a consideration factor - Google Patents

Abstract

This invention relates to a personalized commodity recommendation method mainly comprising the steps: deploying a plurality of locating sensors in a sale field region, and dividing each region of the sale field region by using each of the locating sensors to establish a spatial index value; then, feeding information back to the sale field region by using a recommendation module carried by a consumer, wherein signal strength data between the recommendation module and each sensor can be fed back or data of commodities that the consumer searched online; then, collecting and calculating information such as the traffic flow and staying time of the consumer as well as commodity classification distance by virtue of a recommendation server; and providing a tag conforming to a behavior of the consumer so as to recommend a commodity which is located around the consumer and has the tag conforming to the behavior of the consumer or commodity classification to the consumer in the sale field region in real time. The personalized commodity recommendation method is a recommendation method in which the potential purchase intension of the consumer is used as a consideration factor.

Description

   Personalized product recommendation method   

The invention relates to a digital product recommendation method, in particular to a personalized product recommendation method that collects potential purchase intentions of consumers.

In the modern era of e-commerce, with the prevalence of web advertising and online shopping malls and other virtual stores, consumers ’willingness to spend at physical stores has gradually decreased. One of the main reasons is that the operating mode of physical stores is similar. Compared to mechanisms such as automatic statistical analysis of virtual stores, it is more difficult to actually understand the contours of customers and the products they are interested in, so it is difficult to provide consumers with incentives to browse products in physical stores, or to target individual consumers with higher purchase rates Plan your marketing strategy.

In the past, the method of product recommendation in physical stores was mostly based on the sales terminal records actually purchased by consumers, and then used association rules or collaborative filtering algorithms to calculate the recommendation results. However, this prior technology only considered When it comes to actual sales, it does not take into account consumers ’repeated lingering, hesitation before buying, or searching online for product-related information, which may show behaviors that are interested in products. These behaviors can effectively show consumers’ potential purchase intentions, but Rarely used for analysis, opportunities that may lead to partial upsets, but specific consumers may buy, or opportunities for product categories to be recommended.

In view of the lack of the foregoing prior art, the inventors of the present invention have considered and developed a recommendation method that can be based on consumers' potential purchase intentions. Through the popularization of personal mobile devices in modern society, it is possible to collect consumers from physical stores or surrounding areas. Behavior, and translate it into information that consumers may be interested in making recommendations, is really an effective recommendation method for individuals.

The invention proposes a personalized product recommendation method, which is a method for labeling consumers and recommending products in consideration of consumers' potential purchase intentions, which is to supplement the unfocused part of the prior art.

The personalized product recommendation method of the present invention is applied to a sales field, such as a department store or a sales circle, and a plurality of positioning sensors are deployed in the sales field. Each of the positioning sensors may be WiFi. Or Beacon devices, etc., to provide customers with online browsing information in the sales field, and each positioning sensor can be used as a positioning reference or given coordinates in the sales field; the method of the present invention will be based on each The position of the positioning sensor divides the sales field into regions, and establishes a spatial index value for each region of the sales field, and displays the products to be sold in each region in the sales field. In the foregoing, according to the planning of the sales field, the spatial index value of each region may correspond to at least one product classification.

Next, since personal mobile devices have become quite popular, the consumer end of the method of the present invention is to install a recommendation module on the mobile device carried by the consumer for deployment, and the consumer will carry the recommendation module installed on the mobile device. Mobile device moves, stays, or searches the Internet for information in the sales field, and the recommendation module mainly returns the recommendation module itself to receive the sensing signals from the positioning sensors in various places in the sales field. The intensity change data is transmitted to a recommendation server of the present invention, and the stay position, time, and movement route of the consumer carrying the recommendation module in the sales field are calculated on the recommendation server, and generated based on the Information on the behavioral attributes of consumers in a first field in the sales field; wherein each of the positioning sensors of the present invention can be positioned through triangles, indoor positioning based on the strength of the RSSI signal, and quadtree Or R-Tree technology to locate the recommended module in the sales field.

In addition, in the method of the present invention, since a number of each of the positioning sensors can provide a network connection function, when a mobile device installed by the consumer that installs the recommended module and each of the network functions are installed, When the positioning sensor is connected to the network, each of the positioning sensors can further obtain the behavior of consumers querying or browsing the webpage via the mobile device, and transmit the behavior to the recommendation server to generate a The behavioral attribute data in the second field. This step is mainly to collect consumers to query or browse the basic information about the product, other purchasing channels, competitive prices, etc. through the recommendation module to understand how consumers are interested in the sales field. Whether the products in it may be interested or have online price comparison behavior.

Then, the recommendation server obtains the change in the relative position of the recommendation module and the displayed product in the sales field and the spatial index value according to the behavior attribute data in the first field, and converts it into the sales field of the consumer. The possible purchase rate of goods displayed in each area of the domain. The recommendation server mainly receives the strength of the sensing signals of the positioning sensors along the time line according to the recommendation module included in the behavior attribute data in the first field. And continuously consider the distribution of the range of distances that can be received at a certain point on the timeline under the condition of fixed signal strength, in order to comprehensively calculate the distance between the consumer and the commodity, in order to further define the consumer's possible purchase rate for each commodity and commodity category, where , The stronger the sensing signal and the closer the distance under the signal strength, the higher the value that defines the likely purchase rate of the consumer.

In addition, the recommendation server uses a string similarity algorithm to calculate behavior attribute data in the second field to filter and compare the product information contained in the consumer ’s query or browsing of the webpage through a mobile device, for example, through Compare the product information contained in a webpage with a longer browsing time for each consumer or information about the surroundings of the product to understand whether the consumer is interested in certain products or compare prices online; among which, the string of the present invention Similarity algorithm can use Levenshtein Distance algorithm, N-gram algorithm, JaroWinkler algorithm, Soundex algorithm, Morris-Pratt algorithm, Knuth-Morris-Pratt algorithm, Aho-Corasick algorithm, Baker-Bird algorithm , Gusfield's algorithm, Boyer-Moore algorithm or Wu-Manber algorithm.

Then, the recommendation server will collect the possible purchase rate based on the behavior attribute data in the first field, and the effect of the web product comparison result added on the behavior attribute data in the second field to the possible purchase rate to aggregate Probability information for a consumer; In addition, the recommendation server is also connected to sales terminals in each area of the sales area to obtain the actual status of consumers who carry the recommendation module in each area of the sales area. Information on the purchase of goods, and aggregated to adjust the consumer's purchase probability data, so the consumer's purchase probability data covers the consumer's potential purchase intention and actual purchase behavior.

Subsequently, the recommendation server introduces an external definition data, which is mainly related to the sales strategy of the sales field or the sales time factors, such as the floor distribution of the sales field, brand name, brand value, promotion schedule, Seasonal, sales time, or scheduled holiday schedules published by the government, the recommendation server filters the information in the consumer behavior attribute data based on the externally defined data or factors actively deployed by the server administrator, and then assigns After carrying the behavioral label of the recommendation module, and after repeatedly implementing the foregoing collection and labeling steps, the consumers in the sales field can be divided into multiple groups according to their respective behavioral labels, and Corresponding to each group, the recommendation server will establish a product classification profile for the behavior label corresponding to the behavior label, for example, consumers who are lame around products with a high discount rate in the sales field, or go online Consumers who inquire and compare products with a high discount rate may be given the behavior label of the discount family and be recommended by the recommendation server. The device is divided into groups of discount families, or it can be subdivided into groups according to different levels of discount rates.

Among them, the clustering method in which the recommendation server divides consumers into the same type of group according to the behavior label can use DBSCAN (Density-based Spatial Clustering of Applications with Noise), OPTICS (Ordering Points To Identify the Clustering Structure), K-Means, Hierarchical Agglomerative Clustering, Hierarchical Divisive Clustering, STING (Statistical Information Grid), CLIQUE (Clustering High-Dimensional Space), Wave-Cluster, COBWEB, CLASSIT, or SOM (Self-Organizing Maps).

Finally, the recommendation server of the present invention will generate a recommendation list according to a recommendation algorithm, and dynamically obtain the spatial index value of the area where the recommendation module is currently located in the sales field, and then place the recommendation list in the recommendation. The product information around the module is transmitted to the recommendation module of each consumer in real time to recommend the product information that conforms to the consumer behavior label to the consumer. Among them, the recommendation algorithm is Apriori, FP using association rule algorithm. -growth, Eclat, etc., or AprioriSome, DynamicSome, FreeSpan, PrefixSpan, SPADE, SPIRIT, CloSpan, etc. of a temporal correlation algorithm, or User-based, Item-based, Model-based, Content-Based of a collaborative filtering algorithm Collaborative Filtering, etc., or Spark and other technologies for massive data analysis.

In this way, the product information recommended by the personalized product recommendation method of the present invention can comprehensively consider various actual or potential purchase intentions of consumers, and will be more in line with consumer expectations than the prior art recommendation methods.

S100 ~ S400‧‧‧step flow

S100 ~ S104‧‧‧step flow

S201 ~ S205‧‧‧step flow

S301 ~ S303‧‧‧step flow

S401 ~ S404‧‧‧step flow

FIG. 1 is a flowchart of main steps of the personalized product recommendation method of the present invention.

FIG. 2 is a flowchart of a first sub-step of the personalized product recommendation method of the present invention.

FIG. 3 is a flowchart of a second sub-step of the personalized product recommendation method of the present invention.

FIG. 4 is a flowchart of a third sub-step of the personalized product recommendation method of the present invention.

FIG. 5 is a flowchart of a fourth sub-step of the personalized product recommendation method of the present invention.

The present invention will be further described with an embodiment in combination with the drawings below. The background of this embodiment is that a company wants to use the personalized product recommendation method of the present invention in a department store of a company to conduct a user entry into a department store. Products Featured.

Please refer to FIG. 1, which is a flowchart of the main steps of the personalized product recommendation method of the present invention, in which a positioning sensor is deployed in step S100 to obtain consumer behavior data, and then filtered through external statistics in step S200 Consumer labels and clusters are given, and then a recommendation list is generated for the recommendation algorithm of step S300, and finally, dynamic recommendation of step S400 is performed; the detailed sub-step flow of each main step will be described in the following paragraphs.

Please refer to FIG. 2 again, which is a flowchart of the first sub-step of the personalized product recommendation method of the present invention, for explaining the sub-step flowchart of step S100 in FIG. 1, where step S100 includes four sub-steps, respectively S101 ~ S104, where the sub-step S101 is to establish a spatial index. First, in order to obtain consumer behavior data, a plurality of positioning sensors need to be selectively deployed on each floor in the department store first, so that each The positioning sensor is to establish the coordinates of each location in the department store. It uses a recommendation server to correlate the coordinate data of the positioning sensor with the current product location and product classification of each area in the department store to summarize. Find out which positioning sensors are located in the same product classification area, and then group the plane coordinates of each positioning sensor, and calculate the lower left corner and upper right corner of the range of each positioning sensor cluster. And the center coordinates, etc., to plan a rectangular geographical area, the recommendation server uses the geographical area of these different product categories to create a spatial index, that is, There is a spatial index value for each geographical range in the cargo company. In this way, it also indicates that each sensor in the geographical area has a spatial index value. These positioning sensor codes, product classification information, and spatial index values Create a regional index correspondence table. An example of a regional index correspondence table can be shown in the following table.

Next, sub-step S102 is performed to collect and analyze the behavior attribute data of the first field of the consumer. The consumer installs a recommendation module on the department store on the mobile device owned by the consumer. When the user is moving or staying in the department store, the recommendation module transmits the sensing signal data received from the positioning sensors in the current area to the recommendation server. The recommendation server further The triangle positioning technology or indoor positioning technology based on the strength of the RSSI signal locates the nearest positioning sensors of the consumer in a unit time, and then associates the sensor code with the regional index correspondence table. The recommended servo The device can know which product categories the consumer's recommendation module appears in or around the geographic area, that is, to obtain the behavior attribute data of the first field of the consumer; Based on information such as signal strength, distance, or dwell time, the distribution calculation can be further summarized into the possible purchases that consumers have the opportunity to purchase this product category. Rate, which rate may be purchased in the following table with reference to Fan.

Then, the sub-step S103 is to collect and analyze the behavior attribute data of the second field of the consumer, that is, to collect the consumption in the same unit time based on the unit time of collecting the behavior attribute data of the first field of the consumer. The data of a person using a mobile device to connect to the Internet through a positioning sensor that provides Internet access is the behavior attribute data in the second field. After the behavior attribute data in the second field is aggregated and analyzed, the data can be obtained. Know whether the content of the webpages browsed by consumers in this region include product webpages or competitive webpages on the Internet, and calculate the correlation between the content of the webpages and the product categories displayed in department stores or the information of the product itself. Degree, adjust the likely purchase rate of consumers.

Next, it is a sub-step S104, that is, to continuously obtain the behavior attribute data of the first field and the behavior attribute data of the second field of the consumer, and analyze the possible purchases of consumers for each product category in each region. Rate in order to generate a consumer purchase probability data.

Subsequent to sub-step S104, it is the external statistical screening of step S200 of the present invention. For the flow chart of the sub-steps to which it belongs, please refer to FIG. 3, where step S200 includes five sub-steps; The purchase probability data of the retailer is analyzed statistically by introducing externally defined data. For example, whether there are discounts or discounts for certain categories of products in the current marketing strategy of the department store in this embodiment. External data such as degrees are introduced into the recommendation server to perform statistical analysis to generate consumer quantity distribution data.

Then, enter the sub-step S202 to dynamically draw the distribution map of the number of consumers. The recommendation server will continuously collect the consumer behaviors represented by all recommendation modules in the department store and continuously analyze statistics; the decision makers of the department store According to the information on the number of consumers, perform sub-step S203 to set the screening condition value. Then, perform sub-step S204 to segment the customer groups. The recommendation server will select the matching customer groups from the consumers according to the screening condition values, and This customer group sets a behavior label; finally, the sub-step S205 is performed to establish a product classification profile, that is, the recommendation server aggregates the customer groups of the same behavior label and establishes a product classification profile of interest, for example, in this embodiment For example, in the external definition data introduced, there are indeed some active product categories with discounts, etc., and you can analyze some consumers who are satisfied or purchase when the product is discounted. The recommendation server will create a Discount family behavior labels, and aggregate the consumer purchase probability data of these discount family consumers to build discount families for them Correspondence table of classification contours of products that may be of interest. For example, one of the contours of the discount family's classification of interest may be a product with a discount rate of 50%.

Please refer to FIG. 4 again, which is a flowchart of the third sub-step of the personalized product recommendation method of the present invention, used to explain the recommendation algorithm of step S300 of the present invention, which includes three sub-steps; first, obtained in sub-step S301 The actual sales data is a recommendation server that obtains the actual product sales data of the sales terminal in the department store, and associates the sold product classification with its location in the department store. For example, consumers can The possible purchase rate of the actual purchased product category is set to 1; then, the sub-step S302 is performed to adjust the consumer purchase rate. The recommendation server associates the consumer purchase probability data generated by the sub-step S104 with the result of the sub-step S301. Set operations to adjust consumer purchase probability data; then, in step S303, perform a calculation using a recommendation algorithm, and finally generate a recommendation list.

Finally, please refer to FIG. 5, which is a flowchart of the fourth sub-step of the personalized product recommendation method of the present invention, which is used to explain the dynamic recommendation of step S400 of the present invention, which also includes three sub-steps; first, tracking consumption for sub-step S401 Among them, when the consumer's recommendation module will continue to receive the sensing signals of several positioning sensors in the department store, it can be used to locate the current location of the consumer and return the spatial index value of its area to Recommendation server; Next, sub-step S402 is a query of behavior tags. The recommendation server inquires that the consumer belongs to the discount family according to information such as the consumer behavior or the identity code of the recommendation module that is continuously generated by the recommendation module. Sub-step S403 is pairing consumer peripheral products. The recommendation server finds the corresponding product classification information according to the spatial index value returned by the recommendation module, and confirms whether it meets the recommendation list generated after sub-step S303. Finally, it is recommended to the consumer in step S404, which filters the recommendation list near its location according to the discount family behavior label of the consumer. Product classification to recommend to consumers instantly.

After explaining the present invention through the foregoing embodiments and drawings, readers should be able to better understand how the personalized product recommendation method of the present invention is implemented. However, it should also be understood that the above description is for explanation rather than limitation. The scope of protection of the present invention.

In summary, the present invention is technically innovative and has multiple effects that are inferior to the previous technology. It has fully met the novel and progressive statutory invention patent requirements. It has filed a patent application in accordance with the law and urges your office to approve this invention. The patent application encourages invention, and it is a matter of virtue.

Claims (8)

    A personalized product recommendation method includes the steps of: arranging a plurality of positioning sensors in a sales field, dividing the sales field with each of the positioning sensors as a group, and setting the sales field in a recommendation server for the sales. Establish spatial index values for each area of the field and the products displayed in the area; feedback information in the sales field through a recommendation module carried by the consumer, the recommendation module returns the recommendation module itself to receive each of the positioning The intensity change data of the sensing signal sent by the sensor is transmitted to the recommendation server, and the recommendation server calculates the stay position, time and movement route of the recommendation module in the sales field to generate information about the consumer Behavior attribute data in a first field of the device; through each of the positioning sensors, a consumer mobile device is provided to connect to the network, so as to obtain the consumer's behavior of querying or browsing the webpage via the mobile device and transmitting it to the recommendation server Device to generate behavior attribute data of a second field of the consumer; the recommendation server obtains the recommendation module and the sales according to the behavior attribute data of the first field The change in the relative position and spatial index value of the displayed products in the domain is converted into the possible purchase rate of consumers for the displayed products in each area of the sales field. The recommendation server calculates the second through a string similarity algorithm. The behavioral attribute data in the field is used to compare the consumer's query or browse the products contained in the webpage through a mobile device, and then adjust the consumer's possible purchase rate, and then aggregate it into a consumer's purchase probability data; the recommendation server introduces An externally defined data about the sales strategy or time factor of the sales field to statistically filter the information in the consumer's purchase probability data, and give consumers with the recommendation module an action label, and then spend According to the behavior label, the user is classified into a group of the same type, and a corresponding product group's consumption tendency corresponding to the behavior label is established according to the behavior label; the recommendation server calculates the consumer purchase probability data according to a recommendation algorithm to generate A recommendation list; and the recommendation server dynamically based on the vacancy of the area where the recommendation module is currently located in the sales area The index value, the transfer list in line with the recommendation of the commodity classification of goods and contour located around the recommendation module product information to the recommendation module to instantly recommended to consumers.         The personalized product recommendation method described in item 1 of the scope of the patent application, wherein the recommendation server is further connected to sales terminals in each area of the sales field to obtain the information of consumers carrying the recommendation module. Information on the actual purchase of goods in each area of the sales field to adjust the purchase probability data of the consumer.         The personalized product recommendation method according to item 1 of the scope of the patent application, wherein the recommendation server receives each of the positioning sensors according to the recommendation module included in the behavior attribute data in the first field. The strength of the signal, and considering the distribution of the range of distances that can be received under this signal strength, calculate the distance between the consumer and the product, and then define the consumer's possible purchase rate of the product, where the stronger the sensing signal and the lower the signal strength The closer the distance, the higher the value of the defined possible purchase rate.         The personalized product recommendation method as described in item 1 of the scope of patent application, wherein the externally defined data includes the floor distribution, brand name, brand value, sales promotion season, seasonality, sales time or holiday schedule of the sales field Cheng and other information.         The personalized product recommendation method according to item 1 of the scope of the patent application, wherein each of the positioning sensors is through triangulation positioning, indoor positioning based on signal RSSI strength, quadtree or R-Tree. ) And other technologies to determine the position of the recommended module in the sales field.         The personalized product recommendation method according to item 1 of the scope of patent application, wherein the string similarity algorithm uses the Levenshtein Distance algorithm, the N-gram algorithm, the JaroWinkler algorithm, the Soundex algorithm, and the Morris-Pratt algorithm Method, Knuth-Morris-Pratt algorithm, Aho-Corasick algorithm, Baker-Bird algorithm, Gusfield's algorithm, Boyer-Moore algorithm or Wu-Manber algorithm.         The personalized product recommendation method described in item 1 of the scope of patent application, wherein the recommendation server uses the DBSCAN (Density-based Spatial Clustering of DBSCAN (Density-based Spatial Clustering of Applications with Noise), OPTICS (Ordering Points To Identify the Clustering Structure), K-Means, Hierarchical Agglomerative Clustering, Hierarchical Divisive Clustering, STING (Statistical Information Grid), CLIQUE (Clustering High-Dimensional Space), Wave-Cluster, COBWEB, CLASSIT or SOM (Self-Organizing Maps).         The personalized product recommendation method described in item 1 of the scope of the patent application, wherein the recommendation algorithm used by the recommendation server is Apriori, FP-growth, Eclat, etc., which uses an association rule algorithm, or a temporal association AprioriSome, DynamicSome, FreeSpan, PrefixSpan, SPADE, SPIRIT, CloSpan, etc. of the algorithm, or User-based, Item-based, Model-based, Content-Based Collaborative Filtering of the collaborative filtering algorithm, or the analysis of huge amounts of data Spark and other technologies.