WO2017041226A1 - Commodity information pushing method - Google Patents

Abstract

A commodity information pushing method, comprising the following steps: receiving and responding to a request of a consumer for purchasing a commodity (S1); defining a support degree, a confidence degree and an association rule (S2); finding out a frequent item set and a longest frequent item set (S3); forming a plurality of candidate association rules (S4); respectively calculating the confidence degrees of the candidate association rules (S5); generating an association rule (S6); and pushing corresponding commodity information to the consumer according to the association rule (S7). According to the present invention, an association algorithm is used to analyse information about a commodity of a consumer, so that more accurate commodity pushing or advertisement publication can be provided.

Description

Product information pushing method Technical field

The present invention relates to the field of information processing, and in particular, to a method for pushing commodity information.

Background technique

With the development of e-commerce technology, commodity push technology that can accurately locate consumer demand has become a research hotspot. In the prior art, by counting the number of times a consumer purchases a certain type of product or the distribution of a commodity price interval, the method of pushing the product to the consumer can no longer satisfy the demand of the merchant, and can record the situation of the commodity purchased by the consumer and analyze the consumption. The habit of buying, and then intelligently pushing the surrounding products (including spare parts) related to the goods purchased by consumers is the demand of the majority of businesses. For example, when consumers buy mobile phones, the product push system can push charging treasures and headphones; consumers buy development boards, and the product push system can push downloaders and DuPont lines. It can be seen that the method of accurately pushing the surrounding goods related to the purchase of goods not only helps to enhance the sales of the merchants, but also enhances the purchasing experience of the consumers.

Summary of the invention

The technical problem to be solved by the present invention is to provide a method for accurately pushing peripheral goods related to purchased products.

A method for pushing product information includes the following steps:

Receiving a request for a customer to purchase an item, the request for purchasing the item at least including category information of the item, wherein one category of the item includes at least one specific item;

Responding to the client request, obtaining, according to the category information of the purchased product of the customer, information of at least one third-party purchased product, wherein the information of the third-party purchased product includes at least category information of the purchased product of the customer;

Define support, confidence, and association rules;

The information of the third party purchase commodity is combined into at least one candidate item set according to a preset method, wherein the candidate item set whose support degree is greater than or equal to the preset support degree is a frequent item set;

Select the frequent itemsets that contain the most types of products as the longest frequent itemsets;

All the commodities constituting the longest frequent item set are arbitrarily combined to form a plurality of candidate association rules;

Calculating the confidence of the candidate association rule separately;

Selecting a candidate association rule whose candidate association rule satisfies a preset confidence value greater than or equal to a preset confidence degree as a specified association rule;

Pushing corresponding product information to the consumer according to the prescribed association rule.

Further, the method for determining the candidate item set and finding the frequent item set and the longest frequent item set includes the following steps:

Determining a first layer candidate item set, each candidate item set of the layer is an item, and traversing the layer to calculate the support degree of all candidate item sets;

Compare the support of each candidate item set with the preset support level;

If the support degree of the candidate item set is greater than or equal to the preset support degree, extracting the candidate item set as a frequent item set;

The candidate set of the new layer is generated by the frequent set of items found in the previous time, and the support of all candidate sets of the new layer is calculated to generate a new frequent set of items, and so on;

If the support degree of the candidate item set is less than the preset support degree, the generation of the frequent item set is ended, and the last frequent item set containing the most types of goods is the longest frequent item set.

Further, the obtained information of the third party purchase commodity is preprocessed.

Further, the e-commerce pushes related products to the mobile phone number or mailbox registered by the consumer.

Further, pre-processing the obtained information of the third-party purchased product includes: unifying the data format and eliminating duplicate data.

Further, the process of unifying the data format includes: data type conversion, attribute construction, data discretization, and data standardization.

The beneficial effects of the present invention compared with the prior art are: the commodity information pushing method of the present invention uses an association algorithm to analyze consumer product information to provide more accurate product pushing or advertising. For the manufacturer, it is able to sell a certain product to the customer, push the related products or peripheral parts and consumables to the customer, broaden the sales channel, and improve the marketing efficiency. For the customer, according to the new product advertisements released by the merchant and the information pushed by the merchant, more product information is obtained, which facilitates the purchase of the customer and broadens the purchase channel.

DRAWINGS

1 is a flow chart of some embodiments of the present invention.

2 is a flow chart of finding a frequent item set and a longest frequent item set according to the present invention.

detailed description

In order to more fully understand the technical content of the present invention, the technical solutions of the present invention will be further described and illustrated below in conjunction with specific embodiments.

As shown in FIG. 1, in some embodiments, the commodity information pushing method includes the following steps:

S1. Receiving, by the customer, a request for purchasing a commodity, where the request for purchasing the commodity includes at least category information of the commodity, wherein one category of the commodity includes at least one specific commodity; and responding to the customer request, according to the category of the customer purchasing the commodity The information acquires information of at least one third party purchase item, wherein the information of the third party purchase item includes at least category information of the item purchased by the customer.

For step S1, in order to quickly and accurately locate the consumer's shopping demand, when the merchant enters the network platform provided by the e-commerce, the e-commerce classifies the merchant's merchandise, and the common merchandise categories such as clothing, 3C, and daily necessities. After the merchants to be settled are verified by the e-commerce, their product information will be published to the corresponding information release section according to the classification of the e-commerce. After the consumers purchase the goods on the online platform provided by the e-commerce, the relevant product information (including the product category, product name, price, purchase time, etc.) is stored in the back-end server of the e-commerce. For example, if the consumer A currently purchases the apparel product, the product information of the remaining consumers (ie, third parties) who also have the purchase record of the apparel product may be selected.

The product information may also be selected according to the personal information (including age, gender, etc.) of the consumer A. For example, if the consumer A is a female, the product information of the remaining female consumers is selected.

In order to reduce the amount of irrelevant data, in order to improve the efficiency and accuracy of commodity push, the commodity information obtained in step S1 can also be preprocessed.

For example, some items are very time-sensitive, and a time T can be defined. If the purchase time of the item has exceeded the time T today, the item can be deleted from the information of the third-party purchased item obtained in step S1.

In some embodiments, there is a case where product information of a plurality of e-commerce back-end servers is merged. Since the source of the product information is different, the definition of the same item may be different, such as the base of a certain part, Some are represented by "base", some are represented by "base", and some are represented by "base", but they are essentially a commodity, so they need to be uniformly defined. Therefore, it is also necessary to unify the data format and eliminate duplicate data. Those skilled in the art can understand that the process of unifying the data format includes: data type conversion, attribute construction, data discretization, and data standardization.

The data discretization method may be: discretizing the attributes of the continuous values into a plurality of intervals. For example, the price of a commodity is continuous, but consumers tend to choose within a range, not an independent and exact value. Therefore, for the convenience of analysis, the interval can be divided according to the price segment, and each interval is represented by a numerical value. For example, the numbers 1, 2, 3, 4, ... can be used to represent: 0 to 30, 31 to 60, 60 to 90, 90 to 120, ... different price ranges.

S2, defines support, confidence, and association rules.

The two thresholds of support and confidence are two concepts describing association rules. Support supports the importance of association rules in the database. Confidence measures the credibility of association rules. Support (X, Y) ≥ T1 means that among all the purchase records selected, at least the purchase of T1 (available percentage) is presented as two items of X and Y being purchased by the same consumer. Confidence(X,Y) ≥ T2 means that the purchase of at least T2 (available percentage) is presented as the purchase of X by the same consumer.

Set minimum support (min-Support) = F1, minimum confidence (min-Confidence) = F2. The range of F1 and F2 is [0, 1]. If the association rule {X, Y} found by mining finds both the minimum support and the minimum confidence condition, it means that the association rule of {X, Y} is acceptable, and the association rule is generated. It can be described as: Support (X, Y) ≥ F1 and Confidence (X, Y) ≥ F2.

For example, the purchase record of the clothing product in which the consumer A currently purchases the clothing product and the third party purchases the information stored in the background server is as shown in Table 1, wherein one serial number represents a consumer, A, B. , C, D, E, F each represents different commodities of clothing:

Table 1

Serial number	commodity
		1	{A, C}

2	{A, D, E, F}
		3	{B, C, D, E}
4	{A, C, D, E}
		5	{A, B, C, E}

According to Table 1, the degree of support and confidence of consumers purchasing D products in products containing {C, E} are:

Support Support=2/5=0.4 (At least 0.4 of the purchases in the selected purchases are presented as the same consumer who purchased D, C, and E.)

Confidence=2/3≈0.67 (At least 0.67 of the purchases selected are presented as the purchase of D, the same consumer purchased both C and E.)

S3. The information of the third-party purchased product is combined into at least one candidate item set according to a preset method, wherein the candidate item set whose support degree is greater than or equal to the preset support degree is a frequent item set; and the frequent item including the most product type is selected. Set as the longest frequent item set.

Here, the preset support degree is the defined minimum support degree. As shown in Figure 2, the following methods can be used:

S31. Determine a first layer candidate item set (K=1), and each candidate item set of the layer is a commodity, and traverse the layer to calculate the support degree of all candidate item sets.

The project set shown in Table 2 is the first candidate project set, and the count and support of each project set are as follows:

Table 2

Candidate set	count	Support
			{A}	4	80%
{B}	2	40%
			{C}	4	80%
{D}	3	60%
			{E}	4	80%
{F}	1	20%

S32. Compare the support degree of each candidate item set with a minimum support degree threshold.

It is assumed that the minimum support degree threshold F1 is set to 60%, wherein the candidate item set {B}, {F} support degree <F1, candidate item set {A}, {C}, {D}, {E} ≥ F1.

It will be understood by those skilled in the art that if the minimum support threshold F1 satisfies the requirement, the minimum support count value is at least 3. Therefore, it can also be judged based on the minimum support count value.

S33. If the candidate item set has a minimum support degree threshold requirement, the candidate item set is extracted as a frequent item set.

According to step S32, the candidate item sets {A}, {C}, {D}, and {E} are frequent item sets.

S34, the candidate item set of the new layer is generated by the frequent item set found last time, and the support degree of all candidate item sets of the new layer (K=K+1) is calculated to produce a new frequent item set, and so on.

For step S34, it can be seen from step S33 that the situation of the second layer candidate item set is as shown in Table 3:

table 3

Candidate set	count
		{A, C}	3
{A,D}	2
		{A,E}	3
{C,D}	2
		{C,E}	3
{D,E}	3

The candidate project sets {A, D} and {C, D} do not satisfy the minimum support, while the candidate project sets {A, C}, {A, E}, {C, E}, {D, E} are frequent items. set. In this analogy, continue to the next step, the situation of the third-level candidate project set is shown in Table 4:

Table 4

Candidate set	count
		{A, C, E}	3

As can be seen from Table 4, {A, C, E} is a frequent item set. By analogy with this method, continue to the next step:

table 5

Candidate set	count
		{A, C, D, E}	1

Since the count value of the candidate item set {A, C, D, E} is 1, the requirement of the minimum support degree threshold is not satisfied, and it is not a frequent item set.

S35. If the support degree of the candidate item set is less than the minimum support degree threshold, the generation of the frequent item set is ended, and the frequently obtained frequent item set including the most common item type is the longest frequent item set.

For step S35, it is known from step S34 that the longest frequent item set is {A, C, E}.

S4, arbitrarily combine all the commodities constituting the longest frequent item set described above to form a plurality of candidate association rules.

According to the longest frequent items set {A, C, E} obtained by S3, multiple candidate association rules can be obtained, as follows:

(1) Rules {A, C}→{E}

(2) Rules {A, E}→{C}

(3) Rules {C, E}→{A}

(4) Rules {A}→{C,E}

(5) Rules {C}→{A,E}

(6) Rules {E}→{A,C}

Since the candidate association rule is formed by any combination of all the commodities of the longest frequent item set, the support degree of the above association rule satisfies the minimum support degree threshold requirement.

S5, respectively calculating the confidence of the candidate association rule.

(1) Confidence of the rule {A, C}→{E} Confidence==3/3=1

(2) Confidence of the rule {A, E}→{C} Confidence==3/3=1

(3) Confidence of the rule {C, E}→{A} Confidence==3/3=1

(4) Confidence of rule {A}→{C,E} Confidence==3/4=0.75

(5) Confidence of rule {C}→{A, E} Confidence==3/4=0.75

(6) Confidence of rule {E}→{A, C} Confidence==3/4=0.75

S6. Generating an association rule: selecting a candidate association rule whose candidate association rule satisfies a predetermined confidence greater than or equal to a predetermined confidence rule as a specified association rule

The processing reliability is the defined minimum confidence threshold. If the minimum confidence threshold F2 is 0.75, the candidate association rules of step S5 all meet the minimum confidence requirement. The association rule defined by step S2 knows that the candidate shown in step S5 is known. The association rules are all associated rules that meet the requirements.

S7, pushing corresponding product information to consumers according to the association rule specified above.

(1) The rules {A, C} → {E}, if the consumer has purchased the A and C goods, they should push the E product to them.

(2) The rules {A, E} → {C}, if the consumer has purchased the A, E goods, they should push the C goods to them.

(3) The rule {C, E}→{A} means that if the consumer has purchased the C or E goods, they should push the A product to them.

(4) The rule {A}→{C,E} means that if the consumer has purchased the A product, the C and E products should be pushed to it.

(5) The rule {C}→{A, E} means that if the consumer has purchased the C product, the A and E products should be pushed to it.

(6) The rule {E}→{A,C} means that if the consumer has purchased the E product, the A and C products should be pushed to it.

The product information pushing method of the invention analyzes the consumer's product information by using an association algorithm to provide more accurate product pushing or advertising. For the manufacturer, it is able to sell a certain product to the customer, push the related products or peripheral parts and consumables to the customer, broaden the sales channel, and improve the marketing efficiency. For the customer, according to the new product advertisements released by the merchant and the information pushed by the merchant, more product information is obtained, which facilitates the purchase of the customer and broadens the purchase channel.

The above description is only to clarify the technical content of the present invention by way of examples, so that the reader can understand the present invention more easily, but the embodiment of the present invention is not limited thereto, and any technology extension or re-creation according to the present invention is subject to the present invention. Protection of the invention.

Claims (6)

1. A commodity information pushing method, comprising the following steps:

   Receiving a request for a customer to purchase an item, the request for purchasing the item at least including category information of the item, wherein one category of the item includes at least one specific item;

   Responding to the client request, obtaining, according to the category information of the purchased product of the customer, information of at least one third-party purchased product, wherein the information of the third-party purchased product includes at least category information of the purchased product of the customer;

   Define support, confidence, and association rules;

   The information of the third party purchase commodity is combined into at least one candidate item set according to a preset method, wherein the candidate item set whose support degree is greater than or equal to the preset support degree is a frequent item set;

   Select the frequent itemsets that contain the most types of products as the longest frequent itemsets;

   All the commodities constituting the longest frequent item set are arbitrarily combined to form a plurality of candidate association rules;

   Calculating the confidence of the candidate association rule separately;

   Selecting a candidate association rule whose candidate association rule satisfies a preset confidence value greater than or equal to a preset confidence degree as a specified association rule;

   Pushing corresponding product information to the consumer according to the prescribed association rule.
2. The article information pushing method according to claim 1, wherein the method of determining the candidate item set and finding the frequent item set and the longest frequent item set comprises the following steps:

   Determining a first layer candidate item set, each candidate item set of the layer is an item, and traversing the layer to calculate the support degree of all candidate item sets;

   Compare the support of each candidate item set with the preset support level;

   If the support degree of the candidate item set is greater than or equal to the preset support degree, extracting the candidate item set as a frequent item set;

   The candidate set of the new layer is generated by the frequent set of items found in the previous time, and the support of all candidate sets of the new layer is calculated to generate a new frequent set of items, and so on;

   If the support degree of the candidate item set is less than the preset support degree, the generation of the frequent item set is ended, and the last frequent item set containing the most types of goods is the longest frequent item set.
3. The article information pushing method according to claim 1, wherein the acquired information of the third party purchased product is preprocessed.
4. The product information pushing method according to claim 1, wherein the e-commerce pushes the related product to the mobile phone number or the mailbox registered by the consumer.
5. The product information pushing method according to claim 3, wherein the pre-processing of the acquired information of the third-party purchased product comprises: unifying the data format and eliminating duplicate data.
6. The commodity information pushing method according to claim 5, wherein the process of unifying the data format comprises: data type conversion, attribute construction, data discretization, and data standardization.

Images