CN112232896A - Commodity recommendation method and device - Google Patents
Commodity recommendation method and device Download PDFInfo
- Publication number
- CN112232896A CN112232896A CN202010949448.3A CN202010949448A CN112232896A CN 112232896 A CN112232896 A CN 112232896A CN 202010949448 A CN202010949448 A CN 202010949448A CN 112232896 A CN112232896 A CN 112232896A
- Authority
- CN
- China
- Prior art keywords
- target
- commodity
- probability
- account
- list
- 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.)
- Withdrawn
Links
Images
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
- G06Q30/0207—Discounts or incentives, e.g. coupons or rebates
- G06Q30/0222—During e-commerce, i.e. online transactions
-
- 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]
- G06Q30/0623—Item investigation
- G06Q30/0625—Directed, with specific intent or strategy
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a commodity recommendation method and device. Wherein, the method comprises the following steps: acquiring target data characteristics of a target account for browsing a target commodity in a client, wherein the target data characteristics comprise behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity and order characteristics of the target commodity; inputting the target data feature into a first neural network model; inputting the target data features and temporal features into a second neural network model; and under the condition that the first probability is within a first interval and the second probability is within a second interval, pushing a commodity list to the target account, wherein the commodity list comprises commodities related to the target commodity. The invention solves the technical problem of inaccurate commodity pushing.
Description
Technical Field
The invention relates to the field of online shopping, in particular to a commodity recommendation method and device.
Background
In the prior art, in the process of online shopping, commodities can be pushed to a user generally.
In the prior art, when the commodities are pushed to the user, most of the pushed commodities are matched according to the search records of the user. For example, after "bicycle" is searched in the user history, when the user searches for "iron pan", the goods related to "bicycle" are pushed.
The pushing method causes that the accuracy of the commodities pushed to the user is low in the online shopping process of the user.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the invention provides a commodity recommendation method and a commodity recommendation device, which at least solve the technical problem of inaccurate commodity pushing.
According to an aspect of an embodiment of the present invention, there is provided a commodity recommendation method including: acquiring target data characteristics of a target account for browsing a target commodity in a client, wherein the target data characteristics comprise behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity and order characteristics of the target commodity; inputting the target data feature into a first neural network model, wherein the first neural network model is a model for outputting a first probability representing a probability of the target account number purchasing the target commodity; inputting the target data characteristic and a time characteristic into a second neural network model, wherein the second neural network model is a model for outputting a second probability, the second probability is used for representing the probability that the target account number purchases the target commodity in advance, and the time characteristic is the time used by the target account number for browsing the target commodity; and pushing a product list to the target account, wherein the product list includes products related to the target product, when the first probability is within a first interval and the second probability is within a second interval.
As an optional example, in a case that the first probability is within a first interval and the second probability is within a second interval, before pushing the item list to the target account, the method further includes: acquiring a first list, wherein the first list comprises a plurality of commodities; determining the similarity between each commodity in the first list and the target commodity; and determining the goods in the first list, of which the similarity with the target goods is higher than a first threshold value, as the goods in the goods list.
As an optional example, in the case that the first probability is within a first interval and the second probability is within a second interval, pushing the item list to the target account includes: and pushing the commodity list and a target coupon to the target account, wherein the target coupon is a coupon used when the target account purchases commodities in the commodity list.
As an optional example, before the pushing the item list and the target coupon to the target account, the method further includes: adjusting the value of the order coupon in the order characteristics in the target data characteristics to update the target data characteristics to obtain the updated target data characteristics; inputting the updated target data features into the first neural network model and the second neural network model to obtain the first probability and the second probability; and determining the value of the order coupon when the first probability and the second probability are the maximum as the value of the target coupon.
As an optional example, before inputting the target data feature into the first neural network model, the method further includes: acquiring first sample data, wherein the first sample data comprises a first data characteristic of a first account for browsing a first commodity, and the first data characteristic comprises a behavior characteristic of the first account, an account characteristic of the first account, a commodity characteristic of the first commodity and an order characteristic of the first commodity; training the first neural network model using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold.
As an optional example, before inputting the target data feature into the second neural network model, the method further includes: acquiring second sample data, wherein the second sample data comprises second data characteristics and time characteristics of a second commodity browsed by a second account, the second data characteristics comprise behavior characteristics of the second account, account characteristics of the second account, commodity characteristics of the second commodity and order characteristics of the second commodity, and the time characteristics comprise time consumed by the second account for browsing the second commodity; and training the second neural network model by using the second sample data until the identification accuracy of the second neural network model is greater than a third threshold value.
According to another aspect of the embodiments of the present invention, there is also provided a commodity recommending apparatus including: a first obtaining unit, configured to obtain a target data feature of a target commodity browsed by a target account in a client, where the target data feature includes a behavior feature of the target account, an account feature of the target account, a commodity feature of the target commodity, and an order feature of the target commodity; a first input unit configured to input the target data feature into a first neural network model, wherein the first neural network model is a model for outputting a first probability indicating a probability that the target account purchases the target product; a second input unit, configured to input the target data feature and a time feature into a second neural network model, where the second neural network model is a model for outputting a second probability that represents a probability that the target account purchases the target product in advance, and the time feature is a time used by the target account to browse the target product; and a pushing unit configured to push a product list to the target account, where the product list includes a product related to the target product, when the first probability is within a first interval and the second probability is within a second interval.
As an optional example, the apparatus further includes: a second obtaining unit, configured to obtain a first list before pushing the product list to the target account if the first probability is within a first interval and the second probability is within a second interval, where the first list includes a plurality of products; a first determining unit, configured to determine a similarity between each product in the first list and the target product; a second determining unit configured to determine, as a product in the product list, a product in the first list whose similarity with the target product is higher than a first threshold.
As an alternative example, the pushing unit includes: and the pushing module is used for pushing the commodity list and the target coupon to the target account, wherein the target coupon is a coupon used when the target account purchases commodities in the commodity list.
As an optional example, the pushing unit further includes: an adjusting module, configured to adjust a value of an order coupon in the order feature in the target data feature before pushing the commodity list and the target coupon to the target account, so as to update the target data feature, and obtain an updated target data feature; inputting the updated target data features into the first neural network model and the second neural network model to obtain the first probability and the second probability; and the determining module is used for determining the numerical value of the order coupon under the condition that the first probability and the second probability are the maximum as the numerical value of the target coupon.
As an optional example, the apparatus further includes: a third obtaining unit, configured to obtain first sample data before inputting the target data feature into the first neural network model, where the first sample data includes a first data feature of a first account browsing a first commodity, and the first data feature includes a behavior feature of the first account, an account feature of the first account, a commodity feature of the first commodity, and an order feature of the first commodity; a first training unit, configured to train the first neural network model using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold.
As an optional example, the apparatus further includes: a fourth obtaining unit, configured to obtain second sample data before inputting the target data feature into the second neural network model, where the second sample data includes a second data feature and a time feature of a second account browsing a second commodity, the second data feature includes a behavior feature of the second account, an account feature of the second account, a commodity feature of the second commodity, and an order feature of the second commodity, and the time feature includes time consumed by the second account browsing the second commodity; and the second training unit is used for training the second neural network model by using the second sample data until the identification accuracy of the second neural network model is greater than a third threshold value.
In the embodiment of the invention, a target data characteristic of a target commodity is browsed in a client by acquiring a target account, wherein the target data characteristic comprises a behavior characteristic of the target account, an account characteristic of the target account, a commodity characteristic of the target commodity and an order characteristic of the target commodity; inputting the target data feature into a first neural network model, wherein the first neural network model is a model for outputting a first probability representing a probability of the target account number purchasing the target commodity; inputting the target data characteristic and a time characteristic into a second neural network model, wherein the second neural network model is a model for outputting a second probability, the second probability is used for representing the probability that the target account number purchases the target commodity in advance, and the time characteristic is the time used by the target account number for browsing the target commodity; under the condition that the first probability is in a first interval and the second probability is in a second interval, pushing a commodity list to the target account, wherein the commodity list comprises commodities associated with the target commodity.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
FIG. 1 is a schematic diagram of an application environment of an alternative merchandise recommendation method according to an embodiment of the invention;
FIG. 2 is a schematic diagram of an application environment of an alternative merchandise recommendation method according to an embodiment of the invention;
FIG. 3 is a schematic diagram illustrating the flow of an alternative merchandise recommendation method according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a recommended article list of an alternative article recommendation method according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating a determined item list of an alternative item recommendation method according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of determining commodity similarity for an alternative commodity recommendation method according to an embodiment of the present invention;
FIG. 7 is a schematic diagram of a push targeted coupon for an alternative merchandise recommendation method according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of an alternative merchandise recommendation device according to an embodiment of the present invention;
FIG. 9 is a schematic diagram of an alternative merchandise recommendation device according to an embodiment of the present invention;
fig. 10 is a schematic structural diagram of yet another alternative merchandise recommendation device according to an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
According to an aspect of the embodiments of the present invention, there is provided a product recommendation method, which may be, but is not limited to, applied to the environment shown in fig. 1 as an optional implementation manner.
As shown in fig. 1, the terminal device 102 includes a memory 104 for storing terminal device data, a processor 106 for processing the terminal device data, and a display 108 for displaying the data. The terminal device 102 may interact with a server 112 via the network 110, and the server 112 includes a database 114 for storing data and a processing engine 116 for processing data. The target account number may log in a client running on the terminal device 102 to browse the target product. The terminal device sends the target data characteristics of the target commodity browsed by the target account in the client to the server 112, the first and second probabilities are output by the first and second neural network models in the server 112, if the first and second probabilities meet the pushing requirement, the commodity list is pushed to the terminal device 102, and the commodity list is displayed by the terminal device 102.
Alternatively, as an alternative implementation, the above commodity recommendation method may be applied, but not limited, to the environment shown in fig. 2.
As shown in fig. 2, the terminal device 202 includes a memory 204 for storing terminal device data, a processor 206 for processing terminal device data, and a display 208 for displaying data. The target account number may log on a client running on the terminal device 202 to browse the target product. The terminal device 202 inputs the target data characteristics of the target account browsing the target commodity in the client to the first neural network model and the second neural network model in the terminal device 202, the first neural network model and the second neural network model output a first probability and a second probability, and if the first probability and the second probability meet the pushing requirement, the terminal device 102 displays the commodity list.
Optionally, in this embodiment, the terminal device may be a terminal device configured with a target client, and may include, but is not limited to, at least one of the following: mobile phones (such as Android phones, iOS phones, etc.), notebook computers, tablet computers, palm computers, MID (Mobile Internet Devices), PAD, desktop computers, smart televisions, etc. The target client may be a video client, an instant messaging client, a browser client, an educational client, etc. Such networks may include, but are not limited to: a wired network, a wireless network, wherein the wired network comprises: a local area network, a metropolitan area network, and a wide area network, the wireless network comprising: bluetooth, WIFI, and other networks that enable wireless communication. The server may be a single server, a server cluster composed of a plurality of servers, or a cloud server. The above is merely an example, and this is not limited in this embodiment.
Optionally, as an optional implementation manner, as shown in fig. 3, the method for recommending a commodity includes:
s302, acquiring target data characteristics of a target commodity browsed by a target account in a client, wherein the target data characteristics comprise behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity and order characteristics of the target commodity;
s304, inputting the target data characteristics into a first neural network model, wherein the first neural network model is used for outputting a first probability, and the first probability is used for representing the probability of the target account number for purchasing the target commodity;
s306, inputting the target data characteristics and the time characteristics into a second neural network model, wherein the second neural network model is used for outputting a second probability, the second probability is used for representing the probability that the target account number purchases the target commodity in advance, and the time characteristics are the time used by the target account number for browsing the target commodity;
s308, under the condition that the first probability is in the first interval and the second probability is in the second interval, pushing a commodity list to the target account, wherein the commodity list comprises commodities related to the target commodity.
Alternatively, the above method may be applied to any web shopping website or applet or application. The target account can be an account used by the user in the shopping process, the account can be a long-term account registered by the user or a temporary account, and the temporary account can be purchased and paid by binding other applications or jumping to other applications. When a user browses a target commodity, the target data characteristic of the target commodity browsed by the user can be obtained, and the time characteristic can be time consumed by the user for browsing the target commodity. After the target data characteristics and the time characteristics are obtained, the target data characteristics are input into a first neural network model, the target data characteristics and the time characteristics are input into a second neural network model, the first neural network model outputs a first probability, the first probability can be the probability that a user will purchase a target commodity, the second model outputs a second probability, and the second probability can be the probability that the user pre-purchases the target commodity. And if the first probability is in the first interval and the second probability is in the second interval, pushing a commodity list to the user, wherein commodities in the commodity list are associated with the target commodity. By the method, the effect of improving the accuracy of recommending commodities is achieved.
For example, as shown in fig. 4, fig. 4 is a schematic diagram of a recommended commodity list, in fig. 4, a user may search for a toothbrush, and then click on the toothbrush 1 to view specific information, with the toothbrush 1 being a target commodity. And the system recommends a list of goods under toothbrush 1, including toothbrush 2 and toothbrush 3.
Optionally, the target data characteristics in the present application may include behavior characteristics of the target account, where the behavior characteristics refer to operations executed when the target account logs in the client, such as button clicking, returning, sliding, page jumping, determining, and the like, and may be specifically implemented by using functions provided by the application. If the user browses the next page of commodities through sliding operation, the terminal can record the operation of the user and the browsed commodities; the account characteristics of the target account can be attribute characteristics of the target account, such as whether the target account is a member or not, the grade of the account, the historical browsing records of the account, the unique identification of the account, the historical purchasing records of the account and the like; the commodity characteristics of the target commodity can be attributes of the commodity, such as price, type, shelf date, shelf life, commodity displacement identification, commodity sales volume and other characteristics; an order characteristic of the target item, which may be whether the user created an order for the item and if created, the number of orders created, and whether the order was cancelled, etc. The first probability in this application is a probability that the user purchases the target product, for example, 99%, which means that the user may purchase the target product with a probability of 99%. And the second probability is the probability that the user buys the target commodity in advance, and the pre-purchase can be that the user is interested in the target commodity and does not decide to purchase. That is, pre-purchase is a possibility of interest to the target commodity. If the second probability is 80%, it indicates that the user has an 80% probability of being interested in the target item.
The first interval and the second interval in the present application are preset interval ranges, and may include overlapped portions, completely non-overlapped portions, or one of the two may include the other, and may be specifically set according to actual conditions.
Optionally, the article list in the present application may include one or more articles, and the number of the articles is not limited. Each item in the list of items is associated with a target item. The goods list may be a list obtained by filtering the goods in the first list. The first list may include a plurality of products, and the first list may be a preset list or a list set by the system according to a real-time situation. The items in the first list may or may not be associated with the target item. When the commodity list is recommended, determining commodities in the first list, the similarity of which to the target commodity is greater than a first threshold value, as commodities in the commodity list, and pushing the commodity list to the user. Each item in the item list is an item having a similarity greater than a first threshold value with the target item.
For example, as shown in fig. 5, fig. 5 is an alternative schematic diagram of determining the item list according to the first list. In the above process, as the first list includes a plurality of items, then the target item is a bristle toothbrush, the similarity of each item to the target item is determined, and the items with the similarity higher than a first threshold value, such as 80%, are determined as the items in the item list. The determination of the similarity may be determined according to the type of the article. If all the commodities can be classified, the categories can be divided into a large category and a small category under the large category, for example, the large category is food and clothes, the small category of the food is cooked food and uncooked food, and the small category of the clothes is top-up and bottom-up. Of course, smaller categories may be subdivided under the subclasses. If the similarity of the commodities is compared, the similarity of the commodities can be determined by comparing the categories of the commodities, and if the two commodities belong to the same category, the similarity of the commodities is determined according to the level of the category. If two items belong to the same food, the similarity may be low, perhaps only 20%, whereas if two items belong to the same bacon under cooked food, the similarity is high, perhaps 90%. The smaller the category of the two commodities, the higher the similarity. For example, as shown in fig. 6, fig. 6 is an alternative schematic diagram for determining similarity category hierarchy of a commodity. The smaller the class, the higher the similarity.
Optionally, when the commodity list is pushed to the target account, the target coupon may also be pushed to the target account. The targeted coupon is a line that may be discounted if the user purchases an item from the list of items. For example, when the user looks at the target product, the product list is pushed to the user, the product in the product list can use the target coupon, and the target coupon is 10 yuan, so that the user can give a discount for 10 yuan when purchasing the product in the product list, and cannot give a discount for the target coupon when purchasing the target product.
When the target coupon is pushed, the preferential amount can be determined according to the specific information of the commodity. For example, the more expensive the merchandise, the greater the premium amount. Alternatively, the amount of the offer may be determined using a model. For example, the value of the order coupon in the order feature in the target data feature is adjusted, then the target data feature of the value of the different order coupon is input into the first neural network model and the second neural network model, the first probability and the second probability output by the first neural network model and the second neural network model are obtained, and the weighted summation result of the first probability and the second probability is obtained. For example, the weighted sum result corresponding to the order coupon of 10 blocks is 0.8, the weighted sum result corresponding to the order coupon of 6 blocks is 0.6, the order coupon with the largest weighted sum result is used as the target coupon, and when the product list is pushed to the user, the target coupon is pushed. For example, as shown in FIG. 7, FIG. 7 is a schematic diagram of an alternative push targeted coupon. When the toothbrush 2 and the toothbrush 3 are recommended, a 2-yuan coupon and a 5-yuan coupon are respectively associated and can be received for use.
Optionally, the first neural network model and the second neural network model in the present application are models trained in advance. Acquiring first sample data, wherein the first sample data comprises first data characteristics of a first account for browsing a first commodity, and the first data characteristics comprise behavior characteristics of the first account, account characteristics of the first account, commodity characteristics of the first commodity and order characteristics of the first commodity; the first neural network model is trained using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold. Acquiring second sample data, wherein the second sample data comprises second data characteristics and time characteristics of a second commodity browsed by a second account, the second data characteristics comprise behavior characteristics of the second account, account characteristics of the second account, commodity characteristics of the second commodity and order characteristics of the second commodity, and the time characteristics comprise time consumed by the second account for browsing the second commodity; training the second neural network model using the second sample data until the recognition accuracy of the second neural network model is greater than a third threshold.
The second threshold and the third threshold may be preset values, which indicate that the accuracy of the model needs to be higher than the preset values before the model can be put into use.
The above-described product recommendation method is explained below with reference to specific examples. The commodity recommendation method is applied to the shopping process of the user.
Pre-constructing a database, wherein the database comprises:
1. the data of the buried point is the operation of the user on the shopping website or the content on the page, such as clicking, sliding and rolling operations, and records the use of the user on the page or the function on the website, such as determining, canceling, jumping, clicking a link, adding a shopping cart, purchasing, canceling and the like.
2. The user characteristic data comprises attribute information of the user, such as unique identification, historical data, relevance of browsed commodities, repeated times of browsing the commodities and the like, and time characteristics, wherein the time characteristics are the time of the user for browsing a certain commodity.
3. The commodity characteristic data comprises the unique identification of the commodity, the type of the commodity, the price of the commodity, the sales volume, the sales manufacturer, the manufacturer and other attribute information.
4. Order data, which may include data on whether to create an order, whether to cancel the created order, whether to pay for a completed purchase, coupon information in the order, coupon pickup, number of views, etc.
After creating the database in which the above information is recorded, the first neural network model and the second neural network model can be trained using the information in the database. The first neural network model may be trained using data such as buried point data, relevance of browsing a commodity by user behavior data (user characteristic data), number of iterations of browsing a commodity, number of times of adding a similar commodity to a shopping cart, coupon pickup, browsing number, etc., and data in the database is tagged, for example, that a commodity is purchased by a user or that a commodity is not purchased by a user. And training the first neural network model through the data with the labels, so that the accuracy of the first neural network model is improved. The trained first neural network model may be used to determine a likelihood of a purchase of an item by a user during a shopping session. At the same time, a second neural network model is trained. The second neural network model can be trained by using data such as buried point data, user characteristic data, commodity characteristic data and order data, and during training, the influence of time characteristics on the second neural network model is large, so that the weight can be set to be large, the second neural network model is trained, and the second neural network model can be determined as a model whether a user is interested in a commodity. Such as determining whether the user is hesitant to purchase an item or is interested in an item. The second probability output by the second neural network model is a high or low probability of interest.
And accessing the trained model into a recommendation system. In the process of browsing commodities after a user logs in, recording embedded point information of the user, acquiring user characteristics and commodity characteristics of the user and acquiring time characteristics and order characteristics, calling a first neural network model and a second neural network model every few seconds, and inputting the acquired data into the two models to obtain a first probability and a second probability. The predetermined first interval may be 0.8-1 and the second interval may be 0.6-1. If the first probability of the model output belongs to 0.8-1 and the second probability belongs to 0.6-1, it indicates that the probability of the user's purchase of the target commodity is high and the user is interested in the target commodity. In this case, a product list may be recommended, in which a product having a high similarity to the target product is included. If the first list is preset, the first list may be the current hot goods. According to the sales volume of the goods. And comparing each commodity in the first list with the target commodity according to the type to obtain the similarity, determining the commodity, the similarity of which to the target commodity in the first list is greater than a first threshold value, as the commodity in the commodity list, and pushing the commodity to the user. Meanwhile, each item in the item list corresponds to a target coupon, and the current value of the target coupon is the value with the highest possibility that the user purchases the item in the item list. The target coupon may be used if the user selects to purchase the item in the item list, and may not be used if the user purchases the target item.
Through the embodiment, the method achieves the effect of improving the recommendation accuracy of the commodities.
As an optional implementation manner, in a case where the first probability is within the first interval and the second probability is within the second interval, before pushing the item list to the target account, the method further includes:
acquiring a first list, wherein the first list comprises a plurality of commodities;
determining the similarity between each commodity in the first list and the target commodity;
and determining the commodities in the first list, the similarity of which to the target commodity is higher than a first threshold value, as the commodities in the commodity list.
The description continues with the above example. In the process of shopping by a user, if after current data of the user is input into the first neural network model and the second neural network model, the first probability and the second probability output by the models indicate that the purchase probability of the current commodity is high and the user is interested in the current commodity, a first list is obtained, the first list can be a preset list, and the commodity with the similarity to the current commodity being greater than a first threshold value in the commodities in the first list is determined as the commodity recommended by the user. If the current commodity is the red Fuji apple, determining the commodity in the commodity list in the first list, and then pushing the commodity list to the user if the yellow marshal apple is determined to be the apple in the commodity list.
According to the embodiment of the application, the method is used, and therefore the effect of improving the commodity recommendation accuracy is achieved.
As an optional implementation manner, in the case that the first probability is within the first interval and the second probability is within the second interval, pushing the item list to the target account includes:
and pushing the commodity list and the target coupon to the target account, wherein the target coupon is a coupon used when the target account purchases commodities in the commodity list.
The description continues with the above example. In the process of shopping by a user, if after current data of the user is input into the first neural network model and the second neural network model, the first probability and the second probability output by the models indicate that the purchase probability of the current commodity is high and the user is interested in the current commodity, a first list is obtained, the first list can be a preset list, and the commodity with the similarity to the current commodity being greater than a first threshold value in the commodities in the first list is determined as the commodity recommended by the user. If the current commodity is the red Fuji apple, determining the commodity in the commodity list in the first list, and then pushing the commodity list to the user if the yellow marshal apple is determined to be the apple in the commodity list.
Meanwhile, when yellow marshal apples are pushed, target coupons are also pushed. If the target coupon is 1.5 yuan less than yellow marshal apple every jin, red Fuji apples are not preferential. If the user purchases yellow marshal apples, the user can purchase the yellow marshal apples by subtracting 1.5 yuan per jin.
Through the embodiment, the method achieves the effect of improving the commodity recommendation accuracy.
As an optional implementation manner, before pushing the item list and the target coupon to the target account, the method further includes:
adjusting the value of the order coupon in the order characteristics in the target data characteristics to update the target data characteristics to obtain updated target data characteristics; inputting the updated target data characteristics into the first neural network model and the second neural network model to obtain a first probability and a second probability;
and determining the value of the order coupon under the condition that the first probability and the second probability are the maximum as the value of the target coupon.
The description is continued with reference to the above-described embodiments. The value of the target coupon needs to be determined on a case-by-case basis. If the value is too high, the goods will be lost and sold. If too low, the user's desire to purchase cannot be promoted. Therefore, the gear positions of different coupons can be preset, the numerical values of the coupons of different gear positions are input into the two models together as a sample, and if the weighted summation result of the first probability and the second probability output by the models is the maximum, the current gear position of the coupon is quite suitable.
Through the embodiment, the method achieves the effect of improving the commodity recommendation accuracy.
As an optional embodiment, before inputting the target data feature into the first neural network model, the method further comprises:
acquiring first sample data, wherein the first sample data comprises first data characteristics of a first account for browsing a first commodity, and the first data characteristics comprise behavior characteristics of the first account, account characteristics of the first account, commodity characteristics of the first commodity and order characteristics of the first commodity;
the first neural network model is trained using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold.
The description is continued with reference to the above-described embodiments. In the process of training the model, the first neural network model and the second neural network model may be trained by using data of browsed commodities of all users as samples. Specifically, the sample may be divided into different types of users, and then each type of user trains a first neural network model and a second neural network model, and when in use, which of the first neural network model and the second neural network model is used is determined according to the type of the current user.
Through the embodiment, the method achieves the effect of improving the commodity recommendation accuracy.
As an optional embodiment, before inputting the target data feature into the second neural network model, the method further comprises:
acquiring second sample data, wherein the second sample data comprises second data characteristics and time characteristics of a second commodity browsed by a second account, the second data characteristics comprise behavior characteristics of the second account, account characteristics of the second account, commodity characteristics of the second commodity and order characteristics of the second commodity, and the time characteristics comprise time consumed by the second account for browsing the second commodity;
training the second neural network model using the second sample data until the recognition accuracy of the second neural network model is greater than a third threshold.
The description is continued with reference to the above-described embodiments. In the process of training the model, the first neural network model and the second neural network model may be trained by using data of browsed commodities of all users as samples. Specifically, the sample may be divided into different types of users, and then each type of user trains a first neural network model and a second neural network model, and when in use, which of the first neural network model and the second neural network model is used is determined according to the type of the current user.
Through the embodiment, the method achieves the effect of improving the commodity recommendation accuracy.
It should be noted that, for simplicity of description, the above-mentioned method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present invention is not limited by the order of acts, as some steps may occur in other orders or concurrently in accordance with the invention. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required by the invention.
According to another aspect of the embodiment of the invention, a commodity recommending device for implementing the commodity recommending method is also provided. As shown in fig. 8, the apparatus includes:
a first obtaining unit 802, configured to obtain target data characteristics of a target commodity browsed by a target account in a client, where the target data characteristics include behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity, and order characteristics of the target commodity;
a first input unit 804, configured to input the target data feature into a first neural network model, where the first neural network model is a model for outputting a first probability, and the first probability is used to represent a probability that the target account purchases the target product;
a second input unit 806, configured to input the target data feature and the time feature into a second neural network model, where the second neural network model is a model for outputting a second probability, the second probability is used to represent a probability that the target account purchases the target product in advance, and the time feature is time used by the target account to browse the target product;
and a pushing unit 808, configured to push a commodity list to the target account if the first probability is within the first interval and the second probability is within the second interval, where the commodity list includes commodities associated with the target commodity.
Alternatively, the device can be applied to any network shopping website or small program or application. The target account can be an account used by the user in the shopping process, the account can be a long-term account registered by the user or a temporary account, and the temporary account can be purchased and paid by binding other applications or jumping to other applications. When a user browses a target commodity, the target data characteristic of the target commodity browsed by the user can be obtained, and the time characteristic can be time consumed by the user for browsing the target commodity. After the target data characteristics and the time characteristics are obtained, the target data characteristics are input into a first neural network model, the target data characteristics and the time characteristics are input into a second neural network model, the first neural network model outputs a first probability, the first probability can be the probability that a user will purchase a target commodity, the second model outputs a second probability, and the second probability can be the probability that the user pre-purchases the target commodity. And if the first probability is in the first interval and the second probability is in the second interval, pushing a commodity list to the user, wherein commodities in the commodity list are associated with the target commodity. By the method, the effect of improving the accuracy of recommending commodities is achieved.
As an alternative embodiment, as shown in fig. 9, the apparatus further includes:
a second obtaining unit 902, configured to obtain a first list before pushing a commodity list to the target account when the first probability is within a first interval and the second probability is within a second interval, where the first list includes a plurality of commodities;
a first determining unit 904, configured to determine a similarity between each product in the first list and the target product;
a second determining unit 906, configured to determine, as the product in the product list, a product in the first list, for which the similarity to the target product is higher than the first threshold.
As an alternative embodiment, as shown in fig. 10, the pushing unit 808 includes:
the pushing module 1002 is configured to push a commodity list and a target coupon to a target account, where the target coupon is a coupon used when the target account purchases a commodity in the commodity list.
As an optional implementation manner, the pushing unit further includes:
the adjusting module is used for adjusting the numerical value of the order coupon in the order characteristics in the target data characteristics before pushing the commodity list and the target coupon to the target account so as to update the target data characteristics and obtain the updated target data characteristics; inputting the updated target data characteristics into the first neural network model and the second neural network model to obtain a first probability and a second probability;
and the determining module is used for determining the value of the order coupon under the condition that the first probability and the second probability are the maximum as the value of the target coupon.
As an optional implementation, the apparatus further includes:
the third acquiring unit is used for acquiring first sample data before inputting the target data characteristics into the first neural network model, wherein the first sample data comprises first data characteristics of a first account for browsing a first commodity, and the first data characteristics comprise behavior characteristics of the first account, account characteristics of the first account, commodity characteristics of the first commodity and order characteristics of the first commodity;
a first training unit for training the first neural network model using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold.
As an optional implementation, the apparatus further includes:
a fourth obtaining unit, configured to obtain second sample data before inputting the target data feature into the second neural network model, where the second sample data includes a second data feature and a time feature of browsing the second commodity by the second account, the second data feature includes a behavior feature of the second account, an account feature of the second account, a commodity feature of the second commodity, and an order feature of the second commodity, and the time feature includes time consumed by browsing the second commodity by the second account;
and the second training unit is used for training the second neural network model by using the second sample data until the identification accuracy of the second neural network model is more than a third threshold value.
The embodiments in the embodiments of the present application may refer to the embodiments in the above embodiments, which are not described herein again.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
The integrated unit in the above embodiments, if implemented in the form of a software functional unit and sold or used as a separate product, may be stored in the above computer-readable storage medium. Based on such understanding, the technical solution of the present invention may be substantially or partially implemented in the prior art, or all or part of the technical solution may be embodied in the form of a software product stored in a storage medium, and including instructions for causing one or more computer devices (which may be personal computers, servers, or network devices) to execute all or part of the steps of the method according to the embodiments of the present invention.
In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed client may be implemented in other manners. The above-described embodiments of the apparatus are merely illustrative, and for example, a division of a unit is merely a division of a logic function, and an actual implementation may have another division, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A method for recommending an article, comprising:
acquiring target data characteristics of a target account for browsing a target commodity in a client, wherein the target data characteristics comprise behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity and order characteristics of the target commodity;
inputting the target data feature into a first neural network model, wherein the first neural network model is a model for outputting a first probability representing a probability of the target account number purchasing the target commodity;
inputting the target data characteristic and a time characteristic into a second neural network model, wherein the second neural network model is used for outputting a second probability, the second probability is used for representing the probability that the target account pre-purchases the target commodity, and the time characteristic is the time used by the target account for browsing the target commodity;
and under the condition that the first probability is within a first interval and the second probability is within a second interval, pushing a commodity list to the target account, wherein the commodity list comprises commodities related to the target commodity.
2. The method of claim 1, wherein in the event that the first probability is within a first interval and the second probability is within a second interval, prior to pushing the list of items to the target account number, the method further comprises:
acquiring a first list, wherein the first list comprises a plurality of commodities;
determining the similarity between each commodity in the first list and the target commodity;
determining the goods in the first list, of which the similarity with the target goods is higher than a first threshold value, as the goods in the goods list.
3. The method of claim 1, wherein the pushing a list of items to the target account number with the first probability within a first interval and the second probability within a second interval comprises:
and pushing the commodity list and a target coupon to the target account, wherein the target coupon is a coupon used when the target account purchases commodities in the commodity list.
4. The method of claim 3, wherein prior to pushing the list of items and target coupon to the target account number, the method further comprises:
adjusting the value of the order coupon in the order characteristics in the target data characteristics to update the target data characteristics to obtain the updated target data characteristics; inputting the updated target data features into the first neural network model and the second neural network model to obtain the first probability and the second probability;
and determining the value of the order coupon with the maximum first probability and the maximum second probability as the value of the target coupon.
5. The method of any one of claims 1 to 4, wherein prior to inputting the target data feature into the first neural network model, the method further comprises:
acquiring first sample data, wherein the first sample data comprises first data characteristics of a first account for browsing a first commodity, and the first data characteristics comprise behavior characteristics of the first account, account characteristics of the first account, commodity characteristics of the first commodity and order characteristics of the first commodity;
training the first neural network model using the first sample data until the recognition accuracy of the first neural network model is greater than a second threshold.
6. The method of any one of claims 1 to 4, wherein prior to inputting the target data feature into the second neural network model, the method further comprises:
acquiring second sample data, wherein the second sample data comprises second data characteristics and time characteristics of a second commodity browsed by a second account, the second data characteristics comprise behavior characteristics of the second account, account characteristics of the second account, commodity characteristics of the second commodity and order characteristics of the second commodity, and the time characteristics comprise time consumed by the second account for browsing the second commodity;
training the second neural network model using the second sample data until the recognition accuracy of the second neural network model is greater than a third threshold.
7. An article recommendation device, comprising:
the system comprises a first acquisition unit, a second acquisition unit and a third acquisition unit, wherein the first acquisition unit is used for acquiring target data characteristics of a target commodity browsed by a target account in a client, and the target data characteristics comprise behavior characteristics of the target account, account characteristics of the target account, commodity characteristics of the target commodity and order characteristics of the target commodity;
a first input unit, configured to input the target data feature into a first neural network model, where the first neural network model is a model for outputting a first probability, and the first probability is used to represent a probability that the target account purchases the target commodity;
a second input unit, configured to input the target data feature and a time feature into a second neural network model, where the second neural network model is a model for outputting a second probability, the second probability is used to represent a probability that the target account pre-purchases the target product, and the time feature is a time used by the target account to browse the target product;
a pushing unit, configured to push a commodity list to the target account if the first probability is within a first interval and the second probability is within a second interval, where the commodity list includes commodities associated with the target commodity.
8. The apparatus of claim 7, further comprising:
a second obtaining unit, configured to obtain a first list before pushing the commodity list to the target account if the first probability is within a first interval and the second probability is within a second interval, where the first list includes a plurality of commodities;
a first determining unit, configured to determine a similarity between each product in the first list and the target product;
a second determining unit, configured to determine, as a product in the product list, a product in the first list, for which the similarity with the target product is higher than a first threshold.
9. The apparatus of claim 7, wherein the pushing unit comprises:
and the pushing module is used for pushing the commodity list and a target coupon to the target account, wherein the target coupon is a coupon used when the target account purchases commodities in the commodity list.
10. The apparatus of claim 9, wherein the pushing unit further comprises:
the adjusting module is used for adjusting the numerical value of the order coupon in the order characteristics in the target data characteristics before the commodity list and the target coupon are pushed to the target account, so as to update the target data characteristics and obtain the updated target data characteristics; inputting the updated target data features into the first neural network model and the second neural network model to obtain the first probability and the second probability;
and the determining module is used for determining the numerical value of the order coupon under the condition that the first probability and the second probability are maximum as the numerical value of the target coupon.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010949448.3A CN112232896A (en) | 2020-09-10 | 2020-09-10 | Commodity recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010949448.3A CN112232896A (en) | 2020-09-10 | 2020-09-10 | Commodity recommendation method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112232896A true CN112232896A (en) | 2021-01-15 |
Family
ID=74115626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010949448.3A Withdrawn CN112232896A (en) | 2020-09-10 | 2020-09-10 | Commodity recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112232896A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113705782A (en) * | 2021-08-18 | 2021-11-26 | 上海明略人工智能(集团)有限公司 | Model training method and device for media data recommendation |
-
2020
- 2020-09-10 CN CN202010949448.3A patent/CN112232896A/en not_active Withdrawn
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113705782A (en) * | 2021-08-18 | 2021-11-26 | 上海明略人工智能(集团)有限公司 | Model training method and device for media data recommendation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110827114A (en) | Commodity recommendation method and device | |
| US20110282821A1 (en) | Further Improvements in Recommendation Systems | |
| US20020065797A1 (en) | System, method and computer program for automated collaborative filtering of user data | |
| TWI543095B (en) | Information processing apparatus, information processing method, information processing program product and recording medium with information processing program | |
| JP5957164B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN112435067A (en) | Intelligent advertisement putting method and system for cross-e-commerce platform and social platform | |
| JP6976207B2 (en) | Information processing equipment, information processing methods, and programs | |
| KR101369033B1 (en) | Method for regular seller management service commodities in the electronic commerce and computer readable record medium on which a program therefor is recorded | |
| JP6758450B1 (en) | Information processing equipment, information processing methods, and information processing programs | |
| CN110933472B (en) | Method and device for realizing video recommendation | |
| JP5824602B1 (en) | Information processing apparatus, information processing method, and information processing program | |
| US20220414579A1 (en) | Salesperson evaluation apparatus, salesperson evaluation method, and salesperson evaluation program | |
| CN111767458A (en) | Information pushing method, device, system and storage medium | |
| JP6576043B2 (en) | Product demand forecast system | |
| CN111461827A (en) | Product evaluation information pushing method and device | |
| JP6692156B2 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN114463089A (en) | Method and device for improving shop visit degree | |
| KR100891930B1 (en) | Method, system and computer-readable recording medium for providing popular goods by using evaluation data on goods | |
| CN112232896A (en) | Commodity recommendation method and device | |
| JP6424194B2 (en) | Information processing apparatus, information processing method, and information processing program | |
| CN113254780A (en) | Information processing method and device, electronic equipment and computer storage medium | |
| KR20140079752A (en) | Method and apparatus for providing shopping information | |
| JP4649525B1 (en) | RECOMMENDATION DEVICE, RECOMMENDATION METHOD, AND RECOMMENDATION PROGRAM | |
| CN110781399A (en) | Cross-platform information pushing method and device | |
| CN109829797A (en) | Method for pushing, terminal device, storage medium based on market demand analysis |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20210115 |
|
| WW01 | Invention patent application withdrawn after publication |