CN113744002A - Method, device, equipment and computer readable medium for pushing information - Google Patents
Method, device, equipment and computer readable medium for pushing information Download PDFInfo
- Publication number
- CN113744002A CN113744002A CN202010462528.6A CN202010462528A CN113744002A CN 113744002 A CN113744002 A CN 113744002A CN 202010462528 A CN202010462528 A CN 202010462528A CN 113744002 A CN113744002 A CN 113744002A
- Authority
- CN
- China
- Prior art keywords
- commodity
- user
- information
- recommendation information
- pushing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000007246 mechanism Effects 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims description 38
- 238000004590 computer program Methods 0.000 claims description 9
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000006870 function Effects 0.000 description 21
- 238000010586 diagram Methods 0.000 description 18
- 230000006399 behavior Effects 0.000 description 14
- 230000008569 process Effects 0.000 description 9
- 238000012552 review Methods 0.000 description 8
- 230000009471 action Effects 0.000 description 7
- 238000004891 communication Methods 0.000 description 7
- 238000012545 processing Methods 0.000 description 7
- 239000013598 vector Substances 0.000 description 7
- 230000000694 effects Effects 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 5
- 238000003058 natural language processing Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 235000009508 confectionery Nutrition 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 239000000835 fiber Substances 0.000 description 2
- JEIPFZHSYJVQDO-UHFFFAOYSA-N iron(III) oxide Inorganic materials O=[Fe]O[Fe]=O JEIPFZHSYJVQDO-UHFFFAOYSA-N 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 241000191291 Abies alba Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000003915 air pollution Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
Abstract
The invention discloses a method, a device, equipment and a computer readable medium for pushing information, and relates to the technical field of computers. One embodiment of the method comprises: acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information; generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words; and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user. The method and the device have different preferences for different users, and improve the adaptability of the recommendation information to the users.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, and a computer-readable medium for pushing information.
Background
With the development of big data and the internet, more and more users enjoy online shopping. For the e-commerce platform, how to push information and attract more users to click commodities is a significant matter.
In a business scene, the information is handwritten by operators and then matched with commodities. In order to grasp the shopping psychology of the user, the operator needs to spend much effort, and not only the diversity of the information but also the excellent aesthetic feeling of the information needs to be considered.
In the process of implementing the invention, the inventor finds that at least the following problems exist in the prior art: with the increase of commodities, operators are required to spend more time and energy to write out various information suitable for marketing scenarios so as to push the information to users. However, different users have different preferences, and the recommendation information is poorly adapted to the users.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method, an apparatus, a device, and a computer-readable medium for pushing information, which improve the adaptability between recommended information and users according to different preferences of different users.
To achieve the above object, according to an aspect of the embodiments of the present invention, there is provided a method for pushing information, including:
acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information;
generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words;
and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
Before obtaining recommendation information of a commodity according to a search of the commodity in recommendation information of the user under the condition that the characteristic information is matched with the commodity, the method further comprises the following steps:
if the commodity corresponding to the characteristic information is the same as the commodity, matching the characteristic information with the commodity;
or the like, or, alternatively,
and if the commodity application scene corresponding to the characteristic information is the same as the commodity application scene of the commodity, matching the characteristic information with the commodity.
The pushing of the recommendation information of the commodity to the user includes:
under the condition that the characteristic information is matched with the commodity, searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity;
and taking the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
The searching for recommendation information with the maximum similarity in the recommendation information of the user according to the commodity comprises the following steps:
searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity attribute words of the commodities;
or the like, or, alternatively,
and searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity attribute words of the commodities and the using times of the recommendation information.
Before generating the recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words, the method further includes:
configuring the weight of the commodity attribute training words and the weight of the user attribute training words;
and updating a loss function by using the weight of the commodity attribute training word and the weight of the user attribute training word and adopting a mask matrix, and training to obtain the sequence-to-sequence model based on the attention mechanism.
According to a second aspect of the embodiments of the present invention, there is provided an apparatus for pushing information, including:
the information module is used for acquiring user attribute words based on the characteristic information of the user and extracting commodity attribute words from the commodity information;
the generation module is used for generating the recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words;
and the pushing module is used for searching and obtaining the recommendation information of the commodity in the recommendation information of the user according to the commodity under the condition that the characteristic information is matched with the commodity, and pushing the recommendation information of the commodity to the user.
The pushing module is specifically used for matching the characteristic information with the commodity if the commodity corresponding to the characteristic information is the same as the commodity;
or the like, or, alternatively,
and if the commodity application scene corresponding to the characteristic information is the same as the commodity application scene of the commodity, matching the characteristic information with the commodity.
The pushing module is specifically configured to search recommendation information with the largest similarity in the recommendation information of the user according to the commodity under the condition that the feature information is matched with the commodity;
and taking the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
According to a third aspect of the embodiments of the present invention, there is provided an electronic device for pushing information, including:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method as described above.
According to a fourth aspect of embodiments of the present invention, there is provided a computer readable medium, on which a computer program is stored, which when executed by a processor, implements the method as described above.
One embodiment of the above invention has the following advantages or benefits: acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information; generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words; and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
Because the recommendation information is generated by combining the characteristic information of the user, different preferences are provided for different users, the recommendation information conforms to the preferences of the user, and the adaptability of the recommendation information and the user can be improved.
Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
fig. 1 is a schematic diagram of a main flow of a method of pushing information according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of the composition of feature information of a user according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of the composition of merchandise information according to an embodiment of the invention;
FIG. 4 is a schematic flow diagram of a sequence-to-sequence model for training an attention-based mechanism in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of generating recommendation information for a user, according to an embodiment of the present invention;
FIG. 6 is a schematic flow chart illustrating a process of pushing recommendation information to a user according to an embodiment of the present invention;
fig. 7 is a schematic diagram of a main structure of an apparatus for pushing information according to an embodiment of the present invention;
FIG. 8 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;
fig. 9 is a schematic structural diagram of a computer system suitable for implementing a terminal device or a server according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Currently, recommendation information sent to a user is handwritten by an operator to match a corresponding commodity. With the increase of the types of commodities, operators need to spend more time and energy to write various recommendation information suitable for marketing scenes, and the labor cost is high. In the context of big data marketing, it is increasingly important to send recommendation information to users.
The user preferences are different, so that the user characteristics are different, and the same commodity can meet the actual requirements of different users from multiple angles. However, the same information is sent to different users, and the method cannot adapt to the diversification of the user requirements, so that the suitability of the recommendation information and the users is poor.
In order to solve the technical problem of poor adaptability of recommendation information and a user, the following technical scheme in the embodiment of the invention can be adopted.
Referring to fig. 1, fig. 1 is a schematic diagram of a main flow of a method for pushing information according to an embodiment of the present invention, and recommendation information of a user is generated by combining a user attribute word with a commodity attribute word and using a sequence-to-sequence model based on an attention mechanism, so that recommendation information applicable to the user can be pushed to the user. As shown in fig. 1, the method specifically comprises the following steps:
s101, acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information.
In the embodiment of the invention, the user can be identified by the characteristic information. It is understood that the characteristic information of the user is information for characterizing the user.
Referring to fig. 2, fig. 2 is a schematic composition diagram of feature information of a user according to an embodiment of the present invention, where the feature information of the user includes natural feature information, geographical information of the user, and user behavior.
In one embodiment of the invention, the characteristic information of the user comprises one or more of natural characteristic information, geographical information of the user and user behaviors.
The natural feature information is feature information describing a natural state of the user. In particular, the natural characteristic information may include, from the demographic attributes, a user gender, a user age, and a user occupation; from the social attribute, the marital status of the user and whether the user is a campus user; from a family perspective, it may include whether the user lives in the country, whether there are children, and the gender of the children; from an asset perspective, this may include whether or not there is a car and whether or not there is a house.
It is understood that the natural characteristic information includes one or more of the following: user gender, user age, user occupation, user marital status, whether the user is a campus user, whether the user lives in the country, whether there are children, child gender, whether there are cars, and whether there are houses.
The geographic information is information describing the shipping address of the user. As one example, the geographic information includes a receiving district county that the user commonly uses within a year, a receiving city that the user commonly uses within a year, and a receiving province that the user commonly uses within a year.
User behavior is the specific action that describes the user on the relevant website. As one example, the user behavior may be a user's browsing records and purchasing records at a shopping website. Illustratively, user behavior may include user liveness, user value, user actions, and user cycles.
User activity is the activity level that describes the activity level of a user at a relevant website. As an example, the activity of a user at a certain website is counted for nearly 1, 3, 5, 7, 14 and 30 days respectively.
The user value is a value describing the user's presence at the relevant website. As one example, the user value includes a length of time that the user has logged on to the associated network.
The user action is an action that describes the user at the relevant website. As one example, the user actions include one or more of: click, buy, follow, join a shopping cart, search, share, etc.
A user cycle is a time period involved in describing a user action. As an example, the user cycle includes one or more of: the time period that the user is a member, the time period that the user opens other membership, the time period that the user normally pays, the life cycle of the user in the related network and the life cycle of the user in each type of commodity.
The characteristic information of the user comprises one or more of natural characteristic information, geographical information of the user and user behaviors. It is understood that the user can be portrayed in accordance with the user's characteristic information.
In particular, the user may be characterized by words in the characteristic information of the user. The characteristic information of the user comprises a plurality of words, and each word describes the user from different angles and aspects. Words that appropriately describe the user may be selected from the above words as the user attribute words. The user attribute words are words characterizing the characteristics of the user.
In one embodiment of the invention, a list of user attribute words may be established for each user. The items of the user attribute words are the same for each user. As an example, the items of the user attribute word list include: sex; age; whether there is a child or not; the province is that; whether it is a member. According to the characteristic information of the users, a user attribute word list of each user can be automatically generated.
In this way, the user attribute words can be selected according to the feature information of the user according to the preset conditions. As an example, the preset conditions include: gender and age are taken as user attribute words.
In an embodiment of the present invention, for a new user or a user without any user behavior, there may be a case where there is no corresponding user feature information, and a default user attribute word may be set for the user. As an example, default user attribute words include: a woman; age 25.
In the embodiment of the invention, the commodity is displayed on the website, the user browses the commodity in the website, and the user purchases the commodity under the condition that the commodity can meet the requirement of the user. Then, merchandise information is required to be displayed on the website, and the merchandise information is used for characterizing the merchandise.
Referring to fig. 3, fig. 3 is a schematic composition diagram of commodity information according to an embodiment of the present invention, wherein the commodity information includes purchase information, buyer comments, and commodity behavior. In one embodiment of the invention, the merchandise information includes one or more of purchase information, buyer reviews, and merchandise behavior.
The purchase information is information relating to the characterization of the purchased article. The purchase information may include one or more of whether the commodity is a new commodity, whether the commodity is directly sold by a manufacturer, whether the commodity is self-operated, whether the commodity is purchased globally, whether the commodity is picked up by a store, and whether the commodity is a member commodity.
Buyer reviews are reviews of an item by users who have purchased the item. Buyer reviews may include user good reviews, user medium reviews, or user bad reviews.
The commodity behavior is user behavior that the commodity receives over a period of time. As one example, the commodity behavior includes clicks, concerns, purchases, orders, shares, searches, and the like within 5 minutes, 10 minutes, 1 hour, and 6 hours.
In the embodiment of the invention, the existing commodity attribute model can be utilized to extract the commodity attribute words from the commodity information. The commodity attribute model is a model obtained by utilizing commodity information training.
And S102, generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words.
The commodity attribute words may describe commodities, and the user attribute words may describe users. And combining the commodity attribute words with the user attribute words to generate recommendation information for the user. The recommendation information is a character which enables the commodity to better obtain the cognition of the user and more effectively conveys the commodity value to the user.
Specifically, recommendation information for the user is generated by a sequence-to-sequence model based on an attention mechanism in combination with the user attribute words on the basis of the product attribute words.
Sequence-to-sequence models based on the Attention mechanism include a sequence-to-sequence (seq 2seq) Model and an Attention Model.
The seq2seq model includes an encoding network (encoder network) and a decoding network (decoder network). encoder network, which is a Recurrent Neural Network (RNN) architecture. The RNN network outputs a vector representing the input sequence. decoder network, which takes the output of the encode network as input and outputs one translated word at a time.
Illustratively, after receiving the product attribute words and the user attribute words, the words may be converted into vectors using word2 vec. The vector is input into the RNN network of the encoder network. word2vec is a tool for converting words into vector form, and can reduce the processing of text content into vector operation in vector space.
decoder network consists of the output and input text of the encoder network. After the representation of the encoder network is obtained, candidate generated text symbols are obtained through an activation function and a softmax layer, the text symbols with the maximum probability are screened out, and finally a sequence, namely recommendation information, is output.
Considering that the input words whose seq2seq model is closer to the first cell of the decoder network have a greater impact on decoding, this is not the most reasonable. On the basis of a seq2seq Model, an Attention Model is adopted, the importance of each word in an input sequence is detected for each cell output by a decoder network, and the word which has a large influence on the current text is concerned more.
It should be noted that the sequence-to-sequence model based on the attention mechanism needs to be trained. In the process of training the attention-based sequence-to-sequence model, a loss function is adopted to determine whether the attention-based sequence-to-sequence model completes training. The loss function is used for estimating the degree of inconsistency between the predicted value and the actual value of the model, and is a non-negative value function. The smaller the loss function, the better the robustness of the model.
Referring to fig. 4, fig. 4 is a schematic flowchart of a process of training a sequence-to-sequence model based on an attention mechanism according to an embodiment of the present invention, which specifically includes:
s401, configuring the weight of the commodity attribute training words and the weight of the user attribute training words.
In the embodiment of the invention, the weight of the commodity attribute training word and the weight of the user attribute training word are randomly initialized. That is, the weight of the product attribute training word and the weight of the user attribute training word are both random values in the initial state. In a non-initial state, the weights of the commodity attribute training words and the weights of the user attribute training words are updated along with iteration of the model.
S402, updating a loss function by using the weight of the commodity attribute training words and the weight of the user attribute training words and adopting a mask matrix, and training to obtain a sequence-to-sequence model based on an attention mechanism.
On the basis of the current loss function, updating the initial loss function by using a mask matrix by using the weight of the commodity attribute training word and the weight of the user attribute training word, namely: the updated loss function is equal to the product of the weight of the commodity attribute training word and the weight of the user attribute training word, the mask matrix and the current loss function.
With the continuous iteration of the sequence-to-sequence model based on the attention mechanism, the weights of the commodity attribute training words and the weights of the user attribute training words are updated all the time in order to achieve the goal of reducing the loss function.
In Natural Language Processing (NLP), a common problem is that the input sequences are not of equal length. Therefore, the sentence with shorter length is filled with 0 at the end of the sentence. The mask matrix is used to indicate which are real data, 0, filled.
In the embodiment of fig. 4, the loss function is updated with a mask matrix, so that the loss at fill 0 does not need to be calculated during the pass back of the loss function. Thereby reducing the impact of the data at pad 0 on the attention-based sequence-to-sequence model.
In embodiments of the present invention, attention-based sequence-to-sequence models that complete training may also be evaluated.
Illustratively, the evaluation index is divided into a manual evaluation and an automatic evaluation. Manual evaluation: syntactic correctness (1-10 points), content relevance (1-10 points), and poetry richness (1-10 points). Automatic evaluation: perplexity and Rouge Score are indexed using text-generating algorithms.
Perplexity represents the degree of confusion, and is a score calculated to generate a candidate set of word sentences based on the reference sentence learning language model P. And normalizing the calculated scores based on the sentence lengths, wherein the lower the confusion degree is, the better the quality of the generated text is.
The Rouge Score is used to compare the degree of overlap of n-grams in the candidate set and standard answers, with higher degrees of overlap being considered as higher quality text generated, primarily evaluated from a recall perspective. Where n-gram is a very important concept in natural language processing, usually in NLP, based on corpus, a sentence can be predicted or evaluated whether it is reasonable or not by using n-gram.
In addition, the tester will construct a regression test set before using the attention-based sequence-to-sequence model. Such as: certain words are input, a threshold is set, and a determination is made as to whether the text generated based on the sequence-to-sequence model of the attention mechanism is reliable. If the recommendation information in the regression test set that exceeds the set threshold is satisfactory, the recommendation information for the user may be generated using a sequence-to-sequence model based on an attention mechanism.
Data services may also be coordinated in using a sequence-to-sequence model based on an attention mechanism. To ensure stable service performance; feedback effects to improve the attention-based sequence-to-sequence model.
Specifically, the degree to which the tester or the online user finds that the push information matches the user is low, and the push information can be fed back to the tester. From a model perspective and a data perspective, a tester can analyze to determine which dimension causes a problem, and then clean the data or improve the model to continuously optimize iteration.
Referring to fig. 5, fig. 5 is a schematic diagram of generating recommendation information for a user according to an embodiment of the present invention.
In fig. 5, the item attribute word: one commodity class: apparel, one is a product identifier (SKU): 12345. user attribute words: specialization and literature.
And inputting the commodity attribute words and the user attribute words into a sequence pair sequence and an attention mechanism model, and outputting corresponding recommendation information.
In the embodiment of the invention, because the characteristic information of each user is different, the recommendation information belonging to the user is generated according to the characteristic information of the user and the commodity information. It can be understood that each user has corresponding recommendation information, so that in the subsequent information pushing process, pushed recommendation information can be selected from the recommendation information of the user.
S103, under the condition that the characteristic information is matched with the commodity, the recommendation information of the commodity is searched and obtained from the recommendation information of the user according to the commodity, and the recommendation information of the commodity is pushed to the user.
And recording the user behavior in the process of browsing the commodities by the user. When the characteristic information of the user is determined to be matched with the commodity, the situation that the opportunity of pushing the recommendation information to the user is reached is explained, the recommendation information is pushed to the user, and the possibility that the user purchases the commodity is high.
In the embodiment of the present invention, whether the characteristic information of the user matches the commodity may be determined in the following manner.
In a first mode
And if the commodity corresponding to the characteristic information of the user is the same as the commodity, matching the characteristic information of the user with the commodity.
The characteristic information of the user comprises one or more of natural characteristic information, geographical information of the user and user behaviors. The commodity corresponding to the characteristic information of the user is a commodity satisfying the characteristic information of the user.
As an example, the characteristic information of the user includes: sex: female, child gender: a woman. The article corresponding to the characteristic information of the user may be a woman's garment. As another example, the characteristic information of the user includes: and (4) browsing records: browsing the commodity 1 for 10 times; and (4) purchase record: item 1 was purchased 2 times. The product corresponding to the characteristic information of the user may be product 1.
The characteristic information of the user includes various information, and accordingly, the commodity corresponding to the characteristic information of the user also has various kinds.
Matching of the characteristic information of the user with the commodity is exemplarily described below with reference to a specific application scenario.
Application scenario one
The characteristic information of the user includes star a liked by the user. In the case of discount sales promotion of the star a introduction commodity, the characteristic information of the user is matched with the star a introduction commodity. The recommendation information of the commodity is displayed to the user, and the possibility that the user purchases the star a introduction commodity is high.
Application scenario two
The characteristic information of the user comprises information of commodities in the shopping cart, and the commodities corresponding to the characteristic information of the user comprise the commodities in the shopping cart. The commodity of the shopping cart participates in promotion, namely the characteristic information of the user is matched with the commodity of the shopping cart, the recommendation information of the commodity is displayed to the user, and the possibility that the user orders to purchase the commodity of the shopping cart is high.
Mode two
And if the commodity application scene corresponding to the characteristic information of the user is the same as the commodity application scene of the commodity, matching the characteristic information of the user with the commodity.
The commodity application scene is a specific scene used by the commodity. As an example, the commodity application scenario for an umbrella is: raining; the application scenes of the Christmas tree are as follows: christmas.
The commodity application scenario corresponding to the feature information of the user is the same as the commodity application of the commodity, which indicates that the possibility of purchasing the commodity by the user is higher.
As an example, if the application scene of the commodity corresponding to the feature information of the user is snowy days, the commodity is: snow boots, matched with the characteristic information of the user.
As an example, if the application scenario of the commodity corresponding to the feature information of the user is air pollution, the commodity: the mask is matched with the characteristic information of the user.
As an example, if the commodity application scenario corresponding to the feature information of the user is a birthday, the commodity: and the birthday cake is matched with the characteristic information of the user.
And if the characteristic information is matched with the commodity, the recommended information pushed to the user can be obtained by searching in the recommended information of the user according to the commodity.
Referring to fig. 6, fig. 6 is a schematic flowchart of a process of pushing recommendation information to a user according to an embodiment of the present invention, which specifically includes:
s601, under the condition that the characteristic information is matched with the commodity, searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity.
In the case where the characteristic information matches the commodity, the description may push recommendation information to the user. For the user, there are a plurality of pieces of generated recommendation information. It is understood that there are a plurality of generated commodity files for each registered user.
The recommendation information with the largest similarity can be searched in the recommendation information of the user based on the commodities. Specifically, each piece of recommendation information has a corresponding user attribute word and a corresponding commodity attribute word.
As an example, the user attribute word and the product attribute word of the recommendation information may be generated as the user attribute word and the product attribute word of the recommendation information. Such as: adopting user attribute words: female and click, commodity attribute word: purchased candy a and buyer reviews: and (4) well eating the candy A, and using the user attribute words and the commodity attribute words as the user attribute words and the commodity attribute words of the candy A recommendation information.
In an embodiment of the present invention, recommendation information with the largest similarity may be searched for in the recommendation information of the user according to the product attribute words of the product.
Specifically, the commodity attribute words of the commodity can be represented as vectors, and the recommendation information of the user has corresponding user attribute words and commodity attribute words. And calculating the similarity of the commodity attribute words of the commodity, the user attribute words corresponding to the recommendation information of the user and the commodity attribute words, wherein the similarity can be calculated by adopting cosine similarity, a Pearson correlation coefficient or Euclidean distance.
Then, the recommendation information with the largest similarity is searched in the calculated similarities.
In one embodiment of the invention, recommendation information with the maximum similarity is searched in the recommendation information of the user according to the commodity attribute words of the commodities and the using times of the recommendation information.
And searching recommendation information with the maximum similarity by combining the using times of the recommendation information on the basis of the commodity attribute words of the commodities. The commodity attribute words of the commodities can be used firstly, and the similarity between the user attribute words and the commodity attribute words corresponding to the recommendation information of the user can be calculated.
Then, the first N pieces of recommendation information are selected according to the sequence of the similarity from large to small, and the recommendation information with the largest use frequency in the N pieces of recommendation information is taken as the recommendation information with the largest similarity. Wherein N is a preset parameter greater than 0.
That is, the recommendation information having the greatest similarity is related not only to the attribute words of the product but also to the number of times of use of the recommendation information. The more the number of times of use of the recommendation information is, the more likely it is to be the recommendation information having the greatest similarity.
And S602, using the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
The recommendation information with the maximum similarity can be used as recommendation information for pushing the commodity to the user aiming at the advantage that the user fully represents the commodity.
In the embodiment of fig. 6, recommendation information with the largest similarity is pushed to the user, and the recommendation information can clearly describe characteristics of a commodity which conforms to the user, so that the user is prompted to purchase the commodity.
In the embodiment, the user attribute words are obtained based on the characteristic information of the user, and the commodity attribute words are extracted from the commodity information; generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words; and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
Because the recommendation information is generated by combining the characteristic information of the user, different preferences are provided for different users, the recommendation information conforms to the preferences of the user, and the adaptability of the recommendation information and the user can be improved.
Referring to fig. 7, fig. 7 is a schematic diagram of a main structure of an information pushing apparatus according to an embodiment of the present invention, where the information pushing apparatus may implement the information pushing method, and as shown in fig. 7, the information pushing apparatus specifically includes:
the information module 701 is configured to obtain a user attribute word based on the feature information of the user, and extract a product attribute word from the product information.
A generating module 702, configured to generate recommendation information of a user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words.
The pushing module 703 is configured to, under the condition that the feature information is matched with the commodity, obtain recommendation information of the commodity by searching for the recommendation information of the commodity in the user according to the commodity, and push the recommendation information of the commodity to the user.
In an embodiment of the present invention, the pushing module 703 is specifically configured to match the characteristic information with the commodity if the commodity corresponding to the characteristic information is the same as the commodity;
or the like, or, alternatively,
and matching the characteristic information with the commodity if the commodity application scene corresponding to the characteristic information is the same as the commodity application scene of the commodity.
In an embodiment of the present invention, the pushing module 703 is specifically configured to search recommendation information with the largest similarity in the recommendation information of the user according to the commodity under the condition that the feature information is matched with the commodity;
and taking the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
In an embodiment of the present invention, the pushing module 703 is specifically configured to search recommendation information with the largest similarity in the recommendation information of the user according to the commodity attribute words of the commodities;
or the like, or, alternatively,
and searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity attribute words of the commodities and the using times of the recommendation information.
In an embodiment of the present invention, the generating module 702 is specifically configured to configure the weight of the commodity attribute training word and the weight of the user attribute training word;
and updating a loss function by using the weight of the commodity attribute training word and the weight of the user attribute training word and adopting a mask matrix, and training to obtain a sequence-to-sequence model based on an attention mechanism.
Fig. 8 shows an exemplary system architecture 800 of a method or apparatus for pushing information to which embodiments of the present invention may be applied.
As shown in fig. 8, the system architecture 800 may include terminal devices 801, 802, 803, a network 804, and a server 805. The network 804 serves to provide a medium for communication links between the terminal devices 801, 802, 803 and the server 805. Network 804 may include various types of connections, such as wire, wireless communication links, or fiber optic cables, to name a few.
A user may use the terminal devices 801, 802, 803 to interact with a server 805 over a network 804 to receive or send messages or the like. The terminal devices 801, 802, 803 may have installed thereon various communication client applications, such as shopping-like applications, web browser applications, search-like applications, instant messaging tools, mailbox clients, social platform software, etc. (by way of example only).
The terminal devices 801, 802, 803 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The server 805 may be a server that provides various services, such as a back-office management server (for example only) that supports shopping-like websites browsed by users using the terminal devices 801, 802, 803. The backend management server may analyze and perform other processing on the received data such as the product information query request, and feed back a processing result (for example, target push information, product information — just an example) to the terminal device.
It should be noted that the method for pushing information provided by the embodiment of the present invention is generally executed by the server 805, and accordingly, the apparatus for pushing information is generally disposed in the server 805.
It should be understood that the number of terminal devices, networks, and servers in fig. 8 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring now to FIG. 9, shown is a block diagram of a computer system 900 suitable for use with a terminal device implementing an embodiment of the present invention. The terminal device shown in fig. 9 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 9, the computer system 900 includes a Central Processing Unit (CPU)901 that can perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM)902 or a program loaded from a storage section 908 into a Random Access Memory (RAM) 903. In the RAM 903, various programs and data necessary for the operation of the system 900 are also stored. The CPU 901, ROM 902, and RAM 903 are connected to each other via a bus 904. An input/output (I/O) interface 905 is also connected to bus 904.
The following components are connected to the I/O interface 905: an input portion 906 including a keyboard, a mouse, and the like; an output section 907 including components such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 908 including a hard disk and the like; and a communication section 909 including a network interface card such as a LAN card, a modem, or the like. The communication section 909 performs communication processing via a network such as the internet. The drive 910 is also connected to the I/O interface 905 as necessary. A removable medium 911 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 910 as necessary, so that a computer program read out therefrom is mounted into the storage section 908 as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 909, and/or installed from the removable medium 911. The above-described functions defined in the system of the present invention are executed when the computer program is executed by a Central Processing Unit (CPU) 901.
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The modules described in the embodiments of the present invention may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor includes a transmitting unit, an obtaining unit, a determining unit, and a first processing unit. The names of these units do not in some cases constitute a limitation to the unit itself, and for example, the sending unit may also be described as a "unit sending a picture acquisition request to a connected server".
As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise:
acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information;
generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words;
and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
According to the technical scheme of the embodiment of the invention, the user attribute words are obtained based on the characteristic information of the user, and the commodity attribute words are extracted from the commodity information; generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words; and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
Because the recommendation information is generated by combining the characteristic information of the user, different preferences are provided for different users, the recommendation information conforms to the preferences of the user, and the adaptability of the recommendation information and the user can be improved.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A method for pushing information, comprising:
acquiring user attribute words based on the characteristic information of the user, and extracting commodity attribute words from the commodity information;
generating recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words;
and under the condition that the characteristic information is matched with the commodity, searching and obtaining recommendation information of the commodity in the recommendation information of the user according to the commodity, and pushing the recommendation information of the commodity to the user.
2. The method for pushing information according to claim 1, before obtaining recommendation information of a commodity according to the commodity in search of recommendation information of the user if the feature information matches the commodity, further comprising:
if the commodity corresponding to the characteristic information is the same as the commodity, matching the characteristic information with the commodity;
or the like, or, alternatively,
and if the commodity application scene corresponding to the characteristic information is the same as the commodity application scene of the commodity, matching the characteristic information with the commodity.
3. The method for pushing information according to claim 1, wherein the pushing recommendation information of the commodity to the user comprises:
under the condition that the characteristic information is matched with the commodity, searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity;
and taking the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
4. The method for pushing information according to claim 3, wherein the searching for the recommendation information with the greatest similarity from the recommendation information of the user according to the commodity comprises:
searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity attribute words of the commodities;
or the like, or, alternatively,
and searching recommendation information with the maximum similarity in the recommendation information of the user according to the commodity attribute words of the commodities and the using times of the recommendation information.
5. The method for pushing information according to claim 1, wherein before generating the recommendation information of the user according to the commodity attribute words and the user attribute words through a sequence-to-sequence model based on an attention mechanism, the method further comprises:
configuring the weight of the commodity attribute training words and the weight of the user attribute training words;
and updating a loss function by using the weight of the commodity attribute training word and the weight of the user attribute training word and adopting a mask matrix, and training to obtain the sequence-to-sequence model based on the attention mechanism.
6. An apparatus for pushing information, comprising:
the information module is used for acquiring user attribute words based on the characteristic information of the user and extracting commodity attribute words from the commodity information;
the generation module is used for generating the recommendation information of the user through a sequence-to-sequence model based on an attention mechanism according to the commodity attribute words and the user attribute words;
and the pushing module is used for searching and obtaining the recommendation information of the commodity in the recommendation information of the user according to the commodity under the condition that the characteristic information is matched with the commodity, and pushing the recommendation information of the commodity to the user.
7. The information pushing device according to claim 6, wherein the pushing module is specifically configured to, if the commodity corresponding to the feature information is the same as the commodity, match the feature information with the commodity;
or the like, or, alternatively,
and if the commodity application scene corresponding to the characteristic information is the same as the commodity application scene of the commodity, matching the characteristic information with the commodity.
8. The information pushing device according to claim 6, wherein the pushing module is specifically configured to search recommendation information with the largest similarity among the recommendation information of the user according to the commodity when the feature information is matched with the commodity;
and taking the recommendation information with the maximum similarity as recommendation information for pushing the commodity to the user.
9. An electronic device for pushing information, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.
10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010462528.6A CN113744002A (en) | 2020-05-27 | 2020-05-27 | Method, device, equipment and computer readable medium for pushing information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010462528.6A CN113744002A (en) | 2020-05-27 | 2020-05-27 | Method, device, equipment and computer readable medium for pushing information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113744002A true CN113744002A (en) | 2021-12-03 |
Family
ID=78723789
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010462528.6A Pending CN113744002A (en) | 2020-05-27 | 2020-05-27 | Method, device, equipment and computer readable medium for pushing information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113744002A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114265777A (en) * | 2021-12-23 | 2022-04-01 | 北京百度网讯科技有限公司 | Application program testing method and device, electronic equipment and storage medium |
| CN116579827A (en) * | 2023-07-11 | 2023-08-11 | 深圳千岸科技股份有限公司 | Commodity recommendation method and system based on user network behavior portrayal |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100687906B1 (en) * | 2005-12-31 | 2007-02-27 | 주식회사지앤지커머스 | System for recommendation the goods and method therefor |
| TW201216185A (en) * | 2010-10-06 | 2012-04-16 | Chunghwa Telecom Co Ltd | Multi-aspect merchandise recommendation method and system thereof |
| CN107730343A (en) * | 2017-09-15 | 2018-02-23 | 广州唯品会研究院有限公司 | A kind of user's merchandise news method for pushing and equipment based on picture attribute extraction |
| CN110222272A (en) * | 2019-04-18 | 2019-09-10 | 广东工业大学 | A kind of potential customers excavate and recommended method |
| CN110929138A (en) * | 2018-09-04 | 2020-03-27 | 阿里巴巴集团控股有限公司 | Recommendation information generation method, device, equipment and storage medium |
-
2020
- 2020-05-27 CN CN202010462528.6A patent/CN113744002A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100687906B1 (en) * | 2005-12-31 | 2007-02-27 | 주식회사지앤지커머스 | System for recommendation the goods and method therefor |
| TW201216185A (en) * | 2010-10-06 | 2012-04-16 | Chunghwa Telecom Co Ltd | Multi-aspect merchandise recommendation method and system thereof |
| CN107730343A (en) * | 2017-09-15 | 2018-02-23 | 广州唯品会研究院有限公司 | A kind of user's merchandise news method for pushing and equipment based on picture attribute extraction |
| CN110929138A (en) * | 2018-09-04 | 2020-03-27 | 阿里巴巴集团控股有限公司 | Recommendation information generation method, device, equipment and storage medium |
| CN110222272A (en) * | 2019-04-18 | 2019-09-10 | 广东工业大学 | A kind of potential customers excavate and recommended method |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114265777A (en) * | 2021-12-23 | 2022-04-01 | 北京百度网讯科技有限公司 | Application program testing method and device, electronic equipment and storage medium |
| CN116579827A (en) * | 2023-07-11 | 2023-08-11 | 深圳千岸科技股份有限公司 | Commodity recommendation method and system based on user network behavior portrayal |
| CN116579827B (en) * | 2023-07-11 | 2024-01-05 | 深圳千岸科技股份有限公司 | Commodity recommendation method and system based on user network behavior portrayal |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107172151B (en) | Method and device for pushing information | |
| WO2018192491A1 (en) | Information pushing method and device | |
| CN107862553A (en) | Advertisement real-time recommendation method, device, terminal device and storage medium | |
| CN107426328B (en) | Information pushing method and device | |
| CN110196904B (en) | Method and device for acquiring recommendation information and computer readable storage medium | |
| CN105809473B (en) | Training method for matching model parameters, service recommendation method and corresponding device | |
| CN109783539A (en) | Usage mining and its model building method, device and computer equipment | |
| US11263664B2 (en) | Computerized system and method for augmenting search terms for increased efficiency and effectiveness in identifying content | |
| US20160042403A1 (en) | Extraction device, extraction method, and non-transitory computer readable storage medium | |
| JP2012234503A (en) | Recommendation device, recommendation method, and recommendation program | |
| CN106062743A (en) | Systems and methods for keyword suggestion | |
| TWI579787B (en) | Systems and methods for instant e-coupon distribution | |
| US11430049B2 (en) | Communication via simulated user | |
| CN112989169B (en) | Target object identification method, information recommendation method, device, equipment and medium | |
| CN109117442B (en) | Application recommendation method and device | |
| CN111125491A (en) | Commodity information searching method and device, storage medium and electronic device | |
| CN109819002B (en) | Data pushing method and device, storage medium and electronic device | |
| CN111914188A (en) | Method, system, device and storage medium for selecting recommendation target user | |
| CN111104590A (en) | Information recommendation method, device, medium and electronic equipment | |
| CN113744002A (en) | Method, device, equipment and computer readable medium for pushing information | |
| CN110399564B (en) | Account classification method and device, storage medium and electronic device | |
| CN111767457B (en) | Recommendation method and device | |
| CN112288510A (en) | Article recommendation method, device, equipment and storage medium | |
| CN113495991A (en) | Recommendation method and device | |
| CN110971973A (en) | Video pushing method and device and electronic equipment |
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 |