CN115374365A - Information recommendation method and device, storage medium and computer equipment - Google Patents

Information recommendation method and device, storage medium and computer equipment Download PDF

Info

Publication number
CN115374365A
CN115374365A CN202211114494.7A CN202211114494A CN115374365A CN 115374365 A CN115374365 A CN 115374365A CN 202211114494 A CN202211114494 A CN 202211114494A CN 115374365 A CN115374365 A CN 115374365A
Authority
CN
China
Prior art keywords
information
recommended
target
recommendation
friend
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
Application number
CN202211114494.7A
Other languages
Chinese (zh)
Inventor
徐小韵
赵俊龙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Kangjian Information Technology Shenzhen Co Ltd
Original Assignee
Kangjian Information Technology Shenzhen Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Kangjian Information Technology Shenzhen Co Ltd filed Critical Kangjian Information Technology Shenzhen Co Ltd
Priority to CN202211114494.7A priority Critical patent/CN115374365A/en
Publication of CN115374365A publication Critical patent/CN115374365A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • 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 an information recommendation method, an information recommendation device, a storage medium and computer equipment, relates to the technical field of information and digital medical treatment, and mainly aims to improve the recommendation accuracy and recommendation efficiency of information. The method comprises the following steps: determining each friend corresponding to the information recommendation user in a preset communication tool; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict recommendation values to obtain recommendation value prediction results corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; determining a target information processing mode corresponding to the information type to which the information to be recommended belongs; and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to the target friend. The invention is suitable for recommending information, such as information in the medical field.

Description

Information recommendation method and device, storage medium and computer equipment
Technical Field
The present invention relates to the field of information technologies, and in particular, to an information recommendation method, an information recommendation apparatus, a storage medium, and a computer device.
Background
The commodity information sharing is an important commodity popularization mode in an e-commerce platform. For example, when a user browses commodities, commodity information is recommended to friends through a sharing function provided in an e-commerce platform.
At present, when information recommendation is performed, friends needing to be recommended are usually determined manually. However, the information recommendation method requires a user to select a friend who needs information recommendation from all friends, which results in low efficiency of information recommendation, and meanwhile, the artificially determined friend is not a friend who is interested in the commodity, so that the order placing amount of the commodity is not increased even if the commodity information is sent to the friend, which results in that the commodity information cannot be effectively recommended, and further the accuracy of commodity information recommendation is low.
Disclosure of Invention
The invention provides an information recommendation method, an information recommendation device, a storage medium and computer equipment, which mainly aim to effectively recommend commodity information and improve the recommendation accuracy and recommendation efficiency of the information.
According to a first aspect of the present invention, there is provided an information recommendation method including:
determining each friend corresponding to an information recommending user corresponding to the information to be recommended in a preset communication tool;
acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend;
recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values;
judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type;
and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
According to a second aspect of the present invention, there is provided an information recommendation apparatus comprising:
the determining unit is used for determining each friend corresponding to the information recommendation user corresponding to the information to be recommended in a preset communication tool;
the recommendation value prediction unit is used for acquiring the feature data corresponding to each friend, inputting the feature data into a preset recommendation value prediction model to predict the recommendation value of the friend, and obtaining the recommendation value prediction result corresponding to each friend;
a friend recommending unit, configured to recommend a target friend for the information recommending user in each friend based on each recommendation value prediction result;
the judging unit is used for judging the information type of the information to be recommended and determining a target information processing mode corresponding to the information type;
and the information sending unit is used for processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended and sending the processed information to be recommended to the target friend end.
According to a third aspect of the present invention, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
determining each friend corresponding to an information recommending user corresponding to the information to be recommended in a preset communication tool;
acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend;
recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values;
judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type;
and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
According to a fourth aspect of the present invention, there is provided a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the program:
determining each friend corresponding to an information recommending user corresponding to the information to be recommended in a preset communication tool;
acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend;
recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values;
judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type;
and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
According to the information recommendation method, the information recommendation device, the storage medium and the computer equipment, compared with the mode that the friend to be recommended is usually determined manually when information recommendation is carried out at present, the method and the system recommend each friend corresponding to the user in the preset communication tool by determining the information corresponding to the information to be recommended; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; the information to be recommended is processed based on the target information processing mode to obtain processed information to be recommended, the processed information to be recommended is sent to the target friend end, the characteristic data corresponding to each friend of an information recommending user in a preset communication tool is determined, the conversion value corresponding to the characteristic data is predicted by using a preset recommendation value prediction model to obtain a recommendation value prediction result corresponding to each friend, and finally the information recommending user recommends the commodity information to the target friend, so that the information recommending efficiency is improved, the problem that the commodity information is recommended to the friend who is not interested in the commodity to cause the discomfort of the friend, and meanwhile, the commodity quantity is not improved, and the information recommending accuracy is improved.
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 shows a flowchart of an information recommendation method according to an embodiment of the present invention;
FIG. 2 is a flow chart of another information recommendation method provided by the embodiment of the invention;
fig. 3 is a schematic structural diagram illustrating an information recommendation apparatus according to an embodiment of the present invention;
fig. 4 shows a schematic physical structure diagram of a computer device according to an embodiment of the present invention.
Detailed Description
The invention will be described in detail hereinafter with reference to the accompanying drawings in conjunction with embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
At present, a user generally recommends commodity information to friends with a relatively close relationship, so that the efficiency of information recommendation is low, and meanwhile, the manually determined friends are not friends interested in the commodities, so that the ordering amount of the commodities cannot be increased even if the commodity information is sent to the friends, and further the accuracy of commodity information recommendation is low.
In order to solve the above problem, an embodiment of the present invention provides an information recommendation method, as shown in fig. 1, where the method includes:
101. and determining each friend corresponding to the information recommendation user corresponding to the information to be recommended in a preset communication tool.
The information recommendation user may be a user browsing a commodity, and the preset communication tool includes: each friend can be a friend created by an information recommendation user in the information recommendation device for commodity information recommendation provided by the embodiment of the invention, or a friend in a social platform, for example: wechat friends, QQ friends, mobile phone address list friends and the like.
For the embodiment of the invention, in order to overcome the problem of low accuracy of information recommendation in the prior art, the embodiment of the invention determines the feature data corresponding to each friend of an information recommendation user in a preset communication tool, predicts the conversion value corresponding to the feature data by using a preset recommendation value prediction model to obtain the recommendation value prediction result corresponding to each friend, and finally determines the target friend recommended to the information recommendation user based on the recommendation value prediction result, so that the efficiency of information recommendation is improved, meanwhile, the problem that the target friend is disliked when recommending commodities to friends who do not interest the commodities is avoided, and meanwhile, the problem that the quantity of commodities is not increased is solved, so that the accuracy of information recommendation is improved. The embodiment of the invention is mainly applied to a scene of recommending information, and the execution main body of the embodiment of the invention is a device or equipment capable of recommending information, and can be specifically arranged on one side of a client or a server.
Specifically, when a user browses the commodity in the commodity transaction platform, whether the user clicks a sharing identifier in a commodity browsing interface is monitored, if it is monitored that the user clicks the sharing identifier, each preset communication tool identifier is displayed, then, a clicking operation of the user for any one of the preset communication tool identifiers is monitored, each friend corresponding to the user in any one of the preset communication tools is determined, for example, if it is monitored that the user clicks the sharing identifier, a WeChat identifier, a QQ identifier, a mobile phone short message identifier and the like are displayed, if the user selects a WeChat identifier in each identifier, each WeChat of the user in WeChat is determined, then, feature data corresponding to each WeChat friend is determined, a recommendation value prediction result corresponding to each WeChat is predicted by using a preset recommendation value prediction model based on the feature data, a recommendation target is recommended for the information recommendation user, so that the information recommendation user selects a recommendation list needing to be shared by the friend recommendation information recommendation user in the recommended target friends, thereby improving the accuracy of commodity information, and avoiding purchasing commodity information in the recommendation list.
102. And acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict the recommendation value of the friend to obtain a recommendation value prediction result corresponding to each friend.
The feature data may include various feature information such as age, gender, city, relationship with information recommending users, occupation, income level, family members, interests and the like, and it should be noted that the feature data of each friend is obtained after authorization. The recommendation value prediction result may be an order sales amount corresponding to each friend.
In the embodiment of the invention, after the characteristic data of each friend is obtained, the characteristic data of each friend is respectively input into a preset recommendation value prediction model for recommendation value prediction, the preset recommendation value prediction model can be constructed based on a lightGBM model, an FM model or other neural network models, the conversion value of the characteristic data of each friend can be predicted by using the preset recommendation value prediction model, so that the recommendation value prediction result of each friend is obtained, each friend is recommended to an information recommending user according to the magnitude sequence of the recommendation value prediction result, meanwhile, the target friend with the recommendation value prediction result ranked in the top n can be determined as the target friend, only the target friend with the recommendation value in the top n is recommended to the information recommending user, the user is prevented from selecting the friend to be recommended from a large number of friends, and the efficiency of the user in selecting the friend is improved.
103. And recommending the target friend for the information recommendation user in each friend based on each recommendation value prediction result.
The target friend is a friend who has a high commodity order conversion probability for the information to be recommended, or a friend who has a high sales amount for the commodity order.
For the embodiment of the invention, after the recommendation value prediction results corresponding to the friends are obtained, the recommendation value prediction results are ranked from large to small, and a plurality of friends ranked at the previous preset rank are used for determining the target friends of the information recommendation user needing information recommendation, for example, the recommendation value prediction results corresponding to the WeChat friends are ranked from large to small, and the friends corresponding to the first 5 recommendation value prediction results are determined as the target friends for the family milkshake recommendation. Therefore, by determining the feature data corresponding to each friend of the information recommending user in the preset communication tool and predicting the conversion value corresponding to the feature data by using the preset recommendation value prediction model, the recommendation value prediction result corresponding to each friend is obtained, and finally the target friend recommended to the information recommending user is determined based on the recommendation value prediction result, so that the problems that the friend feels dislike when the friend recommends commodities to friends who do not have interest in the commodities and the quantity of orders under the commodities is not improved are avoided, the accuracy of information recommendation is improved, the quantity of orders under the commodities can be improved, the satisfied products are recommended to the friends, and the experience of the friends is enhanced.
104. And judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type.
The information types comprise character information, picture information, file information and the like.
For the embodiment of the present invention, in order to avoid that a friend receiving information can view information more conveniently, first, an information type to which the information to be recommended belongs needs to be determined, because different information types correspond to different information processing manners, and thus a target information processing manner corresponding to the information type needs to be determined, for example, the information to be recommended is medical insurance information in the digital medical field, and the medical insurance information is text information, the medical insurance information is finally processed according to the target information processing manner corresponding to the text information, so as to obtain processed medical insurance information, and the processed medical insurance information is sent to a target friend, so that a reading experience of the target friend for the medical insurance information is improved, and thus, a purchase amount of medical insurance products can be increased.
105. And processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
For the embodiment of the invention, after a target information processing mode corresponding to the information type to which the information to be recommended belongs is determined, the information to be recommended is processed according to the target information processing mode to obtain the processed information to be recommended, for example, the information to be recommended is text information, and the number of words of the text information is large, so that in order to avoid that the text information occupies too large area in a communication tool information display interface, the text information needs to be folded to obtain the folded text information, and finally the folded text information is sent to a target friend, the reading experience of a user for the commodity information can be enhanced, the problem that the user feels tired of reading when facing a large amount of text information, so that information recommendation fails is solved, and information can be effectively recommended.
According to the information recommendation method provided by the invention, compared with the mode that the friend needing to be recommended is usually determined manually when information recommendation is carried out at present, the information recommendation method has the advantages that the information corresponding to the information to be recommended is determined to recommend each corresponding friend of the user in the preset communication tool; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; the information to be recommended is processed based on the target information processing mode to obtain processed information to be recommended, the processed information to be recommended is sent to the target friend end, the characteristic data corresponding to each friend of an information recommending user in a preset communication tool is determined, the conversion value corresponding to the characteristic data is predicted by using a preset recommendation value prediction model to obtain a recommendation value prediction result corresponding to each friend, and finally the information recommending user recommends the commodity information to the target friend, so that the information recommending efficiency is improved, the problem that the commodity information is recommended to friends who do not interest in the commodity to cause the discomfort of the friends is avoided, and meanwhile, the commodity inventory is not improved, and the information recommending accuracy is improved.
Further, in order to better describe the above information recommendation process, as a refinement and an extension of the above embodiment, an embodiment of the present invention provides another information recommendation method, as shown in fig. 2, where the method includes:
201. and determining each friend corresponding to the information recommendation user corresponding to the information to be recommended in the preset communication tool.
Specifically, when a user browses the commodity in the commodity transaction platform, whether the user clicks a sharing identifier in a commodity browsing page is judged, if the sharing identifier is clicked, preset communication tools such as WeChat, QQ and mobile phone short messages are displayed, and if the user selects WeChat in the preset communication tools, each friend of the information recommendation user in WeChat is determined.
202. And acquiring the feature data corresponding to each friend.
Specifically, after determining each friend of the information recommendation user in the preset communication tool, in order to recommend a target user having a purchase desire for the product to the information recommendation user, feature data corresponding to each friend needs to be acquired first, for example, data such as age, income, job nature, position, and hobbies and interests corresponding to each friend are acquired.
203. Historical order data of a plurality of sample friends completing information recommendation are collected, and target order data used for model training are screened from the historical order data.
The finished information recommendation means that the commodity information is recommended to the sample friend, and the friend finishes ordering operation aiming at the commodity information or does not carry out ordering operation aiming at the commodity information. The collection of historical order data of a plurality of sample friends for which information recommendation is completed refers to recommendation records of the sample friends for which information recommendation is completed, and order related information for the commodity, such as commodity name, commodity purchase quantity, order generation time, sales amount corresponding to the order and the like, generated by recommendation.
For the embodiment of the invention, the acquired historical order data can be further screened out to obtain target order data for model training. For example, target order data can be selected from historical order data for subsequent model training according to multiple dimensions such as ages, professions, cities and sales lines corresponding to orders of different users.
204. And acquiring target sample friends and target sample recommended values corresponding to the orders in the target order data.
For the embodiment of the present invention, the target order data used for training and constructing the preset recommendation value prediction model may include a service order of a certain commodity by a plurality of different sample friends, for example, for a service order of a certain medical instrument, for any order, the corresponding target sample friend and the sales amount of the commodity purchased by the target sample friend are recorded. For any order in the target order data, a target sample friend and a target sample recommended value corresponding to the order can be obtained.
When the target sample friend is obtained, the target sample friend corresponding to the order can be determined according to the unique identifier (such as the order number, the user number and the like) recorded corresponding to the order. In addition, the target sample recommendation value can be sales amount corresponding to the order or other reference data which can represent friend value.
205. And acquiring target sample characteristic data of friends of each target sample, establishing a corresponding relation between the target sample characteristic data and corresponding target sample recommendation values, and constructing a training set.
For the embodiment of the invention, when the target sample characteristic data is obtained, the target sample characteristic data can be obtained based on the identification information of friends of the target sample, and the obtained target sample characteristic data can include but is not limited to the information of age, gender, occupation of the city, income level, family members, hobbies and interests and the like. With the combination of the above embodiments, for each order, the corresponding target sample friend and target sample recommendation value may be obtained and a corresponding relationship may be established to generate a data pair, so that a plurality of data pairs corresponding to a plurality of orders may be used as training data of the model.
206. And constructing a preset recommended value prediction model by using the training set.
For the embodiment of the present invention, after determining the training set, the step 206 specifically includes, based on the training set, training and constructing the preset recommendation value prediction model: inputting the target sample characteristic data in the training set into a preset initial recommendation value prediction model for recommendation value prediction to obtain a prediction sample recommendation value corresponding to the target sample friend; determining a loss function corresponding to the preset initial recommended value prediction model based on the target sample recommended value and the prediction sample recommended value; and performing iterative training on the preset initial recommendation value prediction model based on the loss function to construct the preset recommendation value prediction model.
For the embodiment of the invention, when the training set is used for training the preset recommendation value prediction model, 70% of data in the training set can be used as the training set, 30% of data can be used as the test set, the training set is further used for training the preset recommendation value prediction model, and the test set is used for carrying out test optimization on the trained preset recommendation value prediction model.
Specifically, when the preset recommendation value prediction model is constructed, a plurality of (two or more) preset initial recommendation value prediction models can be constructed, the model architectures of the preset initial recommendation value prediction models can be the same or different, and when the model is trained, the training data can be divided into a plurality of groups of training subdata according to the number of the preset initial recommendation value prediction models; and then, training each preset initial recommended value prediction model by using a plurality of data pairs in each group of training subdata, wherein target sample characteristic data is used as input data, and a target sample recommended value is used as output data.
For example, in this embodiment, a preset initial recommended value prediction model 1 and a preset initial recommended value prediction model 2 are respectively constructed, and meanwhile, the training data may be randomly divided into training subdata 1 and training subdata 2, further, the preset initial recommended value prediction model 1 may be trained by using the training subdata 1, and the preset initial recommended value prediction model 2 may be trained by using the training subdata 2.
Further, when each preset initial recommendation value prediction model is trained, a loss function corresponding to each preset initial recommendation value prediction model may be calculated, and each preset initial recommendation value prediction model may be iteratively trained by using the loss function, where a formula for specifically calculating the loss function is as follows:
Figure BDA0003844938670000101
wherein L is 1 Represents a loss function, a i Representing a target sample recommendation value corresponding to the ith target sample friend, c i Represents a prediction sample recommendation value, v, corresponding to the ith target sample friend r Representing a weight balance factor, v r The value of (d) is far less than 1, di represents the model parameter corresponding to the preset initial recommendation value prediction model, and the preset initial recommendation value prediction model is continuously trained based on the loss function until the preset initial recommendation value is obtainedThe value of the loss function corresponding to the recommendation prediction model does not decrease.
Further, in the embodiment of the invention, the prediction accuracy of each preset initial recommendation value prediction model can be calculated; and selecting a target preset initial recommendation value prediction model with the highest prediction accuracy from the plurality of preset initial recommendation value prediction models as a final preset recommendation value prediction model based on the prediction accuracy of each value prediction model. For example, for the trained preset initial recommendation value prediction model 1 and preset initial recommendation value prediction model 2, the corresponding accuracy rates can be respectively calculated, and if the prediction accuracy rate corresponding to the preset initial recommendation value prediction model 1 is 80% and the prediction accuracy rate corresponding to the preset initial recommendation value prediction model 2 is 95%, the preset initial recommendation value prediction model 2 is finally used as the preset recommendation value prediction model to predict the recommendation value prediction results corresponding to the friends.
207. And inputting the characteristic data into the preset recommendation value prediction model to predict recommendation values, and obtaining recommendation value prediction results corresponding to the friends.
For the embodiment of the invention, after the preset recommendation value prediction model is constructed, the acquired feature data corresponding to each friend are respectively input into the preset recommendation value prediction model to perform recommendation value prediction, so that the recommendation value prediction results corresponding to each friend can be obtained, and finally, the top n target friends with higher recommendation value prediction results are selected from the friends to be recommended to the information recommending user based on the recommendation value prediction results.
208. And recommending the target friend for the information recommendation user in each friend based on each recommendation value prediction result.
For the embodiment of the present invention, after the recommendation value prediction result (order sales) corresponding to each friend is obtained, the friends are ranked in order of order sales from large to small, and the ranked friends are displayed to the information recommending user, or the top N target friends with the order sales are displayed to the information recommending user.
209. And judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type.
For the embodiment of the present invention, when recommending a target friend for the information recommendation user, it is further required to determine an information type to which the information to be recommended belongs, and determine that target information corresponding to the information type is processed, based on this, step 209 specifically includes: if the information type of the information to be recommended is the text information, determining a target information processing mode corresponding to the text information; if the information type of the information to be recommended is picture information, determining a target information processing mode corresponding to the picture information; and if the information type of the information to be recommended is file information, determining a target information processing mode corresponding to the file information.
Specifically, the different information types correspond to different information processing manners, for example, a text information corresponding to a text information processing manner, a picture information corresponding to a picture information processing manner, and a file information corresponding to a file information processing manner, if the information to be recommended is text information, the information to be recommended is processed by using the processing manner corresponding to the text information, if the information to be recommended is picture information, the information to be recommended is processed by using the processing manner corresponding to the picture information, and if the information to be recommended is file information, the information to be recommended is processed by using the processing manner corresponding to the file information.
210. And processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
For the embodiment of the present invention, if the target information processing manner is processing corresponding to the text information, then the information to be recommended needs to be processed based on the processing manner corresponding to the text information, and based on this, step 210 specifically includes: and judging whether the character length corresponding to the character information is larger than a preset character length or not, and if so, folding the character information to obtain the character information after folding.
The preset character length can be determined according to the size of a display area of a dialog box in a preset communication tool. For the embodiment of the present invention, if the information to be recommended is text information, when the length of the text corresponding to the text information exceeds the length that can be accepted by the dialog box, that is, when the length of the text corresponding to the text information is greater than the preset length, it is necessary to fold the text that is greater than the preset length in the text information, so as to generate the text information after folding, so as to display the text information after folding to the target friend, for example, if the text information corresponding to the commodity information (information to be recommended) that needs to be sent includes 30 texts, and the length of the text that can be accepted by the dialog box is 20 texts, it is necessary to fold the last 10 texts in the text information, and specifically > can be used to replace the last 10 texts in the text information.
Further, if the commodity information includes the privacy information or sensitive words of the user, in order to avoid others from seeing the messages in the dialog box and to avoid revealing the privacy information of the user or causing other adverse effects, based on this, before judging whether the text length corresponding to the text information is greater than a preset text length, the method includes: performing word segmentation processing on the character information in the commodity information to obtain each word segmentation corresponding to the character information, and detecting whether each word segmentation comprises a sensitive word or not by using a preset sensitive word bank; and if the sensitive words are contained, hiding the sensitive words contained in the text information. The preset sensitive word bank is a pre-constructed word bank, and a large number of sensitive words such as killers, robbers, abductions and the like are contained in the preset sensitive word bank.
Specifically, if the commodity information to be sent is character information, word segmentation processing is performed on the character information to obtain each word segmentation corresponding to the character information, then a preset sensitive word bank is used for detecting whether each word segmentation comprises a sensitive word in the sensitive word bank, if the character information comprises a sensitive word in the sensitive word bank, the sensitive word in the character information is hidden in a form of' word or mosaic, then whether the character length corresponding to the character information with the sensitive word hidden is larger than the preset character length is judged, if the character length is larger than the preset character length, a part larger than the preset character length in a conversation message with the sensitive word hidden is required to be folded to obtain the character information with the sensitive word hidden and subjected to folding processing, and finally the hidden sensitive word and the folded commodity information are sent to a target friend. It should be noted that, in the embodiment of the present invention, a manner of hiding the sensitive word in the text information is not limited to the form of an "a" or a mosaic, and other symbols may also be used to hide the sensitive word, which is not limited in the embodiment of the present invention.
Further, if the target information processing mode is an information processing mode corresponding to picture information, the method further includes: and acquiring a picture identifier corresponding to the picture information, and determining the picture identifier as the processed information to be recommended.
For the embodiment of the present invention, if the commodity information is picture information, the picture information needs to be represented by a picture identifier, the picture identifier may specifically be represented by [ PIC ], but is not limited to be represented by [ PIC ], the picture identifier is sent to the target friend, further, when the target friend clicks the picture identifier, a corresponding picture frame is popped up, picture attributes such as size and occupied space of the picture, and operation items such as downloading, saving, and previewing are recorded on the picture frame, when the target friend wants to view the content in the picture, the preview option in the picture frame is clicked, the content in the picture can be viewed, and meanwhile, the target user may click the saving option in the picture frame to save the picture locally, open the picture locally, and view the content in the picture.
Further, if the target information processing mode is an information processing mode corresponding to file information, the method further includes: and acquiring a file identifier corresponding to the file information, and determining the file identifier as the processed information to be recommended.
The information to be recommended may also be a PDF document, a world document, an excel document, and the like.
For the embodiment of the present invention, since the information to be recommended is limited by the display area during the dialog box display process, the FILE information needs to be formatted, specifically, if the information to be recommended is FILE information, the information to be recommended needs to be represented by a FILE identifier, the FILE identifier may specifically be represented by [ FILE ] but is not limited to [ FILE ], and the FILE identifier is sent to the target friend, further, when the target friend clicks the FILE identifier, a corresponding frame is popped up, FILE attributes such as the size, the occupied space, the FILE type, and the like of the FILE, and operation items such as downloading, saving, and previewing are recorded on the frame, when the target friend wants to view the content in the FILE, the preview option in the frame is clicked, so that the content in the FILE can be viewed, and at the same time, the target friend can click the download option in the frame to download the FILE to the local, open the FILE locally, and view the content in the FILE.
It should be noted that the information to be recommended in the text form, the information to be recommended in the file form, and the information to be recommended in the picture form all have a link address of the commodity, and when the target friend clicks the link address, the target friend can automatically log in the commodity transaction platform.
Further, after the processed information to be recommended corresponding to the information to be recommended is obtained, the processed information to be recommended needs to be sent to the target friend, and based on this, the method includes: and responding to a trigger event of the information recommending user to the identification information corresponding to the target friend, and sending the processed information to be recommended to the target friend end.
The number of the target friends may be 1 or more, the specific number is not specifically limited in the embodiments of the present invention, and the information to be recommended is description information, website information, and the like corresponding to the commodity to be recommended.
Specifically, after recommending a target friend for a user, the target friend may be shown to the user in the form of identification information, the user selects a friend that needs to send commodity information among the target friends, the user may select one friend among the target friends to send the commodity information, or may select a plurality of or all the target friends to send the commodity information, and after the user selects a friend that is to send the commodity information among the target friends, the system automatically sends the processed information corresponding to the commodity to the friend.
Further, after sending the commodity information to the target friend, in order to encourage people leap forward to the commodity information recommendation, a reward mechanism needs to be adopted for the user, and based on this, the method further includes: judging whether the target friend of the target friend end carries out order conversion aiming at the processed information to be recommended or not; if the target friend carries out order conversion on the processed information to be recommended, determining an order conversion value; and distributing corresponding rewards to the information recommendation users based on the order conversion value.
And the order conversion is the order placing operation, and the order conversion value is the commodity transaction amount. Specifically, whether the target friend purchases the recommended commodity based on the received commodity information is judged, and meanwhile, the transaction amount generated when the target friend purchases the commodity is determined, the larger the transaction amount is, the more rewards distributed to the information recommending users are, and meanwhile, the more target users purchasing the commodity according to the commodity recommending information are, the more rewards distributed to the information recommending users are, the reward form may be a form of issuing a coupon, or a credit, or a full reduction, and the specific reward form is not specifically limited in the embodiment of the present invention.
According to another information recommendation method provided by the invention, compared with a mode of manually determining friends needing to be recommended during information recommendation, the method provided by the invention is used for recommending the friends corresponding to the user in a preset communication tool by determining the information corresponding to the information to be recommended; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; the information to be recommended is processed based on the target information processing mode to obtain processed information to be recommended, the processed information to be recommended is sent to the target friend end, the characteristic data corresponding to each friend of an information recommending user in a preset communication tool is determined, the conversion value corresponding to the characteristic data is predicted by using a preset recommendation value prediction model to obtain a recommendation value prediction result corresponding to each friend, and finally the information recommending user recommends the commodity information to the target friend, so that the information recommending efficiency is improved, the problem that the commodity information is recommended to friends who do not interest in the commodity to cause the discomfort of the friends is avoided, and meanwhile, the commodity inventory is not improved, and the information recommending accuracy is improved.
Further, as a specific implementation of fig. 1, an embodiment of the present invention provides an information recommendation apparatus, as shown in fig. 3, the apparatus includes: a determination unit 31, a recommendation value prediction unit 32, a friend recommendation unit 33, a judgment unit 34, and an information transmission unit 35.
The determining unit 31 may be configured to determine each friend corresponding to the information recommendation user corresponding to the information to be recommended in the preset communication tool.
The recommendation value prediction unit 32 may be configured to obtain feature data corresponding to each friend, and input the feature data into a preset recommendation value prediction model to perform friend recommendation value prediction, so as to obtain a recommendation value prediction result corresponding to each friend.
The friend recommending unit 33 may be configured to recommend a target friend for the information recommending user in each friend based on each recommendation value prediction result.
The determining unit 34 may be configured to determine an information type to which the information to be recommended belongs, and determine a target information processing manner corresponding to the information type.
The information sending unit 35 may be configured to process the information to be recommended based on the target information processing manner, obtain processed information to be recommended, and send the processed information to be recommended to a target friend end.
In a specific application scenario, in order to construct the preset recommendation value prediction model, the apparatus further includes: a data screening unit 36 and a model building unit 37.
The data screening unit 36 may be configured to collect historical order data of multiple sample friends for whom information recommendation is completed, and screen target order data for model training from the historical order data.
The determining unit 31 may be further configured to obtain a target sample friend and a target sample recommendation value corresponding to each order in the target order data.
The determining unit 31 may be further configured to obtain target sample feature data of each target sample friend, establish a corresponding relationship between the target sample feature data and a corresponding target sample recommendation value, and construct a training set.
The model building unit 37 may be configured to build the preset recommendation value prediction model by using the training set.
In a specific application scenario, in order to construct the preset recommendation value prediction model by using the training set, the model constructing unit 37 includes a recommendation value prediction module 371, a function determining module 372, and a model constructing module 373.
The recommendation value predicting module 371 may be configured to input the target sample feature data in the training set into a preset initial recommendation value prediction model to perform recommendation value prediction, so as to obtain a prediction sample recommendation value corresponding to the target sample friend.
The function determining module 372 may be configured to determine a loss function corresponding to the preset initial recommended value prediction model based on the target sample recommended value and the predicted sample recommended value.
The model building module 373 may be configured to build the preset recommendation prediction model based on the loss function.
In a specific application scenario, in order to determine an information type to which the information to be recommended belongs and determine a target information processing manner corresponding to the information type, the determining unit 34 may be specifically configured to determine the target information processing manner corresponding to the text information if the information type to which the information to be recommended belongs is text information; if the information type of the information to be recommended is picture information, determining a target information processing mode corresponding to the picture information; and if the information type of the information to be recommended is file information, determining a target information processing mode corresponding to the file information.
In a specific application scenario, in order to process the information to be recommended based on a target information processing manner to obtain the processed information to be recommended, the information sending unit 35 may be specifically configured to process the information to be recommended based on a target information processing manner corresponding to the text information to obtain the processed information to be recommended, and/or; processing the information to be recommended based on a target information processing mode corresponding to the picture information to obtain processed information to be recommended and/or; and processing the information to be recommended based on a target information processing mode corresponding to the file information to obtain the processed information to be recommended.
In a specific application scenario, if the information to be recommended is text information, in order to obtain processed information to be recommended, the information sending unit 35 may be specifically configured to determine whether a text length corresponding to the text information is greater than a preset text length, and if the text length is greater than the preset text length, perform folding processing on the text information to obtain the folded text information.
If the information to be recommended is picture information, in order to obtain processed information to be recommended, the information sending unit 35 may be further configured to specifically obtain a picture identifier corresponding to the picture information, and determine the picture identifier as the processed information to be recommended.
If the information to be recommended is picture information, in order to obtain processed information to be recommended, the information sending unit 35 may be further configured to specifically obtain a file identifier corresponding to the file information, and determine the file identifier as the processed information to be recommended.
In a specific application scenario, in order to send the processed information to be recommended to a target friend end, the information sending unit 35 may be specifically configured to send the processed information to be recommended to the target friend end in response to a trigger event of the information recommendation user on identification information corresponding to the target friend.
In a specific application scenario, in order to distribute a reward to an information recommending user, the apparatus further includes: a decision unit 38 and a bonus distribution unit 39.
The determining unit 38 may be configured to determine whether the target friend performs order conversion for the information to be recommended.
The determining unit 31 may be further configured to determine an order conversion value if the target friend performs order conversion for the information to be recommended.
The reward distribution unit 39 may be configured to distribute a corresponding reward to the information recommendation user based on the order conversion value.
It should be noted that other corresponding descriptions of the functional modules related to the information recommendation device provided in the embodiment of the present invention may refer to the corresponding description of the method shown in fig. 1, and are not described herein again.
Based on the method shown in fig. 1, correspondingly, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the following steps: determining various friends corresponding to information recommendation users corresponding to the information to be recommended in a preset communication tool; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
Based on the above embodiments of the method shown in fig. 1 and the apparatus shown in fig. 3, an embodiment of the present invention further provides an entity structure diagram of a computer device, as shown in fig. 4, where the computer device includes: a processor 41, a memory 42, and a computer program stored on the memory 42 and executable on the processor, wherein the memory 42 and the processor 41 are both arranged on a bus 43 such that when the processor 41 executes the program, the following steps are performed: determining various friends corresponding to information recommendation users corresponding to the information to be recommended in a preset communication tool; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
According to the technical scheme, the method and the system for recommending the friends of the user in the communication tool determine the information corresponding to the information to be recommended, and recommend the friends corresponding to the user in the preset communication tool; acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend; recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values; judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type; the information to be recommended is processed based on the target information processing mode to obtain processed information to be recommended, the processed information to be recommended is sent to the target friend end, the characteristic data corresponding to each friend of an information recommending user in a preset communication tool is determined, the conversion value corresponding to the characteristic data is predicted by using a preset recommendation value prediction model to obtain a recommendation value prediction result corresponding to each friend, and finally the information recommending user recommends the commodity information to the target friend, so that the information recommending efficiency is improved, the problem that the commodity information is recommended to the friend who is not interested in the commodity to cause the discomfort of the friend, and meanwhile, the commodity quantity is not improved, and the information recommending accuracy is improved.
It will be apparent to those skilled in the art that the modules or steps of the present invention described above may be implemented by a general purpose computing device, they may be centralized in a single computing device or distributed across a network of multiple computing devices, and alternatively, they may be implemented by program code executable by a computing device, such that they may be stored in a memory device and executed by a computing device, and in some cases, the steps shown or described may be executed out of order, or separately as individual integrated circuit modules, or multiple modules or steps thereof may be implemented as a single integrated circuit module. Thus, the present invention is not limited to any specific combination of hardware and software.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made without departing from the spirit and principle of the present invention shall fall within the protection scope of the present invention.

Claims (10)

1. An information recommendation method, comprising:
determining each friend corresponding to an information recommending user corresponding to the information to be recommended in a preset communication tool;
acquiring feature data corresponding to each friend, and inputting the feature data into a preset recommendation value prediction model to predict a friend recommendation value to obtain a recommendation value prediction result corresponding to each friend;
recommending target friends for the information recommendation users in the friends based on the prediction results of the recommendation values;
judging the information type of the information to be recommended, and determining a target information processing mode corresponding to the information type;
and processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and sending the processed information to be recommended to a target friend end.
2. The method of claim 1, wherein before the feature data is input into a preset recommendation prediction model for friend recommendation prediction to obtain recommendation prediction results corresponding to the friends, the method further comprises:
collecting historical order data of a plurality of sample friends which have finished information recommendation, and screening target order data for model training from the historical order data;
acquiring target sample friends and target sample recommendation values corresponding to each order in the target order data;
acquiring target sample characteristic data of each target sample friend, establishing a corresponding relation between the target sample characteristic data and a corresponding target sample recommendation value, and constructing a training set;
and constructing the preset recommended value prediction model by using the training set.
3. The method of claim 2, wherein the constructing the preset recommendation prediction model using the training set comprises:
inputting the target sample characteristic data in the training set into a preset initial recommendation value prediction model for recommendation value prediction to obtain a prediction sample recommendation value corresponding to the target sample friend;
determining a loss function corresponding to the preset initial recommended value prediction model based on the target sample recommended value and the prediction sample recommended value;
and performing iterative training on the preset initial recommendation value prediction model based on the loss function to construct the preset recommendation value prediction model.
4. The method according to claim 1, wherein the determining the target information processing manner corresponding to the information type includes:
if the information type of the information to be recommended is the text information, determining a target information processing mode corresponding to the text information;
if the information type of the information to be recommended is picture information, determining a target information processing mode corresponding to the picture information;
if the information type of the information to be recommended is file information, determining a target information processing mode corresponding to the file information;
the processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended, and the processing comprises:
processing the information to be recommended based on a target information processing mode corresponding to the text information to obtain processed information to be recommended and/or;
processing the information to be recommended based on a target information processing mode corresponding to the picture information to obtain processed information to be recommended and/or;
and processing the information to be recommended based on a target information processing mode corresponding to the file information to obtain the processed information to be recommended.
5. The method according to claim 4, wherein the processing the information to be recommended based on the target information processing mode corresponding to the text information to obtain the processed information to be recommended comprises:
judging whether the character length corresponding to the character information is larger than a preset character length or not, if so, folding the character information to obtain the character information after folding;
the processing the information to be recommended based on the target information processing mode corresponding to the picture information to obtain the processed information to be recommended, and the processing comprises:
acquiring a picture identifier corresponding to the picture information, and determining the picture identifier as the processed information to be recommended;
the processing the information to be recommended based on the target information processing mode corresponding to the file information to obtain the processed information to be recommended includes:
and acquiring a file identifier corresponding to the file information, and determining the file identifier as the processed information to be recommended.
6. The method according to claim 1, wherein the sending the processed information to be recommended to a target friend end comprises:
and responding to a trigger event of the information recommendation user to the identification information corresponding to the target friend, and sending the processed information to be recommended to the target friend end.
7. The method according to claim 1, wherein after sending the processed information to be recommended to a target friend end, the method further comprises:
judging whether the target friend of the target friend end carries out order conversion on the processed information to be recommended or not;
if the target friend carries out order conversion on the processed information to be recommended, determining an order conversion value;
and distributing corresponding rewards to the information recommendation users based on the order conversion value.
8. An information recommendation apparatus, comprising:
the determining unit is used for determining each friend corresponding to the information recommendation user corresponding to the information to be recommended in the preset communication tool;
the recommendation value prediction unit is used for acquiring the feature data corresponding to each friend, inputting the feature data into a preset recommendation value prediction model to predict the recommendation value of the friend, and obtaining the recommendation value prediction result corresponding to each friend;
a friend recommending unit, configured to recommend a target friend for the information recommending user in each friend based on each recommendation value prediction result;
the judging unit is used for judging the information type of the information to be recommended and determining a target information processing mode corresponding to the information type;
and the information sending unit is used for processing the information to be recommended based on the target information processing mode to obtain the processed information to be recommended and sending the processed information to be recommended to the target friend end.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.
10. A computer arrangement comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 7 when executed by the processor.
CN202211114494.7A 2022-09-14 2022-09-14 Information recommendation method and device, storage medium and computer equipment Pending CN115374365A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211114494.7A CN115374365A (en) 2022-09-14 2022-09-14 Information recommendation method and device, storage medium and computer equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211114494.7A CN115374365A (en) 2022-09-14 2022-09-14 Information recommendation method and device, storage medium and computer equipment

Publications (1)

Publication Number Publication Date
CN115374365A true CN115374365A (en) 2022-11-22

Family

ID=84070664

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211114494.7A Pending CN115374365A (en) 2022-09-14 2022-09-14 Information recommendation method and device, storage medium and computer equipment

Country Status (1)

Country Link
CN (1) CN115374365A (en)

Similar Documents

Publication Publication Date Title
US11049156B2 (en) Time-decay analysis of a photo collection for automated item listing generation
US9678641B2 (en) Configurable computation modules
JP6767824B2 (en) Judgment device, judgment method and judgment program
CN111651704B (en) Content recommendation method, computing device and storage medium
US8185535B2 (en) Methods and systems for determining unknowns in collaborative filtering
CN109299356B (en) Activity recommendation method and device based on big data, electronic equipment and storage medium
US8793236B2 (en) Method and apparatus using historical influence for success attribution in network site activity
CN110674391A (en) Product data pushing method and system based on big data and computer equipment
CN113051480A (en) Resource pushing method and device, electronic equipment and storage medium
KR102477687B1 (en) Apparatus and method for managing aadvertisement
KR101737424B1 (en) Method and server for providing advertisement based on purchase and participation possibility of user
CN111787042B (en) Method and device for pushing information
US20230085086A1 (en) System and method for application traffic control
US20150046283A1 (en) Computerized sales lead management apparatus and method
KR20190079582A (en) Service mehtod for recommending gifts and server usingn the same
CN110838019A (en) Method and device for determining trial supply distribution crowd
CN115374365A (en) Information recommendation method and device, storage medium and computer equipment
CN114925275A (en) Product recommendation method and device, computer equipment and storage medium
JP2022027520A (en) Systems and methods for obtaining information from digital message
CN107733775A (en) Information-pushing method and device
CN112015970A (en) Product recommendation method, related equipment and computer storage medium
CN110610380A (en) User portrait acquisition method, device, equipment and computer readable storage medium
CN117217852B (en) Behavior recognition-based purchase willingness prediction method and device
KR102591820B1 (en) Method and apparatus for providing a marketing platform for matching adverisers and advertising performers and conducting marketing through online advertising activities of advertisers
JP7407779B2 (en) Information processing device, information processing method, and information processing program

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