This application claims priority to Chinese Patent Application No. 201710501465.9, filed on Jun. 27, 2017 and entitled “INFORMATION PUSHING METHOD AND SYSTEM”, which is incorporated herein by reference in its entirety.
The present application relates to the field of electronic information, and in particular, to information pushing methods and systems.
With the increasing popularity of e-commerce, recommending commodities to users is an important research area. How to improve users' purchasing power by recommending commodities to the users is an urgent problem to be solved.
This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify all key features or essential features of the claimed subject matter, nor is it intended to be used alone as an aid in determining the scope of the claimed subject matter. The term “techniques,” for instance, may refer to device(s), system(s), method(s) and/or processor-readable/computer-readable instructions as permitted by the context above and throughout the present disclosure.
In the process of the research, the applicant found that simply recommending commodities to users did not have a significant effect on improvement of the purchasing power. However, sending user generated content (UGC), such as comments on commodities, to users can increase the purchasing power.
The present application provides information pushing methods and systems, aimed at solving the problem of how to send UGC on a website as pushed content.
In order to achieve the foregoing objective, the present application provides the following technical solutions.
An information pushing method includes:
Optionally, the plurality of pieces of UGCs include high-quality pieces of UGC; and
Optionally, the condition further includes:
Optionally, pushing the candidate piece(s) of UGC to the user includes:
Optionally, a process of generating the user label includes:
Optionally, a method of selecting the high-quality piece of UGCs includes:
Optionally, extracting the feature from the piece of UGC includes:
Optionally, the candidate piece of UGC does not include a piece of UGC of the user.
Optionally, the condition further includes:
An information pushing system, including:
Optionally, the plurality of pieces of UGC include high-quality pieces of UGC; and
Optionally, the recommendation generation module is specifically configured to:
Optionally, the message pushing module is specifically configured to:
Optionally, the system further includes:
Optionally, the system further includes:
Optionally, the high-quality UGC mining module is specifically configured to:
Optionally, the recommendation generation module is specifically configured to:
Optionally, the recommendation generation module is specifically configured to:
An information pushing system, including:
Optionally, the processor is specifically configured to: generate an information pushing list according to the piece of candidate UGC and a user label of the candidate piece of UGC, each piece of UGC in the information pushing list carrying a user label of the respective piece of UGC; and push the information pushing list to the user.
A computer readable storage medium, wherein the computer readable storage medium stores instructions which, when running on a computer, enable the computer to perform the following functions: determining a demand object of a user according to historical behavior data of the user; selecting, from a plurality of UGCs, a piece of UGC that satisfies a condition as a candidate piece of UGC, the condition including being related to the demand object of the user; and pushing the candidate piece of UGC to the user.
An information pushing method, including:
Optionally, forming the recommended piece of UGC based on the candidate piece of UGC includes:
Optionally, the condition further includes:
According to the methods and systems of the present application, a demand object of a user is determined according to historical behavior data of the user, and a piece of UGC related to the demand object of the user is pushed to the user, so that the pushed information is more credible. Further, the present application can be applied to an e-commerce website to increase users' purchasing power.
To illustrate the technical solutions according to the embodiments of the present application more clearly, the accompanying figures required for describing the embodiments introduced briefly below. Apparently, the accompanying drawings in the following description merely represent some embodiments of the present application. One of ordinary skill in the art can further obtain other drawings according to the accompanying drawings without any creative effort.
The information pushing method and system provided in the present application can be applied to a server of a website. A user registered with the website can publish User Generated Content (UGC) for an object displayed on the website. By taking an e-commerce website as an example, a user registered with the e-commerce website, after purchasing a commodity displayed on the e-commerce website, can make a comment on the purchased commodity (the comment is the user's UGC).
The information pushing method and system provided in the present application are aimed at pushing a user's UGC to users (which may also include the user) other than the user.
The information pushing system 100 may further include one or more processors 114, an input/output (I/O) interface 116, a network interface 118, and memory 120.
The memory 120 may include a form of computer readable media such as a volatile memory, a random access memory (RAM) and/or a non-volatile memory, for example, a read-only memory (ROM) or a flash RAM. The memory 120 is an example of a computer readable media.
The computer readable media may include a volatile or non-volatile type, a removable or non-removable media, which may achieve storage of information using any method or technology. The information may include a computer-readable instruction, a data structure, a program module or other data. Examples of computer storage media include, but not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electronically erasable programmable read-only memory (EEPROM), quick flash memory or other internal storage technology, compact disk read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media, which may be used to store information that may be accessed by a computing device. As defined herein, the computer readable media does not include transitory media, such as modulated data signals and carrier waves.
In implementations, the memory 120 may include program modules 122 and program data 124. The program modules 122 may include one or more of the modules as described above.
The functions of the modules in
S202: A user demand mining module determines a demand object of a user A according to historical behavior data of the user A.
The demand object of the user A is an object of an action that the user may perform, that is, an object of an operation instruction that may be issued by the user A. Specifically, in an e-commerce website, the demand object is at least one of a commodity that the user may bookmark, a commodity that the user may purchase, a commodity that the user may click to view, and a commodity that the user may add to a shopping cart.
Whether the user A may perform an action is determined according to historical behavior data of the user A.
For example, behavior data of the user A in the past seven days such as clicking, bookmarking, addition to the cart, searching, and purchase of commodities are collected based on a log of the website in the past seven days. A key product term and a brand term are extracted from the title of a commodity for which a historical action has occurred, to serve as a candidate demand commodity of the user. Different weights are assigned to different action modes. For example, the weight of the addition to the cart is 10, the weight of the bookmarking is 8, and the weight of the clicking is 5. Scores of the candidate demand commodities of the user are calculated according to action weights and action frequencies by using linear weighting, and commodities whose scores are lower than a score threshold are filtered out. Further, commodities that were purchased by the user in the past seven days can also be filtered out. The remaining commodities are demand objects of the user.
Optionally, after the candidate demand commodities of the user are determined in the foregoing example, weighted scoring may not be performed. Rather, commodities for which the action frequencies are lower than a threshold are filtered out from all the demand commodities of the user, and the remaining commodities are demand objects of the user.
S204: A recommendation generation module selects, from multiple pieces of UGC, a piece of UGC related to the demand object of the user A as a candidate piece of UGC.
The multiple pieces of UGC include high-quality pieces of UGC selected from pieces of UGC received by a website. In this embodiment, any piece of UGC in the high-quality pieces of UGC is a piece of UGC that includes key attributes of a target object and has preset sentiment word features. The target object is an object which the high-quality piece of UGC concerns. A high-quality piece of UGC from a user has reference significance to other users.
By taking an e-commerce website as an example, a high-quality piece of UGC is “It seems that Huang Xiaoniu is of little use in removing blackheads but has a really good skin care effect. It is easy to disperse and absorb and is not greasy. One or two drops can prevent the skin from being dry and tight the whole day. I had to like it.”
A non-high quality piece of UGC is “The commodity is of good quality and fast delivery. The seller's service and attitude are good.”
It can be seen that the high-quality piece of UGC includes key attributes “It is easy to disperse and absorb and is not greasy. Prevent from being dry and tight” of the commodity “Huang Xiaoniu” and sentiment word features “It has a really good skin care effect. I had to like it.” The non-high quality piece of UGC does not include key attributes and sentiment word features.
The multiple pieces of UGC can be included in a UGC library. The multiple pieces of UGC or the UGC library are/is created by the high-quality UGC mining module in
First, a piece of UGC received by a website is pre-processed. The pre-processing includes, but is not limited to, word segmentation and word-type marking. Then, key attributes and sentiment word features are extracted from the pre-processed UGC. Optionally, basic features and industry features may also be extracted from the pre-processed UGC.
In particular, the key attributes of the object are key attributes of a category to which the object belongs, which can be preset. Different categories have different key attributes. For example, key attributes of the category women's wear are fabric, color, and so on. Key attributes of the category cosmetics are color fastness and so on.
As shown by the dashed box in
Sentiment word features are terms included in a preset sentiment word dictionary. Generally, the sentiment word dictionary includes positive words, such as very satisfied, excellent value for money, and the like, as well as negative words, such as shedding, swelling, and the like. The specific manner of extracting sentiment word features from the pre-processed UGC is extracting terms that belong to the sentiment word dictionary from the pre-processed UGC.
Basic features include, but are not limited to, sentence sentiment polarity, repetition of a text fragment, sentence length, correlation between the text and the object, similarity between the text and another text, user ratings, number of likes, and so on. In particular, the sentiment polarity refers to a sentiment classification, usually divided into three classifications (positive, negative and neutral). The sentence sentiment polarity is obtained by predicting a sentence based on a common sentiment analysis technology.
The industry features include, but are not limited to, various key attributes and attribute values given in the industry.
The key attributes and the sentiment word features extracted in the foregoing, optionally further including the basic features and the industry features, are input to a trained support vector machine (SVM) to obtain an evaluation value of the UGC. Specifically, the SVM is a linear model as shown in the formula (1), the output evaluation value is the product of a feature vector X and a weight vector W, and the range of the evaluation value is [0, 1].
score=W*X (1)
In particular, X denotes key attributes and sentiment word features, and optionally further includes basic features and industry features. The weight W of each feature is obtained by training the SVM in advance. In the process of training the SVM, features of an input sample include key attributes and sentiment word features, and optionally further include basic features and industry features. Training methods can be referenced to an existing technology.
After the score of a UGC is obtained, it is judged whether the score is greater than a preset threshold. If yes, the UGC is added into a UGC library; otherwise, the UGC is discarded.
The manner of obtaining an evaluation value by using a SVM in this embodiment is not the sole manner of determining the evaluation value, and the evaluation value may also be obtained according to the formula (1) in another manner.
Optionally, the high-quality UGC mining module can also perform a further selection of the multiple pieces of UGC or the pieces of UGC in the UGC library, that is, determine according to a log of a website whether a piece of UGC among the multiple pieces of UGC or the UGC library is shared by another user or has brought backflow (if the user A enters the e-commerce website through sharing of another user, it is referred to as backflow), and if no, delete the piece of UGC from the multiple pieces of UGC or the UGC library to reduce the data volume of the multiple pieces of UGC or the UGC library and increase the subsequent selection speed. Moreover, the quality of the UGC library and its appeal to users are further enhanced.
In S204, the piece of UGC in the UGC library which is related to the demand of the user A is a piece of UGC that is related to an object included in the user's demand. For example, if the demand of the user A is “lipstick”, the piece of UGC that is related to the demand of the user A is a piece of UGC whose content involves “lipstick”.
Optionally, the piece of UGC that is related to the demand of the user A does not include a piece of UGC of the user A, so that a commodity that has not been purchased by the user before can be recommended to the user, to increase the user's purchase probability.
Optionally, the piece of UGC that is associated with the demand of the user A can include a piece of UGC of the user A, that is, the high-quality piece of UGC created by the user A is pushed back to the user A, to promote a second purchase.
S206: A message pushing module pushes the candidate piece of UGC to the user A.
Specifically, an active time period of the user A can be determined according to historical behaviors of the user A, and information is pushed in the time period when the user A is relatively active. If the historical behaviors of the user A are sparse, the information is pushed in a fixed time period. The fatigue of the user A can also be calculated according to opening of messages by the user, to control the message pushing frequency.
It can be seen from the process shown in
S402: A user demand mining module determines a demand object of a user A according to historical behavior data of the user A.
S404: A personalized matching module determines a portrait of the user A.
Specifically, the portrait of the user is labels of a preference of a user calculated according to demographic information and historical behavior data of the user registered in a website, which include, but are not limited to, gender, age, purchasing power, attribute preferences, and the like.
For example, the portrait of the user A is female, having high purchasing power, and having a preference for forest style.
S406: A recommendation generation module selects, from multiple pieces of UGC, pieces of UGC related to the demand object of the user A.
S408: The recommendation generation module selects a piece of UGC matched with the portrait of the user A from the pieces of UGC related to the demand object of the user A, to serve as a candidate piece of UGC.
For example, if the demand of the user A is a one-piece dress and the portrait of the user A is female, having high purchasing power, and having a preference for forest style, the candidate piece of UGC is a piece of UGC made for a one-piece dress by a female user who has high purchasing power and a preference for forest style and/or a piece of UGC made by a female user for a one-piece dress with a high price and in a forest style.
S410: A user label relation calculation module determines a user label of the candidate piece of UGC.
In this embodiment, the user label includes, but is not limited to, a capability label and a relation label. The capability label refers to an experience level of the user in a preset field, for example, “digital expert”, “mother”, “fashionable man” and so on. The relation label refers to a relation between the user of the candidate piece of UGC (that is, the user who generated the candidate piece of UGC) and the user A, for example, “Taobao™ friends”, “users of the same stature”, and so on.
SS412: The recommendation generation module generates an information pushing list according to the candidate piece of UGC and the user label of the candidate piece of UGC.
All the candidate pieces of UGC in the information pushing list can be scored according to various objects based on a preset rule and are sorted according to the scores. Each piece of UGC carries a user label of the UGC.
S414: A message pushing module pushes the information pushing list to the user A.
The method shown in
The information pushing method in this embodiment may push to a user purchase evaluation of other users (the user himself/herself may also be included), thus increasing the credibility of the recommended content. Moreover, it is costly for the user to find the real experience content among massive quantities of commodity UGC content, and the content is likely to be missed. However, the method provided in the present application selects pieces of UGC, which thus helps the user to save the decision-making cost. Further, the pieces of UGC provide information in more dimensions from the perspective of users, which is an advantage that the existing direct commodity recommendation does not have.
Further, in the process shown in
In combination with the process shown in
Pushing the simplified content of the piece of UGC to the user not only can save the quantity of data transmitted but also can help users to understand the pushed content more efficiently. A user can click the simplified content of the piece of UGC to further understand all the content of the piece of UGC if the user is interested in it.
An embodiment of the present application further discloses an information pushing system 600.
The information pushing system 600 may further include one or more processors 610, an input/output (I/O) interface 612, a network interface 614, and memory 618. The memory 618 is configured to store an application and data generated during execution of the application. The processor 610 is configured to execute the application stored in the memory to realize the processes shown in
An embodiment of the present application further discloses a computer readable storage medium, wherein the computer readable storage medium stores instructions which, when running on a computer, enable the computer to perform the processes shown in
The memory 618 may include a form of computer readable media medias described in the foregoing description. In implementations, the memory 618 may include program modules 620 and program data 622. The program modules 620 may include one or more of the modules as described in above.
When the functions in the methods according to the embodiments of the present application are implemented in the form of a software functional unit and sold or used as an independent product, the product may be stored in a computing device readable storage medium. Based on such an understanding, the part of the embodiments of the present application contributing to the prior art ora part of the technical solutions may be embodied in a form of a software product. The software product is stored in a storage medium and includes several instructions for instructing a computing device (which may be a personal computer, a server, a mobile computing device, a network device, or the like) to perform all or a part of the operations of the methods described in the embodiments of the present application.
The embodiments in the specification are described progressively, each embodiment emphasizes a part different from other embodiments, and identical or similar parts of the embodiments may be obtained by reference to each other.
The above descriptions about the disclosed embodiments enable those skilled in the art to implement or use the present application. A variety of modifications to the embodiments will be obvious for those skilled in the art. General principles defined in this text can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown in this text and will be in line with the broadest scope consistent with the principles and novelties disclosed in this text.
Number | Date | Country | Kind |
---|---|---|---|
201710501465.9 | Jun 2017 | CN | national |