CN105701680A - Individualized recommendation method and system based on opposite attribute knowledge base - Google Patents
Individualized recommendation method and system based on opposite attribute knowledge base Download PDFInfo
- Publication number
- CN105701680A CN105701680A CN201511034894.7A CN201511034894A CN105701680A CN 105701680 A CN105701680 A CN 105701680A CN 201511034894 A CN201511034894 A CN 201511034894A CN 105701680 A CN105701680 A CN 105701680A
- Authority
- CN
- China
- Prior art keywords
- attribute
- user
- contrary
- recommendation results
- recommendation
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
- G06Q30/0256—User search
Abstract
The invention discloses an individualized recommendation method and system. The method includes the steps of: obtaining a recommendation result sequence that a current recommendation system recommends to a user; obtaining a preset recommendation number of recommendation results in a preset direction in the recommendation result sequence as a initial recommendation result sequence; inquiring in an opposite attribute knowledge base whether an opposite attribute of the user is stored according to identity information of the user; if a query result is yes, matching each recommendation result in the initial recommendation result sequence with the opposite attribute of the user; deleting recommendation results in the initial recommendation result sequence whose matching results with the opposite attribute of the user satisfy preset conditions; obtaining a final recommendation result sequence according to remaining recommendation results in the initial recommendation result sequence; and outputting the final recommendation result sequence. The individualized recommendation method improves the accuracy rate of recommendation to the user, thereby improving the adoption rate of the recommendation results by the user, and raising the value of a recommendation system to the user.
Description
Technical field
The present invention relates to recommended technology field, particularly relate to a kind of personalized recommendation method based on contrary attribute knowledge base and system。
Background technology
Along with the continuous expansion of ecommerce scale, commodity number and kind quickly increase, and user requires a great deal of time and just can find the commodity oneself wanting to buy。The process Hui Shi consumer undoubtedly browsing a large amount of irrelevant information and product constantly runs off。In order to solve these problems, personalized recommendation technology is arisen at the historic moment。Personalized recommendation technology is built upon mass data and excavates a kind of Advanced Business intelligent platform on basis, to help e-commerce website to provide completely personalized decision support and information service for its customer purchase。
But existing personalized recommendation system user buy commodity historical data analysis foundation on recommend time, in fact it could happen that mistake recommend。Such as, it is recommended that conventional Characteristic of Interest and the purchasing behavior of system discovery party A-subscriber and party B-subscriber are all much like, and nearest party A-subscriber have purchased sanitary towel, result commending system just is recommended that sanitary towel give party B-subscriber, and whether this recommendation is accurate？Party A-subscriber and party B-subscriber why within the time in past Characteristic of Interest and purchasing behavior all much like, it is because party A-subscriber and party B-subscriber is parent elder sister and younger brother, but party A-subscriber is women, nearest menophania in the period, so starting first time to buy sanitary towel, but party B-subscriber is male, and sanitary towel is recommended party B-subscriber, it is clear that be the recommendation of mistake。Visible, the recommendation results that existing recommended technology obtains usually wants with user that the commodity bought are misfitted, the recommendation led to errors, and then reduces user's rate of adopting to recommendation results, reduces the commending system value to user。
Summary of the invention
Based on above-mentioned situation, the present invention proposes a kind of personalized recommendation method and system, improves the accuracy rate that user is recommended, and then improves user's rate of adopting to recommendation results, promotes the commending system value to user。
To achieve these goals, the embodiment of technical solution of the present invention is:
A kind of personalized recommendation method, comprises the following steps:
Obtain the recommendation results sequence that current commending system is recommended to user；
Obtaining the several recommendation results of default recommendation of preset direction in described recommendation results sequence as first recommendation results sequence, described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Identity information according to described user inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user；
When Query Result is for being, respectively each recommendation results in described first recommendation results sequence is mated with the contrary attribute of described user；
Delete in described first recommendation results sequence the matching result with the contrary attribute of described user and meet pre-conditioned recommendation results；
Consequently recommended result sequence is obtained according to the described first remaining recommendation results of recommendation results sequence；
Export described consequently recommended result sequence。
A kind of personalized recommendation system, including:
Recommendation results retrieval module, for obtaining the recommendation results sequence that current commending system is recommended to user；
First recommendation results retrieval module, the individual several recommendation results of default recommendation for obtaining preset direction in described recommendation results sequence are as first recommendation results sequence, and described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Attribute query module, inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user for the identity information according to described user；
Result matching module, for when Query Result is for being, mating each recommendation results in described first recommendation results sequence with the contrary attribute of described user respectively；
Result removing module, meets pre-conditioned recommendation results for the matching result deleted in described first recommendation results sequence with the contrary attribute of described user；
Consequently recommended result retrieval module, for obtaining consequently recommended result sequence according to the described first remaining recommendation results of recommendation results sequence；
Sequence output module, is used for exporting described consequently recommended result sequence。
Compared with prior art, the invention have the benefit that personalized recommendation method of the present invention and system, based on contrary attribute knowledge base, by predetermined number the recommendation results that current commending system is recommended to user is mated with the contrary attribute of user being stored in advance in contrary attribute knowledge base, consequently recommended result sequence is obtained according to matching result, improve the accuracy rate that user is recommended, meet the personalized recommendation needs of user, improve user and recommendation results adopted rate, promote the commending system value to user, be suitable for application。
Accompanying drawing explanation
Fig. 1 is personalized recommendation method schematic flow sheet in one embodiment of the invention；
Fig. 2 is for based on personalized recommendation method flow chart in the concrete example of method one shown in Fig. 1；
Fig. 3 is personalized recommendation system structural representation in one embodiment of the invention。
Detailed description of the invention
For making the purpose of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is described in further detail。Should be appreciated that detailed description of the invention described herein is only in order to explain the present invention, does not limit protection scope of the present invention。
Personalized recommendation method in one embodiment, as it is shown in figure 1, comprise the following steps:
Step S101: obtain the recommendation results sequence that current commending system is recommended to user；
Wherein, current commending system can be existing various commending system, it is also possible to be commending system newly developed；Commending system can be various types of recommendation results to the recommendation results that user recommends, such as, and the recommendation etc. of the recommendation of commodity, the recommendation of clothes, the recommendation of books, the recommendation of video, the recommendation of picture, the recommendation of paper or good friend；
Step S102: obtaining the several recommendation results of default recommendation of preset direction in described recommendation results sequence as first recommendation results sequence, described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Such as commending system is p to the recommendation results number scale that a user recommends, using front n the recommendation results in this p recommendation results as n the first recommendation results, obtain first recommendation results sequence, wherein, p can be natural number, commending system can recommend at least one recommendation results to user, and that chooses in all recommendation results that commending system is recommended to a user is all or part of as the first recommendation results；
Step S103: inquire about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user according to the identity information of described user；
The contrary attribute list of user from contrary attribute knowledge base such as, is retrieved the contrary attribute of this user, by the identity information of user, the contrary attribute list of user is retrieved, when retrieving the identity information of relative users, then take out the contrary attribute of user corresponding to the identity information of this user；The contrary attribute that can gather user in advance is stored in contrary attribute knowledge base；
Step S104: when Query Result is for being, mates each recommendation results in described first recommendation results sequence with the contrary attribute of described user respectively；
One recommendation results and the matching degree of the contrary attribute of this user in contrary attribute knowledge base, be substantially the contradiction degree of this recommendation results and user property；Size from a recommendation results with the matching degree of the contrary attribute of this user, can be seen that the contradiction degree of this recommendation results and the attribute of this user, one recommendation results is more big with the matching degree of the contrary attribute of this user, then show that this recommendation results is more high with the contradiction degree of the attribute of this user；
Step S105: delete in described first recommendation results sequence the matching result with the contrary attribute of described user and meet pre-conditioned recommendation results；
Such as when a recommendation results with the matching degree of the contrary attribute of this user more than matching degree preset value (matching degree preset value is minimum is 0) time, represent that the attribute of this recommendation results and this user is contradictory, delete this recommendation results；The calculating of described matching degree can be converted into the calculating of string matching degree or similarity, existing string matching degree or similarity algorithm can be adopted, such as Edit distance method (editing distance, just it is used to calculate the minimum insertion from required for former string (s) converting into target string (t), the number deleted and replace。Obviously when a statement editing is the minimum insertion needed for another statement, the number deleted and replace is more little, then matching degree is more big), maximum public substring LCS method (the maximum public substring of obvious two statements is more long, then the two statement matching degree is more big)；The calculating of described matching degree can also use the algorithm of new matching degree, such as using the public number of characters of two character strings size as matching degree；
Step S106: obtain consequently recommended result sequence according to the described first remaining recommendation results of recommendation results sequence；
More than matching degree preset value, the matching degree of one recommendation results and the contrary attribute of this user then shows that the attribute of this recommendation results and this user exists contradiction, using remaining recommendation results after deleting more than the recommendation results of matching degree preset value with the matching degree of the contrary attribute of this user in first recommendation results sequence as consequently recommended result sequence；
Step S107: export described consequently recommended result sequence。
Consequently recommended result being exported to the mode of user can be the mode that existing commending system adopts, it would however also be possible to employ other message stream mode, such as, such as the mode of the mode of webpage, file。
It is evidenced from the above discussion that, personalized recommendation method of the present invention, based on contrary attribute knowledge base, greatly eliminate recommendation results conflicting with user property, meet the needs of the personalized recommendation of user, improve the accuracy rate recommended, improve user and recommendation results adopted rate, improve the commending system value to user。
In addition, in a concrete example, the identity information of described user includes ID (identity number), the contrary attribute list of described user includes user field and the contrary attribute field of user, described user field stores ID, the contrary attribute field of described user stores the contrary attribute of user, the contrary attribute of described user obtains according to the attribute of described user, and the attribute of described user includes any one in the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position or combination in any。
The contrary attribute list of user in contrary attribute knowledge base includes user field, the contrary attribute field of user, stores ID, the contrary attribute of user's contrary attribute field storage user in user field。From contrary attribute knowledge base, retrieve the contrary attribute of this user, be by ID, contrary attribute knowledge base is retrieved, when retrieving relative users ID, then take out the contrary attribute of user corresponding to this ID。The contrary attribute of user obtains according to the attribute of user, and the attribute of user can include the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position etc. and user-dependent information, meets multiple application needs。
Obtain the contrary attribute step of user: first inquire about the antonym of keyword in the attribute of user；When antonym can be inquired, using this antonym contrary attribute as user；When antonym can not be inquired, inquire about in data base apart from the farthest same type keyword of the described keyword contrary attribute as user according to keyword in the attribute of user。Wherein, being previously stored all kinds key word and distance between in data base, distance here refers to diversity, for instance, the key word being all educational background type is clearly " post-doctor " from " primary school " apart from farthest。
Additionally, in a concrete example, when Query Result is no, it is judged that whether described user is the registration user of described current commending system；
When result of determination is for being, the attribute of described user is obtained from the log-on message of the described user of described current commending system, attribute according to described user obtains the contrary attribute of described user, is stored in described contrary attribute knowledge base by the contrary attribute of described user；
When result of determination is no, generates an information gathering window, gather the attribute of described user, obtain the contrary attribute of described user according to the attribute of described user, the contrary attribute of described user is stored in described contrary attribute knowledge base。
From contrary attribute knowledge base, such as retrieve the contrary attribute of this user, when retrieving the contrary attribute less than this user or this user from contrary attribute knowledge base, then judge that whether user is the registration user of commending system, when user is registration user, then inquire about the user property in the log-on message of user, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base, when user is not registration user, then pop-up dialogue box inquiry user, can also be that other interactive modes obtain or inquiry mode obtains the attribute of this user, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base, if the log-on message of user does not have customer attribute information, can also inquire that user or other interactive modes obtain the attribute of this user by pop-up dialogue box, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base。
Additionally, in a concrete example, respectively the step that each recommendation results in described first recommendation results sequence carries out mating with the contrary attribute of described user is included:
Respectively the contrary attribute of each recommendation results in described first recommendation results sequence and described user is converted into character string；
Calculate the matching degree of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively。
Each recommendation results in first recommendation results sequence can be converted into into character string with the contrary attribute of user, calculate the calculating that can be converted into string matching degree or similarity of both matching degrees, can be seen that the contradiction degree of this recommendation results and the attribute of this user from the size of a recommendation results with the matching degree of the contrary attribute of this user, a recommendation results is more big with the matching degree of the contrary attribute of this user, shows that this recommendation results is more high with the contradiction degree of the attribute of this user。
The calculating of described matching degree can be converted into the calculating of string matching degree or similarity, existing string matching degree or similarity algorithm can be adopted, such as Edit distance method (editing distance, just it is used to calculate the minimum insertion from required for former string (s) converting into target string (t), the number deleted and replace。Obviously when a statement editing is the minimum insertion needed for another statement, the number deleted and replace is more little, then matching degree is more big), maximum public substring LCS method (the maximum public substring of obvious two statements is more long, then the two statement matching degree is more big)；The calculating of described matching degree can also use the algorithm of new matching degree, such as using the public number of characters of two character strings size as matching degree。
Additionally, in a concrete example, delete the step meeting pre-conditioned recommendation results in described first recommendation results sequence with the matching result of the contrary attribute of described user and include:
Obtain the identical characters number of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively；
Delete the identical characters number recommendation results more than preset value of the character string that attribute contrary to described user in described first recommendation results sequence converts。
The character string identical characters number that the character string attribute contrary to this user of one recommendation results conversion converts is more than preset value (being greater than 0), attribute contradictory of this recommendation results and this user is described, delete and the conflicting recommendation results of user property, improve the accuracy rate recommended。
In order to be more fully understood that said method, the application example of a personalized recommendation method of the present invention detailed below。
As in figure 2 it is shown, this application example may comprise steps of:
Step S201: obtain the recommendation results sequence that the commending system of a shopping website is recommended to user's first；
Step S202: obtaining front 11 recommendation results in above-mentioned recommendation results sequence as first recommendation results sequence, the recommendation results sum in above-mentioned recommendation results sequence is more than or equal to 11；Described 11 recommendation results are: (1) Marubi sunscreen cream female's waterproof certified products swashs white sun-proof elite isolation breast SPF30 anti-ultraviolet whole body 45g；(2) trendy big tennis shoes man's sandals student's athleisure shoe man's screen cloth in the summer footwear man's intensity code Men's Shoes of bag postal；(3) in, song Gionee S7 cell-phone cover ELIFE7 shell GN9006 transparent silica gel protects soft set housing accessory bonnet tide；(4) iphone4s mobile phone shell Fructus Mali pumilae 5s shell super-thin plastic frosted protection duricrust black and white red tide men and women is brief；(5) bag postal man bag adds sailcloth both shoulders bag man's handbag leisure travelling Bao Chaonan bag Korea Spro's version man's knapsack；(6) stamp no-sleeves shirt vest in summer suspender belt 8520300114 is worn outside graceful 2015 summer clothing of mattress trendy vest female's summer；(7) gloomy paddy bird Korea Spro's version tide 2015 female's canvas shoe foot-high shoes high side in spring and autumn increases women's shoes thickness base fabric shoes；(8) big sim Korea S customization money summer clothing indispensability broken hole pure color brief crew neck loose short sleeve shirt female's T-shirt；(9) person in middle and old age female's money summer clothing T-shirt chiffon shirt jacket big code mother fills loose embroidery cotta old people's clothes；(10) after Semen setariae 2s mobile phone protecting case two s, cover mobile phone sleeve tide Semen setariae 2 leather sheath shell m2 is ultra-thin renovates bag postal firmly；(11) Korea S's Dongdaemun 2015 newly goes up women's dress fashion design of scattered small flowers and plants loose short sleeve shirt chiffon cake shirt shortage of money jacket summer；
Step S203: inquire about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing user's first according to the ID of user's first；The contrary attribute list of described user includes user field and the contrary attribute field of user, described user field stores ID, storing the contrary attribute of user in the contrary attribute field of described user, the contrary attribute of user obtains according to the attribute of user, and the attribute of user includes age and the sex of user；Contrary attribute knowledge base can be previously stored the contrary attribute of user；In one embodiment, user attribute table is as shown in table 1, and user is contrary, and attribute list is as shown in table 2；
Table 1 user attribute table
| ID | The attribute of user |
| 14233 | Old male |
| 14234 | Young woman |
| 14235 | Young men |
| 14236 | Young women |
The contrary attribute list of table 2 user
Step S204: when Query Result is for being, is converted into character string by the contrary attribute of above-mentioned 11 recommendation results and user's first respectively；When Query Result is no, it is judged that whether user's first is the registration user of above-mentioned shopping website；When result of determination is for being, from the log-on message of user's first of above-mentioned shopping website, obtains the attribute of user's first, obtain the contrary attribute of user's first according to the attribute of user's first, the contrary attribute of user's first is stored in contrary attribute knowledge base；When result of determination is no, generate an information gathering window, gather the attribute of user's first, obtain the contrary attribute of user's first according to the attribute of user's first, the contrary attribute of user's first is stored in contrary attribute knowledge base；
The ID of known users first is 14235, it is possible to the contrary attribute inquiring user's first from the contrary attribute list of user that above-mentioned contrary attribute knowledge base prestores is " young women "；
If the contrary attribute list of user that above-mentioned contrary attribute knowledge base prestores is inquired about the contrary attribute less than user's first, then judge that whether user's first is the registration user of above-mentioned shopping website, when user's first is registration user, then inquire about the attribute of user's first in the log-on message of user, attribute according to user's first obtains the contrary attribute of user's first and adds contrary attribute knowledge base, when user's first is not registration user, then pop-up dialogue box inquiry user's first, can also be the acquisition of other interactive modes or the attribute of inquiry mode acquisition user's first, attribute according to user's first obtains the contrary attribute of user's first and adds contrary attribute knowledge base, if the log-on message of user does not have the attribute of user's first, pop-up dialogue box inquiry user or the attribute of other interactive modes acquisition user's first can also be passed through, attribute according to user's first obtains the contrary attribute of user's first and adds contrary attribute knowledge base；
The attribute of user's first is " young men ", keyword " youth ", " male ", and inquiry obtains the antonym " old " of above-mentioned keyword, " women ", by " young women " contrary attribute as user's first；
Step S205: calculate the matching degree of the character string of the character string attribute conversion contrary to user's first that above-mentioned 11 recommendation results convert respectively；
The calculating of described matching degree can be converted into the calculating of string matching degree or similarity, it is also possible to using the public number of characters of two character strings size as matching degree；The contradiction degree of this recommendation results and the attribute of this user is can be seen that from the size of a recommendation results with the matching degree of the contrary attribute of this user；One recommendation results is more big with the matching degree of the contrary attribute of this user, shows that this recommendation results is more high with the contradiction degree of the attribute of this user；
Step S206: obtain the identical characters number of the character string of the character string attribute conversion contrary to user's first that above-mentioned 11 recommendation results convert respectively；
The calculation that matching degree adopts: using the identical number of characters of two character strings size as matching degree:
(1) Marubi sunscreen cream female waterproof certified products swashs the identical characters number of the character string that white sun-proof elite isolation breast SPF30 anti-ultraviolet whole body 45g young women attribute contrary to user's first converts is 1；
(2) the identical characters number of the character string that trendy big tennis shoes man's sandals student's athleisure shoe man's screen cloth in the summer footwear man's intensity code Men's Shoes young women of bag postal attribute contrary to user's first converts is 0；
(3) in, song Gionee S7 cell-phone cover ELIFE7 shell GN9006 transparent silica gel protects the identical characters number of the character string of soft set housing accessory bonnet tide young women attribute conversion contrary to user's first to be 0；
(4) the identical characters number of the character string that the iphone4s mobile phone shell Fructus Mali pumilae 5s shell super-thin plastic frosted protection duricrust black and white brief young women of red tide men and women attribute contrary to user's first converts is 1；
(5) bag postal man bag adds the identical characters number of the character string that sailcloth both shoulders bag man's handbag leisure travelling Bao Chaonan bag Korea Spro's version man's knapsack young women attribute contrary to user's first converts is 0；
(6) the identical characters number wearing the character string that stamp no-sleeves shirt vest in summer suspender belt 8520300114 young women attribute contrary to user's first converts outside graceful 2015 summer clothing of mattress trendy vest female's summer is 1；
(7) gloomy paddy bird Korea Spro's version tide 2015 female's canvas shoe foot-high shoes high side in spring and autumn increases the identical characters number of the character string that women's shoes thickness base fabric shoes young women attribute contrary to user's first converts is 2；
(8) the identical characters number of the character string that big sim Korea S customization money summer clothing indispensability broken hole pure color brief crew neck loose short sleeve shirt female's T-shirt young women attribute contrary to user's first converts is 1；
(9) person in middle and old age female's money summer clothing T-shirt chiffon shirt jacket big code mother fills the identical characters number of the character string that loose embroidery cotta old people's clothes young women attribute contrary to user's first converts is 3；
(10) after Semen setariae 2s mobile phone protecting case two s, the ultra-thin identical characters number firmly renovating the character string that bag postal young women attribute contrary to user's first converts of cover mobile phone sleeve tide Semen setariae 2 leather sheath shell m2 is 0；
(11) Korea S's Dongdaemun 2015 newly goes up the identical characters number of the character string that women's dress fashion design of scattered small flowers and plants loose short sleeve shirt chiffon cake shirt shortage of money jacket young women attribute contrary to user's first converts summer is 1；
Step S207: delete the recommendation results that in above-mentioned 11 recommendation results, the identical characters number of the character string that attribute contrary to user's first converts is not zero；
Namely delete:
Marubi sunscreen cream female's waterproof certified products swashs white sun-proof elite isolation breast SPF30 anti-ultraviolet whole body 45g；
Protection duricrust black and white red tide men and women is brief in iphone4s mobile phone shell Fructus Mali pumilae 5s shell super-thin plastic frosted；
Stamp no-sleeves shirt vest in summer suspender belt 8520300114 worn by mattress outside graceful 2015 summer clothing trendy vest female's summer；
Gloomy paddy bird Korea Spro's version tide 2015 female's canvas shoe foot-high shoes high side in spring and autumn increases women's shoes thickness base fabric shoes；
Big sim Korea S customization money summer clothing indispensability broken hole pure color brief crew neck loose short sleeve shirt female's T-shirt；
Person in middle and old age female money summer clothing T-shirt chiffon shirt jacket big code mother fills loose embroidery cotta old people's clothes；
Korea S's Dongdaemun 2015 newly goes up women's dress fashion design of scattered small flowers and plants loose short sleeve shirt chiffon cake shirt shortage of money jacket summer；
Step S208: obtain consequently recommended result sequence according to above-mentioned 11 remaining recommendation results of recommendation results sequence:
(1) trendy big tennis shoes man's sandals student's athleisure shoe man's screen cloth in the summer footwear man's intensity code Men's Shoes of bag postal；
(2) in, song Gionee S7 cell-phone cover ELIFE7 shell GN9006 transparent silica gel protects soft set housing accessory bonnet tide；
(3) bag postal man bag adds sailcloth both shoulders bag man's handbag leisure travelling Bao Chaonan bag Korea Spro's version man's knapsack；
(4) after Semen setariae 2s mobile phone protecting case two s, cover mobile phone sleeve tide Semen setariae 2 leather sheath shell m2 is ultra-thin renovates bag postal firmly；
Step S209: export above-mentioned consequently recommended result sequence。
Consequently recommended result being exported to the mode of user can be the mode that existing commending system adopts, it would however also be possible to employ other message stream mode, such as, such as the mode of the mode of webpage, file。
This application example the recommendation results in 11 recommendation results with the matching degree of the contrary attribute of user's first not being 0 is deleted after remaining recommendation results as consequently recommended result, greatly eliminate recommendation results conflicting with user property, meet the needs of the personalized recommendation of user, improve the accuracy rate recommended, improve user and recommendation results adopted rate, improve the commending system value to user。
Personalized recommendation system in one embodiment, as it is shown on figure 3, include:
Recommendation results retrieval module 301, for obtaining the recommendation results sequence that current commending system is recommended to user；
First recommendation results retrieval module 302, the individual several recommendation results of default recommendation for obtaining preset direction in described recommendation results sequence are as first recommendation results sequence, and described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Attribute query module 303, inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user for the identity information according to described user；
Result matching module 304, for when Query Result is for being, mating each recommendation results in described first recommendation results sequence with the contrary attribute of described user respectively；
Result removing module 305, meets pre-conditioned recommendation results for the matching result deleted in described first recommendation results sequence with the contrary attribute of described user；
Consequently recommended result retrieval module 306, for obtaining consequently recommended result sequence according to the described first remaining recommendation results of recommendation results sequence；
Sequence output module 307, is used for exporting described consequently recommended result sequence。
In addition, in a concrete example, the identity information of described user includes ID, the contrary attribute list of described user includes user field and the contrary attribute field of user, described user field stores ID, the contrary attribute field of described user stores the contrary attribute of user, the contrary attribute of described user obtains according to the attribute of described user, and the attribute of described user includes any one in the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position or combination in any。
The contrary attribute list of user in contrary attribute knowledge base includes user field, the contrary attribute field of user, stores ID, the contrary attribute of user's contrary attribute field storage user in user field。From contrary attribute knowledge base, retrieve the contrary attribute of this user, be by ID, contrary attribute knowledge base is retrieved, when retrieving relative users ID, then take out the contrary attribute of user corresponding to this ID。The contrary attribute of user obtains according to the attribute of user, and the attribute of user can include the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position etc. and user-dependent information, meets multiple application needs。
Obtain the contrary attribute step of user: first inquire about the antonym of keyword in the attribute of user；When antonym can be inquired, using this antonym contrary attribute as user；When antonym can not be inquired, inquire about in data base apart from the farthest same type keyword of the described keyword contrary attribute as user according to keyword in the attribute of user。Wherein, being previously stored all kinds key word and distance between in data base, distance here refers to diversity, for instance, the key word being all educational background type is clearly " post-doctor " from " primary school " apart from farthest。
As it is shown on figure 3, in a concrete example, described system also includes attribute acquisition module 308, for when Query Result is no, it is judged that whether described user is the registration user of described current commending system；
When result of determination is for being, the attribute of described user is obtained from the log-on message of the described user of described current commending system, attribute according to described user obtains the contrary attribute of described user, is stored in described contrary attribute knowledge base by the contrary attribute of described user；
When result of determination is no, generates an information gathering window, gather the attribute of described user, obtain the contrary attribute of described user according to the attribute of described user, the contrary attribute of described user is stored in described contrary attribute knowledge base。
From contrary attribute knowledge base, such as retrieve the contrary attribute of this user, when retrieving the contrary attribute less than this user or this user from contrary attribute knowledge base, then judge that whether user is the registration user of commending system, when user is registration user, then inquire about the user property in the log-on message of user, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base, when user is not registration user, then pop-up dialogue box inquiry user, can also be that other interactive modes obtain or inquiry mode obtains the attribute of this user, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base, if the log-on message of user does not have customer attribute information, can also inquire that user or other interactive modes obtain the attribute of this user by pop-up dialogue box, attribute according to user obtains the contrary attribute of user and adds contrary attribute knowledge base。
As it is shown on figure 3, in a concrete example, described result matching module 304 includes:
Conversion unit 3041, for being converted into character string by the contrary attribute of each recommendation results in described first recommendation results sequence and described user respectively；
Matching unit 3042, the matching degree of the character string that the character string attribute contrary to described user converted for calculating each recommendation results in described first recommendation results sequence respectively converts。
Each recommendation results in first recommendation results sequence can be converted into into character string with the contrary attribute of user, calculate the calculating that can be converted into string matching degree or similarity of both matching degrees, can be seen that the contradiction degree of this recommendation results and the attribute of this user from the size of a recommendation results with the matching degree of the contrary attribute of this user, a recommendation results is more big with the matching degree of the contrary attribute of this user, shows that this recommendation results is more high with the contradiction degree of the attribute of this user。
The calculating of described matching degree can be converted into the calculating of string matching degree or similarity, existing string matching degree or similarity algorithm can be adopted, such as Edit distance method (editing distance, just it is used to calculate the minimum insertion from required for former string (s) converting into target string (t), the number deleted and replace。Obviously when a statement editing is the minimum insertion needed for another statement, the number deleted and replace is more little, then matching degree is more big), maximum public substring LCS method (the maximum public substring of obvious two statements is more long, then the two statement matching degree is more big)；The calculating of described matching degree can also use the algorithm of new matching degree, such as using the public number of characters of two character strings size as matching degree。
As it is shown on figure 3, in a concrete example, described result removing module 305 includes:
Acquiring unit 3051, the identical characters number of the character string that the character string attribute contrary to described user converted for obtaining each recommendation results in described first recommendation results sequence respectively converts；
Delete unit 3052, for deleting the identical characters number recommendation results more than preset value of the character string that attribute contrary to described user in described first recommendation results sequence converts。
The character string identical characters number that the character string attribute contrary to this user of one recommendation results conversion converts is more than preset value (being greater than 0), attribute contradictory of this recommendation results and this user is described, delete and the conflicting recommendation results of user property, improve the accuracy rate recommended。
Based on the system of the present embodiment shown in Fig. 3, a concrete work process can be discussed further below:
First recommendation results retrieval module 301 obtains the recommendation results sequence that current commending system is recommended to user；First recommendation results retrieval module 302 obtains the several recommendation results of default recommendation of preset direction as first recommendation results sequence in described recommendation results sequence, and described first recommendation number of presetting is less than or equal to the recommendation results sum in described recommendation results sequence；Attribute query module 303 inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user according to the identity information of described user；When Query Result is for being, the contrary attribute of each recommendation results in described first recommendation results sequence and described user is converted into character string by conversion unit 3041 in result matching module 304 respectively；Matching unit 3042 calculates the matching degree of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively；When Query Result is no, attribute acquisition module 308 judges that whether described user is the registration user of described current commending system；When result of determination is for being, the attribute of described user is obtained from the log-on message of the described user of described current commending system, attribute according to described user obtains the contrary attribute of described user, is stored in described contrary attribute knowledge base by the contrary attribute of described user；When result of determination is no, generates an information gathering window, gather the attribute of described user, obtain the contrary attribute of described user according to the attribute of described user, the contrary attribute of described user is stored in described contrary attribute knowledge base；Acquiring unit 3051 in result removing module 305 obtains the identical characters number of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively；Delete unit 3052 and delete the identical characters number recommendation results more than preset value of the character string that attribute contrary to described user in described first recommendation results sequence converts；Consequently recommended result retrieval module 306 obtains consequently recommended result sequence according to the described first remaining recommendation results of recommendation results sequence；Sequence output module 307 exports described consequently recommended result sequence。
It is evidenced from the above discussion that, personalized recommendation system of the present invention, based on contrary attribute knowledge base, greatly eliminate recommendation results conflicting with user property, meet the needs of the personalized recommendation of user, improve the accuracy rate recommended, improve user and recommendation results adopted rate, improve the commending system value to user。
Each technical characteristic of embodiment described above can combine arbitrarily, for making description succinct, the all possible combination of each technical characteristic in above-described embodiment is not all described, but, as long as the combination of these technical characteristics is absent from contradiction, all it is considered to be the scope that this specification is recorded。
Embodiment described above only have expressed the several embodiments of the present invention, and it describes comparatively concrete and detailed, but can not therefore be construed as limiting the scope of the patent。It should be pointed out that, for the person of ordinary skill of the art, without departing from the inventive concept of the premise, it is also possible to making some deformation and improvement, these broadly fall into protection scope of the present invention。Therefore, the protection domain of patent of the present invention should be as the criterion with claims。
Claims (10)
1. a personalized recommendation method, it is characterised in that comprise the following steps:
Obtain the recommendation results sequence that current commending system is recommended to user；
Obtaining the several recommendation results of default recommendation of preset direction in described recommendation results sequence as first recommendation results sequence, described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Identity information according to described user inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user；
When Query Result is for being, respectively each recommendation results in described first recommendation results sequence is mated with the contrary attribute of described user；
Delete in described first recommendation results sequence the matching result with the contrary attribute of described user and meet pre-conditioned recommendation results；
Consequently recommended result sequence is obtained according to the described first remaining recommendation results of recommendation results sequence；
Export described consequently recommended result sequence。
2. personalized recommendation method according to claim 1, it is characterized in that, the identity information of described user includes ID, the contrary attribute list of described user includes user field and the contrary attribute field of user, described user field stores ID, the contrary attribute field of described user stores the contrary attribute of user, the contrary attribute of described user obtains according to the attribute of described user, and the attribute of described user includes any one in the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position or combination in any。
3. personalized recommendation method according to claim 1 and 2, it is characterised in that when Query Result is no, it is judged that whether described user is the registration user of described current commending system；
When result of determination is for being, the attribute of described user is obtained from the log-on message of the described user of described current commending system, attribute according to described user obtains the contrary attribute of described user, is stored in described contrary attribute knowledge base by the contrary attribute of described user；
When result of determination is no, generates an information gathering window, gather the attribute of described user, obtain the contrary attribute of described user according to the attribute of described user, the contrary attribute of described user is stored in described contrary attribute knowledge base。
4. personalized recommendation method according to claim 1, it is characterised in that respectively the step that each recommendation results in described first recommendation results sequence carries out mating with the contrary attribute of described user is included:
Respectively the contrary attribute of each recommendation results in described first recommendation results sequence and described user is converted into character string；
Calculate the matching degree of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively。
5. personalized recommendation method according to claim 4, it is characterised in that delete the step meeting pre-conditioned recommendation results in described first recommendation results sequence with the matching result of the contrary attribute of described user and include:
Obtain the identical characters number of the character string of the character string attribute conversion contrary to described user that each recommendation results in described first recommendation results sequence converts respectively；
Delete the identical characters number recommendation results more than preset value of the character string that attribute contrary to described user in described first recommendation results sequence converts。
6. a personalized recommendation system, it is characterised in that including:
Recommendation results retrieval module, for obtaining the recommendation results sequence that current commending system is recommended to user；
First recommendation results retrieval module, the individual several recommendation results of default recommendation for obtaining preset direction in described recommendation results sequence are as first recommendation results sequence, and described default recommendation number is less than or equal to the recommendation results sum in described recommendation results sequence；
Attribute query module, inquires about, in the contrary attribute list of user that contrary attribute knowledge base prestores, the contrary attribute whether storing described user for the identity information according to described user；
Result matching module, for when Query Result is for being, mating each recommendation results in described first recommendation results sequence with the contrary attribute of described user respectively；
Result removing module, meets pre-conditioned recommendation results for the matching result deleted in described first recommendation results sequence with the contrary attribute of described user；
Consequently recommended result retrieval module, for obtaining consequently recommended result sequence according to the described first remaining recommendation results of recommendation results sequence；
Sequence output module, is used for exporting described consequently recommended result sequence。
7. personalized recommendation system according to claim 6, it is characterized in that, the identity information of described user includes ID, the contrary attribute list of described user includes user field and the contrary attribute field of user, described user field stores ID, the contrary attribute field of described user stores the contrary attribute of user, the contrary attribute of described user obtains according to the attribute of described user, and the attribute of described user includes any one in the age of user, sex, occupation, educational background, specialty, speciality, hobby and geographical position or combination in any。
8. the personalized recommendation system according to claim 6 or 7, it is characterised in that also include attribute acquisition module, for when Query Result is no, it is judged that whether described user is the registration user of described current commending system；
When result of determination is for being, the attribute of described user is obtained from the log-on message of the described user of described current commending system, attribute according to described user obtains the contrary attribute of described user, is stored in described contrary attribute knowledge base by the contrary attribute of described user；
When result of determination is no, generates an information gathering window, gather the attribute of described user, obtain the contrary attribute of described user according to the attribute of described user, the contrary attribute of described user is stored in described contrary attribute knowledge base。
9. personalized recommendation system according to claim 6, it is characterised in that described result matching module includes:
Conversion unit, for being converted into character string by the contrary attribute of each recommendation results in described first recommendation results sequence and described user respectively；
Matching unit, the matching degree of the character string that the character string attribute contrary to described user converted for calculating each recommendation results in described first recommendation results sequence respectively converts。
10. personalized recommendation system according to claim 9, it is characterised in that described result removing module includes:
Acquiring unit, the identical characters number of the character string that the character string attribute contrary to described user converted for obtaining each recommendation results in described first recommendation results sequence respectively converts；
Delete unit, for deleting the identical characters number recommendation results more than preset value of the character string that attribute contrary to described user in described first recommendation results sequence converts。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511034894.7A CN105701680B (en) | 2015-12-31 | 2015-12-31 | Personalized recommendation method and system based on opposite attribute knowledge base |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201511034894.7A CN105701680B (en) | 2015-12-31 | 2015-12-31 | Personalized recommendation method and system based on opposite attribute knowledge base |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105701680A true CN105701680A (en) | 2016-06-22 |
| CN105701680B CN105701680B (en) | 2020-08-25 |
Family
ID=56226854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201511034894.7A Active CN105701680B (en) | 2015-12-31 | 2015-12-31 | Personalized recommendation method and system based on opposite attribute knowledge base |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105701680B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111151858A (en) * | 2020-01-13 | 2020-05-15 | 吉利汽车研究院（宁波）有限公司 | Spot welding parameter application system and setting method |
| CN112269928A (en) * | 2020-10-23 | 2021-01-26 | 百度在线网络技术（北京）有限公司 | User recommendation method and device, electronic equipment and computer readable medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101329674A (en) * | 2007-06-18 | 2008-12-24 | 北京搜狗科技发展有限公司 | System and method for providing personalized searching |
| US20130091164A1 (en) * | 2011-10-11 | 2013-04-11 | Microsoft Corporation | Recommending data based on user and data attributes |
-
2015
- 2015-12-31 CN CN201511034894.7A patent/CN105701680B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101329674A (en) * | 2007-06-18 | 2008-12-24 | 北京搜狗科技发展有限公司 | System and method for providing personalized searching |
| US20130091164A1 (en) * | 2011-10-11 | 2013-04-11 | Microsoft Corporation | Recommending data based on user and data attributes |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111151858A (en) * | 2020-01-13 | 2020-05-15 | 吉利汽车研究院（宁波）有限公司 | Spot welding parameter application system and setting method |
| CN112269928A (en) * | 2020-10-23 | 2021-01-26 | 百度在线网络技术（北京）有限公司 | User recommendation method and device, electronic equipment and computer readable medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105701680B (en) | 2020-08-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220121701A1 (en) | Retrieval support system, retrieval support method and retrieval support program | |
| KR101886161B1 (en) | Method for providing clothing management service based on ai | |
| US20180308151A1 (en) | Enhancing revenue of a retailer by making a recommendation to a customer | |
| Vaccaro et al. | The elements of fashion style | |
| US20140032359A1 (en) | System and method for providing intelligent recommendations | |
| CN106649383A (en) | Clothes management method and system | |
| CN106204124A (en) | Personalized commercial coupling commending system and method | |
| CN106991189A (en) | A kind of wardrobe control method, device and wardrobe | |
| CN103337027A (en) | Clothing information pushing method and system | |
| KR102295459B1 (en) | A method of providing a fashion item recommendation service to a user using a date | |
| US20090281922A1 (en) | Method and system for selecting clothing items according to predetermined criteria | |
| CN105678575B (en) | Personalized recommendation method and system based on user attribute knowledge base | |
| CN105701680A (en) | Individualized recommendation method and system based on opposite attribute knowledge base | |
| CN106375544A (en) | Clothing collocation recommendation method and intelligent terminal | |
| CN111612584A (en) | AI intelligent clothing recommendation method based on wearing and putting-on theory | |
| CN111815407A (en) | Method and device for constructing user portrait | |
| JP2014229129A (en) | Combination presentation system and computer program | |
| JP6494837B1 (en) | Cluster generation device, cluster generation method and cluster generation program | |
| CN105677817B (en) | Personalized recommendation method and system based on positive and negative attribute knowledge base | |
| CN105512341B (en) | Personalized recommendation method and system based on big data search | |
| JP2014013460A (en) | Fashion coordinate service system and its method | |
| KR20200084647A (en) | A method of providing a fashion item recommendation service to a user using a date | |
| JP2020064496A (en) | Information processing system and program | |
| KR102178767B1 (en) | Fashion styling-color cosmetics matching system | |
| KR20230030870A (en) | Apparatus and method for apparel recommendation, computer-readable storage medium and computer program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |