CN105809479A - Item recommending method and device - Google Patents
Item recommending method and device Download PDFInfo
- Publication number
- CN105809479A CN105809479A CN201610127208.9A CN201610127208A CN105809479A CN 105809479 A CN105809479 A CN 105809479A CN 201610127208 A CN201610127208 A CN 201610127208A CN 105809479 A CN105809479 A CN 105809479A
- Authority
- CN
- China
- Prior art keywords
- article
- targeted customer
- interest
- user
- similarity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
An embodiment of the invention provides an item recommending method and device. The method includes determining the interest group of a target user according to history purchased items of the target user; obtaining the item similarity between the history purchased items of the target user and history purchased items of other users in the interest group of the target user; according to the item similarity, determining target recommendation items of the target user. By using the method, thereliability of item similarity is improved. Further, items are recommended to the target user according to the item similarity, so that the accuracy of item recommendation is improved and the recommended items are more suitable for the user. Therefore, user experience is improved. Besides, the interest group of the target user is determined first, and then item recommendation is performed based on data in the interest group, so that analysis and calculation for item similarity of all item data in the prior is saved and the calculation amount is reduced substantially.
Description
Technical field
The present invention relates to Internet technology, particularly relate to a kind of item recommendation method and device.
Background technology
Along with popularizing of the Internet, in user's life, a lot of things can carry out on the internet.Relatively conventional is exactly that user can choose some article, information etc. on the net, including: entity or virtual article are bought at shopping website, or all kinds of information etc. on selection the Internet, such as select the news type read, on smart machine, select all kinds of services etc., as the TV programme liked by intelligent television selection.Owing to the article on current internet, information are varied, user to spend the much time to look for the content that oneself needs, and in order to save the time of user, each internet platform all actively recommends some to meet the content of user preferences to user.
In prior art, during to user's content recommendation, the main recommendation method being based on article adopted, specifically, calculate the similarity of article, generate the recommendation list of corresponding user according to the similarity of article with user's historical behavior.Wherein, when prior art calculates the similarity of article, it is assumed that if two article are in the purchase list of multiple users, then the two article just belong to same field.
But, for multiple users, their interest might not be identical, adopts article possibly recommended in the recommendation list that prior art sends to user cannot meet the actual interest needs of user.
Summary of the invention
The present invention provides a kind of item recommendation method and device, for solving the problem that existing recommendation method cannot meet the actual interest needs of user.
First aspect present invention provides a kind of item recommendation method, including:
History according to targeted customer buys article, it is determined that the interest cluster belonging to described targeted customer；
Obtain the history of described targeted customer to buy the history of other user in interest cluster belonging to article and described targeted customer and buy the article similarity between article；
According to described article similarity, it is determined that the target of described targeted customer recommends article.
Second aspect present invention provides a kind of article recommendation apparatus, including:
Determine module, buy article for the history according to targeted customer, it is determined that the interest cluster belonging to described targeted customer；
First acquisition module, the history for obtaining described targeted customer is bought the history of other user in interest cluster belonging to article and described targeted customer and is bought the article similarity between article；
Recommending module, for according to described article similarity, it is determined that the target of described targeted customer recommends article.
In the item recommendation method of embodiment of the present invention offer and device, history according to targeted customer buys article, determine the interest cluster belonging to this targeted customer, obtain the history of above-mentioned targeted customer to buy the history of other user in interest cluster belonging to article and this targeted customer and buy the article similarity between article, according to above-mentioned article similarity, determine that the target of this targeted customer recommends article, namely the interest cluster belonging to targeted customer is first determined, then in affiliated interest cluster, the history of targeted customer is bought other user in article and this interest cluster and is bought the article similarity of article, improve the reliability of article similarity, and then recommend article further according to article similarity to targeted customer, also just improve the accuracy recommending article to user, the article recommended are closer to the demand of user, improve Consumer's Experience.And, first determine the interest cluster belonging to targeted customer, carry out article recommendation further according to the data in interest cluster, it is not necessary to as prior art, carry out the analytical calculation of article similarity for all of product data, be greatly reduced operand.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, the accompanying drawing used required in embodiment or description of the prior art will be briefly described below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skill in the art, under the premise not paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
Fig. 1 is the schematic flow sheet of item recommendation method embodiment one provided by the invention；
Fig. 2 is the schematic flow sheet of item recommendation method embodiment two provided by the invention；
Fig. 3 is the schematic flow sheet of item recommendation method embodiment three provided by the invention；
Fig. 4 is user network structural representation in item recommendation method provided by the invention；
Fig. 5 is the structural representation of article recommendation apparatus embodiment one provided by the invention；
Fig. 6 is the structural representation of article recommendation apparatus embodiment two provided by the invention；
Fig. 7 is the structural representation of article recommendation apparatus embodiment three provided by the invention.
Detailed description of the invention
Article described in the embodiment of the present invention can be entity or virtual article.
Fig. 1 is the schematic flow sheet of item recommendation method embodiment one provided by the invention, and the executive agent of the method can be backstage is the server that user recommends article.As it is shown in figure 1, the method includes:
S101, buy article according to the history of targeted customer, it is determined that the interest cluster belonging to this targeted customer.
Namely, in the present embodiment, first cluster is divided according to the purchase interest of user.Such as certain user likes buying cosmetics, certain user likes buying sports goods, they can be divided into different clusters, illustrate that in a cluster, the purchase interest of user is close, so follow-up further consider that the recommendation accuracy of article can be higher according to the user in cluster.
Specifically, it is possible to calculate the Interest Similarity between two users, the user that Interest Similarity meets predetermined threshold value is divided into same interest cluster.Wherein, a user can belong simultaneously to multiple interest cluster.
In interest cluster belonging to S102, the history purchase article obtaining above-mentioned targeted customer and this targeted customer, the history of other user buys the article similarity between article.
Wherein, the method obtaining article similarity is a lot, it is assumed that targeted customer bought article i, and other users bought commodity j, example, it is possible to adopt formulaCalculate the similarity w between article i and article j in above-mentioned interest clusterij.Wherein, N (i) represents that in above-mentioned interest cluster, history bought user's set of article i, and N (j) represents that in above-mentioned interest cluster, history bought user's set of article j.
All can buy it should be noted that the popular article having are likely to a lot of users, but it is interested in these popular article to represent user, in order to avoid the popular article impact on above-mentioned similarity accuracy, it is also possible to preferably select formulaCalculate the similarity w between article i and article j in above-mentioned interest clusterij。
Where it is assumed that article j is popular article, then N (i) is likely to be the subset of N (j), or the most of users in N (i) can find in N (j), if adopting formulaSo that other are popular not as article j article, article j is higher with their similarity, is thus unfavorable for finding real similar with article i article, because they may not necessarily obtain higher similarity.On the contrary, if adopting formulaSo article j is more popular, and denominator is also more big, if there being real similar with an article i article g, utilizes formulaWords, it is possible on molecule, | N (i) | is more or less the same with the value of | N (g) |, is all approximately equivalent to | N (i) |, owing to article j is more popular,Less thanwijIt is less than wig, also it may determine that the similarity going out article g and article i is higher.So more tally with the actual situation, because the user buying article i also have purchased article j, it is not for the interest to article j, more likely because article j inherently has the necessity of purchase, and article g not to be popular article can be bought by the people much buying article i, have bigger probability to be because article i and article g and can meet the interest of these groups of people.
S103, according to above-mentioned article similarity, it is determined that the target of this targeted customer recommends article.
Specifically, it is possible to according to above-mentioned article similarity, the article higher with the history of targeted customer purchase article similarity being recommended user, this is not restricted.When realizing, it is possible to generate article recommendation list and be sent to targeted customer.
In the present embodiment, history according to targeted customer buys article, determine the interest cluster belonging to this targeted customer, obtain the history of above-mentioned targeted customer to buy the history of other user in interest cluster belonging to article and this targeted customer and buy the article similarity between article, according to above-mentioned article similarity, determine that the target of this targeted customer recommends article, namely the interest cluster belonging to targeted customer is first determined, then in affiliated interest cluster, the history of targeted customer is bought other user in article and this interest cluster and is bought the article similarity of article, improve the reliability of article similarity, and then recommend article further according to article similarity to targeted customer, also just improve the accuracy recommending article to user, the article recommended are closer to the demand of user, improve Consumer's Experience.And, first determine the interest cluster belonging to targeted customer, carry out article recommendation further according to the data in interest cluster, it is not necessary to as prior art, carry out the analytical calculation of article similarity for all of product data, be greatly reduced operand.
Further, the above-mentioned history according to targeted customer buys article, determine the interest cluster belonging to this targeted customer, can be calculate the Interest Similarity between targeted customer and other user, determine the interest cluster belonging to this targeted customer according to the Interest Similarity between this targeted customer and other user.
Specifically, calculate the method for Interest Similarity between user a lot, assume given user u and user v, the history of user u buys the set of article, N (v) represents that the history of user v buys the set of article, at least can adopt a kind of Interest Similarity calculating user u and user v in following several method to make N (u) represent:
(1) formula is adoptedCalculate the Interest Similarity w of user u and user vuv；
(2) adopt cosine similarity computational methods, namely adopt formulaCalculate the Interest Similarity w of user u and user vuv；
(3) in order to avoid the popular article impact on Interest Similarity, it is possible to adopt formulaCalculate the Interest Similarity w of user u and user vuv, wherein, | N (h) | represents that user u and user's v history buy the common factor of article, and h is certain article h in the common factor of user u and user's v history purchase article,For alleviating the popular article impact on Interest Similarity in the interest cluster of user place.
Certainly, it is not limited with above-mentioned computational methods, if the Interest Similarity can expressed between user.
In a kind of optional scheme, above-mentioned Interest Similarity is divided into same interest cluster more than the corresponding user of the first predetermined threshold value and above-mentioned targeted customer.Interest cluster is divided together with other user by targeted customer.
In alternative dispensing means, other user is divided has got well interest cluster, need to determine an interest cluster for targeted customer, so can obtain targeted customer and the direct Interest Similarity of other user, if targeted customer and the first user Interest Similarity in other user are more than the second predetermined threshold value, then this targeted customer belongs to the interest cluster at first user place really.
Correspondingly, also needed to build at least one interest cluster before determining the interest cluster belonging to targeted customer.
Alternatively, build at least one interest cluster, particularly as follows: the Interest Similarity obtained between user, according to the similarity between user, be divided into same interest cluster by meeting pre-conditioned user.
For example, it is possible to Interest Similarity is divided into same interest cluster more than the user of certain predetermined threshold value, in this no limit.
Further, the above-mentioned history according to targeted customer buys article, it is determined that before the interest cluster belonging to this targeted customer, it is also possible to the purchase parameter according to other user, obtain the purchase liveness of other user above-mentioned；The sequence buying liveness according to other user, it is determined that this purchase liveness is from the first predetermined number user of high to low and buys liveness from low the second paramount predetermined number user for treating eliminating user.
Correspondingly, the Interest Similarity between above-mentioned calculating targeted customer and other user, it is possible to for: calculate in targeted customer and other user and get rid of the Interest Similarity before the above-mentioned row for the treatment of each user in addition to the user.
Alternatively, above-mentioned purchase parameter can be: history buys or combination in any in number of times, history purchase total amount, history purchase type of items number etc., and this is not restricted.
When determining the interest cluster belonging to user, user active especially and sluggish especially user are likely to impact and divide the accuracy of interest cluster, therefore just these certain customers were directly foreclosed before dividing interest cluster, user and sluggish especially user that this part is enlivened especially can be not present in arbitrary interest cluster, can also independently becoming interest cluster, this is not restricted.
Illustrate, according to buy each user of parameter acquiring liveness after, these users can be given sequence according to liveness, certain customers minimum to certain customers the highest for liveness and liveness are got rid of, such as, 10% minimum to 10% the highest for liveness in all users user and liveness user is got rid of.But it is not limited.
Fig. 2 is the schematic flow sheet of item recommendation method embodiment two provided by the invention, and on the basis of above-described embodiment, the above-mentioned history according to targeted customer buys article, it is determined that the interest cluster belonging to this targeted customer, it is possible to including:
S201, calculate in above-mentioned targeted customer and other user except the Interest Similarity between the above-mentioned row for the treatment of each user in addition to the user.
S202, Interest Similarity is divided into same interest cluster more than the corresponding user of the first predetermined threshold value and above-mentioned targeted customer.
Or S202 may be replaced by: if targeted customer and the first user Interest Similarity in other user are more than the second predetermined threshold value, then this targeted customer belongs to the interest cluster at first user place really.
Implementing in process, it is assumed that have n user, n is the integer more than 0, it is possible to obtain the article similarity matrix of n × n, and each node of matrix represents a Similarity value.
Further, after interest assemblage classification completes, each interest cluster can have an article similarity matrix, in order to recommend article more accurately, the article similarity matrix of each interest cluster can be normalized, it is assumed that have t interest cluster, for one of them interest cluster b, it is normalized according to maximum, it is possible to adopt formula:Calculate the article similarity matrix after interest cluster b normalized, wherein, wijRepresent article similarity current for article i and article j in interest cluster b, w 'ijRepresent the article similarity after article i and article j normalization, max (w in interest cluster bij)tFor the maximum of similarity between article i and article j in t interest cluster.
Fig. 3 is the schematic flow sheet of item recommendation method embodiment three provided by the invention.Fig. 4 is user network structural representation in item recommendation method provided by the invention.
The above-mentioned history according to targeted customer buys article, it is determined that before the interest cluster belonging to this targeted customer, it is also possible to the history periodically obtaining above-mentioned targeted customer buys article.Correspondingly, it is also possible to the history periodically obtaining other user buys article.
In real life, user likely buys new article at any time, in order to ensure the accuracy that article are recommended, it is possible to the history just obtaining once above-mentioned targeted customer of short time buys the history purchase article of article and other users as far as possible.Along with history buys the renewal of article, interest cluster is it can also happen that change.
Correspondingly, the above-mentioned history according to targeted customer buys article, it is determined that the interest cluster belonging to this targeted customer, specifically may is that the history according to targeted customer buys article, is updated periodically the interest cluster determined belonging to above-mentioned targeted customer.
Alternatively, the cycle updating the interest cluster belonging to this targeted customer can be longer than the cycle of the history purchase article obtaining above-mentioned targeted customer.
As it is shown on figure 3, this method specifically dividing interest cluster may include that
S301, initializing all users, each user each becomes an interest cluster.
Namely the most at first, only one of which user in each interest cluster, each user is an independent interest cluster.
Alternatively, all users here eliminate above-mentioned waiting to get rid of the user after user.
If S302, calculating each user respectively and be divided into the yield value that other each interest cluster obtains.
Specifically, it is possible to calculate each user and be divided into the gain of other each interest cluster, it is also possible to only select to calculate each user and be divided into the gain of the interest cluster high with oneself Interest Similarity.
As shown in Figure 4,10 users (each circle represents a user) are included at the whole network shown in Fig. 4, before two users, Interest Similarity just carries out line more than the words of certain predetermined threshold value, otherwise not line, the user that there is line relation can as neighboring user, namely illustrating that similarity is high, the weight of line represents the similarity of two users.When calculating gain, it is possible to first calculate and user and the user that there is line are divided into a gain that interest cluster is obtained.
If S303 yield value is more than predetermined threshold value, then corresponding user being divided into corresponding interest cluster, if above-mentioned yield value is less than or equal to predetermined threshold value, then corresponding user still falls within the interest cluster being currently located.
Specifically, it is possible to adopt formula:
Calculate yield value Δ Q, wherein, ∑ in represents that user is currently located between the user in interest cluster similarity sum (namely in place interest cluster between two two users similarity sum), ∑ tot represent user be currently located in interest cluster all users respectively with the similarity sum (i.e. user and user's similarity sum between any two in other cluster in this interest cluster) of user, k in other interest clusteruRepresent user u and other users all similarity add and, ku,inRepresenting user u and be currently located in interest cluster between other similarity sum, m represents in all users similarity sum between user.
It should be noted that above-mentioned Q can represent the quality evaluation of interest cluster,Wherein, u and v refers to user u and user v, A respectivelyu,vRepresent the Interest Similarity between user u and user v, kuRepresent user u and other users all Interest Similarity add and, kvRepresent user v and other users all Interest Similarity add and, m represents in all users similarity sum between user.Wherein, cuRepresent the interest cluster of user's u current home, cvRepresent the interest cluster of user's v current home.In order to simplify calculating, it is possible to the computing formula of this Q is reduced toFormula after this simplification eliminates the delta-function judging whether two users belong to same interest cluster.
Above-mentioned u and v can be specifically the numbering of user, and u and v can be positive integer.
Circulation performs S302-S303, until interest cluster no longer converts, performs S304.
S304, obtain update interest cluster.Adopt new interest cluster to return and perform S302-S304, until the above-mentioned yield value calculated is maximum.
The division of above-mentioned this interest cluster is similar to cluster, may be merged by original interest cluster by calculating in the process that user is divided by gain, thus the interest cluster yield value obtained is bigger, until obtaining the maxgain value more than 0.
It should be noted that, the probability changed due to the interest of user is little, performing to pay close attention to liveness from low the second paramount predetermined number user in the process of said method, these users may come to life, and then can enter above-mentioned interest cluster.
It addition, above-mentioned purchase liveness from the first predetermined number user of high to low and is bought liveness and can be become independent two interest clusters from low the second paramount predetermined number user, and obtain target according to preceding method embodiment and recommend article.
Further, above-mentioned according to article similarity, determine that the target of targeted customer recommends article, specifically may is that according to above-mentioned article similarity, determine the article to be recommended of targeted customer, and then calculate targeted customer's interest-degree to these article to be recommended, according to targeted customer's interest-degree to these article to be recommended, it is determined whether these article to be recommended are recommended article as target.
Namely first according to article similarity, using the article high with the article similarity of targeted customer's history purchase article as article to be recommended, can further determine user's interest-degree to article to be recommended, whether recommend these article finally to determine.
Alternatively, the above-mentioned interest-degree according to targeted customer to these article to be recommended, determine whether as target, these article to be recommended are recommended article, can be judge that above-mentioned targeted customer is to whether the interest-degree of above-mentioned article to be recommended meets the 3rd predetermined threshold value, if meeting, recommends article using these article to be recommended as target.
Determine that the interest-degree of article is had a lot of method by user, for instance user buys the information such as the frequency of ware.
It is alternatively possible to the history of the interest-degree and the targeted customer history of described targeted customer being bought article according to targeted customer buys the article similarity between article and article to be recommended, calculate the above-mentioned targeted customer interest-degree to above-mentioned article to be recommended.
Illustrate, it is assumed that targeted customer u bought the similarity of article i, article i and article j and meets predetermined threshold value, using article j as article to be recommended, then formula p can be adopteduj=∑i∈N(u)wjiruiCalculate the targeted customer interest-degree p to article juj, wherein, N (u) represents that targeted customer's u history buys the set of article, wjiRepresent the article similarity between article i and article j, ruiRepresent targeted customer's interest-degree to article i.Namely it is analyzed according to the user u article bought.
Further, in order to reduce amount of calculation, it is also possible to first determine k the article most like with article j, and seek the common factor of the user u article bought and these k article, then further analyze the targeted customer interest-degree p to article juj, illustrate, it is assumed that targeted customer u bought the similarity of article i, article i and article j and meets predetermined threshold value, using article j as article to be recommended, then formula p can be adopteduj=Σi∈N(u)∩S(j,k)wjiruiCalculate the targeted customer interest-degree p to article juj, wherein, N (u) represents that targeted customer's u history buys the set of article, and (j k) represents and the set (namely choosing k the article that the similarity of article between article j is maximum) of most like k the article of article j, w SjiRepresent the similarity between article i and article j, ruiRepresent targeted customer's interest-degree to article i.
Certainly, it is not limited with above-mentioned computational methods.
It should be noted that the interest-degree of article i can be determined by targeted customer according to the history purchasing behavior of targeted customer, but it is not limited.Illustrate, as this kind of article of sports goods, it is possible to most of users can buy, it is necessary to meet some requirements and could illustrate that user interest degree is high, for instance targeted customer buys the number of times of table tennis more than certain threshold value, and targeted customer is to sense of crackling of ping-pong ball interest；And for some unexpected winner article, the hands as certain animation is done, as long as user bought, it is possible to illustrate that the hands of this animation is done interested by user.During quantization, it is possible to represent interested with 0,1 or lose interest in, for instance ruiIt is that 0 expression targeted customer is not interested to article i, ruiIt is that 1 expression targeted customer is interested in article i；Can also more segmenting, for instance adopt integer 1-5 to represent 5 ranks of interest, 1 represents not interested, and 5 represent greatest interest, determine user's interest-degree to certain article according to purchasing behavior parameters such as the purchase frequency of user, quantity.
Can also by article to be recommended, the highest predetermined number article record of targeted customer's interest-degree, in recommendation list, recommends targeted customer together.
For these 7 users of A-G and some common items, the history that table 1 is these 7 users buys article (stain mark bought the article that this row is corresponding),
Table 1
If employing prior art, when two article simultaneously appear in the purchase list of multiple user, just will be considered that the two article are similar, with reference to table 1, motion bracelet and infanette appear in the purchase list of C, D, E, then just will be considered that according to prior art motion bracelet and infanette are similar article, then when after F also purchasing power motion bracelet, it is possible to infanette can be recommended to F and G, it is apparent that this recommendation is not the interest place of F and G.
Adopt the method that the embodiment of the present invention provides, first A-G is divided into different interest clusters, specifically, as can be seen from Table 1, C, D, E buy the registration of article, and higher (C and D has 3 identical items, C and E has 4 identical items), interest is described more closely, then C, D, E to be divided into an interest cluster；Again it can be seen that A, B, F, G buy the registration higher (A and B, G all have 3 identical items, and F and G has 3 identical items) of article, interest is described more closely, then A, B, F, G to be divided into an interest cluster.Final division result is referred to table 2, has 2 interest clusters (stain represents the interest cluster belonging to this user) in table 2.
Table 2
| Interest cluster 1 | Interest cluster 2 | |
| A | ● | |
| B | ● | |
| C | ● | |
| D | ● | |
| E | ● | |
| F | ● | |
| G | ● |
And then adopt the recommendation method of the embodiment of the present invention again, each interest cluster is analyzed the article similarity between article and then the article that certainly directional user recommends.In this community of interest of A, B, F, G, football and medicated beer, all in the purchase list of A, B, F, illustrate that the article similarity of football and medicated beer is higher, and G have purchased medicated beer, thus can recommend football to G；In like manner, sport shoes and medicated beer appear in the purchase list of F and G, illustrate that the article similarity of sport shoes and medicated beer is higher, and A bought medicated beer, recommend sport shoes to A.Without recommending infanette to G as prior art.
Again for these 7 users of A-G and some common items, the history that table 3 is these 7 users buys article (stain mark bought the article that this row is corresponding, and what the numeral in bracket represented is the purchase number of times of corresponding article).
Table 3
Visible, C and D has 4 identical items, and C and E also has 4 identical items, it is possible to C, D, E are divided into an interest cluster；A and B has 3 identical items, and A and F has 3 identical products, and A and G has 3 identical items, and A and C has 3 identical items, and therefore, A, B, C, F, G are divided into an interest cluster, and concrete division result is as shown in table 4.
Table 4
| Interest cluster 1 | Interest cluster 2 | |
| A | ● | |
| B | ● | |
| C | ● | ● |
| D | ● | |
| E | ● | |
| F | ● | |
| G | ● |
And then adopt the recommendation method of the embodiment of the present invention again, each interest cluster is analyzed the article similarity between article and then the article that certainly directional user recommends.In this interest cluster of A, B, C, F, G, football and sport shoes simultaneously appear in the purchase list of 3 users such as A, B, F, football and diaper occur in the purchase list of two users of A and C, football and medicated beer occur in the purchase list of 3 users such as A, C, G, the article similarity of visible football and sport shoes, football and medicated beer is more than the article similarity of football and diaper, and then be likely to recommend medicated beer to B, recommend sport shoes to G, without recommending diaper to B and G.
Additionally, if introducing interest-degree, (bracket representing, history buys number of times referring to the B in table 3, assume to buy number of times evaluation interest-degree with history), B history bought football, sport shoes and motion hands ring, A, B, C, F, in this interest cluster of G, remove the sport shoes and motion hands ring bought, football and medicated beer occur in A, in the purchase list of 2 users such as G, football and infanette, milk powder, diaper all only simultaneously appears in the purchase list of 1 user, illustrate that the article similarity of football and medicated beer is higher, according to the football bought, can using medicated beer as article to be recommended；Remove the football and motion bracelet bought, sport shoes and medicated beer occur in the purchase list of party A-subscriber, sport shoes and diaper occur in the purchase list of party A-subscriber, and sport shoes and other buy article and do not simultaneously appear in the purchase list of same user, can using medicated beer and diaper as article to be recommended thus according to the sport shoes bought；Remove the football and sport shoes bought, motion bracelet and medicated beer occur in the purchase list of 3 users such as A, B, F, motion bracelet and diaper occur in the purchase list of two users of A and C, motion bracelet and infanette, milk powder, only simultaneously appear in the purchase list of 1 user, it is possible to using medicated beer and diaper as article to be recommended.And then bought football 4 times referring to table 3, B history, buy sport shoes and each 1 time of motion bracelet, illustrates that B is higher to the interest-degree of football, thus article to be recommended according to football acquisition recommend B as final recommendation article the most at last, recommend B by medicated beer.
Certainly, not being limited with above-mentioned example, concrete recommendation is referred to preceding method embodiment and performs.
Fig. 5 is the structural representation of article recommendation apparatus embodiment one provided by the invention, this device can be integrated in article and recommend the background server etc. of platform, specifically, as shown in Figure 5, this device comprises determining that module the 501, first acquisition module 502 and recommending module 503, wherein:
Determine module 501, buy article for the history according to targeted customer, it is determined that the interest cluster belonging to described targeted customer.
First acquisition module 502, the history for obtaining described targeted customer is bought the history of other user in interest cluster belonging to article and described targeted customer and is bought the article similarity between article.
Recommending module 503, for according to described article similarity, it is determined that the target of described targeted customer recommends article.
In the present embodiment, history according to targeted customer buys article, determine the interest cluster belonging to this targeted customer, obtain the history of above-mentioned targeted customer to buy the history of other user in interest cluster belonging to article and this targeted customer and buy the article similarity between article, according to above-mentioned article similarity, determine that the target of this targeted customer recommends article, namely the interest cluster belonging to targeted customer is first determined, then in affiliated interest cluster, the history of targeted customer is bought other user in article and this interest cluster and is bought the article similarity of article, improve the reliability of article similarity, and then recommend article further according to article similarity to targeted customer, also just improve the accuracy recommending article to user, the article recommended are closer to the demand of user, improve Consumer's Experience.And, first determine the interest cluster belonging to targeted customer, carry out article recommendation further according to the data in interest cluster, it is not necessary to as prior art, carry out the analytical calculation of article similarity for all of product data, be greatly reduced operand.
Optionally it is determined that module 501, specifically for calculating the Interest Similarity between described targeted customer and other user；According to the Interest Similarity between described targeted customer and other user, it is determined that the interest cluster belonging to described targeted customer.
Optionally it is determined that described Interest Similarity can be divided into same interest cluster more than the corresponding user of the first predetermined threshold value and described targeted customer by module 501.
In another embodiment, it is also possible to include building module.
Specifically, it is determined that module 501 can when the Interest Similarity of the first user in described targeted customer with other user described be more than the second predetermined threshold value, it is determined that described targeted customer belongs to the interest cluster at described first user place.
Before this, build module, be used for building at least one interest cluster.
Alternatively, build module, obtain the Interest Similarity between user；According to the Interest Similarity between described user, it is divided into same interest cluster by meeting pre-conditioned user.
Fig. 6 is the structural representation of article recommendation apparatus embodiment two provided by the invention, and as shown in Figure 6, on the basis of Fig. 5, this device can also include: the second acquisition module 601 and eliminating module 602, wherein:
Second acquisition module 601, for the purchase parameter according to other user, obtains the purchase liveness of other user described.
Get rid of module 602, for the sequence buying liveness according to other user described, it is determined that described purchase liveness from the first predetermined number user of high to low and described purchase liveness from low the second paramount predetermined number user for treating eliminating user.
Correspondingly, above-mentioned determine that module 501 is for calculating the Interest Similarity between described targeted customer and other user, particularly as follows: calculate described targeted customer and other users described remove described in Interest Similarity between the row for the treatment of each user in addition to the user.
Further, above-mentioned determine module 501 can be also used for according to targeted customer history buy article, it is determined that before the interest cluster belonging to described targeted customer, periodically obtain described targeted customer history buy article.
Correspondingly, it is determined that module 501 buys article according to the history of targeted customer, it is determined that the interest cluster belonging to described targeted customer, particularly as follows: the history according to described targeted customer buys article, it is updated periodically the interest cluster determined belonging to described targeted customer.
Fig. 7 is the structural representation of article recommendation apparatus embodiment three provided by the invention, as it is shown in fig. 7, above-mentioned recommending module 503, it is possible to comprise determining that unit 701, computing unit 702 and recommendation unit 703, wherein:
Determine unit 701, for according to described article similarity, it is determined that the article to be recommended of described targeted customer.
Computing unit 702, for calculating the described targeted customer interest-degree to described article to be recommended.
Alternatively, computing unit 702, the history of interest-degree and described targeted customer for the history of described targeted customer being bought article according to described targeted customer buys the article similarity between article and described article to be recommended, calculates the described targeted customer interest-degree to described article to be recommended；
Recommendation unit 703, for according to the described targeted customer interest-degree to described article to be recommended, it is determined whether will described article to be recommended as described target recommendation article.
Said apparatus is used for performing preceding method embodiment, and it is similar with technique effect that it realizes principle, does not repeat them here.
In another embodiment, the present invention also provides for a kind of article recommendation apparatus, including: memorizer and processor, wherein, memorizer is used for storing programmed instruction, and processor performs preceding method embodiment for calling the instruction in memorizer, it is similar with technique effect that it realizes principle, does not repeat them here.
In several embodiments provided by the present invention, it should be understood that disclosed apparatus and method, it is possible to realize by another way.Such as, device embodiment described above is merely schematic, such as, the division of described unit, being only a kind of logic function to divide, actual can have other dividing mode when realizing, for instance multiple unit or assembly can in conjunction with or be desirably integrated into another system, or some features can ignore, or do not perform.Another point, shown or discussed coupling each other or direct-coupling or communication connection can be through INDIRECT COUPLING or the communication connection of some interfaces, device or unit, it is possible to be electrical, machinery or other form.
The described unit illustrated as separating component can be or may not be physically separate, and the parts shown as unit can be or may not be physical location, namely may be located at a place, or can also be distributed on multiple NE.Some or all of unit therein can be selected according to the actual needs to realize the purpose of the present embodiment scheme.
It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it is also possible to be that unit is individually physically present, it is also possible to two or more unit are integrated in a unit.Above-mentioned integrated unit both can adopt the form of hardware to realize, it would however also be possible to employ hardware adds the form of SFU software functional unit and realizes.
The above-mentioned integrated unit realized with the form of SFU software functional unit, it is possible to be stored in a computer read/write memory medium.Above-mentioned SFU software functional unit is stored in a storage medium, including some instructions with so that a computer equipment (can be personal computer, server, or the network equipment etc.) or processor (English: processor) perform the part steps of method described in each embodiment of the present invention.And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory are (English: Read-OnlyMemory, be called for short: ROM), random access memory (English: RandomAccessMemory, RAM), the various media that can store program code such as magnetic disc or CD be called for short:.
Last it is noted that various embodiments above is only in order to illustrate technical scheme, it is not intended to limit；Although the present invention being described in detail with reference to foregoing embodiments, it will be understood by those within the art that: the technical scheme described in foregoing embodiments still can be modified by it, or wherein some or all of technical characteristic is carried out equivalent replacement；And these amendments or replacement, do not make the essence of appropriate technical solution depart from the scope of various embodiments of the present invention technical scheme.
Claims (15)
1. an item recommendation method, it is characterised in that including:
History according to targeted customer buys article, it is determined that the interest cluster belonging to described targeted customer；
Obtain the history of described targeted customer to buy the history of other user in interest cluster belonging to article and described targeted customer and buy the article similarity between article；
According to described article similarity, it is determined that the target of described targeted customer recommends article.
2. method according to claim 1, it is characterised in that the described history according to targeted customer buys article, it is determined that the interest cluster belonging to described targeted customer, including:
Calculate the Interest Similarity between described targeted customer and other user；
According to the Interest Similarity between described targeted customer and other user, it is determined that the interest cluster belonging to described targeted customer.
3. method according to claim 2, it is characterised in that described according to the Interest Similarity between described targeted customer and other user, it is determined that the interest cluster belonging to described targeted customer, including:
Described Interest Similarity is divided into same interest cluster more than the corresponding user of the first predetermined threshold value and described targeted customer.
4. method according to claim 2, it is characterised in that buy article according to the history of targeted customer, it is determined that before the interest cluster belonging to described targeted customer, also include:
History according to other user buys parameter, obtains the purchase liveness of other user described；
The sequence buying liveness according to other user described, it is determined that described purchase liveness from the first predetermined number user of high to low and described purchase liveness from low the second paramount predetermined number user for treating eliminating user；
Correspondingly,
Interest Similarity between the described targeted customer of described calculating and other user, including:
Calculate the Interest Similarity between the row for the treatment of each user in addition to the user described in removing in described targeted customer and other users described.
5. method according to claim 2, it is characterised in that described according to the Interest Similarity between described targeted customer and other user, it is determined that the interest cluster belonging to described targeted customer, including:
If the Interest Similarity of the first user in described targeted customer and other user described is more than the second predetermined threshold value, it is determined that described targeted customer belongs to the interest cluster at described first user place.
6. method according to claim 5, it is characterised in that described according to the Interest Similarity between described targeted customer and other user, it is determined that before the interest cluster belonging to described targeted customer, also include:
Build at least one interest cluster.
7. method according to claim 6, it is characterised in that at least one interest cluster of described structure, including:
Obtain the Interest Similarity between user；
According to the Interest Similarity between described user, it is divided into same interest cluster by meeting pre-conditioned user.
8. the method according to any one of claim 1～7, it is characterised in that the described history according to targeted customer buys article, it is determined that before the interest cluster belonging to described targeted customer, also include:
The history periodically obtaining described targeted customer buys article；
Correspondingly, the described history according to targeted customer buys article, it is determined that the interest cluster belonging to described targeted customer, including:
History according to described targeted customer buys article, is updated periodically the interest cluster determined belonging to described targeted customer.
9. method according to claim 1, it is characterised in that described according to described article similarity, it is determined that the target of described targeted customer recommends article, including:
According to described article similarity, it is determined that the article to be recommended of described targeted customer；
Calculate the described targeted customer interest-degree to described article to be recommended；
According to the described targeted customer interest-degree to described article to be recommended, it is determined whether described article to be recommended are recommended article as described target.
10. method according to claim 9, it is characterised in that the described calculating described targeted customer interest-degree to described article to be recommended, including:
The history of the interest-degree and described targeted customer of the history of described targeted customer being bought article according to described targeted customer buys the article similarity between article and described article to be recommended, calculates the described targeted customer interest-degree to described article to be recommended.
11. method according to claim 9, it is characterised in that the described interest-degree according to described targeted customer to described article to be recommended, it is determined whether described article to be recommended are recommended article as described target, including:
Judge that described targeted customer is to whether the interest-degree of described article to be recommended meets the 3rd predetermined threshold value, if meeting, then recommends article using described article to be recommended as target.
12. an article recommendation apparatus, it is characterised in that including:
Determine module, buy article for the history according to targeted customer, it is determined that the interest cluster belonging to described targeted customer；
First acquisition module, the history for obtaining described targeted customer is bought the history of other user in interest cluster belonging to article and described targeted customer and is bought the article similarity between article；
Recommending module, for according to described article similarity, it is determined that the target of described targeted customer recommends article.
13. device according to claim 12, it is characterised in that described determine module, specifically for calculating the Interest Similarity between described targeted customer and other user；
According to the Interest Similarity between described targeted customer and other user, it is determined that the interest cluster belonging to described targeted customer.
14. device according to claim 12, it is characterised in that also include:
Second acquisition module, for the purchase parameter according to other user, obtains the purchase liveness of other user described；
Get rid of module, for the sequence buying liveness according to other user described, it is determined that described purchase liveness from the first predetermined number user of high to low and described purchase liveness from low the second paramount predetermined number user for treating eliminating user；
Correspondingly, described determine that module is for calculating the Interest Similarity between described targeted customer and other user, particularly as follows: calculate described targeted customer and other users described remove described in Interest Similarity between the row for the treatment of each user in addition to the user.
15. device according to claim 12, it is characterised in that described recommending module, including:
Determine unit, for according to described article similarity, it is determined that the article to be recommended of described targeted customer；
Computing unit, the history of interest-degree and described targeted customer for the history of described targeted customer being bought article according to described targeted customer buys the article similarity between article and described article to be recommended, calculates the described targeted customer interest-degree to described article to be recommended；
Recommendation unit, for according to the described targeted customer interest-degree to described article to be recommended, it is determined whether will described article to be recommended as described target recommendation article.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610127208.9A CN105809479A (en) | 2016-03-07 | 2016-03-07 | Item recommending method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610127208.9A CN105809479A (en) | 2016-03-07 | 2016-03-07 | Item recommending method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105809479A true CN105809479A (en) | 2016-07-27 |
Family
ID=56466769
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610127208.9A Pending CN105809479A (en) | 2016-03-07 | 2016-03-07 | Item recommending method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105809479A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296270A (en) * | 2016-07-29 | 2017-01-04 | 北京小米移动软件有限公司 | Method of Commodity Recommendation and device |
| CN106372961A (en) * | 2016-08-23 | 2017-02-01 | 北京小米移动软件有限公司 | Commodity recommendation method and device |
| CN107016589A (en) * | 2016-08-10 | 2017-08-04 | 阿里巴巴集团控股有限公司 | The determination method and device of recommended products |
| CN108734549A (en) * | 2018-04-27 | 2018-11-02 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN108734548A (en) * | 2018-04-27 | 2018-11-02 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN108960958A (en) * | 2018-04-27 | 2018-12-07 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN109859004A (en) * | 2019-01-10 | 2019-06-07 | 珠海金山网络游戏科技有限公司 | A kind of Method of Commodity Recommendation and system based on historical data |
| CN109872206A (en) * | 2017-12-04 | 2019-06-11 | 北京京东尚科信息技术有限公司 | Assess determination method for parameter, device and computer readable storage medium |
| CN110110206A (en) * | 2018-01-17 | 2019-08-09 | 阿里巴巴集团控股有限公司 | Relation excavation and recommended method, device, calculating equipment, storage medium between article |
| CN110222254A (en) * | 2019-04-26 | 2019-09-10 | 平安科技（深圳）有限公司 | Course recommended method, device, equipment and computer readable storage medium |
| WO2019205795A1 (en) * | 2018-04-26 | 2019-10-31 | 腾讯科技（深圳）有限公司 | Interest recommendation method, computer device, and storage medium |
| CN110473043A (en) * | 2018-05-11 | 2019-11-19 | 北京京东尚科信息技术有限公司 | A kind of item recommendation method and device based on user behavior |
| CN110611743A (en) * | 2018-06-15 | 2019-12-24 | 夏普株式会社 | Image forming apparatus, image forming system, and display control method |
| CN110751498A (en) * | 2018-07-24 | 2020-02-04 | 北京京东尚科信息技术有限公司 | Article recommendation method and system |
| CN110795620A (en) * | 2019-09-25 | 2020-02-14 | 平安科技（深圳）有限公司 | Product recommendation method, device, equipment and storage medium based on same type of users |
| CN110992137A (en) * | 2019-11-27 | 2020-04-10 | 上海风秩科技有限公司 | Product recommendation method and device, electronic equipment and storage medium |
| CN111414554A (en) * | 2020-03-26 | 2020-07-14 | 透明生活(武汉)信息科技有限公司 | Commodity recommendation method, system, server and storage medium |
| CN111694906A (en) * | 2019-03-14 | 2020-09-22 | 北京搜狗科技发展有限公司 | Method, device and equipment for displaying POI in map |
| CN111949893A (en) * | 2020-08-21 | 2020-11-17 | 深圳市活力天汇科技股份有限公司 | Air ticket recommendation method based on collaborative filtering algorithm |
| CN112330405A (en) * | 2020-11-10 | 2021-02-05 | 深圳大学 | Recommendation method, terminal and storage medium for explicit feedback of item sets |
| CN113674037A (en) * | 2021-10-21 | 2021-11-19 | 西安超嗨网络科技有限公司 | Data acquisition and recommendation method based on shopping behaviors |
| TWI787196B (en) * | 2016-11-29 | 2022-12-21 | 香港商阿里巴巴集團服務有限公司 | Method, device and system for generating business object attribute identification |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102479366A (en) * | 2010-11-25 | 2012-05-30 | 阿里巴巴集团控股有限公司 | Commodity recommending method and system |
| CN103106600A (en) * | 2012-11-15 | 2013-05-15 | 深圳中兴网信科技有限公司 | Commodity information push system and commodity information push method |
| CN103309967A (en) * | 2013-06-05 | 2013-09-18 | 清华大学 | Collaborative filtering method and system based on similarity propagation |
| CN104598643A (en) * | 2015-02-13 | 2015-05-06 | 成都品果科技有限公司 | Article similarity contribution factor, similarity acquiring method, as well as article recommendation method and system thereof |
| US20150379609A1 (en) * | 2014-06-30 | 2015-12-31 | Kobo Incorporated | Generating recommendations for unfamiliar users by utilizing social side information |
-
2016
- 2016-03-07 CN CN201610127208.9A patent/CN105809479A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102479366A (en) * | 2010-11-25 | 2012-05-30 | 阿里巴巴集团控股有限公司 | Commodity recommending method and system |
| CN103106600A (en) * | 2012-11-15 | 2013-05-15 | 深圳中兴网信科技有限公司 | Commodity information push system and commodity information push method |
| CN103309967A (en) * | 2013-06-05 | 2013-09-18 | 清华大学 | Collaborative filtering method and system based on similarity propagation |
| US20150379609A1 (en) * | 2014-06-30 | 2015-12-31 | Kobo Incorporated | Generating recommendations for unfamiliar users by utilizing social side information |
| CN104598643A (en) * | 2015-02-13 | 2015-05-06 | 成都品果科技有限公司 | Article similarity contribution factor, similarity acquiring method, as well as article recommendation method and system thereof |
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106296270A (en) * | 2016-07-29 | 2017-01-04 | 北京小米移动软件有限公司 | Method of Commodity Recommendation and device |
| CN107016589A (en) * | 2016-08-10 | 2017-08-04 | 阿里巴巴集团控股有限公司 | The determination method and device of recommended products |
| CN106372961A (en) * | 2016-08-23 | 2017-02-01 | 北京小米移动软件有限公司 | Commodity recommendation method and device |
| TWI787196B (en) * | 2016-11-29 | 2022-12-21 | 香港商阿里巴巴集團服務有限公司 | Method, device and system for generating business object attribute identification |
| CN109872206A (en) * | 2017-12-04 | 2019-06-11 | 北京京东尚科信息技术有限公司 | Assess determination method for parameter, device and computer readable storage medium |
| CN110110206A (en) * | 2018-01-17 | 2019-08-09 | 阿里巴巴集团控股有限公司 | Relation excavation and recommended method, device, calculating equipment, storage medium between article |
| US11593894B2 (en) | 2018-04-26 | 2023-02-28 | Tencent Technology (Shenzhen) Company Limited | Interest recommendation method, computer device, and storage medium |
| WO2019205795A1 (en) * | 2018-04-26 | 2019-10-31 | 腾讯科技（深圳）有限公司 | Interest recommendation method, computer device, and storage medium |
| CN108734548B (en) * | 2018-04-27 | 2021-12-17 | 达而观信息科技（上海）有限公司 | Item recommendation method and device |
| CN108960958A (en) * | 2018-04-27 | 2018-12-07 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN108734549A (en) * | 2018-04-27 | 2018-11-02 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN108734548A (en) * | 2018-04-27 | 2018-11-02 | 达而观信息科技（上海）有限公司 | item recommendation method and device |
| CN108960958B (en) * | 2018-04-27 | 2022-03-22 | 达而观信息科技（上海）有限公司 | Item recommendation method and device |
| CN108734549B (en) * | 2018-04-27 | 2021-12-21 | 达而观信息科技（上海）有限公司 | Item recommendation method and device |
| CN110473043A (en) * | 2018-05-11 | 2019-11-19 | 北京京东尚科信息技术有限公司 | A kind of item recommendation method and device based on user behavior |
| CN110611743A (en) * | 2018-06-15 | 2019-12-24 | 夏普株式会社 | Image forming apparatus, image forming system, and display control method |
| CN110751498A (en) * | 2018-07-24 | 2020-02-04 | 北京京东尚科信息技术有限公司 | Article recommendation method and system |
| CN109859004A (en) * | 2019-01-10 | 2019-06-07 | 珠海金山网络游戏科技有限公司 | A kind of Method of Commodity Recommendation and system based on historical data |
| CN111694906A (en) * | 2019-03-14 | 2020-09-22 | 北京搜狗科技发展有限公司 | Method, device and equipment for displaying POI in map |
| CN110222254A (en) * | 2019-04-26 | 2019-09-10 | 平安科技（深圳）有限公司 | Course recommended method, device, equipment and computer readable storage medium |
| CN110795620A (en) * | 2019-09-25 | 2020-02-14 | 平安科技（深圳）有限公司 | Product recommendation method, device, equipment and storage medium based on same type of users |
| CN110992137A (en) * | 2019-11-27 | 2020-04-10 | 上海风秩科技有限公司 | Product recommendation method and device, electronic equipment and storage medium |
| CN110992137B (en) * | 2019-11-27 | 2023-09-01 | 上海风秩科技有限公司 | Product recommendation method and device, electronic equipment and storage medium |
| CN111414554A (en) * | 2020-03-26 | 2020-07-14 | 透明生活(武汉)信息科技有限公司 | Commodity recommendation method, system, server and storage medium |
| CN111414554B (en) * | 2020-03-26 | 2023-08-22 | 透明生活(武汉)信息科技有限公司 | Commodity recommendation method, commodity recommendation system, server and storage medium |
| CN111949893A (en) * | 2020-08-21 | 2020-11-17 | 深圳市活力天汇科技股份有限公司 | Air ticket recommendation method based on collaborative filtering algorithm |
| CN111949893B (en) * | 2020-08-21 | 2024-03-29 | 深圳市活力天汇科技股份有限公司 | Air ticket recommending method based on collaborative filtering algorithm |
| CN112330405A (en) * | 2020-11-10 | 2021-02-05 | 深圳大学 | Recommendation method, terminal and storage medium for explicit feedback of item sets |
| CN112330405B (en) * | 2020-11-10 | 2023-11-24 | 深圳大学 | Recommendation method, terminal and storage medium for item set explicit feedback |
| CN113674037A (en) * | 2021-10-21 | 2021-11-19 | 西安超嗨网络科技有限公司 | Data acquisition and recommendation method based on shopping behaviors |
| CN113674037B (en) * | 2021-10-21 | 2022-01-25 | 西安超嗨网络科技有限公司 | Data acquisition and recommendation method based on shopping behaviors |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105809479A (en) | Item recommending method and device | |
| CN109087177B (en) | Method, device and computer-readable storage medium for recommending commodities to target user | |
| CN105095267B (en) | Recommendation method and device for user participation items | |
| CN106651542B (en) | Article recommendation method and device | |
| CN106708883B (en) | Recommendation method and device | |
| US8583502B2 (en) | Value maximizing recommendation systems | |
| CA2700030C (en) | Touchpoint customization system | |
| CN104077714B (en) | Access preference acquisition, advertisement sending method and the system of the user of website | |
| CN107103487B (en) | Advertisement pushing method based on big data | |
| CN111091400A (en) | Method and device for generating advertisement conversion prediction model and delivering advertisement | |
| CN102893300A (en) | Methods and apparatus for integrating volumetric sales data, media consumption information, and geographic -demographic data to target advertisements | |
| US10448120B1 (en) | Recommending features for content planning based on advertiser polling and historical audience measurements | |
| CN108205768A (en) | Database building method and data recommendation method and device, equipment and storage medium | |
| EP2481018A1 (en) | Method and apparatus for executing a recommendation | |
| CN108153791A (en) | A kind of resource recommendation method and relevant apparatus | |
| CN108805598A (en) | Similarity information determines method, server and computer readable storage medium | |
| CN106844402B (en) | Data processing method and device | |
| Aharon et al. | Off-set: one-pass factorization of feature sets for online recommendation in persistent cold start settings | |
| CN110933472B (en) | Method and device for realizing video recommendation | |
| CN108876517B (en) | User preference analysis method and system based on artificial intelligence | |
| CN113689259A (en) | Commodity personalized recommendation method and system based on user behaviors | |
| CN111104606B (en) | Weight-based conditional wandering chart recommendation method | |
| CN105809275A (en) | Item scoring prediction method and apparatus | |
| US20160203540A1 (en) | System and method of product and vendor selection | |
| CN105260458A (en) | Video recommendation method for display apparatus and display apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160727 |
|
| RJ01 | Rejection of invention patent application after publication |