Summary of the invention
Technical problems to be solved in this application are to provide a kind of group purchase platform information processing method and device, can solve the processing time that prior art increase purchases by group Platform Server, bring the problem of unnecessary load to server.
In order to solve the problem, this application discloses a kind of group purchase platform information processing method, comprising the following steps:
Its purchase information is obtained at the log-on message purchasing by group platform based on user; Wherein, described in purchase by group platform real time record user's operation information, and store look into use for follow-up in the server; When user's login purchases by group platform, purchase by group platform and identified user by user's registration information, and obtain the purchase information of storage this user in the server according to log-on message, described purchase information is transaction record collection;
Determine the consumption propensity of described user to commodity of all categories according to described transaction record collection, and merchandise classification consumption propensity being greater than threshold value is set to specific dimension;
Described specific dimension and the basic dimension of adding up in advance are combined as dimension set;
Obtain and currently allly purchase by group commodity actual information, and according to purchasing by group the weighting coefficient of each dimension in the set of commodity actual information determination dimension, comprise: obtain a certain actual information purchasing by group each dimension of commodity, be normalized the actual information of each dimension, the value of normalized gained is the weighting coefficient of each dimension;
Calculate the score respectively purchasing by group commodity for active user based on the weighting coefficient determining each dimension and the important coefficient of each dimension preset, and return the ranking results after according to score sequence.
Further, the weighting coefficient of described basic dimension is determined in the following way:
According to basic dimension set standard value;
The actual value of this basic dimension and standard value are normalized, the value of normalized gained is the weighting coefficient of described basic dimension.
Further, the weighting coefficient of described specific dimension is determined in the following way:
Determine to purchase by group commodity generic;
Calculate the degree of correlation of described classification and specific dimension, the described degree of correlation is the weighting coefficient of specific dimension.
Further, described basic dimension is determined according to one or more analyses in following historical data:
Purchase by group the time, purchase by group the temperature of commodity, discount information and price.
Described method also comprises further:
Dimension set is upgraded.
Further, describedly renewal carried out to dimension set comprise:
Determine new specific dimension;
Calculate the correlativity of each dimension in new specific dimension and existing dimension set, if the degree of correlation of new specific dimension and any dimension is all less than threshold value, then new specific dimension is added in dimension set.
In order to solve the problem, present invention also provides a kind of group purchase platform information treating apparatus, comprising:
Buy data obtaining module, for obtaining its purchase information based on user at the log-on message purchasing by group platform; Wherein, described in purchase by group platform real time record user's operation information, and store look into use for follow-up in the server; When user's login purchases by group platform, purchase by group platform and identified user by user's registration information, and obtain the purchase information of storage this user in the server according to log-on message, described purchase information is transaction record collection;
Specific dimension determination module, for determining the consumption propensity of described user to commodity of all categories according to described transaction record collection, and merchandise classification consumption propensity being greater than threshold value is set to specific dimension;
Dimension set determination module, for being combined as dimension set by described specific dimension and the basic dimension of adding up in advance;
Weighting coefficient determination module, currently allly purchases by group commodity actual information for obtaining, and according to purchasing by group the weighting coefficient of each dimension in the set of commodity actual information determination dimension; Described weighting coefficient determination module comprises: normalized unit, and for being normalized the actual information of each dimension, the value of normalized gained is the weighting coefficient of each dimension;
Order module, for calculating the score respectively purchasing by group commodity for active user based on the weighting coefficient determining each dimension and the important coefficient of each dimension preset, and returns the ranking results after according to score sequence.
Further, described normalized unit comprises:
Basis dimension weighting coefficient determination subelement, for according to basic dimension set standard value, and the actual value of this basic dimension and standard value are normalized, the value of normalized gained is the weighting coefficient of described basic dimension.
Further, described normalized unit comprises:
Specific dimension weighting coefficient determination subelement, purchases by group commodity generic for determining, and calculates the degree of correlation of described classification and specific dimension, and the described degree of correlation is the weighting coefficient of specific dimension.
Further, described device also comprises:
Update module, for upgrading dimension set.
Further, described update module comprises:
Correlation determination unit, calculates the correlativity of each dimension in new specific dimension and existing dimension set, if the degree of correlation of new specific dimension and any dimension is all less than threshold value, then new specific dimension is added in dimension set.
Compared with prior art, the application has the following advantages:
The group purchase platform information processing method of the application and device are by analyzing the purchase information purchasing by group platform registered user, thus determine the specific dimension for each specific user, then the dimension set of ranking results is affected in conjunction with predetermined basic dimension composition, the weighting coefficient that each purchases by group each dimension of commodity is calculated by the actual information purchasing by group commodity, determine and respectively purchase by group the new order models of commodity, thus the personalized ordering realized for each user, make information can pass to user more accurately, avoid the impact of the user of invalid or impurity information.Also be, purchase by group platform and show the sequence of user because consider user preference, also just more meet the expectation of user, thus make information display more accurate, avoid the information display that occurs single mode and bring unnecessary load because user again searches to server.
In addition, obtained by accumulation user being bought to information, when user buys after information changes, can continue according to predetermined method to determine new specific dimension, thus ranking results is upgraded at any time.
Further, when new specific dimension is formed, by judging with the correlativity of existing dimension, when correlativity is less, newer specific dimension being added dimension set, avoiding the accuracy affecting ranking results because two dimension correlativitys are higher.
Embodiment
For enabling above-mentioned purpose, the feature and advantage of the application more become apparent, below in conjunction with the drawings and specific embodiments, the application is described in further detail.
The application purchases by group the information promulgating platform that platform refers to the group purchase information gathering a large amount of group buying websites, by each group buying websites and the cooperation purchasing by group platform, purchase by group platform obtain the group purchase information of each group buying websites and on platform, carry out issue return, the registered user purchasing by group platform purchases by group by logging in the group purchase information that the platform place page then can retrieve each group buying websites, and realizes purchasing by group.
With reference to Fig. 1, the system architecture diagram of the group purchase platform information process of the application is shown.First, group purchase platform information treating apparatus obtains user profile (log-on message, purchase information etc.) from server, then user profile is refined, and determine user's specific dimension (dimension format) with reference to the classifying rules preset, the weighting coefficient of specific dimension and predetermined basis dimension is determined according to preordering method (as normalized), then the important coefficient of each dimension is determined according to modes such as expert along training, finally form order models, sort to purchasing by group commodity according to fixing order models and return.For the order models determined, because user buys information and can change along with the time, therefore have new specific dimension to produce, now then first need the correlativity judging new specific dimension and existing dimension, when correlativity is less, be classified as new dimension again, and redefine weighting coefficient and important coefficient etc., to form new order models.
With reference to Fig. 2, the group purchase platform information processing method embodiment one of the application is shown, comprises the following steps:
Step 101, obtains its purchase information based on user at the log-on message purchasing by group platform.
Purchasing by group platform allows user by purchasing by group platform to realize the purchase purchasing by group commodity by the mode that user registers, namely user is after purchasing by group platform registration, then can the commodity purchasing by group each group buying websites that platform returns be purchased by group, and without the need to again arriving the registration of each group buying websites again.In this way, purchasing by group platform can real time record user's operation information, such as purchase information etc., and can store and look into use for follow-up in the server.
When user's login purchases by group platform, purchase by group platform to be claimed by username, the modes such as account identify user, and can obtain other purchase information of storage this user in the server according to log-on message, as, the level of consumption, purchase merchandise classification, quantity etc.
Step 102, determine the consumption propensity of described user to commodity of all categories, and merchandise classification consumption propensity being greater than threshold value is set to specific dimension according to described purchase information.
Purchase information is a pile transaction record collection, buying the information such as merchandise classification, quantity, the amount of money, time, can obtain the consumption propensity of user to certain classification commodity by analyzing a certain user of acquisition.When realizing, the consumption propensity of user can be represented in the following way: <Category (i), Volume (i) >.Wherein, Category (i) represents the title of a certain commodity or merchandise classification, and Volume (i) represents the consumption propensity of user to these commodity or merchandise classification.Concrete, Volume (i) can be the amount of money that this user buys these commodity or these classification commodity, also can be the quantity that user buys these commodity or these classification commodity.The application does not limit this, as long as can embody user preference.
Threshold value can be determined according to statistics, if consumption propensity adopts the amount of money to represent, because different classes of commodity price differs greatly, so different classes of commodity can set different threshold values according to commodity regular price interval.Such as, electric type commodity threshold value can be 5000 yuan, and clothing commodity threshold value can be 1000 yuan, and food and drink class commodity threshold value also can be 1000 yuan etc.Equally, quantify if consumption propensity is adopted, definite threshold can be carried out according to the specific category of commodity, but because the relation of quantity and merchandise classification is not very large, therefore most of commodity can adopt equal number as threshold value.Such as, electric type commodity, clothing commodity, food and drink class commodity threshold value can be 10 or 20.After the consumption propensity of a certain commodity or certain class I goods is greater than threshold value, then think that user has certain preference to these commodity or such commodity, then using the sequence specific dimension of these these type of commodity as this user, thus can increase the weight of these commodity or these type of commodity when sorting.
Step 103, is combined as dimension set by described specific dimension and the basic dimension of adding up in advance.
Basis dimension refers to some regular dimensions that all users that the data statistics obtained based on historical data or other collections draws may can consider, such as, the time is purchased by group, purchases by group the temperature of commodity (by searched number of times or buy number of times and determine), discount information, price etc.These basic dimensions rule of thumb and can be carried out the mode such as analyze to real data and be pre-determined.
Step 104, obtains and currently allly purchases by group commodity actual information, and according to purchasing by group the weighting coefficient of each dimension in the set of commodity actual information determination dimension.
Weighting coefficient refers to for a certain particular commodity, the particular factor of a certain dimension.Weighting coefficient can be determined in the following way:
S1, obtains a certain actual information purchasing by group each dimension of commodity.
S2, is normalized the weighting coefficient obtaining each dimension to the actual information of each dimension.
Normalized can adopt linearly or non-linear method processes, and the application does not limit this.
Wherein, the normalized for basic dimension can in the following way: set a standard value; The value that after actual value and standard value are carried out related operation, (as add, subtract, multiplication and division or other computings) obtains is the value after normalized.The standard value of each dimension can set according to concrete dimension.Such as, if desired the weighting coefficient of all dimensions is all normalized to the number between 0 to 1.So, for this dimension of discount, can established standards value be 1, the difference of standard value and actual value, as normalized end value, so for a certain commodity, deducts its true discount number with 1, then obtains the weighting coefficient of discount.For price, a mxm. can be preset, the ratio of the real price of commodity and this mxm., finally deduct this ratio with 1, namely obtain the weighting coefficient of price.In like manner, also can adopt similar mode to process for other dimensions.Obtaining weighting coefficient by being normalized each dimension, so that follow-up computing, computation process can be simplified, in addition, because adopt normalized mode to process, make ranking results more accurate.
For specific dimension, representative be user preference, can't there is a fixing standard value, therefore, the normalized of weighting coefficient is then determined to determine according to the degree of correlation, comprising: determine to purchase by group commodity generic; Calculate this and purchase by group the degree of correlation between commodity generic and specific dimension, the value obtained is the weighting coefficient of specific dimension.Such as, for a certain user, its specific dimension is clothes, so purchase by group commodity for a certain, the weighting coefficient deterministic process of this specific dimension for: first obtain this and purchase by group classification belonging to commodity, then calculate the degree of correlation of this classification and clothing, this degree of correlation is this and purchases by group the weighting coefficient of commodity in this specific dimension.
The concrete calculating of the degree of correlation can be in the following way, carry out search according to the concrete class purchasing by group commodity and obtain sequence v1, carry out search according to specific dimension and obtain sequence v2, then by operations such as replacement, deletion or interpolations, map function is carried out to one of them sequence, makes it finally be transformed to another sequence, determine the degree of correlation according to wherein carried out number of operations, number of operations is fewer, then illustrate that the degree of correlation of two sequences is higher.
Step 105, calculates the score respectively purchasing by group commodity for active user based on the weighting coefficient determining each dimension and the important coefficient of each dimension preset, and returns the ranking results after according to score sequence.
Wherein, the important coefficient of each dimension can be empirical value, also can determine according to actual conditions.Specific dimension can by buying information analysis to determine to user, and such as, in all purchase amount of money of user or quantity, a certain specific dimension spends the amount of money or quantity purchase proportion etc.Basis dimension then can be determined by modes such as model trainings a large amount of image data, and the application does not limit this.
Such as, purchase by group commodity for a certain, the weighting coefficient of its each dimension is respectively P
1, P
2, P
3..., P
n, the important coefficient of each dimension is respectively weight
1, weight
2, weight
3..., weight
n.Then can calculate by following formula the score that this purchases by group commodity:
in like manner, for need sort the commodity that purchase by group all calculated by aforementioned manner, finally sort according to score, then can obtain final ranking results.
With reference to Fig. 3, preferably, because the purchase information of user can along with accumulated time, over time, the specific dimension determination mode described by step 102, then new specific dimension also may be had to be formed, and therefore the application can also comprise the following steps:
Step 201, upgrades dimension set.
Renewal comprises: determine new specific dimension according to the mode of step 102; Calculate the correlativity of each dimension in new specific dimension and existing dimension set, if the degree of correlation of new specific dimension and any dimension is all less than threshold value, then new specific dimension is added in dimension set.
Wherein, in new specific dimension and existing dimension set, judge can with reference to aforesaid relatedness computation mode for the correlativity of each dimension, namely carry out search by new specific dimension and obtain sequence, the sequence then obtained with each dimensional searches of existing dimension compares, and detailed process does not repeat them here.
Equally, the weighting coefficient of new specific dimension and the deterministic process of important coefficient are also identical with the weighting coefficient of the specific dimension in embodiment one and important coefficient deterministic process.
With reference to Fig. 4, a kind of group purchase platform information treating apparatus embodiment of the application is shown, comprises and buy data obtaining module 10, specific dimension determination module 20, dimension set determination module 30, weighting coefficient determination module 40 and order module 50.
Buy data obtaining module 10, for obtaining its purchase information based on user at the log-on message purchasing by group platform.
Specific dimension determination module 20, for determining the consumption propensity of described user to commodity of all categories according to described purchase information, and merchandise classification consumption propensity being greater than threshold value is set to specific dimension.
Dimension set determination module 30, for being combined as dimension set by described specific dimension and the basic dimension of adding up in advance.
Weighting coefficient determination module 40, currently allly purchases by group commodity actual information for obtaining, and according to purchasing by group the weighting coefficient of each dimension in the set of commodity actual information determination dimension.
Preferably, weighting coefficient determination module 40 comprises normalized unit, and for being normalized actual information according to predetermined method, the value of normalized is the weighting coefficient of each dimension.Wherein normalized unit comprises basic dimension weighting coefficient determination subelement and specific dimension weighting coefficient determination subelement.Basis dimension weighting coefficient determination subelement, for according to basic dimension set standard value, and the actual value of this basic dimension and standard value are normalized, the value of normalized is the weighting coefficient of described basic dimension.Specific dimension weighting coefficient determination subelement, purchases by group commodity generic for determining, and calculates the degree of correlation of described classification and specific dimension, and the described degree of correlation is the weighting coefficient of specific dimension.
Order module 50, for calculating based on the weighting coefficient determining each dimension and the important coefficient of each dimension preset the score respectively purchasing by group commodity for active user, and returns the ranking results after according to score sequence.
Preferably, this group purchase platform information treating apparatus also comprises update module, for upgrading dimension set.Wherein, update module comprises correlation determination unit, calculates the correlativity of each dimension in new specific dimension and existing dimension set, if the degree of correlation of new specific dimension and any dimension is all less than threshold value, then new specific dimension is added in dimension set.
Each embodiment in this instructions all adopts the mode of going forward one by one to describe, and what each embodiment stressed is the difference with other embodiments, between each embodiment identical similar part mutually see.For device embodiment, due to itself and embodiment of the method basic simlarity, so description is fairly simple, relevant part illustrates see the part of embodiment of the method.
The group purchase platform information processing method provided the application above and device are described in detail, apply specific case herein to set forth the principle of the application and embodiment, the explanation of above embodiment is just for helping method and the core concept thereof of understanding the application; Meanwhile, for one of ordinary skill in the art, according to the thought of the application, all will change in specific embodiments and applications, in sum, this description should not be construed as the restriction to the application.