CN109062994A - Recommended method, device, computer equipment and storage medium - Google Patents
Recommended method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109062994A CN109062994A CN201810725366.3A CN201810725366A CN109062994A CN 109062994 A CN109062994 A CN 109062994A CN 201810725366 A CN201810725366 A CN 201810725366A CN 109062994 A CN109062994 A CN 109062994A
- Authority
- CN
- China
- Prior art keywords
- user
- recommended
- generalization bounds
- preset
- current
- 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.)
- Withdrawn
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Information Transfer Between Computers (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This application involves a kind of recommended method, device, computer equipment and storage mediums.The described method includes: obtaining the current operation behavior set that active user acts on terminal, each operation behavior in current operation behavior set acts on corresponding content substance;At least one Generalization bounds compatible with content substance are selected from preset Generalization bounds set, obtain the corresponding object to be recommended of content substance according to the Generalization bounds of selection;The corresponding current recommendation list of Generalization bounds is updated according to the corresponding object to be recommended of content substance;Object to be recommended is chosen from the corresponding updated current recommendation list of Generalization bounds each in preset Generalization bounds set, target recommendation list is obtained according to the object to be recommended of selection.Using this method can high object and user demand to be recommended matching degree.
Description
Technical field
This application involves Internet technical fields, more particularly to a kind of recommended method, device, computer equipment and storage
Medium.
Background technique
With the rapid development of Internet technology, user can conveniently and efficiently be bought goods by internet.But magnanimity
Occur while merchandise news and mass users, user is on the one hand enabled to be difficult to find the end article of oneself, while also resulting in big
Amount commodity be nobody shows any interest in, so that information utilization is low;On the other hand, the increase of user's amount of access causes raw log files
Increase.In order to targetedly provide a user merchandise news required for it, and then occur related for recommending to user
The recommended method of merchandise news.
However, existing recommended method is normally based on single proposed algorithm, and e.g., content-based recommendation, based on association
With the recommendation of filtering, these algorithms each have limitation, cause the matching degree of object to be recommended and user demand not high, can not
It meets the needs of users.
Summary of the invention
Based on this, it is necessary to which in view of the above technical problems, providing a kind of can be improved object to be recommended and user demand
Recommended method, device, computer equipment and the storage medium of matching degree.
A kind of recommended method, which comprises
The current operation behavior set that active user acts on terminal is obtained, it is each in the current operation behavior set
Operation behavior acts on corresponding content substance;
At least one Generalization bounds compatible with the content substance are selected from preset Generalization bounds set, according to
The Generalization bounds of selection obtain the corresponding object to be recommended of the content substance;
The corresponding current recommendation list of the Generalization bounds is updated according to the corresponding object to be recommended of the content substance;
It is selected from the corresponding updated current recommendation list of Generalization bounds each in the preset Generalization bounds set
Object to be recommended is taken, target recommendation list is obtained according to the object to be recommended of selection.
In one of the embodiments, the Generalization bounds according to selection obtain the content substance it is corresponding to
Recommended, comprising:
The content substance is segmented, word segmentation result is obtained;
Keyword is extracted from the word segmentation result, obtains keyword set;
From the corresponding centre word of each keyword in each keyword set is searched in preset center set of words, obtain
To center set of words, using each centre word in the center set of words as corresponding first user tag of the active user;
Word frequency of the corresponding keyword of first user tag in the content substance is calculated, is obtained according to the word frequency
To the corresponding weight of first user tag;
It is corresponding that the active user is generated according to first user tag and the corresponding weight of first user tag
User portrait, drawn a portrait to obtain the corresponding object to be recommended of the content substance according to the corresponding user of the active user.
The content substance includes merchandise news in one of the embodiments, the recommendation plan according to selection
Slightly obtain the corresponding object to be recommended of the content substance, comprising:
The corresponding goods number of the merchandise news is obtained, corresponding commodity classification, root are searched according to the goods number
Corresponding history weight is obtained according to the commodity classification;
The corresponding current browsing time of the commodity classification is calculated according to the current operation behavior set, it will be described current
Browsing time standardization;
It is corresponding current that the commodity classification is calculated according to the history weight and the standardized current browsing time
Weight;
The commodity classification is ranked up according to the present weight, the commodity of preset quantity are obtained according to ranking results
Classification is as end article classification, using the end article classification as the corresponding second user label of the active user, root
The corresponding user's portrait of active user is generated according to the second user label and corresponding weight;
It is drawn a portrait to obtain the corresponding object to be recommended of the content substance according to the corresponding user of the active user.
It is described in one of the embodiments, to draw a portrait to obtain the content substance according to the corresponding user of the active user
Corresponding object to be recommended, comprising:
Obtain the reference user set for meeting preset rules;
Being drawn a portrait according to the corresponding user of the active user, it is each in the active user and the reference user set to calculate
A similarity with reference to user;
It obtains similarity and is greater than the reference user of preset threshold as object reference user;
The corresponding historical operation behavior set of the object reference user is obtained, it is corresponding according to the object reference user
Historical operation behavior set obtains the corresponding object to be recommended of the content substance.
In one of the embodiments, the method also includes:
When current time reaches the preset triggered time and the corresponding user identifier of the active user is gone through there are corresponding
When history operation behavior set, the corresponding historical content entity of the historical operation behavior set is obtained;
At least one Generalization bounds compatible with the historical content entity are selected from preset Generalization bounds set,
The corresponding current recommendation list of the historical content entity is obtained according to the Generalization bounds.
In one of the embodiments, the method also includes:
When current time reaches preset triggered time and the corresponding user identifier of the active user, there is no corresponding
When historical operation behavior set, acquisition, which is preset, each in object set to be recommended presets object to be recommended corresponding historical viewings time
Number;
It presets the corresponding historical impressions of object to be recommended according to each and presets object to be recommended to each and be ranked up,
The current list to be recommended is obtained as object to be recommended according to the object to be recommended of presetting that ranking results choose preset quantity.
A kind of recommendation apparatus, described device include:
Current operation behavior set obtains module, and the current operation behavior collection of terminal is acted on for obtaining active user
It closes, each operation behavior in the current operation behavior set acts on corresponding content substance;
Generalization bounds choose module, compatible with the content substance for selecting from preset Generalization bounds set
At least one Generalization bounds obtains the corresponding object to be recommended of the content substance according to the Generalization bounds of selection;
Recommendation list update module, for updating the Generalization bounds according to the corresponding object to be recommended of the content substance
Corresponding current recommendation list;
Target recommendation list obtains module, and it is corresponding to be used for each Generalization bounds from the preset Generalization bounds set
Object to be recommended is chosen in updated current recommendation list, target is obtained according to the object to be recommended of selection and recommends column
Table.
The Generalization bounds choose module and are also used to segment the content substance in one of the embodiments,
Word segmentation result is obtained, keyword is extracted from the word segmentation result, obtains keyword set, is looked into from preset center set of words
The corresponding centre word of each keyword in each keyword set is looked for, center set of words is obtained, by the center set of words
In each centre word as corresponding first user tag of the active user, calculate the corresponding key of first user tag
Word frequency of the word in the content substance obtains the corresponding weight of first user tag according to the word frequency, according to described
First user tag and the corresponding weight of first user tag generate the corresponding user's portrait of the active user, according to institute
The corresponding user of active user is stated to draw a portrait to obtain the corresponding object to be recommended of the content substance.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device realizes step described in above-mentioned recommended method when executing the computer program.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
Step described in above-mentioned recommended method is realized when row.
Above-mentioned recommended method, device, computer equipment and storage medium act on the current operation row of terminal by obtaining
To gather, each operation behavior in the current operation behavior set acts on corresponding content substance, from preset recommendation
At least one Generalization bounds compatible with the content substance are selected in strategy set, according to the Generalization bounds pair of selection
The content substance carries out recommending that the corresponding object to be recommended of the content substance is calculated, corresponding according to the content substance
Object to be recommended update the corresponding current recommendation list of the Generalization bounds, updated work as from each Generalization bounds are corresponding
Object to be recommended is chosen in preceding recommendation list, target recommendation list is obtained according to the object to be recommended of selection, due to using
A variety of Generalization bounds are recommended simultaneously, can significantly improve the matching degree of object to be recommended and user demand, thus maximum
Limit is met the needs of users.
Detailed description of the invention
Fig. 1 is the application scenario diagram of recommended method in one embodiment;
Fig. 2 is the flow diagram of recommended method in one embodiment;
Fig. 3 is the flow diagram of step S204 in one embodiment;
Fig. 4 is the flow diagram of step S204 in another embodiment;
Fig. 5 is to be drawn a portrait to obtain the flow diagram of object to be recommended according to user in one embodiment;
Fig. 6 is the structural block diagram of recommendation apparatus in one embodiment;
Fig. 7 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Recommended method provided by the present application can be applied in application environment as shown in Figure 1.Wherein, terminal 102 is logical
Network is crossed to be communicated with server 104 by network.Server 104 obtains the current operation row that active user acts on terminal
For set, and the corresponding content substance of current operation behavior set is obtained, according to the type of the content substance from preset recommendation
At least one Generalization bounds that adaptation is chosen in strategy set carry out that recommending data is calculated according to the Generalization bounds of selection,
The corresponding current recommendation list of Generalization bounds is updated with obtained recommending data, chooses and pushes away from updated each recommendation list
Object is recommended, target recommendation list is generated, is recommended according to the target recommendation list to terminal 102.
Wherein, terminal 102 can be, but not limited to be various personal computers, laptop, smart phone, tablet computer
With portable wearable device, server 104 can use the server set of the either multiple server compositions of independent server
Group realizes.
In one embodiment, as shown in Fig. 2, providing a kind of recommended method, in this way applied to the clothes in Fig. 1
It is illustrated for business device, comprising the following steps:
Step S202 obtains active user and acts on the current operation behavior set of terminal, in current operation behavior set
Each operation behavior act on corresponding content substance.
Specifically, content substance refers to that the content object shown in terminal, content object can be with text, voice, image, sounds
The different expression forms such as frequency and video are shown alone or in combination, for example, content substance can be one with text and figure
Piece combines the medical treatment & health to be formed and pushes away text or medical supplies advertisement etc..Operation behavior refers to the received operation object of acting on
Movement, such as may include click, transmission, sharing, collection, purchase, deletion, top set, one of move down and move up or more
Kind.Operation behavior acts on content substance and refers to the operation behavior for the content substance, and content substance is operation behavior
Operation object.It pushes away the link of text for example, clicking one or has sent the consultation information etc. of an online interrogation.
The time for obtaining current operation behavior set, which can according to need, to be configured, such as be can be set and ought received
Active user also can be set to current operation behavior set is obtained in real time after the once-through operation behavior of content substance every default
Such as every two hour in period obtains once-through operation behavior set, or obtains once-through operation behavior every default operation behavior
Set, it is default before obtaining current time when can also be that the current operation behavior of user meets preset recommendation trigger condition
Operation behavior in period obtains current operation behavior.
It is appreciated that each operation behavior may all act on different content substances in current operation behavior set,
It is also likely to be partly or entirely to act on identical content substance.Therefore, content substance may include one or more.In this reality
It applies in example, after server obtains current operation behavior set, the operation object of each operation behavior is obtained, as corresponding content
Entity.For example, it is assumed that in the first operation behavior include click A push away text operation behavior, send B interrogation information operation behavior with
And the operation behavior of collection C commodity, then first content entity sets include that A pushes away text, B interrogation information and C commodity.
Step S204 selects at least one recommendation plan compatible with content substance from preset Generalization bounds set
Slightly, the corresponding object to be recommended of content substance is obtained according to the Generalization bounds of selection.
Specifically, Generalization bounds refer to calculating reference data according to default rule to obtain object to be recommended
Method.In the present embodiment, a variety of Generalization bounds have been stored in advance on server, including but not limited to, have been based on correlation rule
Generalization bounds, Generalization bounds based on article collaborative filtering, the Generalization bounds based on user collaborative filtering, pushing away based on content
Recommend strategy, Generalization bounds, the total sales volume based on commodity or the Generalization bounds of total browsing time based on user's portrait etc..
The reference data type that each Generalization bounds has it to determine in Generalization bounds set, for example, based on use
The Generalization bounds of family portrait, reference data type can be merchandise news, be also possible to text information, cooperateed with based on article
The Generalization bounds of filter, reference data can be merchandise news, but cannot be text informations.The compatible recommendation plan of content substance
Slightly refer to that content substance reference data type corresponding with Generalization bounds is adapted, a content substance can correspond to one
Or multiple Generalization bounds, for example, the Generalization bounds of adaptation can be based on association when content substance is medical treatment & health commodity
The Generalization bounds of rule, the Generalization bounds based on user collaborative filtering, are based on content at the Generalization bounds based on article collaborative filtering
Generalization bounds etc., when content substance is interrogation text or medical treatment & health text, the content substance of adaptation, which can be, to be based on
The Generalization bounds of user's portrait.
In the present embodiment, after server gets content substance, judge data type belonging to the content substance, according to
Data type selects compatible at least one Generalization bounds as target Generalization bounds, is then pushed away according to the Generalization bounds
Calculating is recommended, the corresponding object to be recommended of content substance is obtained.
Step S206 updates the corresponding current recommendation list of Generalization bounds according to the corresponding object to be recommended of content substance.
Specifically, current recommendation list refer to obtain current operation behavior set before each Generalization bounds it is corresponding to
The list that recommended arranges in a certain order.In one embodiment, current recommendation list can be right by its
The Generalization bounds answered carry out off-line calculation according to historical data and obtain, and historical data can be the historical operation behavior collection of user
It closes, is also possible to historical viewings amount, history sales volume of commodity etc..The corresponding current recommendation list of Generalization bounds is updated to refer to
Object to be recommended in recommendation list is updated, specifically, can be the corresponding object to be recommended of content substance is straight
It connects and is added into recommendation list, reconfigure to obtain recommendation list with object to be recommended original in recommendation list;It is also possible to pair
Object to be recommended in the corresponding object to be recommended of content substance and recommendation list is ranked up according to its corresponding recommendation, choosing
The forward preset quantity that sorts object composition recommendation list to be recommended is taken, wherein recommendation must be that Generalization bounds carry out recommendation meter
The numerical value obtained when calculation can be scoring, similarity, support or the confidence level of object to be recommended according to the difference of Generalization bounds
Deng.
Step S208, the corresponding updated current recommendation list of each Generalization bounds from preset Generalization bounds set
It is middle to choose object to be recommended, target recommendation list is obtained according to the object to be recommended of selection.
In the present embodiment, server can be regular from the corresponding current recommendation of each Generalization bounds according to what is be previously set
Object to be recommended is chosen in list.Specifically, can be chosen from each current list to be recommended equal amount wait push away
Object is recommended, for example, needing 20 objects to be recommended in target recommendation list, there are four recommend altogether in preset Generalization bounds set
Strategy, each corresponding recommendation list of strategy to be recommended, then chosen respectively from each recommendation list respectively 5 it is to be recommended
Object;Or weight is arranged to each Generalization bounds in advance, it is to be recommended right to choose from each list to be recommended according to weight ratio
As for example, the corresponding weight ratio of four Generalization bounds is 1:2:1:1, then respectively from four recommendation lists in the example above
The object number to be recommended chosen is 5,10,5,5.
Further, it after server gets object to be recommended, treats recommended and is arranged according to preset ranking factor
Sequence obtains target recommendation list, wherein ranking factor includes but is not limited to order numbers, pageview, cost rate, amount of collection, click
Amount etc..
In one embodiment, it after server obtains target recommendation list, further obtains each in target recommendation list
Then the inventory data of commodity is filtered target recommendation list according to stockpile number, filters out library in target recommendation list
Deposit the commodity to be recommended that quantity is 0.
Further, target recommendation list is sent to terminal and carries out recommendation by server shows.
In above-mentioned recommended method, server obtains the current operation behavior set for acting on terminal, current operation behavior collection
Each operation behavior in conjunction acts on corresponding content substance, selection and content substance phase from preset Generalization bounds set
At least one Generalization bounds of adaptation carry out recommending that content substance pair is calculated according to the Generalization bounds of selection to content substance
The object to be recommended answered updates the corresponding current recommendation list of Generalization bounds according to the corresponding object to be recommended of content substance, from
Object to be recommended is chosen in the corresponding current recommendation list of each Generalization bounds, target is obtained according to the object to be recommended of selection and is pushed away
List is recommended, is recommended simultaneously due to using a variety of Generalization bounds, object to be recommended and user demand can be significantly improved
Matching degree, to meet the needs of users to greatest extent.
In one embodiment, as shown in figure 3, to obtain content substance according to the Generalization bounds of selection corresponding to be recommended
Object, comprising:
Step S302, segments content substance, obtains word segmentation result.
In the present embodiment, content substance be can be including at least one of interrogation data and medical treatment & health text.Specifically,
Interrogation data refer to that user carries out data caused by online interrogation in terminal, can be text information, voice messaging or view
Frequency information needs first to convert voice messaging or video information, obtains its correspondence when for voice messaging or video information
Text information, specifically, to voice messaging carry out speech recognition obtain text information, video information is first extracted corresponding
Then audio-frequency information carries out speech recognition to audio-frequency information and obtains text information.Medical treatment & health text refers to and medical treatment & health
Relevant text information, such as can be, medical treatment & health is relevant to push away text, the relevant advertisement of medical treatment & health product.
In the present embodiment, server first segments content substance, obtains word segmentation result, and word segmentation result refers to
The sequence of terms of the word composition one by one obtained after participle.Such as, the participle obtained after " open cerebellar hemorrhage " participle
It as a result can be with are as follows: opening/cerebellum/bleeding.Content substance is segmented, specifically, can be incited somebody to action first according to punctuation mark
It is divided into the complete sentence of a rule to content substance, then word segmentation processing is carried out to the sentence of each cutting, such as using character string
Matched segmenting method carries out word segmentation processing to the sentence of each cutting, such as Forward Maximum Method method, the sentence of a cutting
In character string segment from left to right;Alternatively, reversed maximum matching method, the character string in the sentence of a cutting from the right side to
A left side segments;Alternatively, shortest path segments method, it is minimum that the word number cut out is required inside the character string in the sentence of a cutting
's;It is forward and reverse while carrying out participle matching alternatively, two-way maximum matching method.Also using meaning of a word participle method to each cutting
Sentence carries out word segmentation processing, and meaning of a word participle method is a kind of segmenting method of machine talk judgement, is believed using syntactic information and semanteme
Breath segments to handle Ambiguity.Word segmentation processing also is carried out to the sentence of each cutting using statistical morphology, from current
In the historical search record of the historical search record of interrogation user or public users, according to the statistics of phrase, can count some two
The frequency that a adjacent word occurs is more, then can segment using the two adjacent words as phrase.
Step S304, extracts keyword from word segmentation result, obtains keyword set.
Specifically, keyword refers to the corresponding word of the key message in content substance.In one embodiment, it extracts
The step of keyword, is as follows:
1) part-of-speech tagging processing is carried out to each word in word segmentation result, and filters out stop words, only retain specified part of speech
Word obtain candidate keywords such as noun, verb, adjective.
2) candidate keywords figure G=(V, E) is constructed, wherein V is node collection, is made of the candidate keywords that (1) generates, so
The side between two o'clock is appointed using cooccurrence relation (co-occurrence) construction afterwards, there are sides only when they are right between two nodes
The vocabulary answered co-occurrence in the window that length is K, K indicate window size, i.e., most K words of co-occurrence.
3) referring to following formula, the weight of each node of iterative diffusion, until convergence.
Wherein, wherein V indicates that word node, WS indicate word node weights in formula.W indicates side right weight, the word formed according to side
The similarity of node obtains, and d is damped coefficient, and value range is 0 to 1, and it is any other to represent a certain specified point direction from figure
The probability of point, general value are that 0.85, In indicates the point set for being directed toward the word node, and Out indicates the word section that the word node is directed toward
The point set of point.
5) to word node weight carry out Bit-reversed, obtain corresponding word node according to the size of weight, from big to small according to
The word node of secondary selection preset quantity, using the corresponding word of word node selected as keyword.
Step S306 searches the corresponding center of each keyword in each keyword set from preset center set of words
Word obtains center set of words, using centre word each in the set of words of center as corresponding first user tag of active user.
Specifically, one social networks map (igraph) of building in advance, social networks map for descriptor and word it
Between relationship, the corresponding centre word of keyword is then obtained from social networks map using Random Walk Algorithm, wherein center
Word refers to the center contacted between word and word in social networks map, such as " menstruation;Dysmenorrhea;Menstrual period;Big aunt;Physiology phase;Month
Menstrual period;Irregular menstruation;Menstrual blood;Dysmenorrhoea;Official holiday;The onset of ovulation;The corresponding centre word of aunt ... " is " menstruation ".In the present embodiment
In, for each keyword, primary iteration word node x, walk for the first time long λ and control are set in social networks map in advance
Precision ∈ (terminating for control algolithm, a very small positive number) processed.Obtain default iteration control times N and current iteration time
Number K, when current iteration number is less than given the number of iterations, that is, K < N when, the random N-dimensional generated between one (- 1,1) to
Measure u=(u1,u2,...,un) (- 1 < ui< 1, i=1,2 ..., n), and be standardized asEnable x1=x
+ λ u' completes first step migration.Preset function f (x) value is calculated, which is the function of many variables containing n variable, x=
(x1,x2,...,xn) it is N-dimensional vector.As f (x1) < f (x) when, i.e., by the x1Corresponding word node resets K as initial point
=1, x1For x, λ and ∈ is given again and is iterated calculating.As f (x1) > f (x) when, continue migration calculating.It is pre- when reaching
If when the number of iterations and λ < ∈, then current primary iteration word is optimal word, and using the optimal word as the corresponding center of keyword
Word.
Further, using the centre word as corresponding first label of active user.
Step S308 calculates word frequency of the corresponding keyword of the first user tag in content substance, is obtained according to word frequency
The corresponding weight of first user tag.
Specifically, word frequency refers to the frequency that some word occurs in content substance.It in one embodiment, can be to word
Frequency is normalized, it may be assumed that the total word number of total degree/content substance that word frequency=some word occurs.
Further, server obtains the corresponding weight of the user tag according to the corresponding word frequency of the first user tag.One
It, can be directly using the word frequency after normalization as the corresponding weight of the first label in a embodiment.
Step S310 generates the corresponding use of active user according to the first user tag and the corresponding weight of the first user tag
Family portrait, draws a portrait to obtain the corresponding object to be recommended of content substance according to the corresponding user of active user.
Specifically, draw a portrait to obtain the corresponding object to be recommended of content substance according to the corresponding user of active user can for server
Specifically, being ranked up according to weight to label directly according to label recommendations object to be recommended relevant to label,
Then the forward label of selected and sorted is recommended as target labels according to target labels.Such as when target labels are " mother and baby "
When, the relevant commodity of mother and baby can be recommended to user;It is also possible to similar to other users according to label calculating active user
Then degree, the user that similarity is greater than preset threshold obtain the history purchaser record for referring to user as user is referred to, according to
History purchaser record recommends active user.
In the present embodiment, using centre word as user tag, user's portrait is obtained, and recommend according to user's portrait,
Object to be recommended and the matching degree of active user are substantially increased, can preferably meet the needs of active user.
In one embodiment, content substance includes merchandise news, as shown in figure 4, the Generalization bounds according to selection obtain
The corresponding object to be recommended of content substance, comprising:
Step S402, obtains the corresponding goods number of merchandise news, searches corresponding commodity classification, root according to goods number
Corresponding history weight is obtained according to commodity classification.
Specifically, commodity classification refers to classification belonging to commodity, presets between commodity classification and goods number
Mapping relations can be found belonging to the merchandise news after getting merchandise news corresponding goods number according to goods number
Commodity classification have corresponding history weight to each commodity classification, history weight exists for measuring such purpose commodity
In preset time period before current operation behavior set by browsing time number, the number of browsing is more, and history weight is got over
Greatly, if not browsed, history weight is 0, wherein preset time period can be one day before current operation behavior set
Or several days or several hours, it specifically can according to need and set.
It is appreciated that in the present embodiment, merchandise news may include the information of multiple commodity, these commodity can be category
In identical commodity classification, it is also possible to belong to different commodity classifications, therefore, in the present embodiment, commodity classification can be one
It is a to be also possible to multiple, in the present embodiment, it is referred to as with commodity classification, it will be appreciated by those skilled in the art that should
The classification number that commodity classification includes is not defined.
Step S404 calculates the corresponding current browsing time of commodity classification according to current operation behavior set, will be current clear
Number of looking at standardization.
Specifically, according to each operation behavior each commodity corresponding to commodity classification in user's current operation behavior set
Browsing time counted respectively, then add up obtain the corresponding current browsing time of commodity classification.It further, will be current clear
Number of looking at standardization.
Browsing time standardization, which refers to, is allowed to browsing time bi-directional scaling to fall into a small specific sections.One
In a embodiment, browsing classification browsing time is subjected to linear transformation using deviation standardization, the data after being standardized.
Step S406 calculates the corresponding current power of commodity classification according to history weight and standardized current browsing time
Weight.
Specifically, weight calculation formula is used according to history weight and standardized browsing time
The corresponding present weight of commodity classification is obtained, and obtained present weight is saved.Wherein W be commodity classification weight, W' be for
History weight, T are the data after browsing time standardization.
Step S408 is ranked up commodity classification according to present weight, and the quotient of preset quantity is obtained according to ranking results
Category mesh is as end article classification, using end article classification as the corresponding second user label of active user, according to second
User tag and corresponding weight generate the corresponding user's portrait of active user.
Specifically, descending sort can be carried out to commodity classification according to weight size, chooses the preceding preset quantity that sorts
A commodity classification is as end article classification, using these end article classifications as the corresponding second user label of active user.
Step S410 draws a portrait to obtain the corresponding object to be recommended of content substance according to the corresponding user of active user.
Specifically, server can directly be recommended according to the corresponding user tag of user's portrait, because of user tag
For commodity classification, each commodity class has corresponded to a variety of commodity now, therefore, can from the corresponding commodity of commodity classification into
Row is chosen, as object recommendation to be recommended to active user;It is also possible to calculate active user and other users according to label
Then similarity, the user that similarity is greater than preset threshold obtain the history purchaser record of the user, root as user is referred to
Active user is recommended according to history purchaser record.
In the present embodiment, using commodity classification as user tag, user's portrait is obtained, and is pushed away according to user's portrait
It recommends, substantially increases object to be recommended and the matching degree of active user, can preferably meet the needs of active user.
In one embodiment, as shown in figure 5, drawing a portrait to obtain content substance according to the corresponding user of active user corresponding
Object to be recommended, comprising:
Step S502 obtains the reference user set for meeting preset rules.
Specifically, the set being made of the reference user for meeting preset rules, ginseng therein are referred to reference to user's set
User is examined for calculating similarity with active user.Preset rules can be set as needed.It in one embodiment, can be with
It is that the user chosen in a certain period is used as with reference to user.In another embodiment, since user is in registration, server is logical
The attribute information of user, such as gender, age, occupation, residence can be often acquired, accordingly it is also possible to be selection and active user
A certain attribute such as occupation, age identical user be used as refer to user.
Step S504 draws a portrait according to the corresponding user of active user and calculates active user and with reference to each ginseng in user's set
Examine the similarity of user.
Specifically, user's portrait first can be generated with reference to user to each in reference user set, then calculates current use
The corresponding user in family portrait and the similarity drawn a portrait with reference to each in user's set with reference to the corresponding user of user.
It in one embodiment, can be corresponding with reference to user by each in active user and reference user's set respectively
User tag is ranked up according to weight, and according to the maximum identical quantity label of ranking results weight selection, these are marked
Sign a vector of the co-map into vector space, then calculate the corresponding vector of active user it is corresponding with user is referred to
The corresponding user's portrait of cosine similarity between amount, as active user user corresponding with reference user draws a portrait similar
Degree.Wherein, it user tag is mapped into vector space can be and user tag is input to preparatory trained space vector mould
In type, corresponding vector is obtained;It is also possible to calculate the corresponding term vector of each label using word2vec model, then presses
Sequence, which is combined, according to weight size obtains corresponding vector.
In another embodiment, each corresponding user tag of active user can be referred to user with each respectively
Corresponding label is matched one by one, is carried out weight calculation to reference user according to the corresponding weight of the label of successful match, is obtained
To the corresponding weight of reference user, using cumulative obtained weighted value as similarity.For example, it is assumed that active user corresponding two
A user tag A and B, weight are respectively 0.6,0.4, and there are three refer to user: the corresponding user tag of user 1 is A, C, user
2 corresponding user tags are B, D, the corresponding user tag of user 3 is E, F, and the matching degree that pool finally obtains user 1 is 0.6, is used
The matching degree at family 2 is 0.4, and the matching degree of user 3 is 0.
Step S506 obtains similarity and is greater than the reference user of preset threshold as object reference user.
Wherein, preset threshold can be set or be changed in advance as needed.
Step S508 obtains the corresponding historical operation behavior set of object reference user, corresponding according to object reference user
Historical operation behavior set obtain the corresponding object to be recommended of content substance.
Specifically, the corresponding historical operation behavior set of object reference user includes the browsing to commodity, purchase, collection
Deng it is corresponding can to obtain the historical operation behavior set after getting the historical operation behavior set of target user for server
Content substance, i.e. merchandise news, then using the corresponding commodity of the merchandise news as object to be recommended.For example, if with reference to
User 1 has purchased commodity A, commodity B, then can be using commodity A and commodity B as object to be recommended.
In the present embodiment, by calculating the similarity between user's portrait, to select with reference to user, then according to reference
The historical behavior set of user recommends active user, compared to directly according to user draw a portrait label recommend, this
The range that kind mode is recommended is bigger, the potential demand of active user can be excavated, to mention object to be recommended and user significantly
The matching degree of demand.
In one embodiment, the corresponding user identifier of active user is obtained, the corresponding current recommendation of user identifier is searched
List collection, there are corresponding Generalization bounds for each current recommendation list in current recommendation list set.
Wherein, active user refers to the user for currently passing through terminal login service device, wherein login can be to register
The identity of user logs in, and is also possible to log in tourist's identity, when being logged in registering user identity, Yong Hubiao
Know account when can be the user's registration, such as E-mail address, cell-phone number, QQ number, with the user of tourist's identity logs,
Corresponding user identifier is the temporary account that server distributes automatically, can be by the letter of presetting digit capacity, number, symbol or combinations thereof
Composition, user identifier are used for the identity of unique identification active user.
In the present embodiment, for each active user, after server gets its user identifier, user mark is searched
Know corresponding current recommendation list set, includes at least one current recommendation list in current recommendation list set.In a reality
It applies in example, current recommendation list, which can be, to be obtained by its corresponding Generalization bounds according to historical data progress off-line calculation, history
Data can be the historical operation behavior set of user, be also possible to historical viewings amount, history sales volume of commodity etc..In this reality
It applies in example, the Generalization bounds in current recommendation list set in each current recommendation list and preset Generalization bounds set are one by one
Corresponding, i.e., each recommendation list is calculated by a Generalization bounds respectively.
In one embodiment, the above method includes: when current time reaches the preset triggered time and user identifier is deposited
In corresponding historical operation behavior set, the corresponding historical content entity of historical operation behavior set is obtained;It is pushed away from preset
It recommends and selects at least one Generalization bounds compatible with historical content entity in strategy set, obtained in history according to Generalization bounds
Hold the corresponding current recommendation list of entity.
Wherein, the triggered time refers to be previously set in server a time, when reaching the triggered time, service
Device according to can according to historical data carry out off-line calculation obtain recommendation list.Triggered time one is any time in one day,
The triggering period can be set as needed, such as can be triggering in three days is primary, and triggering in one week is primary etc..
In one embodiment, the above method further include: when current time reaches preset triggered time active user couple
When corresponding historical operation behavior set is not present in the user identifier answered, acquisition preset in object set to be recommended it is each it is default to
The corresponding historical impressions of recommended;According to each corresponding historical impressions of object to be recommended of presetting to each default
Object to be recommended is ranked up, and is obtained according to the object to be recommended of presetting that ranking results choose preset quantity as object to be recommended
Current list to be recommended.
Specifically, preset object set to be recommended and refer to the set formed by presetting object to be recommended, wherein it is default to
The preset object possible as object to be recommended that recommended refers to.For example, can be by current all quotient on sale
Product obtain presetting object set to be recommended as object to be recommended is preset.Historical impressions refer to before current time
Preset time period in some preset the total degree that object to be recommended is browsed.
In the present embodiment, historical operation behavior is not present in active user, i.e. the user is first time game server
New user can obtain to preset at this time and each in object set to be recommended preset the corresponding historical impressions of object to be recommended, root
Recommended according to historical impressions.
It should be understood that although each step in the flow chart of Fig. 2-5 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-5
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
In one embodiment, as shown in fig. 6, providing a kind of recommendation apparatus 600, comprising:
Current operation behavior set obtains module 602, and the current operation behavior of terminal is acted on for obtaining active user
Gather, each operation behavior in current operation behavior set acts on corresponding content substance;
Generalization bounds choose module 604, compatible with content substance for selecting from preset Generalization bounds set
At least one Generalization bounds obtains the corresponding object to be recommended of content substance according to the Generalization bounds of selection;
Recommendation list update module 606, it is corresponding for updating Generalization bounds according to the corresponding object to be recommended of content substance
Current recommendation list;
Target recommendation list obtains module 608, and it is corresponding to be used for each Generalization bounds from preset Generalization bounds set
Object to be recommended is chosen in updated current recommendation list, target recommendation list is obtained according to the object to be recommended of selection.
In one embodiment, Generalization bounds choose module 604 and are also used to segment content substance, obtain participle knot
Fruit;Keyword is extracted from word segmentation result, obtains keyword set;Each keyword set is searched from preset center set of words
The corresponding centre word of each keyword, obtains center set of words in conjunction, uses using centre word each in the set of words of center as current
Corresponding first user tag in family;Word frequency of the corresponding keyword of the first user tag in content substance is calculated, according to word frequency
Obtain the corresponding weight of the first user tag;Current use is generated according to the first user tag and the corresponding weight of the first user tag
The corresponding user's portrait in family, draws a portrait to obtain the corresponding object to be recommended of content substance according to the corresponding user of active user.
In one embodiment, content substance includes merchandise news, and Generalization bounds choose module 604 and are also used to obtain commodity
The corresponding goods number of information, searches corresponding commodity classification according to goods number, obtains corresponding history according to commodity classification
Weight;The corresponding current browsing time of commodity classification is calculated according to current operation behavior set, current browsing time is standardized;
The corresponding present weight of commodity classification is calculated according to history weight and standardized current browsing time;According to present weight to quotient
Category mesh is ranked up, according to the commodity classification of ranking results acquisition preset quantity as end article classification, by end article
Classification generates active user couple as the corresponding second user label of active user, according to second user label and corresponding weight
The user's portrait answered;It is drawn a portrait to obtain the corresponding object to be recommended of content substance according to the corresponding user of active user.
In one embodiment, Generalization bounds choose module 604 and are also used to obtain the reference user collection for meeting preset rules
It closes;Being drawn a portrait according to the corresponding user of active user, it is each with reference to the similar of user to reference user's set to calculate active user
Degree;It obtains similarity and is greater than the reference user of preset threshold as object reference user;Obtain that object reference user is corresponding goes through
It is corresponding to be recommended to obtain content substance according to the corresponding historical operation behavior set of object reference user for history operation behavior set
Object.
In one embodiment, above-mentioned apparatus further include: current recommendation list obtains module, for reaching when current time
There are when corresponding historical operation behavior set, obtain history behaviour for preset triggered time and the corresponding user identifier of active user
Make the corresponding historical content entity of behavior set;It is selected from preset Generalization bounds set compatible with historical content entity
At least one Generalization bounds obtain the corresponding current recommendation list of historical content entity according to Generalization bounds.
In one embodiment, current recommendation list obtains module and is also used to reach the preset triggered time when current time
When corresponding historical operation behavior set is not present in the corresponding user identifier of active user, acquisition is preset in object set to be recommended
It is each to preset the corresponding historical impressions of object to be recommended;The corresponding historical impressions of object to be recommended are preset according to each
It presets object to be recommended to each and is ranked up, be used as according to the object to be recommended of presetting that ranking results choose preset quantity wait push away
It recommends object and obtains current list to be recommended.
Specific about recommendation apparatus limits the restriction that may refer to above for recommended method, and details are not described herein.
Modules in above-mentioned recommendation apparatus can be realized fully or partially through software, hardware and combinations thereof.Above-mentioned each module can
It is embedded in the form of hardware or independently of in the processor in computer equipment, computer can also be stored in a software form and set
In memory in standby, the corresponding operation of the above modules is executed in order to which processor calls.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 7.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing operation behavior data, information of goods information data etc..The network interface of the computer equipment is used
It is communicated in passing through network connection with external terminal.To realize a kind of recommended method when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Fig. 7, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment, including memory and processor are provided, which is stored with
Computer program, the processor perform the steps of acquisition active user and act on the current of terminal when executing computer program
Operation behavior set, each operation behavior in current operation behavior set act on corresponding content substance;It is pushed away from preset
It recommends and selects at least one Generalization bounds compatible with content substance in strategy set, obtain content according to the Generalization bounds of selection
The corresponding object to be recommended of entity;The corresponding current recommendation column of Generalization bounds are updated according to the corresponding object to be recommended of content substance
Table;It is to be recommended right to choose from the corresponding updated current recommendation list of Generalization bounds each in preset Generalization bounds set
As obtaining target recommendation list according to the object to be recommended of selection.
In one embodiment, the corresponding object to be recommended of content substance is obtained according to the Generalization bounds of selection, comprising: right
Content substance is segmented, and word segmentation result is obtained;Keyword is extracted from word segmentation result, obtains keyword set;From preset
The corresponding centre word of each keyword in each keyword set is searched in the set of words of center, center set of words is obtained, by center
Each centre word is as corresponding first user tag of active user in set of words;Calculate the corresponding keyword of the first user tag
Word frequency in content substance obtains the corresponding weight of the first user tag according to word frequency;According to the first user tag and first
The corresponding weight of user tag generates the corresponding user's portrait of active user, draws a portrait to obtain according to the corresponding user of active user interior
Hold the corresponding object to be recommended of entity.
In one embodiment, content substance includes merchandise news, obtains content substance pair according to the Generalization bounds of selection
The object to be recommended answered, comprising: obtain the corresponding goods number of merchandise news, corresponding commodity class is searched according to goods number
Mesh obtains corresponding history weight according to commodity classification;It is corresponding current that commodity classification is calculated according to current operation behavior set
Browsing time standardizes current browsing time;Commodity classification is calculated according to history weight and standardized current browsing time
Corresponding present weight;Commodity classification is ranked up according to present weight, the commodity of preset quantity are obtained according to ranking results
Classification is used using end article classification as the corresponding second user label of active user according to second as end article classification
Family label and corresponding weight generate the corresponding user's portrait of active user;It draws a portrait to obtain according to the corresponding user of active user interior
Hold the corresponding object to be recommended of entity.
In one embodiment, it is corresponding to be recommended right to draw a portrait to obtain content substance according to the corresponding user of active user
As, comprising: obtain the reference user set for meeting preset rules;It is drawn a portrait according to the corresponding user of active user and calculates active user
With each similarity with reference to user in reference user set;It obtains similarity and is greater than the reference user of preset threshold as target
With reference to user;The corresponding historical operation behavior set of object reference user is obtained, is grasped according to the corresponding history of object reference user
Make behavior set and obtains the corresponding object to be recommended of content substance.
In one embodiment, it also performs the steps of when processor executes computer program when current time reaches pre-
If triggered time and active user corresponding user identifier there are when corresponding historical operation behavior set, obtain historical operation
The corresponding historical content entity of behavior set;From preset Generalization bounds set select it is compatible with historical content entity to
A kind of few Generalization bounds, obtain the corresponding current recommendation list of historical content entity according to Generalization bounds.
In one embodiment, it also performs the steps of when processor executes computer program when current time reaches pre-
If triggered time and active user corresponding user identifier when corresponding historical operation behavior set is not present, obtain it is default to
It is each in recommended set to preset the corresponding historical impressions of object to be recommended;Object correspondence to be recommended is preset according to each
Historical impressions preset object to be recommended to each and be ranked up, choose the default wait push away of preset quantity according to ranking results
Object is recommended as object to be recommended and obtains current list to be recommended.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor obtains the current operation behavior set that active user acts on terminal, when
Each operation behavior in preceding operation behavior set acts on corresponding content substance;It is selected from preset Generalization bounds set
It is corresponding to be recommended to obtain content substance according to the Generalization bounds of selection at least one Generalization bounds compatible with content substance
Object;The corresponding current recommendation list of Generalization bounds is updated according to the corresponding object to be recommended of content substance;From preset recommendation
In strategy set choose object to be recommended in the corresponding updated current recommendation list of each Generalization bounds, according to selection to
Recommended obtains target recommendation list.
In one embodiment, the corresponding object to be recommended of content substance is obtained according to the Generalization bounds of selection, comprising: right
Content substance is segmented, and word segmentation result is obtained;Keyword is extracted from word segmentation result, obtains keyword set;From preset
The corresponding centre word of each keyword in each keyword set is searched in the set of words of center, center set of words is obtained, by center
Each centre word is as corresponding first user tag of active user in set of words;Calculate the corresponding keyword of the first user tag
Word frequency in content substance obtains the corresponding weight of the first user tag according to word frequency;According to the first user tag and first
The corresponding weight of user tag generates the corresponding user's portrait of active user, draws a portrait to obtain according to the corresponding user of active user interior
Hold the corresponding object to be recommended of entity.
In one embodiment, content substance includes merchandise news, obtains content substance pair according to the Generalization bounds of selection
The object to be recommended answered, comprising: obtain the corresponding goods number of merchandise news, corresponding commodity class is searched according to goods number
Mesh obtains corresponding history weight according to commodity classification;It is corresponding current that commodity classification is calculated according to current operation behavior set
Browsing time standardizes current browsing time;Commodity classification is calculated according to history weight and standardized current browsing time
Corresponding present weight;Commodity classification is ranked up according to present weight, the commodity of preset quantity are obtained according to ranking results
Classification is used using end article classification as the corresponding second user label of active user according to second as end article classification
Family label and corresponding weight generate the corresponding user's portrait of active user;It draws a portrait to obtain according to the corresponding user of active user interior
Hold the corresponding object to be recommended of entity.
In one embodiment, it is corresponding to be recommended right to draw a portrait to obtain content substance according to the corresponding user of active user
As, comprising: obtain the reference user set for meeting preset rules;It is drawn a portrait according to the corresponding user of active user and calculates active user
With each similarity with reference to user in reference user set;It obtains similarity and is greater than the reference user of preset threshold as target
With reference to user;The corresponding historical operation behavior set of object reference user is obtained, is grasped according to the corresponding history of object reference user
Make behavior set and obtains the corresponding object to be recommended of content substance.
In one embodiment, it is also performed the steps of when computer program is executed by processor when current time reaches
There are when corresponding historical operation behavior set, obtain history behaviour for preset triggered time and the corresponding user identifier of active user
Make the corresponding historical content entity of behavior set;It is selected from preset Generalization bounds set compatible with historical content entity
At least one Generalization bounds obtain the corresponding current recommendation list of historical content entity according to Generalization bounds.
In one embodiment, it is also performed the steps of when computer program is executed by processor when current time reaches
When corresponding historical operation behavior set is not present in preset triggered time and the corresponding user identifier of active user, obtain default
It is each in object set to be recommended to preset the corresponding historical impressions of object to be recommended;Object pair to be recommended is preset according to each
The historical impressions answered are preset object to be recommended and are ranked up to each, according to ranking results choose preset quantity it is default to
Recommended obtains current list to be recommended as object to be recommended.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM in a variety of forms may be used
, such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM),
Enhanced SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) are direct
RAM (RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of recommended method, which comprises
Obtain the current operation behavior set that active user acts on terminal, each operation in the current operation behavior set
Behavior acts on corresponding content substance;
At least one Generalization bounds compatible with the content substance are selected from preset Generalization bounds set, according to selection
The Generalization bounds obtain the corresponding object to be recommended of the content substance;
The corresponding current recommendation list of the Generalization bounds is updated according to the corresponding object to be recommended of the content substance;
From the corresponding updated current recommendation list of Generalization bounds each in the preset Generalization bounds set choose to
Recommended obtains target recommendation list according to the object to be recommended of selection.
2. the method according to claim 1, wherein the Generalization bounds according to selection obtain in described
Hold the corresponding object to be recommended of entity, comprising:
The content substance is segmented, word segmentation result is obtained;
Keyword is extracted from the word segmentation result, obtains keyword set;
From searching the corresponding centre word of each keyword in each keyword set in preset center set of words, obtaining
Heart set of words, using each centre word in the center set of words as corresponding first user tag of the active user;
Word frequency of the corresponding keyword of first user tag in the content substance is calculated, institute is obtained according to the word frequency
State the corresponding weight of the first user tag;
The corresponding use of the active user is generated according to first user tag and the corresponding weight of first user tag
Family portrait, draws a portrait to obtain the corresponding object to be recommended of the content substance according to the corresponding user of the active user.
3. described according to choosing the method according to claim 1, wherein the content substance includes merchandise news
The Generalization bounds selected obtain the corresponding object to be recommended of the content substance, comprising:
The corresponding goods number of the merchandise news is obtained, corresponding commodity classification is searched according to the goods number, according to institute
It states commodity classification and obtains corresponding history weight;
The corresponding current browsing time of the commodity classification is calculated according to the current operation behavior set, by the current browsing
Number standardization;
The corresponding present weight of the commodity classification is calculated according to the history weight and the standardized current browsing time;
The commodity classification is ranked up according to the present weight, the commodity classification of preset quantity is obtained according to ranking results
As end article classification, using the end article classification as the corresponding second user label of the active user, according to institute
It states second user label and corresponding weight generates the corresponding user's portrait of active user;
It is drawn a portrait to obtain the corresponding object to be recommended of the content substance according to the corresponding user of the active user.
4. the method according to any one of Claims 2 or 3, which is characterized in that described according to the active user couple
The user answered draws a portrait to obtain the corresponding object to be recommended of the content substance, comprising:
Obtain the reference user set for meeting preset rules;
It is drawn a portrait according to the corresponding user of the active user and calculates each ginseng in the active user and the reference user set
Examine the similarity of user;
It obtains similarity and is greater than the reference user of preset threshold as object reference user;
The corresponding historical operation behavior set of the object reference user is obtained, according to the corresponding history of the object reference user
Operation behavior set obtains the corresponding object to be recommended of the content substance.
5. the method according to claim 1, wherein the method also includes:
When current time reaches preset triggered time and the corresponding user identifier of the active user, there are corresponding history to grasp
When making behavior set, the corresponding historical content entity of the historical operation behavior set is obtained;
At least one Generalization bounds compatible with the historical content entity are selected from preset Generalization bounds set, according to
The Generalization bounds obtain the corresponding current recommendation list of the historical content entity.
6. the method according to claim 1, wherein the method also includes:
When current time reaches preset triggered time and the corresponding user identifier of the active user, there is no corresponding history
When operation behavior set, acquisition, which is preset, each in object set to be recommended presets the corresponding historical impressions of object to be recommended;
It presets the corresponding historical impressions of object to be recommended according to each and presets object to be recommended to each and be ranked up, according to
The object to be recommended of presetting that ranking results choose preset quantity obtains the current list to be recommended as object to be recommended.
7. a kind of recommendation apparatus, which is characterized in that described device includes:
Current operation behavior set obtains module, and the current operation behavior set of terminal, institute are acted on for obtaining active user
The each operation behavior stated in current operation behavior set acts on corresponding content substance;
Generalization bounds choose module, and it is compatible with the content substance at least to be used for the selection from preset Generalization bounds set
One Generalization bounds obtains the corresponding object to be recommended of the content substance according to the Generalization bounds of selection;
Recommendation list update module, it is corresponding for updating the Generalization bounds according to the corresponding object to be recommended of the content substance
Current recommendation list;
Target recommendation list obtains module, for the corresponding update of Generalization bounds each from the preset Generalization bounds set
Object to be recommended is chosen in current recommendation list afterwards, target recommendation list is obtained according to the object to be recommended of selection.
8. being also used to the method according to the description of claim 7 is characterized in that the Generalization bounds choose module to the content
Entity is segmented, and word segmentation result is obtained, and is extracted keyword from the word segmentation result, is obtained keyword set, from preset
The corresponding centre word of each keyword in each keyword set is searched in the set of words of center, obtains center set of words, it will
Each centre word calculates first user as corresponding first user tag of the active user in the center set of words
It is corresponding to obtain first user tag according to the word frequency for word frequency of the corresponding keyword of label in the content substance
Weight generates the corresponding use of the active user according to first user tag and the corresponding weight of first user tag
Family portrait, draws a portrait to obtain the corresponding object to be recommended of the content substance according to the corresponding user of the active user.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 6 the method when executing the computer program.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 6 is realized when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810725366.3A CN109062994A (en) | 2018-07-04 | 2018-07-04 | Recommended method, device, computer equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810725366.3A CN109062994A (en) | 2018-07-04 | 2018-07-04 | Recommended method, device, computer equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109062994A true CN109062994A (en) | 2018-12-21 |
Family
ID=64818764
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810725366.3A Withdrawn CN109062994A (en) | 2018-07-04 | 2018-07-04 | Recommended method, device, computer equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109062994A (en) |
Cited By (36)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109710853A (en) * | 2018-12-27 | 2019-05-03 | 深圳铁盒子文化科技发展有限公司 | A kind of artificial intelligence classified matching method and system |
CN109766492A (en) * | 2018-12-24 | 2019-05-17 | 中国联合网络通信集团有限公司 | Learn recommended method, device, equipment and readable medium |
CN109885770A (en) * | 2019-02-20 | 2019-06-14 | 杭州威佩网络科技有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
CN109933729A (en) * | 2019-03-28 | 2019-06-25 | 广州麦迪森在线医疗科技有限公司 | A kind of academic information recommended method of medical treatment based on user preference and system |
CN110083634A (en) * | 2019-03-19 | 2019-08-02 | 中国平安人寿保险股份有限公司 | Order processing method, apparatus, equipment and storage medium based on data analysis |
CN110413872A (en) * | 2019-01-21 | 2019-11-05 | 北京京东尚科信息技术有限公司 | Method and apparatus for showing information |
CN110533515A (en) * | 2019-09-04 | 2019-12-03 | 深圳创新奇智科技有限公司 | A kind of the electric business personalized recommendation method and device of high low latency of handling up |
CN110659025A (en) * | 2019-08-26 | 2020-01-07 | 联想(北京)有限公司 | Recommendation method and device |
CN110866180A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, server and storage medium |
CN111191129A (en) * | 2019-12-27 | 2020-05-22 | 广东德诚科教有限公司 | Professional recommendation method and device, computer equipment and storage medium |
CN111199428A (en) * | 2020-01-03 | 2020-05-26 | 江苏苏宁物流有限公司 | Commodity recommendation method and device, storage medium and computer equipment |
CN111368181A (en) * | 2018-12-25 | 2020-07-03 | 阿里巴巴集团控股有限公司 | Information recommendation method, information display method and device |
CN111371938A (en) * | 2018-12-26 | 2020-07-03 | 华为终端有限公司 | Fault detection method and electronic equipment |
CN111400546A (en) * | 2020-03-18 | 2020-07-10 | 腾讯科技(深圳)有限公司 | Video recall method and video recommendation method and device |
CN111476639A (en) * | 2020-04-10 | 2020-07-31 | 深圳市物语智联科技有限公司 | Commodity recommendation strategy determining method and device, computer equipment and storage medium |
CN111523010A (en) * | 2019-02-03 | 2020-08-11 | 阿里巴巴集团控股有限公司 | Recommendation method and device, terminal equipment and computer storage medium |
CN111651669A (en) * | 2020-05-20 | 2020-09-11 | 拉扎斯网络科技(上海)有限公司 | Information recommendation method and device, electronic equipment and computer-readable storage medium |
CN111858915A (en) * | 2020-08-07 | 2020-10-30 | 成都理工大学 | Information recommendation method and system based on label similarity |
CN111984837A (en) * | 2019-05-23 | 2020-11-24 | 浙江口碑网络技术有限公司 | Commodity data processing method, device and equipment |
CN112148958A (en) * | 2019-06-26 | 2020-12-29 | 傲基科技股份有限公司 | Method, apparatus, and computer storage medium for information recommendation |
CN112541787A (en) * | 2020-12-09 | 2021-03-23 | 北京奇艺世纪科技有限公司 | Advertisement recommendation method, system, storage medium and electronic device |
CN112686568A (en) * | 2021-01-11 | 2021-04-20 | 南方电网数字电网研究院有限公司 | Operation and maintenance strategy generation processing method, device, system, equipment and storage medium |
CN112733034A (en) * | 2021-01-21 | 2021-04-30 | 腾讯科技(深圳)有限公司 | Content recommendation method, device, equipment and storage medium |
CN112905895A (en) * | 2021-03-29 | 2021-06-04 | 平安国际智慧城市科技股份有限公司 | Similar item recommendation method, device, equipment and medium |
CN113159810A (en) * | 2020-01-22 | 2021-07-23 | 百度在线网络技术(北京)有限公司 | Policy evaluation method, device, equipment and storage medium |
CN113204638A (en) * | 2021-04-23 | 2021-08-03 | 上海明略人工智能(集团)有限公司 | Recommendation method, system, computer and storage medium based on work session unit |
CN113298277A (en) * | 2021-04-25 | 2021-08-24 | 上海淇玥信息技术有限公司 | Target-based continuous reservation information pushing method and device and electronic equipment |
CN113435202A (en) * | 2021-06-28 | 2021-09-24 | 平安科技(深圳)有限公司 | Product recommendation method and device based on user portrait, electronic equipment and medium |
CN113434660A (en) * | 2021-06-28 | 2021-09-24 | 平安银行股份有限公司 | Product recommendation method, device, equipment and storage medium based on multi-domain classification |
CN113722523A (en) * | 2021-09-01 | 2021-11-30 | 北京百度网讯科技有限公司 | Object recommendation method and device |
CN113761327A (en) * | 2020-06-18 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Scheduling method, device and storage medium of recommendation strategy |
CN113779421A (en) * | 2021-08-13 | 2021-12-10 | 咪咕数字传媒有限公司 | Association recommendation method, device, equipment and computer readable storage medium |
CN114429371A (en) * | 2022-04-06 | 2022-05-03 | 新石器慧通(北京)科技有限公司 | Unmanned vehicle-based commodity marketing method and device, electronic equipment and storage medium |
CN114936325A (en) * | 2022-07-20 | 2022-08-23 | 北京数慧时空信息技术有限公司 | Remote sensing image recommendation method and system based on user image |
CN116010588A (en) * | 2023-03-28 | 2023-04-25 | 北京创新乐知网络技术有限公司 | Real-time and offline combined document recommendation method, device, equipment and medium |
CN117522440A (en) * | 2023-10-31 | 2024-02-06 | 深圳市灵智数字科技有限公司 | Marketing strategy generation method and device and electronic equipment |
-
2018
- 2018-07-04 CN CN201810725366.3A patent/CN109062994A/en not_active Withdrawn
Cited By (51)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109766492A (en) * | 2018-12-24 | 2019-05-17 | 中国联合网络通信集团有限公司 | Learn recommended method, device, equipment and readable medium |
CN111368181A (en) * | 2018-12-25 | 2020-07-03 | 阿里巴巴集团控股有限公司 | Information recommendation method, information display method and device |
CN111371938A (en) * | 2018-12-26 | 2020-07-03 | 华为终端有限公司 | Fault detection method and electronic equipment |
CN109710853B (en) * | 2018-12-27 | 2021-05-04 | 深圳铁盒子文化科技发展有限公司 | Artificial intelligence classification matching method and system |
CN109710853A (en) * | 2018-12-27 | 2019-05-03 | 深圳铁盒子文化科技发展有限公司 | A kind of artificial intelligence classified matching method and system |
CN110413872A (en) * | 2019-01-21 | 2019-11-05 | 北京京东尚科信息技术有限公司 | Method and apparatus for showing information |
CN111523010B (en) * | 2019-02-03 | 2023-04-28 | 阿里巴巴集团控股有限公司 | Recommendation method, recommendation device, terminal equipment and computer storage medium |
CN111523010A (en) * | 2019-02-03 | 2020-08-11 | 阿里巴巴集团控股有限公司 | Recommendation method and device, terminal equipment and computer storage medium |
CN109885770A (en) * | 2019-02-20 | 2019-06-14 | 杭州威佩网络科技有限公司 | A kind of information recommendation method, device, electronic equipment and storage medium |
CN110083634B (en) * | 2019-03-19 | 2024-02-06 | 中国平安人寿保险股份有限公司 | Order processing method, device, equipment and storage medium based on data analysis |
CN110083634A (en) * | 2019-03-19 | 2019-08-02 | 中国平安人寿保险股份有限公司 | Order processing method, apparatus, equipment and storage medium based on data analysis |
CN109933729A (en) * | 2019-03-28 | 2019-06-25 | 广州麦迪森在线医疗科技有限公司 | A kind of academic information recommended method of medical treatment based on user preference and system |
CN111984837A (en) * | 2019-05-23 | 2020-11-24 | 浙江口碑网络技术有限公司 | Commodity data processing method, device and equipment |
CN111984837B (en) * | 2019-05-23 | 2021-04-23 | 浙江口碑网络技术有限公司 | Commodity data processing method, device and equipment |
CN112148958A (en) * | 2019-06-26 | 2020-12-29 | 傲基科技股份有限公司 | Method, apparatus, and computer storage medium for information recommendation |
CN110659025A (en) * | 2019-08-26 | 2020-01-07 | 联想(北京)有限公司 | Recommendation method and device |
CN110533515A (en) * | 2019-09-04 | 2019-12-03 | 深圳创新奇智科技有限公司 | A kind of the electric business personalized recommendation method and device of high low latency of handling up |
CN110866180A (en) * | 2019-10-12 | 2020-03-06 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, server and storage medium |
CN110866180B (en) * | 2019-10-12 | 2022-07-29 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, server and storage medium |
CN111191129A (en) * | 2019-12-27 | 2020-05-22 | 广东德诚科教有限公司 | Professional recommendation method and device, computer equipment and storage medium |
CN111199428A (en) * | 2020-01-03 | 2020-05-26 | 江苏苏宁物流有限公司 | Commodity recommendation method and device, storage medium and computer equipment |
CN113159810B (en) * | 2020-01-22 | 2024-06-04 | 百度在线网络技术(北京)有限公司 | Policy evaluation method, device, equipment and storage medium |
CN113159810A (en) * | 2020-01-22 | 2021-07-23 | 百度在线网络技术(北京)有限公司 | Policy evaluation method, device, equipment and storage medium |
CN111400546A (en) * | 2020-03-18 | 2020-07-10 | 腾讯科技(深圳)有限公司 | Video recall method and video recommendation method and device |
CN111476639A (en) * | 2020-04-10 | 2020-07-31 | 深圳市物语智联科技有限公司 | Commodity recommendation strategy determining method and device, computer equipment and storage medium |
CN111651669A (en) * | 2020-05-20 | 2020-09-11 | 拉扎斯网络科技(上海)有限公司 | Information recommendation method and device, electronic equipment and computer-readable storage medium |
CN113761327B (en) * | 2020-06-18 | 2024-06-18 | 北京沃东天骏信息技术有限公司 | Scheduling method, equipment and storage medium of recommendation strategy |
CN113761327A (en) * | 2020-06-18 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Scheduling method, device and storage medium of recommendation strategy |
CN111858915A (en) * | 2020-08-07 | 2020-10-30 | 成都理工大学 | Information recommendation method and system based on label similarity |
CN112541787A (en) * | 2020-12-09 | 2021-03-23 | 北京奇艺世纪科技有限公司 | Advertisement recommendation method, system, storage medium and electronic device |
CN112541787B (en) * | 2020-12-09 | 2023-12-19 | 北京奇艺世纪科技有限公司 | Advertisement recommendation method, advertisement recommendation system, storage medium and electronic equipment |
CN112686568A (en) * | 2021-01-11 | 2021-04-20 | 南方电网数字电网研究院有限公司 | Operation and maintenance strategy generation processing method, device, system, equipment and storage medium |
CN112733034A (en) * | 2021-01-21 | 2021-04-30 | 腾讯科技(深圳)有限公司 | Content recommendation method, device, equipment and storage medium |
CN112733034B (en) * | 2021-01-21 | 2023-08-01 | 腾讯科技(深圳)有限公司 | Content recommendation method, device, equipment and storage medium |
CN112905895A (en) * | 2021-03-29 | 2021-06-04 | 平安国际智慧城市科技股份有限公司 | Similar item recommendation method, device, equipment and medium |
CN112905895B (en) * | 2021-03-29 | 2022-08-26 | 平安国际智慧城市科技股份有限公司 | Similar item recommendation method, device, equipment and medium |
CN113204638A (en) * | 2021-04-23 | 2021-08-03 | 上海明略人工智能(集团)有限公司 | Recommendation method, system, computer and storage medium based on work session unit |
CN113204638B (en) * | 2021-04-23 | 2024-02-23 | 上海明略人工智能(集团)有限公司 | Recommendation method, system, computer and storage medium based on working session unit |
CN113298277A (en) * | 2021-04-25 | 2021-08-24 | 上海淇玥信息技术有限公司 | Target-based continuous reservation information pushing method and device and electronic equipment |
CN113435202A (en) * | 2021-06-28 | 2021-09-24 | 平安科技(深圳)有限公司 | Product recommendation method and device based on user portrait, electronic equipment and medium |
CN113434660A (en) * | 2021-06-28 | 2021-09-24 | 平安银行股份有限公司 | Product recommendation method, device, equipment and storage medium based on multi-domain classification |
CN113779421A (en) * | 2021-08-13 | 2021-12-10 | 咪咕数字传媒有限公司 | Association recommendation method, device, equipment and computer readable storage medium |
CN113722523B (en) * | 2021-09-01 | 2023-11-17 | 北京百度网讯科技有限公司 | Object recommendation method and device |
CN113722523A (en) * | 2021-09-01 | 2021-11-30 | 北京百度网讯科技有限公司 | Object recommendation method and device |
CN114429371B (en) * | 2022-04-06 | 2022-06-28 | 新石器慧通(北京)科技有限公司 | Unmanned vehicle-based commodity marketing method and device, electronic equipment and storage medium |
CN114429371A (en) * | 2022-04-06 | 2022-05-03 | 新石器慧通(北京)科技有限公司 | Unmanned vehicle-based commodity marketing method and device, electronic equipment and storage medium |
CN114936325B (en) * | 2022-07-20 | 2022-10-11 | 北京数慧时空信息技术有限公司 | Remote sensing image recommendation method and system based on user image |
CN114936325A (en) * | 2022-07-20 | 2022-08-23 | 北京数慧时空信息技术有限公司 | Remote sensing image recommendation method and system based on user image |
CN116010588A (en) * | 2023-03-28 | 2023-04-25 | 北京创新乐知网络技术有限公司 | Real-time and offline combined document recommendation method, device, equipment and medium |
CN116010588B (en) * | 2023-03-28 | 2023-08-18 | 北京创新乐知网络技术有限公司 | Real-time and offline combined document recommendation method, device, equipment and medium |
CN117522440A (en) * | 2023-10-31 | 2024-02-06 | 深圳市灵智数字科技有限公司 | Marketing strategy generation method and device and electronic equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109062994A (en) | Recommended method, device, computer equipment and storage medium | |
CN110532451A (en) | Search method and device for policy text, storage medium, electronic device | |
TWI557664B (en) | Product information publishing method and device | |
CN103400286B (en) | A kind of commending system and method carrying out article characteristics mark based on user behavior | |
CA2897886C (en) | Methods and apparatus for identifying concepts corresponding to input information | |
CN104268292B (en) | The label Word library updating method of portrait system | |
US9934293B2 (en) | Generating search results | |
CN110162701A (en) | Content delivery method, device, computer equipment and storage medium | |
CN108897734A (en) | User's portrait generation method, device, computer equipment and storage medium | |
CN109993583B (en) | Information pushing method and device, storage medium and electronic device | |
WO2010014082A1 (en) | Method and apparatus for relating datasets by using semantic vectors and keyword analyses | |
CN111161021B (en) | Quick secondary sorting method for recommended commodities based on real-time characteristics | |
WO2017121272A1 (en) | Method and device for processing user behavior data | |
Shahid et al. | Insights into relevant knowledge extraction techniques: a comprehensive review | |
CN110765348B (en) | Hot word recommendation method and device, electronic equipment and storage medium | |
Baishya et al. | SAFER: sentiment analysis-based fake review detection in e-commerce using deep learning | |
Sharma et al. | Intelligent data analysis using optimized support vector machine based data mining approach for tourism industry | |
CN112989118B (en) | Video recall method and device | |
CN110232071A (en) | Search method, device and storage medium, the electronic device of drug data | |
CN110851708B (en) | Negative sample extraction method, device, computer equipment and storage medium | |
Shoja et al. | Tags-aware recommender systems: a systematic review | |
Meng et al. | Variational Bayesian representation learning for grocery recommendation | |
Ozsoy | Utilizing fasttext for venue recommendation | |
Pongnumkul et al. | Random walk-based recommendation with restart using social information and bayesian transition matrices | |
Rajabi Kouchi et al. | A Movie Recommender System Based on User Profile and Artificial Bee Colony Optimization |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20181221 |
|
WW01 | Invention patent application withdrawn after publication |