CN103377242A - User behavior analysis method, user behavior analytical prediction method and television program push system - Google Patents
User behavior analysis method, user behavior analytical prediction method and television program push system Download PDFInfo
- Publication number
- CN103377242A CN103377242A CN2012101274423A CN201210127442A CN103377242A CN 103377242 A CN103377242 A CN 103377242A CN 2012101274423 A CN2012101274423 A CN 2012101274423A CN 201210127442 A CN201210127442 A CN 201210127442A CN 103377242 A CN103377242 A CN 103377242A
- Authority
- CN
- China
- Prior art keywords
- user
- user behavior
- data
- behavior
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses a user behavior analysis method, a user behavior analytical prediction method and a television program push system. The user behavior analysis method comprises firstly analyzing data of historical user behavior, extracting structured data of user behavior patterns and storing the structured data in a database of user behavior patterns; secondly, leading the structured data of the user behavior patterns to be subjected to a first clustering according to behavior type features and generating a user data set of similar clustering by types; thirdly, leading the user data of similar clustering by types to be subjected to a second clustering according to change structure features and generating a cluster data set of users with similar behavior changes; finally, outputting cluster result data of the users. Due to the fact that time-order characters of the changes of user behavior are taken into consideration in the second clustering, so that the cluster data set of the users contains change information of the user behavior which pure statistical data does not contain, and description of users can be more complete; the cluster result set obtained finally can be conveniently applied to collaborative analysis among users and to promotion field of television programs for pushing potential interested programs for users.
Description
Technical field
The present invention relates to the data mining technology field, particularly a kind of user behavior analysis method, analyzing and predicting method and TV programme supplying system.
Background technology
At present, most of algorithm all usefulness be that statistics is carried out preliminary data and processed, such data process user behavioral data can be lost behavior sequential and characteristics periodically, exactly because losing of these data characteristics can cause the accuracy of user profile undesirable.And same user's behavior has too many unpredictability, analyze iff the statistics for the user, then be difficult to the behavior of next time of complete predictive user, for example: be all Monday user's statistical law and can remove to see the category-A program, owing to the reason of the unknown removes to have seen the category-B program, but you can't find to be hidden in the user according to the description of statistics and watch behavior transition information in the history, thus be to remove to recommend the category-B program, thus the consumer goods and the service of hommization can't be provided for the user.
Therefore, prior art is still waiting to improve and improve.
Summary of the invention
The object of the present invention is to provide a kind of user behavior analysis method, analyzing and predicting method and TV programme supplying system, inaccurate to user behavior analysis in the prior art to solve, can lose behavior sequential and characteristics periodically, the relatively poor problem of accuracy that causes user behavior to describe.Described user behavior analysis method is based on the data analysing method that changes structure, and described analyzing and predicting method is based on described user behavior analysis method, and described TV programme supplying system is based on described analyzing and predicting method.
In order to achieve the above object, the present invention has taked following technical scheme:
A kind of user behavior analysis method wherein, may further comprise the steps:
ST1, according to historical user behavior data, extract the user behavior pattern structured data, and be stored in the user behavior pattern database;
ST2, the user behavior pattern structured data that will be stored in the user behavior pattern database carry out the cluster first time by the behavior type feature, and generate similar by type cluster user data set;
ST3, described by type similar cluster user data is carried out the second time carry out cluster by changing architectural feature, and generate the user's bunch collection data set with similar behavior transition;
ST4, output user clustering bunch assembly fruit data.
Described user behavior analysis method, wherein, described step ST1 analyzes historical user behavior data, extract the user behavior pattern structured data, and be stored in the user behavior pattern database, the user behavior pattern structured data be D1=(U, P) wherein, U={ user 1, the set of user 2...... user n} representative of consumer, P=P (c, s, n, m, k), wherein, c is behavior type, s is the behavior zero-time, n is behavior interval time, and m is that cycle times appears in behavior, and k is time of the act length.
Described user behavior analysis method, wherein, described step ST2, the user behavior pattern structured data that is stored in the user behavior pattern database is carried out the cluster first time by the behavior type feature, and generate similar by type cluster user data and concentrate, describedly described by type user behavior data carried out first time cluster specifically be:
The user behavior type similarity expression formula of definition cluster is
Wherein, C
i, C
jI, a j user's behavior pattern class set among the expression U;
To structured data D1=(U, P), the behavior type c of P (c, s, n, m, k) carries out cluster analysis by described user behavior type similarity expression formula.
Described user behavior analysis method wherein, at described step ST3, is carried out the second time to described by type similar cluster user data and is carried out cluster by changing architectural feature, and generates the user's bunch collection data centralization with similar behavior transition, specifically is:
Define described by type user behavior data and integrate as D2=(U, S), wherein U={ user 1, user 2...... user n}; S={ (s
Ij, s
Ij+ n
Ij* m
Ij), s
IjBe pattern starting point, n
IjBe pattern cycle, m
IjBe periodicity;
Defining then, the variation characteristic similarity expression formula of cluster is
Wherein, w
IjValue be that 1 interval scale user i is identical with j user behavior pattern type and order is identical, otherwise be 0; | (C
i∪ C
j) | be the total number of types of elements of user i and user j existence;
To structured data D2=(U, S), S={ (s
Ij, s
Ij+ n
Ij* m
Ij), carry out cluster analysis by described variation characteristic similarity expression formula.
Described user behavior analysis method, wherein, described cluster analysis is the cluster analysis based on MST.Described cluster analysis based on MST specifically may further comprise the steps:
STA, use the similarity expression formula, calculate the weights between having a few;
STB, use MST cluster construction algorithm generate minimum spanning tree;
STC, set tolerable difference weights, large with or the limit that equals its value be defined as boundary edge;
STD, interrupt all boundary edge; Determine the final cluster numbers that generates according to interrupting quantity.
Described user behavior analysis method, wherein, the minimum spanning tree construction algorithm based on the MST cluster among the described step STB is prim algorithm or kruskal algorithm.
A kind of user behavior analysis Forecasting Methodology wherein, said method comprising the steps of:
STA1 also sets up the user behavior analysis prognoses system, and described user behavior analysis prognoses system stores by changing architectural feature user behavior data collection;
STB1, collection active user behavioral data;
STC1, described user behavior data is inputted described user behavior analysis prognoses system, described user behavior analysis prognoses system generates the data clusters bunch collection that carries out cluster by the variation characteristic user behavior, and passes through the possibility of other user in predicting user behavior transition in the cluster result.
A kind of TV programme supplying system, wherein, described system comprises:
Intelligent television terminal is used for broadcast program and gathers instant user behavior data;
User behavior cluster analysis engine, be used for according to described instant user behavior data comprise by type, by twice cluster analysis that changes architectural feature, the user that will have identical historical behavior transition is divided into a class and forms bunch collection, as later stage recommending data collection;
The program push cloud server is used for pushing the potential interest programs of user according to the cluster result collection.
Described TV programme supplying system, wherein,
Described Intelligent television terminal comprises: the instant behavioral data collection of user and transmitting device, be used for setting up network connection with described user behavior cluster analysis engine, and gather the instant behavioral data with the transmission user;
Described user behavior cluster analysis engine comprises: data sink, be used for setting up network connection with described Intelligent television terminal, and receive the instant behavioral data with the described transmission user of storage; The data clusters analytical equipment is used for according to described user behavior analysis method the instant behavioral data of described user being carried out cluster analysis, generates by bunch collection that changes architectural feature user behavior cluster;
Described program push cloud server comprises: the potential interest programs excavating gear of user, described excavating gear and described user behavior cluster analysis engine are set up network connection, and adopt described user behavior analysis Forecasting Methodology, carry out mining analysis according to described by changing the potential program of watching of architectural feature user behavior data set pair user, and output Result data set; The program push device is used for setting up network connection with described Intelligent television terminal, and pushes the potential interest programs of user.
Described TV programme supplying system, wherein,
Described program push device through collaborative filtering potential interest programs is sorted and then the concrete steps that push as follows:
STA2, list potential interest programs;
STB2, according to program recommendation degree potential interest programs is sorted;
STC2, the potential interest programs after will sorting are recommended the user.
Described TV programme supplying system, wherein,
The computing method of program recommendation degree P are as follows among the described step STB2:
Wherein, the span of P is between 0 to 1;
J represents number of users, and its span is between 1 to n;
I
jExpression user j watches possibility, I when watching
jBe 1, otherwise be 0;
Sim
IjExpression targeted customer i and compare similarity between the object j
Beneficial effect:
User behavior analysis method provided by the invention, analyzing and predicting method and TV programme supplying system, by in changing structure, containing the user behavior transition information that does not have in the simple statistics, so that more complete to user's description meeting, and can carry out very easily Cooperative Analysis (being collaborative filtering) between the user, the variation by approximate user comes may changing of predictive user.
Description of drawings
Fig. 1 is the process flow diagram of user behavior analysis embodiment of the method for the present invention.
Fig. 2 is the structured flowchart of the data structure of the user behavior pattern among Fig. 1.
Fig. 3 is the vectorial difference matrix diagram that a specific embodiment forms among Fig. 2.
The non-directed graph that Fig. 4 (a) forms for Fig. 3.
Fig. 4 (b) is the minimum spanning tree of Fig. 4 (a).
Fig. 4 (c) is 4 (b) " tolerable difference " weights greater than 0.3 cluster result figure.
And Fig. 4 (d) is 4 (b) " tolerable difference " weights greater than 0.27 cluster result figure.
Fig. 5 is the process flow diagram of user behavior analysis Forecasting Methodology of the present invention.
Fig. 6 is the structured flowchart of intelligent television program push of the present invention system.
Embodiment
For making purpose of the present invention, technical scheme and effect clearer, clear and definite, the present invention is described in more detail referring to the accompanying drawing examples.
See also Fig. 1, it is the process flow diagram of user behavior analysis embodiment of the method for the present invention, as shown in the figure, said method comprising the steps of:
S1, according to historical user behavior data, extract the user behavior pattern structured data, and be stored in the user behavior pattern database;
S2, the user behavior pattern structured data that will be stored in the user behavior pattern database carry out the cluster first time by the behavior type feature, and generate similar by type cluster user data set;
S3, described by type similar cluster user data is carried out the second time by changing the architectural feature cluster, and generate the user's bunch collection with similar behavior transition;
S4, output user clustering bunch assembly fruit data.
The below is elaborated for above-mentioned steps respectively:
We know, can from historical user behavior, infer and user behavior pattern, therefore, in first step S1, by historical user behavior data is analyzed, extract the behavioral pattern data structure of user behavior pattern corresponding to user behavior data, the described user behavior data of record in behavior mode data structured data table, and be stored in the database.For example: the passing program viewing of recording user historical (being the analysis of history user behavior data), concentratedly excavate out main user's watching mode, behavior pattern structured data table is watched in foundation, watch in the behavior pattern structured data table the described user of record to watch behavioral data described, select the key point data thereby identify for the later stage user behavior pattern.In first embodiment of the invention, the user behavior pattern structured data be D1=(U, P) wherein, U={ user 1, the set of user 2...... user n} representative of consumer, P=P (c, s, n, m, k), wherein, c is behavior type, and s is the behavior zero-time, n is behavior interval time, m is that cycle times appears in behavior, and k is time of the act length, and its structure as shown in Figure 2.This data structure records start time, can get access to like this starting point of each synchronism behavior, thereby can analyze the temporal aspect that obtains each behavior.Obtain user's characteristic and the behavior end point same period by n and m.And from the angle of statistics, only be concerned about in the data occurrence number what, namely several of the ratio maximum like this is sequential, cycle and the behavior variation characteristic that does not have behavior.Be not only the program that the user often sees and we are concerned about, more pay close attention to the periodicity of user behavior, with the modificability of behavior.The very large probability that has in same user class has identical behavior modificability like this.Certainly, we also can make up with other mode the behavior pattern structured data, as long as described behavior pattern structured data comprises behavior type.
It should be noted that here described structural data (being structured data) refer to that the user behavior that is drawn by P (c, s, n, m, k) structure user data analysis changes structure, the data set that obtains is (a
1, a
2..., a
n), its definition is identical with linear list, has order idol relation between n the element.Wherein a is the behavior classification, between the data that we obtain like this sequential precedence relationship is arranged, and has wherein just comprised the user behavior variation characteristic.Certainly we are available linear memory mode when realizing, also can use the chain type storage mode.
In second step S2, the user behavior pattern structured data that is stored in the user behavior pattern database is carried out the cluster first time by the behavior type feature, and generate similar by type cluster user data set.Cluster is with the process of Data classification to different classes, makes the object in the same class that larger similarity be arranged, and inhomogeneous individuality has larger otherness.In first embodiment of the invention, we will be stored in the behavior type (being c) that comprises in the user behavior pattern structured data in the user behavior pattern database and carry out cluster analysis.Its concrete setting can for: be used for the data source D1=(U, P) of cluster in the row user behavior pattern database wherein, U={ user 1, the set of user 2...... user n} representative of consumer; Then behavior type carries out the similarity expression formula of cluster and is
Wherein, C
i, C
jI, a j user's behavior pattern class set among the expression U.In a word, for the first time cluster analysis is to structured data D1=(U, P), and the behavior type c of P (c, s, n, m, k) carries out cluster analysis by described user behavior type similarity expression formula, thereby generates similar by type cluster user data set.
Then, similar by type cluster user data is carried out the second time by variation architectural feature cluster, and generate user bunch collection data set, i.e. the third step S3 with similar behavior transition.In first embodiment of the invention, it specifically is: define described by type user behavior data and integrate as D2=(U, S), wherein U={ user 1, user 2...... user n}; S={ (s
Ij, s
Ij+ n
Ij* m
Ij), s
IjBe i user's j behavior pattern starting point, n
IjI user's j behavior is pattern cycle, m
IjI user's j behavior is periodicity; Defining then, the variation characteristic similarity expression formula of cluster is
Wherein, w
IjValue be that 1 interval scale user i is identical with j user behavior pattern type and order is identical, otherwise be 0; | (C
i∪ C
j) | be the total number of types of elements of user i and user j existence.For example for (A-＞B-＞C-＞D) is with (sim (i, j)=3/4 of A-＞B-＞D) has 3 because add up to 4 the similar node of node.In a word, described second time cluster namely to structured data D2=(U, S), S={ (s
Ij, s
Ij+ n
Ij* m
Ij), carry out cluster analysis by described variation characteristic similarity expression formula.
At last, described step is S4, output user clustering bunch assembly fruit data.Thereby the user behavior transition information that does not have in the simple statistics that is containing in the data that can pass through to export, so that more complete to user's description meeting, and can carry out very easily Cooperative Analysis between the user, the variation by approximate user comes may changing of predictive user.
Further, we can adopt minimum spanning tree clustering method (MST) in the described first time, for the second time cluster analysis.It specifically can adopt following steps to carry out:
A, use the similarity expression formula, calculate the weights between having a few;
B, use MST cluster construction algorithm generate minimum spanning tree;
C, set tolerable difference weights, large with or the limit that equals its value be defined as boundary edge;
D, interrupt all boundary edge; Determine the final cluster numbers that generates according to interrupting quantity.
Wherein, among the described step B based on minimum spanning tree construction algorithm Ke Yi Wei Pu Limu (Prim) algorithm (time complexity is O (N2)) of MST cluster, Kruskal (Kruskal) algorithm (time complexity is that O (eloge) e is figure limit number) etc.
The below describes above-mentioned clustering algorithm based on MST with an object lesson.The below has provided a difference matrix about 7 points, and as shown in Figure 3: 7 points represent respectively 7 users,
In described diversity factor matrix, the digitized representation of the capable j of i row be diversity factor between user i and the user j.As: numeral second user of 0.27 expression of the 2nd row the 3rd row and the diversity factor of third party are 0.27.
In order to say something this example, the quantity on limit is limited to even number (this can't affect net result), Fig. 4 (a) non-directed graph that vectorial difference rectangular becomes of serving as reasons, X1 represents respectively that to X7 first is to the 7th user among the figure.Fig. 4 (b) is the minimum spanning tree of Fig. 4 (a), and Fig. 4 (c) is " tolerable difference " weights greater than 0.3 cluster result figure, and Fig. 4 (d) is greater than 0.27 cluster result figure by " tolerable difference " weights.
Can find out that by cluster result figure data are generated the process of cluster feature data by cluster, data are carried out the user clustering bunch collection that twice MST cluster scheme just can obtain to have identical behavior transition.The advantage that solves clustering problem with non-directed graph is that it is more directly perceived to think over a problem from space angle, and intelligibility is stronger; To new user data update the time, only need to recomputate new user data on the other hand, need not to calculate former user data, speed can be very fast when carrying out so new user's classification.Certainly use this algorithm also can self limitation when calculated difference is spent, data volume be in very little, and the variation meeting of a point has considerable influence to it.
Here illustrate, (action, comedy, love ...) be the fundamental type collection of existing video frequency program.In analysis, user's the program data collection of watching has comprised the behavior of these types, and then user's behavior just is divided into this set of types.Such as: the film that the party A-subscriber saw within a period of time only has { comedy, love } this user's set of types is exactly: { comedy, love }, the film that the party B-subscriber saw within a period of time only has { action, terrified } this user's set of types is exactly: { moves, terrified }, in like manner C user's { comedy, action, love }, D user's { animation, science fiction }, ..., party A-subscriber and C user belong to a fundamental type collection exactly here, namely describedly carry out the cluster first time by the behavior type feature that comprises in the behavior pattern, can arrive with among the cluster collection user A and C are poly-because they two be that similarity is higher.
Further, to having the user of same basic type collection, be a class from watching sequential to calculate its similarity user that similarity is higher poly-again, historical behavior transition degree similarity is higher between the user in the same class, a user's next time behavior just is likely other user behavior next time so, for example user one is in same class with user two, both historical serieses of watching are: user one { action, love, comedy }, user two { action, love } because both are in same class, so the behavior of user one comedy is likely user two behavior next time, certainly this is a simple example, and the contact between the record of might this user watching is very complicated.
Such as, certain user likes Zhao Benshan, can analyze so one and describe collection, what this described set representations is that the unit that this new description possesses have those, defines one such as us and new is described as Zhao Jiajun={ Zhao Benshan, little Shenyang, song-and-dance duet }, the meaning is that this user likes this type of army of Zhao family, and the key element of this type is: Zhao Benshan, little Shenyang, song-and-dance duet.
This method has been quoted the similitude sum and has been described the degree that the user watches program, and ((the similitude sum is 3 A-＞B-＞D), wherein the ABCD representative of consumer type of watching for A-＞B-＞C-＞D) and user2 such as user1.If (B-＞A-＞D) requirement is not behind A to user2 because B does not satisfy sequence, so the similitude sum is 2.This method is carried out further cluster according to this similitude sum to data set, namely the user behavior data cluster second time of carrying out by type.
In addition, the present invention also provides a kind of user behavior analysis Forecasting Methodology, as shown in Figure 5, said method comprising the steps of:
A1, set up the user behavior analysis prognoses system, described user behavior analysis prognoses system stores by changing architectural feature user behavior data collection; Described by changing architectural feature user behavior data collection, adopt the user behavior analysis method generation of stating in this way, just repeated no more here.
B1, collection active user behavioral data;
C1, described user behavior data is inputted described user behavior analysis prognoses system, described user behavior analysis prognoses system generates the data clusters bunch collection that carries out cluster by the variation characteristic user behavior, and passes through the possibility of other user in predicting user behavior transition in the cluster result.
In described user behavior analysis prognoses system, by storing by changing architectural feature user behavior data collection, instant user behavior data to input is analyzed, predict by approximate user's variation input instant user behavior data the user may change the possibility of i.e. predictive user behavior transition.
When concrete the application, we can be applied to TV programme disseminate technology field with above-mentioned user behavior analysis Forecasting Methodology, and the consumer goods and the service of hommization are provided for the user.
Please continue to consult Fig. 6, it is the structured flowchart of intelligent television program push of the present invention system, and as shown in the figure, described system comprises: Intelligent television terminal 100, user behavior cluster analysis engine 200 and program push cloud server 300.Described Intelligent television terminal 100, user behavior cluster analysis engine 200 are connected with the program push cloud server successively and are connected.
Wherein, described Intelligent television terminal 100 is used for broadcast program and gathers instant user behavior data; In the present embodiment, described Intelligent television terminal 100 comprises the instant behavioral data collection of user and transmitting device, is used for setting up network connection with described user behavior cluster analysis engine, gathers the instant behavioral data with the transmission user.For example: can or watch statistic device to be used as the instant behavioral data collection of user by image collecting device, then set up network connection by coupled transmitting device (wireless, wired etc.) and described user behavior cluster analysis engine, gather and send the instant behavioral data of user.
Described user behavior cluster analysis engine 200 be used for according to described instant user behavior data comprise by type, by twice cluster analysis that changes architectural feature, the user that will have identical historical behavior transition is divided into a class and forms bunch collection, as later stage recommending data collection, its concrete grammar can be referring to above-mentioned user behavior analysis method.In the present embodiment, described user behavior cluster analysis engine 200 comprises: data sink and data clusters analytical equipment, described data sink is used for setting up network connection with described Intelligent television terminal, receives the instant behavioral data with the described transmission user of storage; Described data clusters analytical equipment is used for according to the user behavior analysis method the instant behavioral data of described user being carried out cluster analysis, and the user is classified, and having the user of similar transition behavior poly-is a class.
Described program push cloud server 300 is used for pushing the potential interest programs of user according to the cluster result collection.In the present embodiment, described program push cloud server 300 comprises the potential interest programs excavating gear of user and program push device, the potential interest programs excavating gear of described user and described user behavior cluster analysis engine are set up network connection, and adopt above-mentioned user behavior analysis Forecasting Methodology, go out the user clustering collection according to mining analysis, and output Result data set; The program push device is used for setting up network connection with described Intelligent television terminal, directly pushes the potential interest programs of user or passes through collaborative filtering (also claiming Cooperative Analysis) again and potential interest programs is sorted and then push.
Further, described program push device through collaborative filtering potential interest programs is sorted and then the concrete steps that push as follows:
A2, list potential interest programs;
B2, according to program recommendation degree potential interest programs is sorted;
C2, the potential interest programs after will sorting are recommended the user.
Wherein, the computing method of program recommendation degree P are as follows among the described step B:
The span of described P is between 0 to 1; J represents number of users, and its span is between 1 to n; I
jWhether expression user j watched I when watching
jBe 1, otherwise be 0; Sim
IjExpression targeted customer i and compare similarity between the object j.This recommend method carries out in the result of above-mentioned secondary cluster, thereby (algorithm of the recommendation degree of prior art has multiple to have greatly reduced the operand of algorithm, it is a lot of that its algorithm has been compared the complexity height with the application), and guaranteed the programs recommended accuracy to the user, for the user has brought better service.
In sum, user behavior analysis method provided by the invention, analyzing and predicting method and TV programme supplying system at first, are analyzed historical user behavior data, extract the user behavior pattern structured data, and are stored in the user behavior pattern database; Then, the user behavior pattern structured data that is stored in the user behavior pattern database is carried out the cluster first time by the behavior type feature, and generate similar by type cluster user data set; Again described by type similar cluster user data is carried out the second time and carry out cluster by changing architectural feature, and generate the user's bunch collection data set with similar behavior transition; Export at last the user clustering result data.Because cluster has been considered the transition temporal aspect of user behavior for the second time, wherein containing the user behavior transition information that does not have in the simple statistics, so that more complete to user's description meeting, and the cluster result collection that obtains at last can carry out the Cooperative Analysis between the user very easily, variation by approximate user comes may changing of predictive user, and in TV programme popularization field, for the user pushes potential interest programs.
Need to prove that this user behavior analysis method is based on the data analysing method that changes structure, this analysis Forecasting Methodology is based on described user behavior analysis method, and this TV programme supplying system is based on described analyzing and predicting method.
Be understandable that, for those of ordinary skills, can be equal to replacement or change according to technical scheme of the present invention and inventive concept thereof, and all these changes or replacement all should belong to the protection domain of the appended claim of the present invention.
Claims (11)
1. a user behavior analysis method is characterized in that, may further comprise the steps:
ST1, according to historical user behavior data, extract the user behavior pattern structured data, and be stored in the user behavior pattern database;
ST2, the user behavior pattern structured data that will be stored in the user behavior pattern database carry out the cluster first time by the behavior type feature, and generate similar by type cluster user data set;
ST3, described by type similar cluster user data is carried out the second time carry out cluster by changing architectural feature, and generate the user's bunch collection data set with similar behavior transition;
ST4, output user clustering bunch assembly fruit data.
2. user behavior analysis method according to claim 1 is characterized in that, among the described step ST1, the user behavior pattern structured data be D1=(U, P) wherein, the set of U={ user 1, user 2...... user n} representative of consumer, P=P (c, s, n, m, k), wherein, c is behavior type, and s is the behavior zero-time, and n is behavior interval time, m is that cycle times appears in behavior, and k is time of the act length.
3. user behavior analysis method according to claim 2 is characterized in that, among the described step ST2, describedly described by type user behavior data is carried out first time cluster specifically is:
The user behavior type similarity expression formula of definition cluster is
Wherein, C
i, C
jI, a j user's behavior pattern class set among the expression U;
To structured data D1=(U, P), the behavior type c of P (c, s, n, m, k) carries out cluster analysis by described user behavior type similarity expression formula.
4. user behavior analysis method according to claim 3 is characterized in that, described step ST3 specifically is:
Define described by type user behavior data and integrate as D2=(U, S), wherein U={ user 1, user 2...... user n}; S={ (s
Ij, s
Ij+ n
Ij* m
Ij), s
IjBe i user's j behavior pattern starting point, n
IjI user's j behavior is pattern cycle, m
IjI user's j behavior is periodicity;
Defining then, the variation characteristic similarity expression formula of cluster is
Wherein, w
IjValue be that 1 interval scale user i is identical with j user behavior pattern type and order is identical, otherwise be 0; | (C
i∪ C
j) | be the total number of types of elements of user i and user j existence;
To structured data D2=(U, S), S={ (s
Ij, s
Ij+ n
Ij* m
Ij), carry out cluster analysis by described variation characteristic similarity expression formula.
5. user behavior analysis method according to claim 4 is characterized in that, described cluster analysis specifically may further comprise the steps based on the cluster analysis of MST:
STA, use the similarity expression formula, calculate the weights between having a few;
STB, use MST cluster construction algorithm generate minimum spanning tree;
STC, set tolerable difference weights, large with or the limit that equals its value be defined as boundary edge;
STD, interrupt all boundary edge; Determine the final cluster numbers that generates according to interrupting quantity.
6. user behavior analysis method according to claim 5 is characterized in that, the minimum spanning tree construction algorithm based on the MST cluster among the described step STB is prim algorithm or kruskal algorithm.
7. a user behavior analysis Forecasting Methodology is characterized in that, said method comprising the steps of:
STA1, set up the user behavior analysis prognoses system, described user behavior analysis prognoses system, store by changing architectural feature user behavior data collection, described by changing architectural feature user behavior data collection, adopt such as each described user behavior analysis method of claim 1 to 6 to generate;
STB1, collection active user behavioral data;
STC1, described user behavior data is inputted described user behavior analysis prognoses system, described user behavior analysis prognoses system generates the data clusters bunch collection that carries out cluster by the variation characteristic user behavior, and passes through the possibility of other user in predicting user behavior transition in the cluster result.
8. a TV programme supplying system is characterized in that, described system comprises:
Intelligent television terminal is used for broadcast program and gathers instant user behavior data;
User behavior cluster analysis engine, be used for according to described instant user behavior data comprise by type, by twice cluster analysis that changes architectural feature, the user that will have identical historical behavior transition is divided into a class and forms bunch collection, as later stage recommending data collection;
The program push cloud server is used for pushing the potential interest programs of user according to the cluster result collection.
9. TV programme supplying system as claimed in claim 8 is characterized in that,
Described Intelligent television terminal comprises: the instant behavioral data collection of user and transmitting device, be used for setting up network connection with described user behavior cluster analysis engine, and gather the instant behavioral data with the transmission user;
Described user behavior cluster analysis engine comprises: data sink, be used for setting up network connection with described Intelligent television terminal, and receive the instant behavioral data with the described transmission user of storage; The data clusters analytical equipment, be used for according to such as each described user behavior analysis method of claim 1 to 6 the instant behavioral data of described user being carried out cluster analysis, the user that will have identical historical behavior transition is divided into a class and forms bunch collection, as later stage recommending data collection;
Described program push cloud server comprises: the potential interest programs excavating gear of user, described excavating gear and described user behavior cluster analysis engine are set up network connection, and adopt user behavior analysis Forecasting Methodology as claimed in claim 7, go out the user clustering collection according to mining analysis, and output Result data set; The program push device is used for setting up network connection with described Intelligent television terminal, directly pushes the potential interest programs of user or through collaborative filtering potential interest programs is sorted and then push.
10. TV programme supplying system as claimed in claim 9 is characterized in that,
Described program push device through collaborative filtering potential interest programs is sorted and then the concrete steps that push as follows:
STA2, list potential interest programs;
STB2, according to program recommendation degree potential interest programs is sorted;
STC2, the potential interest programs after will sorting are recommended the user.
11. TV programme supplying system as claimed in claim 10 is characterized in that,
The computing method of program recommendation degree are as follows among the described step STB2:
Wherein, the span of program recommendation degree P is between 0 to 1;
J represents number of users, and its span is between 1 to n;
I
jWhether expression user j watched I when watching
jBe 1, otherwise be 0;
Sim
IjExpression targeted customer i and compare similarity between the object j.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210127442.3A CN103377242B (en) | 2012-04-25 | 2012-04-25 | User behavior analysis method, analyzing and predicting method and television program push system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210127442.3A CN103377242B (en) | 2012-04-25 | 2012-04-25 | User behavior analysis method, analyzing and predicting method and television program push system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103377242A true CN103377242A (en) | 2013-10-30 |
| CN103377242B CN103377242B (en) | 2016-06-22 |
Family
ID=49462368
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210127442.3A Expired - Fee Related CN103377242B (en) | 2012-04-25 | 2012-04-25 | User behavior analysis method, analyzing and predicting method and television program push system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103377242B (en) |
Cited By (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678647A (en) * | 2013-12-20 | 2014-03-26 | Tcl集团股份有限公司 | Method and system for recommending information |
| CN104320704A (en) * | 2014-11-13 | 2015-01-28 | 北京国双科技有限公司 | Method and device for detecting network television installation states |
| CN104899331A (en) * | 2015-06-24 | 2015-09-09 | Tcl集团股份有限公司 | Television used behavior data clustering method and device and Spark big data platform |
| CN105142025A (en) * | 2015-07-16 | 2015-12-09 | Tcl集团股份有限公司 | Information push method and system based on intelligent television terminal |
| CN105512914A (en) * | 2015-12-09 | 2016-04-20 | 联想(北京)有限公司 | Information processing method and electronic device |
| CN105740268A (en) * | 2014-12-10 | 2016-07-06 | 阿里巴巴集团控股有限公司 | Information pushing method and apparatus |
| CN105791349A (en) * | 2014-12-23 | 2016-07-20 | 中兴通讯股份有限公司 | Data analysis method, device and system, terminal and server |
| CN105868318A (en) * | 2016-03-25 | 2016-08-17 | 海信集团有限公司 | Multimedia data type prediction method and device |
| CN105898412A (en) * | 2016-04-26 | 2016-08-24 | 湖北睛彩视讯科技有限公司 | Internet tv cloud platform |
| CN105912708A (en) * | 2016-04-28 | 2016-08-31 | 南京信息工程大学 | Meso-and micro-scale meteorological product integration display platform and recommendation method thereof |
| CN106028126A (en) * | 2016-05-17 | 2016-10-12 | Tcl集团股份有限公司 | Program pushing method and system |
| CN106028071A (en) * | 2016-05-17 | 2016-10-12 | Tcl集团股份有限公司 | Video recommendation method and system |
| CN106294434A (en) * | 2015-05-26 | 2017-01-04 | 联想(北京)有限公司 | A kind of information processing method, it is recommended that method and electronic equipment |
| CN107071578A (en) * | 2017-05-24 | 2017-08-18 | 中国科学技术大学 | IPTV program commending methods |
| CN107451832A (en) * | 2016-05-30 | 2017-12-08 | 北京京东尚科信息技术有限公司 | The method and apparatus of pushed information |
| CN107659855A (en) * | 2017-11-07 | 2018-02-02 | 山东浪潮商用系统有限公司 | A kind of program push system and its program push method |
| CN107864234A (en) * | 2017-12-18 | 2018-03-30 | 广东省电信规划设计院有限公司 | The acquisition methods and device of address renewed treaty time |
| CN107864405A (en) * | 2017-11-14 | 2018-03-30 | 国家新闻出版广电总局广播科学研究院 | A kind of Forecasting Methodology, device and the computer-readable medium of viewing behavior type |
| CN108053262A (en) * | 2017-12-28 | 2018-05-18 | 北京小度信息科技有限公司 | Data push method and device |
| CN108122123A (en) * | 2016-11-29 | 2018-06-05 | 华为技术有限公司 | A kind of method and device for extending potential user |
| CN108924100A (en) * | 2018-06-20 | 2018-11-30 | 广东电网有限责任公司 | A kind of abnormal user recognition methods |
| CN109740559A (en) * | 2019-01-10 | 2019-05-10 | 珠海格力电器股份有限公司 | Personal identification method, apparatus and system |
| CN109934648A (en) * | 2017-12-15 | 2019-06-25 | 中国移动通信集团公司 | A kind of group's Products Show method and device based on Collaborative Filtering algorithm |
| CN111814903A (en) * | 2020-07-21 | 2020-10-23 | 上海数鸣人工智能科技有限公司 | Method for analyzing user sensitivity to marketing activities based on DPI clustering |
| CN111860598A (en) * | 2020-06-18 | 2020-10-30 | 中国地质大学（武汉） | Data analysis method and electronic equipment for identifying sports behaviors and relationships |
| CN112085099A (en) * | 2020-09-09 | 2020-12-15 | 西南大学 | Distributed student clustering integration method and system |
| CN113037840A (en) * | 2021-03-08 | 2021-06-25 | 中国联合网络通信集团有限公司 | Communication data transmission method, communication terminal and communication platform |
| CN113411645A (en) * | 2021-06-16 | 2021-09-17 | 北京百度网讯科技有限公司 | Information recommendation method and device, electronic equipment and medium |
| CN115953166A (en) * | 2022-12-27 | 2023-04-11 | 鑫恒绅企业服务(无锡)有限公司 | Customer information management method and system based on big data intelligent matching |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107894998B (en) * | 2017-10-24 | 2019-04-26 | 迅雷计算机（深圳）有限公司 | Video recommendation method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030101451A1 (en) * | 2001-01-09 | 2003-05-29 | Isaac Bentolila | System, method, and software application for targeted advertising via behavioral model clustering, and preference programming based on behavioral model clusters |
| CN101094335A (en) * | 2006-06-20 | 2007-12-26 | 株式会社日立制作所 | TV program recommender and method thereof |
| CN101702653A (en) * | 2009-10-27 | 2010-05-05 | 中国科学院声学研究所 | Message announcing system based on locating user behavior and method thereof |
| CN101945400A (en) * | 2009-07-10 | 2011-01-12 | 中国移动通信集团公司 | User dynamic behavior analysis method and analysis device |
-
2012
- 2012-04-25 CN CN201210127442.3A patent/CN103377242B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030101451A1 (en) * | 2001-01-09 | 2003-05-29 | Isaac Bentolila | System, method, and software application for targeted advertising via behavioral model clustering, and preference programming based on behavioral model clusters |
| CN101094335A (en) * | 2006-06-20 | 2007-12-26 | 株式会社日立制作所 | TV program recommender and method thereof |
| CN101945400A (en) * | 2009-07-10 | 2011-01-12 | 中国移动通信集团公司 | User dynamic behavior analysis method and analysis device |
| CN101702653A (en) * | 2009-10-27 | 2010-05-05 | 中国科学院声学研究所 | Message announcing system based on locating user behavior and method thereof |
Non-Patent Citations (2)
| Title |
|---|
| CHIN-FENG LAI等: "CPRS: A cloud-based program recommendation system for digital TV platforms", 《FUTURE GENERATION COMPUTER SYSTEMS》 * |
| YI-HUNG WU等: "Enabling Personalized Recommendation on the Web Based on User Interests and Behaviors", 《PROCEEDINGS ELEVENTH INTERNATIONAL WORKSHOP ON RESEARCH ISSUES IN DATA ENGINEERING》 * |
Cited By (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103678647A (en) * | 2013-12-20 | 2014-03-26 | Tcl集团股份有限公司 | Method and system for recommending information |
| CN103678647B (en) * | 2013-12-20 | 2017-12-19 | Tcl集团股份有限公司 | A kind of method and system for realizing information recommendation |
| CN104320704A (en) * | 2014-11-13 | 2015-01-28 | 北京国双科技有限公司 | Method and device for detecting network television installation states |
| CN104320704B (en) * | 2014-11-13 | 2017-08-01 | 北京国双科技有限公司 | The detection method and device of Web TV installation state |
| CN105740268A (en) * | 2014-12-10 | 2016-07-06 | 阿里巴巴集团控股有限公司 | Information pushing method and apparatus |
| CN105740268B (en) * | 2014-12-10 | 2019-04-09 | 阿里巴巴集团控股有限公司 | A kind of information-pushing method and device |
| CN105791349A (en) * | 2014-12-23 | 2016-07-20 | 中兴通讯股份有限公司 | Data analysis method, device and system, terminal and server |
| CN106294434A (en) * | 2015-05-26 | 2017-01-04 | 联想(北京)有限公司 | A kind of information processing method, it is recommended that method and electronic equipment |
| CN104899331A (en) * | 2015-06-24 | 2015-09-09 | Tcl集团股份有限公司 | Television used behavior data clustering method and device and Spark big data platform |
| CN105142025A (en) * | 2015-07-16 | 2015-12-09 | Tcl集团股份有限公司 | Information push method and system based on intelligent television terminal |
| CN105512914A (en) * | 2015-12-09 | 2016-04-20 | 联想(北京)有限公司 | Information processing method and electronic device |
| CN105868318A (en) * | 2016-03-25 | 2016-08-17 | 海信集团有限公司 | Multimedia data type prediction method and device |
| CN105898412A (en) * | 2016-04-26 | 2016-08-24 | 湖北睛彩视讯科技有限公司 | Internet tv cloud platform |
| CN105912708A (en) * | 2016-04-28 | 2016-08-31 | 南京信息工程大学 | Meso-and micro-scale meteorological product integration display platform and recommendation method thereof |
| CN106028071A (en) * | 2016-05-17 | 2016-10-12 | Tcl集团股份有限公司 | Video recommendation method and system |
| CN106028126A (en) * | 2016-05-17 | 2016-10-12 | Tcl集团股份有限公司 | Program pushing method and system |
| CN107451832B (en) * | 2016-05-30 | 2023-09-05 | 北京京东尚科信息技术有限公司 | Method and device for pushing information |
| CN107451832A (en) * | 2016-05-30 | 2017-12-08 | 北京京东尚科信息技术有限公司 | The method and apparatus of pushed information |
| CN108122123A (en) * | 2016-11-29 | 2018-06-05 | 华为技术有限公司 | A kind of method and device for extending potential user |
| CN108122123B (en) * | 2016-11-29 | 2021-08-20 | 华为技术有限公司 | Method and device for expanding potential users |
| CN107071578A (en) * | 2017-05-24 | 2017-08-18 | 中国科学技术大学 | IPTV program commending methods |
| CN107071578B (en) * | 2017-05-24 | 2019-11-22 | 中国科学技术大学 | IPTV program commending method |
| CN107659855A (en) * | 2017-11-07 | 2018-02-02 | 山东浪潮商用系统有限公司 | A kind of program push system and its program push method |
| CN107864405A (en) * | 2017-11-14 | 2018-03-30 | 国家新闻出版广电总局广播科学研究院 | A kind of Forecasting Methodology, device and the computer-readable medium of viewing behavior type |
| CN107864405B (en) * | 2017-11-14 | 2020-12-01 | 国家新闻出版广电总局广播科学研究院 | Viewing behavior type prediction method, device and computer readable medium |
| CN109934648A (en) * | 2017-12-15 | 2019-06-25 | 中国移动通信集团公司 | A kind of group's Products Show method and device based on Collaborative Filtering algorithm |
| CN107864234A (en) * | 2017-12-18 | 2018-03-30 | 广东省电信规划设计院有限公司 | The acquisition methods and device of address renewed treaty time |
| CN107864234B (en) * | 2017-12-18 | 2020-12-11 | 广东省电信规划设计院有限公司 | Method and device for acquiring address continuation time |
| CN108053262A (en) * | 2017-12-28 | 2018-05-18 | 北京小度信息科技有限公司 | Data push method and device |
| CN108924100A (en) * | 2018-06-20 | 2018-11-30 | 广东电网有限责任公司 | A kind of abnormal user recognition methods |
| CN108924100B (en) * | 2018-06-20 | 2020-12-01 | 广东电网有限责任公司 | Abnormal user identification method |
| CN109740559A (en) * | 2019-01-10 | 2019-05-10 | 珠海格力电器股份有限公司 | Personal identification method, apparatus and system |
| CN111860598B (en) * | 2020-06-18 | 2023-02-28 | 中国地质大学（武汉） | Data analysis method and electronic equipment for identifying sports behaviors and relationships |
| CN111860598A (en) * | 2020-06-18 | 2020-10-30 | 中国地质大学（武汉） | Data analysis method and electronic equipment for identifying sports behaviors and relationships |
| CN111814903A (en) * | 2020-07-21 | 2020-10-23 | 上海数鸣人工智能科技有限公司 | Method for analyzing user sensitivity to marketing activities based on DPI clustering |
| CN111814903B (en) * | 2020-07-21 | 2022-04-01 | 上海数鸣人工智能科技有限公司 | Method for analyzing user sensitivity to marketing activities based on DPI clustering |
| CN112085099A (en) * | 2020-09-09 | 2020-12-15 | 西南大学 | Distributed student clustering integration method and system |
| CN112085099B (en) * | 2020-09-09 | 2022-05-17 | 西南大学 | Distributed student clustering integration method and system |
| CN113037840A (en) * | 2021-03-08 | 2021-06-25 | 中国联合网络通信集团有限公司 | Communication data transmission method, communication terminal and communication platform |
| CN113037840B (en) * | 2021-03-08 | 2022-06-10 | 中国联合网络通信集团有限公司 | Communication data transmission method, communication terminal and communication platform |
| CN113411645B (en) * | 2021-06-16 | 2022-12-06 | 北京百度网讯科技有限公司 | Information recommendation method and device, electronic equipment and medium |
| CN113411645A (en) * | 2021-06-16 | 2021-09-17 | 北京百度网讯科技有限公司 | Information recommendation method and device, electronic equipment and medium |
| CN115953166A (en) * | 2022-12-27 | 2023-04-11 | 鑫恒绅企业服务(无锡)有限公司 | Customer information management method and system based on big data intelligent matching |
| CN115953166B (en) * | 2022-12-27 | 2024-04-02 | 鑫恒绅企业服务(无锡)有限公司 | Customer information management method and system based on big data intelligent matching |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103377242B (en) | 2016-06-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103377242A (en) | User behavior analysis method, user behavior analytical prediction method and television program push system | |
| Costa-Montenegro et al. | Which App? A recommender system of applications in markets: Implementation of the service for monitoring users’ interaction | |
| CN103686237B (en) | Recommend the method and system of video resource | |
| US9277264B2 (en) | System, method, and software application for targeted advertising via behavioral model clustering, and preference programming based on behavioral model clusters | |
| CN106028071A (en) | Video recommendation method and system | |
| CN103678647A (en) | Method and system for recommending information | |
| CN102473174A (en) | A method and arrangements for analysing multimedia content | |
| CN109977296B (en) | Information pushing method, device, equipment and storage medium | |
| EP2652909B1 (en) | Method and system for carrying out predictive analysis relating to nodes of a communication network | |
| US10911821B2 (en) | Utilizing combined outputs of a plurality of recurrent neural networks to generate media consumption predictions | |
| CN102486796A (en) | Method and device for acquiring video information | |
| CN103997662A (en) | Program pushing method and system | |
| CN109978575B (en) | Method and device for mining user flow operation scene | |
| CN111259195A (en) | Video recommendation method and device, electronic equipment and readable storage medium | |
| CN102957949A (en) | Device and method for recommending video to user | |
| CN113918577B (en) | Data table identification method and device, electronic equipment and storage medium | |
| CN108182600A (en) | A kind of method and system that extending user is determined according to weighted calculation | |
| Bakar et al. | Application of graph theory and matrix approach as decision analysis tool for smartphone selection | |
| CN108053260A (en) | A kind of method and system that extending user is determined according to statistics interest-degree | |
| CN111353052B (en) | Multimedia object recommendation method and device, electronic equipment and storage medium | |
| CN108038739A (en) | A kind of method and system that extending user is determined according to the statistics degree of association | |
| CN115080824A (en) | Target word mining method and device, electronic equipment and storage medium | |
| CN113792211A (en) | Resource pushing processing method and device, electronic equipment and storage medium | |
| CN113761272A (en) | Data processing method, data processing equipment and computer readable storage medium | |
| Mitsis et al. | Social media analytics in support of documentary production |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160622 |