CN105528659A - Mobile terminal APP usage prediction method combining with time-context based on sequence mode - Google Patents
Mobile terminal APP usage prediction method combining with time-context based on sequence mode Download PDFInfo
- Publication number
- CN105528659A CN105528659A CN201610056153.7A CN201610056153A CN105528659A CN 105528659 A CN105528659 A CN 105528659A CN 201610056153 A CN201610056153 A CN 201610056153A CN 105528659 A CN105528659 A CN 105528659A
- Authority
- CN
- China
- Prior art keywords
- app
- record
- usage behavior
- behavior pattern
- bar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention discloses a mobile terminal APP usage prediction method combining with time-context based on a sequence mode, comprising steps of using a sequence mode to extract the private app usage relevance of the user from a user history app usage sequence, which is the user application usage mode, combining with the fact whether the user is in the mobile phone usage active period, and comprehensively consider the recent behavior mode of the user during the prediction and the time-context information that the user currently use so as to enable the predicted application usage to accord with the real-time demand and preference of the user.
Description
Technical field
The invention belongs to data mining and recommended technology field, be specifically related to a kind of based on sequence pattern binding time contextual mobile terminal APP usage forecastings method.
Background technology
In recent years in order to realize better Consumer's Experience, improve the object such as usage rate of the user and directed marketing promotion, each large platform manufacturer all wished to segment user by large data technique, digging user attribute, drew user's portrait.And facing to the rise of mobile Internet, is emerging in large numbers of mobile large data along with what drive, mobile large data have higher commercial value, according to the APP service condition of mobile object can be objective reflect hobby, the behavioural habits of user.Add the abundant physical location information of mobile object, make the data of user more plentiful, this has also just attracted academia and engineering circles to the concern of the data mining of mobile object.Behavior pattern (BehaviorModel) namely refers to include a personalized habitually basic model, and namely the excavation of behavior pattern excavates fixing behavior pattern from a pile historical data.By can the behavioural habits of clear consumer positioning to the excavation of user behavior pattern under mobile environment, capture user preferences, thus complete personalized customization and Consumer's Experience.Apple and Google be development of user forecasting techniques competitively, as GoogleNowonTap is released in Google plan, understands user how to use application by word and photographic intelligence, and predicts the next behavior of user.
Undertaken applying usage behavior to user by digging user behavior pattern to predict, system can be facilitated to be that the application that user's next one uses prepares running environment in advance, reduce user operation time and misuse rate etc., promote Consumer's Experience.But, seldom the prediction user next one is used to the prediction of APP in existing research.In conventional statistics prediction algorithm, solely use the application that user's frequency of utilization is the highest as prediction, the predicted value of this pattern has certain blindness and stationarity, can not carry out perception to the behavior intention of user.The people such as Yen-SsuChou are by collecting in the quaternary data (user, geographic position, time, service) of most users, build four-layer structure figure, utilization MJMF (MatchJoinsUsingMaxFlows) algorithm carries out the APP service that the next place of target of prediction user can use, but obviously have ignored in this algorithm and all user behavior patterns have been carried out unifiedly to treat, lack the isolation between mobile user data, have ignored its people's use habit etc. of single mobile subscriber.
Summary of the invention
For the above-mentioned technological deficiency existing for prior art, the invention provides a kind of based on sequence pattern binding time contextual mobile terminal APP usage forecastings method, consider the time contextual information of the nearest behavior pattern of user and the current use of user when prediction, thus the application of prediction can be allowed to use the real-time requirement and the preference that more meet user.
A kind of based on sequence pattern binding time contextual mobile terminal APP usage forecastings method, comprise the steps:
(1) obtain APP history and use records series, described APP history uses records series to comprise user's a period of time interior every bar use record for APP in the past;
(2) records series is used to be divided into multiple relevant subsequence whole APP history according to the behavioural characteristic of user;
(3) add up the number of times that often kind of APP usage behavior pattern occurs in all relevant subsequence, and the form that all APP usage behavior patterns are set by trie is carried out structured storage;
(4) obtain APP prediction and use records series, described APP prediction uses records series to comprise user T before current time
stevery bar for APP in large minor time slice uses record, T
stfor the active time window size preset;
(5) by searching the APP usage behavior pattern predicted to APP and use records series relevant from trie tree, and then corresponding the predicting the outcome of generation is extracted according to the number of times that these APP usage behavior patterns occur in all relevant subsequence.
Every bar uses record to include start time, end time and corresponding ID and the title using APP.
The specific implementation process in described step (2), whole APP history use records series being divided into multiple relevant subsequence is as follows:
Arbitrary in records series is used to use record S for APP history
i, newly-built one comprises the relevant subsequence that this uses record, according to following relational expression from use record H
i-1start to judge that every bar uses record whether to record S with use one by one
ithere is correlativity:
Wherein: H
i-1for recording S for use
i-1deterministic process in first bar with use record S
i-1not there is the use record of correlativity, S
iand S
i-1being respectively APP history uses i-th in records series and the i-th-1 to use record, and i is natural number and 1≤i≤n, n is the use record sum in APP history use records series, F (S
i| H
i-1) for using record H
i-1s is recorded with use
icorrelativity, F (S
i| H
i-1)=1 represents to have correlativity, F (S
i| H
i-1)=0 represents not have correlativity, T
in(H
i-1) for using record H
i-1start time, T
out(S
i) for using record S
iend time;
If use record H
i-1s is recorded with use
ithere is correlativity, then by use record H
i-1include this relevant subsequence in and then judge that next uses record, by that analogy until occur that a certain bar uses record to record S with use
inot there is correlativity, then use record to be defined as H this
iand establish by use record S
ito H
ilast bar use this relevant subsequence recording and form;
If use record H
i-1s is recorded with use
ido not have correlativity, then this relevant subsequence is cancelled;
Traveling through APP history according to this uses the every bar in records series to use record, obtains multiple relevant subsequence.
If the APP that customer mobile terminal is installed adds up to m, then the species number of APP usage behavior pattern is
m is relevant subsequence maximum length.
The number of times that in described step (3), statistics often kind of APP usage behavior pattern occurs in all relevant subsequence, specific implementation process is as follows:
APP usage behavior pattern of choosing any one kind of them and a relevant subsequence, get the app [1] in this APP usage behavior pattern, makes the every bar in this relevant subsequence use record to mate with app [1] one by one:
All do not find one to use APP and the app [1] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [1] corresponding to record to mate identical, then use record to be defined as G this
1, and then from G
1start to make every bar use record to mate with app [2] one by one:
All do not find one to use APP and the app [2] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [2] corresponding to record to mate identical, then use record to be defined as G this
2, and then from G
2start to make every bar use record to mate with app [3] one by one; The rest may be inferred, until judge to perform from G
p-1start to make every bar use record to mate with app [p] one by one:
All do not find one to use APP and the app [p] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [p] corresponding to record to mate identical, then judge that this relevant subsequence comprises this APP usage behavior pattern i.e. this APP usage behavior pattern and occurs in this relevant subsequence; Wherein: app [j] is the APP in this APP usage behavior pattern corresponding to the usage behavior of jth position, j is natural number and 1≤j≤p, p is the length of this APP usage behavior pattern, G
p-1for the use record that bar first in the matching process for app [p-1] is identical with app [p-1] corresponding A PP;
Travel through all relevant subsequence according to this, just can count the number of times that this APP usage behavior pattern occurs in all relevant subsequence.
The specific implementation process of described step (5) is as follows:
From trie tree, search length is N+1 and top N usage behavior and APP predicts that using every bar of records series to use records APP usage behavior pattern one to one, and N is that APP predicts that the use record in use records series is total;
If APP prediction use records series be sky, then from APP history use records series extract access times maximum before k APP be supplied to user as predicting the outcome, k be greater than 1 natural number;
If have found the several APP usage behavior pattern meeting above-mentioned condition by search, the number of times then occurred in all relevant subsequence according to these APP usage behavior patterns sorts, and chooses the maximum front k kind APP usage behavior pattern of occurrence number and the APP extracted in this k kind APP usage behavior pattern corresponding to last usage behavior is supplied to user as predicting the outcome;
If do not find any APP usage behavior pattern meeting above-mentioned condition by search, then records series is used to decompose to APP prediction, namely predict to use records series to remove after arbitrary use is recorded from APP and obtain a predictor sequence, traversal removes each bar use record and obtains N number of predictor sequence according to this; Arbitrary predictor sequence search length from trie tree is N and every bar of front N-1 position usage behavior and predictor sequence uses and records APP usage behavior pattern one to one, travels through all predictor sequences according to this; For searching for all APP usage behavior patterns obtained, sorting according to the number of times that these APP usage behavior patterns occur in all relevant subsequence, choosing the maximum front k kind APP usage behavior pattern of occurrence number and the APP extracted in this k kind APP usage behavior pattern corresponding to last usage behavior is supplied to user as predicting the outcome.
Advantageous Effects of the present invention is as follows:
(1) the present invention utilizes series model to use from the historical usage of user the use habit obtaining user's uniqueness record, wherein proposing the obtain manner of mobile subscriber's behavior sequence especially, using the problem of relevance to provide a kind of solution reliably for how obtaining application reliably.
(2) the present invention obtains the preposition use sequence of user in predicting according to the restriction of the effective active time of user, set the thought of predicting candidate value longest-prefix sequence, for providing a kind of feasible thinking in the problem of user's continuous print application usage forecastings difficulty simultaneously.
(3) the present invention proposes one and can use active period in conjunction with user, consider the Forecasting Methodology of Classical forecast pattern and the set of continuous application predictive mode, the application predicted can be made more to meet the current preference of targeted customer, thus reduce searching cost and the misoperation probability of user, and improve the mobile experience of user.
Accompanying drawing explanation
Fig. 1 is the system architecture schematic diagram of mobile terminal APP usage forecastings method of the present invention.
Embodiment
In order to more specifically describe the present invention, below in conjunction with the drawings and the specific embodiments, technical scheme of the present invention is described in detail.
The concrete steps that the present invention is based on the binding time context applied forcasting method of series model are as follows:
Obtain the application use record of user's the previous day, every bar uses record to comprise mobile phone application ID, service time, Apply Names.
History app according to mobile subscriber's behavioural characteristic processing target user uses record, and the effective one section of continuous print application obtained uses sequence, and the application comprised in this sequence is regarded as to have and effectively used correlativity.The behavior, the objective function Equation of sequence Slicing Model for Foreign was:
Wherein, F (i|Sj) represents whether app [i] should add this sequence with in the sequence that existing app [j] starts, and 1 expression adds in this subsequence, and 0 expression does not add; T (i_in) represents that the time that app [i] brings into use, T (j_out) represent that app [j] exits the time of use; St represents the session-time of setting, represents active time window size.The behavior sequence s form then obtained is as follows:
s
j=(x
1x
2...x
m),x
k∈I,1≤k≤m
Wherein, x is the application identities id in sequence, I is all set of applying id, i.e. I={i
1, i
2..., i
n, wherein i
jan app application in unique identification subscriber equipment, each s
jcorrespond to the unique sequence code mark sid that this sequence data is concentrated.
The behavior that can be obtained under different active time window definition by the setting of st uses sequence, user's historical usage can be used record cut into many and use sequence.Wherein, the application be close in each sequence is no more than active time window st at the active mistiming used, and namely has within the st time between adjacent application and effectively use correlativity.
Use sequence from the history app of the targeted customer obtained, use sequential mining to go out all app usage behavior patterns of user by sequence pattern from the app of user.Namely excavate all supports higher than 1 Frequent episodes, wherein, frequent sequence is defined as given positive integer support threshold ξ, if sequence α in sequence data set D support higher than ξ, i.e. support
d(α)>=ξ, then title sequence α is the Frequent episodes in D.Support defined formula is:
Wherein, in D, each row of data is made up of two tuple < sid, s >, and wherein sid represents the unique identification of sequence, and s represents that this line order arranges.Then the support of sequence α in D is the sequence number comprising α in the sequence of all two tuples of D.
Sequence comprises and is defined as given sequence α=< a
1a
2... a
n>, β=< b
1b
2... b
m, there is 1≤j in >
1< j
2< ... < j
n< m, makes
then claim α to be the subsequence of β, β is father's sequence of α, is called that β comprises α, is denoted as
The effective application that can be obtained all appearance by setting ξ=1 uses correlativity.Namely the application that may occur drawn from historical data uses combination continuously.By the excavation to the continuous print app application sequence used in effective active time, obtain the continuous print application usage behavior pattern of user, the use of the low frequency applications simultaneously used based on mobile subscriber app is often drunk with other has this feature of High relevancy, in Frequent episodes excavates, arrange support threshold is 1, ensures the seizure of the using forestland of low frequency applications.
The user behavior pattern of acquisition is carried out structured storage with trie tree.
Fig. 1 illustrates the binding time context-prediction Mobile solution usage forecastings method system framework based on sequence pattern.This recommend method is divided into three main modular: pretreatment module, pattern extraction module, prediction module.In pretreatment module, the historical usage first obtaining user uses record; Usage behavior sequence is applied again by cutting into continuous print user in user's history use record.Pattern extraction module comprises the behavior derivation of two kinds of patterns: sequence pattern is used for excavating continuous print application using forestland, and reflect user in certain user's active period and use the relevance of app, the behavior pattern excavated is with trie storage of data structure; Statistical model then reflects generally user and uses the frequent degree of application, namely uses record from the history app of user, by statistics application frequency of occurrence, extracts user and the most often uses top-n to apply.
In prediction module, first according to active time windowed time, obtain in active period corresponding within the current predictive time, the app sequence that user has used, i.e. prefix prediction sequence.
If prefix prediction sequence is empty, then represent that this user is not in a continuous print application and uses in region, Using statistics pattern is predicted; If prefix prediction sequence is not empty, then represent that user uses in application state at continuous print, the behavior pattern using series model to excavate is predicted, is undertaken predicting and according to corresponding sequence by longest-prefix prediction algorithm.According to emerging application record, real-time update is carried out to the training set existed; The formula that wherein sorts is as follows:
Wherein, the definition of degree of confidence (conf) and computing formula as follows:
Wherein,
represent measurable to being a to n-th from preposition n-1 item derivation sequence
nderivation rule;
represent prediction rule
degree of confidence, illustrate from preposition derivation sequence <a
1a
2a
n-1> predicts the confidence level that next step is applied as an, is the benchmark foundation of prediction sequence.
Above-mentioned is can understand and apply the invention for ease of those skilled in the art to the description of embodiment.Person skilled in the art obviously easily can make various amendment to above-described embodiment, and General Principle described herein is applied in other embodiments and need not through performing creative labour.Therefore, the invention is not restricted to above-described embodiment, those skilled in the art are according to announcement of the present invention, and the improvement made for the present invention and amendment all should within protection scope of the present invention.
Claims (6)
1., based on a sequence pattern binding time contextual mobile terminal APP usage forecastings method, comprise the steps:
(1) obtain APP history and use records series, described APP history uses records series to comprise user's a period of time interior every bar use record for APP in the past;
(2) records series is used to be divided into multiple relevant subsequence whole APP history according to the behavioural characteristic of user;
(3) add up the number of times that often kind of APP usage behavior pattern occurs in all relevant subsequence, and the form that all APP usage behavior patterns are set by trie is carried out structured storage;
(4) obtain APP prediction and use records series, described APP prediction uses records series to comprise user T before current time
stevery bar for APP in large minor time slice uses record, T
stfor the active time window size preset;
(5) by searching the APP usage behavior pattern predicted to APP and use records series relevant from trie tree, and then corresponding the predicting the outcome of generation is extracted according to the number of times that these APP usage behavior patterns occur in all relevant subsequence.
2. mobile terminal APP usage forecastings method according to claim 1, is characterized in that: described every bar uses record to include start time, end time and corresponding ID and the title using APP.
3. mobile terminal APP usage forecastings method according to claim 1, is characterized in that: the specific implementation process in described step (2), whole APP history use records series being divided into multiple relevant subsequence is as follows:
Arbitrary in records series is used to use record S for APP history
i, newly-built one comprises the relevant subsequence that this uses record, according to following relational expression from use record H
i-1start to judge that every bar uses record whether to record S with use one by one
ithere is correlativity:
Wherein: H
i-1for recording S for use
i-1deterministic process in first bar with use record S
i-1not there is the use record of correlativity, S
iand S
i-1being respectively APP history uses i-th in records series and the i-th-1 to use record, and i is natural number and 1≤i≤n, n is the use record sum in APP history use records series, F (S
i| H
i-1) for using record H
i-1s is recorded with use
icorrelativity, F (S
i| H
i-1)=1 represents to have correlativity, F (S
i| H
i-1)=0 represents not have correlativity, T
in(H
i-1) for using record H
i-1start time, T
out(S
i) for using record S
iend time;
If use record H
i-1s is recorded with use
ithere is correlativity, then by use record H
i-1include this relevant subsequence in and then judge that next uses record, by that analogy until occur that a certain bar uses record to record S with use
inot there is correlativity, then use record to be defined as H this
iand establish by use record S
ito H
ilast bar use this relevant subsequence recording and form;
If use record H
i-1s is recorded with use
ido not have correlativity, then this relevant subsequence is cancelled;
Traveling through APP history according to this uses the every bar in records series to use record, obtains multiple relevant subsequence.
4. mobile terminal APP usage forecastings method according to claim 1, it is characterized in that: if the APP that customer mobile terminal is installed adds up to m, then the species number of APP usage behavior pattern is
m is relevant subsequence maximum length.
5. mobile terminal APP usage forecastings method according to claim 1, is characterized in that: the number of times that in described step (3), statistics often kind of APP usage behavior pattern occurs in all relevant subsequence, and specific implementation process is as follows:
APP usage behavior pattern of choosing any one kind of them and a relevant subsequence, get the app [1] in this APP usage behavior pattern, makes the every bar in this relevant subsequence use record to mate with app [1] one by one:
All do not find one to use APP and the app [1] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [1] corresponding to record to mate identical, then use record to be defined as G this
1, and then from G
1start to make every bar use record to mate with app [2] one by one:
All do not find one to use APP and the app [2] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [2] corresponding to record to mate identical, then use record to be defined as G this
2, and then from G
2start to make every bar use record to mate with app [3] one by one; The rest may be inferred, until judge to perform from G
p-1start to make every bar use record to mate with app [p] one by one:
All do not find one to use APP and the app [p] corresponding to record identical if traveled through all records that uses, then judge that this relevant subsequence does not comprise this APP usage behavior pattern;
If occur, a certain bar uses APP and the app [p] corresponding to record to mate identical, then judge that this relevant subsequence comprises this APP usage behavior pattern i.e. this APP usage behavior pattern and occurs in this relevant subsequence; Wherein: app [j] is the APP in this APP usage behavior pattern corresponding to the usage behavior of jth position, j is natural number and 1≤j≤p, p is the length of this APP usage behavior pattern, G
p-1for the use record that bar first in the matching process for app [p-1] is identical with app [p-1] corresponding A PP;
Travel through all relevant subsequence according to this, just can count the number of times that this APP usage behavior pattern occurs in all relevant subsequence.
6. mobile terminal APP usage forecastings method according to claim 1, is characterized in that: the specific implementation process of described step (5) is as follows:
From trie tree, search length is N+1 and top N usage behavior and APP predicts that using every bar of records series to use records APP usage behavior pattern one to one, and N is that APP predicts that the use record in use records series is total;
If APP prediction use records series be sky, then from APP history use records series extract access times maximum before k APP be supplied to user as predicting the outcome, k be greater than 1 natural number;
If have found the several APP usage behavior pattern meeting above-mentioned condition by search, the number of times then occurred in all relevant subsequence according to these APP usage behavior patterns sorts, and chooses the maximum front k kind APP usage behavior pattern of occurrence number and the APP extracted in this k kind APP usage behavior pattern corresponding to last usage behavior is supplied to user as predicting the outcome;
If do not find any APP usage behavior pattern meeting above-mentioned condition by search, then records series is used to decompose to APP prediction, namely predict to use records series to remove after arbitrary use is recorded from APP and obtain a predictor sequence, traversal removes each bar use record and obtains N number of predictor sequence according to this; Arbitrary predictor sequence search length from trie tree is N and every bar of front N-1 position usage behavior and predictor sequence uses and records APP usage behavior pattern one to one, travels through all predictor sequences according to this; For searching for all APP usage behavior patterns obtained, sorting according to the number of times that these APP usage behavior patterns occur in all relevant subsequence, choosing the maximum front k kind APP usage behavior pattern of occurrence number and the APP extracted in this k kind APP usage behavior pattern corresponding to last usage behavior is supplied to user as predicting the outcome.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610056153.7A CN105528659A (en) | 2016-01-27 | 2016-01-27 | Mobile terminal APP usage prediction method combining with time-context based on sequence mode |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610056153.7A CN105528659A (en) | 2016-01-27 | 2016-01-27 | Mobile terminal APP usage prediction method combining with time-context based on sequence mode |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105528659A true CN105528659A (en) | 2016-04-27 |
Family
ID=55770873
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610056153.7A Pending CN105528659A (en) | 2016-01-27 | 2016-01-27 | Mobile terminal APP usage prediction method combining with time-context based on sequence mode |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105528659A (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106227844A (en) * | 2016-07-27 | 2016-12-14 | 深圳市金立通信设备有限公司 | The method of a kind of application recommendation and terminal |
CN106651606A (en) * | 2016-11-29 | 2017-05-10 | 河南科技大学 | Multimedia social network user behavior pattern discovery method |
CN107728874A (en) * | 2017-09-06 | 2018-02-23 | 阿里巴巴集团控股有限公司 | The method, apparatus and equipment of user prompt operation are provided |
CN108268322A (en) * | 2018-02-26 | 2018-07-10 | 北京小米移动软件有限公司 | The method, apparatus and computer readable storage medium of internal memory optimization |
CN110166529A (en) * | 2019-04-16 | 2019-08-23 | 平安普惠企业管理有限公司 | It keeps logging in state method, apparatus, equipment and storage medium |
CN111459797A (en) * | 2020-02-27 | 2020-07-28 | 上海交通大学 | Method, system and medium for detecting abnormity of developer behaviors in open source community |
CN111797318A (en) * | 2020-07-01 | 2020-10-20 | 喜大(上海)网络科技有限公司 | Information recommendation method, device, equipment and storage medium |
CN115562967A (en) * | 2022-11-10 | 2023-01-03 | 荣耀终端有限公司 | Application program prediction method, electronic device and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102231757A (en) * | 2011-06-29 | 2011-11-02 | 浙江大学 | On-line service combination recommendation system and recommendation method thereof |
CN104166696A (en) * | 2014-08-01 | 2014-11-26 | 小米科技有限责任公司 | Method and device for pushing application programs |
CN104427118A (en) * | 2013-09-06 | 2015-03-18 | 华为技术有限公司 | Method for recommending contents and mobile terminal |
CN105117424A (en) * | 2015-07-31 | 2015-12-02 | 中国科学院软件研究所 | Dwell-time-based moving object semantic behavior pattern mining method |
-
2016
- 2016-01-27 CN CN201610056153.7A patent/CN105528659A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102231757A (en) * | 2011-06-29 | 2011-11-02 | 浙江大学 | On-line service combination recommendation system and recommendation method thereof |
CN104427118A (en) * | 2013-09-06 | 2015-03-18 | 华为技术有限公司 | Method for recommending contents and mobile terminal |
CN104166696A (en) * | 2014-08-01 | 2014-11-26 | 小米科技有限责任公司 | Method and device for pushing application programs |
CN105117424A (en) * | 2015-07-31 | 2015-12-02 | 中国科学院软件研究所 | Dwell-time-based moving object semantic behavior pattern mining method |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106227844A (en) * | 2016-07-27 | 2016-12-14 | 深圳市金立通信设备有限公司 | The method of a kind of application recommendation and terminal |
CN106651606A (en) * | 2016-11-29 | 2017-05-10 | 河南科技大学 | Multimedia social network user behavior pattern discovery method |
CN106651606B (en) * | 2016-11-29 | 2021-02-05 | 河南科技大学 | Method for discovering user behavior pattern of multimedia social network |
CN107728874A (en) * | 2017-09-06 | 2018-02-23 | 阿里巴巴集团控股有限公司 | The method, apparatus and equipment of user prompt operation are provided |
CN108268322A (en) * | 2018-02-26 | 2018-07-10 | 北京小米移动软件有限公司 | The method, apparatus and computer readable storage medium of internal memory optimization |
CN110166529A (en) * | 2019-04-16 | 2019-08-23 | 平安普惠企业管理有限公司 | It keeps logging in state method, apparatus, equipment and storage medium |
CN111459797A (en) * | 2020-02-27 | 2020-07-28 | 上海交通大学 | Method, system and medium for detecting abnormity of developer behaviors in open source community |
CN111797318A (en) * | 2020-07-01 | 2020-10-20 | 喜大(上海)网络科技有限公司 | Information recommendation method, device, equipment and storage medium |
CN111797318B (en) * | 2020-07-01 | 2024-02-23 | 喜大(上海)网络科技有限公司 | Information recommendation method, device, equipment and storage medium |
CN115562967A (en) * | 2022-11-10 | 2023-01-03 | 荣耀终端有限公司 | Application program prediction method, electronic device and storage medium |
CN115562967B (en) * | 2022-11-10 | 2023-10-13 | 荣耀终端有限公司 | Application program prediction method, electronic device and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105528659A (en) | Mobile terminal APP usage prediction method combining with time-context based on sequence mode | |
CN108521439B (en) | Message pushing method and device | |
CN110928993B (en) | User position prediction method and system based on deep cyclic neural network | |
CN100412870C (en) | Gateway personalized recommendation service method and system introduced yuan recommendation engine | |
CN106600052B (en) | User attribute and social network detection system based on space-time trajectory | |
Abrol et al. | Tweethood: Agglomerative clustering on fuzzy k-closest friends with variable depth for location mining | |
CN106846061A (en) | Potential user's method for digging and device | |
CN104679743A (en) | Method and device for determining preference model of user | |
CN110149530A (en) | A kind of method for processing video frequency and device | |
CN104008203A (en) | User interest discovering method with ontology situation blended in | |
CN105335391A (en) | Processing method and device of search request on the basis of search engine | |
CN102646114A (en) | News topic timeline abstract generating method based on breakthrough point | |
Saad et al. | Archiving the web using page changes patterns: a case study | |
CN105159930A (en) | Search keyword pushing method and apparatus | |
CN104598583A (en) | Method and device for generating query sentence recommendation list | |
CN109947902B (en) | Data query method and device and readable medium | |
CN109800879A (en) | Construction of knowledge base method and apparatus | |
CN106202563A (en) | A kind of real time correlation evental news recommends method and system | |
CN106407381A (en) | Method and device for pushing information based on artificial intelligence | |
CN103500163B (en) | The method and apparatus of identification event key development | |
CN112084320A (en) | Test question recommendation method and device and intelligent equipment | |
Liang et al. | Exploring large language models for human mobility prediction under public events | |
CN116756576B (en) | Data processing method, model training method, electronic device and storage medium | |
Parker et al. | Named entity recognition through deep representation learning and weak supervision | |
CN115408095B (en) | Method and system for displaying tree data with large data volume based on dynamic paging technology |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160427 |