CN113127696B - Method for improving accuracy of influence measurement based on behaviors - Google Patents
Method for improving accuracy of influence measurement based on behaviors Download PDFInfo
- Publication number
- CN113127696B CN113127696B CN202110299342.8A CN202110299342A CN113127696B CN 113127696 B CN113127696 B CN 113127696B CN 202110299342 A CN202110299342 A CN 202110299342A CN 113127696 B CN113127696 B CN 113127696B
- Authority
- CN
- China
- Prior art keywords
- influence
- user
- users
- time
- behavior
- 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.)
- Active
Links
- 230000006399 behavior Effects 0.000 title claims abstract description 56
- 238000000034 method Methods 0.000 title claims abstract description 26
- 238000005259 measurement Methods 0.000 title abstract description 9
- 230000001105 regulatory effect Effects 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 4
- 230000002238 attenuated effect Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000009471 action Effects 0.000 claims description 2
- 230000008859 change Effects 0.000 claims description 2
- 238000005303 weighing Methods 0.000 abstract 1
- 230000000694 effects Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/901—Indexing; Data structures therefor; Storage structures
- G06F16/9024—Graphs; Linked lists
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Primary Health Care (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method for improving accuracy of influence measurement based on behaviors. And then calculating the position popularity according to the check-in place of the target user. Meanwhile, the total influence of the friend circle is redistributed according to the time model, so that influence factors of the friend circle are eliminated. And measuring the position similarity by using a Jaccard algorithm, and finally obtaining the influence of the specified users. The invention considers various reasons for causing similar sign-in behaviors among users, personal preference of the users, common influence of the friends of the users, position hot factors and the like, thereby ensuring that the influence weighing result between two users is more accurate.
Description
Technical Field
The invention relates to the technical field of Internet, belongs to the field of influence measurement, and in particular relates to a method for improving accuracy of influence measurement based on behaviors.
Background
From the end of the last century, web 2.0 technology has led to rapid development of online social networks, such as Twitter, facebook, microblog, etc. Thereafter, researchers can collect a large amount of real world-compliant data through online websites for the first time. The correlation between social influence and a plurality of reasons such as network structure, user behavior, public opinion and the like in the social network can be studied in depth. The method has the advantages that a large number of achievements are obtained on the aspect of influence force measurement of the social network, and the method is widely applied to the fields of expert investigation, recommendation systems, commodity marketing and the like. Meanwhile, people are used to share daily life on a social network platform, people are not simply receivers of the content, and the people also become producers and propagators of the content. Studying the impact metrics between users has a tremendous effect.
Currently, location-based social networks are becoming more popular, and users are enthusiastically sharing their geographic locations and signing in, posting comments and opinions, and the like in the social networks. At present, many researches on the aspect of a social network based on the position exist, but a part of methods for measuring influence in the researches are simply measured from a network topology structure, so that key interaction content information among users is lacked, and the influence is poor in accuracy; and part of the user behavior and the user-related social text information are considered, so that the user characteristics are enriched, and the influence and measurement effect is better than that based on the network topology structure. However, the above measurement methods are mainly based on the behavior of similarity between users to determine the influence, and the criterion of this determination is too single, so that for the measurement of influence of users based on sign-in information, it is easy for two users to sign in at the same location but there is no relation between them.
In summary, the problems of the prior art are: the existing influence quantity based on the user check-in information mainly determines influence among users through the similarity of the user check-in positions or related check-in information. The existing measuring method has single standard and poor measuring precision of influence.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides an influence measuring method based on sign-in information, which has the following technical scheme:
a method for improving accuracy of a behavior-based impact metric, comprising the steps of:
Step 1, constructing an initial network;
step 2, measuring the influence of the user m on the single behavior of the target user n, and judging whether a connecting edge exists between the user m and the target user n; the two users m and n have no connecting edges, and the influence of the two users m and n under the sign-in action is 0; m, n users have a connecting edge, then carry out step 3;
step 3, eliminating the influence of personal preference of the user;
step 4, calculating the influence of the influence factors of the masses of the society;
Step 5, eliminating the influence from the circle factors of friends;
Step 6, calculating influence similar based on user preference;
step 7, correcting the influence of the user behavior similarity on the change of the user behavior similarity along with time;
And 8, calculating the influence of m on the target user n on the basis of the steps 3 to 7.
Preferably, the undirected graph G (V, E, C) is used in step 1 to represent a location-based social network, V representing a set of nodes in the network, i.e. a set of social network users; e represents a set of edges in the network, namely relationships between social users; check-in recording for arbitrary user xRepresenting an ith check-in record where i represents user x,Indicating that the user x checked-in location,Indicating that user x is atIs a check-in time of (a).
Preferably, in step2 the following sub-steps are included:
step 2.1: if m and n users sign in at the same position and the sign-in time of m is earlier than that of n, the influence exists, and the step 3 is directly entered;
step 2.2: if m and n users check in at the same position, but the check-in time of m is later than that of n, or mn users do not check in at the same position, no influence exists;
Preferably, in step 3, the influence of personal preference of the user is removed according to the check-in records of all users Is to check-in time of (f)Extracting the same position of the userThe earliest sign-in record is then deleted and the user sign-in position is deletedAnd check-in time is atAnd finally, constructing a single-user primary behavior record set for the primary behavior records extracted by each user, wherein the single-user primary behavior record set is represented as follows:
Preferably, in step 4, the social masses are defined herein as indirect friends of the user, as opposed to social circles, i.e. the influence of all first order friends; the influence factors of the masses in the society are calculated by using popularity, and the influence factors of the masses in the society show an inverse proportion relation; the popularity of a place refers to the degree that a place is touted by the masses of the society in a current period of a certain length, and the popularity of a place is as follows: the number of times that the position is checked in by all users in the current time period is the proportion of the total number of times that the position is checked in by all positions in the time period; the method comprises the following two substeps:
Step 4.1: calculating a position According to the sign-in record of the user after the primary behavior extraction in the step 2, calculating the proportion of the number of times that the position is signed in by all users in the current time period to the total sign-in number of times, wherein the proportion is expressed as follows:
Wherein, Representing the positionAt the moment of timeIs the hot degree of (k) is the time period threshold,Representing the presence behavior of user mOr alternativelyCheck-in timeCheck-in time of ratio nEarly k, || represents the size of the collection;
Step 4.2: the total influence from all neighbor friend users to which user n is subjected after the hot influence factors of the position are removed is expressed as follows:
g 1 is the parameter to be regulated, Representing the behavior occurring at nFriend user who has been previously in active stateRepresenting the total influence suffered by user n from all first-order friend users;
Preferably, in step 5, the circle of friends is composed of a plurality of neighbor friends of the target user n, firstly, according to a theory that the influence of behaviors between users is attenuated along with the extension of time intervals, an influence time attenuation model is built, and then, according to influence factors of the circle of friends of the user, the influence of the target user is redistributed according to time sequence; the influence of different users in the friend circle on the target user is redistributed according to the influence of m on n by the softmax function, and the method mainly comprises the following 2 substeps:
Step 5.1: the decay rate is quantified by using an e-exponential time decay model in probability theory, which is expressed as follows:
Wherein I (n|m) represents the influence of user m on n, sigma is the attenuation coefficient, Between 0 and 1, I' (n|m) represents the influence before the time decay;
step 5.2: under the time attenuation model, the influence of different users in the friend circle on the target user is different due to different activation time intervals; and (3) carrying out influence reassignment of m to n according to a softmax function, and removing the influence factors of the friend circle, wherein an influence model of m to n is as follows:
Wherein, The group influence probability of m's circle of friends on n is represented.
Preferably, in step 6, all check-in positions of users m, n are extracted, expressed as:
Preferably, in step 7, the user similarity is time-varying, the user's position is less in the initial stage and more in the later stage, and the results of the two behavior preference similarity calculations are different, so that the Jacquard similarity coefficient is improved in consideration of the time-varying user behavior similarity; suppose that users n and m are respectively And (3) withSimilar behavior occurs, namely going to the same place, then the check-in position records of users n and m are written as follows: Wherein the method comprises the steps of The improved jaccard similarity coefficient is expressed as follows:
Preferably, in step 8, the influence of m on the target user n is calculated, and the influence between m and n is separated from the confusion factor according to the preference similarity between m and n, the location hot influence factor and the influence factor model of the friend circle of the target user n in steps 3-7; wherein the influence is inversely proportional to the position similarity and the position popularity, and the influence of the target user n on m is redistributed by the position similarity based on the influence of the friend circle on the total influence of n; m versus n single pass behavior based The influence of (2) is as follows:
Wherein g is a regulating factor.
The technical effects are as follows:
The technical method of the invention considers various reasons for causing similar sign-in behaviors among users, personal preference of the users, common influence of the friends of the users, position hot factors and the like, thereby ensuring that the influence and the measurement result between two users are more accurate.
Drawings
Fig. 1 is a flow chart of the method of the present invention.
DETAILED DESCRIPTION OF EMBODIMENT (S) OF INVENTION
The following describes the technical scheme of the present invention in detail with reference to examples.
Examples:
By applying the invention to the social network dataset Gowalla, which is a provider of a large location social service, a user can punch a card at a location, and thus the Gowalla dataset is well suited for building a location-based social network. The impact of a given user m on the target user n under a single location access activity Q m is calculated on golella, where m is earlier than n is the same location Q. The specific steps are carried out according to the following steps:
Step 1: the check-in record of each user in the social network consists of a check-in position and a check-in time, and the check-in record of any user x is expressed as a record Where i represents the ith check-in record for user x,Indicating that the user x checked-in location,Indicating that user x is atIs a check-in time of (a).
Step 2: the metric specifies the influence between the two users m, n and determines whether there is influence between user m and target user n.
Step 2.1: if m and n users check in at the same position and the check-in time of m is earlier than that of n, the influence exists, and the step 3 is directly entered.
Step 2.2: if m and n users check in at the same position, but the check-in time of m is later than that of n, or if m and n users do not check in at the same position, no influence exists:
Step 3: user repeated behavior removal according to sign-in records of all users Is to check-in time of (f)Extracting the same position of the userThe earliest sign-in record is then deleted and the user sign-in position is deletedAnd check-in time is atAnd finally, constructing a single-user primary behavior record set for the primary behavior record extracted by each user. The expression is as follows:
Step 4: and calculating the influence of the hot factors of the positions, wherein the influence of the hot factors of the positions is different from the influence of the circle of friends (all neighbor friends), and the influence and the hot influence factors of the positions are calculated by using the hot degree, so that the influence and the hot influence factors of the positions have an inverse relation. The hot degree of a position refers to the degree that a certain position is tracked by the hot door in a certain current length period, and the hot degree of the certain position is as follows: the number of times a location is checked in by all users in the current time period is a proportion of the total number of times the location is checked in for all locations in the time period. The method mainly comprises the following two substeps:
Step 4.1: calculating a position According to the sign-in records of the users after the primary behavior extraction in the step 2, calculating the proportion of the number of times that the positions are signed in by all users in the current time period to the total sign-in number. The expression is as follows:
Wherein, Representing the positionAt the moment of timeIs the hot degree of (k) is the time period threshold,Representing the presence behavior of user m(Or) Check-in timeCheck-in time of ratio nEarly k, || represents the size of the collection.
Step 4.2: the total influence from all neighbor friend users to which user n is subjected after the hot influence factors of the position are removed is expressed as follows:
g 1 is the parameter to be regulated, Representing the behavior occurring at nFriend users who have been previously in an active state.
Step 5: removing the influence from the factors of the friend circle, wherein the friend circle consists of a plurality of neighbor friends of the target user n, firstly, establishing an influence time attenuation model according to the theory that the influence of behaviors among users is attenuated along with the extension of time intervals, and then, reallocating the influence of the target user according to the influence factors of the friend circle and the time sequence. The influence of different users in the friend circle on the target user is redistributed according to the influence of m to n by the softmax function, wherein the influence of different users in the friend circle on the target user is different according to different activation time intervals. The method mainly comprises the following 2 substeps:
Step 5.1: the decay rate is quantified by using an e-exponential time decay model in probability theory, which is expressed as follows:
Wherein I (n|m) represents the influence of user m on n, sigma is the attenuation coefficient, Between 0 and 1, I' (n|m) represents the influence before the lapse of time.
Step 5.2: under the time decay model, the influence of different users in the friend circle on the target user is different due to different activation time intervals. And (3) carrying out influence reassignment of m to n according to a softmax function, and removing the influence factors of the friend circle, wherein an influence model of m to n is as follows:
Wherein, The group influence probability of m's circle of friends on n is represented.
Step 6: calculating the influence based on the position similarity, and extracting all check-in positions of the users m and n, wherein the positions are expressed as follows:
Step 7: as the user similarity changes with time, the position behavior records of the user are fewer in the initial stage, the position behavior records of the user are more in the later stage, the two behavior preference similarity calculation results are different, and the Jacquard similarity coefficient is improved by considering the fact that the user behavior similarity changes with time. Suppose that users n and m are respectively And (3) withSimilar behavior (going to the same place) occurs, then the check-in position records of users n and m are written as: Wherein the method comprises the steps of The improved jaccard similarity coefficient is expressed as follows:
Step 8: and (3) calculating the influence of m on the target user n, and separating the influence of m and n from the confusion factor according to the preference similarity between m and n in the steps (3-7), the position hot influence factor and the influence factor model of the friend circle of the target user n. Wherein the influence is inversely proportional to the location similarity and the location popularity, and the influence of the target user n by m is a redistribution of the influence based on the total influence of the circle of friends on n divided by the location similarity. m versus n single pass behavior based The influence of (2) is as follows:
Wherein g is a regulating factor.
It should be understood that parts of the specification not specifically set forth herein are all prior art.
It should be understood that the foregoing description of the preferred embodiments is not intended to limit the scope of the invention, but rather to limit the scope of the claims, and that those skilled in the art can make substitutions or modifications without departing from the scope of the invention as set forth in the appended claims.
Claims (5)
1. A method for improving accuracy of a behavior-based impact metric, comprising the steps of:
Step 1, constructing an initial network;
Step 2, measuring the influence of the user m on the single behavior of the target user n, and judging whether a connecting edge exists between the user m and the target user n; the influence of the m and n users under the sign-in behavior is 0 when the m and n users have no connecting edges; m, n users have a connecting edge, then carry out step 3;
step 3, eliminating the influence of personal preference of the user;
step 4, calculating the influence of the influence factors of the masses of the society;
Step 5, eliminating the influence from the circle factors of friends;
Step 6, calculating influence similar based on user preference;
step 7, correcting the influence of the user behavior similarity on the change of the user behavior similarity along with time;
Step 8, calculating the influence of m on the target user n on the basis of the steps 3 to 7;
Using an undirected graph G (V, E) in step1 to represent a location-based social network, V representing a set of nodes in the network, i.e. a set of social network users; e represents a set of edges in the network, namely relationships between social users; check-in recording for arbitrary user x Representing, where i represents the ith check-in record for user x,Indicating that the user x checked-in location,Indicating that user x is atIs checked in time of (1);
In step 3, eliminating the influence of personal preference of the user according to the check-in records of all users Is to check-in time of (f)Extracting the same position of the userThe earliest sign-in record is then deleted and the user sign-in position is deletedAnd check-in time is atAnd finally, constructing a single-user primary behavior record set for the primary behavior records extracted by each user, wherein the single-user primary behavior record set is represented as follows:
in step 4, the social masses are defined as indirect friends of the users; the influence factors of the masses in the society are calculated by using popularity, and the influence factors of the masses in the society show an inverse proportion relation; the popularity of a place refers to the degree that a place is touted by the masses of the society in a current period of a certain length, and the popularity of a place is as follows: the number of times that the position is checked in by all users in the current time period is the proportion of the total number of times that the position is checked in by all positions in the time period; the method comprises the following two substeps:
Step 4.1: calculating a position According to the sign-in record of the user after the initial behavior extraction in the step 3, calculating the proportion of the number of times that the position is signed in by all users in the current time period to the total sign-in number of times, wherein the proportion is expressed as follows:
Wherein, Representing the positionAt the moment of timeIs the hot degree of (k) is the time period threshold,Representing the position of user mOr alternativelyCheck-in timeCheck-in time of ratio nEarly k, || represents the size of the collection;
Step 4.2: the total influence from all neighbor friend users to which user n is subjected after the hot influence factors of the position are removed is expressed as follows:
g 1 is the parameter to be regulated, Indicating that user n is in positionA friend user who has been active prior to the occurrence of the action,Representing the total influence suffered by user n from all first-order friend users;
In step 5, the friend circle is composed of a plurality of neighbor friends of the target user n, firstly, according to a theory that the influence of behaviors among users is attenuated along with the extension of time intervals, an influence time attenuation model is built, and then, according to influence factors of the friend circle of the user, the influence of the target user is redistributed according to time sequence; the influence of different users in the friend circle on the target user is redistributed according to the influence of m on n by the softmax function, and the method comprises the following 2 substeps:
Step 5.1: the decay rate is quantified by using an e-exponential time decay model in probability theory, which is expressed as follows:
Wherein I (n|m) represents the influence of user m on n, sigma is the attenuation coefficient, Between 0 and 1, I ′ (n|m) represents the influence before the time decay;
step 5.2: under the time attenuation model, the influence of different users in the friend circle on the target user is different due to different activation time intervals; and (3) carrying out influence reassignment of m to n according to a softmax function, and removing the influence factors of the friend circle, wherein an influence model of m to n is as follows:
Wherein, The group influence probability of m's circle of friends on n is represented.
2. The method for improving accuracy of behavior-based impact metrics according to claim 1, characterized in that in step 2 the following sub-steps are included:
step 2.1: if m and n users sign in at the same position and the sign-in time of m is earlier than that of n, the influence exists, and the step 3 is directly entered;
Step 2.2: if m and n users check in at the same position, but the check-in time of m is later than that of n, or if m and n users do not check in at the same position, no influence exists;
3. the method for improving accuracy of behavior-based impact metrics according to claim 2, characterized in that in step 6, all check-in positions of users m, n are extracted, expressed as:
4. A method for improving accuracy of influence on a performance-based measure according to claim 3, wherein in step 7, the user similarity is time-varying, there are fewer records of the user's position and behavior in the initial stage and more records of the user's position and behavior in the later stage, the results of the calculation of the behavior preference similarity are different, and the user's behavior similarity is considered to be time-varying, so that the jekcard similarity coefficient is improved; suppose that users n and m are respectively And (3) withSimilar behavior occurs, namely going to the same place, then the check-in position records of users n and m are written as follows:
Wherein the method comprises the steps of The improved jaccard similarity coefficient is expressed as follows:
5. The method for improving accuracy of influence metrics based on behaviors according to claim 4, wherein in step 8, the influence of m on the target user n is calculated, and the influence between m and n is separated from the confounding factors according to the preference similarity between m and n, the positional hot influence factors and the influence factor model of the friend circle of the target user n in steps 3-7; wherein the influence is inversely proportional to the position similarity and the position popularity, and the influence of the target user n on m is redistributed by the position similarity based on the influence of the friend circle on the total influence of n; m versus n single pass behavior based The influence of (2) is as follows:
Wherein g is a regulating factor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110299342.8A CN113127696B (en) | 2021-03-21 | 2021-03-21 | Method for improving accuracy of influence measurement based on behaviors |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110299342.8A CN113127696B (en) | 2021-03-21 | 2021-03-21 | Method for improving accuracy of influence measurement based on behaviors |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113127696A CN113127696A (en) | 2021-07-16 |
CN113127696B true CN113127696B (en) | 2024-07-12 |
Family
ID=76773636
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110299342.8A Active CN113127696B (en) | 2021-03-21 | 2021-03-21 | Method for improving accuracy of influence measurement based on behaviors |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113127696B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113239285A (en) * | 2021-04-16 | 2021-08-10 | 江汉大学 | Processing method, device and processing equipment for social network influence |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106919564A (en) * | 2015-12-24 | 2017-07-04 | 天津科技大学 | A kind of influence power measure based on mobile subscriber's behavior |
CN106952166A (en) * | 2016-01-07 | 2017-07-14 | 腾讯科技(深圳)有限公司 | The user force evaluation method and device of a kind of social platform |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109885797B (en) * | 2019-02-18 | 2020-12-01 | 武汉大学 | Relational network construction method based on multi-identity space mapping |
CN109919459B (en) * | 2019-02-21 | 2022-05-13 | 武汉大学 | Method for measuring influence among social network objects |
-
2021
- 2021-03-21 CN CN202110299342.8A patent/CN113127696B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106919564A (en) * | 2015-12-24 | 2017-07-04 | 天津科技大学 | A kind of influence power measure based on mobile subscriber's behavior |
CN106952166A (en) * | 2016-01-07 | 2017-07-14 | 腾讯科技(深圳)有限公司 | The user force evaluation method and device of a kind of social platform |
Also Published As
Publication number | Publication date |
---|---|
CN113127696A (en) | 2021-07-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20170364933A1 (en) | User maintenance system and method | |
CN103533390B (en) | The method and system of television program recommendations are carried out based on social network information | |
CN113568819B (en) | Abnormal data detection method, device, computer readable medium and electronic equipment | |
JP7271529B2 (en) | Automated attribution modeling and measurement | |
CN104281882A (en) | Method and system for predicting social network information popularity on basis of user characteristics | |
CN109784997B (en) | Short video active user prediction method based on big data | |
CN107633035B (en) | Shared traffic service reorder estimation method based on K-Means and LightGBM model | |
US9147161B2 (en) | Determining geo-locations of users from user activities | |
CN107423859A (en) | A kind of built-up pattern modeling method and system | |
CN109409393A (en) | A method of User Activity track is modeled using track insertion | |
US20140214480A1 (en) | Determining a customer profile state | |
CN110134883B (en) | Heterogeneous social network location entity anchor link identification method | |
CN112612942B (en) | Social big data-based fund recommendation system and method | |
CN103136331A (en) | Micro blog network opinion leader identification method | |
CN113127696B (en) | Method for improving accuracy of influence measurement based on behaviors | |
CN109783805A (en) | A kind of network community user recognition methods and device | |
CN115422441A (en) | Continuous interest point recommendation method based on social space-time information and user preference | |
CN116578726A (en) | Personalized book recommendation system | |
CN110807667A (en) | Method and device for activating sleeping customers | |
CN113886697A (en) | Clustering algorithm based activity recommendation method, device, equipment and storage medium | |
CN112819499A (en) | Information transmission method, information transmission device, server and storage medium | |
CN111221915B (en) | Online learning resource quality analysis method based on CWK-means | |
Yan et al. | User recommendation with tensor factorization in social networks | |
CN114676324B (en) | Data processing method, device and equipment | |
Gençer et al. | A new framework for increasing user engagement in mobile applications using machine learning techniques |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant |