CN103678474B - A kind of method of a large amount of hot issue of quick obtaining in social networks - Google Patents

Abstract

A method to quickly obtain a large number of hot topics in the social network, capture the forwarding records of the "status" published by the user in the social network; cluster all the "status" content through a clustering algorithm, and each class is defined as a Events; by analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the friend group; put these K friends in a specific friend group and recommend them to the target user . The advantage of this method is that by analyzing the historical forwarding conditions of the user's friends in the social network, the friends who can best cover all the latest news are collected and placed in a specific group. When the time is limited or the number of "statuses" is too large, the user only needs to quickly browse all the news of this group, and can quickly grasp current affairs hot spots and hot topics.

Description

A way to quickly get a lot of trending topics in social networks

technical field

The invention relates to the technical field of friend group optimization in social networks, in particular to optimize friend group from the perspective of helping users obtain more latest news in a short time.

Background technique

In recent years, with the rapid development of the Internet, people's circle of making friends has also begun to shift from reality to the Internet, and the rise of social networks has greatly expanded the scope of people's making friends. From relatives and friends around you to strangers in the industry and entertainment stars, social networks provide ordinary users with a broader platform for making friends and an effective way to obtain information. Users can obtain a large amount of data information in social networks every day, and the amount of information dissemination in social networks has far exceeded that of traditional news media such as radio, television, and newspapers. However, most messages are time-sensitive. With the increase of followers, the number of messages will increase sharply. Everyone’s time and energy are limited, so how to filter out more latest news from a large number of messages in the shortest possible time? Messaging is a burning problem.

Many social networks provide the function of grouping friends, which can be selectively read by only displaying the "status" of a certain group. Users can be grouped according to the relationship between friends and themselves, such as relatives and friends; they can also be grouped according to the professional status of friends, such as movie stars, computer engineers, etc. In order to help users obtain a large amount of latest news in the shortest time, the present invention proposes a new friend grouping method. By analyzing the historical forwarding conditions of user friends, the friends who can best cover all the latest news are collected and placed in one in a specific group. When the time is limited or the number of "statuses" is too large, the user only needs to quickly browse all the news of this group, and can quickly grasp current affairs hot spots and hot topics.

Contents of the invention

In order to facilitate users to obtain a large amount of latest news in the shortest time, and grasp current current affairs hot spots and hot topics, the present invention proposes a method for quickly obtaining a large number of hot topics in social networks:

1. The method comprises the following steps:

1) Grab the forwarding record of the "status" posted by the user in the social network, including the user name, forwarded content, forwarding time, forwarding amount, original author and original publishing time of the "status";

2) Cluster all "status" content through a clustering algorithm, and each class is defined as an event;

3) By analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the friend group;

4) Put these K friends in a specific friend group and recommend them to target users.

2. In step 2), all "status" contents are clustered through the clustering algorithm, and each class is defined as an event, which is characterized in that:

1) Each type of "state" is defined as an event representation, and obtaining any piece of "state" information in the event means obtaining news of this type of topic.

3. In step 3), by analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the "status" friend group, which is characterized in that:

3.1 Assuming that the user reposts any "status" in an event, it means that the user has covered the event;

3.2 Arbitrarily select K friends of the target user to form a set A, define y=T(i, A), which means the time when the set A covers the event i, that is, the minimum value of all the times when all users in A cover the event i, if Set A does not cover event i, then record T(i, A) = ∞;

3.3 Definition is a penalty function, mapping time t to a real number, representing the loss caused by covering the event at time t,

where m _i is the number of forwarding times of all "states" in event i, sum(i) is the number of forwarding times of all "states", we assume that the importance of an event is proportional to the proportion of forwarding, so the loss is the product of the coverage time and the importance coefficient In direct proportion, the penalty function f _i (t) can be changed according to the actual situation. If T(ui, A)=∞, then f _i (t) takes the maximum value of the function FMax (artificially set);

3.4 Traverse all events and define the penalty function of the entire network:

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

in Indicates the probability of occurrence of event i, m _i is the number of forwarding times of all "states" in event i, and total(i) is the number of all "states";

3.5 Assuming that a certain behavior of user b is directly or indirectly affected by user a, and the impact factor is greater than a certain threshold, it is considered that a covers b. In addition to the influence of the reposter, it may also be potentially influenced by other users. The original author and the reposter may only affect user b on this event, while user a may play a more critical role in other events. We This process will be represented by a probabilistic model;

3.6 Definition σ(A) represents the number of users covered by set A, that is, all users in set A cover the total number of non-repeated users respectively, σ(A) has multiple calculation methods, and the present invention adopts a linear threshold model calculation method, definition:

$\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; v</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; w</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&Greater\; Equal;</span>{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; )</span>$

In this model, b _{v, w} = e ^-a(tv-tw) (t _v > t _w and there is a friend relationship between v and w, otherwise the value is 0) represents the influence factor of user w on v in event i, t _w , t _w are respectively w

and v cover the time of event i, a, θ _v are adjustable parameters, and I is an indicative function.

3.7 Define the objective function:

$\mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; A</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&beta;\&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span>$

Among them, F(A) is the penalty function of the entire network, σ(A) represents the number of users covered by set A, and β is an adjustable parameter. By solving the above objective function, K friends who can cover the most events in the shortest time are obtained. K is artificially set, and the users in set A have the following characteristics: a) they have great influence in the user group of the objective function, and the forwarding rate of published messages is high; important news;

3.8 Minimizing the objective function G(A) is an NP-hard problem. Define R _i (A)=fi(∞)-f _i (T(i,A)), then

Minimizing G(A) is equivalent to maximizing H(A)=R(A)+βσ(A). It can be proved that H(A) is a submodular function, and an approximate solution can be obtained by a greedy algorithm, and the approximate ratio Greater than 1-1/e=0.63.

The invention proposes a new friend grouping method in a social network, which has the advantage of collecting friends who can best cover all the latest news by analyzing the historical forwarding conditions of user friends and putting them in a specific group. When the time is limited or the number of "statuses" is too large, the user only needs to quickly browse all the news of this group, and can quickly grasp current affairs hot spots and hot topics.

Description of drawings

Fig. 1 is a flow chart of the method of the present invention.

detailed description

With reference to accompanying drawing, further illustrate the present invention:

A way to quickly get a lot of trending topics in social networks:

1. The method comprises the following steps:

1) Grab the forwarding record of the "status" posted by the user in the social network, including the user name, forwarded content, forwarding time, forwarding amount, original author and original publishing time of the "status";

2) Cluster all "status" content through a clustering algorithm, and each class is defined as an event;

3) By analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the friend group;

4) Put these K friends in a specific friend group and recommend them to target users.

2. In step 2), all "status" contents are clustered through the clustering algorithm, and each class is defined as an event, which is characterized in that:

1) Each type of "state" is defined as an event representation, and obtaining any piece of "state" information in the event means obtaining news of this type of topic.

3. In step 3), by analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the "status" friend group, which is characterized in that:

1) Assuming that the user forwards any "status" in a certain event, it means that the user has covered the event;

2) Arbitrarily select K friends of the target user to form a set A, define t=T(1, A), which means the time when the set A covers event i, that is, the minimum value of all the times when all users in A cover event i, If set A does not cover event i, record T(i, A) = ∞;

3) Definition is a penalty function, mapping time t to a real number, representing the loss caused by covering the event at time t, is the importance coefficient of event i, where m _i is the number of forwarding times of all "states" in event i, sum(i) is the number of forwarding times of all "states", we assume that the importance of an event is proportional to the proportion of forwarding, so the loss is equal to The coverage time is proportional to the product of important coefficients. Here, the penalty function f _i (t) can be changed according to the actual situation. If T(i, A) = ∞, then f _i (t) takes the maximum value of the function FMax (artificially set Certainly);

4) Traverse all events and define the penalty function of the entire network:

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$

in Indicates the probability of occurrence of event i, m _i is the number of forwarding times of all "states" in event i, and total(t) is the number of all "states";

5) Assuming that a certain behavior of user b is directly or indirectly affected by user a, and the impact factor is greater than a certain threshold, it is considered that a covers b. In addition to being influenced by the forwarded person, it may also be potentially influenced by other users. The original author and the forwarded person may only affect user b on this event, while user a may play a more critical role in other events. We will represent this process with a probabilistic model;

6) Define σ(A) to represent the number of users covered by set A, that is, all users in set A cover the total number of unique users. There are many calculation methods for σ(A). The present invention adopts the linear threshold model calculation method ,definition:

$\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; v</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; w</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&Greater\; Equal;</span>{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; )</span>$

In this model, b _{v, w} = e ^-a(tv-tw) (t _v > t _w and there is a friend relationship between v and w, otherwise the value is 0) represents the influence factor of user w on v in event i, t _w , t _v are the time when w and v cover event i respectively, a and θ _v are adjustable parameters, and I is an indicative function.

7) Define the objective function:

$\mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; A</span>}}{\mathrm{<span\; class="notranslate">\; min</span>}}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&beta;\&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span>$

Among them, F(A) is the penalty function of the entire network, σ(A) represents the number of users covered by set A, and β is an adjustable parameter. By solving the above objective function, K friends who can cover the most events in the shortest time are obtained. K is artificially set, and the users in set A have the following characteristics: a) they have great influence in the user group of the objective function, and the forwarding rate of published messages is high; important news;

8) Minimizing the objective function G(A) is an NP-hard problem. Define R _i (A)=f _i (∞)-f _i (T(i,A)), then

Minimizing G(A) is equivalent to maximizing H(A)=R(A)+βσ(A) It can be proved that H(A) is a submodular function, and an approximate solution can be obtained by a greedy algorithm, and the approximate ratio is greater than 1-1/e=0.63.

The content described in the embodiments of this specification is only an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms stated in the embodiments. Equivalent technical means that a person can think of based on the concept of the present invention.

Claims (1)

1. A method for quickly obtaining a large number of hot topics in a social network, the method is characterized in that: 1) Grab the forwarding record of the "status" posted by the user in the social network, including the user name, forwarded content, forwarding time, forwarding amount, original author and original publishing time of the "status"; 2) Clustering all "status" content by clustering algorithm, each class is defined as an event; described in step 2) clustering all "status" content by clustering algorithm, each class is defined as An event characterized by: Each type of "state" is defined as an event representation, and obtaining any piece of "state" information in the event means obtaining news of this type of topic; 3) By analyzing the "status" forwarding records, for the target user, select K friends who can cover the most events in the shortest time in the friend group; Described in step 3) by analyzing "state" forwarding record, for the target user, select K good friends who can cover the most events in the shortest time in its "state" friend group, it is characterized in that: 3.1 Assuming that the user reposts any "status" in an event, it means that the user has covered the event; 3.2 Arbitrarily select K friends of the target user to form a set A, define t=T(i, A), which means the time when the set A covers the event i, that is, the minimum value of all the times when all users in A cover the event i, if Set A does not cover event i, then record T(i, A) = ∞; 3.3 Definition is a penalty function, mapping time t to a real number, representing the loss caused by covering the event at time t, is the importance coefficient of event i, where m _i is the number of forwarding times of all "states" in event i, sum(i) is the number of forwarding times of all "states", we assume that the importance of an event is proportional to the proportion of forwarding, so the loss is equal to The coverage time is directly proportional to the product of the important coefficients. Here, the penalty function f _i (t) can be changed according to the actual situation. If T(i, A)=∞, then f _i (t) takes the maximum value of the function FMax; 3.4 Traverse all events and define the penalty function of the entire network: $\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>$ in Indicates the probability of occurrence of event i, m _i is the number of forwarding times of all "states" in event i, and total(i) is the number of all "states"; 3.5 Assuming that a certain behavior of user b is directly or indirectly affected by user a, and the impact factor is greater than a certain threshold, it is considered that a covers b. In addition to the influence of the reposter, it may also be potentially influenced by other users. The original author and the reposter may only affect user b on this event, while user a may play a more critical role in other events. A probabilistic model represents this process; 3.6 Definition σ(A) represents the number of users covered by set A, that is, all users in set A cover the total number of non-repeated users respectively, σ(A) has multiple calculation methods, and the present invention adopts a linear threshold model calculation method, definition: $\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\underset{\mathrm{<span\; class="notranslate">\; v</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\underset{\mathrm{<span\; class="notranslate">\; w</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; b</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; w</span>}}<span\; class="notranslate">\; \&Greater\; Equal;</span>{\mathrm{<span\; class="notranslate">\; \&theta;</span>}}_{\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; )</span>$ in the model (t _v >t _w and there is a friend relationship between v and w, otherwise the value is 0) indicates the influence factor of user w on v in event i, t _w and t _v are the time when w and v cover event i respectively, a. θ _v is an adjustable parameter, and I is an indicative function; 3.7 Define the objective function: $\mathrm{<span\; class="notranslate">\; G</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; A</span>}}{\mathrm{<span\; class="notranslate">\; m</span>}\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; no</span>}}\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&beta;</span>}\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; A</span>}<span\; class="notranslate">\; )</span>$ Among them, F(A) is the penalty function of the entire network, σ(A) represents the number of users covered by set A, and β is an adjustable parameter. By solving the above objective function, K friends who can cover the most events in the shortest time are obtained. K is artificially set, and users in set A have the following characteristics: a) It has great influence in the user group of the target function, and the forwarding rate of published messages is very high; b) In a relatively short period of time, forward a large number of important news released by others; 3.8 Minimizing the objective function G(A) is an NP-hard problem. Define R _i (A)=f _i (∞)-f _i (T(i,A)), then Minimizing G(A) is equivalent to maximizing H(A)=R(A)+βσ(A). It can be proved that H(A) is a submodular function, and an approximate solution can be obtained by a greedy algorithm, and the approximate ratio Greater than 1-1/e=0.63; 4) Put these K friends in a specific friend group and recommend them to target users.