WO2011076070A1

WO2011076070A1 - Method and system for recommending network information

Info

Publication number: WO2011076070A1
Application number: PCT/CN2010/079789
Authority: WO
Inventors: 陈敏; 梁柱; 张博; 郑志昊
Original assignee: 腾讯科技（深圳）有限公司
Priority date: 2009-12-25
Filing date: 2010-12-14
Publication date: 2011-06-30
Also published as: CN102110098B; CN102110098A

Abstract

A method and a system for recommending network information are provided. The method comprises: obtaining the set of attributes of a network information during a specified period of time; performing a weighting process on the attribute values of said set of attributes to obtain the total attribute value of the network information during the specified time; then recommending, by means of a friends list, either the network information whose total attribute value is not less than a set attribute threshold value, or a preset amount of network information selected from the total of attribute values arranged in descending order of magnitude. The method and the system for recommending network information can improve the accuracy of recommending specific network information.

Description

Network information recommendation method and system

Technical field

The present invention relates to the field of mass information processing and data mining, and more particularly to a network information recommendation method and system.

Background technique

With the continuous popularization of the network, the network has become a necessary part of people's lives, and the network can provide users with a variety of information. However, with the continuous development of the network, the amount of network information has exploded, and users need to face a large amount of network information every day. A large part of the network information is often information that people are interested in. This information is concerned by a large number of people, also known as "hot spot information." How to help users get hot information has always been the focus of attention.

At present, hot search information is usually extracted by using a search method or a recommendation method. The search method is an active information extraction method, which searches for hotspot information through a search engine; the recommended method is to obtain hotspot information and then push it to the user. However, since the attribute of the hotspot information is difficult to obtain personally, the search method is difficult for the user to actively provide information effective for a certain user, and the recommendation method pushes the blindness greatly, and the information garbage is easily harassed.

technical problem

Based on this, it is necessary to provide a network information recommendation method, which can improve the accuracy of recommending specific network information. In addition, it is also necessary to provide a network information recommendation system that can improve the accuracy of recommending specific network information.

Technical solution

A method for recommending network information, the method comprising: acquiring an attribute set of network information in a specified time period; performing weighting processing on the attribute set to obtain a total attribute value of network information in a specified time; The network information whose value is not less than the set attribute threshold, or the preset number of network information is selected in descending order of the total attribute value, and the recommendation is performed through the friend relationship chain.

The method can also include the step of setting an attribute threshold for network information over a specified time period.

The step of setting the attribute threshold of the network information in the specified time period may be specifically: setting the frequency of the network information to be launched; randomly extracting a part of the users, and obtaining network information samples published by all the friends of the part of the users in the specified time period. a set of attributes; determining a single attribute threshold of the attribute set of the network information sample according to the frequency of the network information; and obtaining an attribute threshold of the network information in the specified time period according to the single attribute threshold and the attribute weight of the network information.

The step of determining a single attribute threshold may be: acquiring a single attribute value of the attribute set of the network information sample; establishing a two-dimensional graph of the single attribute value and the number of network information samples; determining according to the two-dimensional graph A single attribute threshold.

The step of setting an attribute threshold of the network information within the specified time period may further include: acquiring an attribute weight of the network information within the specified time period according to the single attribute threshold.

The step of obtaining the total attribute value of the network information may be: obtaining a single attribute value of the attribute set; and calculating a total attribute value of the network information in the specified time according to the single attribute value and the attribute weight.

In addition, it is also necessary to provide a network information recommendation system that can improve the accuracy of recommending specific network information.

A network information recommendation system, the system comprising: a network information database, storing network information; an attribute extraction module, obtaining an attribute set of network information within a specified time period from the network information database; an attribute value processing module, The attribute set performs attribute value weighting processing to obtain the total attribute value of the network information in the specified time period; the information recommendation module sets the network information whose total attribute value is not less than the set attribute threshold, or according to the total attribute value from large to small. The order selects a preset number of network information and recommends through the friend relationship chain.

The system may further include: a threshold setting module, configured to set a frequency of ejecting network information; the attribute extraction module is further configured to randomly extract a part of users from the network information database, and obtain all friends of the part of the users in a specified time period. a set of attributes of the network information sample that is published; the threshold setting module is further configured to determine a single attribute threshold of the attribute set of the network information sample according to the network information push frequency, and according to the single attribute threshold and the attribute of the network information The weight obtains the attribute threshold of the network information within the specified time period.

The threshold setting module is further configured to obtain a single attribute value of the attribute set of the network information sample, establish a two-dimensional graph of the single attribute value and the number of network information samples, and determine a single attribute threshold according to the two-dimensional graph.

The threshold setting module is further configured to acquire attribute weights of network information within a specified time period according to a single attribute threshold.

The attribute value processing module is further configured to obtain a single attribute value of the attribute set, and calculate a total attribute value of the network information in the specified time period according to the single attribute value and the attribute weight.

Beneficial effect

The above-mentioned network information recommendation system and method obtains the total attribute value by performing attribute value weighting processing on the attribute set of the network information, and the network information whose total attribute value is not less than the set attribute threshold is recommended through the friend relationship chain. The network information recommended to the user's friends through the friend relationship chain more accurately conforms to the content of interest of the user's friends, and improves the accuracy of recommending specific network information.

In addition, by randomly extracting a part of users, the network information published by all the friends of the part of the user in the specified time period is obtained as a network information sample, and the attribute threshold of the network information is set, and the network information is filtered according to the attribute threshold, and the network information is filtered out. The information is spread in the friendship relationship chain, so that users can more accurately obtain the network information that they are interested in, and satisfy the user's experience requirements.

DRAWINGS

1 is a flow chart of a method for recommending network information in an embodiment;

2 is a flow chart of a method for obtaining a total attribute value of network information in a specified time period in an embodiment;

3 is a flow chart of a method for setting a total attribute threshold of network information in an embodiment;

4 is a schematic structural diagram of a network information recommendation system in an embodiment.

Embodiments of the invention

FIG. 1 shows a flow of a network information recommendation method in an embodiment, and the specific process of the method flow is as follows:

In step S10, an attribute set of network information within a specified time period is acquired. Network information can be obtained from the network information database, including a series of massive information such as web logs, network photos, and news. Each network information object has its own attributes, such as the title of the information, the creator of the information, the amount of information accessed, the amount of reply, the amount of reload, and the time at which the information is generated. The set of these attributes is called the network. A collection of attributes for the message. Since the information in a certain period of time is more valuable to the user, for example, the network information within one week is more valuable than the network information within one month or one year, the specified time period can be set in advance as needed. In one embodiment, attribute analysis is performed on network information within a specified time period to obtain a list of attributes of the network information.

In step S20, the attribute set is subjected to attribute weighting processing to obtain a total attribute value of the network information within the specified time. In an embodiment, as shown in FIG. 2, the specific process of obtaining the total attribute value of the network information in a specified time is:

In step S202, a single attribute value of the attribute set is obtained. A single attribute value refers to a single attribute component of the network information in the specified time period. For example, for a network log in a specified time period, a single attribute value includes the number of accesses of the network log, the number of replies, and the number of reprints.

In step S204, the total attribute value of the network information in the specified time is calculated according to the single attribute value and the attribute weight. In one embodiment, each attribute in the attribute list of the network information is respectively assigned a weight, which may be determined empirically or may be determined when the attribute threshold is set. For example, for the network log in the specified time period, the attribute set includes the access amount, the reply amount, and the reload amount of the network log, and the attribute weights of the access amount, the reply amount, and the reload amount respectively assigned to the network log are K1, K2, and K3, and satisfies K1+K2+K3=1. In one embodiment, the calculation formula of the total attribute value of the network information in the specified time period is: V=K1*P1+...+KN*PN Where K1 is the attribute value of attribute 1, KN is the attribute value of attribute N (eg, K1, K2, K3 are 0.4, 0.3, 0.3 in order), P1 is a single attribute value of attribute 1, and PN is a single attribute value of attribute N (For example, P1 represents 37 visits, P2 represents 5 hits, P3 represents 10 reloads, and N is the total number of attributes in the attribute set (for example, N is 3).

In step S30, the network information of the total attribute value is not less than the set attribute threshold, or the preset number of network information is selected in descending order of the total attribute value, and the recommendation is performed through the friend relationship chain. According to the calculated total attribute value of the network information in the specified time period, the total attribute value is compared with the size of the attribute threshold set in advance, and the network information whose total attribute value is not less than the set attribute threshold is hotspot information, that is, It is considered to be the network information that the user is interested in. The obtained hotspot information can be pushed to all friends of the creator of the network information through the friend relationship chain. The network information recommended to the user's friends can accurately match the content that the user's friends are interested in, thus improving the accuracy of recommending specific network information.

In an embodiment, the network information recommendation method further includes the step of setting an attribute threshold of the network information within the specified time period. FIG. 3 shows a method flow for setting an attribute threshold in an embodiment, and the specific process is as follows:

In step S100, the push frequency of the network information is set. The frequency at which network information is introduced is the amount of network information that needs to be pushed out to users in an average time. For example, if it is required to recommend 10 hotspot information to the user every day, the frequency of launching the hotspot information is: 10 pieces/day.

In step S200, a part of the users are randomly selected, and the attribute sets of the network information samples published by all the friends of the part of the users in the specified time period are obtained. In one embodiment, attribute analysis is performed on the obtained network information samples, thereby obtaining an attribute list of the network information samples. Using the network information published by the friends of some users randomly selected in the specified time period as a sample, the attribute threshold is set to filter the network information in the specified time period, and the filtered information is performed in the friend relationship chain. Spread, enabling users to more accurately obtain hotspot information of their own interest.

In step S300, a single attribute threshold of the attribute set of the network information sample is determined according to the push frequency of the network information. In one embodiment, the specific process of determining a single attribute threshold of the attribute set of the network information sample includes: (1) acquiring a single attribute value of the attribute set of the network information sample; (2) establishing a single attribute value and a network information sample number a two-dimensional graph; (2) determining a single attribute threshold based on the two-dimensional graph. In one embodiment, the M users are randomly selected, and all the friends of the M users are obtained a total of X web logs published within a specified time (for example, within one week). The attribute set of the network log includes the amount of access and the amount of the reply. , reload amount, etc. Obtaining a single attribute value (such as the amount of access) of the network log and making a two-dimensional graph with the number of network information samples corresponding to the single attribute value (such as the number of network log articles), according to the introduction of the two-dimensional graph and network information The frequency determines a single attribute threshold. For example, suppose that 10 hotspot information needs to be recommended to the user every day. For 1000 users, at least 10*1000 web logs are needed. According to the above two-dimensional graph, when a single attribute value is greater than or equal to P, 10*1000 articles can be obtained. For the above network log, P is the threshold of the single attribute value, that is, a single attribute threshold.

In step S400, the attribute threshold of the network information in the specified time period is obtained according to the single attribute threshold and the attribute weight of the network information.

After the single attribute threshold is determined, the attribute threshold of the network information in the specified time period can be calculated according to the attribute weight of the network information. For example, when the single attribute threshold of the access amount of the network log in the specified time period is P, the single attribute threshold of the reply quantity is L, the single attribute threshold of the reload amount is J, and the attribute weight of the access quantity is K1, The attribute weight of the response quantity is K2, and the attribute weight of the reload quantity is K3. The attribute threshold of the network information in the specified time period is the multiplication value of each individual attribute threshold and the attribute weight of the single attribute, that is, network information. The attribute threshold R = K1 * P = K2 * L = K3 * J. Since the sum of all attribute weights is 1, that is, K1+K2+K3=1, the attribute threshold R of the network information in the specified time period can be obtained according to the above quaternary equation.

In an embodiment, the network information is filtered according to the attribute threshold of the network information in the specified time period, and the network information whose total attribute value is not less than the attribute threshold is hot spot information, and when the hot spot information is determined to be greater than the set launch When the number of messages is set, for example, if N hotspot information is pushed every day, the first N pieces of the determined hotspot information can be pushed.

In an embodiment, the attribute weights of the network information within the specified time period may also be obtained according to the determined single attribute threshold. For example, the single attribute threshold of the access volume of the network log in the specified time period is P, the single attribute threshold of the reply quantity is L, and the single attribute threshold of the reload quantity is J, then the attribute of the network information in the specified time period The threshold is the multiplication of the individual attribute thresholds with the attribute weights of the single attribute. For example, if the attribute threshold of the network log in the specified time period is R, and the attribute weight of the access quantity is K1, the attribute weight of the reply quantity is K2, and the attribute weight of the reload quantity is K3, the attribute threshold of the network information R = K1 * P = K2 * L = K3 * J. Since the sum of all attribute weights is 1 (ie, K1+K2+K3=1), the attribute weights of a single attribute can be calculated. The obtained attribute weights can be used to calculate the total network attribute value in the specified time period.

FIG. 4 shows a network information push system in an embodiment, which includes a network information database 100, an attribute extraction module 200, an attribute value processing module 300, a threshold setting module 400, and an information recommendation module 500. among them:

The network information database 100 is configured to store network information, including network logs, network photos, news information, and the like, and different types of network information. The attribute extraction module 200 is configured to acquire, from the network information database 100, a set of attributes of network information in a specified time period; The attribute value processing module 300 is configured to perform attribute value weighting processing on the attribute set to obtain a total attribute value of the network information in the specified time period; the threshold setting module 400 is configured to set an attribute threshold value of the network information in the specified time period; The recommendation module 500 is configured to select network information whose total attribute value is not less than the set attribute threshold, or select a preset number of network information in descending order of the total attribute value, and perform recommendation through the friend relationship chain.

According to the calculated total attribute value of the network information in the specified time period, the total attribute value is compared with the size of the attribute threshold set in advance, and the network information whose total attribute value is not less than the set attribute threshold is hotspot information, that is, It is considered to be the network information that the user is interested in. The obtained hotspot information can be pushed to all friends of the creator of the network information through the friend relationship chain. The network information recommended to the user's friends can accurately match the content that the user's friends are interested in, thus improving the accuracy of recommending specific network information.

In one embodiment, the attribute value processing module 300 is further configured to obtain a single attribute value of the attribute set, and calculate a total attribute value of the network information within the specified time period according to the single attribute value and the attribute weight. In an embodiment, each attribute in the attribute list of the network information may be respectively assigned a weight, which may be determined empirically, or may be determined when the attribute threshold is set, and the single attribute value refers to A single attribute component of the network information for the specified time period. In one embodiment, the calculation formula of the total attribute value of the network information in the specified time period is: V=K1*P1+...+KN*PN Where K1 is the attribute value of attribute 1, KN is the attribute value of attribute N, P1 is a single attribute value of attribute 1, PN is a single attribute value of attribute N, and N is the total number of attributes in the attribute set.

In an embodiment, the threshold setting module 400 is further configured to set a frequency of ejecting network information. The attribute extraction module 200 is further configured to randomly extract a part of users from the network information database 100, and obtain all friends of the part of the users at a specified time. a set of attributes of the network information sample published in the segment; the threshold setting module 400 is configured to determine a single attribute threshold of the attribute set of the network information sample according to the network information push frequency, and according to the single attribute threshold and the attribute weight of the network information Gets the attribute threshold of the network information for the specified time period. Using the network information published by the friends of some users randomly selected in the specified time period as a sample, the attribute threshold is set to filter the network information in the specified time period, and the filtered information is performed in the friend relationship chain. Spread, enabling users to more accurately obtain hotspot information of their own interest.

In an embodiment, the threshold setting module 400 is further configured to acquire a single attribute value of the attribute set of the network information sample, establish a two-dimensional graph of the single attribute value and the number of network information samples, and determine according to the two-dimensional graph. A single attribute threshold. After determining a single attribute threshold, the attribute threshold of the network information in the specified time period can be calculated. The network information is filtered according to the attribute threshold of the network information in the specified time period, and the network information whose total attribute value is not less than the attribute threshold is obtained as hotspot information.

In an embodiment, the threshold setting module 400 is further configured to acquire attribute weights of network information within a specified time period according to a single attribute threshold. As described above, the attribute weight of the network information can be set in advance by experience, or can be calculated in the process of setting the attribute threshold.

The above-mentioned embodiments are merely illustrative of several embodiments of the present invention, and the description thereof is more specific and detailed, but is not to be construed as limiting the scope of the invention. It should be noted that a number of variations and modifications may be made by those skilled in the art without departing from the spirit and scope of the invention. Therefore, the scope of the invention should be determined by the appended claims.

Claims

A network information recommendation method, the method comprising:

Get the set of attributes of the network information within the specified time period;

Performing attribute value weighting on the attribute set to obtain a total attribute value of network information within a specified time;

The network information with the total attribute value not less than the set attribute threshold is selected, or the preset number of network information is selected in descending order of the total attribute value, and the recommendation is performed through the friend relationship chain.
The network information recommendation method according to claim 1, wherein the method further comprises the step of setting an attribute threshold of the network information within the specified time period.
The network information recommendation method according to claim 2, wherein the step of setting an attribute threshold of the network information in the specified time period is specifically:

Set the frequency of launching network information;

Randomly extracting a part of users, and obtaining attribute sets of network information samples published by all friends of the part of the users in a specified time period;

Determining a single attribute threshold of the attribute set of the network information sample according to a frequency of elapse of the network information;

Obtaining an attribute threshold of the network information in the specified time period according to the single attribute threshold and the attribute weight of the network information.
The network information recommendation method according to claim 3, wherein the step of determining a single attribute threshold is specifically:

Obtaining a single attribute value of the attribute set of the network information sample;

Establishing a two-dimensional graph of the single attribute value and the number of network information samples;

A single attribute threshold is determined from the two-dimensional graph.
The network information recommendation method according to claim 3, wherein the step of setting an attribute threshold of the network information in the specified time period further comprises: Obtaining attribute weights of network information within a specified time period according to the single attribute threshold.
The network information recommendation method according to claim 1 or 5, wherein the step of calculating the total attribute value of the network information is specifically:

Get a single attribute value for a property collection;

Calculating a total attribute value of the network information in the specified time according to the single attribute value and the attribute weight.
A network information recommendation system, characterized in that the system comprises:

Network information database, storing network information;

An attribute extraction module, configured to acquire, from the network information database, a set of attributes of network information within a specified time period;

The attribute value processing module performs weighting processing on the attribute set to obtain a total attribute value of the network information in the specified time period;

The information recommendation module selects the network information whose total attribute value is not less than the set attribute threshold, or selects a preset number of network information in descending order of the total attribute value, and performs recommendation through the friend relationship chain.
The network information recommendation system according to claim 7, wherein the system further comprises:

a threshold setting module, configured to set a frequency of pushing network information;

The attribute extraction module is further configured to randomly extract a part of users from the network information database, and obtain an attribute set of network information samples published by all the friends of the part of the user in a specified time period; the threshold setting module is further configured to use the network according to the network. The information derivation frequency determines a single attribute threshold of the attribute set of the network information sample, and acquires an attribute threshold of the network information within the specified time period according to the single attribute threshold and the attribute weight of the network information.
The network information recommendation system according to claim 8, wherein the threshold setting module is further configured to acquire a single attribute value of the attribute set of the network information sample, and establish a single attribute value and a number of network information samples. A dimensionality diagram that determines a single attribute threshold based on the two-dimensional graph.
The network information recommendation system according to claim 8, wherein the threshold setting module is further configured to acquire attribute weights of network information within a specified time period according to a single attribute threshold.
The network information recommendation system according to claim 7 or 10, wherein the attribute value processing module is further configured to acquire a single attribute value of the attribute set, and calculate according to the single attribute value and the attribute weight The total attribute value of the network information for the specified time period.