CN106611353A - Audience obtaining method and server equipment - Google Patents

Abstract

Embodiments of the invention provide an audience obtaining method and server equipment. The audience obtaining method comprises the steps of obtaining entity information of advertised goods and user offline behaviors to form offline information, obtaining online browsing behaviors of users to form online information, and storing the offline information and the online information in a user behavior data warehouse; obtaining first offline information and first online information of a first user specific to at least one advertised goods from the user behavior data warehouse; performing normalization on the first offline information and the first online information; and working out interest index of the first user on the at least one advertised goods in a current sampling period according to the normalized data. Compared with the conventional Internet-based advertisement audience positioning technology, the audience obtaining method provided by the invention is suitable for operators and also can be fused with the offline commercial system to form all-around excavating of a single content source, so as to realize transformation of the operators.

Description

A kind of method and server apparatus for obtaining audient

Technical field

The present invention relates to Internet technology, particularly relates to the method and server apparatus of a kind of acquisition audient.

Background technology

The earning model of the Internet includes：Game, ecommerce and advertisement.Internet advertising and conventional ads Maximum attraction be exactly precision marketing.By various audient's directional technologies (Audience Targeting), analysis Numerous labels of the attribute (User Profile) of user, digging user interest or demand, push most worthy, The advertisement that user needs most.

From audient's location technology distinguishing Internet advertisement, being not concerned with the Internet advertising of audient's positioning includes：Bar Width advertisement (Banner Ad), mail direct marketing advertisement (EDM, Email Direct Marketing), FLASH Advertisement (Rich Media Ad), video ads (Video Ad) etc.；Search advertisements (Search Ad), such as google, Baidu etc. is oriented according to the search word of user, then throws in the advertisement of correlation；Contextual advertisement (Contextual Advertising), according to the current webpage of user, it is assumed that the point of interest of user is with current net Page content is related, then pushes the advertisement related to webpage context；Social advertisement (Social Ad), It is comprehensive intimate at social platform (wechat, microblogging etc.) with the rise of social networkies (Social Network) Relation, context, historical information etc. orient advertisement, and carry out customization marketing activity.

Existing audient's location technology as shown in figure 1, including：User's history is collected according to means such as cookie Information；According to user history information and scene information, an advertisement Candidate Set for being adapted to this people is found；Knot Close characteristic of advertisement and expectation, sort inside advertisement Candidate Set, if find best, value highest one or Dry advertisement.

There are problems in audient's location technology of existing Internet advertising.Scene information is not complete：Present receives Many location technologies, completely using information on line, cause that associated scenario on line, such as electronics business can only be confined to Business, search, social activity etc., but do not associate with scene, the position of reality, time under user's line, therefore nothing Method realizes that " user has arrived meal point in a certain layer of shopping mall, and system is actively customized together to him The O2O advertisement forms such as one layer or in walking several minutes distance, certain taste restaurant ".It is existing Similar functions, such as popular comment and Baidu map etc. are required for user to initiate demand, just can know that concrete position Put, if indoor, Internet firm cannot position floor.User history information is not complete：Present receives Many location technologies, completely using information on line, do not merge where user has gone down online, where stop Wait so long information more, make audient be positioned with limitation.Cannot with other, present commercial system merge：The Internet is wide A part for only whole commercial system is accused, because present audient's location technology is completely using on line Information, has isolated commercial system under the lines such as lamp box advertisement, solid shop/brick and mortar store, LED advertisements in reality.It is existing Technology relies on content source excavation substantially, and this is unfavorable to operator：User under line is with the network linking of operator Application on line, Internet firm, by existing audient's location technology, is user's advertisement, is obtained greatest Income, and operator is only into real pipeline.

How to excavate by pipeline, highlight operator O＆O (online＆offline) scene and control power, hold and arrive Advantages for development under end data and line, realize that pipeline value is cashed, and become operator's problem demanding prompt solution.

The content of the invention

The technical problem to be solved in the present invention is to provide a kind of method and server apparatus of acquisition audient, is used for Solve the problems, such as that the audient in existing audient's location technology is positioned with limitation, excavate by pipeline and found for advertisement Suitable audient.

To solve above-mentioned technical problem, embodiments of the invention provide a kind of method for obtaining audient, method bag Include：Obtain behavior under the entity information and user's line of promotion item and form information under line, obtain the online of user Navigation patterns form information on line, and information on information under the line and line is stored in into user behavior data warehouse In；First user is obtained for believing under at least one of the user behavior data warehouse First Line of promotion item Information on breath and First Line；Information on information under the First Line and the First Line is normalized；Root According to the data after normalization the first user is calculated in current sample period at least one advertiser The interest index of product.

In described method, the behavior under the entity information and user's line of promotion item that obtains forms information under line, Obtain and also include before information on the behavior formation line that surfs the web of user：Build wireless access control server Log server, collect wireless access control server log, according to wireless access control server log Obtain the track of behavior under user's line；Build the topological relation that WLAN webmasters collect wlan network；Take The distributed downlink deep packet inspection device of single wireless access control server is built, user is restored and is surfed the web Target web, obtain user's each application name for using of online.

In described method, the behavior under the entity information and user's line of promotion item that obtains forms packet under line Include：Each access is formed according to the track of behavior, the topological relation of wlan network under user's line Information under the line of node AP, sets up information labels stock under a line and puts information under the line；Obtain user Surf the web behavior formed line on information include：According to the target web and application name, will be every The behavior of surfing the web of position user is encoded into digital label, the logical order of digital label be classification, source, Classification and details, form information on line；Set up information labels stock on a line and put information on the line.

In described method, information on information under the line and line is stored in into user behavior data warehouse Zhong Bao Include：According to user with time point by information under the line with information fusion on line into behavioral data, by the row It is stored in user behavior data warehouse for data；Information includes number of clicks on line, packet under line on line Include and reach under line stay time under number of times and line；Obtain first user to be directed in user behavior data warehouse extremely Information includes on information and First Line under the First Line of a few promotion item：First user is obtained to particular row To reach stay time under number of times and line under number of clicks, line on the line of data.

In described method, information on information under the First Line and the First Line is normalized including： χ₁, χ₂, χ₃..., χ_nFor in the sampling period, n names user is respectively to point on the line of specific behavior data Hit number of times；δ₁, δ₂, δ₃..., δ_nFor in the sampling period, n names user is respectively to specific behavior data Line under reach number of times；φ₁, φ₂, φ₃..., φ_nFor in the sampling period, n names user is respectively to specific Number of times of the stay time more than minute threshold value m under the line of behavioral data；MAX₁For in the sampling period, n The maximum of number of clicks on name user's center line；MAX₂In the sampling period, to arrive under n name user's center lines Up to the maximum of number of times；MAX₃For in the sampling period, stay time is more than minute under n name user's center lines The number of times maximum of threshold value m；Using min-max standardized models, min=0 is taken, in the sampling period First user, to clicking on index λ on specific behavior data, normalized lines₁, index λ is reached under line₂With Index λ is stopped under line₃RespectivelyWith

In described method, the first user is calculated in current sample period according to the data after normalization In the interest index of at least one promotion item is included：Calculate interest of the first user in the i-th sampling period Index θ_i=η × λ₁+β×λ₂+γ×λ₃, η, β, γ are coefficient；Calculate the note in early stage x sampling period The time degeneration factor recalled；The interest index that first user is then calculated in current i-th sampling period isX is represented and review forward x sampling period.

A kind of server apparatus for obtaining audient, including：Behavioral data unit, for obtaining promotion item Behavior under entity information and user's line forms information under line, and the behavior that surfs the web for obtaining user forms letter on line Breath, information on information under the line and line is stored in user behavior data warehouse；Information index unit, Under for obtaining First Line of the first user at least one of user behavior data warehouse promotion item Information in information and First Line；Normalization unit, for information under the First Line and the First Line Information is normalized；

Interest index unit, for calculating the first user in present sample according to the data after normalization To the interest index of at least one promotion item in cycle.

Described server apparatus, behavioral data unit includes：Information labels library module under line, for basis Information under the line of clear and definite each accessed node AP in track of behavior under user's line, under setting up a line Information labels stock puts information under the line；Information labels library module on line, for according to the target web And application name, the behavior of surfing the web of every user is encoded into into digital label, digital label is patrolled It is classification, source, classification and details to collect order, forms information on line；Set up information mark on a line Sign stock and put information on the line.

Described server apparatus, behavioral data unit includes：Behavioral data unit includes data warehouse module, Data warehouse module is used for information under the line with information fusion on line into behavioral data, by the behavior number According to being stored in user behavior data warehouse；Information includes number of clicks on line on line, and information includes line under line Stay time under lower arrival number of times, line；Single user behavior acquisition module, for obtaining first user to specific Stay time under number of times, line is reached on the line of behavioral data under number of clicks, line.

Described server apparatus, normalization unit includes：Module is realized in normalization, for adopting min-max Standardized model, takes min=0, and the first user in the sampling period is to specific behavior data, normalized Index λ is clicked on line₁, index λ is reached under line₂With stop index λ under line₃Respectively χ₁, χ₂, χ₃..., χ_nFor in the sampling period, n name users Respectively to number of clicks on the line of specific behavior data；δ₁, δ₂, δ₃..., δ_nFor in the sampling period, N names user under the line of specific behavior data respectively to reaching number of times；φ₁, φ₂, φ₃..., φ_nFor sampling In cycle, n names user is more than respectively the number of times of minute threshold value m to stay time under the line of specific behavior data； MAX₁For in the sampling period, the maximum of number of clicks on n name user's center lines；MAX₂For sampling week In phase, the maximum of number of times is reached under n name user's center lines；MAX₃For in the sampling period, in n name users Number of times maximum of the stay time more than minute threshold value m under line.

Described server apparatus, interest index unit includes：Monocycle interest index module, for calculating Interest index θ of the first user in the i-th sampling period_i=η × λ₁+β×λ₂+γ×λ₃, η, β, γ are coefficient； Early stage time degeneration factor module, for calculating the time degeneration factor of the memory in early stage x sampling period； Interest Index for Calculation module, be for calculating interest index of the first user in current i-th sampling periodX is represented and review forward x sampling period.

The above-mentioned technical proposal of the present invention has the beneficial effect that：Compare traditional Internet advertising audient positioning Technology, is not only more suitable for operator, moreover it is possible to merge with commercial system under line, is formed and single content source is excavated Comprehensive compacting, realize operator make the transition.

Description of the drawings

Fig. 1 represents the operation principle schematic diagram of existing audient's directional technology；

Fig. 2 represents a kind of schematic flow sheet of the method for acquisition audient；

Fig. 3 represents the schematic diagram of the application scenarios for obtaining audient.

Specific embodiment

To make the technical problem to be solved in the present invention, technical scheme and advantage clearer, below in conjunction with attached Figure and specific embodiment are described in detail.

A set of zero hardware cost of embodiment of the present invention offer, practicality are wide, audient's location technology of carrier-class, Merge O＆O scenes, end-to-end data comprehensively, the user interest for excavating in real time simultaneously quantifies it.

The embodiment of the present invention provides a kind of method for obtaining audient, as shown in Fig. 2 including：

Step 201, the behavior under the entity information and user's line of promotion item that obtains forms information under line, obtains The behavior that surfs the web of user forms information on line, and information on information under the line and line is stored in into user's row For in data warehouse；

Step 202, obtains first user at least one of user behavior data warehouse promotion item Information on information and First Line under First Line；

Step 203, is normalized to information on information under the First Line and the First Line；

Step 204, calculates the first user right in current sample period according to the data after normalization The interest index of at least one promotion item.

Using technology provided in an embodiment of the present invention, comprehensively fusion O＆O scenes, end-to-end data, realize The real-time excavation of user interest simultaneously quantifies, be a set of brand-new, zero hardware cost, practicality is wide, fortune Audient's location technology of battalion's business's level.

Wherein, interest index refers to interest of the user to promotion item；Interest index is in just with the size of interest Correlation, numerical value is high then to represent that interest is big.

In a preferred embodiment, as shown in figure 3, step 101 obtain promotion item entity information and Behavior under user's line forms information under line, obtains and also wrapped before information on the behavior formation line that surfs the web of user Include：Gather each data source.Specifically, hardware net is built, three data sources is gathered in hardware net, That is, the log server of wireless access control server is built, wireless access control server log is collected, The track of behavior under user's line is obtained according to wireless access control server log；

Build topological relation, resource information and KPI Key Performance Indicator that WLAN webmasters collect wlan network；

The distributed downlink deep packet inspection device of single wireless access control server is built, is restored on user The target web that net is browsed, obtains each application name that user's online is used.The step for it is concrete Including：

Step 11, builds collection wireless access control server (AC, Wireless Access Point Controller) the log server of daily record, collects AC daily records, and these AC daily records are used to complete collection use Family track：Syslog functions are all set on all AC, on log server two ends, by AC daily record realities When transmit to log server, and be stored on log server.User trajectory is each of user's experience AP。

Step 12, builds WLAN webmasters, collect the topological relation of wlan network, resource information and KPI Key Performance Indicator (KPI, Key Performance Indicator)；MIB of the WLAN webmasters from equipment manufacturer Storehouse, gathers the apparatus for transmitting signal (AP, Access Point) of WLAN and the topological relation of AC, record Resource information and alteration.

Step 13, builds the distributed downlink deep-packet detection (DPI, Deep Packet Inspection) of single AC Equipment, in the downlink traffic mouth of AC, carries out DPI, restores each webpage of user's online, obtains Each APP title that user's online is used.Here obtain be user online content, such as AC Daily record provides the route map of your driving, provided herein is that the seen landscape of driving.

In a preferred embodiment, also include：According to track, the WLAN of behavior under user's line The topological relation of network, or even resource information and KPI Key Performance Indicator KPI form each accessed node AP Line under information, set up information labels stock under a line and put information under the line；According to the target web And application name, the behavior of surfing the web of every user is encoded into into digital label, digital label is patrolled It is classification, source, classification and details to collect order, and according to digital label information on line is formed；Set up one Information labels stock puts information on the line on individual line.As shown in figure 3, the step for specifically include：

Step 21, sets up information labels storehouse under line, special counter, shop, advertisement position under each AP is covered, The information such as feature, are encoded into digital label, the logical order of digital label be project-building-floor-special counter/ Shop/advertisement position, specifies the entity information of promotion item under the line of each AP；

For example, certain user goes to 4 building, silver-colored Thailand Building A, under residing in Younger special counter, the information quilt Logic participle is " building (14) of (B)-Yin Tai (01)-Building A (02)-4-Younger special counter (58) under line ", therefore this The digital label of information is B01021458.

Step 22, sets up information labels storehouse on line, by every user browse including webpage, FTP, EMAIL, APP etc. is encoded into digital label in interior internet content, the logical order of digital label be classification-source-point Class-details, according to information on the line of digital label each internet content of clearly every user.

For example, first user skims through the shop of Younger by Taobao；The internet content is by logic point Word is (A)-electric business (00)-Taobao (01)-clothing (89)-Younger (32) on line, therefore, information on the line Digital label is A00018932.

In a preferred embodiment, information on information under the line and line is stored in into user behavior data storehouse Storehouse includes：According to user and time point by information under the line with information fusion on line into behavioral data, will The behavioral data is stored in user behavior data warehouse；Information includes number of clicks on line on line, under line Information includes reaching stay time under number of times and line under line；

First user is obtained under at least one of the user behavior data warehouse First Line of promotion item Information includes in information and First Line：First user is obtained to number of clicks, line on the line of specific behavior data Stay time under lower arrival number of times and line.

As shown in figure 3, including step 3, according to the time point that behavior under user and user's line occurs, by line Information in lower information and line, is fused into the O＆O user behavior datas warehouse with user as axle.

For example, -9 point of 20 minutes January 1 day 9 point in 2015 28 minutes, user X 4 building, silver-colored Thailand Building A, 8 minutes (digital label is B01021458) is resided in Younger special counter, during this period, user X exists 9 points 10 seconds 22 minutes also by Taobao browsed Younger shop-digital labels be A00018932,9 Point 22 minutes and 35 seconds, user X also completes other operations by Taobao APP.

In a preferred embodiment, first user is obtained at least one of user behavior data warehouse Information includes on information and First Line under the First Line of promotion item：

Obtain first user stop under number of times, line to reach under number of clicks, line on the line of specific behavior data Duration.In the embodiment of the present invention, for convenience of describing, by information under line with information fusion on line into behavioral data, The behavioral data of some promotion item is then the specific behavior data of the promotion item, therefore, when user exists When information on the line of some promotion item is clicked on line, clicked on the line to some specific behavior data, When online lower arrival accesses some promotion item to user, then under being the line to some specific behavior data Reach.

In an application scenarios, as shown in figure 3, specific flow process includes：

Step 41～step 43, obtain first user to specific behavior data-include certain brand, certain content, Line on stay time under number of times and line is reached under number of clicks, line；

Step 44～step 45, obtains full dose user to point on associated user's number of the specific behavior data, line Hit total degree, line and be issued under total degree, line the information such as total duration that stops, these information can be used for Technology provided in an embodiment of the present invention is upgraded afterwards, it is improved during use.

In an application scenarios, " the 8th He of digital label on line during needs assessment Younger, is filtered out 9 is 32 " and " under line the 8th and 9 of digital label be 58 " user's experience, obtain first and use Family and full dose user, to reaching the letter such as stay time under number of times, line under number of clicks, line on the line of Younger Breath.

In a preferred embodiment, using min-max standardization, the normalization of three information indexes is realized. As shown in figure 3, this process is specifically included：

χ₁, χ₂, χ₃..., χ_nFor in the sampling period, n names user is respectively to the line of specific behavior data Upper number of clicks；

δ₁, δ₂, δ₃..., δ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Reach number of times；

φ₁, φ₂, φ₃..., φ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Number of times of the stay time more than minute threshold value m；

MAX₁For in the sampling period, the maximum of number of clicks on n name user's center lines；

MAX₂In the sampling period, the maximum of number of times to be reached under n name user's center lines；

MAX₃For in the sampling period, under n name user's center lines stay time more than minute threshold value m number of times most Big value；

Step 51～step 53, with reference to min-max standardization, take min=0, it is to avoid refers to during number of times=min Count the problem for 0, the first user (a) in the sampling period, to specific behavior data, normalized lines Upper click index λ₁, index λ is reached under line₂, index λ is stopped under line₃Respectively With

In a preferred embodiment, as shown in figure 3, calculating first user according to the data after normalization Step 6 is included to the interest index of the first behavioral data in current sample period：

In the interest index in the i-th sampling period, η, β, γ are coefficient to first user；

The interest of user is decided by the x sampling period before the information acquisition on the same day, and the i-th sampling period Since memory, therefore the memory in early stage x sampling period also needs to consideration time degeneration factor；

Then interest index of the first user in current i-th sampling period isX is represented forward Review x sampling period.

The embodiment of the present invention provides a kind of server apparatus for obtaining audient, including：

Behavioral data unit, forms for behavior under the entity information and user's line that obtain promotion item and believe under line Breath, the behavior that surfs the web for obtaining user forms information on line, and information on information under the line and line is deposited In user behavior data warehouse；

Information index unit, it is wide at least one of user behavior data warehouse for obtaining first user Accuse under the First Line of commodity information on information and First Line；

Normalization unit, for being normalized to information on information under the First Line and the First Line；

Interest index unit, for calculating the first user in present sample according to the data after normalization To the interest index of at least one promotion item in cycle.

Traditional Internet advertising audient's location technology is compared, is not only more suitable for operator, moreover it is possible to business under line Industry system merges, and forms the comprehensive compacting excavated to single content source, realizes that operator makes the transition.

In a preferred embodiment, behavioral data unit includes：

Information labels library module under line, for according to clear and definite each access in the track of behavior under user's line Information under the line of node AP, sets up information labels stock under a line and puts information under the line；

Information labels library module on line, for according to the target web and application name, will per use The behavior of surfing the web at family is encoded into digital label, and the logical order of digital label is classification, source, classification And details, form information on line；Set up information labels stock on a line and put information on the line.

In a preferred embodiment, behavioral data unit includes：

Behavioral data unit includes data warehouse module, and data warehouse module is used for information and line under the line Upper information fusion is stored in the behavioral data in user behavior data warehouse into behavioral data；Believe on line Breath includes number of clicks on line, and information includes reaching stay time under number of times, line under line under line；

Single user behavior acquisition module, for obtain number of clicks on line of the first user to specific behavior data, Stay time under number of times and line is reached under line.

In a preferred embodiment, normalization unit includes：

Module is realized in normalization, for using min-max standardized models, min=0 being taken, in the sampling period First user, on specific behavior data, normalized lines click on index λ₁, index λ is reached under line₂ With stop index λ under line₃RespectivelyWithχ₁, χ₂, χ₃..., χ_nFor in the sampling period, n names user is respectively to number of clicks on the line of specific behavior data；

δ₁, δ₂, δ₃..., δ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Reach number of times；

φ₁, φ₂, φ₃..., φ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Number of times of the stay time more than minute threshold value m；

MAX₁For in the sampling period, the maximum of number of clicks on n name user's center lines；

MAX₂In the sampling period, the maximum of number of times to be reached under n name user's center lines；

MAX₃For in the sampling period, under n name user's center lines stay time more than minute threshold value m number of times most Big value.

In a preferred embodiment, interest index unit includes：

Monocycle interest index module, for calculating interest index of the first user in the i-th sampling period θ_i=η × λ₁+β×λ₂+γ×λ₃, η, β, γ are coefficient；

Early stage time degeneration factor module, degenerates for calculating the time of memory in early stage x sampling period Factor；

Interest Index for Calculation module, for calculating interest index of the first user in current i-th sampling period ForX is represented and review forward x sampling period.

It is using the advantage after this programme：By the content clicked on user's line and the position of scene under line, arrive Blend up to time and the time of staying, so, when a certain layer of user (shopping mall) at the mall When, system can be at table in the time period actively push in same layer or walking several minutes distance to user, The O2O advertisements such as restaurant of certain taste；Compared with traditional Internet advertising audient's location technology, the present invention Embodiment has merged where user has gone down online, where stops information of waiting so long more, extends the complete of analysis Face property, is not only more suitable for operator, moreover it is possible to merge with commercial system under line, is formed and single content source is excavated Comprehensive compacting, realize operator make the transition.

The above is the preferred embodiment of the present invention, it is noted that for the common skill of the art For art personnel, on the premise of without departing from principle of the present invention, some improvements and modifications can also be made, These improvements and modifications also should be regarded as protection scope of the present invention.

Claims (11)

1. it is a kind of obtain audient method, it is characterised in that include：

Obtain behavior under the entity information and user's line of promotion item and form information under line, obtain the online of user Navigation patterns form information on line, and information on information under the line and line is stored in into user behavior data warehouse In；

First user is obtained under at least one of the user behavior data warehouse First Line of promotion item Information in information and First Line；

Information on information under the First Line and the First Line is normalized；

According to the data after normalization calculate the first user in current sample period to it is described at least The interest index of one promotion item.

2. method according to claim 1, it is characterised in that obtain the entity information of promotion item Information under line is formed with behavior under user's line, is obtained and gone back before information on the behavior formation line that surfs the web of user Including：

The log server of wireless access control server is built, wireless access control server log is collected, The track of behavior under user's line is obtained according to wireless access control server log；

Build the topological relation that WLAN webmasters collect wlan network；

The distributed downlink deep packet inspection device of single wireless access control server is built, is restored on user The target web that net is browsed, obtains each application name that user's online is used.

3. method according to claim 2, it is characterised in that

Behavior under the entity information and user's line of acquisition promotion item forms information under line and includes：According to the use The track of behavior under the line of family, the topological relation of wlan network are formed believes under the line of each accessed node AP Breath, sets up information labels stock under a line and puts information under the line；

Obtaining information on the behavior formation line that surfs the web of user includes：According to the target web and apply journey Sequence title, by the behavior of surfing the web of every user the digital label with logical order is encoded into, according to institute State digital label and form information on line；Set up information labels stock on a line and put information on the line.

4. method according to claim 1, it is characterised in that

Information on information under the line and line is stored in into user behavior data warehouse includes：According to user and Time point by information under the line with information fusion on line into behavioral data, the behavioral data is stored in into use In the behavioral data warehouse of family；Information includes number of clicks on line on line, and information includes reaching number of times under line under line With stay time under line；

First user is obtained under at least one of the user behavior data warehouse First Line of promotion item Information includes in information and First Line：First user is obtained to number of clicks, line on the line of specific behavior data Stay time under lower arrival number of times and line.

5. method according to claim 4, it is characterised in that to information under the First Line and institute State information on First Line be normalized including：

χ₁, χ₂, χ₃... ..., χ_nFor in the sampling period, n names user is respectively to the line of specific behavior data Upper number of clicks；

δ₁, δ₂, δ₃... ..., δ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Reach number of times；

φ₁, φ₂, φ₃... ..., φ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Number of times of the stay time more than minute threshold value m；

MAX₁For in the sampling period, the maximum of number of clicks on n name user's center lines；

MAX₂In the sampling period, the maximum of number of times to be reached under n name user's center lines；

MAX₃For in the sampling period, under n name user's center lines stay time more than minute threshold value m number of times most Big value；

Using min-max standardized models, min=0 is taken, the first user in the sampling period is to specific behavior Index λ is clicked on data, normalized line₁, index λ is reached under line₂With stop index λ under line₃Respectively For ${\mathrm{\λ}}_1=\frac{{\mathrm{\χ}}_a}{{\mathrm{MAX}}_1},{\mathrm{\λ}}_2=\frac{{\mathrm{\δ}}_a}{{\mathrm{MAX}}_2}$ With ${\mathrm{\λ}}_3=\frac{{\mathrm{\φ}}_a}{{\mathrm{MAX}}_3}.$

6. method according to claim 5, it is characterised in that calculated according to the data after normalization Go out the first user includes in current sample period to the interest index of at least one promotion item：

Calculate interest index θ of the first user in current i-th sampling period_i=η × λ₁+β×λ₂+γ×λ₃, η, β, γ are coefficient；

Calculate the time degeneration factor of the memory in early stage x sampling period；

The interest index that first user is then calculated in current i-th sampling period isX tables Show and review forward x sampling period.

7. it is a kind of obtain audient server apparatus, it is characterised in that include：

Behavioral data unit, forms for behavior under the entity information and user's line that obtain promotion item and believe under line Breath, the behavior that surfs the web for obtaining user forms information on line, and information on information under the line and line is deposited In user behavior data warehouse；

Information index unit, it is wide at least one of user behavior data warehouse for obtaining first user Accuse under the First Line of commodity information on information and First Line；

Normalization unit, for being normalized to information on information under the First Line and the First Line；

Interest index unit, for calculating the first user in present sample according to the data after normalization To the interest index of at least one promotion item in cycle.

8. server apparatus according to claim 7, it is characterised in that behavioral data unit includes：

Information labels library module under line, for according to clear and definite each access in the track of behavior under user's line Information under the line of node AP, sets up information labels stock under a line and puts information under the line；

Information labels library module on line, for according to the target web and application name, will per use The behavior of surfing the web at family is encoded into digital label, and the logical order of digital label is classification, source, classification And details, form information on line；Set up information labels stock on a line and put information on the line.

9. server apparatus according to claim 7, it is characterised in that behavioral data unit includes：

Behavioral data unit includes data warehouse module, and data warehouse module is used for will according to user and time point The behavioral data is stored in user behavior number by information and information fusion on line into behavioral data under the line According in warehouse；Information includes number of clicks on line on line, and information includes being reached under line and stops under number of times and line under line Stay duration；

Single user behavior acquisition module, for obtain number of clicks on line of the first user to specific behavior data, Stay time under number of times and line is reached under line.

10. server apparatus according to claim 7, it is characterised in that normalization unit includes：

Module is realized in normalization, for using min-max standardized models, min=0 being taken, in the sampling period First user, on specific behavior data, normalized lines click on index λ₁, index λ is reached under line₂ With stop index λ under line₃Respectively ${\mathrm{\λ}}_1=\frac{{\mathrm{\χ}}_a}{{\mathrm{MAX}}_1},{\mathrm{\λ}}_2=\frac{{\mathrm{\δ}}_a}{{\mathrm{MAX}}_2}$ With ${\mathrm{\λ}}_3=\frac{{\mathrm{\φ}}_a}{{\mathrm{MAX}}_3};$ χ₁, χ₂, χ₃... ..., χ_nFor in the sampling period, n names user is respectively to number of clicks on the line of specific behavior data；

δ₁, δ₂, δ₃... ..., δ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Reach number of times；

φ₁, φ₂, φ₃... ..., φ_nFor in the sampling period, n names user is respectively under the line of specific behavior data Number of times of the stay time more than minute threshold value m；

MAX₁For in the sampling period, the maximum of number of clicks on n name user's center lines；

MAX₂In the sampling period, the maximum of number of times to be reached under n name user's center lines；

MAX₃For in the sampling period, under n name user's center lines stay time more than minute threshold value m number of times most Big value.

11. server apparatus according to claim 7, it is characterised in that interest index unit includes：

Monocycle interest index module, for calculating interest index of the first user in current i-th sampling period θ_i=η × λ₁+β×λ₂+γ×λ₃, η, β, γ are coefficient；

Early stage time degeneration factor module, degenerates for calculating the time of memory in early stage x sampling period Factor；

Interest Index for Calculation module, for calculating interest index of the first user in current i-th sampling period ForX is represented and review forward x sampling period.