CN106529233A - A return visit-paying user identification algorithm based on browser fingerprint differences - Google Patents

Abstract

The invention relates to a return visit-paying user identification algorithm based on browser fingerprint differences. The method comprises the steps of collecting fingerprint feature parameters and making a fingerprint by using a Hash function; judging whether the fingerprint exists in a fingerprint database, judging the fingerprint as the fingerprint of a return visit-paying user if yes, and performing the next step if no; calculating the global difference values between the fingerprint and all fingerprints in the fingerprint database one by one; if one global difference value is less than a threshold value, determining that the fingerprint is an upgraded fingerprint of a return visit-paying user, and updating the current fingerprint corresponding to the global difference value in the fingerprint database to the upgrade fingerprint; otherwise, determining that the fingerprint is the fingerprint of a new user and storing the fingerprint into the fingerprint database. The identification rate of the return visit-paying user identification algorithm is increased and the misjudgment rate of identification is reduced.

Description

A kind of return visit user identification arithmetic based on browser fingerprint diversity factor

Technical field

The present invention relates to a kind of return visit user identification arithmetic based on browser fingerprint diversity factor.

Background technology

Browser fingerprint is mainly used in the fields such as user tracking, authentication, user's identification.Browser fingerprint is a kind of base In the recognition methods of browser information combination, by browser platform from layers such as configuration information, software composition and hardware compositions The characteristic parameters such as the screen message of equipment, plugin information, font information being got on secondary, and combines formation can be used with unique identification The finger print data at family.Wherein, the fingerprint characteristic parameter of collection is the intrinsic status information of user, rather than the privacy letter of user Breath.

Following achievement in research is achieved with regard to browser fingerprint technique at present：Peter Eckersley et al. propose one kind With the intimate user identification methods of cookie, referred to as browser fingerprint, and by experimental verification browser fingerprint tool There is higher discrimination.It is right that Nikiforakis et al. has been carried out to the characteristic parameter acquisition modes in three big business fingerprint algorithms Than analysis, it is found that the function of limiting Flash and JavaScript can effectively suppress fingerprint to generate, and then it is clear to propose interference The method that device fingerprint of looking at is generated.It is excellent that Keaton Movwery et al. propose a kind of API and Javascript for combining HTML5 The fingerprint identification method of gesture, realizes the fingerprint recognition based on HTML 5, and wherein most typically HTML5 fingerprint mode refers to for canvas Line.Other this technology of JavaScript engine fingerprint, due to different browsers JavaScript engine with trickle Difference, therefore fingerprint is made using small difference.In prior art also by character string between successive appraximation fingerprint phase Like spending, in certain similar range, then it is assumed that fingerprint is the renewal fingerprint for paying a return visit user.

It is comprehensive it, contemporary literature mainly studies the side that the operation principle of different types of browser fingerprint and interference fingerprint are generated Method, causes the mode of fingerprint static matching recognize completely after rarer document concern characteristic parameter change and pays a return visit asking for user Topic.

The content of the invention

In view of this, it is an object of the invention to provide a kind of return visit user's identification based on browser fingerprint diversity factor is calculated Method, compares the diversity factor between browser fingerprint by 8 fingerprint characteristic parameters, efficiently identifies the return visit user of website.

For achieving the above object, the present invention is adopted the following technical scheme that：A kind of return visit based on browser fingerprint diversity factor User identification arithmetic, it is characterised in that comprise the following steps：

Step S1：Collection fingerprint characteristic parameter simultaneously makes fingerprint using hash function；

Step S2：Judge that the fingerprint whether there is in fingerprint base, if existing, the fingerprint is judged as the finger of user is paid a return visit Line, otherwise carries out step S3；

Step S3：The global disparity angle value of the fingerprint and all fingerprints in fingerprint base is calculated one by one；

Step S4：If there is a certain global disparity angle value less than threshold value, the fingerprint is judged as the upgrading of user is paid a return visit Fingerprint, and existing fingerprint corresponding with the global disparity angle value in fingerprint base is updated to into the upgrading fingerprint；Otherwise judge institute The fingerprint that fingerprint is new user is stated, and the fingerprint is stored in fingerprint base.

Further, the computational methods of the global disparity angle value are as follows：

Wherein, D (BF₁,BF₂) represent fingerprint BF₁With fingerprint BF₂Global disparity angle value；d(attr_i(BF₁,BF₂)) represent Characteristic parameter diversity factor, when i takes 1 to 8, represents User agent diversity factoies, plugin information diversity factor, font information difference respectively Degree, screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies, video diversity factoies and Cookie diversity factoies； W_attriThe weight of each characteristic parameter diversity factor is represented, comentropy of its value for character pair parameter.

Further, the User agent diversity factoies are by browser related data diversity factor and device architecture difference Degree two parts composition：

d(attr₁(BF₁,BF₂))=0.5 × F_br+0.5×F_archi

Wherein, F_brFor browser related data diversity factor, F_archiFor device architecture diversity factor,For referring to Line BF₁Browser title,For fingerprint BF₂Browser title,For fingerprint BF₁Browser version Number,For fingerprint BF₂Browser version number；d(attr₁(BF₁,BF₂)) it is fingerprint BF₁With fingerprint BF₂User Agent diversity factoies.

Further, the circular of the plugin information diversity factor is as follows：Make fingerprint BF₁Plug-in unit list be LP (BF₁), fingerprint BF₂Plug-in unit list be LP (BF₂), then：

FU=| (LP (BF₁)\(LP(BF₁)∩LP(BF₂)))∪LP(BF₂)|

Wherein, FU is fingerprint BF₁With fingerprint BF₂Two plug-in unit lists in plug-in unit sum；

For fingerprint BF₁With fingerprint BF₂, the ratio of distinctive plug-in unit in two plug-in unit lists is calculated, its result uses F respectively₁With F₂Represent, solution procedure is as follows：

Wherein, LP_{=name ,=ver}The plugin name plug-in unit number consistent with plug-in version number in indication plug unit list LP, LP_{=name, ≠ ver}In indication plug unit list LP, plugin name is identical, the different plug-in unit number of plug-in version number；

Calculate LP (BF₁) and LP (BF₂) plugin name is identical in two plug-in unit lists, the different plug-in unit ratio of plug-in version number And use F₃Represent, solution procedure is as follows：

Calculate LP (BF₁) and LP (BF₂) plugin name and version number are the same from plug-in unit ratio and use table in two plug-in unit lists Show F₄, solution procedure is as follows：

For fingerprint BF₁With fingerprint BF₂, its plugin information diversity factor d (attr₂(BF₁,BF₂)) solution it is as follows：

Further, the circular of the font information diversity factor is as follows：Make fingerprint BF₁List of fonts be LF (BF₁), fingerprint BF₂List of fonts be LF (BF₂), then：

FU=| (LF (BF₁)\(LF(BF₁)∩LF(BF₂)))∪LF(BF₂)|

Wherein, FU fingerprints BF₁With fingerprint BF₂Two list of fonts in font sum；

List of fonts LF (BF₁) and list of fonts LF (BF₂) in the ratio of distinctive font use F respectively₁And F₂Represent, ask Solution preocess is as follows：

List of fonts LF (BF₁) and list of fonts LF (BF₂) in same font ratio F₃Represent, solution procedure is such as Under：

For fingerprint BF₁With fingerprint BF₂, its font information diversity factor d (attr₃(BF₁,BF₂)) solution it is as follows：

Further, the screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies, video differences The circular of degree and Cookie diversity factoies is as follows：

Wherein, x is natural number and 4≤x≤8；As x=4, d (attr₄(BF₁,BF₂)) represent screen resolution difference Degree, attr₄(BF₁) represent fingerprint BF₁The screen resolution of middle characteristic parameter, attr₄(BF₂) represent fingerprint BF₂Screen Resolution ratio；As x=5, d (attr₅(BF₁,BF₂)) represent time zones differences degree, attr₅(BF₁) represent fingerprint BF₁Time zone ginseng Number, attr₅(BF₂) represent fingerprint BF₂Time zone parameter；Work as x=6, d (attr₆(BF₁,BF₂)), represent HTTP Accept differences Degree, attr₆(BF₁) represent fingerprint BF₁HTTP Accept parameters, attr₆(BF₂) represent fingerprint BF₂HTTP Accept ginseng Number；As x=7, d (attr₇(BF₁,BF₂)) represent video diversity factoies, attr₇(BF₁) represent fingerprint BF₁Video parameters, attr₇(BF₂) represent fingerprint BF₂Video parameters；As x=8, d (attr₈(BF₁,BF₂)) Cookie diversity factoies are represented, attr₈(BF₁) represent fingerprint BF₁Cookie parameters, attr₈(BF₂) fingerprint BF₂Cookie parameters

Further, the computational methods of the comentropy of the characteristic parameter are as follows：A browser fingerprint algorithm is made to be BF (), gives a new browser information x, generates browser fingerprint BF (x), and its discrete probability density function is P (f_n),n∈[0,1,...,N]；Then for characteristic parameter a, its comentropy is：

Wherein, f_n=BF (x) represents a browser fingerprint.

Further, the threshold value is 0.025,0.05 or 0.1.

The present invention is had the advantages that compared with prior art：The present invention first defines 8 fingerprint characteristic parameters Local diversity factor computational methods, weighted sum obtain weigh two fingerprints between diversity factor formula, it is then, complete by what is tried to achieve Office's difference angle value is contrasted with threshold value set in advance, and then judges whether user is to pay a return visit user；Return visit can not only be improved The discrimination of user identification arithmetic, while the False Rate for recognizing can also be reduced.

Description of the drawings

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

Fig. 2 is the fingerprint recognition rate under difference diversity factor threshold value of the invention.

Fig. 3 is the fingerprint False Rate under difference diversity factor threshold value of the invention.

Specific embodiment

Below in conjunction with the accompanying drawings and embodiment the present invention will be further described.

Fig. 1 is refer to, the present invention provides a kind of return visit user identification arithmetic based on browser fingerprint diversity factor, its feature It is to comprise the following steps：

Step S1：Collection fingerprint characteristic parameter simultaneously makes fingerprint using hash function；

Step S2：Judge that the fingerprint whether there is in fingerprint base, if existing, the fingerprint is judged as the finger of user is paid a return visit Line, otherwise carries out step S3；

Step S3：The global disparity angle value of the fingerprint and all fingerprints in fingerprint base is calculated one by one；

Step S4：If there is a certain global disparity angle value less than threshold value (threshold value is 0.025,0.05 or 0.1), sentence The fixed fingerprint is the upgrading fingerprint for paying a return visit user, and existing fingerprint corresponding with the global disparity angle value in fingerprint base is updated For the upgrading fingerprint；Otherwise judge fingerprint of the fingerprint as new user, and the fingerprint is stored in fingerprint base.

Further, the computational methods of the global disparity angle value are as follows：

Wherein, D (BF₁,BF₂) represent fingerprint BF₁With fingerprint BF₂Global disparity angle value, interval be [0,1], if referring to Line BF₁With fingerprint BF₂Completely the same, functional value is 0, and if both are entirely different, functional value is 1；d(attr_i(BF₁,BF₂)) table Show characteristic parameter diversity factor, when i takes 1 to 8, represent User agent diversity factoies, plugin information diversity factor, font information respectively poor Different degree, screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies and Cookie diversity factoies；Represent each The weight of characteristic parameter diversity factor, comentropy of its value for character pair parameter.

Further, the User agent diversity factoies are by browser related data diversity factor and device architecture difference Degree two parts composition：

d(attr₁(BF₁,BF₂))=0.5 × F_br+0.5×F_archi

Wherein, F_brFor browser related data diversity factor, F_archiFor device architecture diversity factor,For referring to Line BF₁Browser title,For fingerprint BF₂Browser title,For fingerprint BF₁Browser version Number,For fingerprint BF₂Browser version number；d(attr₁(BF₁,BF₂)) it is fingerprint BF₁With fingerprint BF₂User Agent diversity factoies；Fingerprint BF₁With fingerprint BF₂Browser title and during completely the same version number, F_brValue is 0；If browser Title is identical and version number is inconsistent, F_brThen value is 0.125；If both are entirely different, value is 1；F_archiValue principle With F_brIt is identical.

Diversity factor function d (attr₁(BF₁,BF₂)) interval be [0,1].If fingerprint BF₁With fingerprint BF₂User Agent is identical, and functional value is then 0, if the User agent of two fingerprints are entirely different, functional value is then 1.

Further, the circular of the plugin information diversity factor is as follows：Make fingerprint BF₁Plug-in unit list be LP (BF₁), fingerprint BF₂Plug-in unit list be LP (BF₂), then：

FU=| (LP (BF₁)\(LP(BF₁)∩LP(BF₂)))∪LP(BF₂)|

Wherein, FU is fingerprint BF₁With fingerprint BF₂Two plug-in unit lists in plug-in unit sum；

For fingerprint BF₁With fingerprint BF₂, the ratio of distinctive plug-in unit in two plug-in unit lists is calculated, its result uses F respectively₁With F₂Represent, solution procedure is as follows：

Wherein, LP_{=name ,=ver}The plugin name plug-in unit number consistent with plug-in version number in indication plug unit list LP, LP_{=name, ≠ ver}In indication plug unit list LP, plugin name is identical, the different plug-in unit number of plug-in version number；

Calculate LP (BF₁) and LP (BF₂) plugin name is identical in two plug-in unit lists, the different plug-in unit ratio of plug-in version number And use F₃Represent, solution procedure is as follows：

Calculate LP (BF₁) and LP (BF₂) plugin name and version number are the same from plug-in unit ratio and use table in two plug-in unit lists Show F₄, solution procedure is as follows：

For fingerprint BF₁With fingerprint BF₂, its plugin information diversity factor d (attr₂(BF₁,BF₂)) solution it is as follows：

Diversity factor function d (attr₂(BF₁,BF₂)) interval be [0,1].If fingerprint BF₁With BF₂Plug-in unit list it is complete Identical, difference angle value is then 0, if the plugin information of two fingerprints is entirely different, difference angle value is then 1.

Further, the circular of the font information diversity factor is as follows：Make fingerprint BF₁List of fonts be LF (BF₁), fingerprint BF₂List of fonts be LF (BF₂), then：

FU=| (LF (BF₁)\(LF(BF₁)∩LF(BF₂)))∪LF(BF₂)|

Wherein, FU fingerprints BF₁With fingerprint BF₂Two list of fonts in font sum；

List of fonts LF (BF₁) and list of fonts LF (BF₂) in the ratio of distinctive font use F respectively₁And F₂Represent, ask Solution preocess is as follows：

List of fonts LF (BF₁) and list of fonts LF (BF₂) in same font ratio F₃Represent, solution procedure is such as Under：

For fingerprint BF₁With fingerprint BF₂, its font information diversity factor d (attr₃(BF₁,BF₂)) solution it is as follows：

Diversity factor function d (attr₃(BF₁),attr₃(BF₂)) interval be [0,1].

Further, the screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies, video differences The circular of degree and Cookie diversity factoies is as follows：

Wherein, x is natural number and 4≤x≤8；As x=4, d (attr₄(BF₁,BF₂)) represent screen resolution difference Degree, attr₄(BF₁) represent fingerprint BF₁The screen resolution of middle characteristic parameter, attr₄(BF₂) represent fingerprint BF₂Screen Resolution ratio；As x=5, d (attr₅(BF₁,BF₂)) represent time zones differences degree, attr₅(BF₁) represent fingerprint BF₁Time zone ginseng Number, attr₅(BF₂) represent fingerprint BF₂Time zone parameter；Work as x=6, d (attr₆(BF₁,BF₂)), represent HTTP between fingerprint Accept diversity factoies, attr₆(BF₁) represent fingerprint BF₁HTTP Accept parameters, attr₆(BF₂) represent fingerprint BF₂HTTP Accept parameters；As x=7, d (attr₇(BF₁,BF₂)) represent video diversity factoies between fingerprint, attr₇(BF₁) represent fingerprint BF₁Video parameters, attr₇(BF₂) represent fingerprint BF₂Video parameters；As x=8, d (attr₈(BF₁,BF₂)) represent Cookie diversity factoies, attr₈(BF₁) represent fingerprint BF₁Cookie parameters, attr₈(BF₂) fingerprint BF₂Cookie parameters.Difference Different degree function d (attr_x(BF₁,BF₂)) interval be [0,1]

Further, the computational methods of the comentropy of the characteristic parameter are as follows：A browser fingerprint algorithm is made to be BF (), gives a new browser information x, generates browser fingerprint BF (x), and the discrete probabilistic of the browser fingerprint is close Degree function is P (f_n),n∈[0,1,...,N]；First, the concept of self-information I is introduced, which is defined as follows shown：

I (BF (x)=f_n)=- log (P (f_n))

Wherein, self-information amount I characterizes bit number of the browser fingerprint comprising information.

P(f_n) comentropy H (BF) for browser fingerprint self-information amount desired value, shown in which is defined as follows：

Wherein, the value of H (BF) is bigger, and the accuracy for distinguishing different browsers is higher.

Browser fingerprint is combined by different characteristic parameters, the information content of each characteristic parameter is individually discussed and is determined The comentropy of adopted fingerprint characteristic parameter.If certain fingerprint characteristic parameter is a, its self-information amount is with comentropy computational methods respectively such as Shown in lower.

I(f_n,a)=- log (P (f_n,a))

The characteristic parameter separate for two, self-information amount is calculated can be according to the direct linear, additive of formula.

In order to allow those skilled in the art to be better understood from technical scheme, enter below in conjunction with three values of threshold value Row is further illustrated.

In Fig. 2, what three curves were represented is diversity factor threshold value difference value when being 0.025,0.05 and 0.1, and fingerprint is known Not rate changes over situation.Understand that the time interval that user accesses is longer by experimental result, the discrimination for paying a return visit user is lower, The reason for producing the result is that the amplitude of variation of fingerprint characteristic parameter increases as user pays a return visit interval time growth.Wherein, The setting of diversity factor threshold value affects the discrimination for paying a return visit user.The fingerprint diversity factor threshold value of setting is bigger, then it represents that allow Difference degree between fingerprint is bigger, that is, the accuracy for paying a return visit user's identification is lower.When fingerprint diversity factor threshold value is set to When 0.05, the discrimination of user is paid a return visit when time interval is one month up to more than 86%.

False Rate is to assess another index of the algorithm performs effect in this experiment.Wherein, what False Rate was represented is to return The probability of the wrong identification occurred during visiting user's identification.The setting of diversity factor threshold value can not only affect to pay a return visit the identification of user Rate, can also affect False Rate.Wherein, diversity factor threshold value arranges larger, can be effectively reduced the False Rate of user's identification, but The accuracy rate of identification can also be decreased.And diversity factor threshold value arranges less than normal, the discrimination of user can be effectively improved, but together When also increase False Rate.For, under the different diversity factor threshold condition in 0.025,0.05 and 0.1 3 kind, statistics pays a return visit user The False Rate of identification, its experimental result are as shown in Figure 3.

Fig. 3 shows, under three kinds of different diversity factor threshold conditions, in one month, pays a return visit the False Rate of user's identification Increase with time interval and increase, but the False Rate of identification is all below 7%.

In sum, diversity factor threshold value setting and fingerprint diversity factor calculate be pay a return visit user identification arithmetic in it is most important Two links.Select appropriate diversity factor threshold value improve the discrimination for paying a return visit user identification arithmetic, while can also Reduce the False Rate of identification.

The foregoing is only presently preferred embodiments of the present invention, all impartial changes done according to scope of the present invention patent with Modification, should all belong to the covering scope of the present invention.

Claims (8)

1. a kind of return visit user identification arithmetic based on browser fingerprint diversity factor, it is characterised in that comprise the following steps：

Step S1：Collection fingerprint characteristic parameter simultaneously makes fingerprint using hash function；

Step S2：Judge that the fingerprint whether there is in fingerprint base, if existing, the fingerprint is judged as the fingerprint of user is paid a return visit, Step S3 is carried out otherwise；

Step S3：The global disparity angle value of the fingerprint and all fingerprints in fingerprint base is calculated one by one；

Step S4：If there is a certain global disparity angle value less than threshold value, the fingerprint is judged as the upgrading fingerprint of user is paid a return visit, And existing fingerprint corresponding with the global disparity angle value in fingerprint base is updated to into the upgrading fingerprint；Otherwise judge the fingerprint For the fingerprint of new user, and the fingerprint is stored in fingerprint base.

2. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 1, it is characterised in that：Institute The computational methods for stating global disparity angle value are as follows：

$D({\mathrm{BF}}_1,{\mathrm{BF}}_2)=\frac{\underset{i=1}{\overset{8}{\Σ}}{W}_{{\mathrm{attr}}_i}\×d\left({\mathrm{attr}}_i\right({\mathrm{BF}}_1,{\mathrm{BF}}_2\left)\right)}{\underset{i=1}{\overset{8}{\Σ}}{W}_{{\mathrm{attr}}_i}}$

Wherein, D (BF₁,BF₂) represent fingerprint BF₁With fingerprint BF₂Global disparity angle value；d(attr_i(BF₁,BF₂)) represent feature Parameter differences degree, when i takes 1 to 8, represent respectively User agent diversity factoies, plugin information diversity factor, font information diversity factor, Screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies, video diversity factoies and Cookie diversity factoies；Table Show the weight of each characteristic parameter diversity factor, comentropy of its value for character pair parameter.

3. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute State User agent diversity factoies to be made up of browser related data diversity factor and device architecture diversity factor two parts：

$F_{archi}=\left\{\begin{array}{cc}0& {\mathrm{archi}}_{{\mathrm{BF}}_1}={\mathrm{archi}}_{{\mathrm{BF}}_2}\\ 1& {\mathrm{archi}}_{{\mathrm{BF}}_1}\≠{\mathrm{archi}}_{{\mathrm{BF}}_2}\end{array}\right.$

d(attr₁(BF₁,BF₂))=0.5 × F_br+0.5×F_archi

Wherein, F_brFor browser related data diversity factor, F_archiFor device architecture diversity factor,For fingerprint BF₁ Browser title,For fingerprint BF₂Browser title,For fingerprint BF₁Browser version number,For fingerprint BF₂Browser version number；d(attr₁(BF₁,BF₂)) it is fingerprint BF₁With fingerprint BF₂User agent Diversity factor.

4. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute The circular for stating plugin information diversity factor is as follows：Make fingerprint BF₁Plug-in unit list be LP (BF₁), fingerprint BF₂Plug-in unit List is LP (BF₂), then：

FU=| (LP (BF₁)\(LP(BF₁)∩LP(BF₂)))∪LP(BF₂)|

Wherein, FU is fingerprint BF₁With fingerprint BF₂Two plug-in unit lists in plug-in unit sum；

For fingerprint BF₁With fingerprint BF₂, the ratio of distinctive plug-in unit in two plug-in unit lists is calculated, its result uses F respectively₁And F₂Table Show, solution procedure is as follows：

$F_1=\frac{\left|\left(LP\right({\mathrm{BF}}_1)\backslash ({\mathrm{LP}}_{=name,=ver}\∪{\mathrm{LP}}_{=name,\≠ver}\left)\right)\right|}{FU}$

$F_2=\frac{\left|\left(LP\right({\mathrm{BF}}_2\left)N\right({\mathrm{LP}}_{=name,=ver}\∪{\mathrm{LP}}_{=name,\≠ver}\left)\right)\right|}{FU}$

Wherein, LP_{=name ,=ver}The plugin name plug-in unit number consistent with plug-in version number, LP in indication plug unit list LP_{=name, ≠ ver} In indication plug unit list LP, plugin name is identical, the different plug-in unit number of plug-in version number；

Calculate LP (BF₁) and LP (BF₂) plugin name is identical in two plug-in unit lists, the different plug-in unit ratio of plug-in version number is used in combination F₃Represent, solution procedure is as follows：

$F_3=\frac{\left|{\mathrm{LP}}_{=name,\≠ver}\right|}{FU}$

Calculate LP (BF₁) and LP (BF₂) plugin name and version number are the same from plug-in unit ratio and with representing in two plug-in unit lists F₄, solution procedure is as follows：

$F_4=\frac{\left|{\mathrm{LP}}_{=name,=ver}\right|}{FU}$

For fingerprint BF₁With fingerprint BF₂, its plugin information diversity factor d (attr₂(BF₁,BF₂)) solution it is as follows：

$d\left({\mathrm{attr}}_2\right({\mathrm{BF}}_1,{\mathrm{BF}}_2\left)\right)=\frac{F_1+F_2-0.75\×F_3-F_4+1}{2}.$

5. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute The circular for stating font information diversity factor is as follows：Make fingerprint BF₁List of fonts be LF (BF₁), fingerprint BF₂Font List is LF (BF₂), then：

FU=| (LF (BF₁)\(LF(BF₁)∩LF(BF₂)))∪LF(BF₂)|

Wherein, FU fingerprints BF₁With fingerprint BF₂Two list of fonts in font sum；

List of fonts LF (BF₁) and list of fonts LF (BF₂) in the ratio of distinctive font use F respectively₁And F₂Represent, solution procedure It is as follows：

$F_1=\frac{\left|\left(LF\right({\mathrm{BF}}_1)\backslash (LF\left({\mathrm{BF}}_1\right)\∩LF\left({\mathrm{BF}}_2\right)\left)\right)\right|}{FU}$

$F_2=\frac{\left|\left(LF\right({\mathrm{BF}}_2)\backslash (LF\left({\mathrm{BF}}_2\right)\∩LF\left({\mathrm{BF}}_2\right)\left)\right)\right|}{FU}$

List of fonts LF (BF₁) and list of fonts LF (BF₂) in same font ratio F₃Represent, solution procedure is as follows：

$F_3=\frac{|\left(LF\right({\mathrm{BF}}_1\left)\right)\∩LF\left({\mathrm{BF}}_2\right)|}{FU}$

For fingerprint BF₁With fingerprint BF₂, its font information diversity factor d (attr₃(BF₁,BF₂)) solution it is as follows：

$d\left({\mathrm{attr}}_3\right({\mathrm{BF}}_1,{\mathrm{BF}}_2\left)\right)=\frac{F_1+F_2-F_3+1}{2}.$

6. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute State the tool of screen resolution diversity factor, time zones differences degree, HTTP Accept diversity factoies, video diversity factoies and Cookie diversity factoies Body computational methods are as follows：

$d\left({\mathrm{attr}}_x\right({\mathrm{BF}}_1,{\mathrm{BF}}_2\left)\right)=\left\{\begin{array}{cc}0& {\mathrm{attr}}_x\left({\mathrm{BF}}_1\right)={\mathrm{attr}}_x\left({\mathrm{BF}}_2\right)\\ 1& {\mathrm{attr}}_x\left({\mathrm{BF}}_2\right)\≠{\mathrm{attr}}_x\left({\mathrm{BF}}_2\right)\end{array}\right.$

Wherein, x is natural number and 4≤x≤8；As x=4, d (attr₄(BF₁,BF₂)) screen resolution diversity factor is represented, attr₄(BF₁) represent fingerprint BF₁The screen resolution of middle characteristic parameter, attr₄(BF₂) represent fingerprint BF₂Screen differentiate Rate；As x=5, d (attr₅(BF₁,BF₂)) represent time zones differences degree, attr₅(BF₁) represent fingerprint BF₁Time zone parameter, attr₅(BF₂) represent fingerprint BF₂Time zone parameter；Work as x=6, d (attr₆(BF₁,BF₂)), HTTP Accept diversity factoies are represented, attr₆(BF₁) represent fingerprint BF₁HTTP Accept parameters, attr₆(BF₂) represent fingerprint BF₂HTTP Accept parameters； As x=7, d (attr₇(BF₁,BF₂)) represent video diversity factoies, attr₇(BF₁) represent fingerprint BF₁Video parameters, attr₇(BF₂) represent fingerprint BF₂Video parameters；As x=8, d (attr₈(BF₁,BF₂)) Cookie diversity factoies are represented, attr₈(BF₁) represent fingerprint BF₁Cookie parameters, attr₈(BF₂) fingerprint BF₂Cookie parameters.

7. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute The computational methods for stating the comentropy of characteristic parameter are as follows：A browser fingerprint algorithm is made for BF (), give one it is new clear Look at device information x, generate browser fingerprint BF (x), its discrete probability density function is P (f_n),n∈[0,1,...,N]；Then For characteristic parameter a seeks comentropy, its comentropy is：

$H\left({\mathrm{BF}}_a\right)=-\underset{n=1}{\overset{N}{\Σ}}P\left(f_{n,a}\right)log\left(f_{n,a}\right)$

Wherein, f_n=BF (x) represents a browser fingerprint.

8. the return visit user identification arithmetic based on browser fingerprint diversity factor according to claim 2, it is characterised in that：Institute Threshold value is stated for 0.025,0.05 or 0.1.