CN104636947A - Calculation method of confidence degree of recommendation information in Internet of Things - Google Patents

Abstract

The invention relates to a calculation method of confidence degree of recommendation information in Internet of Things. The calculation method includes the steps of firstly, calculating the source confidence degree of the recommendation information RI; secondly, calculating the recommender confidence degree of the recommendation information RI; thirdly, calculating the word-of-mouth confidence degree of the recommendation information RI; fourthly, calculating the adoption confidence degree of the recommendation information RI; fifthly, acquiring the confidence degree of the recommendation information RI of the weighted sum of the confidence degrees. Compared with the prior art, the calculation method has the advantages that the recommendation information is evaluated and calculated from four different aspects of the source confidence degree, the recommender confidence degree, the word-of-mouth confidence degree and the adoption confidence degree, the confidence degree of the recommendation information in the Internet of Things is obtained by synthesizing the four aspects, and the method is fast, accurate, high in expansibility and capable of being effectively applied to fields such as E-commerce, electronic government affairs, public opinion monitoring and information services.

Description

The computing method of recommendation information degree of confidence in a kind of Internet of Things

Technical field

The present invention relates to field of computer technology, especially relate to the computing method of recommendation information degree of confidence in a kind of Internet of Things, be applicable to the fields such as ecommerce, E-Government, public sentiment monitoring and information service.

Background technology

In the Internet of Things full-blown epoch, people can obtain rich in natural resources and information from disparate networks.At this wherein, there is much information, such as merchandise news, information on services, public feelings information and news etc., all recommend by specific mechanism or personnel and obtain.When people face these recommendation informations time, owing to lacking enough cognitive methods and distinguishing rule, the content whether very difficult decision can accept and believe and use these recommendation informations to comprise.

In general, in environment of internet of things, a recommendation information includes following additional data: information source, nominator, evaluation situation (public praise) and passingly adopted situation.Under the prerequisite that cannot judge recommendation information content, whether reliably these addition records have just become judgement recommendation information important evidence.Reliable information source, reliably nominator, detailed favorable comment and passing repeatedly by the record adopted, all mean that this recommendation information is a trustworthy object: contrary, anonymous source, the nominator of prestige difference, difference are repeatedly commented and seldom by the record adopted, mean that the information that this is recommended is incredible.Based on above-mentioned consideration, in environment of internet of things, need for computing machine provides one automatically according to the relevant additional data of recommendation information and the method for passing record calculating assessment recommendation information degree of confidence, thus help internet-of-things terminal user to judge whether this recommendation information is worth believing, and used.

Summary of the invention

Object of the present invention is exactly provide the computing method of recommendation information degree of confidence in a kind of Internet of Things to overcome defect that above-mentioned prior art exists.

Object of the present invention can be achieved through the following technical solutions: the computing method of recommendation information degree of confidence in a kind of Internet of Things, is characterized in that, comprise the following steps:

(1) calculated recommendation information RI carry out source confidence T_source (RI);

(2) the nominator degree of confidence T_recommend (RI) of calculated recommendation information RI;

(3) the public praise degree of confidence T_praise (RI) of calculated recommendation information RI;

(4) calculated recommendation information RI adopt degree of confidence T_adopt (RI);

(5) the degree of confidence T (RI) of calculated recommendation information RI,

T(RI)=σ1×T_source(RI)+σ2×T_recommend(RI)

+σ3×T_praise(RI)+σ4×T_adopt(RI)

In formula, σ 1, σ 2, σ 3 and σ 4 is weight parameter, meets σ 1, σ 2, σ 3, σ 4 ∈ [0,1], σ 1+ σ 2+ σ 3+ σ 4=1.

The source confidence T_source (RI) that comes of the calculated recommendation information RI described in step (1) is specially,

$T\_\mathrm{source}\left(\mathrm{RI}\right)=\left\{\begin{array}{cc}{\mathrm{\β}}^{\frac{1}{n}}/2\×(r(\mathrm{RI}.\mathrm{source})+\frac{{\mathrm{\Σ}}_{i=1}^{n}T(\mathrm{RI}.\mathrm{source}.{\mathrm{past}}_i)}{n})& n\≠0\\ r(\mathrm{RI}.\mathrm{source})& n=0\end{array}\right.,$

In formula, RI.source is the source of recommendation information RI, and r (RI.source) ∈ [0,1] is the credit worthiness of this information source in Internet of Things, and this information source has issued n recommendation information in the past altogether, RI.source.past _irepresent the recommendation information issued for i-th time, T (RI.source.past _i) ∈ [0,1] expression recommendation information RI.source.past _idegree of confidence, β ∈ [0,1] is factor of influence.

The nominator degree of confidence T_recommend (RI) of the calculated recommendation information RI described in step (2) is specially,

$T\_\mathrm{recommend}\left(\mathrm{RI}\right)=\frac{{\mathrm{\Σ}}_{j=1}^{m}r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)}{m}\×{\mathrm{\β}}^{l/m},$

In formula, there is m nominator in recommendation information RI, RI.recommender _jrepresent a jth nominator, r (RI.recommender _j) ∈ [0,1] expression information recommendation person RI.recommender _jcredit worthiness, β ∈ [0,1] is factor of influence.

The public praise degree of confidence T_praise (RI) of the calculated recommendation information RI described in step (3) is specially,

$\begin{array}{c}T\_\mathrm{praise}\left(\mathrm{RI}\right)=\\ {\mathrm{\ω}}_1\×(\frac{{\mathrm{\Σ}}_{j=1}^m(r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)\×\mathrm{judgment}\left({\mathrm{RI}.\mathrm{recommender}}_j\right))}{{\mathrm{\Σ}}_{j=1}^{m}r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)}\×{\mathrm{\β}}^{p/m})\\ +{\mathrm{\ω}}_2\×(\frac{{\mathrm{\Σ}}_{k=1}^u(r\left({\mathrm{RI}.\mathrm{user}}_k\right)\×\mathrm{judgment}\left({\mathrm{RI}.\mathrm{user}}_k\right))}{{\mathrm{\Σ}}_{k=1}^{u}r\left({\mathrm{RI}.\mathrm{user}}_k\right)}\×{\mathrm{\β}}^{(1-q/u)})\end{array},$

In formula, judgment (RI.recommender _j) ∈ [0,1] expression information recommendation person RI.recommender _jto the public praise evaluation of estimate that recommendation information RI provides, when public praise evaluation of estimate is lower than public praise evaluation of estimate threshold value, represent that information recommendation person does not recommend other people to use, p represents the evaluation number of times not recommending other people to use;

Recommendation information RI receives u domestic consumer's public praise evaluation, RI.user _j, represent kGe domestic consumer, r (RI.user _j) ∈ [0,1] represents domestic consumer RI.user _jcredit worthiness, judgment (RI.user _j) ∈ [0,1] represents domestic consumer RI.user _jthe public praise evaluation of estimate provided, when public praise evaluation of estimate is higher than public praise evaluation of estimate threshold value, represent that deserving public praise is evaluated as favorable comment, q represents the favorable comment number of times in u domestic consumer's public praise is evaluated, ω ₁and ω ₂for weight parameter, meet ω ₁, ω ₂∈ [0,1], ω ₁+ ω ₂=1.

The degree of confidence T_adopt (RI) that adopts of the calculated recommendation information RI described in step (4) is specially,

$T\_\mathrm{adopt}\left(\mathrm{RI}\right)=\frac{{\mathrm{\Σ}}_{l=1}^{s}r\left({\mathrm{RI}.\mathrm{user}}_l\right)}{s}\×{\left(\frac{s}{y}\right)}^{l/s},$

In formula, recommendation information RI is recommended gives y user, adopts user s altogether of this recommendation information, RI.user _lrepresent l the user accepting this recommendation information, r (RI.user _l) ∈ [0,1] expression user RI.user _lcredit worthiness.

Compared with prior art, the present invention's first always source confidence, referrer's degree of confidence, public praise degree of confidence and adopt degree of confidence four different aspects evaluates calculation is carried out to recommendation information, comprehensively above-mentioned four aspects obtain the degree of confidence of recommendation information in Internet of Things again, there is the advantages such as quick, accurate and high scalability, effectively can be applied to the fields such as ecommerce, E-Government, public sentiment monitoring and information service.

Embodiment

Below in conjunction with specific embodiment, the present invention is described in detail.

Embodiment 1,

The information source of table 1 recommendation information RI1, nominator, public praise and passingly adopted situation

Illustrate: during the public praise of the present embodiment calculates, public praise evaluation of estimate thresholds is 0.5, the public praise evaluation of estimate that nominator provides lower than 0.5 time, not think to recommend everybody to use; When domestic consumer's public praise evaluation of estimate is greater than 0.5, think favorable comment.

According to above-mentioned parameter situation, can calculate as follows:

$T\_\mathrm{source}\left(\mathrm{RI}1\right)=\frac{{0.8}^{\frac{1}{3}}}{2}\×(0.95+\frac{0.85+0.9+0.9}{3})=0.85;$

$T\_\mathrm{recommend}\left(\mathrm{RI}1\right)={0.8}^{\frac{1}{4}}\×\left(\frac{0.9+0.6+0.8+0.5}{4}\right)=0.66;$

$\begin{array}{c}T\_\mathrm{praise}\left(\mathrm{RI}1\right)\\ =0.6\×(\frac{0.9\×0.8+0.6\×0.4+0.8\×0.8+0.5\×0.9}{0.9+0.8+0.6+0.5}\×{0.8}^{1/4})\\ +0.4\×(\frac{0.7\×0.8+0.8\×0.9+0.95\×1+0.85\×0.6+0.4\×0.5+0.8\×0.8}{0.8+0.9+1+0.6+0.5+0.8}\×{0.8}^{(1-5/6)})\\ =0.65\end{array};$

$T\_\mathrm{adopt}\left(\mathrm{RI}1\right)=\left(\frac{0.8+0.9+1+0.6+0.5+0.5+0.7+0.9}{4}\right)\×{(8/20)}^{1/8}=0.69;$

Thus the degree of confidence obtaining recommendation information RI1 is,

T(RI1)=σ1×T_source(RI1)+σ2×T_recommend(RI1)

+σ3×T_praise(RI1)+σ4×T_adopt(RI1)。

=0.25×(0.85+0.66+0.65+0.69)=0.713

Embodiment 2, the information source of table 2 recommendation information RI2, nominator, public praise and passingly adopted situation

Illustrate: during the public praise of the present embodiment calculates, public praise evaluation of estimate threshold value is 0.6, the public praise evaluation of estimate that nominator provides lower than 0.6 time, not think to recommend everybody to use; When domestic consumer's public praise evaluation of estimate is greater than 0.6, think favorable comment.

According to above-mentioned parameter situation, can calculate as follows:

$T\_\mathrm{source}\left(\mathrm{RI}2\right)=\frac{{0.9}^{\frac{1}{4}}}{2}\×(0.98+\frac{0.8+0.95+0.86+0.9}{4})=0.90;$

$T\_\mathrm{recommend}\left(\mathrm{RI}2\right)={0.9}^{\frac{1}{4}}\×\left(\frac{0.85+0.65+0.9+0.5+0.75}{5}\right)=071;$

$\begin{array}{c}T\_\mathrm{praise}\left(\mathrm{RI}2\right)\\ =0.5\×\left(\frac{0.9\×0.8+0.6\×0.4+0.8\×0.8+0.5\×0.9}{0.9+0.8+0.6+0.5}{\×0.9}^{1/4}\right)\\ +0.5\×(\frac{0.7\×0.8+0.8\×0.9+0.95\×1+0.85\×0.6+0.4\×0.5+0.8\×0.8}{0.8+0.9+1+0.6+0.5+0.8}\×{0.9}^{(1-5/6)})\\ =0.74\end{array},;$

$T\_\mathrm{adopt}\left(\mathrm{RI}2\right)=\left(\frac{0.8+0.9+1+0.6+0.5+0.8+0.7+0.9}{8}\right)\×{(8/30)}^{1/8}=0.66;$

Thus the degree of confidence obtaining recommendation information RI2 is,

T(RI2)=σ1×T_source(RI2)+σ2×T_recommend(RI2)

+σ3×T_praise(RI2)+σ4×T_adopt(RI2)。

=0.25×0.90+0.2×0.71+0.4×0.74+0.15×0.66=0.762

Embodiment 1 and 2 shows, the confidence evaluation that technology involved in the present invention may be used for recommendation information in Internet of Things calculates, and by calculating the confidence value obtained, user therefrom can select the higher information of degree of confidence and adopt.

The present invention can be widely used in, in the Internet of Things behaviors such as ecommerce, E-Government, public sentiment monitoring and information service, comprising the much information of commodity, service etc. for Calculation Estimation.Simultaneously, those of ordinary skill in the art will be appreciated that, above embodiment is only used to object of the present invention is described, and be not used as limitation of the invention, as long as in essential scope of the present invention, the change of the above embodiment, modification all will be dropped in the scope of claim of the present invention.

Claims (5)

1. the computing method of recommendation information degree of confidence in Internet of Things, is characterized in that, comprise the following steps:

(1) calculated recommendation information RI carry out source confidence T_source (RI);

(2) the nominator degree of confidence T_recommend (RI) of calculated recommendation information RI;

(3) the public praise degree of confidence T_praise (RI) of calculated recommendation information RI;

(4) calculated recommendation information RI adopt degree of confidence T_adopt (RI);

(5) the degree of confidence T (RI) of calculated recommendation information RI,

T(RI)=σl×T_source(RI)+σ2×T_recommend(RI)

+σ3×T_praise(RI)+σ4×T_adopt(RI)’

In formula, σ 1, σ 2, σ 3 and σ 4 is weight parameter, meets σ 1, σ 2, σ 3, σ 4 ∈ [0,1], σ 1+ σ 2+ σ 3+ σ 4=1.

2. the computing method of recommendation information degree of confidence in a kind of Internet of Things according to claim 1, it is characterized in that, the source confidence T_source (RI) that comes of the calculated recommendation information RI described in step (1) is specially,

$T\_\mathrm{source}\left(\mathrm{RI}\right)=\left\{\begin{array}{cc}{\mathrm{\β}}^{\frac{1}{n}}/2\×(r(\mathrm{RI}.\mathrm{source})+\frac{{\mathrm{\Σ}}_{i=1}^{n}T(\mathrm{RI}.\mathrm{source}.{\mathrm{past}}_i)}{n})& n\≠0\\ r(\mathrm{RI}.\mathrm{source})& n=0\end{array}\right.,$

In formula, RI.source is the source of recommendation information RI, and r (RI.source) ∈ [0,1] is the credit worthiness of this information source in Internet of Things, and this information source has issued n recommendation information in the past altogether, RI.source.past _irepresent the recommendation information issued for i-th time, T (RI.source.past _i) ∈ [0,1] expression recommendation information RI.source.past _idegree of confidence, β ∈ [0,1] is factor of influence.

3. the computing method of recommendation information degree of confidence in a kind of Internet of Things according to claim 1, it is characterized in that, the nominator degree of confidence T_recommend (RI) of the calculated recommendation information RI described in step (2) is specially,

$T\_\mathrm{recommend}\left(\mathrm{RI}\right)=\frac{{\mathrm{\Σ}}_{j=1}^{m}r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)}{m}\×{\mathrm{\β}}^{l/m},$

In formula, there is m nominator in recommendation information RI, RI.recommender _jrepresent a jth nominator, r (RI.recommender _j) ∈ [0,1] expression information recommendation person RI.recommender _jcredit worthiness, β ∈ [0,1] is factor of influence.

4. the computing method of recommendation information degree of confidence in a kind of Internet of Things according to claim 3, it is characterized in that, the public praise degree of confidence T_praise (RI) of the calculated recommendation information RI described in step (3) is specially, $\begin{array}{c}T\_\mathrm{praise}\left(\mathrm{RI}\right)=\\ {\mathrm{\ω}}_1\×(\frac{{\mathrm{\Σ}}_{j=1}^m(r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)\×\mathrm{jundgment}\left({\mathrm{RI}.\mathrm{recommender}}_j\right))}{{\mathrm{\Σ}}_{j=1}^{m}r\left({\mathrm{RI}.\mathrm{recommender}}_j\right)}\×{\mathrm{\β}}^{p/m})\\ +{\mathrm{\ω}}_2\×(\frac{{\mathrm{\Σ}}_{k=1}^u(r\left({\mathrm{RI}.\mathrm{user}}_k\right)\×\mathrm{judgment}\left({\mathrm{RI}.\mathrm{user}}_k\right))}{{\mathrm{\Σ}}_{k=1}^{u}r\left({\mathrm{RI}.\mathrm{user}}_k\right)}\×{\mathrm{\β}}^{(1-q/u)})\end{array};$

In formula, judgment (RI.recommender _j) ∈ [0,1] expression information recommendation person RI.recommender _jto the public praise evaluation of estimate that recommendation information RI provides, when public praise evaluation of estimate is lower than public praise evaluation of estimate threshold value, represent that information recommendation person does not recommend other people to use, p represents the evaluation number of times not recommending other people to use;

Recommendation information RI receives u domestic consumer's public praise evaluation, RI.user _jrepresent kGe domestic consumer, r (RI.user _j) ∈ [0,1] represents domestic consumer RI.user _jcredit worthiness, judgment (RI.user _j) ∈ [0,1] represents domestic consumer RI.user _j, the public praise evaluation of estimate provided, when public praise evaluation of estimate is higher than public praise evaluation of estimate threshold value, represent that deserving public praise is evaluated as favorable comment, q represents the favorable comment number of times in u domestic consumer's public praise is evaluated, ω ₁and ω ₂for weight parameter, meet ω ₁, ω ₂∈ [0,1], ω ₁+ ω ₂=1.

5. the computing method of recommendation information degree of confidence in a kind of Internet of Things according to claim 1, it is characterized in that, the degree of confidence T_adopt (RI) that adopts of the calculated recommendation information RI described in step (4) is specially, $T\_\mathrm{adopt}\left(\mathrm{RI}\right)=\frac{{\mathrm{\Σ}}_{l=1}^{s}r\left({\mathrm{RI}.\mathrm{user}}_l\right)}{s}\×{\left(\frac{s}{y}\right)}^{l/s},$

In formula, recommendation information RI is recommended gives y user, adopts user s altogether of this recommendation information, RI.user _lrepresent l the user accepting this recommendation information, r (RI.user _l) ∈ [0,1] expression user RI.user _lcredit worthiness.