CN109360058A - Method for pushing, device, computer equipment and storage medium based on trust network - Google Patents
Method for pushing, device, computer equipment and storage medium based on trust network Download PDFInfo
- Publication number
- CN109360058A CN109360058A CN201811191704.6A CN201811191704A CN109360058A CN 109360058 A CN109360058 A CN 109360058A CN 201811191704 A CN201811191704 A CN 201811191704A CN 109360058 A CN109360058 A CN 109360058A
- Authority
- CN
- China
- Prior art keywords
- user
- trust
- value
- comment
- row vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003860 storage Methods 0.000 title claims abstract description 55
- 238000000034 method Methods 0.000 title claims abstract description 49
- 239000011159 matrix material Substances 0.000 claims abstract description 139
- 239000013065 commercial product Substances 0.000 claims abstract description 65
- 230000008909 emotion recognition Effects 0.000 claims description 65
- 230000008451 emotion Effects 0.000 claims description 38
- 230000011218 segmentation Effects 0.000 claims description 28
- 230000001186 cumulative effect Effects 0.000 claims description 26
- 238000004590 computer program Methods 0.000 claims description 14
- 238000012163 sequencing technique Methods 0.000 claims description 14
- 230000007935 neutral effect Effects 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 6
- 235000013399 edible fruits Nutrition 0.000 claims description 5
- 238000009825 accumulation Methods 0.000 claims 1
- 230000035508 accumulation Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 14
- 239000000047 product Substances 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000001914 filtration Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000005520 cutting process Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0609—Buyer or seller confidence or verification
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses method for pushing, device, computer equipment and storage mediums based on trust network.This method parses comment on commodity information and determines there is mutually comment relationship between user, to obtain the users to trust matrix of current time storage, it obtains from current and trusts target user corresponding to the row vector that matrix is chosen, according to trust value size each in row vector, obtains the trust user before trust value size is located at rank threshold and trust user's cluster to form;User's corresponding scoring row vector in user-rating matrix is respectively trusted in user's cluster according to trusting, and is obtained trust user and is formed commercial product recommending row vector to the trust score value of each commodity;Commercial product recommending list is obtained by commercial product recommending row vector, commercial product recommending list is pushed into the corresponding receiving end of target user.This method is realized using intelligent recommendation technology trusts user's cluster by determining between user there are mutual comment relationship, and the commodity recommended according to trust user are precisely to recommend target user.
Description
Technical field
The present invention relates to information advancing technique field more particularly to it is a kind of by the method for pushing of trust network, device, based on
Calculate machine equipment and storage medium.
Background technique
Currently, progress shopping at network is more and more frequent on online store Internet-based, these online stores pair
When user carries out commercial product recommending, usually used is that (collaborative filtering, principle are users to the proposed algorithm based on collaborative filtering
The commodity that those users with similar interests liked are liked, for example your friend likes film Harry Potter I, then will
You is recommended, this is the simplest collaborative filtering based on user).When using the proposed algorithm based on collaborative filtering,
The data set of use is mainly based upon user-commodity rating matrix, and then calculates the similitude between user, belongs to stealth
Trust network.But the similitude between user is calculated using rating matrix and there are problems that cold start-up, influence recommendation
Process.
Summary of the invention
The embodiment of the invention provides a kind of method for pushing based on trust network, device, computer equipment and storages to be situated between
Matter, it is intended to solve to calculate between user using rating matrix using the proposed algorithm based on collaborative filtering in the prior art
Similitude easily leads to the cold start-up for recommending operation, thus the problem of influencing recommendation process.
In a first aspect, the embodiment of the invention provides a kind of method for pushing based on trust network comprising:
If the comment on commodity information and the comment on commodity information comparison of last moment at current time are to determine to deposit between user
In newly-increased mutual comment relationship, current trust is obtained according to the corresponding comment content of the mutual comment relationship increased newly between user
Value;
The users to trust matrix that last moment stores is updated according to the current trust value, obtaining current time deposits
The users to trust matrix of storage;Wherein, the users to trust matrix of the users to trust matrix of last moment storage and current time storage
In, each value indicates that value is expert at corresponding user to the trust value of the corresponding user of value column;
Target user corresponding to the row vector chosen from the users to trust matrix that the current time stores is obtained, according to
Each trust value size in the corresponding row vector of target user obtains the trust before trust value size is located at default rank threshold and uses
User's cluster is trusted to form in family;
According to user's corresponding scoring row vector in user-rating matrix is respectively trusted in trust user's cluster, obtain
Trust user and forms commercial product recommending row vector to the trust score value of each commodity;And
Commercial product recommending list is obtained by commercial product recommending row vector, it is corresponding that the commercial product recommending list is pushed to target user
Receiving end.
Second aspect, the embodiment of the invention provides a kind of driving means based on trust network comprising:
Current trust value acquiring unit, if believing for the comment on commodity information at current time and the comment on commodity of last moment
Breath is compared to determine to have newly-increased mutual comment relationship between user, corresponding according to the mutual comment relationship increased newly between user
Comment content obtain current trust value;
Users to trust matrix update unit, the users to trust square for being stored according to the current trust value to last moment
Battle array is updated, and obtains the users to trust matrix of current time storage;Wherein, the users to trust matrix and work as that last moment stores
In the users to trust matrix of preceding moment storage, each value indicates that the value corresponding user that is expert at is corresponding to value column
The trust value of user;
Trust user's cluster acquiring unit, for obtains from the row chosen of users to trust matrix of current time storage to
The corresponding target user of amount obtains trust value size and is located at according to trust value size each in the corresponding row vector of target user
Trust user before default rank threshold trusts user's cluster to form;
Recommend row vector acquiring unit, for respectively trusting user in user-rating matrix according in trust user's cluster
In corresponding scoring row vector, obtain trust user commercial product recommending row vector is formed to the trust score value of each commodity;
Push unit pushes the commercial product recommending list for obtaining commercial product recommending list by commercial product recommending row vector
To the corresponding receiving end of target user.
The third aspect, the embodiment of the present invention provide a kind of computer equipment again comprising memory, processor and storage
On the memory and the computer program that can run on the processor, the processor execute the computer program
Method for pushing based on trust network described in the above-mentioned first aspect of Shi Shixian.
Fourth aspect, the embodiment of the invention also provides a kind of computer readable storage mediums, wherein the computer can
It reads storage medium and is stored with computer program, it is above-mentioned that the computer program when being executed by a processor executes the processor
Based on the method for pushing of trust network described in first aspect.
The embodiment of the invention provides a kind of method for pushing based on trust network, device, computer equipment and storages to be situated between
Matter.This method determines there is mutually comment relationship between user by parsing comment on commodity information, obtains current time to adjust
The users to trust matrix of storage obtains mesh corresponding to the row vector chosen from the users to trust matrix that the current time stores
User is marked, according to trust value size each in the corresponding row vector of target user, trust value size is obtained and is located at default rank threshold
Trust user before trusts user's cluster to form;User is respectively trusted in user-rating matrix according in trust user's cluster
In corresponding scoring row vector, obtain trust user commercial product recommending row vector is formed to the trust score value of each commodity;By quotient
Product recommend row vector to obtain commercial product recommending list, and the commercial product recommending list is pushed to the corresponding receiving end of target user.It should
Method is realized using intelligent recommendation technology trusts user's cluster by determining between user there are mutual comment relationship, according to trust
The commodity that user is recommended are precisely to recommend target user.
Detailed description of the invention
Technical solution in order to illustrate the embodiments of the present invention more clearly, below will be to needed in embodiment description
Attached drawing is briefly described, it should be apparent that, drawings in the following description are some embodiments of the invention, general for this field
For logical technical staff, without creative efforts, it is also possible to obtain other drawings based on these drawings.
Fig. 1 is the flow diagram of the method for pushing provided in an embodiment of the present invention based on trust network;
Fig. 2 is the sub-process schematic diagram of the method for pushing provided in an embodiment of the present invention based on trust network;
Fig. 3 is another sub-process schematic diagram of the method for pushing provided in an embodiment of the present invention based on trust network;
Fig. 4 is another sub-process schematic diagram of the method for pushing provided in an embodiment of the present invention based on trust network;
Fig. 5 is the schematic block diagram of the driving means provided in an embodiment of the present invention based on trust network;
Fig. 6 is the subelement schematic block diagram of the driving means provided in an embodiment of the present invention based on trust network;
Fig. 7 is another subelement schematic block diagram of the driving means provided in an embodiment of the present invention based on trust network;
Fig. 8 is another subelement schematic block diagram of the driving means provided in an embodiment of the present invention based on trust network;
Fig. 9 is the schematic block diagram of computer equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
It should be appreciated that ought use in this specification and in the appended claims, term " includes " and "comprising" instruction
Described feature, entirety, step, operation, the presence of element and/or component, but one or more of the other feature, whole is not precluded
Body, step, operation, the presence or addition of element, component and/or its set.
It is also understood that mesh of the term used in this description of the invention merely for the sake of description specific embodiment
And be not intended to limit the present invention.As description of the invention and it is used in the attached claims, unless on
Other situations are hereafter clearly indicated, otherwise " one " of singular, "one" and "the" are intended to include plural form.
It will be further appreciated that the term "and/or" used in description of the invention and the appended claims is
Refer to any combination and all possible combinations of one or more of associated item listed, and including these combinations.
Referring to Fig. 1, Fig. 1 is the flow diagram of the method for pushing provided in an embodiment of the present invention based on trust network,
The method for pushing based on trust network is applied in management server, and this method passes through the application that is installed in management server
Software is executed, and management server is the enterprise terminal for carrying out the push based on trust network.
As shown in Figure 1, the method comprising the steps of S110~S150.
If the comment on commodity information comparison of S110, the comment on commodity information at current time and last moment with determine user it
Between there is newly-increased mutual comment relationship, obtained currently according to the corresponding comment content of the mutual comment relationship increased newly between user
Trust value.
In the present embodiment, when management server in having crawled default URL address list the corresponding commodity of each webpage
After comment information, can according between user and user whether there is comment relationship to judge the trusting relationship between user, and
Current trust value is obtained according to the corresponding comment content of the mutual comment relationship increased newly between user.Such customer relationship is aobvious
Property, such as in comment on commodity, party A-subscriber has affirmed the comment of the user of B, it is shown that party A-subscriber trusts party B-subscriber, or in community
In space, party A-subscriber has affirmed the content of party B-subscriber.It, can be in addition existing historical data based on this explicit trust network
The problem of better solving cold start-up.
In one embodiment, as shown in Fig. 2, step S110 includes;
S111, obtain user between increase newly the corresponding comment content of mutual comment relationship, by the comment content into
Row participle obtains word segmentation result;
S112, it is obtained in word segmentation result by Word2Vec model and respectively comments on the corresponding term vector of keyword, tied by participle
The corresponding acquisition text vector of the corresponding term vector of keyword is respectively commented in fruit;
The input of S113, the model-naive Bayesian for obtaining text vector as preparatory training obtain and comment content
Corresponding emotion recognition result;Wherein, if emotion recognition result value is 1 when emotion recognition result is positive emotion result, if
Emotion recognition result value is -1 when emotion recognition result is negative emotion result, if emotion recognition result is neutral emotion result
When emotion recognition result value be 0;
S114, emotion recognition result is multiplied with attenuation coefficient, obtains current trust value;Wherein attenuation coefficient is e-λ(t-t0), λ is preset adjustment parameter and value range is (0,1), and mutual comment relationship of the t-t0 between user is corresponding to be commented
By time interval.
In the present embodiment, by being parsed to comment on commodity information, mainly judge between user with the presence or absence of phase
Mutually comment relationship carries out positive emotion, neutral emotion and negative emotion to comment content if there are comment relationships between two users
Judgement.When carrying out emotion recognition to comment content, comment content is first subjected to participle and is then converted to text vector, by text
Input of the vector as model-naive Bayesian, obtains emotion recognition result.But in order in view of being commented between user and user
Relationship is influenced by time decaying, then is needed emotion recognition result multiplied by an attenuation coefficient (it can be appreciated that the time
The decline factor) obtain more objective and accurate current trust value.Wherein, Word2Vec is from a large amount of corpus of text with unsupervised
Mode learn a kind of model of semantic knowledge.
If current trust value of the user i to user j is denoted as aij, then aij+=Sentimentij*e-λ(t-t0), namely it is current
The trust value at moment is to carry out that (such as i user is every to the comment of j user from the adjustment increased on the basis of the trust value of last moment
Every all delivering primary comment for a period of time, then the comment delivered every time can all have an impact current trust value), wherein
SentimentijIndicate that the comment content that i user delivers the comment of j user carries out the emotion recognition knot that emotion recognition obtains
Fruit, if Sentiment when emotion recognition result is positive emotion resultij=1, if emotion recognition result is negative emotion result
Sentimentij=-1, if Sentiment when emotion recognition result is neutral emotion resultij=0.
For example, if preset adjustment parameter λ=0.5, the additional comment below the comment of user 2 of user 1, and the comment is
The difference of the comment of positive emotion, the comment time t0 of the additional comment time t and user 2 of user 1 is 1 day, then user 1 is to user
2 current trust value is denoted as a12=0+1*e-0.5*(1-0)=0.607;The additional comment below the comment of user 3 of user 1, and should
Comment is the comment of negative emotion, and the difference of the comment time t0 of the additional comment time t and user 3 of user 1 is 1 day, then user 1
A is denoted as to the current trust value of user 313=0-1*e-0.5*(1-0)=-0.607;The addition below the comment of user 4 of user 1 is commented
By, and the comment is the comment of neutral emotion, the difference of the comment time t0 of the additional comment time t and user 4 of user 1 is 1 day,
Then user 1 is denoted as a to the current trust value of user 415=0-0*e-0.5*(1-0)=0.
Using the corresponding emotion recognition of comment content between user as a result, as the calculating basis of trust value between user,
It is that the explicit and non-implicit trust network of one kind builds mode, can solve the problem of being cold-started in recommendation process.
In one embodiment, as shown in figure 4, step S111 includes:
S1111, candidate word is taken out from comment content by sequence from left to right;
S1112, probability value corresponding with each candidate word is inquired in pre-stored dictionary, and record each candidate word
Left adjacent word;
S1113, the cumulative probability for obtaining each candidate word is calculated, and it is each to obtain the corresponding multiple left adjacent words of each candidate word
From cumulative probability, if in multiple left adjacent words of each candidate word there are in the cumulative probability that cumulative probability is multiple left adjacent words most
The left adjacent word being worth greatly, using the left adjacent word of maximum value in cumulative probability as best left adjacent word corresponding with candidate word;
S1114, using the terminal word for commenting on content as starting point, be sequentially output from right to left corresponding with each candidate word best
Left neighbour's word, obtains word segmentation result.
In the present embodiment, to comment content segment when, be by the segmenting method based on probability statistics model into
Row participle.For example, enabling C=C1C2...Cm, C is Chinese character string to be slit, and to enable W=W1W2...Wn, W be cutting as a result, Wa,
Wb ... .Wk is all possible cutting scheme of C.So, the segmentation model based on probability statistics is to find purpose word
W is gone here and there, so that W meets: P (W | C)=MAX (P (Wa | C), P (Wb | C) ... P (Wk | C)) participle model, above-mentioned participle model obtains
The word string W i.e. estimated probability arrived is the word string of maximum.
I.e. to a substring S to be segmented, according to sequence from left to right take out whole candidate word w1, w2 ..., wi ...,
wn;The probability value P (wi) of each candidate word is found in dictionary, and the left adjacent word of the whole for recording each candidate word;It calculates each
The cumulative probability of candidate word, while comparing the best left adjacent word for obtaining each candidate word;If current word wn is the tail of word string S
Word, and cumulative probability P (wn) is maximum, then wn is exactly the terminal word of S;It, successively will be each according to sequence from right to left since wn
The best left adjacent word output of word, the i.e. word segmentation result of S.
S120, the users to trust matrix that last moment stores is updated according to the current trust value, is obtained current
The users to trust matrix of moment storage;Wherein, user's letter of the users to trust matrix of last moment storage and current time storage
Appoint in matrix, each value indicates that value is expert at corresponding user to the trust value of the corresponding user of value column.
In the present embodiment, users to trust matrix indicates the trust value between user, horizontal axis in users to trust matrix and
The longitudinal axis is user list, users to trust matrix An*nMiddle aijRefer to that user i to the trust value of user j, carves storage at the beginning
Users to trust matrix in all values be 0, later every period of time T rejudge user i whether have to user j it is newly-increased
Comment content is to determine whether to adjust user i to the trust value of user j.The users to trust matrix of last moment storage is management clothes
Business device has crawled the corresponding quotient of each webpage in default URL address list in last moment (when such as X month Y day Z1 in 2018 Z2 divide)
After product comment information, can according between user and user whether there is comment relationship to judge the trusting relationship between user,
And the trust value of last moment is obtained according to the corresponding comment content of the mutual comment relationship between user.
If corresponding with last moment interval time cycle T is to crawl at current time (when such as X month Y day Z3 in 2018 Z4 points)
In default URL address list after the corresponding comment on commodity information of each webpage, whether can be deposited according between user and user
In newly-increased comment relationship to judge whether the trusting relationship between user changes, and according to the mutual comment between user
The corresponding comment content of relationship obtains the trust value at current time.That is user i is denoted as a to the current trust value of user jij, then aij
+=Sentimentij*e-λ(t-t0)Namely the trust value at current time is carried out on the basis of the trust value of last moment from increasing
Adjustment (such as i user delivers primary comment to the comment of j user at regular intervals, then the comment delivered every time is all
Current trust value can be had an impact).
Target corresponding to the row vector that S130, acquisition are chosen from the users to trust matrix that the current time stores is used
Family obtains trust value size and is located at before default rank threshold according to trust value size each in the corresponding row vector of target user
Trust user with form trust user's cluster.
In the present embodiment, such as the users to trust matrix of current time storage is as follows:
And in the row vector that the row vector that the users to trust matrix of current time storage is chosen is the first row, the row of the first row
The corresponding target user of vector is user 1, is respectively worth the trust indicated between user 1 and other users in the row vector of the first row
Value.If default rank threshold is 3, it is 3 and 1 that trust value size, which (arranges preceding 2 trust values) before being located at the 3rd, is divided
User 2 and user 4 are not corresponded to, then trusting includes user 2 and user 4 in user's cluster.
S140, according to respectively trusting user's corresponding scoring row vector in user-rating matrix in trust user's cluster,
It obtains trust user and forms commercial product recommending row vector to the trust score value of each commodity.
In the present embodiment, user-rating matrix indicates user to project (project can be understood as specific commodity)
Scoring, user-rating matrix horizontal axis are project, and the longitudinal axis is user, and value in the middle is scoring of the user i to project j.Such as with
The matrix that family-rating matrix S is 4 × 6, such as:
In user-rating matrix S the first row row vector indicate user 1 respectively be directed to project 1- project 5 scoring, second
Capable row vector indicates that user 2 is directed to the scoring of project 1- project 5 respectively, and the row vector of the third line indicates that user 3 is directed to respectively
The scoring of project 1- project 5.
Such as, it is determined that trusting includes user 2 and user 4 in user's cluster, then corresponding obtain is used in user-rating matrix
The row vector of the scoring row vector namely the second row of family 2 and user 4 and the row vector of fourth line.Comprehensively considering target user
With trust the trust value in user's cluster between each user, and trust user to the score value of each commodity, can operation obtain quotient
Product recommend row vector.
In one embodiment, as shown in figure 3, step S140 includes:
S141, it obtains in trust user's cluster and respectively trusts user's corresponding scoring row vector in user-rating matrix
The rating matrix of composition;
S142, it obtains by trust user's cluster letter that each trust value size forms in the corresponding row vector of target user
Appoint family row vector;
S143, the trust user row vector is multiplied with the rating matrix, to obtain commercial product recommending row vector.
In this embodiment, such as selected having determined is trusted including user 2 and user 4 in user's cluster, then in user-scoring
The row vector of the row vector and fourth line of the corresponding scoring row vector for obtaining user 2 and user 4 namely the second row in matrix, then
User 2 and user 4 are as follows for the rating matrix for the composition that scores between projects (namely each commodity):
In one embodiment, it is specifically included in step S141: obtaining in trust user's cluster and respectively trust user in user-
Corresponding scoring row vector in rating matrix, according to each trust user in user-rating matrix in corresponding scoring row vector
The sequencing of row serial number is arranged, and rating matrix is obtained.According to above-mentioned each user, each row occurs in user-rating matrix
Sequencing, sequentially to obtain each trust user corresponding scoring row vector in user-rating matrix, aforesaid way can essence
The rating matrix for really obtaining the corresponding scoring row vector composition of each trust user, convenient for subsequent calculating user to the trust of each commodity
Score value.
The trust user's row being made of trust user's cluster each trust value size in the corresponding row vector of target user
Vector is [31].
In one embodiment, it is specifically included in step S142: obtaining trust value size and be located at before default rank threshold
Trust user, trust value size is located to the trust user before the rank threshold according in the corresponding row vector of target user
The sequencing of middle column serial number is arranged, user's row vector of establishing trust.It is corresponding in target user according to above-mentioned each user
Existing sequencing is respectively listed in row vector, sequentially to obtain each trust value size in the corresponding row vector of target user, with
Trust user's row vector of composition, aforesaid way, which can be obtained accurately, trusts user's row vector, convenient for subsequent calculating user to each quotient
The trust score value of product.
User's row vector will be trusted to be multiplied with rating matrix, i.e.,
I.e. commercial product recommending row vector is [2 13 8 13 7], and scoring in this commercial product recommending row vector, ranking is higher to be commented
Commodity corresponding to point can be used as trusting the commodity that target user recommends in group.By the way that user's row vector and scoring will be trusted
The mode of matrix multiple obtains commercial product recommending row vector, can be effectively with reference to trusting user institute to the comprehensive scores of each commodity, with work
For the reference index for carrying out commercial product recommending to target user.
S150, commercial product recommending list is obtained by commercial product recommending row vector, the commercial product recommending list is pushed into target and is used
The corresponding receiving end in family.
In the present embodiment, after obtaining commercial product recommending row vector, user is respectively trusted you can learn that trusting in user's cluster
To the comprehensive score of each commodity, scoring ranking in the corresponding each scoring of commercial product recommending row vector may be selected at this time and be located at default ranking
Scoring before value (such as preset rank value is 4), then obtain commenting before the ranking that scores in each scoring is located at default rank value
Divide corresponding merchandise news, merchandise news is formed into commercial product recommending list and pushes to the corresponding receiving end of target user.Using
The mode of commercial product recommending row vector is calculated to obtain commercial product recommending list, the accuracy of recommendation can be effectively improved.
For example, scoring is sorted in descending order in 5 commodity that above-mentioned trust group (user 2 and user 4) is recommended to user 1
It is commodity 2, commodity 3 and commodity 4 respectively that heel row name, which is located at preceding 3 commodity, at this time using above-mentioned 3 commodity as commercial product recommending row
Vector.
This method is realized using intelligent recommendation technology trusts user by determining between user there are mutual comment relationship
Cluster, the commodity recommended according to trust user are precisely to recommend target user.
The embodiment of the present invention also provides a kind of driving means based on trust network, should the driving means based on trust network
For executing any embodiment of the aforementioned method for pushing based on trust network.Specifically, referring to Fig. 5, Fig. 5 is of the invention real
The schematic block diagram of the driving means based on trust network of example offer is provided.The driving means 100 based on trust network can be with
It is configured in management server.
As shown in figure 5, the driving means 100 based on trust network includes current trust value acquiring unit 110, users to trust
Matrix update unit 120 trusts user's cluster acquiring unit 130, recommends row vector acquiring unit 140 and push unit 150.
Current trust value acquiring unit 110, if the comment on commodity information and the commodity of last moment for current time are commented
By information comparison to determine to have newly-increased mutual comment relationship between user, according to the mutual comment relationship increased newly between user
Corresponding comment content obtains current trust value.
In the present embodiment, when management server in having crawled default URL address list the corresponding commodity of each webpage
After comment information, can according between user and user whether there is comment relationship to judge the trusting relationship between user, and
Current trust value is obtained according to the corresponding comment content of the mutual comment relationship increased newly between user.Such customer relationship is aobvious
Property, such as in comment on commodity, party A-subscriber has affirmed the comment of the user of B, it is shown that party A-subscriber trusts party B-subscriber, or in community
In space, party A-subscriber has affirmed the content of party B-subscriber.It, can be in addition existing historical data based on this explicit trust network
The problem of better solving cold start-up.
In one embodiment, as shown in fig. 6, current trust value acquiring unit 110 includes:
Participle unit 111 will be described for obtaining the corresponding comment content of mutual comment relationship increased newly between user
Comment content is segmented to obtain word segmentation result;
Text vector acquiring unit 112 respectively comments on keyword pair for obtaining in word segmentation result by Word2Vec model
The term vector answered, by respectively commenting in word segmentation result, the corresponding term vector of keyword is corresponding to obtain text vector;
Emotion recognition unit 113, the input of the model-naive Bayesian for obtaining text vector as preparatory training,
Obtain emotion recognition result corresponding with comment content;Wherein, if emotion recognition result be positive emotion result when emotion recognition
As a result value is 1, if emotion recognition result value is -1 when emotion recognition result is negative emotion result, if emotion recognition result
Emotion recognition result value is 0 when for neutral emotion result;
Trust value computing unit 114 obtains current trust value for emotion recognition result to be multiplied with attenuation coefficient;Its
Middle attenuation coefficient is e-λ(t-t0), λ is preset adjustment parameter and value range is (0,1), t-t0 mutually commenting between user
By the corresponding comment time interval of relationship.
In the present embodiment, by being parsed to comment on commodity information, mainly judge between user with the presence or absence of phase
Mutually comment relationship carries out positive emotion, neutral emotion and negative emotion to comment content if there are comment relationships between two users
Judgement.When carrying out emotion recognition to comment content, comment content is first subjected to participle and is then converted to text vector, by text
Input of the vector as model-naive Bayesian, obtains emotion recognition result.But in order in view of being commented between user and user
Relationship is influenced by time decaying, then is needed emotion recognition result multiplied by an attenuation coefficient (it can be appreciated that the time
The decline factor) obtain more objective and accurate current trust value.
If current trust value of the user i to user j is denoted as aij, then aij+=Sentimentij*e-λ(t-t0), namely it is current
The trust value at moment is to carry out that (such as i user is every to the comment of j user from the adjustment increased on the basis of the trust value of last moment
Every all delivering primary comment for a period of time, then the comment delivered every time can all have an impact current trust value), wherein
SentimentijIndicate that the comment content that i user delivers the comment of j user carries out the emotion recognition knot that emotion recognition obtains
Fruit, if Sentiment when emotion recognition result is positive emotion resultij=1, if emotion recognition result is negative emotion result
Sentimentij=-1, if Sentiment when emotion recognition result is neutral emotion resultij=0.
For example, if preset adjustment parameter λ=0.5, the additional comment below the comment of user 2 of user 1, and the comment is
The difference of the comment of positive emotion, the comment time t0 of the additional comment time t and user 2 of user 1 is 1 day, then user 1 is to user
2 current trust value is denoted as a12=0+1*e-0.5*(1-0)=0.607;The additional comment below the comment of user 3 of user 1, and should
Comment is the comment of negative emotion, and the difference of the comment time t0 of the additional comment time t and user 3 of user 1 is 1 day, then user 1
A is denoted as to the current trust value of user 313=0-1*e-0.5*(1-0)=-0.607;The addition below the comment of user 4 of user 1 is commented
By, and the comment is the comment of neutral emotion, the difference of the comment time t0 of the additional comment time t and user 4 of user 1 is 1 day,
Then user 1 is denoted as a to the current trust value of user 415=0-0*e-0.5*(1-0)=0.
Using the corresponding emotion recognition of comment content between user as a result, as the calculating basis of trust value between user,
It is that the explicit and non-implicit trust network of one kind builds mode, can solve the problem of being cold-started in recommendation process.
In one embodiment, as shown in figure 8, participle unit 111 includes:
Candidate word selection unit 1111, for taking out candidate word from comment content by sequence from left to right;
Initial left adjacent word acquiring unit 1112, it is corresponding with each candidate word general for being inquired in pre-stored dictionary
Rate value, and record the left adjacent word of each candidate word;
Best left adjacent word acquiring unit 1113, for calculating the cumulative probability for obtaining each candidate word, and obtains each time
The corresponding multiple left adjacent respective cumulative probabilities of word of word are selected, if there are cumulative probability being more in multiple left adjacent words of each candidate word
The left adjacent word of maximum value in the cumulative probability of a left adjacent word, using the left adjacent word of maximum value in cumulative probability as corresponding with candidate word
Best left adjacent word;
Word segmentation result output unit 1114, for using the terminal word for commenting on content as starting point, be sequentially output from right to left with
The corresponding best left adjacent word of each candidate word, obtains word segmentation result.
In the present embodiment, to comment content segment when, be by the segmenting method based on probability statistics model into
Row participle.For example, enabling C=C1C2...Cm, C is Chinese character string to be slit, and to enable W=W1W2...Wn, W be cutting as a result, Wa,
Wb ... .Wk is all possible cutting scheme of C.So, the segmentation model based on probability statistics is to find purpose word
W is gone here and there, so that W meets: P (W | C)=MAX (P (Wa | C), P (Wb | C) ... P (Wk | C)) participle model, above-mentioned participle model obtains
The word string W i.e. estimated probability arrived is the word string of maximum.
I.e. to a substring S to be segmented, according to sequence from left to right take out whole candidate word w1, w2 ..., wi ...,
wn;The probability value P (wi) of each candidate word is found in dictionary, and the left adjacent word of the whole for recording each candidate word;It calculates each
The cumulative probability of candidate word, while comparing the best left adjacent word for obtaining each candidate word;If current word wn is the tail of word string S
Word, and cumulative probability P (wn) is maximum, then wn is exactly the terminal word of S;It, successively will be each according to sequence from right to left since wn
The best left adjacent word output of word, the i.e. word segmentation result of S.
Users to trust matrix update unit 120, for being believed according to the current trust value the user that last moment stores
Appoint matrix to be updated, obtains the users to trust matrix of current time storage;Wherein, the users to trust matrix of last moment storage
In the users to trust matrix of current time storage, each value indicates that value is expert at corresponding user to value column pair
The trust value of the user answered.
In the present embodiment, users to trust matrix indicates the trust value between user, horizontal axis in users to trust matrix and
The longitudinal axis is user list, users to trust matrix An*nMiddle aijRefer to that user i to the trust value of user j, carves storage at the beginning
Users to trust matrix in all values be 0, later every period of time T rejudge user i whether have to user j it is newly-increased
Comment content is to determine whether to adjust user i to the trust value of user j.The users to trust matrix of last moment storage is management clothes
Business device has crawled the corresponding quotient of each webpage in default URL address list in last moment (when such as X month Y day Z1 in 2018 Z2 divide)
After product comment information, can according between user and user whether there is comment relationship to judge the trusting relationship between user,
And the trust value of last moment is obtained according to the corresponding comment content of the mutual comment relationship between user.
If corresponding with last moment interval time cycle T is to crawl at current time (when such as X month Y day Z3 in 2018 Z4 points)
In default URL address list after the corresponding comment on commodity information of each webpage, whether can be deposited according between user and user
In newly-increased comment relationship to judge whether the trusting relationship between user changes, and according to the mutual comment between user
The corresponding comment content of relationship obtains the trust value at current time.That is user i is denoted as a to the current trust value of user jij, then aij
+=Sentimentij*e-λ(t-t0)Namely the trust value at current time is carried out on the basis of the trust value of last moment from increasing
Adjustment (such as i user delivers primary comment to the comment of j user at regular intervals, then the comment delivered every time is all
Current trust value can be had an impact).
Trust user's cluster acquiring unit 130, is chosen for obtaining from the users to trust matrix of current time storage
Target user corresponding to row vector obtains trust value size according to trust value size each in the corresponding row vector of target user
Trust user before default rank threshold trusts user's cluster to form.
In the present embodiment, such as the users to trust matrix of current time storage is as follows:
And in the row vector that the row vector that the users to trust matrix of current time storage is chosen is the first row, the row of the first row
The corresponding target user of vector is user 1, is respectively worth the trust indicated between user 1 and other users in the row vector of the first row
Value.If default rank threshold is 3, it is 3 and 1 that trust value size, which (arranges preceding 2 trust values) before being located at the 3rd, is divided
User 2 and user 4 are not corresponded to, then trusting includes user 2 and user 4 in user's cluster.
Recommend row vector acquiring unit 140, for respectively trusting user in user-scoring square according in trust user's cluster
Corresponding scoring row vector in battle array obtains trust user and forms commercial product recommending row vector to the trust score value of each commodity.
In the present embodiment, user-rating matrix indicates user to project (project can be understood as specific commodity)
Scoring, user-rating matrix horizontal axis are project, and the longitudinal axis is user, and value in the middle is scoring of the user i to project j.Such as with
The matrix that family-rating matrix S is 4 × 6, such as:
In user-rating matrix S the first row row vector indicate user 1 respectively be directed to project 1- project 5 scoring, second
Capable row vector indicates that user 2 is directed to the scoring of project 1- project 5 respectively, and the row vector of the third line indicates that user 3 is directed to respectively
The scoring of project 1- project 5.
Such as, it is determined that trusting includes user 2 and user 4 in user's cluster, then corresponding obtain is used in user-rating matrix
The row vector of the scoring row vector namely the second row of family 2 and user 4 and the row vector of fourth line.Comprehensively considering target user
With trust the trust value in user's cluster between each user, and trust user to the score value of each commodity, can operation obtain quotient
Product recommend row vector.
In one embodiment, as shown in fig. 7, recommendation row vector acquiring unit 140 includes:
Rating matrix acquiring unit 141 respectively trusts user in user-rating matrix for obtaining in trust user's cluster
In it is corresponding scoring row vector composition rating matrix;
User's row vector acquiring unit 142 is trusted, for obtaining by the trust user cluster in the corresponding row of target user
Trust user's row vector of each trust value size composition in vector;
Matrix calculation unit 143, for the trust user row vector to be multiplied with the rating matrix, to obtain commodity
Recommend row vector.
In this embodiment, such as selected having determined is trusted including user 2 and user 4 in user's cluster, then in user-scoring
The row vector of the row vector and fourth line of the corresponding scoring row vector for obtaining user 2 and user 4 namely the second row in matrix, then
User 2 and user 4 are as follows for the rating matrix for the composition that scores between projects (namely each commodity):
In one embodiment, rating matrix acquiring unit 141 is also used to: being obtained in trust user's cluster and is respectively trusted user
The corresponding scoring row vector in user-rating matrix, according to each trust user, corresponding scoring is gone in user-rating matrix
The sequencing of row serial number is arranged in vector, obtains rating matrix.It is each in user-rating matrix according to above-mentioned each user
The sequencing that row occurs, sequentially to obtain each trust user corresponding scoring row vector, above-mentioned side in user-rating matrix
Formula can accurately obtain the rating matrix of the corresponding scoring row vector composition of each trust user, convenient for subsequent calculating user to each commodity
Trust score value.
The trust user's row being made of trust user's cluster each trust value size in the corresponding row vector of target user
Vector is [31].
In one embodiment, trust user's row vector acquiring unit 142 to be also used to: obtaining trust value size and be located at default row
Trust value size is located at the trust user before the rank threshold according in target user by the trust user before name threshold value
The sequencing of column serial number is arranged in corresponding row vector, user's row vector of establishing trust.According to above-mentioned each user in mesh
Existing sequencing is respectively listed in the corresponding row vector of mark user, is respectively believed in the corresponding row vector of target user sequentially to obtain
Appoint value size, with the trust user's row vector formed, aforesaid way, which can be obtained accurately, trusts user's row vector, is convenient for subsequent calculating
Trust score value of the user to each commodity.
User's row vector will be trusted to be multiplied with rating matrix, i.e.,
I.e. commercial product recommending row vector is [2 13 8 13 7], and scoring in this commercial product recommending row vector, ranking is higher to be commented
Commodity corresponding to point can be used as trusting the commodity that target user recommends in group.By the way that user's row vector and scoring will be trusted
The mode of matrix multiple obtains commercial product recommending row vector, can be effectively with reference to trusting user institute to the comprehensive scores of each commodity, with work
For the reference index for carrying out commercial product recommending to target user.
Push unit 150 pushes away the commercial product recommending list for obtaining commercial product recommending list by commercial product recommending row vector
Send receiving end corresponding to target user.
In the present embodiment, after obtaining commercial product recommending row vector, user is respectively trusted you can learn that trusting in user's cluster
To the comprehensive score of each commodity, scoring ranking in the corresponding each scoring of commercial product recommending row vector may be selected at this time and be located at default ranking
Scoring before value (such as preset rank value is 4), then obtain commenting before the ranking that scores in each scoring is located at default rank value
Divide corresponding merchandise news, merchandise news is formed into commercial product recommending list and pushes to the corresponding receiving end of target user.Using
The mode of commercial product recommending row vector is calculated to obtain commercial product recommending list, the accuracy of recommendation can be effectively improved.
For example, scoring is sorted in descending order in 5 commodity that above-mentioned trust group (user 2 and user 4) is recommended to user 1
It is commodity 2, commodity 3 and commodity 4 respectively that heel row name, which is located at preceding 3 commodity, at this time using above-mentioned 3 commodity as commercial product recommending row
Vector.
The device is realized using intelligent recommendation technology trusts user by determining between user there are mutual comment relationship
Cluster, the commodity recommended according to trust user are precisely to recommend target user.
The above-mentioned driving means based on trust network can be implemented as the form of computer program, which can be with
It is run in computer equipment as shown in Figure 9.
Referring to Fig. 9, Fig. 9 is the schematic block diagram of computer equipment provided in an embodiment of the present invention.
Refering to Fig. 9, which includes processor 502, memory and the net connected by system bus 501
Network interface 505, wherein memory may include non-volatile memory medium 503 and built-in storage 504.
The non-volatile memory medium 503 can storage program area 5031 and computer program 5032.The computer program
5032 are performed, and processor 502 may make to execute the method for pushing based on trust network.
The processor 502 supports the operation of entire computer equipment 500 for providing calculating and control ability.
The built-in storage 504 provides environment for the operation of the computer program 5032 in non-volatile memory medium 503, should
When computer program 5032 is executed by processor 502, processor 502 may make to execute the method for pushing based on trust network.
The network interface 505 is for carrying out network communication, such as the transmission of offer data information.Those skilled in the art can
To understand, structure shown in Fig. 9, only the block diagram of part-structure relevant to the present invention program, is not constituted to this hair
The restriction for the computer equipment 500 that bright scheme is applied thereon, specific computer equipment 500 may include than as shown in the figure
More or fewer components perhaps combine certain components or with different component layouts.
Wherein, the processor 502 is for running computer program 5032 stored in memory, to realize following function
Can: if the comment on commodity information at current time and the comment on commodity information comparison of last moment are newly-increased to determine to exist between user
Mutual comment relationship, according to the current trust value of mutual comment relationship corresponding comment content acquisition increased newly between user;Root
The users to trust matrix that last moment stores is updated according to the current trust value, obtains user's letter of current time storage
Appoint matrix;Wherein, in the users to trust matrix of last moment storage and the users to trust matrix of current time storage, each value
Indicate that value is expert at corresponding user to the trust value of the corresponding user of value column;It obtains and is stored from the current time
The row vector chosen of users to trust matrix corresponding to target user, according to each trust value in the corresponding row vector of target user
Size obtains the trust user before trust value size is located at default rank threshold and trusts user's cluster to form;According to the letter
It appoints and respectively trusts user's corresponding scoring row vector in user-rating matrix in the cluster of family, obtain and trust user to each commodity
Score value is trusted to form commercial product recommending row vector;Commercial product recommending list is obtained by commercial product recommending row vector, the commodity are pushed away
It recommends list and pushes to the corresponding receiving end of target user.
In one embodiment, processor 502 is described corresponding according to the mutual comment relationship increased newly between user in execution
It when comment content obtains the step of current trust value, performs the following operations: obtaining the mutual comment relationship increased newly between user
Corresponding comment content is segmented the comment content to obtain word segmentation result;Participle knot is obtained by Word2Vec model
The corresponding term vector of keyword is respectively commented in fruit, the corresponding term vector of keyword is corresponding to obtain text by respectively commenting in word segmentation result
Vector;The input for the model-naive Bayesian that text vector is obtained as preparatory training obtains feelings corresponding with comment content
Feel recognition result;Wherein, if emotion recognition result value is 1 when emotion recognition result is positive emotion result, if emotion recognition
Emotion recognition result value is -1 when being as a result negative emotion result, if emotion is known when emotion recognition result is neutral emotion result
Other result value is 0;Emotion recognition result is multiplied with attenuation coefficient, obtains current trust value;Wherein attenuation coefficient is e-λ(t-t0), λ is preset adjustment parameter and value range is (0,1), and mutual comment relationship of the t-t0 between user is corresponding to be commented
By time interval.
In one embodiment, processor 502 described respectively trusts user in user-according in trust user's cluster executing
Corresponding scoring row vector in rating matrix, obtain trust user the trust score value of each commodity is formed commercial product recommending row to
When the step of amount, perform the following operations: it is corresponding in user-rating matrix respectively to trust user in acquisition trust user's cluster
The rating matrix of the row vector that scores composition;Acquisition each trust value in the corresponding row vector of target user by the trust user cluster
Trust user's row vector of size composition;The trust user row vector is multiplied with the rating matrix, is pushed away with obtaining commodity
Recommend row vector.
In one embodiment, processor 502 described is segmented the comment content to obtain word segmentation result executing
It when step, performs the following operations: taking out candidate word from comment content by sequence from left to right;In pre-stored dictionary
Probability value corresponding with each candidate word is inquired, and records the left adjacent word of each candidate word;It calculates and obtains the tired of each candidate word
Product probability, and the respective cumulative probability of the corresponding multiple left adjacent words of each candidate word is obtained, if multiple left neighbours of each candidate word
There are the left adjacent words of maximum value in the cumulative probability that cumulative probability is multiple left adjacent words in word, by a left side for maximum value in cumulative probability
Adjacent word is as best left adjacent word corresponding with candidate word;Using the terminal word for commenting on content as starting point, be sequentially output from right to left with
The corresponding best left adjacent word of each candidate word, obtains word segmentation result.
In one embodiment, processor 502 respectively trusts user in user-in executing acquisition trust user's cluster
It in rating matrix when the step of the rating matrix of corresponding scoring row vector composition, performs the following operations: obtaining the trust and use
User's corresponding scoring row vector in user-rating matrix is respectively trusted in the cluster of family, according to each trust user in user-scoring square
The sequencing of row serial number is arranged in corresponding scoring row vector in battle array, obtains rating matrix.
In one embodiment, processor 502 is corresponding in target user by the trust user cluster in the execution acquisition
In row vector when the step of trust user's row vector of each trust value size composition, performs the following operations: obtaining trust value size
Trust user before default rank threshold, trust value size is located at trust user before the rank threshold according to
The sequencing of column serial number arranges in the corresponding row vector of target user, user's row vector of establishing trust.
It will be understood by those skilled in the art that the embodiment of computer equipment shown in Fig. 9 is not constituted to computer
The restriction of equipment specific composition, in other embodiments, computer equipment may include components more more or fewer than diagram, or
Person combines certain components or different component layouts.For example, in some embodiments, computer equipment can only include depositing
Reservoir and processor, in such embodiments, the structure and function of memory and processor are consistent with embodiment illustrated in fig. 9,
Details are not described herein.
It should be appreciated that in embodiments of the present invention, processor 502 can be central processing unit (Central
Processing Unit, CPU), which can also be other general processors, digital signal processor (Digital
Signal Processor, DSP), specific integrated circuit (Application Specific Integrated Circuit,
ASIC), ready-made programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic
Device, discrete gate or transistor logic, discrete hardware components etc..Wherein, general processor can be microprocessor or
Person's processor is also possible to any conventional processor etc..
Computer readable storage medium is provided in another embodiment of the invention.The computer readable storage medium can be with
For non-volatile computer readable storage medium.The computer-readable recording medium storage has computer program, wherein calculating
If machine program performs the steps of the comment on commodity information and the comment on commodity of last moment at current time when being executed by processor
Information comparison is to determine to have newly-increased mutual comment relationship between user, according to the mutual comment relationship pair increased newly between user
The comment content answered obtains current trust value;The users to trust matrix that last moment stores is carried out according to the current trust value
It updates, obtains the users to trust matrix of current time storage;Wherein, the users to trust matrix of last moment storage and current time
In the users to trust matrix of storage, each value indicates that value is expert at corresponding user to the corresponding user's of value column
Trust value;Target user corresponding to the row vector chosen from the users to trust matrix that the current time stores is obtained, according to
Each trust value size in the corresponding row vector of target user obtains the trust before trust value size is located at default rank threshold and uses
User's cluster is trusted to form in family;According to respectively trusting user's corresponding scoring in user-rating matrix in trust user's cluster
Row vector obtains trust user and forms commercial product recommending row vector to the trust score value of each commodity;And by commercial product recommending row
Vector obtains commercial product recommending list, and the commercial product recommending list is pushed to the corresponding receiving end of target user.
In one embodiment, described to be obtained currently according to the corresponding comment content of the mutual comment relationship increased newly between user
Trust value, comprising: obtain the corresponding comment content of mutual comment relationship increased newly between user, the comment content is carried out
Participle obtains word segmentation result;It is obtained in word segmentation result by Word2Vec model and respectively comments on the corresponding term vector of keyword, by dividing
The corresponding acquisition text vector of the corresponding term vector of keyword is respectively commented in word result;Text vector is obtained as preparatory training
The input of model-naive Bayesian obtains emotion recognition result corresponding with comment content;Wherein, if emotion recognition result is positive
Emotion recognition result value is 1 when the emotion result of face, if emotion recognition result takes when emotion recognition result is negative emotion result
Value is -1, if emotion recognition result value is 0 when emotion recognition result is neutral emotion result;By emotion recognition result and decaying
Multiplication obtains current trust value;Wherein attenuation coefficient is e-λ(t-t0), λ be preset adjustment parameter and value range be (0,
1), mutual comment relationship corresponding comment time interval of the t-t0 between user.
In one embodiment, described corresponding in user-rating matrix according to user is respectively trusted in trust user's cluster
Scoring row vector, obtain trust user commercial product recommending row vector is formed to the trust score value of each commodity, comprising: obtain institute
State the rating matrix trusted and respectively trust user's corresponding scoring row vector composition in user-rating matrix in user's cluster;It obtains
The trust user's row vector being made of trust user's cluster each trust value size in the corresponding row vector of target user;By institute
It states trust user's row vector to be multiplied with the rating matrix, to obtain commercial product recommending row vector.
In one embodiment, described segmented the comment content to obtain word segmentation result, comprising: by from left to right
Sequence takes out candidate word from comment content;Probability value corresponding with each candidate word is inquired in pre-stored dictionary, and
Record the left adjacent word of each candidate word;The cumulative probability for obtaining each candidate word is calculated, and it is corresponding more to obtain each candidate word
A left adjacent respective cumulative probability of word, if there are cumulative probability being the tired of multiple left adjacent words in multiple left adjacent words of each candidate word
The left adjacent word of maximum value in product probability, using the left adjacent word of maximum value in cumulative probability as best left neighbour corresponding with candidate word
Word;Using the terminal word for commenting on content as starting point, it is sequentially output best left adjacent word corresponding with each candidate word from right to left, obtains
Word segmentation result.
In one embodiment, respectively trust user in acquisition trust user's cluster to correspond in user-rating matrix
Scoring row vector composition rating matrix, comprising: obtain in trust user's cluster and respectively trust user in user-rating matrix
In corresponding scoring row vector, according to each trust user in user-rating matrix row serial number in corresponding scoring row vector
Sequencing is arranged, and rating matrix is obtained.
In one embodiment, it is described acquisition by the trust user cluster in the corresponding row vector of target user each trust value
Trust user's row vector of size composition, comprising: the trust user that trust value size is located at before default rank threshold is obtained, it will
Trust value size is located at the trust user before the rank threshold according to the column serial number in the corresponding row vector of target user
Sequencing is arranged, user's row vector of establishing trust.
It is apparent to those skilled in the art that for convenience of description and succinctly, foregoing description is set
The specific work process of standby, device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
Those of ordinary skill in the art may be aware that unit described in conjunction with the examples disclosed in the embodiments of the present disclosure and algorithm
Step can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and software
Interchangeability generally describes each exemplary composition and step according to function in the above description.These functions are studied carefully
Unexpectedly the specific application and design constraint depending on technical solution are implemented in hardware or software.Professional technician
Each specific application can be used different methods to achieve the described function, but this realization is it is not considered that exceed
The scope of the present invention.
In several embodiments provided by the present invention, it should be understood that disclosed unit and method, it can be with
It realizes by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unit
It divides, only logical function partition, there may be another division manner in actual implementation, can also will be with the same function
Unit set is at a unit, such as multiple units or components can be combined or can be integrated into another system or some
Feature can be ignored, or not execute.In addition, shown or discussed mutual coupling, direct-coupling or communication connection can
Be through some interfaces, the indirect coupling or communication connection of device or unit, be also possible to electricity, mechanical or other shapes
Formula connection.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
In network unit.Some or all of unit therein can be selected to realize the embodiment of the present invention according to the actual needs
Purpose.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, is also possible to two or more units and is integrated in one unit.It is above-mentioned integrated
Unit both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in one storage medium.Based on this understanding, technical solution of the present invention is substantially in other words to existing
The all or part of part or the technical solution that technology contributes can be embodied in the form of software products, should
Computer software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be
Personal computer, server or network equipment etc.) execute all or part of step of each embodiment the method for the present invention
Suddenly.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), magnetic disk or
The various media that can store program code such as person's CD.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (10)
1. a kind of method for pushing based on trust network characterized by comprising
If the comment on commodity information and the comment on commodity information comparison of last moment at current time are new to determine to exist between user
The mutual comment relationship increased obtains current trust value according to the corresponding comment content of the mutual comment relationship increased newly between user;
The users to trust matrix that last moment stores is updated according to the current trust value, obtains current time storage
Users to trust matrix;Wherein, in the users to trust matrix of last moment storage and the users to trust matrix of current time storage, often
One value indicates that value is expert at corresponding user to the trust value of the corresponding user of value column;
Target user corresponding to the row vector chosen from the users to trust matrix that the current time stores is obtained, according to target
Each trust value size in the corresponding row vector of user, obtain the trust user that trust value size is located at before default rank threshold with
Composition trusts user's cluster;
According to user's corresponding scoring row vector in user-rating matrix is respectively trusted in trust user's cluster, obtains and trust
User forms commercial product recommending row vector to the trust score value of each commodity;And
Commercial product recommending list is obtained by commercial product recommending row vector, the commercial product recommending list pushes to target user is corresponding to be connect
Receiving end.
2. the method for pushing according to claim 1 based on trust network, which is characterized in that described according to new between user
The corresponding comment content of the mutual comment relationship increased obtains current trust value, comprising:
The corresponding comment content of mutual comment relationship increased newly between user is obtained, the comment content is segmented to obtain
Word segmentation result;
It is obtained in word segmentation result by Word2Vec model and respectively comments on the corresponding term vector of keyword, by respectively being commented in word segmentation result
The corresponding term vector of keyword is corresponding to obtain text vector;
The input for the model-naive Bayesian that text vector is obtained as preparatory training obtains emotion corresponding with comment content
Recognition result;Wherein, if emotion recognition result value is 1 when emotion recognition result is positive emotion result, if emotion recognition knot
Emotion recognition result value is -1 when fruit is negative emotion result, if emotion recognition when emotion recognition result is neutral emotion result
As a result value is 0;
Emotion recognition result is multiplied with attenuation coefficient, obtains current trust value;Wherein attenuation coefficient is e-λ(t-t0), λ is default
Adjustment parameter and value range be (0,1), mutual comment relationship corresponding comment time interval of the t-t0 between user.
3. the method for pushing according to claim 1 based on trust network, which is characterized in that described to be used according to the trust
User's corresponding scoring row vector in user-rating matrix is respectively trusted in the cluster of family, is obtained and is trusted user to the trust of each commodity
Score value is to form commercial product recommending row vector, comprising:
Obtain the scoring that user's corresponding scoring row vector composition in user-rating matrix is respectively trusted in trust user's cluster
Matrix;
Obtain the trust user's row being made of trust user's cluster each trust value size in the corresponding row vector of target user
Vector;
The trust user row vector is multiplied with the rating matrix, to obtain commercial product recommending row vector.
4. the method for pushing according to claim 2 based on trust network, which is characterized in that described by the comment content
It is segmented to obtain word segmentation result, comprising:
Candidate word is taken out from comment content by sequence from left to right;
Probability value corresponding with each candidate word is inquired in pre-stored dictionary, and records the left adjacent word of each candidate word;
The cumulative probability for obtaining each candidate word is calculated, and it is general to obtain the corresponding multiple left adjacent respective accumulations of word of each candidate word
Rate, if there are the left neighbours of maximum value in the cumulative probability that cumulative probability is multiple left adjacent words in multiple left adjacent words of each candidate word
Word, using the left adjacent word of maximum value in cumulative probability as best left adjacent word corresponding with candidate word;
Using the terminal word for commenting on content as starting point, it is sequentially output best left adjacent word corresponding with each candidate word from right to left, obtains
To word segmentation result.
5. the method for pushing according to claim 3 based on trust network, which is characterized in that described to obtain the trust use
The rating matrix of user's corresponding scoring row vector composition in user-rating matrix is respectively trusted in the cluster of family, comprising:
It obtains and respectively trusts user's corresponding scoring row vector in user-rating matrix in trust user's cluster, according to each letter
Appointing family, the sequencing of row serial number arranges in corresponding scoring row vector in user-rating matrix, obtains scoring square
Battle array.
6. the method for pushing according to claim 3 based on trust network, which is characterized in that the acquisition is by the trust
Trust user's row vector of user's cluster each trust value size composition in the corresponding row vector of target user, comprising:
The trust user that trust value size is located at before default rank threshold is obtained, trust value size is located at the rank threshold
Trust user before arranges according to the sequencing of the column serial number in the corresponding row vector of target user, use of establishing trust
Family row vector.
7. a kind of driving means based on trust network characterized by comprising
Current trust value acquiring unit, if for the comment on commodity information at current time and the comment on commodity information ratio of last moment
To determine to have newly-increased mutual comment relationship between user, commented according to the mutual comment relationship increased newly between user is corresponding
Current trust value is obtained by content;
Users to trust matrix update unit, users to trust matrix for being stored according to the current trust value to last moment into
Row updates, and obtains the users to trust matrix of current time storage;Wherein, last moment storage users to trust matrix and it is current when
In the users to trust matrix for carving storage, each value indicates that value is expert at corresponding user to the corresponding user of value column
Trust value;
User's cluster acquiring unit is trusted, for obtaining the row vector institute chosen from the users to trust matrix that the current time stores
Corresponding target user obtains trust value size positioned at default according to trust value size each in the corresponding row vector of target user
Trust user before rank threshold trusts user's cluster to form;
Recommend row vector acquiring unit, for right in user-rating matrix according to user is respectively trusted in trust user's cluster
The scoring row vector answered obtains trust user and forms commercial product recommending row vector to the trust score value of each commodity;
The commercial product recommending list is pushed to mesh for obtaining commercial product recommending list by commercial product recommending row vector by push unit
Mark the corresponding receiving end of user.
8. the driving means according to claim 7 based on trust network, which is characterized in that the current trust value obtains
Unit, comprising:
Participle unit will be in the comment for obtaining the corresponding comment content of mutual comment relationship increased newly between user
Appearance is segmented to obtain word segmentation result;
Text vector acquiring unit, for by Word2Vec model obtain word segmentation result in respectively comment on the corresponding word of keyword to
Amount, by respectively commenting in word segmentation result, the corresponding term vector of keyword is corresponding to obtain text vector;
Emotion recognition unit, the input of the model-naive Bayesian for obtaining text vector as preparatory training, obtain with
Comment on the corresponding emotion recognition result of content;Wherein, if emotion recognition result takes when emotion recognition result is positive emotion result
Value is 1, if emotion recognition result value is -1 when emotion recognition result is negative emotion result, if emotion recognition result is neutrality
Emotion recognition result value is 0 when emotion result;
Trust value computing unit obtains current trust value for emotion recognition result to be multiplied with attenuation coefficient;Wherein decaying system
Number is e-λ(t-t0), λ is preset adjustment parameter and value range is (0,1), mutual comment relationship pair of the t-t0 between user
The comment time interval answered.
9. a kind of computer equipment, including memory, processor and it is stored on the memory and can be on the processor
The computer program of operation, which is characterized in that the processor realizes such as claim 1 to 6 when executing the computer program
Any one of described in the method for pushing based on trust network.
10. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence, the computer program make the processor execute such as base as claimed in any one of claims 1 to 6 when being executed by a processor
In the method for pushing of trust network.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811191704.6A CN109360058A (en) | 2018-10-12 | 2018-10-12 | Method for pushing, device, computer equipment and storage medium based on trust network |
PCT/CN2018/124954 WO2020073526A1 (en) | 2018-10-12 | 2018-12-28 | Trust network-based push method, apparatus, computer device, and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811191704.6A CN109360058A (en) | 2018-10-12 | 2018-10-12 | Method for pushing, device, computer equipment and storage medium based on trust network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109360058A true CN109360058A (en) | 2019-02-19 |
Family
ID=65349036
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811191704.6A Pending CN109360058A (en) | 2018-10-12 | 2018-10-12 | Method for pushing, device, computer equipment and storage medium based on trust network |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN109360058A (en) |
WO (1) | WO2020073526A1 (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111460819A (en) * | 2020-03-31 | 2020-07-28 | 湖南大学 | Personalized comment text recommendation system and recommendation method based on fine-grained sentiment analysis |
CN111597220A (en) * | 2019-02-21 | 2020-08-28 | 北京沃东天骏信息技术有限公司 | Data mining method and device |
KR20210046594A (en) * | 2020-04-01 | 2021-04-28 | 바이두 온라인 네트웍 테크놀러지 (베이징) 캄파니 리미티드 | Method and device for pushing information |
CN113837846A (en) * | 2021-10-27 | 2021-12-24 | 武汉卓尔数字传媒科技有限公司 | Commodity recommendation method and device, computer equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106682114A (en) * | 2016-12-07 | 2017-05-17 | 广东工业大学 | Personalized recommending method fused with user trust relationships and comment information |
CN107025606A (en) * | 2017-03-29 | 2017-08-08 | 西安电子科技大学 | The item recommendation method of score data and trusting relationship is combined in a kind of social networks |
CN107506480A (en) * | 2017-09-13 | 2017-12-22 | 浙江工业大学 | A kind of excavated based on comment recommends method with the double-deck graph structure of Density Clustering |
CN108228867A (en) * | 2018-01-15 | 2018-06-29 | 武汉大学 | A kind of theme collaborative filtering recommending method based on viewpoint enhancing |
CN108573411A (en) * | 2018-04-17 | 2018-09-25 | 重庆理工大学 | Depth sentiment analysis and multi-source based on user comment recommend the mixing of view fusion to recommend method |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103559623A (en) * | 2013-09-24 | 2014-02-05 | 浙江大学 | Personalized product recommendation method based on combined non-negative matrix decomposition |
CN107273438B (en) * | 2017-05-24 | 2021-02-23 | 深圳大学 | Recommendation method, device, equipment and storage medium |
CN107967641A (en) * | 2017-10-18 | 2018-04-27 | 美的智慧家居科技有限公司 | Method of Commodity Recommendation, device and computer-readable recording medium |
-
2018
- 2018-10-12 CN CN201811191704.6A patent/CN109360058A/en active Pending
- 2018-12-28 WO PCT/CN2018/124954 patent/WO2020073526A1/en active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106682114A (en) * | 2016-12-07 | 2017-05-17 | 广东工业大学 | Personalized recommending method fused with user trust relationships and comment information |
CN107025606A (en) * | 2017-03-29 | 2017-08-08 | 西安电子科技大学 | The item recommendation method of score data and trusting relationship is combined in a kind of social networks |
CN107506480A (en) * | 2017-09-13 | 2017-12-22 | 浙江工业大学 | A kind of excavated based on comment recommends method with the double-deck graph structure of Density Clustering |
CN108228867A (en) * | 2018-01-15 | 2018-06-29 | 武汉大学 | A kind of theme collaborative filtering recommending method based on viewpoint enhancing |
CN108573411A (en) * | 2018-04-17 | 2018-09-25 | 重庆理工大学 | Depth sentiment analysis and multi-source based on user comment recommend the mixing of view fusion to recommend method |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111597220A (en) * | 2019-02-21 | 2020-08-28 | 北京沃东天骏信息技术有限公司 | Data mining method and device |
CN111597220B (en) * | 2019-02-21 | 2024-03-05 | 北京沃东天骏信息技术有限公司 | Data mining method and device |
CN111460819A (en) * | 2020-03-31 | 2020-07-28 | 湖南大学 | Personalized comment text recommendation system and recommendation method based on fine-grained sentiment analysis |
CN111460819B (en) * | 2020-03-31 | 2023-06-20 | 湖南大学 | Personalized comment text recommendation system and recommendation method based on fine granularity emotion analysis |
KR20210046594A (en) * | 2020-04-01 | 2021-04-28 | 바이두 온라인 네트웍 테크놀러지 (베이징) 캄파니 리미티드 | Method and device for pushing information |
CN113495942A (en) * | 2020-04-01 | 2021-10-12 | 百度在线网络技术(北京)有限公司 | Method and device for pushing information |
KR102606175B1 (en) * | 2020-04-01 | 2023-11-24 | 바이두 온라인 네트웍 테크놀러지 (베이징) 캄파니 리미티드 | Method and device for pushing information |
CN113837846A (en) * | 2021-10-27 | 2021-12-24 | 武汉卓尔数字传媒科技有限公司 | Commodity recommendation method and device, computer equipment and storage medium |
CN113837846B (en) * | 2021-10-27 | 2023-09-22 | 武汉卓尔数字传媒科技有限公司 | Commodity recommendation method, commodity recommendation device, computer equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
WO2020073526A1 (en) | 2020-04-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109360058A (en) | Method for pushing, device, computer equipment and storage medium based on trust network | |
CN105335519B (en) | Model generation method and device and recommendation method and device | |
CN109999504B (en) | Game prop recommendation method, device, server and storage medium | |
CN103678672B (en) | Method for recommending information | |
CN105095219B (en) | Micro-blog recommendation method and terminal | |
CN109325182B (en) | Information pushing method and device based on session, computer equipment and storage medium | |
CN109360057A (en) | Information-pushing method, device, computer equipment and storage medium | |
CN109166017A (en) | Method for pushing, device, computer equipment and storage medium based on reunion class | |
EP3139288A1 (en) | Item recommendation method and device | |
CN109862432A (en) | Clicking rate prediction technique and device | |
US10846332B2 (en) | Playlist list determining method and device, electronic apparatus, and storage medium | |
US20150169758A1 (en) | Multi-partite graph database | |
CN107895038B (en) | Link prediction relation recommendation method and device | |
CN109102371A (en) | Method of Commodity Recommendation, device, computer equipment and storage medium | |
CN106372101B (en) | A kind of video recommendation method and device | |
US20180024989A1 (en) | Automated building and sequencing of a storyline and scenes, or sections, included therein | |
CN107545451B (en) | Advertisement pushing method and device | |
CN106446189A (en) | Message-recommending method and system | |
CN103793476A (en) | Network community based collaborative filtering recommendation method | |
CN109241451B (en) | Content combination recommendation method and device and readable storage medium | |
CN106168980A (en) | Multimedia resource recommends sort method and device | |
CN111400546B (en) | Video recall method and video recommendation method and device | |
EP4026070A1 (en) | Cluster and image-based feedback system | |
CN109241410A (en) | A kind of article recommended method and device | |
CN112989211B (en) | Method and system for determining information similarity |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |