CN111507804A - Emotion perception commodity recommendation method based on mixed information fusion - Google Patents
Emotion perception commodity recommendation method based on mixed information fusion Download PDFInfo
- Publication number
- CN111507804A CN111507804A CN202010319045.0A CN202010319045A CN111507804A CN 111507804 A CN111507804 A CN 111507804A CN 202010319045 A CN202010319045 A CN 202010319045A CN 111507804 A CN111507804 A CN 111507804A
- Authority
- CN
- China
- Prior art keywords
- user
- commodity
- similarity
- information
- score
- 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.)
- Granted
Links
- 230000008451 emotion Effects 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 32
- 230000004927 fusion Effects 0.000 title claims abstract description 13
- 230000008447 perception Effects 0.000 title claims abstract description 9
- 230000002996 emotional effect Effects 0.000 claims abstract description 11
- 238000012937 correction Methods 0.000 claims description 26
- 238000012552 review Methods 0.000 claims description 7
- 238000012935 Averaging Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000036651 mood Effects 0.000 description 1
Images
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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Game Theory and Decision Science (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides an emotion perception commodity recommendation method based on mixed information fusion, which comprises the following steps: calculating user similarity between users; calculating initial commodity recommendation degree of the commodity to the user based on the user similarity degree; correcting the initial commodity recommendation degree of the commodity to the user by utilizing the emotional information to obtain the final commodity recommendation degree of the commodity to the user; and for each commodity in the range of the unpurchased commodities, obtaining the final recommendation value of the commodity to the user, arranging the final recommendation values corresponding to the commodities in the range of the unpurchased commodities in the sequence from high to low, and selecting a plurality of commodities in the front sequence to recommend to the user. The method comprehensively considers the scoring data and the comment data of all users who have purchased the commodity, the user similarity between the user and each user who has purchased the commodity and the consultation comment data of the user before purchasing each commodity, thereby obtaining the commodity recommendation degree.
Description
Technical Field
The invention belongs to the technical field of commodity recommendation, and particularly relates to an emotion perception commodity recommendation method based on mixed information fusion.
Background
The appearance and popularization of the internet bring a great deal of information to users, and the requirement of the users on the information in the information age is met, but with the rapid development of the network, the amount of information on the internet is greatly increased, so that the users cannot obtain information really useful for the users when facing a great amount of information, the use efficiency of the information is reduced on the contrary, and the problem of information overload is solved.
One very potential solution to the information overload problem is the recommendation system, which is now widely used in many fields, the most typical application of which is the e-commerce field.
In the field of electronic commerce, the main implementation manner of the existing commodity recommendation system is as follows: and realizing personalized commodity recommendation for the user according to the comment data of the purchased commodity of the user, the interest of the user and other information. The following problems mainly exist in the commodity recommendation system: the commodity recommendation system has the problem of low commodity recommendation precision due to less user information considered.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides the emotion perception commodity recommendation method based on mixed information fusion, which can effectively solve the problems.
The technical scheme adopted by the invention is as follows:
the invention provides an emotion perception commodity recommendation method based on mixed information fusion, which comprises the following steps of:
step 1, the recommendation platform establishes a commodity set P ═ (P)1,P2,...,Pn) And user set Q ═ Q (Q)1,Q2,...,Qm) (ii) a The commodity set P is a commodity range capable of recommending commodities on the e-commerce platform, and n is the quantity of commodities capable of recommending commodities; the user set Q is a user set registered to the e-commerce platform, and m is the number of users in the user set;
a recommendation platform establishes a scoring database; whenever arbitrary user QiPurchasing arbitrary goods P from the e-commerce platformjAnd for the commodity PjWhen making post-purchase scores, the raw scores are Score (Q)i,Pj) Storing the score into the scoring database; wherein, i is 1, 2.. times, m; j is 1,2,. n; raw Score (Q)i,Pj) Representing user QiFor commodity PjOriginal review ofDividing;
step 2, when any user QiWhen accessing the electronic commerce platform, the electronic commerce platform searches the grading database to obtain a user QiThe commodity range which is originally scored after the purchase is carried out is removed from the commodity set P, and the user Q is obtainediThe commodity range without original scoring is the user QiFor each product P in the range of unpurchased productskAll adopt the method of step 3-step 5 to obtain the commodity PkFinal recommendation value of (a):
step 3, for the commodity PkCounting to obtain a pair of products PkSet of users who score and post-purchase review data for post-purchase (R ═ R1,R2,...,Rz) (ii) a The user set R is a subset of the user set Q;
for each user R in the set R of usersa1, 2.. z, each performing the steps of:
step 3.1, calculate user R by the following methodaAnd QiUser feature similarity between US (Q)i,Ra):
Step 3.1.1, collect user RaThe social network data of (1) to construct a user RaTag feature set U L (R)a) (ii) a Collecting user QiTo construct a user QiTag feature set of U L (Q)i) (ii) a The label feature set comprises a plurality of user labels for representing user features;
step 3.1.2, according to U L (R)a) And U L (Q)i) The number of the same user tags in the user R is calculated by adopting the following formulaaAnd QiUser feature tag similarity between U L (Q)i,Ra):
Wherein:
|UL(Qi)∩UL(Ra) I represents U L (Q)i) And U L (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UL(Qi)∪UL(Ra) I represents U L (Q)i) And U L (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.3, analyze user RaExtracting the social network data to obtain the user RaFriend feature set UF (R)a) (ii) a Wherein, the user RaFriend feature set UF (R)a) Is a user Ra(ii) a focused friend ID;
analyzing user QiExtracting to obtain the user QiFriend feature set of UF (Q)i) (ii) a Wherein, user QiFriend feature set of UF (Q)i) Is user Qi(ii) a focused friend ID;
step 3.1.4, according to UF (Q)i) And UF (R)a) The number of the same elements in the same group is calculated by adopting the following formula to obtain the user RaAnd QiUser characteristic friend similarity UF (Q) betweeni,Ra):
Wherein:
|UF(Qi)∩UF(Ra) I represents UF (Q)i) And UF (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UF(Qi)∪UF(Ra) I represents UF (Q)i) And UF (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.5, fusing and overlapping the similarity U L (Q) of the user feature label by using the multiplication principlei,Ra) Friend similarity UF (Q) with user characteristicsi,Ra) Constructing the similarity US (Q) of the obtained user featuresi,Ra)=UL(Qi,Ra)*UF(Qi,Ra);
Step 3.2, integrating the scores after correction and the similarity of the user characteristics, and calculating the user RaAnd QiUser similarity sim (Q) betweeni,Ra) The method comprises the following steps:
step 3.2.1, searching the scoring database, and counting to obtain the user RaAnd user QiCommon score set of goods I ═ I (I)1,I2,...,Ih) (ii) a The commodity set I is a subset of the commodity set P; h is the number of the common scoring commodities;
step 3.2.2, searching the scoring database to obtain the user RaFor commodity IuRaw Score of (R)a,Iu) (ii) a Wherein, u is 1,2,. and h;
using affective information for user RaFor commodity IuRaw Score of (R)a,Iu) Corrected to obtain a corrected Score (R)'a,Iu) The method specifically comprises the following steps:
1) collect user RaFor commodity IuAll the published comment information items are provided with w comment information items, and each comment C item is provided with w comment information itemsb(Ra,Iu) Wherein, b is 1,2b(Ra,Iu) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)b(Ra,Iu) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)b(Ra,Iu) 1); make statistics of the obtained comment Cb(Ra,Iu) Number of middle and positive emotion words Pos (C)b(Ra,Iu) And the number of negative emotion words Neg (C)b(Ra,Iu));
2) User R is obtained by adopting the following formulaaFor commodity IuCorrected Score of (R) ((R))a,Iu):
3) Calculating to obtain the userQiFor commodity IuCorrected Score' of (Q)i,Iu);
Step 3.2.3, calculating in sequence to obtain the user RaAnd (3) scoring each commodity in the set I after correction, so that h first scores after correction are obtained in total, and averaging the h first scores after correction to obtain the user RaCorrected score mean of
Sequentially calculating to obtain a user QiScoring each commodity in the set I after correction, thereby obtaining h second scores after correction in total, and then averaging the h second scores after correction to obtain the user QiCorrected score mean of
Step 3.2.4, calculate user R using the following equationaAnd QiUser similarity sim (Q) betweeni,Ra):
Step 4, adopting the following formula to calculate the commodity P based on the user similaritykFor user QiInitial commodity recommendation degree Rec (Q)i,Pk):
Wherein, Score' (R)a,Pk) Finger user RaFor commodity PkScoring after correction of (1);
step 5, correcting the commodity P by utilizing the emotional informationkFor user QiInitial commodity recommendation degree Rec (Q)i,Pk) To obtain a commodity PkFor user QiFinal commodity recommendation degree of (Q) FREci,Pk) The method comprises the following steps:
step 5.1, collecting to obtainUser QiIn purchasing an item PkAll the comment information which is consulted and discussed in the comment area of the commodity is provided with AT comment information, and each comment C is provided withT(Qi,Pk) Wherein T ═ 1, 2.., AT, statistical review CT(Qi,Pk) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)T(Qi,Pk) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)T(Qi,Pk) 1); make statistics of the obtained comment CT(Qi,Pk) Number of middle and positive emotion words Pos (C)T(Qi,Pk) And the number of negative emotion words Neg (C)T(Qi,Pk));
Step 5.2, the commodity P is obtained by adopting the following formulakFor user QiFinal commodity recommendation degree of (Q) FREci,Pk):
Step 6, therefore, for each item P in the range of unpurchased itemskAll obtain a commodity PkFor user QiThe final recommendation values corresponding to all the commodities in the range of the commodities not purchased are arranged from high to low, and a plurality of commodities which are sorted in the front are selected and recommended to the user Qi。
Preferably, in step 3.1.1, the user tags characterizing the user include the gender of the user, the age of the user, the occupation of the user, the location of the user, and social information of the user.
Preferably, the user social information includes user interest and hobby information and user character information.
The emotion perception commodity recommendation method based on mixed information fusion provided by the invention has the following advantages:
the emotion perception commodity recommendation method based on mixed information fusion comprehensively considers the scoring data and the comment data of all users who have purchased the commodity on an e-commerce platform, the user similarity between the user and each user who has purchased the commodity, and the consultation comment data of the user A before purchasing each commodity, so that the commodity recommendation degree of the commodity to the user A is obtained.
Drawings
Fig. 1 is a schematic flow diagram of an emotion-aware commodity recommendation method based on mixed information fusion provided by the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantageous effects solved by the present invention more clearly apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Referring to fig. 1, the invention provides an emotion-aware commodity recommendation method based on mixed information fusion, which includes the following steps:
step 1, the recommendation platform establishes a commodity set P ═ (P)1,P2,...,Pn) And user set Q ═ Q (Q)1,Q2,...,Qm) (ii) a The commodity set P is a commodity range which can be recommended by commodities on the electronic commerce platform, generally all commodities on the electronic commerce platform, and n is the quantity of the commodities which can be recommended by the commodities; the user set Q is a user set registered to the e-commerce platform, and m is the number of users in the user set;
a recommendation platform establishes a scoring database; whenever arbitrary user QiPurchasing arbitrary goods P from the e-commerce platformjAnd for the commodity PjWhen making post-purchase scores, the raw scores are Score (Q)i,Pj) Storing the score into the scoring database; wherein, i is 1, 2.. times, m; j is 1,2,. n; raw Score (Q)i,Pj) Representing user QiFor commodity PjRaw scoring is performed; in general, the system defaults to purchasing a commodity only when the commodity is purchasedAnd (6) grading the lines.
Step 2, when any user QiWhen accessing the electronic commerce platform, the electronic commerce platform searches the grading database to obtain a user QiThe commodity range which is originally scored after the purchase is carried out is removed from the commodity set P, and the user Q is obtainediThe commodity range without original scoring is the user QiFor each product P in the range of unpurchased productskAll adopt the method of step 3-step 5 to obtain the commodity PkFinal recommendation value of (a):
step 3, for the commodity PkCounting to obtain a pair of products PkSet of users who score and post-purchase review data for post-purchase (R ═ R1,R2,...,Rz) (ii) a The user set R is a subset of the user set Q;
for each user R in the set R of usersa1, 2.. z, each performing the steps of:
step 3.1, calculate user R by the following methodaAnd QiUser feature similarity between US (Q)i,Ra):
Step 3.1.1, collect user RaThe social network data of (1) to construct a user RaTag feature set U L (R)a) (ii) a Collecting user QiTo construct a user QiTag feature set of U L (Q)i) (ii) a The label feature set comprises a plurality of user labels for representing user features;
in practical applications, the user tags characterizing the user characteristics include the user gender, the user age, the user occupation, the user location, and the user social information. The user social information includes, but is not limited to, user interest and hobby information and user personality information. The user social information is obtained by analyzing user social network data, and the user social network data is from social network platforms, such as microblogs, network communities and other social network platforms.
Specifically, the similarity of the traditional user feature tags is calculated based on the scoring data of the user on the commodities. However, the user score may only represent an emotional bias at a certain time. Therefore, when calculating the similarity of the user feature labels, more stable information that can reflect the user features is needed.
In the invention, stable tag information representing user characteristics is firstly extracted from user social network data, wherein the tag information comprises gender S, age A, occupation W, place P and social tag group { L1, L2, … and L V }, the social tag group generally selects keywords capable of stably representing user characteristics, such as 'Kairang', 'love tour', 'like basketball' and the like, from a keyword set provided by a social network site, and a tag characteristic set U L (Q) of a user is constructedi)={S,A,W,P,L1,L2,…,LV}。
Step 3.1.2, according to U L (R)a) And U L (Q)i) The number of the same user tags in the user R is calculated by adopting the following formulaaAnd QiUser feature tag similarity between U L (Q)i,Ra):
Wherein:
|UL(Qi)∩UL(Ra) I represents U L (Q)i) And U L (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UL(Qi)∪UL(Ra) I represents U L (Q)i) And U L (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.3, analyze user RaExtracting the social network data to obtain the user RaFriend feature set UF (R)a) (ii) a Wherein, the user RaFriend feature set UF (R)a) Is a user Ra(ii) a focused friend ID;
analyzing user QiExtracting to obtain the user QiFriend feature set of UF (Q)i) (ii) a Wherein the userQiFriend feature set of UF (Q)i) Is user Qi(ii) a focused friend ID;
step 3.1.4, according to UF (Q)i) And UF (R)a) The number of the same elements in the same group is calculated by adopting the following formula to obtain the user RaAnd QiUser characteristic friend similarity UF (Q) betweeni,Ra):
Wherein:
|UF(Qi)∩UF(Ra) I represents UF (Q)i) And UF (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UF(Qi)∪UF(Ra) I represents UF (Q)i) And UF (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.5, fusing and overlapping the similarity U L (Q) of the user feature label by using the multiplication principlei,Ra) Friend similarity UF (Q) with user characteristicsi,Ra) Constructing the similarity US (Q) of the obtained user featuresi,Ra)=UL(Qi,Ra)*UF(Qi,Ra);
Therefore, when the user feature similarity is calculated, the user feature label similarity and the user feature friend similarity are comprehensively considered, and the value of the user feature similarity representing the similarity between users can be obtained more accurately due to comprehensive considered information, so that the product recommendation accuracy of the recommendation system is improved.
Step 3.2, integrating the scores after correction and the similarity of the user characteristics, and calculating the user RaAnd QiUser similarity sim (Q) betweeni,Ra) The method comprises the following steps:
step 3.2.1, searching the scoring database, and counting to obtain the user RaAnd user QiCommon score set of goods I ═ I (I)1,I2,...,Ih) (ii) a Wherein,the commodity set I is a subset of the commodity set P; h is the number of the common scoring commodities;
step 3.2.2, searching the scoring database to obtain the user RaFor commodity IuRaw Score of (R)a,Iu) (ii) a Wherein, u is 1,2,. and h;
using affective information for user RaFor commodity IuRaw Score of (R)a,Iu) Corrected to obtain a corrected Score (R)'a,Iu) The method specifically comprises the following steps:
1) collect user RaFor commodity IuAll the published comment information items are provided with w comment information items, and each comment C item is provided with w comment information itemsb(Ra,Iu) Wherein, b is 1,2b(Ra,Iu) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)b(Ra,Iu) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)b(Ra,Iu) 1); make statistics of the obtained comment Cb(Ra,Iu) Number of middle and positive emotion words Pos (C)b(Ra,Iu) And the number of negative emotion words Neg (C)b(Ra,Iu));
2) User R is obtained by adopting the following formulaaFor commodity IuCorrected Score of (R) ((R))a,Iu):
3) Calculating to obtain a user QiFor commodity IuCorrected Score' of (Q)i,Iu);
The reason for this step is: in an actual e-commerce platform, it is often seen that the score data of the user and the comment data of the user cannot be completely consistent, for example, the score of the user a for the commodity B is 7 points, but a negative emotional word, such as "no so", appears in the comment of the user a for the commodity B. The main reason is that the non-professional users areThe scoring may be influenced by a variety of factors, such as the current mood, the current point of interest, etc. Therefore, it is necessary to correct the score data using the emotion information of the comment data and recommend the score data. In the invention, the emotional information is analyzed and extracted from the user comment data, and the emotional information is calculated based on the statistics of an emotional dictionary. Then, the emotional information is adopted to the user RaFor commodity IuThe raw scores of (a) are corrected.
Step 3.2.3, calculating in sequence to obtain the user RaAnd (3) scoring each commodity in the set I after correction, so that h first scores after correction are obtained in total, and averaging the h first scores after correction to obtain the user RaCorrected score mean of
Sequentially calculating to obtain a user QiScoring each commodity in the set I after correction, thereby obtaining h second scores after correction in total, and then averaging the h second scores after correction to obtain the user QiCorrected score mean of
Step 3.2.4, calculate user R using the following equationaAnd QiUser similarity sim (Q) betweeni,Ra):
Specifically, similarity calculation among users is based on key steps in a user collaborative filtering algorithm, and similarity calculation in traditional collaborative filtering is based on scoring data of the users on commodities. However, the conventional similarity calculation has the following problems: first, the emotional information in the user comment on the product is not considered. Second, the similarity information implied in the user's social network data is not considered.
Therefore, the user similarity is calculated by fusing the user correction scoring data and the user feature similarity. Thereby obtaining a more accurate value of the user similarity.
Step 4, adopting the following formula to calculate the commodity P based on the user similaritykFor user QiInitial commodity recommendation degree Rec (Q)i,Pk):
Wherein, Score' (R)a,Pk) Finger user RaFor commodity PkScoring after correction of (1);
step 5, correcting the commodity P by utilizing the emotional informationkFor user QiInitial commodity recommendation degree Rec (Q)i,Pk) To obtain a commodity PkFor user QiFinal commodity recommendation degree of (Q) FREci,Pk) The method comprises the following steps:
step 5.1, collect and get user QiIn purchasing an item PkAll the comment information which is consulted and discussed in the comment area of the commodity is provided with AT comment information, and each comment C is provided withT(Qi,Pk) Wherein T ═ 1, 2.., AT, statistical review CT(Qi,Pk) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)T(Qi,Pk) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)T(Qi,Pk) 1); make statistics of the obtained comment CT(Qi,Pk) Number of middle and positive emotion words Pos (C)T(Qi,Pk) And the number of negative emotion words Neg (C)T(Qi,Pk));
Step 5.2, the commodity P is obtained by adopting the following formulakFor user QiFinal commodity recommendation degree of (Q) FREci,Pk):
In particular, in practical applications, it is often found that a user consults and discusses in the comment area of a commodity before purchasing a commodity. The conventional recommendation method usually only focuses on the comment data of the purchased goods, and does not fully utilize the comment data of the user on the goods which are not purchased.
In the method, emotion information is analyzed and extracted from user comment data before a user purchases a certain commodity, and then initial commodity recommendation degree is corrected by using the emotion information, so that more accurate commodity recommendation degree is obtained.
According to the correction process, even if the initial recommendation degrees of the users to the two commodities are the same, if the user frequently pays attention to comment and consults one commodity, and the comment is positive, the commodity is recommended preferentially.
Step 6, therefore, for each item P in the range of unpurchased itemskAll obtain a commodity PkFor user QiThe final recommendation values corresponding to all the commodities in the range of the commodities not purchased are arranged from high to low, and a plurality of commodities which are sorted in the front are selected and recommended to the user Qi。
Therefore, in the invention, when commodity recommendation needs to be carried out on a certain user A, each commodity which is not purchased by the user on an electronic commerce platform is obtained firstly; and then, comprehensively considering the rating data and the comment data of all users who have purchased the commodity on the e-commerce platform, the user similarity between the user and each user who has purchased the commodity and the consultation comment data of the user A before purchasing each commodity, so as to obtain the commodity recommendation degree of the commodity to the user A.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and improvements can be made without departing from the principle of the present invention, and such modifications and improvements should also be considered within the scope of the present invention.
Claims (3)
1. A mixed information fusion-based emotion perception commodity recommendation method is characterized by comprising the following steps:
step 1, the recommendation platform establishes a commodity set P ═ (P)1,P2,...,Pn) And user set Q ═ Q (Q)1,Q2,...,Qm) (ii) a The commodity set P is a commodity range capable of recommending commodities on the e-commerce platform, and n is the quantity of commodities capable of recommending commodities; the user set Q is a user set registered to the e-commerce platform, and m is the number of users in the user set;
a recommendation platform establishes a scoring database; whenever arbitrary user QiPurchasing arbitrary goods P from the e-commerce platformjAnd for the commodity PjWhen making post-purchase scores, the raw scores are Score (Q)i,Pj) Storing the score into the scoring database; wherein, i is 1, 2.. times, m; j is 1,2,. n; raw Score (Q)i,Pj) Representing user QiFor commodity PjRaw scoring is performed;
step 2, when any user QiWhen accessing the electronic commerce platform, the electronic commerce platform searches the grading database to obtain a user QiThe commodity range which is originally scored after the purchase is carried out is removed from the commodity set P, and the user Q is obtainediThe commodity range without original scoring is the user QiFor each product P in the range of unpurchased productskAll adopt the method of step 3-step 5 to obtain the commodity PkFinal recommendation value of (a):
step 3, for the commodity PkCounting to obtain a pair of products PkSet of users who score and post-purchase review data for post-purchase (R ═ R1,R2,...,Rz) (ii) a The user set R is a subset of the user set Q;
for each user R in the set R of usersa1, 2.. z, each performing the steps of:
step 3.1, calculate user R by the following methodaAnd QiUser feature similarity between US (Q)i,Ra):
Step 3.1.1, collect user RaThe social network data of (1) to construct a user RaTag feature set U L (R)a) (ii) a Collecting user QiTo construct a user QiTag feature set of U L (Q)i) (ii) a The label feature set comprises a plurality of user labels for representing user features;
step 3.1.2, according to U L (R)a) And U L (Q)i) The number of the same user tags in the user R is calculated by adopting the following formulaaAnd QiUser feature tag similarity between U L (Q)i,Ra):
Wherein:
|UL(Qi)∩UL(Ra) I represents U L (Q)i) And U L (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UL(Qi)∪UL(Ra) I represents U L (Q)i) And U L (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.3, analyze user RaExtracting the social network data to obtain the user RaFriend feature set UF (R)a) (ii) a Wherein, the user RaFriend feature set UF (R)a) Is a user Ra(ii) a focused friend ID;
analyzing user QiExtracting to obtain the user QiFriend feature set of UF (Q)i) (ii) a Wherein, user QiFriend feature set of UF (Q)i) Is user Qi(ii) a focused friend ID;
step 3.1.4, according to UF (Q)i) And UF (R)a) The number of the same elements in the same group is calculated by adopting the following formula to obtain the user RaAnd QiIn betweenUser feature friend similarity UF (Q)i,Ra):
Wherein:
|UF(Qi)∩UF(Ra) I represents UF (Q)i) And UF (R)a) Performing set intersection operation to obtain the number of elements in the set;
|UF(Qi)∪UF(Ra) I represents UF (Q)i) And UF (R)a) After the collection and the operation are carried out, the number of elements in the collection is obtained;
step 3.1.5, fusing and overlapping the similarity U L (Q) of the user feature label by using the multiplication principlei,Ra) Friend similarity UF (Q) with user characteristicsi,Ra) Constructing the similarity US (Q) of the obtained user featuresi,Ra)=UL(Qi,Ra)*UF(Qi,Ra);
Step 3.2, integrating the scores after correction and the similarity of the user characteristics, and calculating the user RaAnd QiUser similarity sim (Q) betweeni,Ra) The method comprises the following steps:
step 3.2.1, searching the scoring database, and counting to obtain the user RaAnd user QiCommon score set of goods I ═ I (I)1,I2,...,Ih) (ii) a The commodity set I is a subset of the commodity set P; h is the number of the common scoring commodities;
step 3.2.2, searching the scoring database to obtain the user RaFor commodity IuRaw Score of (R)a,Iu) (ii) a Wherein, u is 1,2,. and h;
using affective information for user RaFor commodity IuRaw Score of (R)a,Iu) Corrected to obtain a corrected Score (R)'a,Iu) The method specifically comprises the following steps:
1) collect user RaFor commodity IuAll the published comment information are provided with w pieces of comment information, and the comment information is pairedAt each comment Cb(Ra,Iu) Wherein, b is 1,2b(Ra,Iu) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)b(Ra,Iu) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)b(Ra,Iu) 1); make statistics of the obtained comment Cb(Ra,Iu) Number of middle and positive emotion words Pos (C)b(Ra,Iu) And the number of negative emotion words Neg (C)b(Ra,Iu));
2) User R is obtained by adopting the following formulaaFor commodity IuCorrected Score of (R) ((R))a,Iu):
3) Calculating to obtain a user QiFor commodity IuCorrected Score' of (Q)i,Iu);
Step 3.2.3, calculating in sequence to obtain the user RaAnd (3) scoring each commodity in the set I after correction, so that h first scores after correction are obtained in total, and averaging the h first scores after correction to obtain the user RaCorrected score mean of
Sequentially calculating to obtain a user QiScoring each commodity in the set I after correction, thereby obtaining h second scores after correction in total, and then averaging the h second scores after correction to obtain the user QiCorrected score mean of
Step 3.2.4, calculate user R using the following equationaAnd QiUser similarity sim (Q) betweeni,Ra):
Step 4, adopting the following formula to calculate the commodity P based on the user similaritykFor user QiInitial commodity recommendation degree Rec (Q)i,Pk):
Wherein, Score' (R)a,Pk) Finger user RaFor commodity PkScoring after correction of (1);
step 5, correcting the commodity P by utilizing the emotional informationkFor user QiInitial commodity recommendation degree Rec (Q)i,Pk) To obtain a commodity PkFor user QiFinal commodity recommendation degree of (Q) FREci,Pk) The method comprises the following steps:
step 5.1, collect and get user QiIn purchasing an item PkAll the comment information which is consulted and discussed in the comment area of the commodity is provided with AT comment information, and each comment C is provided withT(Qi,Pk) Wherein T ═ 1, 2.., AT, statistical review CT(Qi,Pk) The number of negative words is in, if the number of negative words is odd, the emotion polarity Pol (C)T(Qi,Pk) 1 is ═ 1; if the number of negative words is even, the emotion polarity Pol (C)T(Qi,Pk) 1); make statistics of the obtained comment CT(Qi,Pk) Number of middle and positive emotion words Pos (C)T(Qi,Pk) And the number of negative emotion words Neg (C)T(Qi,Pk));
Step 5.2, the commodity P is obtained by adopting the following formulakFor user QiFinal commodity recommendation degree of (Q) FREci,Pk):
Step 6, therefore, for each item P in the range of unpurchased itemskAll obtain a commodity PkFor user QiThe final recommendation values corresponding to all the commodities in the range of the commodities not purchased are arranged from high to low, and a plurality of commodities which are sorted in the front are selected and recommended to the user Qi。
2. The mixed information fusion-based emotion-aware commodity recommendation method of claim 1, wherein in step 3.1.1, the user tags representing the user characteristics include user gender, user age, user occupation, user location and user social information.
3. The mixed information fusion-based emotion-aware commodity recommendation method of claim 2, wherein the user social information includes user interest and hobby information and user personality information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010319045.0A CN111507804B (en) | 2020-04-21 | 2020-04-21 | Emotion perception commodity recommendation method based on mixed information fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010319045.0A CN111507804B (en) | 2020-04-21 | 2020-04-21 | Emotion perception commodity recommendation method based on mixed information fusion |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111507804A true CN111507804A (en) | 2020-08-07 |
CN111507804B CN111507804B (en) | 2022-05-13 |
Family
ID=71871236
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010319045.0A Active CN111507804B (en) | 2020-04-21 | 2020-04-21 | Emotion perception commodity recommendation method based on mixed information fusion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111507804B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112767085A (en) * | 2021-01-22 | 2021-05-07 | 武汉蔚来能源有限公司 | Commodity similarity analysis and commodity recommendation method, commodity similarity analysis and commodity recommendation device, and computer storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103886074A (en) * | 2014-03-24 | 2014-06-25 | 江苏名通信息科技有限公司 | Commodity recommendation system based on social media |
CN105069072A (en) * | 2015-07-30 | 2015-11-18 | 天津大学 | Emotional analysis based mixed user scoring information recommendation method and apparatus |
US20160196580A1 (en) * | 2015-01-02 | 2016-07-07 | Sk Planet Co., Ltd. | System for content recommendation service, content recommendation device and method of operating the same |
CN107077486A (en) * | 2014-09-02 | 2017-08-18 | 菲特尔销售工具有限公司 | Affective Evaluation system and method |
CN110297939A (en) * | 2019-06-21 | 2019-10-01 | 山东科技大学 | A kind of music personalization system of fusion user behavior and cultural metadata |
-
2020
- 2020-04-21 CN CN202010319045.0A patent/CN111507804B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103886074A (en) * | 2014-03-24 | 2014-06-25 | 江苏名通信息科技有限公司 | Commodity recommendation system based on social media |
CN107077486A (en) * | 2014-09-02 | 2017-08-18 | 菲特尔销售工具有限公司 | Affective Evaluation system and method |
US20160196580A1 (en) * | 2015-01-02 | 2016-07-07 | Sk Planet Co., Ltd. | System for content recommendation service, content recommendation device and method of operating the same |
CN105069072A (en) * | 2015-07-30 | 2015-11-18 | 天津大学 | Emotional analysis based mixed user scoring information recommendation method and apparatus |
CN110297939A (en) * | 2019-06-21 | 2019-10-01 | 山东科技大学 | A kind of music personalization system of fusion user behavior and cultural metadata |
Non-Patent Citations (3)
Title |
---|
HAN-CHIEH CHAO ET AL.: "A M-Learning Content Recommendation Service by Exploiting Mobile Social Interactions", 《IEEE TRANSACTIONS ON LEARNING TECHNOLOGIES》 * |
扈中凯等: "基于用户评论挖掘的产品推荐算法", 《浙江大学学报(工学版)》 * |
谭云志等: "基于用户评分和评论信息的协同推荐框架", 《模式识别与人工智能》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112767085A (en) * | 2021-01-22 | 2021-05-07 | 武汉蔚来能源有限公司 | Commodity similarity analysis and commodity recommendation method, commodity similarity analysis and commodity recommendation device, and computer storage medium |
CN112767085B (en) * | 2021-01-22 | 2024-05-24 | 武汉蔚来能源有限公司 | Commodity similarity analysis and commodity recommendation method and device and computer storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN111507804B (en) | 2022-05-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110222272B (en) | Potential customer mining and recommending method | |
US20130275269A1 (en) | Searching supplier information based on transaction platform | |
CN109711955B (en) | Poor evaluation early warning method and system based on current order and blacklist base establishment method | |
CN110020128B (en) | Search result ordering method and device | |
CN109584006B (en) | Cross-platform commodity matching method based on deep matching model | |
CN107194769A (en) | A kind of Method of Commodity Recommendation that content is searched for based on user | |
CN111460819A (en) | Personalized comment text recommendation system and recommendation method based on fine-grained sentiment analysis | |
Prasetyo | Searching cheapest product on three different e-commerce using k-means algorithm | |
CN112801803B (en) | Financial product recommendation method and device | |
CN111507804B (en) | Emotion perception commodity recommendation method based on mixed information fusion | |
CN110287410A (en) | The fusion method of a variety of proposed algorithms of user under a kind of O2O electric business scene | |
CN113837844A (en) | Multi-cascade downstream enterprise recommendation system and method and storage medium | |
CN108171589A (en) | Verification method and device | |
Pasaribu | The Shopee Application User Reviews Sentiment Analysis Employing Naïve Bayes Algorithm | |
CN115511582A (en) | Artificial intelligence based Commodity recommendation system and method | |
Hoiriyah et al. | Lexicon-Based and Naive Bayes Sentiment Analysis for Recommending the Best Marketplace Selection as a Marketing Strategy for MSMEs | |
CN115131108A (en) | E-commerce commodity screening system | |
Krisdhamara et al. | Improvement of collaborative filtering recommendation system to resolve sparsity problem using combination of clustering and opinion mining methods | |
CN114861079A (en) | Collaborative filtering recommendation method and system fusing commodity features | |
Nainggolan et al. | Sentiment; Clustering; K-Means Analysis Sentiment in Bukalapak Comments with K-Means Clustering Method | |
Xiao et al. | Hybrid Embedding of Multi-Behavior Network and Product-Content Knowledge Graph for Tourism Product Recommendation. | |
GB2498384A (en) | System and methods to generate a list of recommended items | |
CN116402565B (en) | Commodity recommendation method and system based on big data analysis | |
Eide et al. | Five lessons from building a deep neural network recommender for marketplaces | |
CN109214848B (en) | Method and system for analyzing influence similarity of virtual commodities on recommendation system |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CP02 | Change in the address of a patent holder | ||
CP02 | Change in the address of a patent holder |
Address after: 321004 No.588, Yingbin Avenue, Wucheng District, Jinhua City, Zhejiang Province Patentee after: Mo Yuchang Address before: 362021 No. 269 North China Road, Fengze District, Quanzhou City, Fujian Province Patentee before: Mo Yuchang |