CN108428156A - Net purchase user buys the generation method and system of false exponential model - Google Patents
Net purchase user buys the generation method and system of false exponential model Download PDFInfo
- Publication number
- CN108428156A CN108428156A CN201810238533.1A CN201810238533A CN108428156A CN 108428156 A CN108428156 A CN 108428156A CN 201810238533 A CN201810238533 A CN 201810238533A CN 108428156 A CN108428156 A CN 108428156A
- Authority
- CN
- China
- Prior art keywords
- false
- user
- comment
- index
- commodity
- 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.)
- Withdrawn
Links
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
- G06Q30/018—Certifying business or products
- G06Q30/0185—Product, service or business identity fraud
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention provides a kind of net purchase the user generation method and system of buying false exponential model, it generates respectively and buys false comment index, buys false Activity Index and buy false public opinion index, it wherein generates and buys false comment index, especially by the user's comment on commodity text data and open press media data obtained on electric business platform;Pretreatment through data, after obtaining user's comment on commodity text and open medium public sentiment text data, carry out data prediction, after identifying and excluding all kinds of interference data, buy false comment identification, buy false behavior type and judge to show that net purchase user buys false comment index after buying false public sentiment attitude trend analysis with different subjects;The calculating data that the net purchase user that this method is generated buys false exponential model easily obtain, it is unique, relatively stable that model exports result, and can be in the dynamic monitoring and prediction section time, particular commodity, the specific consumer group buy false comment state, buy and false behavior and buy false public sentiment distribution characteristics and overall development state.
Description
Technical field
The present invention relates to generation methods and system that a kind of net purchase user buys false exponential model.
Background technology
According to relevant search and discovery is compared, the existing big index design of net purchase user has with establishment technique:
2015, Alibaba intercepted restocking based on actively discovering in annual risk control on line, consumer complains
In the sell-fake-products goods links that are verified, and close up to a million are punished by Ali behind and are accused of sell-fake-products commodity data, it is constructed
China fake products active index map.But the data of the fake products active index mainly pass through the basic letter to publishing commodity
Breath is obtained with consumer's complaint, for the essential information of publishing commodity, including text information and pictorial information, does not include key
Word, and consumer do not complain buy vacation, also referred to as know that vacation is bought in vacation, or buy vacation and do not know vacation etc., cannot be monitored.
The above problem is should to pay attention to and solve the problems, such as in the generating process that net purchase user buys false exponential model.
Invention content
The object of the present invention is to provide generation methods and system that a kind of net purchase user buys false exponential model, can concentrated expression
Different platform, different regions, different commodity classification, different crowd consumer buy false behavior and buy false Behavior and attitude tendency sequential
Index model is used by buying false comment index, buying false Activity Index and buying false public opinion index come comprehensive monitoring and prediction net purchase
The development trend for buying false behavior and its attitude at family, solve the essential information existing in the prior art by publishing commodity with
Consumer's complaint obtains, for the essential information of publishing commodity, including text information and pictorial information, does not include keyword
, and consumer do not complain the problem of buying vacation, cannot be monitored.
Technical solution of the invention is:
A kind of net purchase user buys the generation method of false exponential model, generates respectively and buys false comment index, buys false Activity Index
With buy false public opinion index, include the following steps:
S1, generation buy false comment index, specially:
User's comment on commodity text data on S11, acquisition electric business platform;
The pretreatment of S12, data after obtaining user's comment on commodity text data, carry out data prediction, identify and exclude
The brush comment data of the existing promise breaking in violation of rules and regulations of shop businessman, to obtain validated user comment data;
S13, the identification for buying false comment text obtain validated user comment data and then according to user comment sentence
Similarity algorithm come calculate and identify these user's comment on commodity data content whether be buy false comment, and count buy false comment
Quantity;
S14, the generation for buying false comment index;
False Activity Index is bought in S2, generation, specially:
S21, data acquisition is complained, including the user's Merchandise Complaint data for obtaining electric business platform and user's Merchandise Complaint under line
Data;
The pretreatment of S22, data after the complaint data for obtaining step S21, determine to complain and belong to net purchase user in data and buy
False report user, and relevant rudimentary information of putting on record (including net purchase user buys false report user's name, gender, age, platform account
ID, retail shop's title where platform names, shopping where trade name, shopping), it is standby that false report user is then bought by net purchase user
Case basic information traces and identifies that non-fraudulent buys false behavior and fraudulent buys false behavior;
S23, production net purchase user buy false Activity Index, and establishment buys false Activity Index with net purchase user's fraudulent is calculated;
False public opinion index is bought in S3, generation, specially:
S31, it carries out buying false public sentiment data acquisition, and collected network public-opinion data is subjected to text analyzing and theme point
Class;
S32, it is divided into fraudulent according to different themes and buys false and non-fraudulent and buy this false two principal themes to count three classes master
Body includes that mainstream news, entrepreneur and common net this three main bodies buy false public sentiment, and pass through network public-opinion fuzzy evaluation mould
Type is judged into row index, calculates the public opinion index of each index;
S33, different themes, the network public-opinion composite index of different subjects are calculated by weighting algorithm.
Further, in step S12, data prediction is specially:
S121, businessman is removed from comment data;
S122, the removal comment nonstandard comment data of number of words;
The identical comment data of S123, removal comment word;
S124, the comment data for removing violation brush screen.
Further, in step S13, similarity algorithm by counterfeit goods Type division structure net purchase user specifically, bought
False corpus is realized by Keywords matching and is identified.
Further, it in the similarity algorithm of step S13, for complex text, establishes and bogus subscriber is bought based on corpus
Comment identification case based reasoning model is identified.
Further, step S14 specifically,
S141, the net purchase user for generating the shops f classification i commodity j buy false comment index:After false comment entry is bought in acquisition, remove
All users to get the net purchase commodity are gone out are commented on validated user and buy false comment ratio value, are equal toWherein,
mBCNet purchase user to buy the commodity buys false number of reviews, and m is the effective number of reviews of net purchase user for buying the commodity;
S142, the net purchase user for generating f classification i commodity buy false comment index:It is equal to all under the i commodity of f classifications
Tired and average value, the calculation formula that the net purchase user in shop buys the ratio value of false comment be:
Wherein, n is f classes all shop quantity of i commodity now;
S143, the net purchase user for generating f classification commodity buy false comment index:It is equal to the net of all commodity of f classes now
Tired and average value, the calculation formula that purchase user buys the ratio value of false comment be:
Wherein, n is f classes all shop quantity of i commodity now, and k is the quantity of f classes all commodity now;
S144, generation net purchase user buy false comment index:It is bought vacation equal to the net purchase user of all classification net purchase commodity and comments
Tired and average value, the calculation formula of the ratio value of opinion be:
Wherein, 19 be commodity classification main body quantitative value, and n is f classes all shop quantity of i commodity now, and k is f classes institute now
There is the quantity of commodity.
Further, in step S22, data prediction is specially:
S221, confirm that true net purchase user's buys false complaint, that completes these true net purchase users buys what vacation was complained
Record information acquires, including the historical trading of the basic information of individual subscriber, the basic information of complained businessman, individual subscriber,
The historical review information of commodity corresponding to the historical transactional information of complained businessman, complained businessman, complained businessman its
All user bases letter in the historical review information of commodity corresponding to the historical review information of his commodity, complained businessman
Breath;
S222, identify and distinguish that being complained non-fraudulent in comment on commodity buys by text mining and machine learning algorithm
Bogus subscriber buys bogus subscriber with fraudulent, and show that non-fraudulent buys bogus subscriber's ratio in complained comment on commodity and fraudulent buys vacation
User's ratio;Wherein identify and distinguish that being complained non-fraudulent in comment on commodity buys with machine learning algorithm by text mining
Bogus subscriber buys bogus subscriber with fraudulent, specially:
First, it is determined that whether net purchase user is comment user, then continue to judge whether net purchase user is that difference comments use in this way
Family, in this way it is poor comment user then net purchase user be fraudulent buy bogus subscriber, if net purchase user is not that difference comments user, then the net purchase
User be favorable comment user or in comment user, at this moment need to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, judgement obtain whether
Bogus subscriber is bought for non-fraudulent;
Such as judge net purchase user not and be to comment on user, determines whether customer service report user, customer service report user in this way,
Then net purchase user is that fraudulent buys bogus subscriber;If not being customer service report user, then net purchase user is not find to be spoofed use
Family or non-fraudulent buy bogus subscriber, at this moment need to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, and judgement obtains whether be non-
Fraudulent buys bogus subscriber or does not find that the fraudulent being spoofed buys bogus subscriber.
Further, step S23, specifically,
S231, generation f classification i commodity net purchase user's fraudulents buy false complaint index
Wherein,The complaint after f classification i counterfeit goods, non-favorable comment and ignorant are bought for t phase users
Total number of persons,The total number of persons of f classification i counterfeit goods is bought for t phase users,For the base period, user buys f
Complaint, non-favorable comment after classification i counterfeit goods and unwitting total number of persons,F classifications i is bought for base period user
The total number of persons of counterfeit goods;
S232, generation f classification commodity net purchase user's fraudulents buy false complaint index:
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S233, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value, can refer to Fig. 7 commodity classifications and draws to calculate;
S234, the generation non-fraudulent of f classification i commodity net purchase users buy false index
Wherein,Not complaining after f classification i counterfeit goods is bought for t phase users, is not commented on and non-difference is commented
By total number of persons,The total number of persons after f classification i counterfeit goods is bought for t phase users,T buys for base period user
Not complaining after to f classification i counterfeit goods is not commented on and non-difference comment total number of persons,F classes are bought for base period user
Total number of persons after mesh i counterfeit goods;
S235, the generation non-fraudulent of f classification commodity net purchase users buy false index,
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S236, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value, can refer to Fig. 7 commodity classifications and draws to calculate;
Further, step S31 to buying false public sentiment data specifically, carry out two themes and three main bodies classification, specifically
For, text analyzing pretreatment is carried out to buying false public sentiment data, according to the once division methods of the theme in public sentiment subject classification system,
Carry out automatic Text Categorizations using 1 grade and 2 grades themes of support vector machines pair, using 3 grades of descriptor of Keywords matching algorithm pair into
The automatic classification for search of row, includes the following steps,
S311, according to different types of network information classification, advance collecting part text data respectively, and carry out keyword
Language and abstract extraction;
S312, the mode based on man-computer cooperation, the key words to acquiring text in advance carry out two theme class with abstract
It Biao Ji not be marked with three principal classes, constantly training SVM buys false public sentiment subject classification device, and forms svm classifier model;
S313, it after test text is carried out key words abstract extraction, is directed respectively into svm classifier model, carries out automatic
Buy false public sentiment classification.
Further, step S33 specifically,
S331, mainstream news public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is trade name
Home identity,False public sentiment event mainstream news quantity is bought for the base period,False public sentiment event mainstream news number is bought for the t phases
Amount, mainstream news buy false public sentiment combined index and buy false public sentiment thing equal to area, platform, firm name, commodity classification and trade name
Part mainstream news public sentiment separate index number adding up and being averaged;
S332, entrepreneur's public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is trade name
Home identity,False public sentiment event entrepreneur is bought for the base period to express an opinion quantity,False public sentiment event enterprise is bought for the t phases
Industry personage expresses an opinion quantity, entrepreneur buy false public sentiment combined index be equal to area, platform, firm name, commodity classification and
Trade name buys the cumulative and average of pseudo event public sentiment separate index number;
S333, common netizen's public opinion index is generated:
Wherein, wherein i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is commodity
Title home identity,False public sentiment event money order receipt to be signed and returned to the sender is bought for base period common netizen, is posted, the cumulative and average value of touching quantity,For t phases common netizen buy false public sentiment event money order receipt to be signed and returned to the sender, be posted, touching quantity it is cumulative and average, common netizen buys false public sentiment
Combined index is equal to the cumulative peace that area, platform, firm name, commodity classification and trade name buy pseudo event public sentiment separate index number
.
A kind of net purchase user of generation method that buying false exponential model using net purchase user described in any one of the above embodiments buys vacation
The generation system of exponential model, including buy false comment index generation module, buy false Activity Index generation module and buy false public sentiment and refer to
Number generation module, includes the following steps:
Buy false comment index generation module:Obtain user's comment on commodity text data on electric business platform;Obtain user quotient
After judging paper notebook data, data prediction is carried out, identifies and exclude the brush comment data of the existing promise breaking in violation of rules and regulations of shop businessman,
To obtain validated user comment data;The identification of false comment text is bought, obtains validated user comment data and then according to user
The similarity algorithm of sentence is commented on come whether calculate and identify the content of these user's comment on commodity data be to buy false comment, and count
Number buys false number of reviews;False comment index is bought in generation.
Buy false Activity Index generation module:Data acquisition is complained, including obtains user's Merchandise Complaint data of electric business platform
With user's Merchandise Complaint data under line;The pretreatment of data after the complaint data for obtaining step S21, determines to complain and belongs in data
False complaint is bought in net purchase user;The generation of false Activity Index is bought, including f classification i commodity net purchase users fraudulent, non-fraudulent are bought
False Activity Index, f classification commodity net purchase users fraudulent, non-fraudulent buy false Activity Index, net purchase user fraudulent, non-fraud
Property buys false Activity Index;
Buy false public opinion index generation module:Buy false public sentiment data acquisition, and by collected network public-opinion data into
Row text analyzing and subject classification;Referred to according to three human subjects and two principal themes, and by network public-opinion Fuzzy evaluation mode
Mark is judged, and the public opinion index of each index is calculated;The public sentiment composite index of each subject network is calculated by weighting algorithm.
The beneficial effects of the invention are as follows:This kind of net purchase user buys the generation method and system of false exponential model, is generated
The calculating data that net purchase user buys false exponential model easily obtain, model export result it is unique, relatively stable, and can dynamic monitoring with
It predicts in certain time, particular commodity, the specific consumer group buy false comment state, buy and false behavior and buy false public sentiment distribution characteristics
And overall development state.
Description of the drawings
Fig. 1 is the flow diagram for the generation method that net purchase user of the embodiment of the present invention buys false exponential model.
Fig. 2 is the flow diagram of data acquisition in embodiment.
Fig. 3 is that the pretreated flow diagram of false comment data is bought in embodiment.
Fig. 4 is that the flow diagram for buying false public opinion index is generated in embodiment.
Fig. 5 is the schematic diagram palmed off (counterfeit) type of merchandise in embodiment and divide table.
Fig. 6 be in embodiment net purchase user buy false index index system illustrate schematic diagram.
Fig. 7 is the schematic diagram of net purchase commodity scheme of classes in embodiment.
Fig. 8 is the schematic diagram that fraudulent buys bogus subscriber's recognizer with non-fraudulent in embodiment.
Specific implementation mode
The preferred embodiment that the invention will now be described in detail with reference to the accompanying drawings.
Embodiment
In embodiment, it refers to buying the user of fake products to express and itself see by way of comment on commodity to buy false comment index
The index model that the quantitative proportion of method is changed over time with distribution.It refers to that net purchase user's fraudulent buys false behavior to buy false Activity Index
The ratio of false behavior is bought with non-fraudulent and index model that distribution changes over time.It refers to different society master to buy false public opinion index
Body (consumer, news media and businessman enterprise) buys the comment of false behavior to net purchase user and attitude is inclined to evaluation model.
In embodiment, since comment buys false index, buys false Activity Index and buys the required data of false public opinion index not
Together, and specific establishment and computational methods are there are larger difference, and embodiment will generate three above index respectively.
A kind of net purchase user buys the generation method of false exponential model, generates respectively and buys false comment index, buys false Activity Index
With buy false public opinion index, such as Fig. 1, include the following steps:
S1, generation buy false comment index, specially:
User's comment on commodity text data on S11, acquisition electric business platform;User's comment on commodity text data includes user
Log-on message, customer transaction information, platform name, trade name, commodity category name, store name, wherein user's registration information is inclusive
Not, age, area, customer transaction information include transaction count, trading volume, turnover.
User's comment on commodity text data is user's comment on commodity text on major electric business platform such as Ali, Jingdone district, Suning
Notebook data.These data can provide specific data interface API by various regions electric business and obtain, can also utilize web crawlers technology into
Row obtains.Phase is retrieved and obtained according to this to the flow chart that specific data acquisition step can be given according to Fig. 2 according to keyword
Close data, wherein the oriented arrow of solid line is data retrieval and obtains route, and dotted line indicates bi-directional association, that is, obtains corresponding data
After two-way can retrieve.
The pretreatment of S12, data after obtaining user's comment on commodity text data, carry out data prediction, identify and exclude
The brush comment data of the existing promise breaking in violation of rules and regulations of shop businessman, to obtain validated user comment data;
Obtain particular platform, specific classification, particular commodity, specific shop the useful comment data of institute after, first shown in Fig. 3
Data prediction is carried out, emphasis is the brush comment data for identifying and excluding shop businessman violation promise breaking that may be present, to carry
The accuracy that high index calculates.To obtain effective comment data.Really, the discriminating of effective comment data further includes deleting businessman
Answer, advertisement and the comment data unrelated with this commodity.
In step S12, data prediction is specially:
S121, businessman is removed from comment data;
S122, the removal comment nonstandard comment data of number of words;
The identical comment data of S123, removal comment word;
S124, the comment data for removing violation brush screen.
S13, the identification for buying false comment text obtain validated user comment data and then according to user comment sentence
Similarity algorithm come calculate and identify these user's comment on commodity data content whether be buy false comment, and count buy false comment
Quantity;
It obtains validated user comment data and then is calculated according to the similarity algorithm of user comment sentence (keyword)
With identify whether the content that is illustrated of these user's comments on commodity is to buy false comment, and count and buy false number of reviews.Here knowledge
Other method need to divide table by personation (counterfeit) type of merchandise in Fig. 5, and a constructed net purchase user buys false corpus, passes through pass
Keyword matching achieves that identification.For complex text, a bogus subscriber that buys based on corpus can be established and comment on identification
Case based reasoning model is identified.
S14, the generation for buying false comment index;Step S14 specifically,
S141, the net purchase user for generating the shops f classification i commodity j buy false comment index:After false comment entry is bought in acquisition, remove
All users to get the net purchase commodity are gone out are commented on validated user and buy false comment ratio value, are equal toWherein,
mBCNet purchase user to buy the commodity buys false number of reviews, and m is the effective number of reviews of net purchase user for buying the commodity;
S142, the net purchase user for generating f classification i commodity buy false comment index:It is equal to all under the i commodity of f classifications
Tired and average value, the calculation formula that the net purchase user in shop buys the ratio value of false comment be:
Wherein, n is f classes all shop quantity of i commodity now;
S143, the net purchase user for generating f classification commodity buy false comment index:It is equal to the net of all commodity of f classes now
Tired and average value, the calculation formula that purchase user buys the ratio value of false comment be:
Wherein, n is f classes all shop quantity of i commodity now, and k is the quantity of f classes all commodity now;
S144, generation net purchase user buy false comment index:It is bought vacation equal to the net purchase user of all classification net purchase commodity and comments
Tired and average value, the calculation formula of the ratio value of opinion be:
Wherein, n is f classes all shop quantity of i commodity now, and 19 be commodity classification main body quantitative value, and k is f classes institute now
There is the quantity of commodity.
In step S14, other attributes net purchase user buys the calculating of false comment index, follow-up group index and composite index
Establishment is consistent with calculating with the above process.The difference is that region-by-region index calculates, the area for distinguishing net purchase user is needed
Belong to ID, the index that crowd characteristic belongs to is divided to calculate the gender for needing to distinguish user and age.
False Activity Index is bought in S2, generation, specially:
S21, data acquisition is complained, including the user's Merchandise Complaint data for obtaining electric business platform and user's Merchandise Complaint under line
Data;Complaint data needed for the index construction are still the user that electric business platform is put on record on major line such as Ali, Jingdone district, Suning
User's Merchandise Complaint data under the line that Merchandise Complaint data and 315 equal consumers' right-safeguarding mechanisms put on record.
The pretreatment of S22, data after the complaint data for obtaining step S21, determine to complain and belong to net purchase user in data and buy
False report user, and relevant rudimentary information of putting on record (including net purchase user buys false report user's name, gender, age, platform account
ID, retail shop's title where platform names, shopping where trade name, shopping), it is standby that false report user is then bought by net purchase user
Case basic information traces and identifies that non-fraudulent buys false behavior and fraudulent buys false behavior;
In step S22, data prediction is specially:
S221, confirm that true net purchase user's buys false complaint, that completes these true net purchase users buys what vacation was complained
Record information acquires, including the historical trading of the basic information of individual subscriber, the basic information of complained businessman, individual subscriber,
The historical review information of commodity corresponding to the historical transactional information of complained businessman, complained businessman, complained businessman its
All user bases letter in the historical review information of commodity corresponding to the historical review information of his commodity, complained businessman
Breath;Wherein, the basic information of individual subscriber includes the information such as User ID, age, purchase commodity, is complained the basic information of businessman
Including Merchant ID, shop title etc..
S222, identify and distinguish that being complained non-fraudulent in comment on commodity buys by text mining and machine learning algorithm
Bogus subscriber buys bogus subscriber with fraudulent, and show that non-fraudulent buys bogus subscriber's ratio in complained comment on commodity and fraudulent buys vacation
User's ratio;Wherein identify and distinguish that being complained non-fraudulent in comment on commodity buys with machine learning algorithm by text mining
Bogus subscriber buys bogus subscriber with fraudulent, specially:
First, it is determined that whether net purchase user is comment user, then continue to judge whether net purchase user is that difference comments use in this way
Family, in this way it is poor comment user then net purchase user be fraudulent buy bogus subscriber, if net purchase user is not that difference comments user, then the net purchase
User be favorable comment user or in comment user, at this moment need to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, judgement obtain whether
Bogus subscriber is bought for non-fraudulent;
Such as judge net purchase user not and be to comment on user, determines whether customer service report user, customer service report user in this way,
Then net purchase user is that fraudulent buys bogus subscriber;If not being customer service report user, then net purchase user is not find to be spoofed use
Family or non-fraudulent buy bogus subscriber, at this moment need to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, and judgement obtains whether be non-
Fraudulent buys bogus subscriber or does not find that the fraudulent being spoofed buys bogus subscriber.
It in step S22, obtains after complaining data, it is thus necessary to determine which is complained data to belong to net purchase user and buys false complaint.
At this point, correlation functional department of government is needed to cooperate with each electric business platform, further determines that these net purchases user buys and false which complains be
True.And put on record clue based on these true case history, it traces and put on record its history merchandise news, Transaction Information, comment
Information and calling information etc..Then bogus subscriber is bought by distinguishing non-fraudulent according to the algorithm of Fig. 8 and text mining algorithm
(user knows that purchased commodity are fake products) ratio and fraudulent buy bogus subscriber (it is fake products that user, which is unaware of purchased commodity) ratio.This
Shi Yiran can be distinguished the counterfeit commodity of personation, counterfeit or inferior quality goods, commodity of evading the tax, violated commodity and non-conformity article etc. by Fig. 5
Fraudulent buys false and non-fraudulent and buys false ratio and distribution situation.
In view of the consumer that majority is spoofed can be complained or be commented on afterwards afterwards, therefore this part only provides fraudulent
Buy establishment and the calculating process of false Activity Index, particular content and buying false comment index establishment and calculating in step S14
Journey is similar.It is all scouted it is true and decide on a verdict all sell-fake-products businessman historical reviews on the basis of, exclude and be spoofed user
It is remaining to substantially belong to know that vacation is bought after (can be commented by whether carrying out difference afterwards, if subsequent to carry out complaint and right-safeguarding to exclude)
Bogus subscriber.The calculating of its calculating process and fraudulent user index is identical.
S23, the generation for buying false Activity Index, including f classification i commodity net purchase users fraudulent, non-fraudulent buy false behavior
Index, f classification commodity net purchase users fraudulent, non-fraudulent buy false Activity Index, net purchase user fraudulent, non-fraudulent and buy vacation
Activity Index;Step S23, specifically,
S231, generation f classification i commodity net purchase user's fraudulents buy false complaint index
Wherein,The complaint after f classification i counterfeit goods, non-favorable comment and ignorant are bought for t phase users
Total number of persons,The total number of persons of f classification i counterfeit goods is bought for t phase users,For the base period, user buys f
Complaint, non-favorable comment after classification i counterfeit goods and unwitting total number of persons,F classifications i is bought for base period user
The total number of persons of counterfeit goods;
S232, generation f classification commodity net purchase user's fraudulents buy false complaint index:
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S233, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value, can refer to Fig. 7 commodity classifications and draws to calculate;
S234, the generation non-fraudulent of f classification i commodity net purchase users buy false index
Wherein,Not complaining after f classification i counterfeit goods is bought for t phase users, is not commented on and non-difference is commented
By total number of persons,The total number of persons after f classification i counterfeit goods is bought for t phase users,T buys for base period user
Not complaining after to f classification i counterfeit goods is not commented on and non-difference comment total number of persons,F classes are bought for base period user
Total number of persons after mesh i counterfeit goods;
S235, the generation non-fraudulent of f classification commodity net purchase users buy false index,
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S236, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value, can refer to Fig. 7 commodity classifications and draws to calculate;
In step S23, other attribute net purchase user's fraudulents buy the false calculating for complaining index, follow-up group index and synthesis
The establishment of index is consistent with calculating with the above process.The difference is that region-by-region index calculates, need to distinguish net purchase user
Area ownership ID, divide crowd characteristic belong to index calculate need distinguish user gender and the age.
False public opinion index, such as Fig. 4 are bought in S3, generation, specially:
S31, it carries out buying false public sentiment data acquisition, and collected network public-opinion data is subjected to text analyzing and theme point
Class;Buying false public sentiment data includes:News, user and enterprise post content, time of posting, quantity of posting, money order receipt to be signed and returned to the sender time, money order receipt to be signed and returned to the sender number
Amount, hits, money order receipt to be signed and returned to the sender number, forwarding number.
Step S31 to buying false public sentiment data specifically, carry out two themes and three main bodies classification, specifically, to buying vacation
Public sentiment data carries out text analyzing pretreatment, according to the once division methods of the theme in public sentiment subject classification system, using support
1 grade and 2 grades themes of vector machine pair carry out automatic Text Categorization, are searched automatically using 3 grades of descriptor of Keywords matching algorithm pair
Rope is classified, and is included the following steps,
S311, according to different types of network information classification, advance collecting part text data respectively, and carry out keyword
Language and abstract extraction;
S312, the mode based on man-computer cooperation, the key words to acquiring text in advance carry out two theme class with abstract
It Biao Ji not be marked with three principal classes, constantly training SVM buys false public sentiment subject classification device, and forms svm classifier model;
S313, it after test text is carried out key words abstract extraction, is directed respectively into svm classifier model, carries out automatic
Buy false public sentiment classification.
It should be noted that unified coding mode must be used by buying the process of false public sentiment classification, need to carry out text
Participle, the crucial table and text snippet of generation are also required to carry out the filtering of stop words.
S32, it is divided into fraudulent according to different themes and buys false and non-fraudulent and buy this false two principal themes to count three classes master
Body includes that mainstream news, entrepreneur and common net this three main bodies buy false public sentiment, and pass through network public-opinion fuzzy evaluation mould
Type is judged into row index, calculates the public opinion index of each index.
The calculating of false public opinion index three-level index, including two types are bought, the first is directly to pass through network data acquisition
Data are counted.Such as buy the quantity that false public sentiment is released news.Another kind is to need to carry out the text-processing of depth and semanteme
The calculating of corresponding index is carried out after identification again.Such as buy the calculating of false affective disposition.
Net purchase user buy false public sentiment affective disposition calculating mainly respectively to news media, enterprise and consumer this three
The support of false public sentiment is bought in the net purchase of human subject, the neutral tendentiousness with opposition is quantified, calculated and judged.Such as to each whole nation
The net purchase of big mainstream media is bought false evental news report text and is segmented, after obtaining Feature Words (keyword) and news in brief,
Part-of-speech tagging and sentiment analysis are carried out again, you can calculate each news media buys false public sentiment attitude tendency.Net purchase to enterprise
The calculating that user buys false public sentiment mainly knows the advanced pipe such as news media of enterprise itself distribution platform and its CEO to area in all parts of the country
The news report text for buying false problem of reason personnel segments and sentiment analysis, calculates its public sentiment tendency.To all parts of the country
The calculating of false public sentiment is bought in area's consumer's net purchase, mainly carries out buying false attitude tendency with money order receipt to be signed and returned to the sender information according to posting for the network user
It calculates.
Step S33 specifically,
S331, mainstream news public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is trade name
Home identity,False public sentiment event mainstream news quantity is bought for the base period,False public sentiment event mainstream news number is bought for the t phases
Amount, mainstream news buy false public sentiment combined index and buy false public sentiment thing equal to area, platform, firm name, commodity classification and trade name
Part mainstream news public sentiment separate index number adding up and being averaged;
S332, entrepreneur's public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is trade name
Home identity,False public sentiment event entrepreneur is bought for the base period to express an opinion quantity,False public sentiment event enterprise is bought for the t phases
Industry personage expresses an opinion quantity, entrepreneur buy false public sentiment combined index be equal to area, platform, firm name, commodity classification and
Trade name buys the cumulative and average of pseudo event public sentiment separate index number;
S333, common netizen's public opinion index is generated:
Wherein, wherein i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is commodity
Title home identity,False public sentiment event money order receipt to be signed and returned to the sender is bought for base period common netizen, is posted, the cumulative and average value of touching quantity,For t phases common netizen buy false public sentiment event money order receipt to be signed and returned to the sender, be posted, touching quantity it is cumulative and average, common netizen buys false public sentiment
Combined index is equal to the cumulative peace that area, platform, firm name, commodity classification and trade name buy pseudo event public sentiment separate index number
.
False exponential model design, which is bought, about net purchase user is related safeguarding-rights act index to the maximum difficult point of establishment technique
It selects, refine, embodying and designed and establishment with the computation model of specific sub- index.Embodiment is directed to these technological difficulties, provides
Each index Design and detailed computation model solve the technological difficulties that specific sub- index is calculated with combined index.
The index system that net purchase user buys false index is first divided into and buys false comment index, buys false Activity Index and buy false carriage
Three two-level index of feelings index, wherein each two-level index is divided into combined index and group index again, number when minimum index granularity
Amount, the amount of money and ratio, and this four lateral indexs can be belonged to from area ownership, platform ownership, classification ownership and crowd and handed over
Fork, particular content are as shown in Figure 6.Wherein separate index number and combined index are in contrast, mainly to indicate by what classification dimension
Degree is divided, and combined index is only the cumulative mean of its separate index number.For example, the input data that model calculates, is marked through area
After label identification, the commodity ship-to for all pointing to user is from some identical area i, and press step S1, S2, S3 and other
Calculate separately out step by step and buy false Index A i, the as areas i net purchase user buys false separate index number.Pass through above-mentioned multiple areas
Ai index cumulative mean mathematics calculate, that is, calculate all regions and buy false combined index A.It is such, by platform ownership, quotient
Category mesh belongs to and the calculating logic of the relationship of the defined separate index number of user crowd's ownership and combined index is also identical.
It should be noted that area ownership, is to be divided by national province, city, county regional areas, is divided into county's level-one;Platform
Ownership divides, and is divided according to the platform enterprises title such as Suning, Jingdone district and Ali, and commodity classification ownership is according to national work
The commodity classification for the commodity classification and several electric business platforms that quotient office provides is modified the criteria for classifying obtained later, such as Fig. 7;
Crowd, which belongs to, to be divided, and is mainly divided from two dimensions of the age of user and gender.
A kind of net purchase user of generation method that buying false exponential model using net purchase user described in any one of the above embodiments buys vacation
The generation system of exponential model, including buy false comment index generation module, buy false Activity Index generation module and buy false public sentiment and refer to
Number generation module, includes the following steps:
Buy false comment index generation module:Obtain user's comment on commodity text data on electric business platform;Obtain user quotient
After judging paper notebook data, data prediction is carried out, identifies and exclude the brush comment data of the existing promise breaking in violation of rules and regulations of shop businessman,
To obtain validated user comment data;The identification of false comment text is bought, obtains validated user comment data and then according to user
The similarity algorithm of sentence is commented on come whether calculate and identify the content of these user's comment on commodity data be to buy false comment, and count
Number buys false number of reviews;False comment index is bought in generation.
Buy false Activity Index generation module:Data acquisition is complained, including obtains user's Merchandise Complaint data of electric business platform
With user's Merchandise Complaint data under line;The pretreatment of data after the complaint data for obtaining step S21, determines to complain and belongs in data
False complaint is bought in net purchase user;The generation of false Activity Index is bought, including f classification i commodity net purchase users fraudulent, non-fraudulent are bought
False Activity Index, f classification commodity net purchase users fraudulent, non-fraudulent buy false Activity Index, net purchase user fraudulent, non-fraud
Property buys false Activity Index;
Buy false public opinion index generation module:Buy false public sentiment data acquisition, and by collected network public-opinion data into
Row text analyzing and subject classification;Referred to according to three human subjects and two principal themes, and by network public-opinion Fuzzy evaluation mode
Mark is judged, and the public opinion index of each index is calculated;The public sentiment composite index of each subject network is calculated by weighting algorithm.
The net purchase user of embodiment buys the generation method and system of false exponential model, and obtained net purchase user buys false index
The calculating data of model easily obtain, model export result it is unique, relatively stable, and can in the dynamic monitoring and prediction section time,
Particular commodity, the specific consumer group buy false comment state, buy and false behavior and buy false public sentiment distribution characteristics and overall development state.
Claims (10)
1. a kind of net purchase user buys the generation method of false exponential model, it is characterised in that:It generates respectively and buys false comment index, buys vacation
Activity Index and false public opinion index is bought, included the following steps:
S1, generation buy false comment index, specially:
User's comment on commodity text data on S11, acquisition electric business platform;
The pretreatment of S12, data after obtaining user's comment on commodity text data, carry out data prediction, identify and exclude shop
The brush comment data of the existing promise breaking in violation of rules and regulations of businessman, to obtain validated user comment data;
S13, the identification for buying false comment text obtain validated user comment data and then according to the similar of user comment sentence
Algorithm is spent to calculate and identify whether the content of these user's comment on commodity data is to buy false comment, and count and buy false comment number
Amount;
S14, the generation for buying false comment index;
False Activity Index is bought in S2, generation, specially:
S21, data acquisition is complained, including the user's Merchandise Complaint data for obtaining electric business platform and user's Merchandise Complaint number under line
According to;
The pretreatment of S22, data after the complaint data for obtaining step S21, determine to complain and belong to net purchase user in data and buy false throwing
It tells user, and relevant rudimentary information of putting on record, false report user is then bought by net purchase user and is put on record basic information, traces and identifies
Go out that non-fraudulent buys false behavior and fraudulent buys false behavior;
S23, production net purchase user buy false Activity Index, and establishment buys false Activity Index with net purchase user's fraudulent is calculated;
False public opinion index is bought in S3, generation, specially:
S31, it carries out buying false public sentiment data acquisition, and collected network public-opinion data is subjected to text analyzing and subject classification;
S32, it is divided into fraudulent according to different themes and buys false and non-fraudulent and buy this false two principal themes to count three human subject packets
Include the false public sentiment of buying of mainstream news, entrepreneur and common net this three main bodies, and by network public-opinion Fuzzy evaluation mode into
Row index is judged, and the public opinion index of each index is calculated;
S33, different themes, the network public-opinion composite index of different subjects are calculated by weighting algorithm.
2. net purchase user as described in claim 1 buys the generation method of false exponential model, it is characterised in that:In step S12, number
Data preprocess is specially:
S121, businessman is removed from comment data;
S122, the removal comment nonstandard comment data of number of words;
The identical comment data of S123, removal comment word;
S124, the comment data for removing violation brush screen.
3. net purchase user as described in claim 1 buys the generation method of false exponential model, it is characterised in that:In step S13, phase
Like degree algorithm specifically, buying false corpus by counterfeit goods Type division structure net purchase user, is realized and known by Keywords matching
Not.
4. net purchase user as claimed in claim 3 buys the generation method of false exponential model, it is characterised in that:The phase of step S13
Like in degree algorithm, for complex text, establishes bogus subscriber's comment identification case based reasoning model of buying based on corpus and be identified.
5. net purchase user as described in claim 1 buys the generation method of false exponential model, it is characterised in that:Step S14 is specific
For,
S141, the net purchase user for generating the shops f classification i commodity j buy false comment index:After false comment entry is bought in acquisition, Chu Yiyou
Effect user comment buys false comment ratio value to get all users for going out the net purchase commodity, is equal toWherein, mBCFor
The net purchase user for buying the commodity buys false number of reviews, and m is the effective number of reviews of net purchase user for buying the commodity;
S142, the net purchase user for generating f classification i commodity buy false comment index:It is equal to all shops under the i commodity of f classifications
Tired and average value, the calculation formula of the net purchase user ratio value of buying false comment be:
Wherein, n is f classes all shop quantity of i commodity now;
S143, the net purchase user for generating f classification commodity buy false comment index:It is equal to the net purchase of all commodity of f classes now and uses
Tired and average value, the calculation formula that the ratio value of false comment is bought at family be:
Wherein, n is f classes all shop quantity of i commodity now, and k is the quantity of f classes all commodity now;
S144, generation net purchase user buy false comment index:It is equal to the net purchase user of all classification net purchase commodity and buys false comment
Tired and average value, the calculation formula of ratio value be:
Wherein, 19 be commodity classification main body quantitative value, wherein n is f classes all shop quantity of i commodity now, and k is f classes institute now
There is the quantity of commodity.
6. net purchase user as claimed in claim 5 buys the generation method of false exponential model, it is characterised in that:In step S22, number
Data preprocess is specially:
S221, confirm that true net purchase user's buys false complaint, that completes these true net purchase users buys putting on record for false complaint
Information collection, including the historical trading of the basic information of individual subscriber, the basic information of complained businessman, individual subscriber, thrown
Tell historical review information, complained businessman other quotient of the commodity corresponding to the historical transactional information of businessman, complained businessman
All user base informations in the historical review information of commodity corresponding to the historical review information of product, complained businessman;
S222, identify and distinguish that being complained non-fraudulent in comment on commodity buys false use by text mining and machine learning algorithm
Bogus subscriber is bought in family with fraudulent, and show that non-fraudulent buys bogus subscriber's ratio in complained comment on commodity and fraudulent buys bogus subscriber
Ratio;Wherein identify and distinguish that being complained non-fraudulent in comment on commodity buys false use by text mining and machine learning algorithm
Bogus subscriber is bought in family with fraudulent, specially:
First, it is determined that whether net purchase user is comment user, then continue to judge whether net purchase user is that difference comments user in this way, such as
It is that difference comments user then net purchase user is that fraudulent buys bogus subscriber, if net purchase user is not that difference comments user, then net purchase user
For favorable comment user or in comment user, at this moment need to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, judgement obtain whether be non-
Fraudulent buys bogus subscriber;
It is to determine whether that customer service report user, customer service report user in this way then should to comment on user such as to judge net purchase user not
Net purchase user is that fraudulent buys bogus subscriber;If not being customer service report user, then net purchase user be do not find to be spoofed user or
Non- fraudulent buys bogus subscriber, at this moment needs to buy bogus subscriber's behavioural characteristic library with vacation is known and be compared, and judgement obtains whether be non-fraud
Property buys bogus subscriber or does not find that the fraudulent being spoofed buys bogus subscriber.
7. net purchase user as claimed in claim 6 buys the generation method of false exponential model, it is characterised in that:Step S23, specifically
For,
S231, generation f classification i commodity net purchase user's fraudulents buy false complaint index
Wherein,The complaint after f classification i counterfeit goods, non-favorable comment and unwitting total are bought for t phase users
Number,The total number of persons of f classification i counterfeit goods is bought for t phase users,F classifications i is bought for base period user
Complaint, non-favorable comment after counterfeit goods and unwitting total number of persons,F classifications i, which is bought, for base period user palms off quotient
The total number of persons of product;
S232, generation f classification commodity net purchase user's fraudulents buy false complaint index:
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S233, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value;
S234, the generation non-fraudulent of f classification i commodity net purchase users buy false index
Wherein,Not complaining after f classification i counterfeit goods is bought for t phase users, is not commented on and the comment of non-difference is total
Number,The total number of persons after f classification i counterfeit goods is bought for t phase users,F classes are bought for base period user
Not complaining after mesh i counterfeit goods is not commented on and non-difference comment total number of persons,It is false that f classifications i is bought for base period user
Emit the total number of persons after commodity;
S235, the generation non-fraudulent of f classification commodity net purchase users buy false index,
Wherein, n be from studied and judged for be personation and be f classifications commodity in sampling obtain commodity total quantity;
S236, generation net purchase user's fraudulent buy false complaint index:
Wherein, 19 be commodity classification quantitative value.
8. net purchase user as claimed in claim 7 buys the generation method of false exponential model, it is characterised in that:Step S31 is specific
Two themes and three main bodies classification to be carried out to buying false public sentiment data, specifically, carrying out text analyzing to buying false public sentiment data
Pretreatment, according to the once division methods of the theme in public sentiment subject classification system, using 1 grade and 2 grades themes of support vector machines pair
Automatic Text Categorization is carried out, automatic classification for search is carried out using 3 grades of descriptor of Keywords matching algorithm pair, includes the following steps,
S311, according to different types of network information classification, advance collecting part text data respectively, and carry out key words with
Abstract extraction;
S312, the mode based on man-computer cooperation, the key words to acquiring text in advance carry out two subject categories marks with abstract
Note and three principal classes mark, and constantly training SVM buys false public sentiment subject classification device, and forms svm classifier model;
S313, it after test text is carried out key words abstract extraction, is directed respectively into svm classifier model, progress buys vacation automatically
Public sentiment is classified.
9. net purchase user as claimed in claim 8 buys the generation method of false exponential model, it is characterised in that:Step S33 is specific
For,
S331, mainstream news public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z belongs to for trade name
Mark,False public sentiment event mainstream news quantity is bought for the base period,False public sentiment event mainstream news quantity is bought for the t phases, it is main
Stream news buys false public sentiment combined index and buys false public sentiment event mainstream equal to area, platform, firm name, commodity classification and trade name
News public sentiment separate index number adding up and being averaged;
S332, entrepreneur's public opinion index is generated:
Wherein, i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z belongs to for trade name
Mark,False public sentiment event entrepreneur is bought for the base period to express an opinion quantity,False public sentiment event Enterprise Human is bought for the t phases
Scholar expresses an opinion quantity, and entrepreneur buys false public sentiment combined index and is equal to area, platform, firm name, commodity classification and commodity
Title buys the cumulative and average of pseudo event public sentiment separate index number;
S333, common netizen's public opinion index is generated:
Wherein, wherein i is regional home identity, and j is platform home identity, and k is retail shop's title home identity, and z is trade name
Home identity,False public sentiment event money order receipt to be signed and returned to the sender is bought for base period common netizen, is posted, the cumulative and average value of touching quantity,
For t phases common netizen buy false public sentiment event money order receipt to be signed and returned to the sender, be posted, touching quantity it is cumulative and average, common netizen buys false public sentiment and always refers to
Number is equal to area, platform, firm name, commodity classification and trade name and buys the cumulative and average of pseudo event public sentiment separate index number.
10. a kind of net purchase of generation method for being bought false exponential model using claim 1-9 any one of them net purchase users is used
Buy the generation system of false exponential model in family, it is characterised in that:It is generated including buying false comment index generation module, buying false Activity Index
Module and false public opinion index generation module is bought, included the following steps:
Buy false comment index generation module:Obtain user's comment on commodity text data on electric business platform;User's commodity are obtained to comment
After paper notebook data, data prediction is carried out, identifies and exclude the brush comment data of the existing promise breaking in violation of rules and regulations of shop businessman, with
To validated user comment data;The identification of false comment text is bought, obtains validated user comment data and then according to user comment
The similarity algorithm of sentence come calculate and identify these user's comment on commodity data content whether be buy false comment, and count buy
False number of reviews;False comment index is bought in generation.
Buy false Activity Index generation module:Data acquisition is complained, including obtains the user's Merchandise Complaint data and line of electric business platform
Lower user's Merchandise Complaint data;The pretreatment of data after the complaint data for obtaining step S21, determines to complain in data and belongs to net
Purchase user buys false complaint;The generation of false Activity Index is bought, including f classification i commodity net purchase users fraudulent, non-fraudulent buy false row
It buys false Activity Index, net purchase user fraudulent, non-fraudulent for index, f classification commodity net purchase users fraudulent, non-fraudulent and buys
False Activity Index;
Buy false public opinion index generation module:Buy false public sentiment data acquisition, and by collected network public-opinion data into style of writing
This analysis and subject classification;It is commented into row index according to three human subjects and two principal themes, and by network public-opinion Fuzzy evaluation mode
Sentence, calculates the public opinion index of each index;The public sentiment composite index of each subject network is calculated by weighting algorithm.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810238533.1A CN108428156A (en) | 2018-03-21 | 2018-03-21 | Net purchase user buys the generation method and system of false exponential model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810238533.1A CN108428156A (en) | 2018-03-21 | 2018-03-21 | Net purchase user buys the generation method and system of false exponential model |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108428156A true CN108428156A (en) | 2018-08-21 |
Family
ID=63159311
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810238533.1A Withdrawn CN108428156A (en) | 2018-03-21 | 2018-03-21 | Net purchase user buys the generation method and system of false exponential model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108428156A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109657124A (en) * | 2018-12-14 | 2019-04-19 | 成都德迈安科技有限公司 | Public sentiment monitoring system based on consumer behaviour |
CN111311411A (en) * | 2020-02-14 | 2020-06-19 | 北京三快在线科技有限公司 | Illegal behavior identification method and device |
CN112132368A (en) * | 2019-06-06 | 2020-12-25 | 阿里巴巴集团控股有限公司 | Information processing method and device, computing equipment and storage medium |
-
2018
- 2018-03-21 CN CN201810238533.1A patent/CN108428156A/en not_active Withdrawn
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109657124A (en) * | 2018-12-14 | 2019-04-19 | 成都德迈安科技有限公司 | Public sentiment monitoring system based on consumer behaviour |
CN112132368A (en) * | 2019-06-06 | 2020-12-25 | 阿里巴巴集团控股有限公司 | Information processing method and device, computing equipment and storage medium |
CN111311411A (en) * | 2020-02-14 | 2020-06-19 | 北京三快在线科技有限公司 | Illegal behavior identification method and device |
CN111311411B (en) * | 2020-02-14 | 2022-03-08 | 北京三快在线科技有限公司 | Illegal behavior identification method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Ahmed et al. | Detection of online fake news using n-gram analysis and machine learning techniques | |
Zhang et al. | Detecting spammer groups from product reviews: a partially supervised learning model | |
Tufail et al. | The effect of fake reviews on e-commerce during and after Covid-19 pandemic: SKL-based fake reviews detection | |
Baj-Rogowska | Sentiment analysis of Facebook posts: The Uber case | |
TW201115370A (en) | Systems and methods for capturing and managing collective social intelligence information | |
Bajaj et al. | A novel user-based spam review detection | |
Markou et al. | Predicting taxi demand hotspots using automated internet search queries | |
CN108428156A (en) | Net purchase user buys the generation method and system of false exponential model | |
Cherqi et al. | Analysis of hacking related trade in the darkweb | |
Kim et al. | Text mining for the evaluation of public services: the case of a public bike-sharing system | |
Sinha et al. | Fake product review monitoring using opinion mining | |
Ishfaq et al. | Identifying the Influential Bloggers: A modular approach based on Sentiment Analysis. | |
Wang et al. | Fake Review Identification Methods Based on Multidimensional Feature Engineering | |
CN110213342A (en) | A kind of information push method of purchasing system, device, storage medium and terminal | |
Zhañay et al. | A Text Mining Approach to Discover Real-Time Transit Events from Twitter | |
Zhang et al. | Data mining algorithm for demand forecast analysis on flash sales platform | |
Singh et al. | Model for detecting fake or spam reviews | |
Alsubari et al. | Rule-based classifiers for identifying fake reviews in e-commerce: a deep learning system | |
Mei et al. | Research on e-commerce coupon user behavior prediction technology based on decision tree algorithm | |
Han et al. | User requirements dynamic elicitation of complex products from social network service | |
Kaur et al. | Multi-layered sentiment analytical model for product review mining | |
Maître et al. | Event detection and time series alignment to improve stock market forecasting | |
Gao et al. | Identification of Deceptive Reviews by Sentimental Analysis and Characteristics of Reviewers. | |
Kotriwal et al. | Deceptive reviews detection in E-commerce websites using machine learning | |
Kausar et al. | Sentiment Classification based on Machine Learning Approaches in Amazon Product Reviews |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20180821 |
|
WW01 | Invention patent application withdrawn after publication |