CN107766316A - The analysis method of evaluating data, apparatus and system - Google Patents
The analysis method of evaluating data, apparatus and system Download PDFInfo
- Publication number
- CN107766316A CN107766316A CN201610670873.2A CN201610670873A CN107766316A CN 107766316 A CN107766316 A CN 107766316A CN 201610670873 A CN201610670873 A CN 201610670873A CN 107766316 A CN107766316 A CN 107766316A
- Authority
- CN
- China
- Prior art keywords
- evaluation
- score value
- essential elements
- user
- cluster
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
Abstract
The invention provides a kind of analysis method of evaluating data, apparatus and system, belong to natural language processing field.Wherein, the analysis method of evaluating data includes:The evaluating data of user is obtained, the evaluating data includes the essential elements of evaluation of product and user evaluates score value to the first of the essential elements of evaluation;The essential elements of evaluation is clustered;Parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, the second evaluation score value of each essential elements of evaluation cluster is obtained, the second evaluation score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user.It can be obtained closer to evaluation of the user to product or the true idea of a certain feature of product by technical scheme.
Description
Technical field
The present invention relates to natural language processing field, particularly relates to a kind of analysis method of evaluating data, apparatus and system.
Background technology
Evaluation of the user to product at present is expressed typically by text, in order to more fully understand and assay
Viewpoint in text, the opining mining of essential elements of evaluation turn into the major subjects in evaluation analysis field.The opining mining of essential elements of evaluation
It is main to include two steps, the differentiation of extraction emotion tendency corresponding with its of essential elements of evaluation.
Essential elements of evaluation can be a certain feature for being evaluated product, for example, in mobile phone productses field, " battery " and " screen "
It can be essential elements of evaluation.User generally can be to evaluate score value to represent to the emotion tendency of essential elements of evaluation, such as+N usual
Front evaluation is represented, 0 represents neutral evaluation, and-N represents unfavorable ratings, wherein, N is positive integer.Analysis user to product or
During the viewpoint of a certain feature of product, the evaluation score value of the multiple users of comprehensive analysis is obviously than analyzing the evaluation score value of sole user more
It is significant.Therefore, prior art is usually to obtain evaluation score value of multiple users to same essential elements of evaluation, then is averaged value work
For the authentic assessment score value of the essential elements of evaluation.
But evaluation score value of the user to essential elements of evaluation often has certain biasing.Such as same Mobile phone
Product, user A relatively take notice of mobile phone screen, less take notice of that then user A is to the mobile phone productses for the endurance of battery of mobile phone
When being evaluated, the marking to mobile phone screen is harsher for -4 points, and the more tolerant marking to battery of mobile phone is+5 points;And use
Family B is on the contrary, compare the endurance for taking notice of battery of mobile phone, and less take notice of for mobile phone screen, then user B is to the mobile phone productses
When being evaluated, the marking to battery of mobile phone is harsher for -3 points, and the more tolerant marking to mobile phone screen is+4 points.It is aobvious
So, the arithmetic average for simply to evaluation score value of multiple users to same essential elements of evaluation calculate acquisition can not represent use
Family is to product or the authentic assessment of a certain feature of product.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of analysis method of evaluating data, apparatus and system, can obtain
Obtain closer to evaluation of the user to product or the true idea of a certain feature of product.
In order to solve the above technical problems, embodiments of the invention offer technical scheme is as follows:
On the one hand, there is provided a kind of analysis method of evaluating data, including:
The evaluating data of user is obtained, the evaluating data includes the essential elements of evaluation of product and user will to the evaluation
First evaluation score value of element;
The essential elements of evaluation is clustered;
Parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, obtains each essential elements of evaluation
Second evaluation score value of cluster, the second evaluation score value are the first evaluation score value and use of the essential elements of evaluation according to each cluster
The weighted average that the scoring weight at family obtains.
Further, the evaluating data for obtaining user specifically includes:
Capture the evaluation text of user;
The evaluation text is identified, extracts essential elements of evaluation and the user of the product that the evaluation text is included
Score value is evaluated to the first of the essential elements of evaluation, according to the first evaluation of the essential elements of evaluation of product and user to the essential elements of evaluation
Score value is to generate the evaluating data.
Further, it is described that the essential elements of evaluation is clustered, using the described first evaluation score value to each cluster
Essential elements of evaluation carries out parameter Estimation, obtains the second evaluation score value of each essential elements of evaluation cluster and includes:
The evaluating data of acquisition is divided into multiple set, the essential elements of evaluation in each set belongs to the same of identical product
Cluster;
In each set, establish using the scoring weight of the second evaluation score value of corresponding essential elements of evaluation cluster and user as ginseng
Several maximal possibility estimation equations;
Iteration second, which evaluates score value estimate and the scoring weight estimate of user, makes the maximal possibility estimation equation
Desired value is maximized to convergence, the corresponding essential elements of evaluation of set where the obtain second evaluation score value estimate is used as cluster the
Two evaluation score values.
Further, it is described to obtain also including after the second evaluation score value of each essential elements of evaluation cluster:
Filter out multiple products for including same Cluster Assessment key element;
The the second evaluation score value clustered according to the essential elements of evaluation is ranked up to the multiple product.
Further, it is described to obtain also including after the second evaluation score value of each essential elements of evaluation cluster:
The the second evaluation score value clustered according to the different evaluation key element of identical product calculates average value, obtains the product
3rd evaluation score value;
Multiple products are ranked up using the described 3rd evaluation score value.
Further, fourth evaluation score value of the evaluating data also including product, the 3rd evaluation score value of product is obtained
Also include afterwards:
Average value is calculated according to the 4th of identical product the evaluation score value and the 3rd evaluation score value, obtains the 5th of the product
Evaluate score value;
Multiple products are ranked up using the described 5th evaluation score value.
The embodiment of the present invention additionally provides a kind of analytical equipment of evaluating data, including:
Acquisition module, for obtaining the evaluating data of user, the evaluating data includes the essential elements of evaluation and use of product
Score value is evaluated to the first of the essential elements of evaluation in family;
Cluster module, for being clustered to the essential elements of evaluation;
Computing module, for carrying out parameter Estimation to the essential elements of evaluation of each cluster using the described first evaluation score value, obtain
To the second evaluation score value of each essential elements of evaluation cluster, the second evaluation score value is the according to the essential elements of evaluation of each cluster
The weighted average that the scoring weight of one evaluation score value and user obtain.
Further, described device also includes:
Handling module, for capturing the evaluation text of user;
Identification module, for the evaluation text to be identified, extract the product that the evaluation text is included
Essential elements of evaluation and user evaluate score value to the first of the essential elements of evaluation, according to the essential elements of evaluation of product and user to the evaluation
The first of key element evaluates score value to generate the evaluating data.
Further, the computing module includes:
Division unit, for the evaluating data of acquisition to be divided into multiple set, the essential elements of evaluation in each set belongs to
The same cluster of identical product;
Establishing equation unit, in each set, establishing with the second evaluation score value of corresponding essential elements of evaluation cluster and
The scoring weight of user is the maximal possibility estimation equation of parameter;
Unit is solved, the scoring weight estimate of score value estimate and user are evaluated for iteration second makes the maximum seemingly
So the desired value of estimation equation is maximized to convergence, and the obtain second evaluation score value estimate is gathered into corresponding evaluation as where
Second evaluation score value of key element cluster.
The embodiment of the present invention additionally provides a kind of analysis system of evaluating data, including:
Network interface, processor, input equipment, memory and the display device interconnected by bus architecture;Wherein,
The network interface, for being connected to network;
The input equipment, for receiving input instruction, and it is sent to computing device;
The memory, for the intermediate data in storage program area and application program, and processor calculating process;
The display device, the result for processor to be obtained are shown;
The processor, for obtaining the evaluating data of user, the evaluating data include product essential elements of evaluation and
User evaluates score value to the first of the essential elements of evaluation, and the essential elements of evaluation is clustered, and utilizes the described first evaluation score value
Parameter Estimation is carried out to the essential elements of evaluation of each cluster, obtains the second evaluation score value of each essential elements of evaluation cluster, described second
Evaluation score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user
Value.
Embodiments of the invention have the advantages that:
In such scheme, parameter Estimation is carried out to the essential elements of evaluation of each cluster using the first evaluation score value, obtained each
Second evaluation score value of essential elements of evaluation cluster, the second evaluation score value are the first evaluation score value according to the essential elements of evaluation of each cluster
The weighted average obtained with the scoring weight of user, technical scheme are not simply to calculate multiple users to same
The average value of the evaluation score value of Cluster Assessment key element, and the scoring weight of user is allowed for, added using the scoring weight of user
The evaluation score value that weight average obtains can more accurately express authentic assessment of the user to the Cluster Assessment key element.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of the analysis method of evaluating data of the embodiment of the present invention;
Fig. 2 is the structured flowchart of the analytical equipment of evaluating data of the embodiment of the present invention;
Fig. 3 is the structured flowchart of the analytical equipment of another embodiment of the present invention evaluating data;
Fig. 4 is the structured flowchart of computing module of the embodiment of the present invention;
Fig. 5 is the structured flowchart of the analysis system of evaluating data of the embodiment of the present invention;
Fig. 6 is the schematic flow sheet of the analysis method of a specific embodiment evaluating data of the invention.
Embodiment
To make embodiments of the invention technical problems to be solved, technical scheme and advantage clearer, below in conjunction with
Drawings and the specific embodiments are described in detail.
Embodiments of the invention are directed in the prior art simply to evaluation score value of multiple users to same essential elements of evaluation
Arithmetic average calculating is carried out, the arithmetic average of acquisition can not represent user and product or the true of a certain feature of product are commented
The problem of valency, there is provided a kind of analysis method of evaluating data, apparatus and system, can obtain closer to user to product or production
The evaluation of the true idea of a certain feature of product.
Embodiment one
The present embodiment provides a kind of analysis method of evaluating data, as shown in figure 1, the present embodiment includes:
Step 101:Obtain user evaluating data, the evaluating data include product essential elements of evaluation and user to institute
State the first evaluation score value of essential elements of evaluation;
Preferably, evaluating data can include the essential elements of evaluation of all products and user is commented the first of the essential elements of evaluation
Valency score value.
Step 102:The essential elements of evaluation is clustered;
Step 103:Parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, obtained each
Second evaluation score value of essential elements of evaluation cluster, the second evaluation score value are the first evaluation according to the essential elements of evaluation of each cluster
The weighted average that the scoring weight of score value and user obtain.
In the present embodiment, the biasing of user's scoring is embodied using the scoring weight of user, utilizes the first evaluation score value
Parameter Estimation is carried out to the essential elements of evaluation of each cluster, obtains the second evaluation score value of each essential elements of evaluation cluster, the second evaluation
Score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user, this
The technical scheme of invention is not simply to calculate the average value of evaluation score value of multiple users to same Cluster Assessment key element, but
In view of the scoring weight of user, the evaluation score value averagely obtained using the scoring Weight of user can be expressed more accurately
Authentic assessment of the user to the Cluster Assessment key element.
As an example, the evaluating data for obtaining user includes:
Capture the evaluation text of user;
The evaluation text is identified, extracts essential elements of evaluation and the user of the product that the evaluation text is included
Score value is evaluated to the first of the essential elements of evaluation, according to the first evaluation of the essential elements of evaluation of product and user to the essential elements of evaluation
Score value is to generate the evaluating data.
It is described that the essential elements of evaluation is clustered as an example, using the described first evaluation score value to each poly-
The essential elements of evaluation of class carries out parameter Estimation, obtains the second evaluation score value of each essential elements of evaluation cluster and includes:
The evaluating data of acquisition is divided into multiple set, the essential elements of evaluation in each set belongs to the same of identical product
Cluster;
In each set, establish using the scoring weight of the second evaluation score value of corresponding essential elements of evaluation cluster and user as ginseng
Several maximal possibility estimation equations;
Iteration second, which evaluates score value estimate and the scoring weight estimate of user, makes the maximal possibility estimation equation
Desired value is maximized to convergence, the corresponding essential elements of evaluation of set where the obtain second evaluation score value estimate is used as cluster the
Two evaluation score values.
As an example, the maximal possibility estimation equation is:
Iterative parameter is:
Wherein, rijkThe first evaluation score value for user j to product i k Cluster Assessment key elements,Will for k Cluster Assessments
The scoring weight of the user of element,For the scoring weight estimate of the user of k Cluster Assessment key elements, qikEvaluated for product i k
Second evaluation score value of key element cluster,The the second evaluation score value estimate clustered for product i k essential elements of evaluations.
Further, it is described to obtain also including after the second evaluation score value of each essential elements of evaluation cluster:
Filter out multiple products for including same Cluster Assessment key element;
The the second evaluation score value clustered according to the essential elements of evaluation is ranked up to the multiple product.
So when user plane is to multiple products including same Cluster Assessment key element, it can be selected according to above-mentioned ranking results
One in multiple products is selected, the second evaluation score value of product is better, and it is better to the Cluster Assessment factors evaluation of product to illustrate.
For example user compares concern essential elements of evaluation " battery ", and mobile phone productses A, mobile phone productses B and mobile phone productses C include same gather
The essential elements of evaluation " battery " of class, mobile phone productses A, mobile phone productses B, mobile phone productses C are carried out according to the second obtained evaluation score value
Sequence, with for reference, user can select the second evaluation score value highest product, because the second evaluation score value highest represents
The performance comparision of the Cluster Assessment key element of the product is superior.
Further, it is described to obtain also including after the second evaluation score value of each essential elements of evaluation cluster:
The the second evaluation score value clustered according to the different evaluation key element of identical product calculates average value, obtains the product
3rd evaluation score value;
Multiple products are ranked up using the described 3rd evaluation score value.
3rd evaluation score value of product is not only related to a certain Cluster Assessment key element, but comprehensive multiple essential elements of evaluations gather
The result that class obtains, product is ranked up according to the 3rd of product the evaluation score value, can so when user plane is to multiple products
Product is selected to evaluate the height of score value according to the 3rd, the 3rd evaluation score value of product is higher, illustrates the combination property of product
Better.
Further, fourth evaluation score value of the evaluating data also including product, as the first evaluation score value is evaluation
For user to the initial evaluation score value of essential elements of evaluation, the 4th evaluation score value is that user is overall to product in evaluating data in data
Initial evaluation score value, obtain also including after the 3rd evaluation score value of product:
Average value is calculated according to the 4th of identical product the evaluation score value and the 3rd evaluation score value, obtains the 5th of the product
Evaluate score value;
Multiple products are ranked up using the described 5th evaluation score value.
, can be not only according to the second evaluation of the different evaluation key element of product cluster when obtaining the evaluation score value of product
Score value calculates, and can be combined with user's initial evaluation score value overall to product to calculate the evaluation score value of product, so
The accuracy of product evaluation score value can be further lifted, the 5th evaluation score value finally given not only will with a certain Cluster Assessment
It is plain related, but comprehensive multiple essential elements of evaluation clusters and user evaluate obtained result, according to the 5th of product the evaluation score value pair
Product is ranked up, and so when user plane is to multiple products, can be selected product according to the height of the 5th evaluation score value, be produced
5th evaluation score value of product is higher, illustrates that the combination property of product is better.
Embodiment two
The present embodiment additionally provides a kind of analytical equipment of evaluating data, as shown in Fig. 2 the present embodiment includes:
Acquisition module 21, for obtaining the evaluating data of user, the evaluating data include product essential elements of evaluation and
User evaluates score value to the first of the essential elements of evaluation;
Cluster module 22, for being clustered to the essential elements of evaluation;
Computing module 23, for carrying out parameter Estimation to the essential elements of evaluation of each cluster using the described first evaluation score value,
The second evaluation score value of each essential elements of evaluation cluster is obtained, the second evaluation score value is the essential elements of evaluation according to each cluster
The weighted average that the scoring weight of first evaluation score value and user obtain.
In the present embodiment, the biasing of user's scoring is embodied using the scoring weight of user, utilizes the first evaluation score value
Parameter Estimation is carried out to the essential elements of evaluation of each cluster, obtains the second evaluation score value of each essential elements of evaluation cluster, the second evaluation
Score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user, this
The technical scheme of invention is not simply to calculate the average value of evaluation score value of multiple users to same Cluster Assessment key element, but
In view of the scoring weight of user, the evaluation score value averagely obtained using the scoring Weight of user can be expressed more accurately
Authentic assessment of the user to the Cluster Assessment key element.
Further, as shown in Fig. 2 the analytical equipment of evaluating data also includes:
Input module 20, for providing the evaluating data of user to acquisition module 21;
Output module 24, for exporting the result of calculation of computing module 23.
Further, as shown in figure 3, described device also includes:
Handling module 25, for capturing the evaluation text of user;
Identification module 26, for the evaluation text to be identified, extract the product that the evaluation text is included
Essential elements of evaluation and user evaluate score value to the first of the essential elements of evaluation, according to the essential elements of evaluation of product and user to institute's commentary
The first of valency key element evaluates score value to generate the evaluating data.
Further, as shown in figure 4, the computing module 23 includes:
Division unit 231, for the evaluating data of acquisition to be divided into multiple set, the essential elements of evaluation category in each set
In the same cluster of identical product;
Establishing equation unit 232, in each set, establishing the second evaluation score value with corresponding essential elements of evaluation cluster
Scoring weight with user is the maximal possibility estimation equation of parameter;
Solve unit 233, for iteration second evaluate score value estimate and user scoring weight estimate make it is described most
The desired value of maximum-likelihood estimation equation is maximized to convergence, and using the obtain second evaluation score value estimate, the set as where is corresponding
Second evaluation score value of essential elements of evaluation cluster.
Embodiment three
The present embodiment additionally provides a kind of analysis system 50 of evaluating data, as shown in figure 5, the present embodiment includes:
Network interface 51, processor 52, input equipment 53, memory 54, hard disk 55 and display device 56.It is above-mentioned each
It can be interconnected between interface and equipment by bus architecture.Bus architecture can be the bus that can include any number of interconnection
And bridge.One or more central processing unit (CPU) specifically represented by processor 52, and one represented by memory 54
Or the various of multiple memories are electrically connected to together.Bus architecture can also will such as ancillary equipment, voltage-stablizer and power
The various other of management circuit or the like are electrically connected to together.It is appreciated that bus architecture is used to realize between these components
Connection communication.Bus architecture is in addition to including data/address bus, in addition to power bus, controlling bus and status signal bus in addition,
These are all it is known in the art, being therefore no longer described in greater detail herein.
The network interface 51, network (such as internet, LAN) can be connected to, dependency number is obtained from network
According to, such as the evaluation text of user, and can be stored in hard disk 55.
The input equipment 53, can receive the various instructions of operating personnel's input, and be sent to processor 52 for holding
OK.The input equipment 53 can include keyboard or pointing device (for example, mouse, trace ball (trackball), touch-sensitive plate
Or touch-screen etc..
The display device 56, the result that the execute instruction of processor 52 obtains can be shown.The display device
56 can include display, projector equipment etc..
The memory 54, program and data necessary to being run for storage program area, and processor 52 calculate
During the data such as intermediate result.
It is appreciated that the memory 54 in the embodiment of the present invention can be volatile memory or nonvolatile memory,
Or it may include both volatibility and nonvolatile memory.Wherein, nonvolatile memory can be read-only storage (ROM),
Programmable read only memory (PROM), Erasable Programmable Read Only Memory EPROM (EPROM), Electrically Erasable Read Only Memory
Or flash memory (EEPROM).Volatile memory can be random access memory (RAM), and it is used as External Cache.Herein
The memory 54 of the apparatus and method of description is intended to the memory of including but not limited to these and any other suitable type.
In some embodiments, memory 54 stores following element, can perform module either data structure or
Their subset, or their superset:Operating system 541 and application program 542.
Wherein, operating system 541, comprising various system programs, such as ccf layer, core library layer, driving layer etc., for reality
The hardware based task of existing various basic businesses and processing.Application program 542, include various application programs, such as browser
(Browser) etc., for realizing various applied business.Realize that the program of present invention method may be embodied in application program
In 542.
Above-mentioned processor 52, when calling and perform the application program and data that are stored in the memory 54, specifically,
When can be the program or instruction that are stored in application program 542, it can be used for the evaluating data for obtaining user, the evaluating data
The essential elements of evaluation and user for including product evaluate score value to the first of the essential elements of evaluation, and the essential elements of evaluation is gathered
Class, parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, obtains each essential elements of evaluation cluster
The second evaluation score value, the second evaluation score value is according to the first evaluation score value of the essential elements of evaluation of each cluster and user
The weighted average that scoring weight obtains.
The method that the above embodiment of the present invention discloses can apply in processor 52, or be realized by processor 52.Place
It is probably a kind of IC chip to manage device 52, has the disposal ability of signal.In implementation process, each step of the above method
It can be completed by the integrated logic circuit of the hardware in processor 52 or the instruction of software form.Above-mentioned processor 52 can
To be general processor, digital signal processor (DSP), application specific integrated circuit (ASIC), ready-made programmable gate array (FPGA)
Either other PLDs, discrete gate or transistor logic, discrete hardware components, it is possible to achieve or perform
Disclosed each method, step and logic diagram in the embodiment of the present invention.General processor can be microprocessor or this at
It can also be any conventional processor etc. to manage device.The step of method with reference to disclosed in the embodiment of the present invention, can directly embody
Completion is performed for hardware decoding processor, or completion is performed with the hardware in decoding processor and software module combination.Software
Module can be located at random access memory, flash memory, read-only storage, programmable read only memory or electrically erasable programmable storage
In the ripe storage medium in this areas such as device, register.The storage medium is located at memory 54, and processor 52 reads memory 54
In information, with reference to its hardware complete the above method the step of.
It is understood that embodiments described herein can use hardware, software, firmware, middleware, microcode or its
Combine to realize.Realized for hardware, processing unit can be realized in one or more application specific integrated circuits (ASIC), numeral letter
Number processor DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), general processor, controller, microcontroller, microprocessor, other electronics lists for performing herein described function
In member or its combination.
Realize, can be realized herein by performing the module (such as process, function etc.) of function described herein for software
Described technology.Software code is storable in memory and passes through computing device.Memory can within a processor or
Realized outside processor.
Further, processor 52 is additionally operable to capture the evaluation text of user;The evaluation text is identified, extracted
The essential elements of evaluation and user for going out the product that the evaluation text is included evaluate score value to generate to the first of the essential elements of evaluation
The evaluating data.
Specifically, the evaluating data of acquisition is divided into multiple set by processor 52, the essential elements of evaluation category in each set
In the same cluster of identical product;In each set, establish with the second evaluation score value of corresponding essential elements of evaluation cluster and user
Scoring weight be parameter maximal possibility estimation equation;Iteration second evaluates score value estimate and the scoring weight estimation of user
Value makes the desired value of the maximal possibility estimation equation maximize to convergence, using the obtain second evaluation score value estimate as institute
In the second evaluation score value of the corresponding essential elements of evaluation cluster of set.
Alternatively, processor 52 filters out multiple products for including same Cluster Assessment key element;According to the essential elements of evaluation
Second evaluation score value of cluster is ranked up to the multiple product.
Alternatively, the second evaluation score value that processor 52 clusters according to the different evaluation key element of identical product calculates average
Value, obtain the 3rd evaluation score value of the product;Multiple products are ranked up using the described 3rd evaluation score value.
Alternatively, processor 52 calculates average value according to the 4th of identical product the evaluation score value and the 3rd evaluation score value, obtains
The 5th to the product evaluates score value;Multiple products are ranked up using the described 5th evaluation score value.
In the present embodiment, the biasing of user's scoring is embodied using the scoring weight of user, utilizes the first evaluation score value
Parameter Estimation is carried out to the essential elements of evaluation of each cluster, obtains the second evaluation score value of each essential elements of evaluation cluster, the second evaluation
Score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user, this
The technical scheme of invention is not simply to calculate the average value of evaluation score value of multiple users to same Cluster Assessment key element, but
In view of the scoring weight of user, the evaluation score value averagely obtained using the scoring Weight of user can be expressed more accurately
Authentic assessment of the user to the Cluster Assessment key element.
Example IV
Specifically, as shown in fig. 6, the analysis method of the evaluating data of the present embodiment specifically includes following steps:
Step 601:Capture the evaluation text of user;
Specifically, the evaluation text of user can be captured from multiple enriched data sources, data source includes but is not limited to produce
Product information page, editorial, blog articles, product news and forum etc., the evaluation text of user include certain of product or product
One feature, i.e. essential elements of evaluation, in addition to evaluation content and emotion tendency to essential elements of evaluation, the emotion tendency can be with tables
The now initial evaluation score value (i.e. above-mentioned first evaluation score value) for user to essential elements of evaluation.In addition, evaluating text can be with
Including other information, such as user profile, temporal information, evaluation serviceability information etc..
Exemplarily, the user evaluation text of crawl is:" the relatively good use of button of mobile phone, is quick on the draw.It is but electric
Pond heating problem is than more serious, and RAM card can not be general, does not recommend to buy.Corresponding star is evaluated as RAM card -5, is
System fluency+4, operability+1 ".Wherein ,+N represents front evaluation, and 0 represents neutral evaluation, and-N represents unfavorable ratings, wherein, N
For positive integer.In viewpoint of the analysis user to essential elements of evaluation, the obvious score of initial evaluation score value of the multiple users of comprehensive analysis
The initial evaluation score value for analysing sole user is more meaningful.Therefore, user as much as possible should be captured and evaluate text to be analyzed.
Step 602:Evaluation text to user is identified, extract product essential elements of evaluation and user to essential elements of evaluation
Initial evaluation score value;
Evaluation text to user carries out essential elements of evaluation analysis, and the evaluating data extracted can be that { product, evaluation will
Element, emotion tendency } or { essential elements of evaluation, emotion tendency }, wherein product and essential elements of evaluation is noun, and emotion tendency is one
Individual score value, its span are { -1,0 ,+1 } or { -5, -4 ...+4 ,+5 } etc., and the implication of emotion tendency is positive emotion, are born
Face emotion and neutral emotion (also referred to as mixed feeling), such as in { -1,0 ,+1 } ,+1 represents positive emotion, and -1 represents negative emotion,
0 represents neutral emotion;In { -5, -4 ...+4 ,+5 }, -5 represent extreme negative emotion, and -1 represents slight negative emotion etc..This implementation
Above-mentioned score value is referred to as initial evaluation score value of the user to essential elements of evaluation by example.
Wherein, the method that essential elements of evaluation is extracted from the evaluation text of user, includes but is not limited to:Based on sequence labelling
Method, the method based on topic model, the method based on dictionary, method based on syntax etc..Taken out from the evaluation text of user
Method of the family to the initial evaluation score value of essential elements of evaluation is taken, is included but is not limited to:The method of supervised learning, the side based on dictionary
Method, method based on document sets etc..
Step 603:Essential elements of evaluation is clustered;
When the simulation authentic assessment score value (i.e. above-mentioned second evaluation score value) to essential elements of evaluation is analyzed, first have to by
The essential elements of evaluation extracted is clustered, and each Cluster Assessment key element belongs to identical product, and the then evaluation to each cluster will
The simulation authentic assessment score value of element is analyzed by cluster.For example the essential elements of evaluation of the mobile phone productses extracted includes " battery
Endurance " and " battery-heating problem ", then according to application purpose, " battery durable ability " can be established and " battery-heating is asked
The two small clusters of topic ", further, can also be equal by the essential elements of evaluation of " battery durable ability ", " battery-heating problem "
It is grouped under " battery " this big cluster.Exemplarily, digital camera essential elements of evaluation cluster system be " outward appearance ", " battery ",
" annex ", " image quality ", " camera lens ", " feature ", " operability ", " cost performance " and " storage card ".
The biasing (the scoring weight of i.e. above-mentioned user) evaluated due to user essential elements of evaluation is typically implemented in product
In a certain feature, rather than on all products.For example mobile phone and digital camera belong to electronic product, evaluation all occur will
Element cluster:" battery ", " outward appearance ", " operability " etc., user can be in the evaluations of same cluster when being given a mark to essential elements of evaluation
Occur identical biasing in key element, rather than biasing property is embodied on all products, such as the heavier electricity regarding electronic product of user A
Pond performance, then when user A gives a mark to " battery " of mobile phone sum code-phase machine, it may appear that identical biases, rather than opponent
There is identical biasing in machine or digital camera the two products.Therefore, essential elements of evaluation is clustered, according to cluster to evaluation
The simulation authentic assessment score value of key element is analyzed, and can obtain user to product or the authentic assessment of a certain feature of product.
Wherein, the method clustered to essential elements of evaluation, is included but are not limited to:Method based on field priori,
Method based on topic model, general Text Clustering Method such as K-Means algorithms etc..Result after cluster can regard multiple as
Set, each set include multiple evaluating datas, and the essential elements of evaluation of evaluating data belongs to the same of identical product in each set
Cluster.
Step 604:Calculate the simulation authentic assessment score value of each essential elements of evaluation cluster.
User ujTo product oiEssential elements of evaluation akInitial evaluation score value be denoted as rijk, rijkOne is modeled as with " simulation
Authentic assessment score value " qikWith the scoring weight of userFor the stochastic equation of parameter.
Present embodiment assumes that rijkNormal Distribution:
Then parameter q (qik) and σ (σjk) maximal possibility estimation equation be:
The solution of equation, iterative parameterWithSo that above-mentioned desired value maximizes until the convergence, " simulation that will be obtained
Authentic assessment score value " qikEstimateSimulation authentic assessment score value as essential elements of evaluation cluster.
Wherein,For all users a is clustered in essential elements of evaluationkThe weighted average of upper marking:
Meanwhile the scoring weight of user is the standardization of its biasing, it is presented as initial score score value relative to " truly commenting
The variance of point score value ":
In above-mentioned parameter, set ri*kAll users are represented to product oiEssential elements of evaluation cluster akMarking;Set r*jkGeneration
All user u of tablejU is clustered in essential elements of evaluationjOn marking,For the scoring weight estimation of the user of k Cluster Assessment key elements
Value.
By above-mentioned steps 601-604, you can obtain the simulation authentic assessment score value of each essential elements of evaluation cluster.
Further, can be true by simulation after the simulation authentic assessment score value of each essential elements of evaluation cluster is drawn
Score value is evaluated to be applied in following scene:
Scene one, multiple products for including same Cluster Assessment key element are filtered out, the simulation clustered according to the essential elements of evaluation
Authentic assessment score value is ranked up to multiple products.
For example mobile phone productses A, mobile phone productses B and mobile phone productses C include the essential elements of evaluation " battery " of same cluster, then
Simulation authentic assessment point of the user to above-mentioned mobile phone productses " battery " can be calculated respectively by above-mentioned steps 601-604
Value, can be ranked up according to obtained simulation authentic assessment score value to mobile phone productses A, mobile phone productses B, mobile phone productses C, for
User refers to.
Further, for example mobile phone productses A, mobile phone productses B and digital camera product D include the evaluation of same cluster
Key element " battery ", then mould of the user to above-mentioned electronic product " battery " can be calculated respectively by above-mentioned steps 601-604
Intend authentic assessment score value, mobile phone productses A, mobile phone productses B, digital camera can be produced according to obtained simulation authentic assessment score value
Product D is ranked up, with for reference.
Scene two, the simulation authentic assessment score value clustered according to the different evaluation key element of identical product calculate average value, obtain
To the simulation authentic assessment score value of product, multiple products are ranked up according to the simulation authentic assessment score value of product.
For example mobile phone productses A multiple Cluster Assessment key elements include " screen ", " battery " and " operation fluency ", then lead to
Crossing above-mentioned steps 601-604, the simulation authentic assessment score value that user clusters to these three essential elements of evaluations can be calculated respectively,
Arithmetic average is carried out to these three cluster simulation authentic assessment score values or weighted average, the simulation for obtaining mobile phone productses A are truly commented
Valency score value.Similarly, mobile phone productses B and mobile phone productses C simulation authentic assessment score value can be obtained, according to obtaining the simulation of product
Authentic assessment score value is ranked up to mobile phone productses A, mobile phone productses B, mobile phone productses C, with for reference.
Specifically, arithmetic average can be calculated by below equation:
Weighted average can be calculated by below equation:
Wherein, set qi*Representative products oiSimulation authentic assessment score value in each Cluster Assessment key element.
Scene three, initial evaluation score value and simulation authentic assessment score value calculating average value according to identical product, are produced
The mixing evaluation score value of product, multiple products are ranked up according to the mixing evaluation score value of product.
For example after mobile phone productses A, mobile phone productses B, mobile phone productses C simulation authentic assessment score value is obtained, it can combine
User to mobile phone productses A, mobile phone productses B, mobile phone productses C initial evaluation score value, by calculating average value, obtain mobile phone productses
A, mobile phone productses B, mobile phone productses C mixing evaluation score value, further according to mixing evaluation score value to mobile phone productses A, mobile phone productses B,
Mobile phone productses C is ranked up, with for reference.
Described above is the preferred embodiment of the present invention, it is noted that for those skilled in the art
For, on the premise of principle of the present invention is not departed from, some improvements and modifications can also be made, these improvements and modifications
It should be regarded as protection scope of the present invention.
Claims (10)
- A kind of 1. analysis method of evaluating data, it is characterised in that including:Obtain user evaluating data, the evaluating data include product essential elements of evaluation and user to the essential elements of evaluation First evaluation score value;The essential elements of evaluation is clustered;Parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, obtains each essential elements of evaluation cluster The second evaluation score value, the second evaluation score value is according to the first evaluation score value of the essential elements of evaluation of each cluster and user The weighted average that scoring weight obtains.
- 2. the analysis method of evaluating data according to claim 1, it is characterised in that the evaluating data for obtaining user Specifically include:Capture the evaluation text of user;The evaluation text is identified, extracts the essential elements of evaluation for evaluating the product that text is included and user to institute State the first of essential elements of evaluation and evaluate score value to generate the evaluating data.
- 3. the analysis method of evaluating data according to claim 1, it is characterised in that described to be carried out to the essential elements of evaluation Cluster, parameter Estimation is carried out to the essential elements of evaluation of each cluster using the described first evaluation score value, each essential elements of evaluation is obtained and gathers Second evaluation score value of class includes:The evaluating data of acquisition is divided into multiple set, the essential elements of evaluation in each set belongs to the same poly- of identical product Class;In each set, establish using the scoring weight of the second evaluation score value of corresponding essential elements of evaluation cluster and user as parameter Maximal possibility estimation equation;The evaluation score value estimate of iteration second and the scoring weight estimate of user make the target of the maximal possibility estimation equation Value is maximized to restraining, and the obtain second evaluation score value estimate is gathered into corresponding essential elements of evaluation cluster as where second is commented Valency score value.
- 4. the analysis method of the evaluating data according to any one of claim 1-3, it is characterised in that it is described obtain it is each Also include after second evaluation score value of essential elements of evaluation cluster:Filter out multiple products for including same Cluster Assessment key element;The the second evaluation score value clustered according to the essential elements of evaluation is ranked up to the multiple product.
- 5. the analysis method of the evaluating data according to any one of claim 1-3, it is characterised in that it is described obtain it is each Also include after second evaluation score value of essential elements of evaluation cluster:The the second evaluation score value clustered according to the different evaluation key element of identical product calculates average value, obtains the 3rd of the product Evaluate score value;Multiple products are ranked up using the described 3rd evaluation score value.
- 6. the analysis method of evaluating data according to claim 5, it is characterised in that the evaluating data also includes product The 4th evaluation score value, obtain product the 3rd evaluation score value after also include:Average value is calculated according to the 4th of identical product the evaluation score value and the 3rd evaluation score value, obtains the 5th evaluation of the product Score value;Multiple products are ranked up using the described 5th evaluation score value.
- A kind of 7. analytical equipment of evaluating data, it is characterised in that including:Acquisition module, for obtaining the evaluating data of user, the evaluating data includes essential elements of evaluation and the user couple of product First evaluation score value of the essential elements of evaluation;Cluster module, for being clustered to the essential elements of evaluation;Computing module, for carrying out parameter Estimation to the essential elements of evaluation of each cluster using the described first evaluation score value, obtain every Second evaluation score value of one essential elements of evaluation cluster, the second evaluation score value are commented for first of the essential elements of evaluation according to each cluster The weighted average that the scoring weight of valency score value and user obtain.
- 8. the analytical equipment of evaluating data according to claim 7, it is characterised in that described device also includes:Handling module, for capturing the evaluation text of user;Identification module, for the evaluation text to be identified, extract the evaluation for the product that the evaluation text is included Key element and user evaluate score value to the first of the essential elements of evaluation, according to the essential elements of evaluation of product and user to the essential elements of evaluation The first evaluation score value to generate the evaluating data.
- 9. the analytical equipment of evaluating data according to claim 7, it is characterised in that the computing module includes:Division unit, for the evaluating data of acquisition to be divided into multiple set, the essential elements of evaluation in each set belongs to same The same cluster of product;Establishing equation unit, in each set, establishing with the second evaluation score value of corresponding essential elements of evaluation cluster and user Scoring weight be parameter maximal possibility estimation equation;Unit is solved, the scoring weight estimate of score value estimate and user are evaluated for iteration second estimates the maximum likelihood The desired value of meter equation is maximized to convergence, and the obtain second evaluation score value estimate is gathered into corresponding essential elements of evaluation as where Second evaluation score value of cluster.
- A kind of 10. analysis system of evaluating data, it is characterised in that including:Pass through the network interface of bus architecture interconnection, processing Device, input equipment, memory and display device;Wherein,The network interface, for being connected to network;The input equipment, for receiving input instruction, and it is sent to computing device;The memory, for the intermediate data in storage program area and application program, and processor calculating process;The display device, the result for processor to be obtained are shown;The processor, for obtaining the evaluating data of user, the evaluating data includes essential elements of evaluation and the user of product Score value is evaluated to the first of the essential elements of evaluation, the essential elements of evaluation is clustered, using the described first evaluation score value to every The essential elements of evaluation of one cluster carries out parameter Estimation, obtains the second evaluation score value of each essential elements of evaluation cluster, second evaluation Score value is the weighted average obtained according to the first evaluation score value of the essential elements of evaluation of each cluster and the scoring weight of user.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610670873.2A CN107766316B (en) | 2016-08-15 | 2016-08-15 | Evaluation data analysis method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610670873.2A CN107766316B (en) | 2016-08-15 | 2016-08-15 | Evaluation data analysis method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107766316A true CN107766316A (en) | 2018-03-06 |
CN107766316B CN107766316B (en) | 2021-03-30 |
Family
ID=61259906
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610670873.2A Active CN107766316B (en) | 2016-08-15 | 2016-08-15 | Evaluation data analysis method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107766316B (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109447767A (en) * | 2018-11-26 | 2019-03-08 | 重庆电子工程职业学院 | A kind of commodity evaluation method and system applied to e-commerce |
CN109559020A (en) * | 2018-11-08 | 2019-04-02 | 南京邮电大学 | A kind of quality testing method mutually commented based on colleague |
CN111325475A (en) * | 2020-03-04 | 2020-06-23 | 国网江苏省电力有限公司扬州供电分公司 | Emergency repair work order evaluation factor analysis method based on total log-likelihood algorithm |
CN111581975A (en) * | 2020-05-09 | 2020-08-25 | 北京明朝万达科技股份有限公司 | Case writing text processing method and device, storage medium and processor |
CN112036157A (en) * | 2020-08-04 | 2020-12-04 | 林树 | Foundation manager tone text analysis method and device |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030033280A1 (en) * | 2001-05-18 | 2003-02-13 | Van Den Hamer Peter | Self-descriptive data tag |
US20030167245A1 (en) * | 2002-01-31 | 2003-09-04 | Communications Research Laboratory, Independent Administrative Institution | Summary evaluation apparatus and method, and computer-readable recording medium in which summary evaluation program is recorded |
CN101572820A (en) * | 2009-05-11 | 2009-11-04 | 宁波海视智能系统有限公司 | Preprocessing method of video signal in detection process of moving target |
US20110314031A1 (en) * | 2010-03-29 | 2011-12-22 | Ebay Inc. | Product category optimization for image similarity searching of image-based listings in a network-based publication system |
CN103530318A (en) * | 2007-01-05 | 2014-01-22 | 雅虎公司 | Clustered search processing |
US20140181100A1 (en) * | 2005-09-14 | 2014-06-26 | Millennial Media, Inc. | Predictive Text Completion For A Mobile Communication Facility |
CN105354208A (en) * | 2015-09-21 | 2016-02-24 | 江苏讯狐信息科技有限公司 | Big data information mining method |
CN105740434A (en) * | 2016-02-01 | 2016-07-06 | 腾讯科技(深圳)有限公司 | Network information scoring method and device |
CN105809379A (en) * | 2014-12-30 | 2016-07-27 | 阿里巴巴集团控股有限公司 | Logistics branch evaluation method, device and electronic device |
-
2016
- 2016-08-15 CN CN201610670873.2A patent/CN107766316B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030033280A1 (en) * | 2001-05-18 | 2003-02-13 | Van Den Hamer Peter | Self-descriptive data tag |
US20030167245A1 (en) * | 2002-01-31 | 2003-09-04 | Communications Research Laboratory, Independent Administrative Institution | Summary evaluation apparatus and method, and computer-readable recording medium in which summary evaluation program is recorded |
US20140181100A1 (en) * | 2005-09-14 | 2014-06-26 | Millennial Media, Inc. | Predictive Text Completion For A Mobile Communication Facility |
CN103530318A (en) * | 2007-01-05 | 2014-01-22 | 雅虎公司 | Clustered search processing |
CN101572820A (en) * | 2009-05-11 | 2009-11-04 | 宁波海视智能系统有限公司 | Preprocessing method of video signal in detection process of moving target |
US20110314031A1 (en) * | 2010-03-29 | 2011-12-22 | Ebay Inc. | Product category optimization for image similarity searching of image-based listings in a network-based publication system |
CN105809379A (en) * | 2014-12-30 | 2016-07-27 | 阿里巴巴集团控股有限公司 | Logistics branch evaluation method, device and electronic device |
CN105354208A (en) * | 2015-09-21 | 2016-02-24 | 江苏讯狐信息科技有限公司 | Big data information mining method |
CN105740434A (en) * | 2016-02-01 | 2016-07-06 | 腾讯科技(深圳)有限公司 | Network information scoring method and device |
Non-Patent Citations (2)
Title |
---|
YANG DAO-JIAN 等: "" Evaluation of Township Competitiveness in China: A Case Study of Zhenjiang City"", 《ASIAN AGRICULTURAL RESEARCH》 * |
郑燕林 等: ""大数据在美国教育评价中的应用路径分析"", 《中国电化教育》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109559020A (en) * | 2018-11-08 | 2019-04-02 | 南京邮电大学 | A kind of quality testing method mutually commented based on colleague |
CN109447767A (en) * | 2018-11-26 | 2019-03-08 | 重庆电子工程职业学院 | A kind of commodity evaluation method and system applied to e-commerce |
CN111325475A (en) * | 2020-03-04 | 2020-06-23 | 国网江苏省电力有限公司扬州供电分公司 | Emergency repair work order evaluation factor analysis method based on total log-likelihood algorithm |
CN111581975A (en) * | 2020-05-09 | 2020-08-25 | 北京明朝万达科技股份有限公司 | Case writing text processing method and device, storage medium and processor |
CN112036157A (en) * | 2020-08-04 | 2020-12-04 | 林树 | Foundation manager tone text analysis method and device |
Also Published As
Publication number | Publication date |
---|---|
CN107766316B (en) | 2021-03-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107766316A (en) | The analysis method of evaluating data, apparatus and system | |
Lund et al. | Likelihood ratio as weight of forensic evidence: a closer look | |
Movahedi et al. | Design and perceptual validation of performance measures for salient object segmentation | |
CN108009228A (en) | A kind of method to set up of content tab, device and storage medium | |
CN108921710A (en) | The method and system of medical insurance abnormality detection | |
CN107463953A (en) | Image classification method and system based on quality insertion in the case of label is noisy | |
CN108319888B (en) | Video type identification method and device and computer terminal | |
CN108961087A (en) | Insure recommended method, device, computer equipment and computer readable storage medium | |
CN107423278A (en) | The recognition methods of essential elements of evaluation, apparatus and system | |
CN104850663A (en) | System and method for inputting and analyzing test scores | |
CN106919551A (en) | A kind of analysis method of emotion word polarity, device and equipment | |
CN112989087B (en) | Image processing method, device and computer readable storage medium | |
CN110362601A (en) | Mapping method, device, equipment and the storage medium of metadata standard | |
CN104199838B (en) | A kind of user model constructing method based on label disambiguation | |
US20210097296A1 (en) | Data processing methods, apparatuses and storage media | |
CN110175990A (en) | Quality of human face image determination method, device and computer equipment | |
CN109215131A (en) | The driving method and device of conjecture face | |
CN105740434B (en) | Network information methods of marking and device | |
CN110489593A (en) | Topic processing method, device, electronic equipment and the storage medium of video | |
CN103529939B (en) | A kind of man-machine interaction method based on color, computer vision and stress and strain model | |
CN106599671B (en) | Method and device for identifying picture based on user filling verification code | |
CN108021941A (en) | Use in medicament-induced hepatotoxicity Forecasting Methodology and device | |
Shen et al. | Predicting usefulness of Yelp reviews with localized linear regression models | |
CN107644102A (en) | Data characteristics building method and device, storage medium, electronic equipment | |
CN105488061B (en) | A kind of method and device of verify data validity |
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 |